You are on page 1of 149

Instrumentation Volume 2

Improving Plant Operation,


Safety and Control
Table of Contents
Sampling Particulate Materials the Right Way...........................................................................................................4
To obtain a representative sample for particle size characterization, adhere to the golden rules of sampling and follow these
best practices

The Direct Integration Method: A Best Practice for Relief Valve Sizing....................................................................12
The approach described here is easier to use, and provides more-accurate results, compared to leading valve-sizing methodologies

Engineering for Plant Safety........................................................................................................................................15


Early process-hazards analyses can lead to potential cost savings in project and plant operations

Managing SIS Process Measurement Risk and Cost.................................................................................................24


With a focus on flowmeters, this article shows how advances in measurement technologies help safety system designers
reduce risk and cost in their safety instrumented systems (SIS) design and lifecycle management

Column Instrumentation Basics..................................................................................................................................32


An understanding of instrumentation is valuable in evaluating and troubleshooting column performance

Control Valve Position Sensors....................................................................................................................................40

Control Valve Performance..........................................................................................................................................41

Common Mistakes When Conducting a HAZOP and How to Avoid Them...............................................................42


An important part of ensuring the success of a HAZOP study is to understand the errors that can cause the team to lose focus

Chemical Process Plants: Plan for Revamps..............................................................................................................47


Follow this guidance to make the most of engineering upgrades that are designed to improve plant operations or boost
throughput capacity

Point-Level Switches for Safety Systems...................................................................................................................53


Industries that manufacture or store potentially hazardous materials need to employ point-level switches to protect people
and the environment from spills

Control Strategies Based On Realtime Particle Size Analysis...................................................................................58


Practical experience illustrates how to achieve better process control

Process Hazards Analysis Methods.............................................................................................................................62

Aging Relief Systems — Are they Working Properly?................................................................................................63


Common problems, cures and tips to make sure your pressure relief valves operate properly when needed

Overpressure Protection: Consider Low Temperature Effects in Design..................................................................69


Understanding the inherent limitations of current over-pressure protection analyses is key to developing a more robust heuristic

Things You Need to Know Before Using an Explosion-Protection Technique..........................................................73


Understanding the different classification methods is necessary to better select the explosion-protection techniques that will be used

Cybersecurity Defense for Industrial Process- Control Systems..............................................................................78


Security techniques widely used in information technology (IT) require special considerations to be useful in operational settings.
Here are several that should get closer attention

Plant Functional Safety Requires IT Security.............................................................................................................84


Cybersecurity is critical for plant safety. Principles developed for plant safety can be applied to the security of IT systems

Dilute-phase Pneumatic Conveying: Instrumentation and Conveying Velocity......................................................91


Follow these guidelines to design a well-instrumented and controlled system, and to optimize its conveying velocity

Alarm Management By the Numbers.........................................................................................................................94


Deeper understanding of common alarm-system metrics can improve remedial actions and result in a safer plant

Understand and Cure High Alarm Rates...................................................................................................................100


Alarm rates that exceed an operator’s ability to manage them are common. This article explains the causes for high alarm
rates and how to address them
Wireless Communication in Hazardous Areas.........................................................................................................105
Consider these criteria in deciding where wireless fits in today’s CPI plants and the explosive atmospheres that permeate them

Piping-System Leak Detection and Monitoring for the CPI.....................................................................................110


Eliminating the potential for leaks is an integral part of the design process that takes place at the very onset of facility design

Monitoring Flame Hazards In Chemical Plants........................................................................................................117


The numerous flame sources in CPI facilities necessitate the installation of advanced flame-detection technologies

Integrated Risk-Management Matrices.....................................................................................................................121


An overview of the tools available to reliability professionals for making their organization the best-in-class

Process Safety and Functional Safety in Support of Asset Productivity and Integrity.........................................126
Approaches to plant safety continue to evolve based on lessons learned, as well as new automation standards and technology

Improving the Operability of Process Plants............................................................................................................131


Turndown and rangeability have a big impact on the flexibility and efficiency of chemical process operations

Solids Discharge: Characterizing Powder and Bulk Solids Behavior.....................................................................138


How shear-cell testing provides a basis for predicting flow behavior

Advantages Gained in Automating Industrial Wastewater Treatment Plants........................................................142


Process monitoring and automation can improve efficiencies in wastewater treatment systems. A number of parameters
well worth monitoring, as well as tips for implementation are described
Feature Report

Sampling Particulate
Materials the Right Way
Remi Trottier and
Shrikant Dhodapkar To obtain a representative sample for particle size
The Dow Chemical Company
characterization, adhere to the golden rules of

I
n the chemical process industries
(CPI) it is often necessary to verify sampling and follow these best practices
material specification at various
points in the process. In that effort,
it is usually impossible — or at the analysis. In this example, we can all sample that has a distribution of par-
very least impractical — to measure appreciate the two main, necessary ticle sizes. This notion that individual
the whole production. Instead, small characteristics of the sample: particles are not identical is referred
samples must be extracted from a 1. That a relatively small sample is to as constitutional heterogeneity
parent population. Such is the case in taken (much smaller than the total (CH). The higher the upper end of the
particle size characterization of bulk available) distribution, the higher the hetero-
solids, process streams and slurries. 2. That the sample be representative geneity. The Gy sampling theory can
While truly representative sam- of the whole (so that the correct di- estimate the variance of this funda-
pling has long been an important goal, agnosis can be made) mental sampling error due to the CH,
a number of current trends are driving Although both points are extremely using Equation (1), [2]:
the incentive for rapid implementation simple concepts, a great deal of dili-
of top-notch sampling strategies to be gence is usually necessary to achieve
the standard, rather than the excep- them. Careless sampling of powders
tion. These trends include the ever-in- or slurries often results in a faulty
creasing demand for superior material conclusion, regardless of whether good  (1)
quality in the high-technology indus- analytical techniques are employed. In
tries, more-stringent pharmaceutical that respect, the first item that should Where MS is the mass of the sample,
regulations and higher environmental be considered for a particle-character- ML is the mass of the parent popula-
standards, to name a few. ization study is a sampling protocol tion from which the sample is taken,
Unfortunately many sampling that insures a representative sample ƒ is a shape factor (0.5 for spheres, 1
strategies in use today do not take of the proper size. for cubes, 0.1 for flakes), ρ is the par-
into account the most modern sam- ticle density, cL is the mass fraction of
pling theories (for more on the history Statistics of sampling material in the size class of interest,
of sampling strategies, see box, p. 45), The first necessary step for a good sam- d1 is the average particle diameter
which leads to inaccurate test results pling program is to define the sample in the size class of interest, g is the
and unrealistic material specifications that is needed and clearly specify how granulometric factor [ratio of the di-
that are impossible to verify properly. the sample is taken, including equip- ameter corresponding to the 5th per-
The best practices outlined in this ment specification. It is important to centile of the size distribution to the
article provide guidelines for collect- keep in mind that in particulate ma- diameter corresponding to the 95th
ing representative samples from most terial sampling, the best we can ever percentile of the size distribution (d05/
solids handling and processing equip- achieve is a random sample where d95)], d is the diameter corresponding
ment and then reducing the sample all particles within the parent popu- to the 95th percentile of the distribu-
to the proper size for the analytical lation have an equal chance of being tion (d95). This allows the calculation
technique used in the measurement. sampled, thereby assuming that no of the fundamental error for any size
In addition, an assessment of sam- systematic bias exists in the sampling class in a distribution. If the mass of
pling errors, based on simple statisti- process. Since there is no such thing as the parent population is much greater
cal theories, illustrates the pitfalls of two identical samples, a perfectly ex- than the sample mass, the term 1/ML
sampling methods. tracted sample (random sample) will can be dropped from the equation. A
One of the everyday examples of always be inflicted by a residual error, few important highlights from the
sampling that all of us can relate to called the fundamental error (FE), as above equation:
is when a medical doctor orders blood first postulated by Gy [1]. This is due 1. The variance of the fundamental
to be drawn for routine laboratory to the heterogeneity of any particulate error decreases as the sample size
42 Chemical Engineering www.che.com april 2012
Example of a Sampling problem, with solution

A
fter several customer complaints, an engineer is assigned the responsibility of
setting up a sampling protocol for a ground product that frequently does not d05 100
= = 0.40
meet the specification that no more than 5% of the mass, or volume distribution d95 250
should be greater than 250 microns (Figure 1). This product is sold in lots consisting of
several tons. This specification should be measured at the 99% confidence level. This 5th percentile
product has a density of 2.5 g/mL. Assuming that correct sampling techniques were - 100 µm
used to obtain a random sample, what is the minimum sample size that needs to be 95th percentile
collected and analyzed? -250 µm

Solution: Material
1. Since the mass of the sample is much smaller that the mass of the lot, the equation specification
for the fundamental error estimation [Equation (1)] can be rearranged as follows to < 5% greater
than 250 µm
solve for the minimum sample mass

100 250 275


 (4) Diameter (micron)

2. Measure the size distribution on a volume, or mass basis to obtain the diameters Figure 1. Example of size distribution
with information necessary to calculate
corresponding to the 5th and the 95th percentile (Figure 1)
minimum sample mass
3. The 99% confidence level implies that the value of FE is 0.01. The variance of the
2
fundamental error, Var(FE), is 0.01 , or 0.0001. The shape factor (ƒ) can be set at
0.5, assuming that the particles can be approximated by spheres. The particle density (ρ) is 2.5 g/cm3. The fraction of material in
the size class of interest cL is 0.05 (5% > 250 microns). The average diameter in the size class if interest (d1) can be taken as 275
microns (see Figure 1). The granulometric factor (g) is defined as d05/d95 (see Figure 1) is 0.40 for this distribution. Finally, d,
defined as the 95th percentile of the distribution (see Figure 1) is 250 microns. Changing all units to CGS units to obtain the sample
mass (MS) in grams, we obtain the following:

 (5)

Please note that not only a sample of 4.8 g is needed, but an analysis technique that can analyze the whole sample needs to be utilized.
 ❏

increases. Since the variance is curacy and precision. In addition to size, sampling typically becomes less
equal to the square of the funda- sampling error, analytical error will of an issue, and analytical errors take
mental sampling error, the funda- also add to the uncertainty of the over. Table 2 outlines the basic steps
mental sampling error decreases in measurement. With modern particle- for correct sampling.
proportion to the square root of the characterization instrumentation, Grab samples should not be used
sample mass the sampling error will typically be- even if one attempts to mix the bulk
2. The variance of the fundamental come much larger than the expected specimen prior to sampling — for ex-
error is a strong function of the analytical error as the top end of the ample, bulk bags or perhaps a sample
coarse end (95th percentile) of the distribution (95th percentile) exceeds brought to the laboratory. It is simply
size distribution as dictated by the 100 microns. Gy defined each of the not possible to obtain a homogeneous
d3 term. seven error components as an additive mix from blending alone, and there-
The above equation can easily be rear- model where the variance of the total fore such a practice should not be used
ranged to provide the minimum sam- error is as follows: to properly minimize grouping and
ple mass to be used in an analysis. The segregation errors. Pitard [2] showed
TE = FE + GE + CE2 + CE3 + DE +
sample mass estimate is the minimum that the variance of the grouping error
EE + PE (2)
sample size, since additional sources can be compared to the variance of the
of error will contribute to the variance If correct sampling practices are uti- fundamental error as follows:
of the total sampling error. It should lized, the terms GE, CE2, CE3, DE,
be noted that these additional con- EE, and PE are minimized, and are
tributors can be minimized through much smaller that the FE term, for  (3)
good sampling practices, and therefore particles sized greater than about
are controllable to a large extent. Gy 100 microns. This minimization of As a rule of thumb, at least 30 sample
broke down the total sampling errors the sampling error can only be accom- increments (N) are recommended to
into seven basic components as listed plished through appropriate selection minimize GE errors.
in Table 1. of sampling equipment for all phases
The mass required to meet a prod- of the sampling and sub-sampling pro- Correct Sampling
uct specification is related to the in- cess. For smaller particle sizes, where Correct sampling implies following
herit degree of heterogeneity in the the heterogeneity of the system de- a few simple rules throughout the
material and the desired level of ac- creases as the third power of particle sampling process as well as using ap-
Chemical Engineering www.che.com april 2012 43
Feature Report

propriate sampling tools to minimize


Table 1. Seven basic sampling errors
the errors identified in the previous
section. Correct sampling practices in- Name Description / Mitigation
clude the following:
• Taking many samples at regular or 1 Fundamental Caused by constitutional heterogeneity (CH). Reduce
random time intervals (>30 sam- Error (FE) FE by increasing sample size. Note that this is the
sample size that not only needs to be sampled, but
ples), and sub-dividing into smaller analyzed in its entirety
samples for analysis to minimize
grouping and segregation error (GE) 2 Grouping and Incremental samples can be different from each other.
• Using correctly designed sampling Segregation Reduce GE by collecting and combining several ran-
Error (GE) dom sub-samples, taken correctly from the parent lot
tools to minimize delimination and
extraction errors (DE and EE). 3 Long-Range Fluctuations in size distribution over time contribute to
• Using common sense and diligence Heterogeneity the heterogeneity. Reduce CE2 by collecting a large
to minimize sample preparation and Fluctuation Error number of sub-samples at random or regular intervals
(CE2) to form a composite
analysis errors (avoid particle set-
tling, agglomeration, dissolution, 4 Periodic Periodic fluctuations in size distribution over time con-
and swelling) (PE) Heterogeneity tribute to the heterogeneity. Reduce CE3 by collecting
In this section, we will introduce sam- Fluctuation Error a large number of sub-samples at random or regular
pling equipment designed to sample (CE3) intervals to form a composite
from various solids systems including 5 Increment Delimitation errors occur when the sampling process
static bulk materials, gravity flow sys- Delimitation does not give an equal probability of selecting all
tems, mechanical conveying systems, Error (DE) parts of the parent lot. As an example, a grab sample
pneumatic conveying systems, solids- will only sample from accessible parts of the lot,
usually the surface. Reduce DE by using properly de-
processing unit operations and slurry
signed sampling tools and strategies
systems. The sampling techniques in
different systems are discussed and 6 Increment Since particles are discrete elements of various sizes,
recommendations for proper sampling Extraction Error they will be forced in or out of the sampling device
are provided. (EE) — even if they are on the sample target boundary. If
a particle’s center of gravity is within the sampling
boundary, it should be part of the sample, otherwise it
Sampling process overview should not be part of the sample. Reduce EE by using
There are usually several stages in properly designed sampling tools
particulate matter sampling, and it
7 Preparation Error Sample degradation error caused by inadequate
is of paramount importance to main-
(PE) preparation where particles settle, dissolve, aggregate,
tain the integrity of the sample until break or swell during preparation or analysis. Use
the analysis is carried out. Figure 2 proper sample handling and dispersion techniques
takes us through the stages of a sam-
pling process. Several increments are
taken from the bulk lot using properly solely governed by the heterogeneity properly designed tools. Each increment
designed sampling equipment as out- of the material — grab sampling at can be defined as the group of particles
lined in the next section. The gross any level will almost guarantee that extracted from the parent lot during a
sample may be too large to be sent this goal will not be achieved. single operation cycle of the sampling
to the laboratory, and may need to be device. The final gross sample should
reduced to a more practical weight. Gross sample extraction consist of at least 30 such increments.
Depending on the measurement tech- Consistent with Gy’s sampling theo-
nique, and the amount of sample re- ries, Allen [3] independently proposed Static material sampling
quired by the instrument sample de- two “Golden Rules” of sampling: Ideally, the sampling should have
livery system, the laboratory sample 1. Sample a moving stream — sam- been carried out before the material
may need to be further sub-divided to pling of bulk solids at rest should became a static bulk, which is much
the test sample to be used in its en- be avoided. more difficult to correctly sample.
tirety by the instrument. Even at the 2. The whole of the stream of powder The degree of inhomogeneity will de-
laboratory-sample level, which is the should be taken for many small in- pend on the powder’s history. In the
last step before analysis, the common crements in time in preference to case of free-flowing material, it is a
practice of simply scooping material part of the stream being taken for safe bet to assume segregation has
out of the container is likely to in- the whole time. taken place during the transfer, and
troduce bias. The overall goal of any Applying Gy’s principles and Allen’s for non-free flowing material, the de-
sampling procedure is simple: it is to recommendations, extraction of a gross gree of inhomogeneity will largely de-
obtain a sample with a total sampling sample consists of properly extracting pend on its history.
error similar to that expected from the several increments from the parent lot The inherent problem with sampling
fundamental sampling error, which is during processing or handling using static material is that no equipment
44 Chemical Engineering www.che.com april 2012
History of sampling techniques

S
ampling became a common, but non-scientific practice first in the mining industry,
then in the pharmaceutical and chemical industries shortly after the industrial revo-
lution. Back in those early days of sampling, although no rigorous theory existed,
scientists and engineers used a common-sense approach based on their intuition and
their experience to guess at the requirements on what constituted a good sample. In
the mid-19th century, Vezin was the first to introduce the concept of a minimum sample
size necessary for obtaining a representative sample, without the benefits of modern
sampling theories. He also invented a sampler that bears his name, and is still in use
today. It was not until the 1950s that the guessing game in sampling was replaced by a
more rigorous discipline, thanks to Gy’s [1] development of the statistical theories behind
sampling. This offered a structured approach to sampling from which all sampling errors
are broken down to basic components.

Gross sample Lab sample Test sample Measured sample


g to mg
> Kg < Kg g
1 2 3
Bulk
or Sample Sample Sample delivery
Increments
division division system
Figure 3. The sampling thief is one
process
of the simplest devices to extract pow-
stream
der from a static bulk
Propagation of errors

Goal: total error ≈ fundamental error through correct sampling


region at the bottom of Figure 3 in-
dicates the region where there is no
chance of sampling, which illustrates
Figure 2. In this sampling process, incremental sampling throughout the sam-
pling and sample reduction process is practiced to minimize propagation of sam- a weakness of this device. Another
pling errors source of error to be aware of when
using this type of device occurs as the
Table 2. Basic steps for correct sampling material is being displaced down by
the probe moving through the bulk
1. Define sample quality Data Quality Objective – precision and accuracy
required for product specification, or quality
material, thereby causing segregation
and preventing equal probability for
2. Define sample size Sample size: gross sample, lab sample, actual all particles to be sampled.
amount analyzed
3. Define sampling strategy Equipment, sampling location, sampling fre- Sampling free-falling streams
quency, sample reduction The rotary chute sampler, also referred
4. Preserve sample integrity Sample reduction, prevent particle aggregation, to as the Vezin sampler, is a multi-pur-
attrition, dissolution, and swelling pose device that collects representative
samples from materials (dry powders or
5. Verify that the required Is the equipment and strategy used adequate to
slurries) that are free-falling from pipes,
data quality can be meet data quality objective? Is the sample size
achieved analyzed large enough? chutes or hoppers. This sampler is gen-
erally a good choice for installation on
loading and unloading equipment, or at
exists that can take a sample where sampling procedure, the inner tube is the entrance or exit of material trans-
every particle has an equal chance of rotated so that the matching groves fer equipment. Various versions of the
being sampled. There will always be are on opposite sides, then the probe Vezin sampler are available in several
parts of the bulk that will not be acces- is inserted in the powder. The second sizes from multiple manufacturers. This
sible to the sampler. step consists of twisting the inner device, shown in Figure 4, operates by
The workhorse of the bulk sampling tube to align the two sets of grooves, one or more cutters revolving on a cen-
domain remains the thief sampler thereby allowing powder to flow into tral shaft, passing through the sample
(Figure 3), which provides several in- the sampler. Thirdly, the inner tube stream and collecting a fixed percentage
crements taken at random throughout is twisted to lock the powder into the of the total material. A Vezin sampler is
the bulk material. This device consists sampler, which is then withdrawn totally enclosed to minimize spillage or
of a co-axial outer sleeve and an inner from the bulk. This procedure is re- leakage problems. The area between the
hollow tube with matching grooves to peated several times to extract sev- sample cutter and the discharge chute
allow powder flow in the core of the eral increments to make up the bulk is sealed to prevent possible contamina-
inner cylinder. In the first step of the sample ready for splitting. The shaded tion or sample loss.
Chemical Engineering www.che.com april 2012 45
Feature Report

Table 3. List of questions to


consider when Selecting A
Sampler
Is the material free-flowing?

Is the material abrasive?

Is the material friable?

Materiel properties
Does the material have a broad
size distribution?

Is the material dusty?

What is the largest particle


diameter

Is the material temperature sen-


sitive?

Is the process enclosed?

Are the particles dispersed in


gas phase?

Process conditions
Intersystems
Figure 4. Rotary chute sampler Is the process in a pressurized
enclosure?

Is the process at elevated


As a rule of thumb, incremental temperature?
extraction errors can be minimized
by limiting the cutter speed to 0.6 Is the process wet or dry?
m/s, an inner-wall sampler three
times the particle diameter (3d) for Is the powder in motion?
coarse material, where d > 3 mm, and
at least 10 mm for finer material. What sample size is required?

Sampling from gravity flow


Sample requirements

Are there any sanitary


As shown in Figure 5, gravity flow can
requirements?
be any free-flowing powder or slurry
from a conveyor, hopper, launder or Is automatic sampling required?
unit operation under the influence of
gravitational forces. When sampling in
Is a composite sample
such systems, each increment should required?
be obtained by collecting the whole of Heath & Sherwood Co.
the stream for a short time. The width Is the sample sensitive to
of the receiver should be made at least moisture?
Figure 5. Linear gravity flow samplers
10 mm or three times the diameter collect samples from free-flowing pow-
of the largest particles — whichever ders under the influence of gravity
is larger. The volume of the receiver
must be large enough to ensure that Mechanical conveying systems noted above. However, if the situation
the receiver is never full of material. The conveyor types for mechanical and is such that samples have to be taken
The length of the receiver should be pneumatic conveying of bulk solids in- directly from within the conveying line,
sufficient to ensure that the full depth clude belt conveyor, screw conveyor, several types of sampler have been de-
of the stream is collected. The ladle or bucket conveyor, vibrating conveyor, veloped. An example of such samplers
receiver should cross the whole stream and dense- or dilute-phase conveyers. designed to extract samples from belt
in one direction at constant velocity. The best position for collecting the conveyor systems is illustrated in Fig-
For heavy mass flow, a traversing cut- samples is where the material falls in ure 6. The mid-belt sampler uses a ro-
ter as a primary sampler together with a stream from the end of the conveyor. tating scoop that makes a pass across
a Vezin sampler as a secondary splitter One can then follow the procedure for the moving belt, thereby cutting a
can usually be applied. gravity flow or free-falling streams as clean cross section of material.
46 Chemical Engineering www.che.com april 2012
Powder
supply

Sample
cups

Rotation

Drive
axis

Direction
of rotation

Intersystems
Figure 8. The spinning riffler is com-
prised of a ring of containers rotating
under a powder stream

Figure 6. Automatic mid-belt samplers are used with belt conveyors

Vp Vp Vp

Vs Vs Vs

Vp = Vs Vp > Vs Vp < Vs

Figure 7. These illustrations of isokenetic sampling from a pipeline show the Figure 9. The chute riffler
sampling velocity (Vs) equal to the process velocity (Vp; left), Vp greater than Vs (mid- splits a sample using a series of
dle), and Vp less than Vs (right) alternate chutes

Slurry sampling dent on such parameters as particle ing fluid velocity (Vp). No sampling bias
The same basic sampling rule where size and density, fluid density and vis- is expected during isokinetic sampling.
all particles have an equal chance of cosity, flowrate and pipe diameter [4]. If the process flow velocity is greater
being sampled must also be followed Turbulent flow, which provides mixing, than the sampling velocity, particle
when sampling from slurries. Knowl- is typically required to keep the slurry inertia causes an excess of larger par-
edge of slurry properties and behavior well mixed before sampling. Pipelines ticles to enter the sampling probe while
is essential to insure proper sampling can be sampled isokinetically using a process flow velocity smaller than
strategies. For instance, sampling a nozzles provided the slurry is well sampling velocity will cause an excess
slurry from a point in a tank, or flowing mixed at the sampling point. Isokinetic of larger particles to avoid the probe.
through a pipeline requires the pres- sampling (Figure 7) occurs when the Therefore, non-isokinetic sampling will
ence of a homogeneous suspension at average fluid velocity in the sampling introduce a bias based on the particle
the point of sampling, which is depen- tube (Vs) is the same as the surround- size distribution.
Chemical Engineering www.che.com april 2012 47
Feature Report

Relevant standards on sampling of particulate materials

ASTM Standards: ASTM D345 - 02(2010) Standard Test Method for Sampling and
Testing Calcium Chloride for Roads and Structural Applications
ASTM B215 - 10 Standard Practices for Sampling Metal Powders
ASTM D346 / D346M - 11 Standard Practice for Collection and
ASTM C322 - 09 Standard Practice for Sampling Ceramic White- Preparation of Coke Samples for Laboratory Analysis
ware Clays
ASTM D460 - 91(2005) Standard Test Methods for Sampling and
ASTM C50 - 00(2006) Standard Practice for Sampling, Sample Chemical Analysis of Soaps and Soap Products
Preparation, Packaging, and Marking of Lime and Limestone
Products ASTM D75 / D75M - 09 Standard Practice for Sampling Aggre-
gates
ASTM C702 / C702M - 11 Standard Practice for Reducing Sam-
ples of Aggregate to Testing Size ASTM D979 / D979M - 11 Standard Practice for Sampling Bitu-
minous Paving Mixtures
ASTM D140 / D140M - 09 Standard Practice for Sampling Bitu-
minous Materials ASTM E105 - 10 Standard Practice for Probability Sampling Of
Materials
ASTM D1799 - 03a(2008) Standard Practice for Carbon Black—
Sampling Packaged Shipments ASTM E122 - 09e1 Standard Practice for Calculating Sample Size
to Estimate, With Specified Precision, the Average for a Charac-
ASTM D1900 - 06(2011) Standard Practice for Carbon Black teristic of a Lot or Process
Sampling Bulk Shipments
ASTM E141 - 10 Standard Practice for Acceptance of Evidence
ASTM D1900-06(2011) Standard Practice for Carbon Black Sam- Based on the Results of Probability Sampling
pling Bulk Shipments
ASTM D197 - 87(2007) Standard Test Method for Sampling and
Fineness Test of Pulverized Coal International Standards:
ASTM D197 - 87(2007) Standard Test Method for Sampling and BS 3406: Part 1: 1986 British Standard Methods for Determina-
Fineness Test of Pulverized Coal tion of particle size distribution Part 1. Guide to Powder Sam-
pling, British Standards Institute, London (1986).
ASTM D2013 / D2013M - 11 Standard Practice for Preparing
Coal Samples for Analysis ISO/WD: 14888 Sample Splitting of Powders for Particle Size
Characterisation International Organization for Standardization,
ASTM D2234 / D2234M - 10 Standard Practice for Collection of Geneva.
a Gross Sample of Coal
ISO 2859-Statistical Sampling. http://www.iso-9000.
ASTM D2590 / D2590M - 98(2011)e1 Standard Test Method for co.uk/9000qfa9.html, International Organization for Standard-
Sampling Chrysotile Asbestos ization, Geneva (2000).

It is better to sample from a verti- number of cuts (>30) for both the pri- widely used for reducing the amount
cal pipe so that particle segregation by mary and secondary samplers needs of powder to be analyzed to a smaller
gravity can be avoided. In such a situ- to be extracted. Not all situations are representative sample. In this com-
ation, the sampler should be located at alike, and therefore, these samplers mercially available device, a ring of
least ten pipe diameters downstream need to be installed and designed containers rotates under a powder
from any bends or elbows in the pipe. properly to fit the application. flow to be sampled, thereby cutting
Particle diameter has a strong influ- Selection of the proper sampling the powder flow into several small in-
ence on particle segregation by grav- equipment may not always be trivial, crements so that each container con-
ity since the settling velocity is pro- and may depend on material proper- sists of a representative sample. The
portional to the square of the particle ties, type of process, and sample re- spinning riffler is a versatile device
diameter. Gravity starts to play an quirements. Table 3 provides a list of that can handle free-flowing powders,
important role at particle diameters questions to consider when designing dusty powders and cohesive powders.
greater than roughly 50 microns. The a sampling protocol. The operating capacity of this device
best approach, if possible, is to sample varies from 25 mL to 40 L. If only
at the discharge where a cross-stream Sample reduction the small capacity spinning riffler is
sampler (Figure 5) may be used as a Powder sampling is typically done available, the Vezin sampler can be
primary sampler followed by a Vezin at two levels, a gross sample taken used to reduce the gross sample to
sampler cutter to reduce sample size. directly from the process, and then the appropriate quantity suitable for
This allows sampling even in the non- sub-divided into samples suitable for the spinning riffler. The spinning rif-
ideal case where some segregation the laboratory. The spinning riffler, fler, when properly used, is the most
may have occurred in the pipe. A large as illustrated in Figure 8, has been efficient sample divider available.
48 Chemical Engineering www.che.com april 2012
Another commonly used device for crease the number of total increments, Summary
sample reduction of free-flowing pow- it will likely not match the number of Appropriate attention to sampling,
ders is the chute riffler as shown in Fig- increments performed by the spinning sample size reduction and data analy-
ure 9. It consists of alternating chutes riffler. As such, the spinning riffler is sis is the first step towards obtaining
where half of the material discharges the best device for sample reduction reliable analytical results from a batch
on one side and the second half on the and should be used whenever possible. [5]. To obtain a representative sample,
other. The total number of chutes rep- Several standards dealing with pow- one must adhere to the golden rules of
resents the number of increments de- der sampling are available from a num- sampling and follow the best practices
fining the sample. Although the sample ber of organizations. A comprehensive as outlined in this article. ■
can be processed several times to in- list is provided in the box, p. 48. Edited by Rebekkah Marshall

Authors
References Remi Trottier is a research Shrikant V. Dhodapkar is a
1. Gy, Pierre, “Sampling Theory and Sampling scientist in the Solids Pro- fellow in the Dow Elastomers
Practice. Heterogeneity, Sampling Correct- cessing Discipline of Engi- Process R&D Group at The
ness, and Statistical Process Control”, 2nd neering & Process Sciences Dow Chemical Co. (Freeport,
Ed., CRC Press, Boca Raton, 1993. at The Dow Chemical Co. TX 77541; Phone: 979-238-
(Phone: 979-238-2908; Email: 7940; Email: sdhodapkar@dow.
2. Pitard, Francis F., “Pierre Gy’s Sampling The- ratrottier@dow.com). He re- com). He received his B.Tech.
ory and Sampling Practice: Heterogeneity, ceived his Ph.D. in chemical in chemical engineering from
Sampling Correctness, and Statistical Process engineering from Loughbor- I.I.T-Delhi (India) and his
Control”, CRC Press, Boca Raton, 1993. ough University of Technol- M.S.Ch.E. and Ph.D. from the
3. Allen, T., “Particle Size Measurement”, 4th ogy, U.K,, and M.S. and B.S. University of Pittsburgh. Dur-
Ed., Chapman & Hall, London, 1990. degrees in Applied Physics ing the past 20 years, he has
at Laurentian University, Sudbury, Ont. He has published numerous papers on particle technology
4. Turian, R.M., and Yuan, T.F., Flow of Sluries more than 20 years of experience in particle and contributed chapters to several handbooks.
in Pipepines, 3, 1977, AIChE J., Vol. 23, 3, pp. characterization, aerosol science, air filtration He has extensive industrial experience in powder
232–243. and solids processing technology. He has au- characterization, fluidization, pneumatic convey-
thored some 20 papers, has been an instructor ing, silo design, gas-solid separation, mixing, coat-
5. Trottier, Remi and Dhodapkar, Shrikant, and of the course on Particle Characterization at the ing, computer modeling and the design of solids
Wood, Steward, Particle Sizing Across the International Powder & Bulk Solids Conference/ processing plants. He is a member of AIChE and
CPI, Chem. Eng., April 2010, pp. 59–65. Exhibition for the past 15 years. past chair of the Particle Technology Forum.

Whether you need to transport, analyze, weigh, batch,


mix, grind, dry, shape or package you’ll find the solution at…

Exhibition & Conference: May 8–10, 2012


Donald E. Stephens Convention Center • Rosemont, IL (adjacent to O’Hare Airport)
Featuring:

Meet Industry Leaders with Thousands of the Latest Processing Solutions:


• Accessories • Feeders • Plant Maintenance/ • Storage
• Conveyors and Elevators • Instrumentation & Controls Safety/Health Products • Thermal Solids Processors
• Dryers • Material Handling & Transportation • Processing/Mixing/ • Weighing Systems & Scales
Blending Equipment
• Dust Collection/Control • Mixers & Blenders …and much more
• Size Reduction
• Energy, Environmental & • Packaging & Bagging
Pollution Control • Particle Enlarger & Formers
• Filtration/Separation

Follow the tag to register or log on to:


Get the free
mobile app at PowderShow.com
Produced and managed by: UBM Canon
http://gettag.mobi 11444 W. Olympic Blvd. • Los Angeles, CA 90064-1549
21377_CH_PTXi12 Tel: 310/445-4200 • Fax: 310/996-9499

Circle 22 on p. 82 or go to adlinks.che.com/40266-22
Chemical Engineering www.che.com april 2012 49
BS&B Safety Systems

Feature Report
Engineering Practice

The Direct Integration Method:


A Best Practice for
Relief Valve Sizing
The approach described here is easier to use,
and provides more-accurate results,
compared to leading valve-sizing methodologies
Mark Siegal Silvan Larson
Consulting Engineer and William Freivald Figure 1. Today, with the help of
spreadsheet programs and simulators,
Valdes Engineering Company the once-cumbersome Direct

W
Integration Method is easier than
hat if someone were to merical integration of the isentropic ever to use to size relief valves
tell you that there is one nozzle equation is now easier, faster,
method available for sizing and more accurate than other meth-
relief valves that applies ods for determining the mass flux pressure at constant entropy over the
to virtually every situation, includ- through a relief valve. This article dis- pressure range encountered in the
ing two-phase flow and supercritical cusses the use of process simulators nozzle. To solve the integral analyti-
fluids? And what if they told you that to simplify the numerical integration cally, an equation of state needs to be
method is more accurate and easier method, and describes the advantages available for the fluid at constant en-
to use than traditional methods or of numerical integration over other tropy. However, for many fluids, such
formulas? As it turns out, both of methods that may be used to calculate an equation is not available for density
these statements are true. The ap- the required relief valve area. as a function of pressure. To overcome
proach described here — the Direct this limitation, various simplifying as-
Integration Method — involves nu- Calculation methods sumptions were traditionally made to
merical integration of the isentropic Isentropic Converging Nozzle allow the integral to be solved analyti-
nozzle equation [1]. Equation. The calculation of the the- cally, rather than by performing a nu-
From as early as 2005, the “method oretical mass flux for homogeneous merical integration.
of choice” for determining the flow fluids through a relief valve is gener- For instance, for non-flashing liq-
through a relief valve has been the Di- ally accepted to be modeled based on uids, the density is assumed to be con-
rect Integration Method [2]. API 520 the isentropic converging nozzle. The stant, and the integral is easily solved.
has also sanctioned this method due isentropic nozzle equation is devel- The traditional vapor-sizing equation
to its general applicability to any situ- oped from the Bernoulli equation by is obtained by assuming the vapor
ation where the fluid is homogeneous assuming that the flow is adiabatic is an ideal gas with a constant heat
[1]. However, because this method is and frictionless [4]. capacity [5]. However, the assump-
perceived to be difficult or time con- tions required by these methods may
suming, many engineers continue to introduce large errors under some
 (1)
opt for older, simplified methods, even conditions. In contrast, the Direct In-
though such methods can produce less- tegration Method has been shown to
accurate results. For instance, without The required nozzle area of the relief produce more-accurate results.
careful analysis, using the traditional valve is calculated using Equation (2). Direct Integration Method. The
gas-phase equation near a fluid’s Direct Integration Method uses a nu-
critical point can yield an undersized merical method to evaluate the inte-
valve [3].  (2) gral in the isentropic nozzle equation
Fortunately, thanks to the wide- [2]. API 520 proposes the use of the
spread availability of process simu- To use Equation (1), the fluid den- Trapezoidal Rule, shown below, to cal-
lators and spreadsheet software, nu- sity must be known as a function of culate the integral:
54 Chemical Engineering www.che.com April 2013
Nomenclature1
G0 Mass flux, Pn Nozzle exit Kd Discharge coefficient,
lb/h • in.2 pressure, psi unitless ing predictions are fairly accurate at
ρ Density, lb/ft3 A Orifice area, in.2 Pi Pressure at stage i, psi conditions well away from the critical
P0 Relieving W Relieving mass rate, ρi Density at stage pressure. However, two-phase models
pressure, psi lb/h i, lb/ft3 are more difficult to implement. Exist-
ing two-phase flow models approxi-
1. Unit conversion may be required, depending on the units selected.
mate the pressure-density relation-
ship of the fluid in order to calculate
mass flux will occur at some pressure the integral in Equation (3).
 (3) above the relief-valve exit pressure. One of the simplest models, the
This is called the choked pressure [6] Omega Method, assumes a linear
pressure-density relationship, with
The method is performed by using a Implementation the omega parameter (ω) represent-
process simulator to generate data It is possible to fully automate the ing the slope of the pressure-density
points for the fluid density at various Direct Integration Method using a curve. An analytical solution to the
pressures, utilizing an isentropic flash spreadsheet program (such as Micro- isentropic nozzle equation was devel-
routine over a pressure range from the soft Excel 2010) and a process simula- oped using the omega parameter to
relieving pressure to the exit pressure. tor (such as AspenTech HYSYS 7.2) [7]. solve the integral [8].
The simulation data are used to deter- Users can automate the process to the The TPHEM Method uses three
mine the theoretical mass flux at each point where all they would need to do is pressure-density points to define co-
point. simply hit a button in the spreadsheet efficients for an empirical equation
Using Equation (3), the maximum program and the numerical integration of state “model” [9]. The empirical
mass flux is determined by calculat- will be performed on an existing stream equation is then used to evaluate the
ing the mass flux over incrementally in the simulator using a VBA (Visual integral numerically. Pressure-den-
larger pressure ranges, beginning at Basic for Applications) program. sity data for these models are often
the relieving pressure, and observing First, the spreadsheet is set up to provided by a process simulator. If
where a maximum flux is reached. If accept the pressure and density data a simulator is available, then it
the maximum occurs at the relief-valve for the numerical integration points. is much simpler to use the Direct
exit pressure (built-up backpressure), The inlet and outlet pressure points, Integration Method.
then the flow is not choked. Generally pressure step size, and name of relief The Direct Integration Method is
accurate results can be obtained with stream in the simulator are placed fundamentally different from the
pressure increments as large as 1 psi, into specific cells in the spreadsheet, other methods described here because
but smaller step sizes can be specified which are referenced in the VBA code. it does not generate an explicit equa-
if desired [2]. Once the mass flux is The VBA code instructs the simulator tion-of-state model to relate pressure
determined, the required relief valve to create a new, ideal expander pro- and density. Instead, pressure and
orifice area* can be determined from cess block and associated streams in density data are generated using the
Equation (2). the simulator. The code then iterates full thermodynamic models available
The value of the discharge coefficient, across the pressure range and modi- in the selected process simulator, and
Kd, depends on the phase of the fluid fies the pressure of the expander prod- these data are then used to solve the
and varies by the manufacturer of the uct stream and automatically exports integral numerically. Since there is
relief valve. The discharge coefficient the pressure and density data to the no reliance on a curve-fit pressure-
corrects for the difference between the Excel spreadsheet. density model, the Direct Integration
theoretical flow and the actual flow For each data point in the spread- Method is more exact and reliable,
through the nozzle. This value is deter- sheet, the summand, cumulative sum, assuming the simulator’s thermody-
mined empirically for liquid and vapor and mass flux are calculated using namic model is accurate. Specifically,
and reported by vendors for each make Equation (3) with typical spreadsheet there is no chance for inaccuracies
and model of relief valve. If vendor formulas. When a maximum mass flux associated with the fluid equation of
data are not available, an initial guess is reached, the spreadsheet uses this state “model” propagating through the
of 0.975 for gases, or 0.65 for liquids maximum flux value to calculate an rest of the calculations resulting in
can be used [1]. orifice size, given the relieving mass inaccurate mass flux estimations and
For two-phase flow, the liquid-dis- rate and coefficients. Alternatively, ultimately an inappropriate relief-
charge coefficient should be used if the data can be collected using the valve area [8, 9, 10].
flow in the valve is not choked and the “databook” feature in the simulator Note that the Direct Integration
maximum mass flux will occur at the and copied into the spreadsheet using Method assumes that the two-phase
relief-valve exit pressure. If the flow is a simple copy-and-paste operation. fluid is homogeneous, and that the
choked, then the gas-discharge coeffi- fluid is in mechanical and thermody-
cient should be used and the maximum Two-phase relief scenarios namic phase equilibrium. The homo-
The existing single-phase vapor and geneous assumption is valid for most
* While relief valves are designed with a nozzle, non-flashing liquid methods are rela- two-phase reliefs due to high velocity
the area at the end of the nozzle is commonly
referred to as the “orifice area”. tively easy to calculate and the result- in the nozzle, which promotes mixing
Chemical Engineering www.che.com April 2013 55
GE/Consolidated and Allied Valve

Engineering Practice

[2]. The mechanical equilibrium as- Closing remarks


sumption is valid for flashing flows Using a spreadsheet to import data
[2]. The thermodynamic equilibrium from a simulator and to calculate the
assumption is valid for nozzles with summation over a range of pressures
a length longer than 10 cm [4]. Most is extremely easy and straightfor-
standard relief valves have a nozzle ward. One simply needs to simulate
that is slightly longer than this [11]. the relieving stream and perform
a flash operation at each pressure
Pros and cons and capture the required data. Not
Advantages of this Method. The Di- only is the Numerical Integration
rect Integration Method is not bound Method much simpler than the al-
by the same constraints as many ternatives for two-phase flow, but it
other models or methods. Using this is also more accurate, since it does
approach, the same method can be not rely on a potentially sensitive
used whether the flow is choked or not equation-of-state model. There is no
choked, flashing or not flashing, single FIGURE 2. The Direct Integration need for a model because physical
Method is not only easy to use, but
or two-phase, close or far from the provides more accurate results when property data are generated for each
critical point, subcooled or supercriti- sizing pressure relief valves, since this data point directly from simulation.
cal. The only assumptions required approach does not rely on a potentially In addition, the Numerical Integra-
for the Direct Integration Method are sensitive equation of state model tion Method can be used for single-
that flow through the relief valve is phase flow and choked or not-choked
isentropic, homogeneous, and in ther- overly conservative results in a cou- conditions. This versatility and ease
modynamic and mechanical equilib- ple of circumstances, which can lead of calculation makes Numerical In-
rium, although it is possible to adjust to under-prediction of the mass flux tegration the obvious choice for any
the method to account for mechanical and selection of an oversized valve. relief valve calculation where physi-
non-equilibrium or slip [6]. This appears to be an issue only when cal property data are available in a
Although most other methods give the fluid is in two-phase frozen flow process simulator. n
unsatisfactory results near the thermo- (no flashing), or the relief valve has Edited by Suzanne Shelley
dynamic critical point, the Direct Inte- a short nozzle and there is flashing
gration Method continues to function flow [2].
Authors
properly [12]. Additionally, many other This potential limitation can be
Mark Siegal (Email: msiegal2
concerns that come up when using compensated for in both situations by @gmail.com), was, until re-
relief-valve model equations, such as applying a slip factor. However, at this cently, a process engineer at
Valdes Engineering Company
determining the heat capacity ratio or time, there is insufficient literature where he was responsible for
process design, process mod-
isentropic expansion coefficients, are available to provide accurate guidance eling, and emergency relief
no longer relevant since they are inher- on the value of a slip factor. The accu- system design. He holds a
B.S.Ch.E. from the University
ent to the simulator itself [3]. racy of the calculation is also limited of Illinois at Urbana-Cham-
Downsides to this Method. The Di- by the accuracy of the physical prop- paign.
rect Integration Method can produce erty data in the simulator.
Silvan Larson is a principal
process engineer at Valdes
Engineering Company (100
References 7. AspenTech, Aspen HYSYS Customization
Guide, Version 7.2, July 2010. W 22nd St., Suite 185, Lom-
bard, IL 60148; Phone: 630-
1. American Petroleum Inst., “Sizing, Selection, 792-1886; Email: slarson@
and Installation of Pressure-Relieving De- 8. Leung, J.C., The Omega Method for Dis-
charge Rate Evaluation, in “International valdeseng.com),where he is
vices in Refineries,” ANSI/API RP 520, 8th responsible for process design
Ed., Part 1: Sizing and Selection, Washing- Symposium on Runaway Reactions and
Pressure Relief Design,” G.A. Melhem and and emergency-relief-system
ton, D.C., Dec. 2008. design. He has more than 30
H.G. Fisher, Eds., pp. 367–393, AIChE., New
2. Darby, R., Size safety-relief valves for any York, N.Y., 1995. years of experience in manu-
conditions, Chem. Eng., pp. 42-50, Sept. facturing and process design
2005. 9. Center for Chemical Process Safety, “Guide- engineering in the chemicals and petroleum re-
lines for Pressure Relief and Effluent Han- fining industries. He holds a B.S.Ch.E. from Uni-
3. Kim, J.S., H J. Dunsheath and N.R. Singh, dling Systems,” AIChE, New York, N.Y., 1998. versity of WisconsinMadison and is a registered
Proper relief-valve sizing requires equation professional engineer in Ill.
mastery, Hydrocarbon Proc., pp. 77–80, Dec. 10. Diener, R., and J. Schmidt, Sizing of throt-
2011. tling device for gas/liquid two-phase flow,
Part 1: Safety valves, Process Safety Prog., William A. Freivald is the
4. Huff, J., Flow through emergency relief de- Vol. 23, No. 4, pp. 335–344, 2004. manager of process engineer-
vices and reaction forces, J. Loss Prev. Pro- ing at Valdes Engineering
cess Ind., Vol. 3, pp. 43–49, 1990. 11. Fisher, H.G., and others, “Emergency Relief Company (Phone: 630-792-
System Design Using DIERS Technology — 1886; Email: wfreivald@
5. Bird, R.B., and others, “Transport Phenom- The Design Institute for Emergency Relief valdeseng.com). He has more
ena,” pp. 481, John Wiley, New York, 1960. Systems (DIERS) Project Manual,” pp. 91, than 17 years of international
Wiley-AIChE, 1992. process design experience in
6. Darby, R., On two-phase frozen and flashing specialty chemicals, gas pro-
flows in safety relief valves: Recommended 12. Schmidt, J., and S. Egan, Case studies of cessing and refining. He holds
calculation method and the proper use of the sizing pressure relief valves for two-phase a B.S.Ch.E. from Northwest-
discharge coefficient, J. Loss Prev. Process flow, Chem. Eng. Technol., Vol. 32, No. 2, pp. ern University and is a regis-
Ind., Vol. 17, pp. 255–259, 2004. 263–272, 2009. tered professional engineer in Illinois.

56 Chemical Engineering www.che.com April 2013


Feature Report

Engineering for Plant Safety


Early process-hazards analyses can lead to potential cost savings in
project and plant operations
Sebastiano
Giardinella
and Alberto
Baumeister
Ecotek group of
companies
Mayra
Marchetti
Consultant

In Brief
CPI project lifecycle

Process Hazard
Identification

T
When to use a given
he chemical process industries (CPI) vamps and decommissioning.
method
handle a wide variety of materials, In each of the plant life phases, different
Safe-design options many of which are hazardous by na- choices are made by engineers that have a
addressing hazards ture (for example, flammable, toxic direct impact on the overall risks in the fa-
early or reactive), or are processed at hazardous cility; however, the highest opportunities for
conditions (such as high pressures or tem- cost-effective risk reduction are present in
final remarks
peratures). The risks associated with CPI the earlier phases of the project. In contrast,
facilities not only extend to the plant person- the cost of implementing changes in the later
nel and assets, but can potentially affect the stages of the project increases dramatically.
surrounding population and environment — Hence, it is important for the design team to
sometimes with consequences having re- identify risks, and implement effective design
gional or international scale, as in the case of solutions as early as possible.
toxic vapor or liquid releases. This article covers some of the typical de-
It is for this reason that process safety is cisions that the project design team has to
recognized as a key element throughout the make over the course of a project, with ex-
entire life of the plant, and several industry and amples of how the incorporation of process
professional associations and government safety throughout the entire design process
authorities have issued norms, standards and can significantly reduce the risk introduced
regulations with regards to this subject. by a new CPI facility, while also avoiding po-
Process safety, as defined by the Center tential cost-overruns, or unacceptable risk
for Chemical Process Safety (CCPS), is “a scenarios at later stages.
discipline that focuses on the prevention and
mitigation of fires, explosions and acciden- CPI project lifecycle
tal chemical releases at process facilities. A project for a new chemical process facility
Excludes classic worker health and safety usually involves different phases, which are
issues involving working surfaces, ladders, outlined here:
protective equipment and so on.” [1] Process A screening or visualization phase. In this
safety involves the entire plant lifecycle: from phase, the business need for the plant is
visualization and concept, through basic assessed. Typical choices at this stage in-
and detailed engineering design, construc- volve defining plant throughput, processing
tion, commissioning, startup, operations, re- technology, main blocks and plant location
50 Chemical Engineering www.chemengonline.com august 2015
Project Life Figure 1. The relative influence of decisions on
total life cost, and cost of implementing changes
throughout the project lifecycle

Influence of
earlier in the project lifecycle have the
decisions on greatest impact on the total plant life
total life cost
cost; in contrast, the cost of imple-
menting changes in the later stages
Cost of of the project increases dramatically,
implementing
changes
as can be seen on Figure 1.
The same holds true for overall
plant risk, as the impact of decisions
on overall facility risk is greatest in
the earliest stages of the project.

Risks and hazards


A risk can be defined by a hazard,
its likelihood (or probability) of occur-
rence, and the magnitude of its con-
sequence (or impact).
Visualization Conceptual Basic Engineering, procurement & construction Operations A hazard, as defined by the Center
engineering engineering
for Chemical Process Safety (CCPS),
(high-level), with the goal of develop- bulk materials, the execution of the is “an inherent chemical or physical
ing a high-level project profile, and a construction work, the pre-commis- characteristic that has the potential
preliminary business case based on sioning, commissioning and startup for causing damage to people, prop-
“ball-park” estimates, benchmarks of the facilities. erty or the environment” [2].
and typical performance ranges, in Table 1 shows typical engineering Process hazards can be classified
order to identify project prospects. deliverables, along with their degree in terms of the following:
A conceptual engineering phase. of completion, for each phase of 1. Their dependence on design
In this phase, the design team fur- project development. choices:
ther develops the concept of the After the plant construction is fin- • Intrinsic — not dependent on
plant, leading to a more-defined ished, the facility enters the opera- design decisions (that is, always
project description, an improved tions phase. At the end of its life, the associated with the operation or
capital-cost estimate, and a more- plant is decommissioned. process). For instance, hazards
developed business model. At this It is a generally accepted fact in proj- associated with the chemistry of
stage, the process scheme is de- ect management that decisions made the materials being handled (flam-
fined, along with the characteristics
of the major pieces of equipment and TABLE 1. Typical Engineering Deliverables and Status per Project Phase
their location on the layout (which Deliverable V CE BE DE
would ideally be set over a selected Project scope, design basis and criteria S P C C
terrain). The needs for raw materials, Soil studies, topography, site preparation S C C
intermediate and final product inven- Construction bid packages P C
tories, as well as utility requirements
Process block diagrams S/P P/C C C
are also established.
Plot plan S P/C C
A basic engineering, or front end
engineering design (FEED) phase. Process and utility flow diagrams (PFDs / UFDs) S/P P/C C
This sets the basis for the future P&IDs S P/C C
engineering, procurement and con- Material & energy, utility balances S P/C C
struction (EPC) phase, by gener- Equipment list S/P P/C C
ating a scope of work that further Single line diagrams S/P P/C C
develops the process engineering, Data sheets, specifications, requisitions S P/C C
and includes the early mechanical,
Mechanical equipment design drawings and documents S P/C C
electrical, instrumentation and civil/
Piping design drawings and documents S/P C
structural documents and drawings.
This phase also serves to generate a Electrical design drawings and documents S/P C
budget for the construction. Automation and control drawings and documents S/P C
An EPC phase. The EPC phase also Civil / structural / architectural design drawings and documents S/P C
includes the detailed engineering for Cost estimate C5 C4 C3 C2/C1
the development of the “for con- Key: V = visualization; CE = conceptual engineering; BE = basic engineering; DE = detailed engineering; S =
struction” engineering deliverables, started; P = preliminary; C = completed; C5, C4, ..., C1 = Class 5, Class 4, ..., Class 1 cost estimate (AACE)
the procurement of equipment and

Chemical Engineering www.chemengonline.com august 2015 51


Studies Scope definition Engineering, procurement & construction

Visualization Conceptual Basic Detail Precomm.,


engineering engineering engineering Construction comm & startup Operations Decommissioning

• Expert judgement • HAZID • HAZOP • HAZOP • Inspections • Preventive • HAZID


• High level risk • What-if • LOPA • Constructability • Materials and and corrective
identification • Consequence • QRA review equipment tests maintenance checks
analysis • FAT & SAT • Periodically check
• Hydrostatic tests instrument and relief-
valve calibration
• Periodic hazards
Figure 2. Typical hazards analyses that are used analysis
throughout a CPI project lifecycle

mability, toxicity, reactivity and so with the operating conditions material embrittlement
on); these properties cannot be (pressure, temperature), and ma- ❍❍ higher material inventories

separated from the chemicals terial inventories. As general rules: increase the impact of poten-
• Extrinsic — dependent on de- ❍❍ higher pressures increase the tial releases, whereas lower
sign decisions. As an example: impact of potential releases, material inventories reduce
hazards associated with heating whereas vacuum pressures response times in abnormal
flammable materials with direct increase the probability of air operating conditions
burners can be avoided by using entering the system • Equipment failures — associated
indirect heating ❍❍ higher temperatures increase with damages to plant equipment
the energy of the system (and • Utility failures — associated with
2. Their source: hazards, especially when failures in utilities supplied to the
• Process chemistry — associated near the flashpoint or self-ig- facility, such as electricity, cooling
with the chemical nature of the nition temperature), whereas water, compressed air, steam, fuel
materials (for example, flammabil- very low temperatures could or others
ity, toxicity, reactivity and
Mu?llerGmbh_Chemical so on)
Engineering e pose the risks of freezing,
86x123_2011.qxd:Mu�ll • Human activity — associated with
• Process variables — associated Chem eng 1-4pgKrytox Ad 11-9-2014.qxp_Layout 2 11/15/14 12:46 PM Page
Chemical Engineering e UC 86x123 formation
03/2011of hydrates, or activities by humans over the facil-

Ultra-Clean Fluorinated Oils, Greases,


The new cGMP-drum PTFE and Dry Film Lubricants
offers process reliability by Miller-Stephenson offers a complete line of inert high
performance fluorinated lubricants that include
validated cleaning procedures DuPont™ Krytox ® oils and greases, as well as a fami-
ly of PTFE Dry Lubricants. They provide superior
lubricity while being thermally stable, nonflammable,
non-migrating, and contain no silicone. Krytox ® lubri-
Details of the Ultra-Clean line:
– Sanitary welded edging
cants offer extreme pressure, anti-corrosion and anti-
wear properties, as well oxygen compatibility and low
– Geometry of beads and bottom
optimized for clean discharge
of product and for drum cleaning outgassing. Our PTFE creates a superior release for
– Body, base and lid in stainless plastics, elastomers and resins with multiple releases
steel AISI 316 between applications.
For technical information and sample, call 800-992-2424 or
– FDA-approved silicone elastomer
seal ring, USP Class VI
– Choose from a range 203-743-4447.
of 20 different sizes
m
TM

s
– Compliant with FDA and cGMP
guidelines

Müller GmbH - 79 618 Rheinfelden (Germany)


Industrieweg 5 - Phone: +49 (0) 76 23 / 9 69 - 0 - Fax: +49 (0) 76 23 / 9 69 - 69 Connecticut - Illinois - California - Canada
A company of the Müller group supportCE@mschem.com
info@mueller-gmbh.com - www.mueller-gmbh.com miller-stephenson.com

Circle 20 on p. 74 or go to adlinks.chemengonline.com/56200-20 Circle 19 on p. 74 or go to adlinks.chemengonline.com/56200-19


52 Chemical Engineering www.chemengonline.com august 2015
Studies Scope definition Engineering, procurement & construction

Visualization Conceptual Basic Detail Precomm.,


engineering engineering engineering Construction comm & startup Operations Decommissioning

• Define plant • Define process


• Select codes • Define • Analyze plant • Develop • Conduct inspections • Perform preventive • Assess hazards
capacity scheme and standards process and hazards and construction • Test piping and and corrective • Follow work
• Select technology • Define for design controls operability drawings materials maintenance procedures
• Define process equipment • Define basis • Define design • Identify layers • Verify plant • Perform factory • Periodically • Document and
blocks and buildings of design conditions of protection hazards and acceptance and site check instrument signal abandoned
• Decide plant location (layout) • Define • Assess risk risks acceptance tests (FAT and relief valve facilities (for example,
location • Define raw electrical area • Identify • Finalize & SAT) calibration underground piping,
materials, classification additional safeguards • Calibrate instruments • Train new ducting, and so on).
products and • Select safeguard design and relief valves operations
intermediate materials needs • Define • Perform hydrostatic personnel
product • Design/ commissioning tests • Follow work
inventories specify and startup • Train operations procedures
equipment procedures personnel • Periodically assess
• Design • Conduct hazards
buildings constructability • Repeat previous
• Define review activities for
control and revamps /
Figure 3. Typical design decisions af- emergency expansions
systems
fecting cost and risk throughout a CPI • Design
project lifecycle preliminary
relief system
ity (for example, operator error, quency or probability of occurrence. Onshore Reliability Data (OREDA),
tampering with facilities, security This likelihood can be either relative American Institute of Chemical En-
threats and so on) (low, medium, high), or quantita- gineers (AIChE), Center for Chemical
• Environmental — associated with tive (for instance, 1 in 10,000 years). Process Safety (CCPS), American
environmental conditions (for ex- Quantitative values of the likelihood of Petroleum Institute (API), U.K. Health
ample, earthquakes, hurricanes, different categories of risk, or equip- and Safety Executive (HSE), Nether-
freezing, sandstorms and so on) ment failures, as well as risk tolerabil- lands Committee for the Prevention
The likelihood of a risk can be ex- ity criteria, can be obtained from liter- of Disasters by Dangerous Materials
pressed in terms of an expected fre- ature sources, such as Offshore and (CPR), or local government agencies,

PROTECT PUMPS
DRY RUNNING • CAVITATION • BEARING FAILURE • OVERLOAD

MONITOR PUMP POWER COMPACT EASY MOUNTING


• Best Sensitivity Only 3.25" x 6.25" x 2"
• Digital Display • Starter Door • Panel
• Raceway • Wall
TWO ADJUSTABLE SET POINTS
• Relay Outputs UNIQUE RANGE FINDER SENSOR
• Adjustable Delay Timers • Works on Wide-range of Motors
• Simplifies Installation
4-20 MILLIAMP ANALOG OUTPUT

WHY MONITOR POWER INSTEAD OF JUST AMPS? PUMP POWER


VALVE CLOSING

PUMPING
POWER

AMPS

VALVE OPENING NO FLUID


No Sensitivity
Power is Linear-Equal Sensitivity For Low Loads
at Both Low and High Loads
NO LOAD FULL LOAD NO LOAD FULL LOAD

CALL NOW FOR YOUR FREE 30-DAY TRIAL 888-600-3247


WWW.LOADCONTROLS.COM

Circle 18 on p. 74 or go to adlinks.chemengonline.com/56200-18

Chemical Engineering www.chemengonline.com august 2015 53


TABLE 2. Examples of Changes in Design as Result of Process Hazards analyses in method to quantitatively assess the
different Project phases consequences of hazardous material
Impact on Project Conceptual Basic Detailed releases. Release rates are calculated
description engineering engineering engineering
for the worst case and alternative sce-
Process Industrial 1. Preliminary 2. Tower diameter, reboiler, 3. Line routings narios, end toxic points are defined,
definition solvents process design condenser, and pump changed after
manufactur- and equipment capacities changed, constructa- and release duration is determined.
ing facility characteristics spare equipment, alter- bility review, Hazard identification analysis
were defined nate lines and valves adding pres- (HAZID). HAZID is a preliminary
based on process added following What-If sure drop,
simulations and analysis. which altered
study that is performed in early proj-
best engineering pumps and ect stages when hazard material,
practices. control valves. process information, flow diagram
Plant layout / High-pressure 1. Preliminary plot 3. Relief systems design 5. Location of and plant location are known. It’s
area gas plant plan was arranged required further modifi- some lines generally used later on to perform
based on available cations to plot plan, and and equip-
terrain and recom- an additional 10% of ment was
other hazard studies and to design
mended equipment space for flare exclusion slightly the preliminary piping and instru-
spacing. area. changed as mentation diagrams (P&IDs).
2. After conse- 4. After QRA, proper safe- result of con- What-if. This is a brainstorming
quence analysis, guards were selected structability
plant area was in order to reduce risk review, to method that uses questions start-
increased by 50% contours to tolerable allow early ing with “What if...,” such as “What if
and equipment and levels in occupied build- operations in the pump stops running” or “What if
buildings were re- ings and public spaces, parallel with
located to prevent hence reducing space construction.
the operator opens or closes a cer-
impact areas from requirement by 25% tain valve?” It has to be held by ex-
reaching occupied versus that required by perienced staff to be able to foresee
buildings and pub- consequence analysis. possible failures and identify design
lic spaces.
alternatives to avoid them.
Automation Crude-oil cen- 1. Only summary de- 2. Control system designed 5. Some addi- Hazard and operability study
and controls tral process- scription of major according to P&IDs. tional modifi-
ing facilities control system 3. Approximately 30% cations were (HAZOP). This technique has been
items developed in more instruments and required after a standard since the 1960s in the
conceptual engi- control loops added as reception of chemical, petroleum and gas Indus-
neering. result of HAZOP. vendor infor-
4. The overall system was mation.
tries. It is based on the assumption
increased from SIL-1 that there will be no hazard if the
to SIL-2 after LOPA, as plant is operated within the design
result of one section of parameters, and analyzes deviations
the plant handling light
ends. of the design variables that might
lead to undesirable consequences
and they can be especially valuable factor in ensuring that plant hazards for people, equipment, environment,
when performing quantitative, or can be avoided or reduced as early plant operations or company image.
semi-quantitative studies. as possible in the design. If a deviation is plausible, its conse-
The consequence of a risk can be Aside from the experience of the quences and probability of occur-
expressed in terms of its impact on team, it is generally accepted that rence are then studied by the HAZOP
several recipients, such as assets, different methodical approaches can team. Usually an external company
personnel, society and environment. be applied in a timely manner to the is hired to interact with the operator
The combination of likelihood and engineering design process, in order company and the engineering com-
consequence defines the risk. The to detect possible hazards that were pany to perform this study. There are
risk is then analyzed versus tolerabil- not addressed by the design team. at least two methods using matrices
ity criteria, either qualitatively (for ex- These structured reviews are called to evaluate the risk (R): one evalu-
ample, in a risk matrix), or quantita- process hazards analyses (PHAs), ates consequence level (C) times
tively (for example, in risk contours). and are often conducted or moder- frequency (F) of occurrence; and the
Company management and the de- ated by a specialist, with participa- other incorporates exposition (E) as a
sign team may then select measures tion of the design team, owner’s em- time value and probability (P) ranging
to eliminate or reduce individual risks, ployees or experienced operators. from practically impossible to almost
if they are not in the tolerable range. Several methodologies exist for sure to happen, in this method, the
conducting a PHA, each suitable for risk is found by Equation (1):
Process hazards identification specific purposes, processes, and
An experienced engineering design for certain phases of project devel- R = E × P × C (1)
team, with proper design basis doc- opment and plant lifecycle (Figure 2).
umentation, and working under ap- Below is a brief description of some of Layer-of-protection analysis
proved industry standards and best the most used PHAs in the CPI. (LOPA). This method analyzes the
engineering practices, is the first Consequence analysis. This is a probability of failure of independent
54 Chemical Engineering www.chemengonline.com august 2015
protection layers (IPLs) in the event example, if a consequence analysis
of a scenario previously studied in is not performed in a conceptual or
a quantitative hazard evaluation like pre-FEED phase, important plot plan
HAZOP. It is used when a plant uses considerations can be missed, such
instrumentation independent from as the need to own more land to
operation, safety instrumented sys- avoid effects over public spaces; or
tems (SIS) to assure a certain safety the fact that the location might have
integrity level (SIL). The study uses a a different height with respect to sea
fault tree to study the probability of level than surrounding public places
failure on demand (PDF) and assigns impacted by a flare plume. PROVEN
a required SIL to a specific instru- Some other studies, like HAZOP,
mentation node. For example in pe- cannot be developed without a con- PERFORMANCE
troleum refineries, most companies trol philosophy or P&IDs, and are
will maintain a SIL equal to or less performed at the end of the FEED or
than 2 (average probability of failure detailed engineering (for best results,
on demand ≥10−3 to <10−2), and a at the end of both) to define and vali-
nuclear plant will tolerate a SIL 4 (av- date pressure safety valves (PSVs)
erage probability of failure on demand location and other process controls ROTOFORM
≥10−5 to <10−4). and instrument safety requirements.
Fault-tree analyses. Fault-tree anal- QRA or LOPA (or both) are done GRANULATION
ysis is a deductive technique that after HAZOP to validate siting and FOR PETROCHEMICALS
uses Boolean logic symbols (that is, define safety instrumented systems AND OLEOCHEMICALS
AND or OR gates) to break down SIL levels, and finally meet the level
the causes of a top event into basic required by the plant.
equipment failures or human errors. Figure 2 shows the typical CPI
The immediate causes of the top project phases, with a general indi-
event are called “fault causes.” The cation of when it is recommended
resulting fault-tree model displays the to conduct each study; however,
logical relationship between the basic this may vary depending on the
events and the selected top event. specific industry, corporate prac-
Quantitative risk assessment tices, project scope and execution
(QRA). QRA is the systematic de- strategy. AIChE’s CCPS [2] has an High productivity solidification of
velopment of numerical estimates of Applicable PHA technique table that products as different as resins, hot
the expected frequency and conse- indicates which study to perform in melts, waxes, fat chemicals and
quence of potential accidents based each project phase, which also in- caprolactam has made Rotoform® the
on engineering evaluation and math- cludes research and development granulation system of choice for
ematical techniques. The numerical (R&D), pilot plant operations, and chemical processors the world over.
estimates can vary from simple val- other phases not covered in the Whatever your solidification
ues of probability or frequency of an present article. requirements, choose Rotoform for
event occurring based on relevant Table 2 includes some real-life ex- reliable, proven performance and a
premium quality end product.
historical data of the industry or amples of how the results of some of
other available data, to very detailed these studies can impact the devel-
 High productivity –
frequency modeling techniques [4]. opment of the plant design at differ- on-stream factor of 96%
The events studied are the release ent project phases.  Proven Rotoform technology –
of a hazardous or toxic material, ex- Out of the previously mentioned nearly 2000 systems installed
plosions or boiling liquid expanded studies, a properly timed HAZOP, in 30+ years
vapor explosion (BLEVE). The results at the end of the basic engineering  Complete process lines or
of this study are usually shown on phase, is key to identifying safety retrofit of existing equipment
top of the plot plan. and operability issues that have been  Global service / spare parts supply
Failure mode and effects analysis overlooked by the engineering de-
(FMEA). This method evaluates the sign team, especially when involving
ways in which equipment fails and the an experienced facilitator and plant
system’s response to the failure. The operators in the study, given that
focus of the FMEA is on single equip- they have a fresh, outsiders’ view
Sandvik Process Systems
ment failures and system failures. of the project, and they can provide Division of Sandvik Materials Technology Deutschland GmbH
input on daily operating experience. Salierstr. 35, 70736 Fellbach, Germany
Tel: +49 711 5105-0 · Fax: +49 711 5105-152
When to use a given method Also, the deviations identified in the info.spsde@sandvik.com

Some studies have more impact HAZOP can serve to detect the need www.processsystems.sandvik.com
in some phases than in others. For for additional safeguards that were
Circle 26 on p. 74 or go to adlinks.chemengonline.com/56200-26

Chemical Engineering www.chemengonline.com august 2015 55


SANDVIK_Chemical_ad_55.6x254_MASTER.indd 1 09/02/2015 14:48
TABLE 3. Additional Costs of Changes Associated with HAZOP Recommendations The design options include, in de-
during EPC Phase
scending order of reliability: inherently
Sample Project description Estimated cost of changes PHAs and proper safer design, engineering controls
project associated with HAZOP safe-design practices
number recommendations in EPC phase implemented in previous (passive and active) and administra-
(as % of approved budget) design phases? tive controls (procedural).
1 Gas dehydration unit 3% Yes Inherently safer design involves
2 Gas compression unit 3% Yes avoiding or reducing the likelihood of
3 Crude oil atmospheric 1% Yes
a hazard in a permanent or insepa-
unit rable fashion. For example, when de-
4 Fuel storage tank farm 2% Yes signing a centrifugal pump discharge
5 Petrochemical plant relief 2% Yes
system, an inherently safer design
and flare systems would be to specify the design pres-
6 Crude oil dehydration 1% Yes sure at the centrifugal pump shut-off
station pressure, thereby largely reducing
7 Crude oil evaluation 1% Yes the risk that an increase in the pump
facilities discharge pressure (for example, due
8 Heavy crude oil dehydra- 3% Yes to a blocked outlet) could cause a
tion unit rupture in the pipes with consequent
9 Propane/air injection 1% Yes loss of containment.
plant Engineering controls are features
10 Oil pipeline + two gas 1% Yes incorporated into the design that re-
compression units duce the impact of a hazard without
11 New flare system in ex- 1% Yes requiring human intervention. These
isting refinery
can be classified as either passive
12 Refinery gas concentra- 7% No (not requiring sensing and or active
tion unit revamp
response to a process variable) or
13 Extra-heavy oil deasfalt- 5% No
ing unit
active (responding to variations in
process conditions). In the previ-
14 Demineralized water 13% No
plant ous centrifugal pump example, a
15 Hydrogen compression 35% No
passive solution would be to con-
unit tain possible leaks within dikes, and
with adequate drainage. Examples
not considered by the design team. to identify design improvements to of active solutions could be: a) pro-
When the recommendations are prevent process hazards in the lat- viding a high-pressure switch asso-
implemented correctly, and no other ter project phases. ciated with an interlock that shuts
changes to the process or plant are the pump down; and b) providing
done between the preparation of the Safe-design options a pressure safety valve (PSV) de-
basic engineering design book and At the early project phases, it is not signed for blocked outlet.
the EPC phase, then a HAZOP sig- possible to identify all possible risk- Administrative controls require
nificantly reduces the probability of reduction measures that could be human intervention. These are the
significant cost impacts in the latter included in the design. However, a least reliable, because they depend
as a result of changes due to addi- safety-oriented design team might on proper operator training and re-
tional PHAs. be able to pinpoint sources of project sponse. In the previous example, an
Even though “what-if,” HAZID risk due to lack of data, and oppor- administrative control would be to
and consequence analyses have tunities for risk reduction that could require operators to verify that the
impact on the capital cost of the be evaluated in later stages, as the valves in the pump discharge lines
project, the cost of implementing design progresses and further details are open.
their modifications to the design are are known. Throughout the engineering
typically included on the EPC bid- Some large organizations have col- phases leading to the EPC phase,
ding process, as they are realized lected the pool of their experiences different safe-design choices can be
at the beginning of the project life- within risk checklists and proprietary made, as further information is made
cycle. Fault-tree analysis and LOPA design standards, thus paving the available. Figure 3 shows some of
are used to define the redundancy way for future work. Where organiza- the typical design choices made by
level of controls and instrumenta- tions have not established their own the engineering team throughout
tion. The changes derived from standards and engineering practices, a chemical process plant lifecycle,
these studies generally represent a the design team should look for ac- which have direct impact on lifecycle
minor portion of the total capital ex- cepted codes and standards that are cost and risk.
penditure. That leaves HAZOP and the result of best engineering prac- In the visualization phase, safety
QRA as the most important studies tices in a particular field or industry. can be included in the analysis as a
56 Chemical Engineering www.chemengonline.com august 2015
factor to decide key items, such as terial inventories to the minimum and $200 million.
production technology and plant lo- required to maintain adequate The projects are divided into two
cation. These key items are typically surge/storage capacity and flex- categories: a) projects where the
selected based on other technical ibility in shipping. design contractor applied best en-
criteria, such as overall efficiency, In the basic engineering or FEED gineering standards and employed
production cost, or vicinity to either phase, many design choices are PHAs at optimum points during the
raw materials, or markets (or export made over the specific mechani- conceptual engineering and FEED
facilities). For instance, when select- cal, piping, electrical, automation phases; and b) projects where ad-
ing a technology, health, safety and and civil design that impact on the equate PHAs and safe-design prac-
environmental concerns could be overall facility risk. The first deci- tices were not applied in the previous
included as a criteria on the evalua- sion involves selecting the codes design phases.
tion matrix, by adding positive points and standards that will be used As can be seen in Table 3, there
to technologies that reduce risks to for design, and defining the design is a significant difference between
their environment by using less-toxic basis and criteria for each engineer- the cost of the changes arising from
materials, operating at lower pres- ing discipline. Then, throughout the HAZOP recommendations when
sures or temperatures, or yielding design, some other decisions may proper safe-design practices and
non-toxic byproducts. When se- include: selecting between auto- PHAs were applied during the FEED
lecting a high-level plant location, mated and manual operation, set- phase, and when they were not.
management could opt to locate ting equipment and piping design For the first category, changes
the plant away from large population conditions, defining the electrical were typically in the range of 1 to
centers, in order to minimize risks to area classification, designing or 3%. In the upper end of this cate-
communities. In this case, planning specifying equipment, structures gory, changes were higher when the
authorities also have an important and buildings, defining control and owner requested some minor modi-
role in defining allowable land-uses. emergency systems (including ap- fications to the FEED design without
In the conceptual engineering phase, propriate redundancy, where appli- properly assessing the risks associ-
safety can be included in the analysis, cable), and designing appropriate ated with said changes.
for example, in the following ways: relief systems, among others. Then, As an example, the heavy crude
1. Defining a simple, yet functional there are equipment and system- oil dehydration unit (Project 8) was
process scheme, as relatively specific hazards and available safe- designed according to best engi-
simple processes have less equip- guards that need to be considered. neering practices, and adequate
ment and consequently lower fail- Ref. 2 contains a comprehensive list analyses (HAZOP, LOPA) were
ure probability (this can conflict of hazards and safeguards for vari- conducted during the engineering
with other design goals); also, the ous types of unit operations. phase. However, the owner de-
types of equipment selected can When hazards have been prop- cided to implement changes in the
have an important effect on pro- erly identified and addressed in the design in order to compress the
cess safety (for example, selecting earlier design phases, this reduces schedule, by removing several long-
indirect over direct heating). the probability of significant costly lead items that included emergency
2. Including safety concerns in the changes being made during the shutdown system (ESD) valves and
early layout definition. For instance, EPC phase as a result of unsafe components, without updating the
a design by blocks — keeping the process conditions. PHAs. With the unit in operation,
main process, storage, and utility the owner asked the contractor to
areas separate from each other — Addressing hazards early include the ESD items that were in
can reduce overall risk. Other good When hazards are identified, and the original design.
practices include: maintaining an proper design choices are taken For the second category, changes
adequate separation between early in the engineering design to ad- exceeded 5%, and in one case
pieces of equipment; separating dress them, significant benefits can reached as high as 35% of the ap-
product inventories taking into ac- be obtained. proved budget. Below is a descrip-
count their flammability, toxicity or Table 3 compares the additional tion of what went wrong in each of
reactivity, and considering dikes cost of changes arising from recom- these projects:
around tanks containing danger- mendations made during a HAZOP The refinery gas concentration unit
ous materials; placing flares and at the EPC phase. The costs are revamp (Project 12) FEED consid-
vents in locations separate from expressed as a percentage of the ered hand operations in key pieces
human traffic, taking into account budget that was approved dur- of equipment. As a result of a HAZOP
wind direction (for example, so that ing the bidding stage, of projects during the EPC, the operations had
flames or plumes are directed far- of different scope and plant type, to be automated, which changed
ther from personnel or population); executed by different companies the equipment specifications and
and allowing sufficient plot space in different countries, including the design. The number of loops added
for an adequate exclusion area. U.S. and Latin America, with ap- after the HAZOP exceeded the ca-
3. Keeping flammable and toxic ma- proved budgets between $5 million pacity of the controller, and another

Chemical Engineering www.chemengonline.com august 2015 57


one had to be installed. 4. American Petroleum Institute (API) Recommended
Practice (RP) 752, Management of Hazards Associ-
The extra-heavy oil deasphalting unit ated with Location of Process Plant Permanent Build-
(Project 13) was designed during the ings, 3rd ed., 2009.
basic engineering phase as a mostly 5. U.S. Environmental Protection Agency (EPA), “Risk Man-
hand-operated facility, with minimum agement Program Guidance For Offsite Consequence
Now Available: supervisory controls. As a result of a
Analysis,” March, 2009.
6. EPA, Chemical Emergency Prevention & Planning News-
Chemical Engineering HAZOP during the EPC, the risk was letter, Process Hazard Analysis, July – August, 2008.
not tolerable to the owner, and the 7. Occupational Safety and Health Administration (OSHA)
Features Report whole unit had to be automated. 29 CFR 1910.119. Process Safety Management of
Guidebook-2014 The demineralized water plant Highly Hazardous Chemicals.
(Project 14) was delivered by the ven-
dor as a package unit, and no PHAs Authors
were conducted by the vendor. When Sebastiano Giardinella is
received, the plant had many safety the vice president and co-owner
of the Ecotek group of companies
Chemical Engineering’s Feature and operability issues and a number (The City of Knowledge, Bldg.
of important modifications had to 239, 3rd floor, offices A and B,
Reports provide concise, factual be made, including: additional lines, Clayton, Panama City, Republic of
Panama; Phone: +507-203-
information that aids in solving block and control valves, relief valves 8490; Email: sgiardinella@
real problems. The practical, and associated lines, among others. ecotekgrp.com). He has experi-
ence in corporate management,
Aside from the costs associated with
how-to orientation of these the changes, the project was delayed
project management, project engineering and process
engineering consulting in engineering projects for the
articles ensures that they can by six months. chemical petrochemical, petroleum-refining, oil-and-
gas and electrical power-generation industries. He is a
be directly applied to chemical The hydrogen compression unit certified project management professional (PMP), has
(Project 15) basic engineering design a M.Sc. in renewable energy development from Heriot-
engineers’ daily jobs. Each article Watt University (Scotland, 2014), a master’s degree in
did not address all of the safety con-
applies to a relatively broad siderations associated with hydro-
project management from Universidad Latina de Pan-
amá (Panama, 2009), and a degree in chemical engi-
section of the chemical process gen handling. Some of the modifica- neering from Universidad Simón Bolívar (Venezuela,
tions recommended by the HAZOP/ 2006). He is also professor of project management at
industries as opposed to focusing Universidad Latina de Panamá, and has written a num-
LOPA studies during the EPC phase ber of technical publications.
on just one niche. included changing the compressor Mayra Marchetti is a senior
specification, and increasing the SIL process engineer, currently working
of the SIS from SIL-1 to SIL-3. as independent consultant (Coral
Springs, Fla.; Email: mmarchetti@
ecotekgrp.com), with more than ten
Final remarks years of experience in the oil-and-
Hazards are present in the CPI; gas, petrochemical, petroleum-re-
fining and pharmaceutical indus-
some are avoidable, while others tries, and has participated in the
cannot be separated from the plant, development of conceptual, basic
and detail engineering projects. She specializes in pro-
as they are tied to the very nature cess simulation, plant debottlenecking and optimization,
of the chemicals or the unit opera- and relief systems design. She has a master’s degree in
ts tions, or both. However, a proper engineering management from Florida International Uni-

r e R ep or n design team, one that is trained to


versity (Florida, 2008), and a degree in chemical engi-

FeatuCompilati
o identify hazards, and address them
neering from Universidad de Buenos Aires (Argentina,
1996). She has published articles and delivered world-

2014 using the best engineering practices wide seminars focused in the use of simulation tools for
the process industry.
in safe-design from early on in the
Alberto Baumeister is the
project lifecycle, along with prop- CEO and co-owner of the Ecotek
erly timed and executed PHAs, can group of companies (same address
be very valuable in avoiding costly as above; Email: abaumeister@
ecotekgrp.com). He has experience
changes during the EPC phase, or in corporate management, project
even worse: potential damages to management, and senior process
persons and the environment. n consulting in engineering projects
for the chemical, petrochemical,
Edited by Gerald Ondrey petroleum-refining, oil-and-gas,
25139
electrical power-generation and agro-industrial indus-
tries. He has a specialization in environmental engineer-
References ing (gas effluents treatment) from the Universidad Miguel
1. Center for Chemical Process Safety (CCPS), “Guidelines de Cervantes (Spain, 2013), a master’s diploma in water
Find this and other related for Investigating Chemical Process Incidents,” 2nd edi- treatment management from Universidad de León (Spain,
tion, CCPS, AIChE, New York, N.Y., 2003. 2011), a specialization in management for engineers at
reference material at Instituto de Estudios Superiores de Administración (Ven-
2. CCPS, “Guidelines for Engineering Design for Process
store.chemengonline.com Safety,” 2nd ed., CCPS, AIChE, New York, N.Y., 2012. ezuela, 1990), and a degree in chemical engineering
from Universidad Metropolitana (Venezuela, 1987). He
3. AACE International Recommended Practice No. 18R- was professor of the Chemical Engineering School at
97, Cost Estimate Classification System – As Applied Universidad Metropolitana between 1995 and 2007, and
in Engineering, Procurement, and Construction for the has written a number of technical publications.
Process Industries.
58 Chemical Engineering www.chemengonline.com august 2015
Feature Report

Managing SIS Process


Measurement Risk and Cost
With a focus on flowmeters, this article shows how advances in measurement
technologies help safety system designers reduce risk and cost in their safety
instrumented systems (SIS) design and lifecycle management
Craig McIntyre
and Nathan
Hedrick
Endress+Hauser

IN BRIEF
RISK SOURCES FOR SIS

MAINTAINING LOW
FAILURE RISK

EXTENDING PROOF-TEST
INTERVALS

TRACEABLE
CALIBRATION
VERIFICATION

REDUNDANT
REFERENCES

LIFECYCLE
MANAGEMENT TOOLS

S
DETECTING PROBLEMS
uccessful implementation and man- FIGURE 1. Flowmeters like the one shown here can play key
CONCLUDING REMARKS roles in reducing risks with safety instrumented systems
agement of a safety instrumented (SIS)
system (SIS) requires designers and
operators to address a range of ISA 84 (International Electrotechnical Com-
risks. First among these involves the specifi- mission; Geneva, Switzerland; www.iec.ch
cation of a proven measurement instrument, and International Society for Automation;
such as a flowmeter (Figure 1), and its proper Research Triangle Park, N.C.; www.isa.org),
installation for a given application, an under- which provides “good engineering practice”
taking that is fundamental to achieving the guidance for SIS development and manage-
initial targeted risk reduction. ment. The emerging IEC 61511 Edition 2 in-
Second is the definition of the support re- troduces some changes to these guidelines,
quired to keep the flowmeter (or other mea- strengthening emphasis on the requirements
surement subsystem) available at that tar- for end users to collect reliability data to
geted level of risk reduction throughout the qualify or justify specifications and designs.
life of the SIS equipment. The support for the This article shows how to address those
flowmeter must be defined in the design and risks and describes several tools, capabilities
implementation phase. and procedures that can be considered for
Third involves following the recommen- designing and managing a SIS installation in
dations found in the standard IEC 61511/ flow-measurement applications.
50 CHEMICAL ENGINEERING WWW.CHEMENGONLINE.COM AUGUST 2016
Risk sources for SIS 3.0E-02
Under IEC 61511-ANSI/ISA 84, op-
erators and SIS designers are re- 2.5E-02
quired to qualify the appropriateness

Probability of Failure on Demand


of a SIS measurement subsystem to 2.0E-02
be effective in addressing an appli- Coriolis flowmeter A - 73 FIT
SIL 1
cation-specific safety instrumented
1.5E-02
function (SIF). This not only includes
the initial design of the SIS itself, but
1.0E-02
the qualification of the measurement Typical SIL capable coriolis flowmeter B - 160 FIT
subsystem used in that service.
SIL 2
The capture and assessment of 5.0E-03

data is used to qualify the use of


measurement instruments in SIS ap- 0.0E+00 SIL 3

plications. Even after this qualifica- 0 1 2 3 4 5 6 7 8


Years
9 10 11 12 13 14 15

tion, operational data and manage-


ment of change of these instruments FIGURE 2. Flowmeters with a lower “dangerous undetected” (ƛdu) FIT and in-situ testing capabilities may
over their lifetimes in SIS applications allow extension of the interval time needed for proof tests
must still be captured and assessed.
SIS measurement subsystems Maintaing low PFD and du determined and executed to keep
are typically exposed to challenging Risk of failure to perform an expected both the probability of failure on de-
process and environmental condi- function can come from probabilis- mand (PFD) average and the lambda
tions, so they tend to contribute a tic failure sources. For example, this dangerous undetected (du; the fail-
higher risk to the availability of the includes the collective probabilistic ure rate for all dangerous undetected
SIS than safety controllers, which failures of electronic components in failures) fault risk (that is outside the
are normally installed in a controlled a transmitter. Required maintenance reach of diagnostics) below a re-
Chemical Eng.1-4pgKrytoxA Ad 2014.qxp_Layout
environment. and2 11/15/14
proof-test 12:49 PM Page 1must be quired average risk-reduction target.
procedures

Krytox ® Lubricants Prefabricated piping and spools


Greases and Oils are: Chemically Inert. Insoluble in common
solvents. Thermally stable (-103°F to 800°F). Nonflammable. The use of sophisticated welding techniques, BUTTING Group
Nontoxic. Oxygen Compatible, Low Vapor Pressure. Low Germany · Brazil ·
the processing of austenitic steels and special Canada · China
Outgassing. No Migration - no silicones or hydrocarbons.
DuPontTM Krytox® Lubricants offer Extreme Pressure, steel grades and the production of stainless Marcel Bartels
Anticorrosion and Anti-wear properties. Mil-spec, Aerospace steel pipes have always been our core
Phone: +49 5834 50-7155
marcel.bartels@butting.de
and Food Grades (H-1 and H-2) available! Widely used in
Vacuum and Cleanroom environments. competences. www.butting.com

We also offer a complete line of inert fluorinated Dry Lubricants


and Release Agents. We also offer customized solutions for piping
For technical information, call 800.992.2424 or 203.743.4447 systems, vessels and special components in
(8AM - 4PM ET).
nickel alloys, duplex and superduplex
Channel Partner Since 1991 materials and clad materials.
m
TM

s Take advantage of our know-how and


contact us.

Connecticut - Illinois - California - Canada


supportCE@mschem.com Progress by Tradition!
miller-stephenson.com/krytox

Circle 31 on p. 82 or go to adlinks.chemengonline.com/61498-31 Circle 07 on p. 82 or go to adlinks.chemengonline.com/61498-07

CHEMICAL ENGINEERING WWW.CHEMENGONLINE.COM AUGUST 2016 51


Prefabricated_piping_&_spools_Chem_Eng_86x123mm.indd 1 05.07.2016 13:57:42
National time Secondary Counter/ increase in PFD over time.
time
standard standard timer
Flow standard
For example, some Coriolis flow-
Flowmeter
International National
(calibration rig) meters have du values in the
mass mass Reference range of 150 to 178 failure in time
standard standard mass
(FIT, where 1 FIT= 1 failure in a bil-
FIGURE 3. The figure shows a traceability chain for a mass flowmeter lion hours). Others, such as two-wire
Coriolis flowmeters, have du val-
Risk of failure to perform an ex- tic coverage, but also redundancy ues in the 73 to 89 FIT range. Vortex
pected function can also come from — meaning the testing functions are flowmeters with du in the 70 to 87
systematic failure sources. This redundant and continuously check- range are also available. If all other
could include damage to a sen- ing each other. This redundancy pro- factors were equal, a measurement
sor while being tested, for example. vides a number of benefits for the subsystem with half the FIT value
Systematic fault risk may be created lifecycle management of instruments could allow a doubling of the proof-
by properties of the process fluids, used in a SIS. test interval time (Figure 2).
operating conditions, build-up, cor- Some measurement subsystems
rosion or other factors. Periodic vi- Extending proof-test intervals offer the capability to remotely invoke
sual field inspections, calibrations Periodic proof-testing of the SIS in-situ proof testing with a high de-
and maintenance that may need to and its measurement subsystems gree of proof-test coverage to reduce
be conducted can introduce failure is required to confirm the contin- the PFD subsystem contribution.
risk. There is some measure of risk ued operation of the required SIF, Given that external visual inspec-
from (and to) personnel who need to and to reduce the probability of tions are sufficient for at least some
follow written procedures to conduct dangerous undetected failures that proof-test events, these measure-
activities in the field and work with are not covered by diagnostics. A ment instruments might be proof-
instruments that may need to be re- proof-test procedure for a flowme- tested in-situ without the need to
moved, transported, repaired, tested ter or other measurement devices remove the instrument from service.
and reinstalled. often requires removal of the instru- Data from these proof-tests can
It has been stated by one of the ment and its wiring, transportation be transmitted via 4–20-mA HART
world’s largest chemical compa- to a testing facility, and reinstallation connections from the instrument to
nies that “2% of every time we afterward. In some cases, modern and through some safety control
have human intervention, we cre- instrumentation may provide the ca- systems to a digital network, such
ate a problem.” Another leading pability to conduct proof testing in- as Ethernet/IP, where these data
specialty chemical company con- situ, thus eliminating the removal of can be captured. In short, the proof-
ducted a study that concluded “4% equipment and risk of wiring, instru- testing event can be invoked, and
of all devices (instruments) that are ment or equipment damage. related data can be captured, man-
proof-tested get damaged during re- Safety Integrity Level (SIL)-ca- aged and reported through safety
installation.” Reducing the need for pable measurement subsystems control systems supporting these
personnel to physically touch a mea- typically have hardware and soft- capabilities.
surement subsystem offers design- ware assessments conducted dur- In-situ proof testing can create
ers an avenue to reduce systematic ing their development to determine documented evidence that diagnos-
failure risk to a SIS. failure mode effects and diagnostic tic checks have been carried out,
IEC 61511 Edition 2 points to the analysis and to manage change and thereby fulfill the requirements
need to specify in the safety require- processes according to IEC 61508- for documentation of proof-testing,
ments specification (SRS) the meth- 2, 3. The du and proof-test cov- in accordance with IEC 61511-1,
ods and procedures required for test- erage values, among other safety Section 16.3.3b, “Documentation
ing SIS diagnostics. SRS clause 10 parameters, are provided in a safety of proof testing and inspections.”
states some of the requirements for function manual and described in a When in-situ proof testing can be
proof-test procedures — including certificate. Lower du values give engineered into an SIS design, cost
scope, duration, state of the tested system designers greater freedom may be reduced compared to the
device, procedures used to test the when setting measurement subsys- expense of periodically removing
diagnostics, state of the process, tem proof-test intervals, because the instrument from service to per-
detection of common cause failures, these intervals contribute a lower form testing.
methods and prevention of errors.
Flowmeter
Measurement subsystems from Measured
Sensor Transmitter
several instrument suppliers are now True mass flow
mass flow AV Analog- AV AV (measurand)
available with integral redundant self- Sensing
Transducer to-digital
Signal Data
element processing display
testing diagnostics that can conduct converter
continuous availability monitoring.
This means a measurement subsys- AV1 auxiliary variable

tem may not only have high diagnos- FIGURE 4. The diagram illustrates the relationship among the various subsystem elements of a flowmeter

52 CHEMICAL ENGINEERING WWW.CHEMENGONLINE.COM AUGUST 2016


(Control of monitoring and measur-
AV1
Maximum permissible error (MPE) [mA] ing devices), which states: “Where
Measurement error [%]

necessary to ensure valid results,


AV2
measuring equipment shall…be
[mV] calibrated or verified at specified
intervals, or prior to use, against
measurement standards traceable
AV3
[Hz] to international or national mea-
Flow [kg/h] Tolerance interval surement standards.”
Some measurement instruments
(a) (b)
provide certified integral and re-
FIGURE 5. All measurements results from a particular instrument need to be within the band between the dundant references that have been
measuring error of the instrument and the maximum permissible error for the verification to be consid-
ered positive (AV = auxiliary variable) calibrated via accredited and trace-
able means, and can thus have their
Traceable calibration verification age during removal, transport and measurement calibration verified in-
Measurement subsystem proof-test reinstallation. There is also a risk in- situ. This eliminates sources of risk
procedures often require calibra- troduced for unrealized damage or and cost associated with removing
tion verification of the measuring in- the introduction of an error due to instruments from service, while still
strument. As operators seek to set process shutdowns, which are often meeting ISO 9001:2008 Clause
proof-test intervals, they also need required when an instrument is re- 7.6a requirements.
to set associated intervals for cali- moved from service.
bration verification. The measurement subsystem Redundant references
Verification and documentation may need to be calibrated or veri- Appointed with the task of coordinat-
to prove that the SIS subsystem fied with traceability to an inter- ing the realization, improvement and
calibration is acceptable normally national standard. If an organiza- comparability of worldwide mea-
requires removal of the subsystem. tion is ISO 9001:2008-certified, surement systems, the International
This exposes the instrument to dam- it needs to address Clause 7.6a Bureau of Weights and Measures

Superior check valves should . . .


Save space
Arrive on time
Withstand rigorous conditions
®
Check-All the above

A superior check valve does all the right


things. By offering a wide selection of spring See us at
settings and seat materials, Check-All® spring WEFTEC
Booth #8521
loaded check valves are assembled to your
exact needs, ensuring absolute precision and
reliability. They work like they should. That’s
what makes Check-All® the only choice. Plus,
most lead times are less than one week. All
the better!

Get me a Check-All® ! Manufactured in


• 515-224-2301
West Des Moines, Iowa, USA
• www.checkall.com • sales@checkall.com
Circle 11 on p. 82 or go to adlinks.chemengonline.com/61498-11 Circle 24 on p. 82 or go to adlinks.chemengonline.com/61498-24

CHEMICAL ENGINEERING WWW.CHEMENGONLINE.COM AUGUST 2016 53


Flowmeter Safety controller/ Lifecycle management/ Figure 4 illustrates the basic con-
logic solver data
subsystem cept and the relation among subsys-
Hart communication tem elements in a flowmeter.
Namur NE107
During the lifecycle of any instru-
ment, it is important to monitor mea-
4-20mAdc LOOP surement performance on a regular
Namur NE43
Device Lifecycle basis (ISO 9001:2008 Chapter 7.6.a),
management
Ethernet/IP
especially if the measurements from
Serial number F the instrument can significantly im-
pact process quality.
Serial number F For example, in Figure 4, the pro-
cess value is defined as mass flow,
Field device and a traceable flow calibration sys-
management
tem can be used to perform a proof
test. Typically, the outcome of this
test is seen in calibration certificates
Serial number F as a graph depicting the relative
FIGURE 6. Cloud- or enterprise network-based lifecycle management tools can provide support docu- measuring error of the instrument
mentation for specific instruments and the maximum permissible error
band. All of the measurement results
(Sèvres, France; www.bipm.org) de- put signal representing the value of are expected to be enclosed within
fines traceability as “the property of a the primary process variable being this band for the verification to be
measurement result to be related to measured (that is, mass flow). considered positive (Figure 5a).
a reference through a documented 2. Auxiliary variable: Signal(s) coming A second approach (Figure 5b)
unbroken chain of calibrations, each either from the instrument’s sensor consists of assessing the function-
contributing to the measurement un- (transducer) or a certain element of ality of an instrument by looking at
certainty.” Figure 3 shows a trace- the transmitter, such as an analog- one or more elements that can sig-
ability chain for a flowmeter. to-digital (A/D) converter, ampli- nificantly impact the process value.
The term “measurement result” fier, signal processing unit and so In this case, verification can assist in
can be used in two different ways to on. This variable is often used to assessing the instrument’s function-
describe the metrological features of transmit current, voltage, time, fre- ality by observing the response of
a measuring instrument: quency, pulse and other informa- the process variable and the auxiliary
1. Measurand (Process Value): Out- tion. variables. The auxiliary variables are
compared to specific reference val-
Current ranges for signal of digital transmitters ues to make sure they are within a
tolerance interval established by the
manufacturer.
Typically, proof testing requires
Failure information Failure information
the flowmeter to be removed from
A
Measurement information
M A the process line and examined with
specific equipment, such as a mobile
mA
calibration rig or a verification unit.
0 3.6 4 20 21 This rig or unit needs to be main-
3.8 20.5 tained and calibrated by qualified
personnel, thus introducing a costly
Current ranges for signal identification in process control systems
and time-consuming procedure.
The process has to be shut down to
perform testing, often resulting in a
Measurement information M
loss of production. If removal and re-
installation of the flowmeter are car-
A:=1 A:=0 A:=0 A:=1 ried out in a hazardous area, safety
issues can arise.
mA Modern instruments, such as
0 3.6 4 20 21 mass flowmeters, typically have in-
3.8 20.5 situ proof testing built into the de-
A =Alarm state (i0,1); M=measurement (analog mA value) vices. While many instrument ven-
dors have similar solutions, there are
FIGURE 7. NAMUR NE43 recommendations for 4–20-mA d.c. transmitters (top) and process control sys- significant differences in how they
tems (bottom) address the risk of mixing different vendor-specific current range signal levels work. In the cases where flowmeter
54 CHEMICAL ENGINEERING WWW.CHEMENGONLINE.COM AUGUST 2016
Status signal Color Symbol drift does occur, it must be detected the frequency of an oscillation
Normal; valid output signal n n immediately. The stability of the ref- (such as the rate of capacitance
erences can be addressed with du- swings by the differential switched
Maintenance required; still n
valid output signal rable and high-quality components. capacitor sensor in vortex flow-
Potential drift can be detected by the meters)
Out of specification; signal out n
of the specified range use of an additional, redundant refer- In flowmeter models where re-
ence, so that each can cross-check dundant references are in place,
Function check; temporary n
non-valid output signal with the other. If one or both refer- observing both references drifting si-
Failure; non-valid output signal n
ences drift out of tolerance, these multaneously in the same manner is
cross-checks can trigger an alarm. very unlikely. On an installed base of
FIGURE 8. Five standard status states are speci- Redundancy of the references is 100,000 flowmeters, such an event
fied by the NAMUR NE 107 recommendation
achieved differently depending upon is anticipated to occur just once
the measurement technology: every 148 years. Put another way, a
hardware and its associated soft- • Electromagnetic flowmeters use device with a typical lifecycle of 20
ware can conduct in-situ testing, the voltage references because the years would have only a 0.007%
approach is often different as well. primary signal generated by the probability of experiencing such a
For example, the authors’ company sensor is a voltage induced by the drift during its life.
embeds the verification functionality conductive fluid passing through a Independent, third-party verifica-
in the device electronics of the flow- magnetic field tion of a particular redundant-refer-
meter, so removal of the flowmeter is • Coriolis, vortex, and ultrasonic ences approach can be obtained by
not required. flowmeters use frequency genera- organizations such as TÜV Rhein-
A key requirement for this type of tors (digital clocks) as references land AG (Cologne, Germany; www.
verification method is high reliability. because the primary signals are tuv.com), and verification reports
The internal references used to verify measured either by a time period thus obtained can satisfy the need to
the auxiliary variables must remain (the phase-shift in a mass flowme- document the approach.
stable and avoid drift during the ter or the time-of-flight differential In practice, a verification report
service life of the instrument. And if in an ultrasonic flowmeter), or by from an independent, third-party or-

I NNOVATION
Low Liquid Rate Technology
Engineered for demanding, very low liquid flow distillation
Many applications, such as vacuum distillation, fatty acid, tall oil, vitamin production and
natural gas dehydration, operate at very low liquid rates. Spreading the liquid evenly
across the tower, which is difficult to do at low liquid rates, is important for the
good liquid distribution needed to achieve superior packing efficiency.

Koch-Glitsch has developed a line of liquid distributors specifically for very


low liquid rate applications. When combined with high performance
FLEXIPAC® and INTALOX® structured packings, these distributors can help
optimize separations in many demanding distillation services.
• Minimize liquid entrainment
• Maximize usable packing surface area
• Flows as low as 0.03 gpm/ft2 [70 liters/h/m2]
• Baffle design helps wet every sheet in the structured packing
• Secondary troughs act as flow multipliers to enhance the spreading of the liquid
• Decrease in number of distribution points results in larger, more fouling resistant orifices

YOU CAN RELY ON US.™

United States (316) 828-5110 | Canada (905) 852-3381 | Italy +39-039-6386010 | Singapore +65-6831-6500
For a complete list of our offices, visit our Web site.
www.koch-glitsch.com
For related trademark information, visit http://www.koch-glitsch.com/trademarks.
This technology is protected by one or more patents in the USA. Other foreign patents may be relevant.

Circle 29 on p. 82 or go to adlinks.chemengonline.com/61498-29

CHEMICAL ENGINEERING WWW.CHEMENGONLINE.COM AUGUST 2016 55


ganization constitutes the front end lifecycle events, such as systematic though the logic solver with the flow-
of an unbroken, documented chain and probabilistic failures in the mea- meter via 4–20-mA HART to moni-
of traceability. Since the internal ref- surement subsystems. Subsystems tor the device, to invoke in situ proof
erences remain valid over the lifetime can be repaired and replaced, and testing and calibration functions,
of the instrument, their own factory then the correct configurations can and to diagnose problems. The
calibration, performed in accredited be uploaded, reducing time and risk field-device-management tool com-
facilities and documented, is the next of errors. municates via Ethernet/IP to a life-
link in this chain. At least one field device manage- cycle management server installed
In addition, a traceable calibration ment tool follows the Field Device within the user’s network or the
of the instrument ensures that the Tool (FDT) standard from the FDT Cloud, where all flowmeter data are
integrity of the device has not de- Group (Jodoigne, Belgium; www. stored in accordance with ISA and
teriorated during assembly or han- fdtgroup.org), which provides a uni- IEC standards. The flowmeter data
dling in the plant. Calibration of the fied structure for accessing mea- are synchronized and maintained
equipment used for calibration in the surement subsystem parameters, from the flowmeter along with all as-
factory can then be traced back to configuring and operating them, and sociated data via its serial numbers.
national standards. diagnosing problems. A logic solver The goal of this kind of field-de-
In-situ verification is therefore com- with HART I/O and HART pass- vice-management software is to en-
pliant with international standards for through management capabilities able plant operators to design a sys-
traceable verification. can allow such a tool to work with tem that provides the following:
the measurement subsystem to in- • Device power and wiring condi-
Lifecycle management tools voke in-situ proof testing and trace- tion monitoring through the logic
ISA 84 and IEC 61511 — in partic- able calibration verification. solver or safety controller
ular, edition 2 clause 11 of the IEC Some field-device-management • Device primary current loop/sec-
document — require end users to tools can be used with device lifecy- ondary HART communication and
collect reliability data to qualify or cle management tools to aid in sub- status management through the
justify specification and design data. system-related data support access logic solver
According to these documents, data and capture. These tools can also be • Device repair/replace manage-
quality and sources: integrated with overall lifecycle-man- ment through the logic solver
• Shall be credible, traceable, docu- agement tools. • Device proof testing management
mented and justified Several instrument suppliers pro- through the logic solver
• Shall be based on the field feed- vide, populate and maintain a real- • Device traceable verification of
back existing on similar devices time Cloud- or enterprise-based calibration management through
used in similar operating device lifecycle-management tool the logic solver
environments connection for individual device-spe- • Capture and management of de-
• Can use engineering judgment to cific support documentation, certifi- vice proof testing, calibration, and
assess missing reliability data or cates, history, changes and calibra- other lifecycle data that may re-
evaluate the impact on reliability tion information. duce risk and cost in SIS designs
data collected in a different oper- For example, Figure 6 illustrates and lifecycle management
ating environment one possible configuration. In this
Collecting reliability data for SIS case, information flows between a Detecting problems
is costly, but lifecycle management flowmeter subsystem through a logic A typical SIL-capable instrument,
tools are available to reduce the risk solver (safety controller) to a field- such as a flowmeter, connects to
and required time for some of these device-management tool and device the logic solver or safety controller
activities. Several vendors offer life- lifecycle-management software in via 4–20-mA or 4–20-mA HART.
cycle management tools that can the Cloud or on a local server. These signals are also used to indi-
work externally or through the safety In this example, the flowmeter cate problems.
system environment. They can also and logic solver both use NAMUR Current signals per NAMUR NE 43
capture lifecycle events, such as NE 43-recommended current loop recommendations (Figure 7) convey
systematic and probabilistic failures. signal settings to reduce systematic measurement and failure information
If anomalies are detected, SIS com- risk from mixing the different vendor- from the flowmeter to the safety con-
ponents can be repaired or replaced. specific current loop signal levels. troller via the 4–20-mA loop. Most
The right configurations can then be Also, the flowmeter and logic solver every instrumentation and control
uploaded, reducing required time both use standard HART Commu- system supplier offers options to
and risk of errors. nication commands including the support this standard. Essentially,
Field device management tools NAMUR NE 107 recommendation, any flowmeter and logic solver that
can work externally or through the which provides five clear actionable follows the NAMUR NE 43 recom-
safety system environment to invoke subsystem status indicators. mendation uses 4–20 mA for the
subsystem proof-testing and cali- In the case of the author’s em- measurement, and signals of less
bration verification, and to capture ployer, the FDT communicates than 3.6 mA or greater than 20.5
56 CHEMICAL ENGINEERING WWW.CHEMENGONLINE.COM AUGUST 2016
mA to indicate failures. The benefit Concluding remarks Authors
of following this practice is reducing Implementation of a SIS requires Nathan Hedrick is the flow prod-
the risk of mixing different instrument process risk protection to a targeted uct marketing manager at
Endress+Hauser (2350 Endress
vendor-specific signal level variations minimum while maintaining design Place, Greenwood, IN. 46143;
with different safety controller sig- and lifecycle costs at a reasonable Phone: 1-888-363-7377; Email:
nal level settings — something that level. Intelligent instruments and life- nathan.hedrick@us.endress.
com). Hedrick has more than six
could happen during a repair or re- cycle management tools can help years of experience consulting on
placement event. process plant personnel reduce risks process automation. He graduated
from the Rose-Hulman Institute of
Figure 8 shows the five stan- and costs associated with a SIS sys- Technology in 2009 with a B.S.Ch.E. He began his ca-
dard status states specified by the tem. They can also aid in capturing reer with Endress+Hauser in 2009 as a technical sup-
NAMUR NE 107 recommendation. reliability data. port engineer. In 2014, Hedrick became the technical
support team manager for flow, where he was respon-
The NE 107 recommendation is now Instrumentation suppliers who sible for managing the technical support team covering
implemented within many HART- serial-number their components are the flow product line. He has recently taken on his cur-
enabled devices for standard sta- able to provide operators a realtime rent position.
tus communication. Under NE 107, Cloud- or enterprise-based connec-
problems are identified as normal, tion between the measurement device Craig McIntyre is the chemical
industry manager with
failure, out-of-specification, mainte- in the field and serial number-based Endress+Hauser (same address
nance required and function check. support documentation, certificates, as above; Phone: 1-888-363-
The purpose of NE 107 is to alert history, changes and calibration in- 7377; Email: craig.mcintyre@
us.endress.com). McIntyre has
systems and operations personnel in formation. These data are maintained held several positions with
an actionable way if a problem ex- by the supplier for the user. Additional Endress+Hauser over the past 17
ists. When the logic controller sees user data can be captured, including years. They include level product
manager, communications prod-
a NE 107 status indication change, service history. This can help reduce uct manager and business development manager. Prior
it notifies the operator. Field-device- the time required to obtain needed in- to joining E+H, he was director of marketing for an Em-
erson Electric subsidiary. McIntyre holds a B.S. degree
management tools can be used to formation, as well as reduce the risk in physics from Greenville College and an MBA from the
provide additional
CIC-10307 diagnostic
halfp page ad.qxd data to
3/25/07 of6:19
usingPMthePage
wrong1 information. n Keller Graduate School of Management.
help identify specific problems. Edited by Scott Jenkins

PLASTIC CONTROL VALVES FOR


ALL YOUR CORROSIVE APPLICATIONS
Collins plastic control valves are
highly responsive control valves
designed for use with corrosive
media and/or corrosive atmos-
pheres.
Collins valves feature all-plastic
construction with bodies in PVDF,
PP, PVC and Halar in various body
styles from 1/2" - 2" with Globe,
Angle or Corner configurations and
many trim sizes and materials.
Valves may be furnished without
positioner for ON-OFF applications.
Call for more information on our
plastic control valves.

P.O. Box 938 • Angleton, TX 77516


Tel. (979) 849-8266 • www.collinsinst.com

Circle 14 on p. 82 or go to adlinks.chemengonline.com/61498-14

CHEMICAL ENGINEERING WWW.CHEMENGONLINE.COM AUGUST 2016 57


Feature Report

Column Instrumentation
Basics An understanding of instrumentation is valuable in
evaluating and troubleshooting column performance
Ruth R. Sands
DuPont Engineering
Research & Technology

I
nstrumentation is critical to un- '5
derstanding and troubleshooting
all processes. Very few engineers '5
specialize in this field, and many
learn about instrumentation through
'5
experience, myth and rumor. A good
understanding of the various types of &SSPS
instrumentation used on columns is a
valuable tool for engineers when evalu- '5
ating column performance, starting up
new towers or troubleshooting any type
of problem. This article gives an over-
view of the common types of instru- Figure 1. Which flowmeter is the most Figure 2. Flush-mounted diaphragm
accurate? What is the source of error in pressure transmitters are common in
ments used for pressure, differential the material balance? low-temperature services
pressure, level, temperature and flow.
A discussion of their accuracy, common 1. Based on the material balance, the as in scrubbers and storage tanks.
installation problems and troubleshoot- engineer concluded that the bottoms The process diaphragm, an integral
ing examples are also included. flowrate must be in error and wrote part of the transmitter, is mounted
The purpose of this article is to pro- a work order to have the flowmeter on a nozzle directly on the vessel, and
vide some basic information regarding recalibrated. The instrument group the transmitter is mounted directly
the common types of instrumentation disagreed heartily. By the end of this on the nozzle.
found on distillation towers so that article, the reader will understand the
process engineers and designers can instrument group’s response. Remote-seal diaphragm
do their jobs more effectively. Used in higher temperature service
Pressure when the electronics must be mounted
Introduction There are three common types of pres- away from the process, a flush-mounted
Anyone trying to complete a simple sure transmitters: flush-mounted diaphragm is installed on a nozzle at
mass balance around a column under- diaphragm transmitters, remote-seal the process vessel. A capillary tube
stands that process data contain some diaphragm transmitters and impulse- filled with hydraulic fluid connects the
error. Closing a mass balance within line transmitters. All use a flexible flush-mounted diaphragm to a second
10% using plant data is usually consid- disk, or diaphragm, as the measuring diaphragm, which is located at the re-
ered very good. Generally, some values element. The deflection of the flexible motely mounted pressure transmitter.
must be thrown out when matching a disk is measured to infer pressure. The hydraulic fluid must be appropri-
model to plant data. Understanding The diaphragm can be made of many ate for the process temperature and
which measured plant data is likely different materials of construction, pressure. Hydraulic fluid leaks will
to be most accurate is invaluable in but the disk is thin and there is little lead to errors in measurement. Cali-
making good decisions about a model tolerance for corrosion. Coating of the bration is complex because the head
of the plant, column performance and diaphragm leads to error in the mea- from the hydraulic fluid must be con-
future designs. surement. The instrument accuracy sidered. The calibration changes if the
The following is a real case and a of all three types of pressure trans- transmitter is moved, the relative po-
telling example of how little the aver- mitters is similar, usually 0.1% of the sition of the diaphragms changes or if
age chemical engineer may understand span, or calibrated range. the hydraulic fluid is changed.
about instrumentation. A process engi-
neer with over 20 years of experience Flush-mounted diaphragms Impulse-line
was doing a material balance around a These pressure transmitters are com- Impulse-line pressure transmitters
distillation tower, illustrated in Figure mon in low-temperature services, such can either be purged or non-purged.
48 Chemical Engineering www.che.com March 2008

20_CHE_031508_FR_KT.indd 48 2/26/08 11:16:10 AM


DEFINITIONS
Instrumentation range • Best-in-class performance with 0.025% accuracy
The instrumentation range, the scale over which the instrument • ±0.10% reference accuracy
is capable of measuring, is built into the device by the manu- • ±0.065% of span
facturer. The purchaser defines the desired measured range, These examples refer to the ideal instrument accuracy, which is
and the vendor should provide a device that is appropriate for only the accuracy of the measuring device itself. The total ac-
the application. curacy, on the other hand, includes the instrument accuracy plus
all other factors that contribute to error in the measured reading
Calibrated range as compared to the actual value. These other factors can include
The calibrated range is the scale over which the instrument is digital to analog conversions, density errors, piping configura-
set to measure at the plant. It is a subset of the instrument range. tions, calibration errors, vibration errors, plugging and more.
The calibration has a zero and a span. The zero is the minimum
reading, while the span is the width of the calibrated range. Turndown ratio
The calibrated range will simply be referred to as the range at The ratio of the maximum to minimum accurate value is an impor-
a plant site. tant factor in considering the total accuracy of a measured value.

Instrument accuracy maximum accurate value


Turndown ratio =
Error minimum accurate value
Accuracy  v100%
Scale of Measurement
For example, an instrument with 100:1 turndown and 0–100-
The instrument accuracy is published by the manufacturer in the psi instrument range would have the stated instrument accuracy
product documentation, which is easily obtained on-line. A few down to 1 psi. Below 1 psi, the instrument might read, but it will
examples of how accuracy can be expressed are: have greater inaccuracy. ❏

Purged impulse-line pressure trans- different type of pressure transmitter. strument mechanics. The operating
mitters measure purge-fluid pressure Adequate freeze protection on the im- range must be clearly defined; for ex-
to infer the process pressure. Most pulse lines is also important to obtain ample, is the range 100-mm Hg vac-
commonly, the purge fluid is nitro- accurate measurements. uum, 100-mm Hg absolute, or 650-mm
gen, but it can also be air or other Example 1. A good example of a prob- Hg absolute? Using different measure-
clean fluids. The purge fluid is added lem with impulse-line pressure trans- ment scales in the same plant is con-
to an impulse line of tubing to detect mitters can be found in Kister’s Distil- fusing, and it can make it very hard
pressure at the desired point in the lation Troubleshooting [2]. Case Study for mechanics to calibrate the pressure
process. The purge fluid enters the 25.3 (p. 354), contributed by Dave transmitters accurately.
process and must be compatible with Simpson of Koch-Glitsch U.K., de- Another issue is measuring the
it. Check valves are required to en- scribes three redundant impulse-line relief pressure. The system designer
sure that process material does not pressure transmitters used to mea- must consider the instrument ranges
back up into the purge-fluid header. sure column head pressure. Following available and the accuracy of the
The system must be designed so that a tray retrofit, operating difficulties measurement for the operating range
the pressure drop through the im- eventually led to suspicion of the head versus the relief pressure range. It is
pulse line is negligible. A pressure pressure readings. The impulse lines good practice to install a second pres-
transmitter measures the purge-fluid and pressure transmitters had been sure transmitter on vacuum towers to
pressure with a diaphragm to infer moved during the turnaround. The measure the relief pressure.
the process pressure. transmitters had been moved below Example 2. An excellent example of
the pressure taps on the vessel. Con- calibration problems is illustrated in
Non-purged, impulse-line densate filled the impulse lines and vacuum service in Reference [2]. Case
Rather than a purge fluid, this type caused a false high reading. Relocat- Study 25.1 (p. 348), contributed by
of pressure transmitter uses process ing the transmitters to the original Dr. G. X. Chen of Fractionation Re-
fluid. Usually, this style is chosen location above the nozzles solved the search, Inc., describes several years
when the process is non-fouling or problem by allowing condensate to of troubleshooting a steam-jet system
it is undesirable to add inerts to the drain back into the tower. in an attempt to achieve 16-mm Hg
process. One example is a situation absolute head pressure on a tower. It
where emissions from an overhead Transmitters in vacuum service was eventually determined that the
condenser vent must be minimized. Pressure transmitters in vacuum ser- calibration of the top pressure trans-
An impulse line is connected from vice are generally the most problem- mitter was wrong, and they had been
the desired measurement point in atic, leading to greater inaccuracy in pulling deeper vacuum than they
the process to a pressure transmitter, the measured value. Damage to the thought. The top pressure transmitter
which measures the process pressure diaphragm can occur from exceeding was calibrated using the local airport
at the remote point. The system must the maximum pressure rating of the barometric pressure, which was nor-
be designed so that the pressure drop instrument. Often, this happens on malized to sea-level pressure and was
through the impulse line is negligible. startup, or it can happen when per- off by 28-mm Hg.
The system designer must consider forming a pressure test of the vessel.
the safety implications of an impulse- The diaphragm deflects permanently Differential pressure
line failure. The consequence of releas- and introduces error. Differential pressure can be measured
ing hazardous material from a tubing Calibration of vacuum pressure either with a differential pressure (dP)
failure may warrant the selection of a transmitters is more difficult for in- meter or by subtracting two pressure
Chemical Engineering www.che.com March 2008 49

20_CHE_031508_FR_KT.indd 49 2/26/08 11:18:36 AM


Feature Report

3FCPJMFS
-5 SFUVSO

Figure 3. (left) Remote-seal Figure 6. Non-contact radar level


diaphragm pressure transmitters transmitters generate waves that are
are used in high-temperature reflected from the surface of the level
service back to the transmitter
Figure 5. Nuclear level transmit-
Figure 4. (above) Location of ters are non-contact devices
reboiler return nozzle does not Example 1. A column in a high-tem-
allow for accurate level reading perature, fouling service began to ex-
perience high pressure drop, and the
measurements. Subtracting two pres- ters are the most common type of plant engineers were concerned that
sure readings is not always accurate level instrument, and an accurate they were flooding the column. Calcu-
enough to obtain a meaningful mea- density is required to convert the dP lations showed that the tower should
surement, so it is important to consider reading to a level reading. In many not be flooding if the trays were not
the span of the anticipated measured cases, froth in the liquid level de- damaged. Downcomer flooding was a
readings. If the dP is a substantial frac- creases the actual density and causes possibility if the cartridge trays had
tion of the top pressure, then it is okay faulty readings. Changes in composi- become dislodged and reduced the
to subtract the readings of two pres- tion or the introduction of a different downcomer clearance. The tower was
sure transmitters. However, if the dP process feed with a different density taken down, and internal inspection
is a small fraction of the top pressure, are cited several times as reasons for revealed no damage to the internals.
then it will be within the instrument level measurement problems. Plug- It was determined that a false low
error of the pressure transmitter. ging of impulse lines and equipment level caused the bottoms flow control-
For example, a column at a plant runs arrangements that make accurate ler to close. This raised the level in the
at 30 psia top pressure. The expected readings impossible are also very tower above the reboiler return line
dP is 2-in. H2O over a few trays. The in- common problems. and above the lower column pressure
strument error for a 0–50 psi pressure Differential pressure transmitters tap. The column dP meter was reading
transmitter is 1.4-in. H2O. The mea- are the most common type of level the height of liquid above the lower-
surement is within the accuracy of the transmitter. The accuracy of the in- column pressure tap. Consultation
pressure transmitters, and a dP meter strument is quite good, at 0.1% of with the instrument manufacturer
is the appropriate meter to obtain an span (calibrated range). Any type of revealed that the remote, seal hydrau-
accurate measurement. The downside dP meter can be used: flush-mounted lic fluid was not appropriate for the
of dP meters is that very long impulse diaphragms, remote-seal diaphragms, high temperature of the process. The
lines are required on tall towers. purged impulse-line, or non-purged hydraulic fluid was boiling in the cap-
impulse-line pressure transmitters. illary tubes and had deformed the dia-
Level The level measurement is dependent phragm, which was also coated from
Level and flow are the hardest basic on the density of the fluid: the fouling service. The level transmit-
things to measure on a distillation P ter was switched to a periodic, purged
tower. Kister reports that tower base  height of liquid, ft impulse-line dP meter. An automated
Rl
level and reboiler return problems high-flow nitrogen purge prevents ac-
rank second in the top ten tower mal- An accurate density is required for cumulation of the solids in the impulse
functions, citing that “Half of the case calibration. Changes in composition or lines and is done once per shift. Logic
studies reported were liquid levels ris- the introduction of a process feed with was added to the control loop to main-
ing above the reboiler return inlet or a different density will cause errone- tain the previous level reading during
the bottom gas feed. Faulty level mea- ous readings. Level transmitters suf- the short nitrogen purge, a method
surement or control tops the causes fer from the same problems that occur that has eliminated the problem with
of these high levels...Results in tower in pressure transmitters. Hydraulic the level.
flooding, instability, and poor separa- fluid leaks, compatibility of the hy- Example 2. Another common ex-
tion...Vapor slugging through the liq- draulic fluid, damage to diaphragms, ample of a level transmitter failure
uid also caused tray or packing uplift and plugging or freezing of impulse is based on the fact that equipment
and damage.” (Reference 2, p. 145) lines are just a few of the problems is designed in such a way that an ac-
One of the main reasons for faulty that can be encountered with dP level curate level reading can never be ob-
level indications is that dP me- transmitters. tained. Though this may be surpris-
50 Chemical Engineering www.che.com March 2008

20_CHE_031508_FR_KT.indd 50 2/26/08 11:19:43 AM


Figure 8. Resis-
tive temperature de-
tectors respond to a
temperature change
with a change in
resistance
Figure 7. Guided wave radar level
transmitter on a distillation
tower level [5]
figure 9. For
vapor or gas applica-
tions, orifice flowme-
ing, it is mentioned in Ref. [3], Case Radar level ters require tempera-
Study 8.4 (p. 149), as illustrated in transmitters ture and pressure
Figure 4. A column that was being This type of level transmit- compensation
retrofitted was originally designed ter has been used in the
so that the reboiler return was intro- chemical processing in-
duced directly between the two liq- dustries (CPI) for the last
uid-level taps. The level in the tower 30 years. They demonstrate high ac- the accuracy of the measurement. In-
could never be accurately measured, curacy on oil tankers and have been strument accuracy is ±0.1% of span.
and it was modified on the retrofit to used frequently in storage-tank appli- Example 3. A reflux tank on a batch
rectify this situation. cations. Radar level transmitters are distillation tower had a non-contact
now being applied to distillation tow- radar level transmitter. The tower
Nuclear level transmitters ers but are still more commonly found stepped through a series of water
Common in polymer, slurry and on auxiliary equipment, like reflux washes, solvent washes, and process
highly corrosive or fouling services, tanks. There are contact and non-con- cuts. The reflux-tank level transmit-
these instruments work by placing a tact types of radar level instruments. ter gave false high readings during the
radioactive source on one side of the A non-contact, radar level transmit- solvent wash cycle, which used tolu-
vessel and a detector on the other side. ter generates an electromagnetic wave ene. The reflux pumps would always
The amount of radiation reaching the from above the level being measured. gas off during this part of the process.
detector depends on how much mate- The wave hits the surface of the level The dielectric constants of the various
rial is inside the vessel. A strip source and is partially reflected to the instru- fluids in the reflux tank, of which tolu-
and strip detector are more accurate ment. The distance to the surface is ene had the lowest dielectric constant,
than a single source, strip detector. A calculated by measuring the time of varied ten times during the cycle, af-
sketch of a single source, strip detector flight, which is the time it takes for the fecting the height of liquid able to be
is shown in Figure 5. The advantage reflected signal to reach the transmit- measured. Larger antennas focus the
of nuclear level transmitters is that ter. Some things that cause inaccuracy signal more and give greater signal
they are non-contact devices, making with non-contact radar are: size of the strength. As the dielectric constant de-
them ideal for services where the pro- cone, heaving foaming, turbulence, creases, a larger antenna is required to
cess fluid would coat or damage other deposits on the antenna, and varying measure the same height of fluid. The
types of level instruments. dielectric constants caused by changes level transmitter used in this service
Nuclear level transmitters are more in composition or service. The instru- was not appropriate for all measured
expensive than other level devices. ment accuracy is reported as ±5 mm. fluids and could not accurately mea-
They also require permits and a radia- Contact radar sends an electromag- sure the liquid level when the reflux
tion safety officer, so they are often only netic pulse down a wire to the vapor- drum was inventoried with toluene.
used as a last resort. The instrument liquid interface. A sudden change in
accuracy is generally ±1% of span. The the dielectric constant between the Temperature
total accuracy depends on how well the vapor and the liquid causes some of There are two common types of tem-
system was understood by the designer the signal to be reflected to the trans- perature transmitters in distillation
and installer. The thickness of the ves- mitter. The time of flight of the re- service — thermocouples and Resis-
sel walls and any other metal protru- flected signal determines the level. tive Temperature Devices (RTDs).
sions in the measuring range, such as Guided wave radar can be used for Both are installed in thermowells.
baffles, must be taken into account in services where the dielectric constant Thermocouples. The most popular
the calibration, along with the correct changes, but is not a good fit for fouling temperature transmitter, thermocou-
rate of decay of the source. Build-up of services. A bridle (Figure 7), is used on ples, consist of two wires of dissimilar
solids in the measuring range will also distillation towers to reduce turbulence metals connected at one end. An elec-
result in error. and foaming and therefore increases tric potential is generated when there
Chemical Engineering www.che.com March 2008 51

20_CHE_031508_FR_KT.indd 51 2/26/08 11:20:19 AM


0SJGJDF1MBUF4RVBSF3PPU'VODUJPO


Feature Report 


'MPXSBUF HQN


is a temperature delta between the

joined end and the reference junction.

Type J thermocouples, made of iron

and Constantine, are commonly used
 #FMPXPGTDBMF FSSPSJTIJHIGPS
in the CPI for measuring tempera- GMPXTFOTPSBOEEQNFBTVSFNFOU
tures under 1,000°C. 
          
0SJGJDFQMBUFQSFTTVSFESPQ JO)0

RTDs
The second most-common type of tem- Figure 10. Volumetric flowrate is proportional to Figure 11. Due to impulse
the square root of the ∆P, causing high error at less line problems, this "clean" ser-
perature transmitter, RTDs consist of than 10% of span vice did not meet standards
a metal wire or fiber that responds to
a temperature change by changing The rest of the column responded to equipment arrangement for an orifice
its resistance. Though RTDs are less the change in boilup in about 3 min. flowmeter with temperature and pres-
rugged than thermocouples, they are The lag in the control point caused sure compensation.
also more accurate. Typically, they are cycling of the steam flowrate and cre- Typical turndown for orifice plates is
made of platinum. The instrument ated an unstable control loop. The 10:1. Below 10% of span, the measure-
accuracy of thermcouples and RTDs cause was determined to be a thermo- ment is extremely erroneous because
is very good in both. However, ther- couple that was too short for its ther- the volumetric flowrate is proportional
mocouples have a higher error than mowell. Normally, thermocouples are to the square root of the ∆P. At 10% of
RTDs. The total accuracy of a thermo- spring-loaded to ensure that the tip span, the meter is only measuring 1%
couple is 1–2°C. There is greater error is touching the end of the thermow- of the ∆P span (Figure 10).
due to calibration errors and cold-ref- ell, but the instrument mechanics had Multiple meters can be used to
erence junction error. installed a thermocouple of the wrong overcome the turndown ratio when
It is important to note that, with length because they lacked the proper high accuracy is required over the
temperature transmitters, there is replacement part. The poor heat entire span. This is often worth the
a lag in the dynamic response to transfer through the air gap between effort when measuring the flowrate
changes in process temperatures. All the end of the thermocouple and the of raw materials or final products. At
temperature measurements have a thermowell caused the delay in tem- one plant, three orifice plates in par-
slow response, because the mass of perature response. Replacing the in- allel were used to measure the plant-
the thermowell must change in tem- stalled thermocouple with one of the boundary steam flowrate due to the
perature before the thermocouple or proper length fixed the problem. large span and the accuracy required
RTD can see the change. The lag time at the low end of the range. This re-
will depend on the thickness of the Flow sulted in a very complicated system.
thermowell and on the installation. There are many different types of There are many common problems
The thermocouple and RTD must be flowmeters. Here, the types commonly that lead to error in the orifice plate
touching the tip of the thermowell for used in plants will be discussed: orifice measurement, including inaccurate
best performance. If there is an air plates, vortex shedding meters, mag- density, impulse-line problems, erosion
gap between the thermowell and the netic flowmeters and mass flowmeters. of the orifice plate, and an inadequate
measuring device, the heat-transfer number of pipe diameters upstream
resistance of the air will add substan- Orifice plates and downstream of the orifice plate.
tially to the lag time, which is also Orifice plates are the most common An accurate density is required to
why temperature transmitters work type of industrial flowmeter. They are obtain an accurate flowrate. In a plant
better in liquid service. The response inexpensive, but they also have the that has a process feed that varies
time for temperature transmitters greatest error of all the common types from as low as 12% to as high as 30%
in liquid service is between 1–10 s, of flowmeters. Orifice plates measure water, the density changes signifi-
whereas the response time for tem- volumetric flowrate according to the cantly, and therefore an orifice meter
perature transmitters in vapor service following equation: will not provide an accurate reading
is about 30 s. Heat-transfer paste is a 1
¥ $P ´ 2 without density compensation.
thermally conductive silicone grease; Q  C v¦ µ Impulse line problems include
it has been used with success in some § R ¶ plugging, freezing due to loss of elec-
plants to improve the response time of Q is the volumetric flowrate, C is a con- tric heat tracing, and leaking. Con-
temperature transmitters. stant, ∆P is the pressure drop across densate filling the impulse lines in
Example. The plant in this example the orifice, and ρ is the fluid density. vapor/gas service and gas bubbles
experienced a temperature lag prob- To obtain an accurate flowrate, an ac- in the impulse lines in liquid service
lem. A thermocouple near the bottom curate fluid density must be known. are also commonly cited. Figure 11
of a large tower controlled the steam Temperature and pressure compen- shows a pipe just upstream of an ori-
to the reboiler. The temperature con- sation are required for vapor or gas fice that was in “clean” water service
trol point had a 10-min delayed re- applications and may be required for two years. There was a filter just
sponse to changes in steam flowrate. for some liquids. Figure 9 shows the upstream of this section of pipe. The
52 Chemical Engineering www.che.com March 2008

20_CHE_031508_FR_KT.indd 52 2/26/08 11:21:43 AM


Figure 13. The magnetic flowmeter prin-
ciple states that the voltage induced across
Figure 12. Vortex meters contain a shedder bar that creates vortices down- a conductor as it moves at right angles
stream when fluid flows past it (left). Depending on the application and pipe size, through a magnetic field is proportional to its
vortex shedding meters are available in a range of sizes and shapes (right) velocity.

impulse lines to the orifice plate flow- the net effect unpredictable. There- for straight runs of pipe upstream and
meter were completely plugged. This fore, under average operating condi- downstream of the meter vary, but
section of pipe was removed and a tions, a typical orifice installation can both are usually longer than for orifice
Teflon-lined magnetic flowmeter was be expected to have an overall inac- plates. In general, 30 pipe diameters
installed instead. curacy in the range of 2 to 5% AR (ac- are required upstream and 15 pipe
Orifice plates can erode, especially tual reading)” [6]. diameters downstream. The upstream
in vapor service with some entrained and downstream piping must be the
liquid. This is common in steam ser- Vortex shedding meters same size pipe as the meter.
vice, and orifice plates should be Vortex shedding meters contain a There are only a few problems com-
checked every three years for wear. bluff body, or a shedder bar, that cre- monly encountered with vortex me-
Orifice plates generally need 20 ates vortices downstream of the object ters. Older models may be sensitive to
pipe diameters upstream and 10 pipe when a fluid flows past it. The meters building vibrations, but newer models
diameters downstream of the orifice utilize the principle that the frequency have overcome this issue. If the shed-
plate for the velocity profile to fully of vortex generation is proportional der bar becomes coated or fouled, the
develop for predictable pressure-drop to the velocity of the fluid. The whis- internal vibration sensor will cease to
measurement. This requirement var- tling sound that wind makes blowing work. This can be avoided by using an
ies with the orifice type and the piping through tree branches demonstrates external vibration sensor. The most
arrangement. This is rarely achieved the same phenomenon. common issue is failing to meet the
in a plant, which introduces error in The fluid’s density and viscosity are Remin requirements over the desired
the measurement. used to set a “k” factor, which is used span. At one plant, every vortex meter
The instrument accuracy of orifice to calculate the fluid velocity from was line-sized, which means it was
plates ranges from ±0.75–2% of the the frequency measurement. The fre- the same size as the surrounding pip-
measured volumetric flowrate. Vari- quency, or vibration, sensor can either ing. The flow went into the laminar
ous problems are encountered with be internal or external to the shedder region in the desired measured range
orifice plate installations, and they bar. The velocity of the fluid is con- in every case. The flow read zero when
have the highest error of all flowme- verted to a mass flowrate using the it transitions to laminar, making the
ters. “Orifice plates are, however, quite fluid density. Therefore, accurate fluid meters useless.
sensitive to a variety of error-induc- density is important for accurate mea- Example. Another good example of
ing conditions. Precision in the bore surements. Vortex meters work well failing to meet the Remin require-
calculations, the quality of the instal- both in liquid and gas service. They ments over the desired span hap-
lation, and the condition of the plate are commonly used in steam service pened on a project where a tower that
itself determine total performance. because they can handle high tem- had been out of service for some time
Installation factors include tap loca- peratures. They are available in many was recommissioned. The distillate
tion and condition, condition of the different materials of construction and flowrate was substantially lower than
process pipe, adequacy of straight can be used in corrosive service. the original tower design and was in
pipe runs, gasket interference, mis- Vortex meters have lower pressure the laminar flow region over the en-
alignment of pipe and orifice bores, drop and higher accuracy than orifice tire operating range. The distillate
and lead line design. Other adverse plates. A minimum Reynolds num- flow was a major control point on the
conditions include the dulling of the ber (Remin) is required to achieve the tower, but the vortex meter could not
sharp edge or nicks caused by corro- manufacturer’s stated accuracy. Vortex read the flowrate. The control strat-
sion or erosion, warpage of the plate meters exhibit non-linear operation egy had to be changed to work around
due to water hammer and dirt, and as they transition from turbulent to this issue until an appropriate meter
grease or secondary phase deposits laminar flow. Typical accuracy above could be installed.
on either orifice surface. Any of the the Remin is 0.65–1.5% of the actual
above conditions can change the ori- reading. In general, the meter size Magnetic flowmeters
fice discharge coefficient by as much must be smaller than the piping size Faraday’s law states that the voltage
as 10%. In combination, these prob- to stay above the Remin throughout induced across any conductor as it
lems can be even more worrisome and the desired span. The requirements moves at right angles through a mag-
Chemical Engineering www.che.com March 2008 53

20_CHE_031508_FR_KT.indd 53 2/26/08 11:22:36 AM


Feature Report

netic field is proportional to the " " "


velocity of that conductor. This
#
is the principle used to measure # #
velocity in magnetic flowmeters,
which are commonly referenced
to as mag meters (Figure 13).
Mag flowmeters measure the    /PGMPX
volumetric flowrate of conduc-
tive liquids. Fluids like pure Figure 14. Mass flowmeters use the Coriolis effect to infer mass flowrate from the mea-
organics or deionized water do surement of flowtube deflection
not have a high enough conduc-
tivity for a mag meter. An ac-
curate density is required to convert flowmeters have the highest ac- the new mass flowmeter had enough
the volumetric flowrate to a mass curacy of all the different types of additional pressure drop to force the
flowrate. The meters are line-sized, flowmeters, usually ±0.1–0.4% of the liquid level into the condenser tubes
but they have a minimum and maxi- actual reading. The measurement is and restrict rates — an expensive les-
mum velocity to achieve the stated independent of the fluid’s physical son for a new engineer.
instrument accuracy. A smaller line properties, making mass flowmeters Example 2. Another tower had a
size may be necessary to achieve the unique in that most flowmeters re- mass flowmeter installed on the bot-
velocity requirements throughout quire the fluid density as an input. toms flow, which was pumped but
the desired span. The instrument Mass flowmeters are insensitive to not cooled. The mass flowmeter al-
accuracy is quite good, generally upstream and downstream pipe con- ways had erratic readings and was
at ±0.5% of the actual reading. The figurations. Practical turndown is never believed. A closer examination
error is very high below the mini- 100:1, although the manufacturers of the system revealed enough pres-
mum velocity. Turndown for newer claim 1,000:1. The density measure- sure drop through the mass flowme-
mag meters is 30:1, but older models ment is not as accurate as a density ter to result in flashing in the flow-
will be closer to 10:1. meter. Mass flowmeters are gener- tube. The two-phase flow caused the
Mag meters do not have a lot of op- ally very reliable and only require erratic readings.
erating problems. They must be liq- periodic calibration to zero them.
uid-full to get an accurate reading and Mass flowmeters are on the expen- Epilogue
are often placed in vertical piping to sive end to purchase and to install. With a knowledge of the basics of
achieve this. They rarely plug as they They require 110-V power. Pressure column instrumentation, the ques-
can be specified with Teflon liners and drop can sometimes be an issue, and tion posed in the introduction should
are often used in slurry service. Mag the meters are only available in line seem trivial. Our experienced en-
flowmeters are more expensive to sizes up to 6 in. Coating of the inside gineer had concluded that the bot-
install because they usually require of the flowtube will result in higher toms flowrate of the column had to
110-V power. pressure drop and can result in loss of be erroneous, but the instrument
range and accuracy if the tube is re- group had disagreed. The flowmeter
Mass flowmeters stricted. Wear and corrosion can result in question was a mass flowmeter in
Mass flowmeters use the Coriolis effect in a gradual change of the mechanical relatively clean and non-corrosive
to measure mass flowrate and density. characteristics of the tube, resulting service. The other three flowmeters
A very small oscillating force is applied in error. Zero stability was an issue on the column were orifice plates
to the meter’s flowtube, perpendicular with older meters but this problem and are known to have a myriad of
to the direction of the flowing fluid. The has been solved in newer units. problems that introduce error.
oscillations cause Coriolis forces in the Example 1. The reflux flowrate on
fluid, which deform or twist the flow- a final product column was an im- Summary
tube. Sensors at the inlet and outlet of portant measurement, and the reli- Some basic knowledge of instrumen-
the flowtube measure the change in the ability of the existing flowmeter was tation can be a very valuable trouble-
geometry of the flowtube, which is used questioned. Product literature for shooting and design tool. Gauging
to calculate the mass flowrate. The os- mass flowmeters promised high ac- whether an instrument installation
cillation frequency is used to measure curacy and low pressure drop. The will ever give accurate readings or
the fluid’s density. The temperature of plant-area engineer coordinated a whether it is an expensive spool piece
the fluid is measured to compensate for small project to replace the existing is useful in itself. Being able to assess
thermal influences and can be chosen orifice plate flowmeter with a mass the relative accuracy of two measure-
as an output of the meter. flowmeter. Column performance was ments will help determine from which
The original mass meters were U- very poor after startup. The new data to draw conclusions. Knowledge
tubes, but several different shapes meter had to be bypassed to operate of common instrument problems can
are now available, including straight the column normally. The overhead help in troubleshooting.
tubes as shown in Figure 14. Mass condenser was gravity drained, and Get to know the instrumentation
54 Chemical Engineering www.che.com March 2008

20_CHE_031508_FR_KT.indd 54 2/26/08 11:23:14 AM


M I C R O F I LT R AT I O N
U LT R A F I LT R AT I O N
0SJGJDF
GMPXNFUFS
N A N O F I LT R A T I O N
'5

'5

Have a bath with a friend


0SJGJDF
0SJGJDF GMPXNFUFS
GMPXNFUFS
'5

/PGMPX 'MPX
.BTT
GMPXNFUFS
Figure 15. With an under-
'5
standing of the accuracies of - 18. APRIL
LEIPZIG 14.
mass flowmeters and orifice
flowmeters, we revisit the STAN D G 9
H
question — Which flowmeter IF AT M U N IC
is the most accurate? 5. - 9. M AY
HA LL A2
on your towers. Gather the manufac- to Nick, the following DuPont col- ST AN D 30 3
.PTUMJLFMZUPCFNPTUBDDVSBUF FWFOBU
MPXGMPXSBUFT*OTFOTJUJWFUPQJQJOHDPO
turer’s information so you can assess leagues contributed their instrument GJHVSBUJPO6OMFTTGMPXJTFSSBUJDEVFUP
the instrument accuracy. Keep in war stories, and the author is grate- GMBTIJOHJOMJOFPSUIFUVCFJTDPSSPEFEPS
mind that the manufacturer’s litera- TFWFSFMZQMVHHFE UIFPSJGJDFQMBUFTBSF
ful for their willingness to share their
NVDINPSFMJLFMZUPIBWFIJHIFSFSSPS
ture refers to the ideal instrument experiences:
accuracy, which is the accuracy of the • Jim England, DuPont Electronic
measuring device itself. There are Technologies (Circleville, Ohio)
many other factors that contribute • Charles Orrock, DuPont Advanced
to the accuracy of the reading that is Fibers Systems (Richmond, Va.)
displayed on the DCS screen or in the • Adrienne Ashley, DuPont Advanced
data historian. The total accuracy in- Fibers Systems (Richmond, Va.)
cludes the instrument accuracy plus • Joe Flowers, DuPont Engineering
all of the other things that contribute Research & Technology (Wilming-
to error in the measured reading as ton, Del.)
compared to the actual value. Other
inaccuracies lie in digital to analog References
conversions, density errors, piping 1. Gillum, Donald R., Industrial Pressure,
Level and Density Measurement. Resources
configurations, calibration errors, vi- for Measurement and Control Series. ISA,
bration errors, and the list goes on 1995.
and on. Check the field installation 2. Kister, Henry Z., “Distillation Troubleshoot-
ing,” John Wiley & Sons, 2006.
to see what types of problems your 3. Spitzer, David W., Industrial Flow Measure-
meters will experience. ment. Resources for Measurement and Con-
trol Series. ISA, 1990
Get to know your mechanics and in-
4. Trevathan, V. L., editor. A Guide to the Auto-
strumentation experts at your plant. mation Body of Knowledge. ISA, 2006.
Now that you know some of the lingo
of instrumentation, you can better
5. emersonprocess.com/rosemount
6. omega.com
7>D"8:Aœ
converse with your instrument engi- 7. efunda.com hjWbZg\ZYBZbWgVcZBdYjaZ
neers and mechanics. 8. us.endress.com [dgW^dad\^XValVhiZlViZgigZVibZci
9. spiraxsarco.com

Acknowledgements Author
This paper is a compilation of in- Ruth Sands is a senior con-
strumentation basics obtained from sulting engineer for DuPont
Engineering Research &
the references listed below, of trou- Technology (Heat, Mass &
Momentum Transfer Group,
bleshooting experience from many 1007 Market St., B8218,
colleagues at DuPont, and of trou- Wilmington, DE 19898; Phone:
302-774-0016; Fax: 302-774-
bleshooting examples from Henry 2457; Email: ruth.r.sands@
Kister’s most recent book, Distilla- usa.dupont.com). She has
specialized for the last nine MICRODYN
tion Troubleshooting. Much of the years in mass transfer unit TECHNOLOGIES INC
technical information and many of operations: distillation, extraction, absorption,
adsorption, and ion exchange. Her activities P. O. Box 98269
the examples come from Nick Sands, include new designs and retrofits, pilot plant Raleigh, NC. 27624
testing, evaluation of flowsheet alternatives, and
Process Control Leader for DuPont troubleshooting. She has 17 years of experience
Phone 001 - 919 - 341-5936
Chemical Solutions Enterprise in with DuPont, which includes assignments in info@microdyn-nadir.com
process engineering, manufacturing, and corpo-
Deepwater, N.J. Nick has worked for rate recruiting. She holds a B.S.Ch.E. from West
DuPont for 17 years and is a special- Virginia University, is a registered professional WWW.MICRODYN-NADIR.COM
engineer in the state of Delaware, and is a mem-
ist in process control. In addition ber of the FRI Executive Committee.
Circle 31 on p. 76 or go to
adlinks.che.com/7370-31
56x254_biocel_messe_usa_rz.indd 1 07.02.2008 15:36:39 Uhr

20_CHE_031508_FR_KT.indd 55 2/26/08 11:23:54 AM


Control valve
position sensors
Department Editor: Scott Jenkins

T
he precise monitoring and control of The Hall effect is created Limit Switch Options
valve position is essential for efficient au- when a magnetic field is
tomation of both discrete and continuous applied perpendicular to
processes. Measurement of valve position the current direction.
provides the data required for the use of A sensor using the
advanced control strategies and predictive Hall effect is a trans-
maintenance algorithms. ducer that returns a volt-

COM
More effective monitoring of valve posi- age output according to
tion has been an area in which consider- changes in the magnetic

NO
NC
able progress has been made in improving field. For valve position
the performance and reliability of control sensing, an integrated
valves. Modern electrical valve-position Hall-effect sensor and
indicators offer either mechanical or non- magnet assembly detect
contact switching. The position indicators the presence, absence
are typically mounted either directly on a and orientation of a
magnetic trigger. The Mechanical Solid state Reed proximity
valve actuator or work indirectly using a switch sensor switch
non-contact remote feedback device. sensor is powered by
a constant current, and
Mechanical switches develops a varying
Most mechanical-switch valve positioners electrical potential
(Figure, top left) utilize some type of rotary that is proportional to
potentiometer for converting linear to rotary the flux density of a
feedback. These widely used devices are magnetic field applied
similar to variable resistors. perpendicular to the
Rotary potentiometers have an arched coil axis of the sensor.
of wire, over which an arm, called a wiper, Hall-effect proxim-
slides. The wiper is attached to the valve ity sensors used for
cam shaft, and as it moves across the coil of valve positioning offer
wire, a differing voltage output is produced. increased reliability in
The voltage output is proportional to the extreme environments.
angle at which the wiper is oriented. These sensors eliminate
Mechanical switches include contact link- all mechanical contact
ages that are subject to wear over time. The between the valve actua-
wear can eventually degrade performance. tor and the transmitter.
Because there are no
Non-contact proximity positioners moving parts within the
Non-contact technology approaches to Hall-effect sensor and magnet, the life Improved reliability — Safety integrity level
valve positioning can provide accurate expectancy is improved compared to a (SIL) ratings are higher with non-contact
valve-position data without the need for the traditional electromechanical switch. sensors and low-power solenoids. SILs are
linkages or levers required by traditional a measure of safety system performance.
systems. Avoiding mechanical contact in the Reed switches Higher SIL numbers mean better safety
valve positioning system addresses some Some non-contact valve positioners are performance and higher confidence in the
of the performance and cost challenges based on reed switches. A reed switch is field device.
associated with control valves, including an electrical switch that is operated by an Lower costs — Non-contact valve position-
mechanical wear, environmental hazards, applied magnetic field. Reed switches have ers have a lower overall total cost of own-
human error and inaccurate readings. a pair of electrical contacts on ferrous metal ership than conventional devices, thanks
Many non-contact proximity positioners reeds in a hermetically sealed glass enve- to the precise positioning capabilities that
(Figure, top right) incorporate a control- lope. An applied magnetic field moves the can be customized by valve application.
loop feedback mechanism based on an reeds, causing the contacts to either touch Also, the cost of ownership is lowered by
analog PID (proportional integral deriva- or move apart. The contacts can either be ease of calibration and service, and rich
tive) algorithm that has been updated for a open normally, closing when a magnetic diagnostics for predictive maintenance
digital device. The algorithm incorporates field is present, or closed normally, opening signatures.
the Ziegler-Nichols (Z-N) tuning procedure, in the presence of a magnetic field. Bifur- Increased versatility — Non-contact valve
a well-known method for tuning automatic cated reed switches can be used in applica- positioners are designed to be compatible
controllers. It is a two-step tuning approach tions where ultralow power or capacitive with most standard industrial communica-
that adjusts how agressively the valve con- discharge consideration are in effect. tions protocols, including HART, Foundation
troller reacts to errors between the process Fieldbus, AS-I, Modbus, DeviceNet and
variable and the desired setpoint. Benefits Profibus. These devices can help engineers
Significant benefits for non-contact valve take advantage of the cost savings and in-
Hall-effect sensors positioners, include the following: creased diagnostic capabilities of networks,
A number of non-contact proximity position- Greater flexibility — Non-contact position- along with the advantages offered by
ers are based on the solid-state Hall effect, ers utilizing Hall-effect sensors provide improved position sensors.
and are used to help improve monitoring feedback on valve position without link-
and control of production processes. The ages, levers or rotary or linear seals. This Notes
Hall effect refers to a potential differ- allows a remote sensor-head assembly to This edition of “Facts at Your Fingertips” was
ence, known as the Hall voltage, between be mounted a considerable distance from adapted from Jack DiFranco’s article, entitled Ad-
the electronics enclosure, giving engineers vances in Valve Position Monitoring, that appeared
opposite sides of an electrical conductor in the December 2007 issue of Chemical Engineer-
through which an electric current is flowing. increased flexibility and improved safety. ing, pp. 46–50.
Control Valve
Performance
Department Editor: Scott Jenkins

M
inimizing process variability is an response time and over-
important component of a plant’s prof- shoot for a control valve
itability. The performance of control response to a step input
valves within process control loops has a change. In this case,
significant impact on maintaining consistent stem position in percent
processes. This refresher outlines some of the of travel is used as the Output
important aspects of control valve perfor- control valve “output.”
mance, including parameters of both the Step-change size. The
static response and the dynamic response. dynamic response of b
a control valve varies a c
d
Static response depending upon the a < resolution ≤ b
A valve’s static response refers to mea- size of the input step
surements that are made with data points change. Four “ranges” c ≤ dead band < d
Input
recorded when the device is at rest. Key of step sizes to help

Amplitude
static-response parameters for control valves understand the static-
include travel gain, dead band and resolu- and dynamic- response
tion (Figure 1). metrics are defined by
Travel gain (Gx). This term represents the ANSI/ISA standards: Time
change in position of the valve closure • Small input steps (Re- Dynamics are not shown
member divided by the change in input gion 1) that result in no
FIGURE 1. Dead band and resolution, illustrated here, are
signal. Both quantities are expressed as measurable movement
of the closure member key static-response parameters for control valves
a percentage of the full valve span. The
within the specified
closure member is part of the valve trim (the wait time 39
combination of flow-control elements inside Initial overshoot to 38.11 = 23% Final steady-state
• Input step changes average values
a valve). Travel gain measures how well the that are large enough input = 37.84, stem = 37.65
valve system positions its closure member to result in some 38 Stem
compared to the input signal it receives. control-valve response
Input, stem %

with each input signal Input


Without signal characterization in the valve
system, the travel gain should be 1.0. [1] change, but the re-
sponse does not satisfy 37 Travel gain = 0.91
Dead band. This term can be defined as Time to steady state, Tss = 18.3 s
the requirements of
the range through which an input signal the specified time and
may be varied, with reversal of direction, 86.5% of response, T86 = 2.06 s
linearity (Region 2) 36
without initiating a response (an observable • Step changes that are
Dead time Td =1.6 s
change in output signal). With respect to large enough to result
control valve performance, if the process in flow coefficient Initial steady state average values, input and stem = 35.67
changes, which sat- 35
controller attempts to reverse the position of 0 10 20 30
the control valve, the valve will not begin isfy both the specified
maximum response Time, s
to move until after the controller output has
time and the specified FIGURE 2. This graph shows the response of a control valve
reversed an amount greater than the dead maximum linearity to a step input (reprinted with permission from EnTech Con-
band. A large dead band will negatively (Region 3) trol Valve Dynamic Specification V3.0)
impact control performance. • Input steps larger than
Resolution. This term can be defined as the in Region 3 where the specified magnitude- istic refers to the curve relating percentage of
minimum amount of change in valve shaft response linearity is satisfied but the speci- flow to percentage of valve travel. Inherent
position when an input is applied. Resolu- fied response time is exceeded (Region 4) flow characteristic applies when constant
tion will cause the control valve to move Region 1 is directly related to dead band pressure drop is maintained across the valve.
and resolution. Region 2 is a highly nonlinear
in discrete steps in response to small, step Typically linear, quick opening or equal per-
region that causes performance problems and
input changes in the same direction. This should be minimized. Region 3 is the range of centage, this will impact both the magnitude
occurs as the valve travel sticks (when the input movements that are important to control and the consistency of the process gain over
starting friction on the valve shaft is greater performance [1]. the operating range [1]. Good control-valve
than the friction when the shaft is in motion). performance depends on proper valve sizing
Similar to dead band, a larger resolution Process gain and trim characteristics.
will negatively impact control performance. Process gain is the ratio of the change in
a given process variable to the change in References
Dynamic response controller output that caused the change. 1. Beall, James, Improving Control Valve Perfor-
Dynamic response for a control valve is the To achieve effective process control, the pro- mance, Chem. Eng., Oct. 2010, pp. 41–45.
time-dependent response resulting from a cess gain should ideally fall within a certain 2. Emerson Entech, Control Valve Dynamic Specifi-
time-varying input signal. range, and should be consistent throughout cation, Version 3.0, November 1998.
Dead time. This term refers to the time after the operating range of the valve. When the 3. Hoop, Emily, Control Valves: An Evolution in
the initiation of an input change and before process gain is too high, valve non-lineari- Design, Chem. Eng., August 2012, pp. 48–51.
the start of the resulting observable response. ties are amplified by the process gain and 4. Ruel, M., A simple method to determine control
Step response time. This term represents valve performance and its impacts on control
process control performance deteriorates.
loop performance, Top Control Inc., Swanton,
the interval of time between initiation of an When the process gain is too low, the Vt., white paper, 2001.
input-signal step change and the moment range of control is reduced. Changes in the 5. International Society of Automation (ISA) and
that the dynamic response reaches 86.5% process gain over the range of operation American National Standards Institute (ANSI).
of its full, steady-state value [1]. The step result in poorly performing regions in the ANSI-ISA-TR75-25-02-2000, Control Valve Re-
response time includes the dead time before closed-loop controller response. sponse Measurement from Step Inputs, 2000.
the dynamic response. Two control-valve features impact process 6. Neles-Jamesbury Inc., “The Valve Book,” Neles-
Overshoot. This term is the amount by which gain: the size of the valve trim and the inher- Jamesbury, Worchester, Mass., 1990.
a step response exceeds its final, steady-state ent flow characteristic of the valve. If the 7. Skousen, Philip L., Valve Handbook, McGraw
value. Overshoot is usually expressed as a valve trim is oversized, the process gain will Hill, New York, 1998.
percentage of the full change in steady-state be higher than it would be for an appropri- Editor’s note: Portions of this page were adapted
value. Figure 2 shows the dead time, step ately sized valve. The valve’s flow character- from the article in Ref. 1.
Environmental Manager

Common Mistakes When Conducting a


HAZOP and How to Avoid Them
An important part of ensuring the success of a HAZOP study is to understand the
errors that can cause the team to lose focus
Arturo Trujillo, Walter S. Kessler
and Robert Gaither
Chilworth, a DEKRA Company

S
ince its inception in the 1960s
and its first official publica-
tion in 1977, the Hazard and
Operability Study (HAZOP)
has become one of the most power-
ful tools for identifying process haz-
ards in the chemical process indus-
tries (CPI). Utilizing systems that are
qualitative or even simplified semi-
quantitative, the HAZOP method has
been increasingly used, not only as a
tool for identifying process hazards,
equipment deficiencies or failures and
operability problems and assessing
their risks, but also as a tool for priori-
tizing actions and recommendations
for process-risk reduction. Reducing
risk is especially important in ensur-
ing the safety of the personnel who
must work in the plant environment
each day (Figure 1).
The HAZOP methodology is a sys-
tematic team-based technique that
can be used to effectively identify
and analyze the risks of potentially
hazardous process operations. It is
the most widely used process hazard
analysis (PHA) technique in numer-
ous industries worldwide, including
petrochemicals, pharmaceuticals,
oil-and-gas and nuclear, and is used
during the design stages of new pro-
cesses or projects, for major process
modifications and for periodic review
of existing operations. FIGURE 1. HAZOP studies are useful tools in reducing process risk, and they provide safeguards against
A HAZOP is a time-consuming ex- hazardous scenarios for the personnel who must maintain and operate the plant
ercise and should be conducted in
such a way to ensure that the results to correct these mistakes if or when Benefits of a HAZOP
justify the effort. This article presents they occur during the course of the The advantages offered by HAZOP
some common mistakes that can HAZOP study. Therefore, the selec- over other process-risk analysis
jeopardize a HAZOP team’s task. Fre- tion of an experienced facilitator is tools are numerous, and include
quent or chronic occurrence of these an essential element for assuring the the following:
mistakes indicates potential gaps in success of the HAZOP. Without an • It is a rigorous process; it is structured,
the site’s process-management sys- adequate depth of knowledge and systematic and comprehensive
tem. However, it is ultimately the re- experience, the HAZOP can become • It is adaptable to the majority of CPI
sponsibility of the HAZOP facilitator a “check the box” exercise. and manufacturing operations, in-
54 Chemical Engineering www.chemengonline.com december 2015
cluding those in petroleum refiner-
ies (Figure 2) and other oil-and-gas
processing plants, nuclear facili-
ties, and specialty chemical, phar-
maceutical and even high-speed
manufacturing plants
• It is team-based and allows the
interchange of knowledge and ex-
perience between the participants
• It helps to anticipate potential ac-
cidents or harm to employees, the
facility, the environment and the
surrounding community
• It functions as a type of training for
the team’s participants and leader,
who are required by the nature of
the method to look at the process
from a new perspective — not just
from the perspective of “how should
it run?,” but also “how can it fail to FIGURE 2. Many processes in the CPI are potentially hazardous if not managed correctly. HAZOP studies
run correctly?” seek to prioritize actions to reduce process risks, and are adaptable across a wide range of
A HAZOP is time-consuming be- industrial sectors
cause it requires the participation estimate of the time required for the Mistake 3: Incorrect size of HAZOP
of a multi-disciplinary team over ex- team based on the process descrip- team. The HAZOP team should be
tended timeframes. This investment tion and preliminary count of HAZOP limited in size, ideally five to seven
of time and personnel, often involv- nodes (specific portions or topics of people, excluding the HAZOP facili-
ing third parties, means that the per- the study process) so that managers tator and the HAZOP scribe or sec-
formance of the HAZOP needs to be are aware of the degree of personnel retary. A team that is too large can
optimized to maximize its value. The commitment that will be required. easily lose focus, dwell on a subject
following sections detail some com- Mistake 2: Incomplete, inaccurate or issue too long, or be disruptive. It
monly found mistakes that occur or unavailable process safety in- is human nature that all participants
during the planning, execution and formation. Another common mis- seek to present their perspectives,
followup stages of a HAZOP. take during a HAZOP is not having but this can lead to excessive dis-
all the prerequisite process safety cussion. A group that is too small will
Planning stage information (PSI) and other valuable not likely include the right expertise
Mistake 1: Mismanagement of information available, including out- or provide enough different perspec-
time-allotment issues. One of the of-date or incomplete information. tives to evaluate the process haz-
most frequent mistakes of a HAZOP This is especially critical regarding ards and controls adequately or in
is failure to manage the time allot- piping and instrumentation diagrams the right detail.
ted for the study. A HAZOP is often (P&IDs), current standard operating
scheduled for a set amount of time, procedures (SOPs) and appropriate Execution stage
neither by the HAZOP facilitator nor data on flammability, combustibility, Mistake 4: Lack of focus during
the team, and sufficient time may not reactivity, toxicity and electrostatic the meeting. A HAZOP is a com-
have been allocated. Furthermore, properties of materials in all forms plex exercise that requires the con-
there may be little or no flexibility in and phases, as well as compatibil- centrated and coordinated contribu-
the schedule. An insufficient amount ity of chemicals with each other and tion of all the members of the team.
of time for the HAZOP limits discus- with the processing equipment. If the Distractions should be minimized
sion and brainstorming and reduces HAZOP is conducted by an external in order to ensure and maintain the
the quality of the analysis, in turn facilitator, it is the responsibility of the team’s focus. Therefore, team mem-
leading to some of the mistakes dis- owner of the process to verify the in- bers should not be allowed to come
cussed in more detail below. tegrity of the PSI. and go into and out of the meeting,
Estimating the duration of a Related to this, it is not acceptable take phone calls, answer emails,
HAZOP is not an exact science, and that participants attend the HAZOP or discuss issues not related to the
it requires a good knowledge of the for the purpose of obtaining informa- HAZOP during the sessions. Use of
methodology, the complexity of the tion on a process or project. HAZOP an offsite venue may be helpful to
process, the nature of the risks that participants should be well prepared prevent plant operations from be-
can be identified up front and the id- to contribute to the discussion and coming a distraction.
iosyncrasies of the group. Although a have all requisite background infor- It is the responsibility of the HAZOP
HAZOP should not be open-ended in mation with them. It is the respon- facilitator to maintain the focus of the
time allotment, the ideal HAZOP has sibility of the facilitator to instruct all group and keep the HAZOP process
some flexibility built into the sched- participants that they must come to moving by allowing some open dis-
ule. The team leader should make an the HAZOP prepared. cussion on the issue, node and con-

Chemical Engineering www.chemengonline.com december 2015 55


the responsibility of the HAZOP fa-
cilitator to manage the team and
the HAZOP study process to ensure
that the team stays focused and that
no nodes or hazards are missed by
the team.
Mistake 6: Mistaking the tools for
the process. The HAZOP spread-
sheet should not be viewed as a
questionnaire whose boxes all have
to be filled in, even with numerous
repetitions of scenarios. The combi-
nation of pairs of key words and pa-
rameters is not intended to be an end
in itself, but to encourage discussion
and identify deviations from the de-
sired state. As would be expected,
the same deviation generally causes
FIGURE 3. It is crucial that a HAZOP be explicitly targeted for the specific process in question, and not
based on previous HAZOPs for similar processes, as process safety information and controls may have the alteration of more than one pro-
recently changed cess parameter, and therefore could
be entered in more than one place in
sequence at hand, but not letting it be identified that constitutes the the spreadsheet. An obvious exam-
get out of control. Sufficient (but not purpose of the HAZOP, and this is ple is a distillation column, in which
excessive) breaks for participants to where it demonstrates its power pressure, temperature, composition
eat and drink and conduct activities • Carrying out a superficial review of and flowrate (of reflux, for example)
not related to the HAZOP, such as the combinations of key words and are clearly interrelated. Hence, any
checking their emails and voice- parameters, listing the most obvi- change in one of the parameters
mails, should be planned and coor- ous, and often repetitive, causes of automatically causes responses and
dinated. The HAZOP room should deviation without going into detail. changes in the others.
be free from cellphones, and distrac- In other words, repeating the same It is not as important for all the
tions like texting during the HAZOP causes, parameter after parameter spreadsheet “boxes” to be filled in
exercise should be forbidden. and node after node, instead of as it is for the HAZOP group to work
Mistake 5: Preventing the team conducting a more in-depth analy- effectively in identifying all the pos-
from brainstorming. Another fre- sis and discussion sible deviations. A HAZOP table is
quent mistake in HAZOPs is to re- • Carrying out HAZOPs using some not and should not be a form-filling
strict the brainstorming exercise, form of prior information — pre- exercise. Rather, it should guide and
which is, after all, the basis (and the built templates or the HAZOP structure strategic brainstorming dis-
power) of the method. The most from a similar project, for example. cussion with the intent of identifying
common issues in this area include Again, what the HAZOP is meant all hazards and operability problems
the following: to do is analyze the possible spe- that may injure employees (Figure 4),
• Omitting key words, parameters cific risk scenarios (especially the cause damage to property and as-
or even nodes, with the argument non-obvious ones) of the process sets, impact the community or cause
that an upper bound for the con- or project being studied at the time environmental damage.
sequences in this node can be of the HAZOP (Figure 3). While one Mistake 7: Misrepresenting or
easily identified, and these maxi- can refer to, or reference previous misunderstanding safeguards.
mum consequences are protected material, the HAZOP is to be con- Documentation of effective and ap-
by safeguards. This clearly means ducted based upon the current propriate safeguards is a key step
that steps or phases of the HAZOP facility or process, and the equip- in the PHA team’s decision whether
procedure will be skipped, and ment, process or controls may have additional process-risk reduction
some process hazards may not be changed since the last HAZOP is required for a specific scenario.
identified. This violates the HAZOP In practice, the quality of a HAZOP Examples of safeguards that are
methodology and overall purpose is influenced by the ability of the neither effective nor appropriate are
of conducting the HAZOP in the HAZOP leader to ask the appro- given below:
first place. Although on many oc- priate questions to ensure that the • Local instruments that are never
casions, strict application of the team identifies all the hazards of the checked by field operators
methodology will not identify any process being studied, not only the • Alarms that fail to give the operator
hazardous scenarios other than the most obvious hazards. This ability sufficient time to effectively halt the
obvious ones, which have already is based on the leader’s experience consequences of the deviation.
been listed up front and used as with the HAZOP technique and his or Examples include the following:
an argument for omitting any fur- her technical skills in process-hazard ❍❍ Alarms that fail
ther analysis. Nevertheless, some- identification, as well as human error ❍❍ Very generic alarms that are
times a non-obvious scenario will and equipment failure potential. It is activated in numerous differ-
56 Chemical Engineering www.chemengonline.com december 2015
ent situations. In this case, evaluated and ruled out for various
the operator has to diagnose reasons. While a HAZOP can and
which of the multiple op- should include recommendations re-
tions he or she is faced with, lated to operational and maintenance
thereby losing valuable time issues, the HAZOP’s sole intent is for
for action the identification of issues, not to find
❍❍ Alarms that are activated fre- a solution to the problems or rede-
quently, often for trivial reasons, sign the facility. All recommendations
and that therefore tend to be are made for further investigation and
ignored by the operators design considerations. Therefore, the
❍❍ Alarms where no specific actual HAZOP is not the best time FIGURE 4. HAZOP studies intend to provide a com-
operator response has or place to deal with these types of prehensive index of the hazards and operability
been given in procedures problems that may cause damage or put employ-
issues. They should be further inves- ees in danger
and training tigated offline in the correct setting,
❍❍ Cascades of alarms, where and should include the appropriate able operating range. Utilize this
“first-in” is not obvious personnel in the discussions. chart to set appropriate process
or indicated Mistake 10: Excessively lax rec- alarm and shutdown points
• Pressure-relief systems (such as ommendations. When making rec- It should be noted how these two
safety valves and rupture discs) that ommendations in a HAZOP, it is very recommendations are very specific
were not designed for the case and important to utilize the proper word- action items and also include the
process conditions being studied. ing. Since the HAZOP team is com- reason for the action.
Obviously, the purpose of a HAZOP posed of knowledgeable people, On some occasions, there may
is not to verify the correct design of recommendations should be made be two or more divergent opinions,
pressure-relief systems. Neverthe- that involve action. Two words that and a consensus cannot be reached
less, if there is reasonable doubt, are highly over-utilized are “recom- during the HAZOP itself. In this case,
a recommendation should be is- mend” and “consider.” “Recom- both recommendations should be
sued to check that the scenario for mend” is already used in the title for included in the HAZOP and left for
which it was listed as a safeguard the column and most of the time, the further investigation or evaluation
was one of the cases of design for team’s brainstorming makes up the by the company, based upon the
the relief device or system “consideration” aspect of the recom- information from the HAZOP. For
Operating procedures cannot be mendation being proposed. If addi- scenarios such as these, the best
considered safeguards when the tional risk analysis is required, “con- solution — after further investigation
cause giving rise to the scenario is sider” is an appropriate phrase. and research is completed — may
human error, which presupposes There are often multiple ways to re- be something not even mentioned or
that the procedure has not been fol- duce risk and the team’s time should thought about in the HAZOP itself.
lowed properly not be spent analyzing alternatives. Again, it should be reiterated that
Mistake 8: Excessive recommen- Another common phrase seen in except for a few unique situations,
dations. Some HAZOP groups be- many HAZOP recommendations is such as the divergent opinion case,
lieve that they should issue a recom- “Further study on what needs to be recommendations should be clear,
mendation for any scenario that has done in order to...” — which in reality specific, not open to interpretations
negative consequences, whether a is not specific and can be left open and include the reasoning at the time
hazard scenario, equipment failure or for interpretation. Most of the time, that the HAZOP was conducted.
operability problem. This is not in the recommendations that involve an Mistake 11: Trying to solve the rec-
spirit of the HAZOP method. What a action and have a specific purpose ommendation or design the solu-
HAZOP aims to do is identify all of should be made. Start recommen- tion during the HAZOP. Another
the hazardous scenarios, determine dations with strong action words, common mistake that can delay the
the associated risk for each particu- such as “install,” “investigate,” HAZOP and cause the group to lose
lar scenario and check whether the “graph,” or “add.” Additionally, when its focus is trying to solve the prob-
process has been duly protected by wording recommendations, if a rec- lem or redesign the process listed
the safeguards, and only if there is ommendation is being made for a in the recommendation during the
not adequate protection, propose specific reason, include that reason HAZOP study itself. This is most
recommendations for doing so. in the recommendation so it is not common when process-design en-
Mistake 9: Irrelevant recommen- forgotten when the HAZOP report gineers are team members and they
dations. Sometimes, people will is written or is being reviewed. The desire to make the process perfect.
suggest and utilize HAZOP recom- following are good examples of well- Unless it is a clear and easy solution,
mendations as a way to obtain ap- worded recommendations: many recommendations require fur-
proval for an operational or plant • Install a pressure gage and trans- ther investigation or other actions to
design improvement that is not nec- mitter on the overhead line “L12” of complete the task, alleviate or mini-
essarily directly related to the safety the distillation column to increase mize the hazard, and close out the
of personnel or the release of a haz- the SIL level from 1 to 2 action item based upon the recom-
ardous chemical. In many cases, • Graph the P/T curve for the reac- mendation. It must be remembered
these changes have already been tion process and add the accept- that a HAZOP is a brainstorming

Chemical Engineering www.chemengonline.com december 2015 57


exercise with knowledgeable pro- Mistake 13: Failure to update sents an extremely powerful tool for
cess personnel from different areas HAZOPs when process knowledge the identification, semi-quantification
of the plant, whose task is to identify changes. A HAZOP worksheet is a liv- and mitigation of risks in CPI produc-
hazards or hazardous scenarios and ing document. Ideally, it reflects man- tion facilities with continuous, batch
make practical recommendations agement’s current knowledge of the or semi-batch processes.
to alleviate or minimize the hazard- process hazards, the consequences The biggest inconvenience of this
ous scenarios or consequences. of those hazards and the controls technique is its relatively high cost, in
As previously stated, not all recom- necessary to reduce the process risk terms of time and people who need
mendations have clear-cut solutions, to a tolerable level. HAZOPs lose their to be involved and participate in the
and the HAZOP time should not be effectiveness over time when they are brainstorming sessions. This high
wasted with actions that may require not updated promptly. cost means that the HAZOP needs
research and further investigation Changes in process safety infor- to be carried out to optimum effect,
that only one of the participants, or mation should result in a PHA review avoiding the sorts of mistakes that
a qualified expert, can resolve in the through the site MOC procedure. The have been discussed in this article. It
quiet of his or her own office. Even review will identify any new causes of is the responsibility of the HAZOP fa-
HAZOP-recommended changes to a a process deviation or operability is- cilitator to make sure the group stays
process should be subjected to the sues, changes in safeguards for previ- focused and does not commit any
site’s management-of-change (MOC) ously documented hazard scenarios, of these mistakes. Finally, the selec-
process to prevent the introduction and possibly new or revised recom- tion of a knowledgeable and experi-
of new hazards. It is not uncommon mendations to address the hazards. enced PHA facilitator is a crucial ele-
for an incident to be triggered by a Recent accidents or near misses ment for assuring the success of the
change made for safety reasons. on a site process, or a similar process HAZOP process. n
The HAZOP can and should result in elsewhere, should trigger a HAZOP Edited by Mary Page Bailey
a list of actions or recommendations, review to ensure that the same or
with the designation of someone re- similar scenario has already been Authors
sponsible for carrying them out, but considered and documented dur- Arturo Trujillo is managing direc-
not necessarily the final solution or ing the most recent HAZOP and that tor of Chilworth Amalthea, the
Spanish subsidiary of the process
re-engineering of the plant. effective controls are in place to pre- safety division of DEKRA (Nàpols
vent a similar incident from occurring 249, 4ª planta 08013 Barcelona,
Followup stage in the future. Spain; Phone: +34-931-426-029;
The output of the HAZOP study is Email: arturo.trujillo@dekra.com).
He has facilitated more than 200
the set of recommendations that are Additional applications HAZOPs, and his specialities in-
usually presented to management in For the sake of simplicity, this article clude SIL and LOPA. Prior to work-
a standardized report format. At this has focused on common mistakes ing at Chilworth, he served as a division manager at
Technip Iberia and as engineering director at Asesoría
stage, site management is respon- observed during the use of the HAZOP Energética. He attended Universitat Politècnica de Catalu-
sible for responding to each recom- methodology. The discussion in this nya and received a Ph.D. from Johns Hopkins University.
mendation according to local or site article can be equally applied to other Walter S. Kessler is a senior pro-
requirements and the requirements scenario-based methodologies, such cess safety consultant at Chilworth
of applicable standards, such as the as “what-if” analyses, which can be Technology Inc. (113 Campus
Drive, Princeton, NJ 08540;
U.S. Occupational Safety and Health carried out at very early stages of the Phone: 832-492-4358; Email:
Administration (OSHA) Process Safety process lifecycle — HAZOP is typi- walter.kessler@dekra.com).
Management (PSM) standard Title cally reserved for late-design stage Kessler has 20 years of experi-
ence in the petroleum refinery,
29, CFR Part 1910.119. Site proce- or later-lifecycle stages when more gas-processing, specialty-chemi-
dures should include regular followup detailed PSI is available. The specific cal, pharmaceutical, manufactur-
reports to track recommendations PSI that is available and the expertise ing and HVACR (heating, venting, air conditioning and
refrigeration) industries, including five years performing
to their resolution. needed for other hazard evaluation process-safety engineering functions. He was instru-
Mistake 12: Failure of manage- methodologies may be different, but mental in the design and construction of several refin-
ment to act promptly on each rec- the types of mistakes discussed here, ery, gas and chemical processing facilities, designing a
ommendation. Site management and their prevention, are very similar. pharmaceutical filling process and also has experience
in Six Sigma and lean manufacturing.
must evaluate each recommendation
Robert L. Gaither is a senior pro-
according to its technical feasibility, Closing thoughts cess safety specialist at Chilworth
the risk-reduction benefit versus total OSHA recognizes the HAZOP tech- Technology Inc. (113 Campus
cost of implementation, availability of nique as an acceptable methodology Drive, Princeton, NJ 08540;
Phone: 732-589-6940; Email:
alternative solutions and other fac- for conducting PHAs of processes robert.gaither@dekra.com).
tors. The PSM standard allows rejec- covered by the PSM standard. Other Gaither has more than 28 years of
tion of a PHA recommendation only regulators around the world also ac- experience in company operations,
for specific causes. Good industry cept the HAZOP methodology as regulatory compliance, manage-
ment consulting and process
practices dictate that management appropriate for analyzing the exist- safety and risk management. He has led organizations at
takes prompt action on each recom- ing and potential hazards of a com- site, division and corporate levels to achieve record
mendation and ensures that all rec- plex process that involves a highly safety performance and significant cost savings. Gaither
is trained in HAZOP and SIL/LOPA facilitation. He holds a
ommendations are tracked to final hazardous substance. Ph.D. and is a certified safety professional (CSP).
resolution and closure. The HAZOP methodology repre-
58 Chemical Engineering www.chemengonline.com december 2015
Engineering Practice

Chemical Process Plants: Plan for Revamps


Follow this guidance to make the most of engineering upgrades that are designed to
improve plant operations or boost throughput capacity
Koya Venkata Reddy 600
590 mm (max)
80
FACT Engineering and Design Organization Efficiency %
570 mm (rated) 70

T
500
he chemical process industries (CPI) are func- 530 mm (min)
60
tioning in an era of globalization, and between the
400
prevailing economic conditions and upheavals 50
in the energy sector, the number of new invest-

Efficiency, %
Head, m
ments in CPI facilities has fallen in recent years. Many 300 40
industries are seeking cost reductions by revamping
existing plants with minimum investment. The objective 30
is to reduce the cost of production through the use of 200
upgrades and new technologies, to remain competitive 20
in the market. By way of example, if one wants to set 100
up a new complex to produce ammonia and urea, the NPSHR 10
specific capital cost will be on the order of $666/ton of
urea. By comparison, if an existing plant is revamped to 0 0
0 250 500 750 1,000 1,250 1,500 1,750
raise the existing production from 100% to 120% (that Capacity, m3
is, adding 20% additional capacity), this can be done at
an expenditure that is closer to $300/ton to achieve this FIGURE 1. Shown here are typical pump characteristic curves, with three
different impeller sizes, showing capacity versus head, and NPSHR versus
incremental production capacity
This article reviews key concepts, objectives and pro-
cedures that are needed to successfully carry out vari- a turnkey basis. Meticulous planning related to the hook-
ous types of CPI plant revamps. up of tie-in points arising out of expansion schemes can
help to reduce the amount of downtime required to ex-
The need for revamps ecute the revamp schemes and put the plant back on-
Chemical process plant revamps are typically under- line.
taken for the following reasons:
• To change in feedstock composition Targeted revamp capacity, change in process
• To adopt energy-conserving processes in light of in- In general, it is possible to increase the rated capacity
creasing energy costs of a plant by 10%, with very little added expenditure.
• To reduce the fixed-cost components of production, But to increase capacity by 20–50% over the nameplate
by increasing capacity within the existing facility capacity, substantial modifications must be taken into
• To extend the life of a well-maintained process plant consideration that often involve implementing different
Similarly, there are many benefits to conducting an ap- technologies from the ones already applied in the exist-
propriate plant revamp. These include the ability to: ing plant. When seeking such notable increases in pro-
• Increase the reliability of equipment, leading to reduced duction capacity, plant operators and managers must
downtime and maintenance costs not only verify the soundness of the economics, but also
• Reduce energy consumption carefully evaluate the potential drawbacks, if any.
• Extend useful plant life Sometimes the existing process path may have to be
• Reduce the cost of production, thereby improving the changed to enhance the capacity of the plant, since the
overall bottom line for the facility current process may not yield the desired efficiency or
However, experience shows that inefficient implementa- conversion rates. Two cases are discussed below.
tion of proposed revamp options can lead to failure, so Example 1. In the case of units to recover liquefied pe-
care must be taken to avoid this by building the right troleum gas (LPG) from natural gas, such units are de-
team of experts. This team typically includes represen- signed for a certain composition of feed gas. The need
tatives of the process licensor company, engineering for a revamp often arises if the gas composition has
and project-management consultants, and experts from changed and the expected recovery of C3/C4 and higher
the owner company representing diverse fields, such as compounds has become unprofitable. In this case, the
operations, project management and maintenance. If expected recovery of LPG and natural gas liquids (NGLs)
sufficient expertise for the proposed revamp is not avail- can be achieved by compressing the feedstock to higher
able internally, one can hire consultants to carry out the pressures than present levels, or by spiking heavier NGLs
feasibility studies and implementation of the revamp on back to the feed gas stream. Thus, such a revamp re-
48 Chemical Engineering www.chemengonline.com december 2015
quires a study to assess the technical and economic fea- raw materials, utilities and energy per unit of production
sibility of the different process paths being considered. are tabulated. The material-and-energy balance of the
Example 2. A feedstock change from naphtha to natural existing operation, and the required revamp plant load,
gas in ammonia plants, hydrogen plants and methanol are prepared.
plants also necessitates a need for revamp of the re- The existing equipment components are rated for
former section and front end, but in many cases, the the revamp conditions, and then changes and required
existing process path can be retained. In this case, the new equipment are identified. Cost estimates of various
absorbed duty of the reformer — which tends to be the schemes are prepared (after consultation with various
major energy-consuming equipment found in the system vendors). Feasibility studies, followed by detailed project
— and the burner duties required vis-a-vis the required reports (DPR), are also prepared. The potential rates of
reformer absorbed duty are calculated to check their suit- return of various options are studied. The best option
ability. The maximum skin temperature of the reformer available (on the grounds of economic sustainability and
tubes for the feedstock change must be checked. technical feasibility) is then selected, so that the basic
In all cases, the existing process path, along with other engineering design package (BEDP) can be prepared,
options, must be studied in detail to arrive at the most and the revamp project implemented.
economical and technically feasible revamp option. As noted, successful revamps require assembling
the right revamp team. Typically, such a team consists
Lifecycle of the plant of individuals from the process licensor company, con-
The different phases of a plant’s lifecycle must be taken sultants for basic engineering and detailed engineering
into consideration when planning a revamp. Such phases services, contractors for specific electrical-, mechanical-
include the following: and instrumentation-related aspects of the project, and
1. Incubation stage — Initial stabilization period various engineers from the owner’s group (for instance,
2. Growth stage — Optimization and debottlenecking of those who represent specific disciplines and have a con-
operations to improve the efficiency crete understanding of the current operation).
3. Maturity stage — Attainment of stable operation The following planning steps should be undertaken:
4. Declining stage — Realization that plant capacity is 1. Estimate the plant’s inherent capacity from past and re-
not sustainable because of frequent equipment fail- cent data. This can be done by identifying weak areas
ures or excessive maintenance requirements in the plant (for instance, those that are contributing
Revamping the plant during Phases 1, 2 or 3 is relatively to non-realization of rated or required plant capacity),
easy, whereas revamping a plant during Phase 4, when or by conducting an end-to-end survey of the plant.
the facility is already in decline, requires the engineering Once such a study is carried out, efforts should be
team to adapt many of the modern technology options made to predict the potential performance improve-
to an aging infrastructure, and to replace many equip- ments of the plant if the weak areas are rectified.
ment components. 2. Prepare the process scheme and the equipment data
sheets. Carry out feasibility studies of all options (in-
Objectives of a revamp cluding both technical and financial aspects of the pro-
The objectives of a plant revamp should be spelled out posed revamps) and then develop the detailed project
prior to studying the options. Possible objectives could report. Set the target of the revamp in terms of time
be the following: (schedule) and cost.
• Enhance capacity from the present operating level to 3. Implement the approved revamp. Ideally, the revamp
expand capacity to, say, 110%, 120%, 130% of rated activities should be carried out during the annual
capacity scheduled turnaround period for the plant, to minimize
• Reduce production costs unscheduled downtime.
• Reduce pollution
• Reduce the consumption ratios of various raw materi- Estimate plant capacity
als and utilities Many older CPI plants can run at or above the rated ca-
• Reduce maintenance costs and increase the onstream pacity continuously for a week or a month. But due to cer-
factor tain operating limitations, and downtime that may arise
• Upgrade the technology to keep pace with the new from some underperforming equipment, the annual rated
developments, and to increase the plant life capacity is seldom achieved. Analyzing past operating
• Minimize plant shutdown data on a monthly basis (for the past 10 years or so) will
These objectives can be achieved by maximizing efficiency, reveal which equipment components are most often to
yield and conversion of raw materials in various sections. blame for downtime, and are thus affecting overall capac-
Specifically, plant revamps are often implemented to im- ity utilization. Such a study of past data is often called a
prove process optimization, increase energy conserva- weak-area analysis. Similarly, sometimes an end-to-end
tion, improve product quality and expand capacity. survey of the plant (from the plant commissioning to the
present day) is also conducted.
Key revamp procedures Existing equipment poses both opportunities (in the
Every revamp project should start by identifying the goals form of underutilized capabilities) and challenges (in terms
and actual bottlenecks. A material-and-energy balance of limitations). The ability to identify problem areas can
for the base case should be developed to reflect the ac- help the team to prioritize their debottlenecking efforts in
tual operating conditions. The consumption of various order to improve capacity utilization more quickly.

Chemical Engineering www.chemengonline.com december 2015 49


The weak-area analysis TABLE 1. A typical calculation of Cv, before and after a
revamp
Understanding current operation is very important for
Unit Before After
the successful revamp of a plant. The plant performance
can be evaluated based on the performance data for the Flowrate m3/h 80 100
past 10 years, if the plants are relatively old. Otherwise Density kg/m3 950 950
the plant performance is studied from the beginning to P kPa 49.03 49.03
the present day (using the end-to-end survey). P kg/cm2 0.5 0.5
Two indices, the plant load factor (PLF), and the on-
N1 unitless 0.0865 0.0865
stream factor (OSF), are important to scientifically evalu-
ate the plant performance. Cv unitless 128.73 160.92
Control valve size in. 4 6
Actual production 100 Pipeline size in. 6 6
PLF =
(Actual stream days) (Daily rated capacity) (1) • Internal reasons: Recurring. Examples include pro-
cess problems, mechanical breakdown of equipment,
Actual stream days 100 planned shutdowns and more
OSF = • Internal reasons: Non-recurring. Examples include
Annual design on stream days (2) lack of finished product sales, effluent treatment, lack
of byproduct sales and more
PLF OSF • External reasons: Recurring. Examples include utility
Overall capacity utilization = failure, raw-material shortages and more
100 (3) • External reasons: Non-recurring. Examples include
worker strikes, natural calamities and more
Actual annual production 100
=
Annual design onstream days x daily rated capacity (4) FFS and RLA analysis
In a chemical process plant, critical equipment and pip-
The performance of the plant is studied based on the ing must be evaluated for their fitness for service (FFS),
highest PLF and OSF, on both a yearly and monthly basis. according to API 579 [1], and their potential residual
Data on the highest daily production that is achieved with life analysis (RLA) must also be assessed. The API 579
the present hardware should also be captured. guidelines are designed to ensure that pressurized criti-
In addition to the past production performance of the cal equipment are operated safely. The ability to establish
units, a breakdown of individual equipment must be as- the minimum years of residual life of the critical equip-
sessed to identify the weak areas and arrive at the pre- ment is essential to justify the revamp of the old and well-
dicted performance in the post-revamp implementation maintained plants.
scenario. The best yearly, monthly and daily performance
must be considered in order to find the target capacity of Use of simulation software
the plant and identify the number of stream days that this Simulation software can play an important role during the
target capacity is likely to achieve. evaluation of potential revamp options, so its use is rec-
Analysis of historic downtime factors can also pro- ommended to study the competing process-revamp op-
vide insight. To assess the feasibility of the plant oper- tions. Such modeling can help the team to substantially
ating at higher capacity, the best-achieved PLF (on a reduce the time needed to study the technical feasibility
monthly basis), and the highest load achieved, should of revamp options. However, great care must be taken to
be considered. ensure the use of most appropriate thermodynamic mod-
In any process plant, onstream days are lost due to eling options that are suitable for the plant and its com-
various factors — including process problems, mechani- ponents, fluid properties, process conditions and so on;
cal breakdown of equipment, raw material shortages, otherwise the results can be wrong. Appropriate use of
planned shutdowns, finished product sales, effluent simulation software can reduce the time required to carry
treatment and byproduct sales (if any). Such lost days — out the revamp projects, and help the team to identify an
which contribute to a loss of overall capacity utilization optimized, cost-effective process path, based on an eval-
— should be tabulated, and the associated causative uation of proposed process sequence changes given the
factors noted and tabulated. various constraints.
From the weak-area analysis, one can estimate the The various revamp options are studied from a techni-
inherent capacity potential of the plant and identify in- cal and financial point of view, a suitable process path
dividual equipment components or sections that are is selected and the equipment that create a bottleneck
becoming a bottleneck to maximum capacity utilization. for the desired revamp option are identified. Once the
Sometimes the plant capacity is affected by external additional equipment and piping are identified (per the
circumstances, such as feedstock supply issues (for in- proposed expansion schemes), the required hookup
stance, urea plant capacity is impacted by the capacity points and tie-in connections must be identified. As
of upstream ammonia plants) utility supplies and more. noted, to reduce the impact of these hookups, they should
Dividing these factors into recurring and non-recurring — wherever possible — be undertaken in conjunction
factors will also provide insight into the priorities needed with short shutdowns that are planned for preventive
to address the problem. maintenance.
50 Chemical Engineering www.chemengonline.com december 2015
Environmental and safety impacts TABLE 2. Typical Design Velocities of Fluids in CPI PIPELINEs
Environmental-impact assessment studies should be Type of line Allowable velocity (max), m/s
conducted during the conceptual stage to evaluate the Suction lines for the pump 1
positive and negative impacts of the proposed engineer- Discharge lines for the pump 2–3
ing changes on the environment, and to arrive at the so- Fire water 5
lutions to mitigate the adverse impacts, if any.
Gravity lines 0.6–0.7
Safety is always a paramount consideration. The team
must ensure that the proposed plant revamp, and all re- Low-pressure gas 20
vised process schemes, conform to the latest codes and High-pressure gas 15
safety norms. Hazard operability (Hazop) studies of the Low-pressure steam 20
process schemes during the basic engineering-design High-pressure steam 15
package stage, front-end engineering-design stage, and
the detailed engineering stage should be conducted. Dur-
ing the implementation stage, periodic technical audits the percent flooding velocity with the revamped through-
should be conducted to see that the construction is pro- put. If the flooding velocity is greater than 80%, the pack-
gressing according to design intentions. ings are replaced with ones that offer lower packing factors
Hazardous-area classification drawings of the plant and higher surface area per specified volume. However,
are developed, and existing electrical considerations and adequate wetting of the packing must be ensured, ac-
other instruments are evaluated and changed accord- cording to design guidelines, and circulation rates of liq-
ing to the modified hazardous area classification of the uids must be enhanced accordingly, if needed.
plant. Quantitative risk analysis (QRA) is also conducted Packed towers that contain ceramic packings have a
to submit to the statutory authorities, and any onsite and tendency to flood at lower gas velocities. Hence, in some
offsite emergency plans must be revised, as needed. cases, such packings may be replaced with steel packings
Similarly, a safety integrity level (SIL) analysis should (after conducting the technical suitability check) to help re-
also be conducted according to BS IEC 61511[3] and duce the flooding velocity and increase throughput.
BS IEC 61508 [4]. And, all safety-instrumented functions Pumps. Pumps are very important and often provide a
(SIF) of the instruments are to be SIL 2 (minimum). relatively simple revamp opportunity, to take advantage
of advancements in pump technology. The throughput
Debottlenecking individual equipment systems required at desired plant capacity is determined, and
Different strategies are available to debottleneck different the characteristic head-versus-capacity curves, re-
equipment components and systems. Some examples quired net positive suction head (NPSH), and other key
are discussed below: characteristics should be studied. Normally, pump
Trayed columns. The design data of the distillation manufacturers indicate three impellers (mini-
column should be studied, preferably using process mum, normal, maximum) that are suitable for
simulation software. The column is simulated for both any duty. The possibility of using a larger-sized
the existing operating conditions, and for desired higher impeller diameter should be studied, considering the
throughput or changed feed composition. The liquid and head and capacity requirements (Figure 1).
vapor rates for each tray, along with their physical proper- As the pump capacity increases, required NPSH
ties, are obtained. After obtaining the column profile and (NPSHR) increases. Hence, the available NPSH
liquid-vapor-traffic details in the column, the tray hydrau- (NPSHA)should be checked, to avoid cavitation of the
lics are calculated and suitable recommendations are pump at higher flows. The motor’s suitability should
made, regarding changes made to the weir height, the also be verified. Many successful revamps were car-
number of holes, pitch, the diameter of the holes (con- ried out by changing the impellers to those with larger
sidering the flooding conditions) and more. Tray vendors diameters. The team should also carry out a design
should be contacted when considering revamping the check to ensure that the piping material classification
distillation column trays. The team should ensure that is still suitable for the pump’s discharge piping.
the reboiler and condenser are rated for the maximum Instruments. Instruments such as flowmeters (orifice,
throughput expected. venturi and mass flowmeters), pressure indicators, tem-
Many advanced separation technologies that are avail- perature transmitters, level instruments and so on should
able today allow for higher-capacity trays to be retrofit- be rated and studied in detail for the proposed changed
ted into distillation columns. Similarly, the suitability of condition. Since orifice meters often give rise to higher
advanced structured packings can also be considered pressure drop, they may be replaced with mass flow-
when planning a revamp of distillation columns in petro- meters. Similarly, level instruments based on differential
leum refinery and other critical CPI applications. Many pressure can be replaced with non-contact type, radar-
present-day structured packings can help revamped type level instruments, which tend to be more accurate.
columns to improve capacity by 40–50%, while reducing Normally, the orifice plates in flowmeters are maintained
pressure drop across the column. with  ratios — that is, the ratio of orifice plate bore di-
Packed columns. In the late 1980s, Raschig rings were ameter (d) to pipeline diameter (D) — of 0.3 (minimum) to
popular in chemical process operations. A study of pres- 0.7 (maximum). The orifice meters are rated for the tar-
sure drop of the packed column at the rated capacity get throughput and the pressure drop across the orifice
should be carried out to determine the pressure drop per element is determined. If the pressure drop is too high,
foot of packed column. Such a study should also identify the orifice plates are changed to those of higher  ratios,

Chemical Engineering www.chemengonline.com december 2015 51


to address the pressure drop issue without changing the TABLE 3. Allowable Pressure and Temperature ratings, per [7]
transmitter. To keep the  ratio less than 0.7 for a given Flange rating, Allowable pressure Allowable temperature (max)
pressure drop across primary element, either or both the per ANSI B16.5 (max) kg/cm2
orifice plate and the transmitter is changed. 150 class 18.3 93.3°C /200°F
Control valves. The flow through a control valve depends 300 class# 47.8 93.3°C /200°F
on its capacity, or so-called CV value (Equation 5), which is
defined as the flowrate in m3/h of water at a temperature
of 60°F with a pressure drop across the valve of 1 psi. The with changing the heat exchanger. Also, increasing the
rule-of-thumb rule is that the CV is roughly 10D2 (where number of baffles on the shell side to increase the heat
D is the size of the control valve in inches). For example, transfer coefficient should be considered. In the case of
the CV of a 2-in. control valve is roughly 40. The CV value plate heat exchangers, additional plates can be added
is recalculated according to ISA 75.01.01[2] with the new to increase the heat transfer, in consultation with original
flowrate, inlet pressure and allowable pressure drop. equipment manufacturer.
Normally, the control valves in the original design of the Limitation in line sizes. All of the line sizes are checked
plant are kept one size lower than the pipe line diameter, using the standard velocity criterion. Typical standard ve-
and their rated flow is specified as 1.7 times the normal locity criteria are shown in Table 2.
target flow, or 1.3 times the maximum target flowrate. The lines are checked for pressure drop. In case the line
Since the flowrate is specified as 70% higher normal pressure drop is high, the lines are changed to provide
flowrate, or 30% higher maximum flowrate, the control larger-diameter pipes. Special attention must be given for
valves will be suitable to handle the revamped target gravity-flow lines, as the allowable velocity is in the range
flow, which is 20–30% more than the design flowrate. of 0.6–0.7 m/s and sufficient slope must be ensured.
Hence, for a proposed 20–30% plant load increase, the The piping material thickness (according to ANSI B
existing control valve will normally be sufficient. If the 31.3) and flange ratings (ANSI B16.5) are checked to be
CV of the control valve is not sufficient, the team may sure they comply with higher pressure. In some cases,
consider either changing the trim of the control valve, the flange rating will be sufficient, as there is often a wide
or installing one with a higher CV. Equation 5 is used to margin available, as shown in Table 3.
calculate the CV . Thus, if a line that was designed for 10 kg/cm2 is going
to experience a pressure of 12 kg/cm2 at 90°C, then the
Q / 0 flange rating of 150# need not be changed. However, the
CV = 1
actual pipe thickness should be measured and checked
N1 P (5) for its suitability in the revamped design pressure condi-
tion. Sometimes no piping needs to be changed — for
Where: instance, if the design pressure in the revamped condition
Q = the flowrate through the control valve, m3/h is less than that of the original process. One example is an
N1 = a constant (8.65 x 10-2), from ISA 75.01.01-2007 ammonia synthesis section, where pressures have come
(IEC 60534-2-1 Mod), Table 1 [2] down from 200 kg/cm2 to 140 kg/cm2.
1 = density of the fluid, kg/m3 Pressure safety valve (PSVs). When the plant runs
0 = density of the water at 15°C, kg/m3 at higher revamped capacity, all of the PSVs must be
P = differential pressure, kPa checked according to API 520 [5]. The team must evalu-
Table 1 shows a typical calculation of CV before and ate the nozzle area suitability and the rating of the inlet
after revamp flowrates, and shows how the existing con- and outlet piping, after recalculating the fluid-relieving
trol valve must be changed to the pipeline size for a 20% rates associated with the new throughput. PSVs are
increase in flowrate. changed if they are found to be unsuitable. In the case of
Control valves should also be checked for noise lev- feedstock changeover, PSVs must be also be checked
els. Controllability and rangeabilty are also important for for changes in fluid properties such as molecular weight,
revamping the valve. Revamps involving control valves compressibility factors and so on.
should always involve vendor cooperation. If the revamp Compressors. Various options for revamping the com-
is not able to bring the process into the controllability pressors should be studied initially. Various revamp op-
range, either the valve should be replaced with one of tions include the following:
higher size, or fine feed-control valve can be added par- 1. Installation of a suction booster
allel to the existing control valves. 2. Installation of a parallel compressor
Heat exchangers. The existing heat exchangers should 3. Changing internals in the low-pressure and high-pres-
be checked for any excess available surface area, by rat- sure casing, along with steam turbine upgrading
ing them using standard software modeling packages. In 4. Providing a chiller at the suction inlet and changing the
general, an existing heat exchanger provides enhanced intercoolers. A chiller can be installed to reduce the
heat exchanging capacity if the pressure drop across the gas temperature and increase the volumetric capac-
tube side or shell side is increased. ity of the gas and reduces the power requirement. In
If the heat exchanger is downstream of a pump, the cases where the drive needs to be changed, this can
team should consider increasing pump head, which be applied.
would increase the allowable pressure drop across the 5. Change of compressor type. In older-generation urea
heat exchanger. There may be a tradeoff between the plants, urea reactors operated at 200 kg/cm2, and
operating cost of the pump and fixed cost associated they fed the CO2 to the urea reactor; Historically, CO2
52 Chemical Engineering www.chemengonline.com december 2015
compressors have been reciprocating-type, which sis converter pressures were reduced to 135 kg/cm2
incur high energy costs. As the pressures in present- (from an initial level of 200 kg/cm2), as a result of the
day urea reactors have come down to 135 kg/cm2, introduction of radial basket converters instead of the
centrifugal compressors can be used instead, which older-generation axial converters. By retaining the same
helps to reduce operating costs as well as mainte- high-pressure converter shell, one can change the con-
nance costs). verter baskets to radial ones, which helps to reduce
Effluent treatment plants (ETP). Worldwide, waste- pressure drop.
water-treatment plants are typically designed with high Catalysts play a vital role in enhancing the reaction rate.
safety margins, to cater to shock loading or sudden The use of advanced catalysts should be considered,
peak loading of effluents containing high chemical oxy- where possible. For example, in sulfuric acid plants, va-
gen demand (COD). However, when a plant is stabilized nadium pentoxide (V2O5) is typically used as a catalyst.
and optimized, the generation of wastewater containing If an improved cesium catalyst is added to the reactor,
high COD is drastically reduced. the SO2 to SO3 conversion can be increased, and the
The following methodology should be adopted while emission of SO2 can be reduced, generally to far below
checking the capacity of ETP that are based on an acti- the statutory limits.
vated sludge process during revamp planning: Storage tanks. If the process revamp is based on a
1. Evaluate existing facilities by collecting operating data “more in/more out” concept — that is, more fluids will be
for one month and developing a statistical analysis of flowing into and out of storage tanks — then the team
various parameters. must check the capacity of “breather” valves and emer-
2. Check the design basis and the design volume of the gency vents according to API 2000 [8]. If the breather
aeration basin, thickener and clarifier. valves need to be replaced, the pressure settings may
3. Evaluate the operating case using the above design be adjusted in consultation with vendors, according to
basis. the applicable codes.
4. Calculate the energy requirements of the design and Utilities. During any plant revamp, the capacity of key
operating cases, and quantify the potential for reduc- plant utilities, such as demineralized water, instrument
tion of electrical energy at various loads. air, plant air, steam plants, power, and cooling tower
Flares and knockout drums. Flare systems, including should also be checked to be sure they will support the
knockout drums, must be checked before embarking on proposed revamp. Offsite facilities related to raw-mate-
a plant revamp. Flares are used to ensure plant safety, rials receiving, tank farms, and product-storage capaci-
by flaring hydrocarbons in case of emergency conditions ties must also be studied and related personnel require-
such as power outages, fire or blocked discharge. ments must be ascertained. n
While converting the ammonia plant from a liquid fuel Edited by Suzanne Shelley
(such as naphtha) to natural gas, the properties of the
fluid (such as molecular weight, compressibility factor), References
viscosity and density undergo a drastic change and 1. American Petroleum Inst., API 579: Recommended Practice for Fitness for Service, 2nd
have profound effects on height, flare diameter and flare Ed., July 2007.
tip suitability. Calculations must be performed to verify 2. Instrument Soc. of America, ISA 75.01.01-2007 (IEC 60534-2-1 Mod): Flow Equations
for Sizing Control Valves, 2007.
the new case, according to API 521[6]. The goal is to
see whether the existing flare is suitable to handle the 3. International Electrotechnical Commission (IEC), BS IEC 61511: Functional Safety –
Safety Instrumented Systems for the Process Industry, 2003.
changed load and fluid conditions associated with the
4. International Electrotechnical Commission (IEC), BS IEC 61508: Standard for Functional
proposed revamp. Vendor support should be sought, Safety of Electrical/Electronic/Programmable Electronic Safety-Related Systems, 2010.
if needed, and the flare design can be checked using 5. American Petroleum Inst., API 520: Sizing, Selection and Installation of Pressure Relieving
manual calculations, spreadsheet calculations and flare- Devices, Part 1, 8th Ed., 2008, and Part 2, 5th Ed., 2003.
specific computer software. 6. American Petroleum Inst., API 521: Pressure Relieving and Depressurizing Systems, 5th
Reactors. Reactors are the heart of chemical process Ed., 2007.
operations. Efforts should be made to maximize yield and 7. ASME/ANSI B16.5: Pipe Flanges and Flanged Fittings, April 2013.
conversion rates in the revamp scheme. If, following the 8. American Petroleum Inst., API 2000: Venting Atmospheric and Low Pressure Storage
reaction, raw materials remain unconverted, they must Tanks, 7th Ed., March 2014.
be separated and recycled back to the reactors. This
consumes utilities, thereby increasing energy consump- Author
tion. If conversion rates in the reaction are increased via Koya Venkata Reddy is senior manager, process engineer-
a revamp, the recycle ratios will be drastically reduced. ing, at FACT Engineering & Design Organization (FEDO), a div. of
In one urea plant, a revamp involved the following Fertilizers and Chemicals Travancore Ltd. (FACT; Udyogamandal
683501, Kochi, Kerala, India; Phone: +91-484-2568763; Email:
changes: Introduction of higher-capacity trays in the koyareddy@yahoo.com). He has 24 years of experience in chem-
urea reactor in the ammonia plant; changing the con- ical plant operations, including expertise in the fields of process
verter baskets from axial- to radial-type in the ammonia control, process design, process risk analysis, Hazop analysis, pro-
cess simulations, environmental management and plant revamps.
converter in the caprolactam plant; using an enriched- He is a recipient of FACT’s Merit Award. Reddy holds a Bachelor of
oxygen supply to the cyclohexanone reactors with in- Technology degree from Andhra University (Visakhapatnam) and a
troduction of improved safety features. These changes Master of Technology degree in project management from Cochin University of Science
and Technology. He also received an M.B.A. in finance from Indira Gandhi National Open
were able to increase the conversion rate, increase over- University (IGNOU; Delhi). He is a lifetime member of the Indian Inst. of Chemical Engi-
all production and decrease energy consumption. neers (IIChE) and a member of the Institution of Engineers (India).
In the ammonia plant’s synthesis section, the synthe-

Chemical Engineering www.chemengonline.com december 2015 53


Feature Report

Point-Level Switches for


Safety Systems
Industries that manufacture or store potentially hazardous materials need to employ
point-level switches to protect people and the environment from spills
Bill Sholette
Endress + Hauser

In Brief
The need for point
level switches

The need for testing

Types of point-level
switches

Summary

S
afety is an important and common Figure 1. Preventing overfilling of chemical storage tanks
subject of discussion in the chemi- requires proper selection of high-high point-level switches
cal process industries (CPI) today. model. Determining what may go wrong and
Conversations on safety include then taking steps to reduce the possibility by
many topics, such as risk assessment, risk adding safety procedures, retention dikes,
mitigation, and tolerable risk. Acronyms safety instrumentation and so on, are univer-
like SIS (safety instrumented systems), SIL sal to any safety program.
(safety integrity level) and PFD (probability of The following are assessments of point
failure on demand) and others have become level switches as they are used in over-
part of the safety lexicon in CPI facilities fill prevention safety programs. We review
throughout North America and the world. All some basic concepts and look at some of
of these terms and acronyms can be con- the common technologies used to prevent
fusing, complicating what steps need to be overfilling of vessels (Figure 1). The positive
taken to make a facility safe. and negative aspects of each technology are
Regardless of how the safety concepts are also considered.
labeled, there are a few principles that form
the basis for all safety models. Whether you The need for point-level switches
subscribe completely to the SIS concept or Point-level switches are often used in appli-
have developed your own safety procedures cations designed to prevent accidents. In-
internally, risk assessment and risk mitiga- dustries that manufacture or store materials
tion are the two key concepts in any safety that are potentially hazardous employ point-

Chemical Engineering www.chemengonline.com december 2015 43


level switches to protect people and poses can potentially cause a spill High-high level overfill
the environment. These industries and therefore is considered bad prevention point-level
switch
include oil-and-gas, chemical and practice. Raising the product level to
petrochemical manufacturing. Some test the high-high switch is also spe-
examples of where these safety cifically not permitted per API 2350
switches are used include overfill (American Petrochemical Institute)
and spill prevention on tanks, reten- recommended practices for above-
tion dike level alarms, and seal pot ground storage tanks. API 2350
Normal stop-fill
low-level indication. These critical states that high-high level switches point-level
switch
safety applications require careful must be tested on a regular basis
consideration to make certain the without raising the level to a danger-
best technology is provided for the ously high condition.
given application. Technologies that Depending on the type of point-
are robust with few or no moving level switch being used, the only Figure 2. Located above the normal stop-fill
control, high-high level switches may go for years
parts are preferred. Additionally, a accepted method to ensure the per- without seeing the liquid in the vessel
procedure for testing the integrity of formance of the switch may be to
the switch is critical. remove it from the vessel for testing. ing any maintenance. Trying to start
Providing safe and reliable facilities Removing a switch for testing incurs that car after a year may be difficult.
is a moral and financial responsibility. cost through downtime and lost pro- This represents a higher PFD. PFD
Accidents such as Buncefield 2005, duction, as well as the time and ex- increases with time. Testing provides
Texas City petroleum refinery 2005, pense for personnel to remove the a way to return to a lower PFD.
and the Elk River spill 2014 can switch, perform the test and reinstall Generally speaking, in-situ testing
and must be avoided. By providing the switch. There is also the concern (testing with the unit installed in the
safety systems and instruments to that the switch could be damaged process) may only validate a per-
prevent or mitigate accidental spills during removal and reinstallation, centage of the potential failures. This
and releases, we protect against or that the switch is not reinstalled is known as a partial proof test. As
injury to people, damage to equip- correctly. Either of these scenar- such, the PFD recovery is dependent
ment, environmental damage, and ios would negate the test and the on the proof test coverage (PTC).
ensure that the availability of the switch failure may not be detected. The PTC is based on the percentage
process is maintained. For these reasons, employing a of failures exercised by the proof test.
Due to the nature of the safety re- point-level switch that can be tested The higher the PTC percentage, the
quirement for point-level switches, in-situ (Figure 3) should be the first higher the recovery, and the result is a
they are typically placed in a position choice for safety applications. lower PFD. Since the partial proof test
where they may never be used. That Testing the switch exercises the recovers a percentage of the PFD, it
is, for example, a switch for high-level point-level switch and may bring to does not return to the entire original
overfill prevention is located above light any potential failures. The intent installed state. The result is that with
the highest point the level should ever is to validate the switch, with the goal each partial proof test, the PFD will be
reach. These switches are often called being to return the switch as close a little higher after each test. To cor-
high-high level because they are as possible to its original installed rect for this “drift,” a full proof test will
above the stop-fill high-level instru- condition. That is, the switch should be required after a determined num-
ment in the vessel (Figure 2). These be validated to “new” condition or ber of partial proof tests. A full proof
high-high switches may go for years as close as is reasonably possible. test will typically require removing the
without ever having the level reach Conceptually, this refers to probabil- switch from the process for testing.
them because reaching the high-high ity of failure on demand (PFD). When Clearly, a level switch product with a
switch is an accidental occurrence or a point level switch is first installed, high PTC will allow for a longer period
noncompliant condition. it has a low PFD. Over time the PFD of in-situ partial proof testing resulting
of the switch increases. Testing the in savings and process availability.
The need for testing switch re-establishes the PFD to a
Because of this, it is imperative that lower number. Types of point-level switches
the safety switch has a means of A good analogy of the PFD con- There are many point technologies
testing on a regular basis to ensure it cept would be the purchase of a available that can be used for level
will operate in the event of an actual brand new car. You park the car in indication. Because of the critical
emergency. This test must exercise your driveway and retire for the eve- nature of safety switches, some
the entire switch — not just the con- ning. The next morning you get up technologies are better suited than
tact closure or output — to expose and go to start your car. The expec- others for this task. Let’s take a look
any potential failures, and should tation is that the car will start. This at some of these technologies and
not require raising the product level represents a low PFD. Now, leave why they may or may not be good
to the switch point. Raising the level that same car in your driveway for choices for safety applications.
up to a high-high switch for test pur- a year without starting it or perform- Float point-level switch. Float
44 Chemical Engineering www.chemengonline.com december 2015
PROVEN
PERFORMANCE

ROTOFORM
GRANULATION
FOR PETROCHEMICALS
Figure 3. The ability to test in-situ validates the functionality of the safety switch while reducing main-
tenance and downtime
AND OLEOCHEMICALS

switches, as the name implies, utilize parts and the potential for a lack of
a float that changes position due to buoyancy are critical failure points.
buoyancy and indicates presence From the standpoint of safety appli-
of a liquid. The float may move on a cations, floats should be avoided.
vertical shaft and trip a magnetically Ultrasonic gap point-level switch.
coupled reed switch or may pivot Ultrasonic gap switches are com-
on an access providing a mechani- prised of two piezoelectric crystals
cal internal switch to activate. The situated on opposite sides of a gap.
appeal of float switches is that they One crystal is excited electrically and High productivity solidification of
are simple devices and relatively in- generates acoustic energy that is products as different as resins, hot
expensive. However, the mechani- directed across the gap toward the melts, waxes, fat chemicals and
cal nature of a float and the moving second crystal. With air or gas in the caprolactam has made Rotoform® the
granulation system of choice for
parts that can hang up or bind due to gap, the energy is not strong enough
chemical processors the world over.
coatings makes them questionable to reach the second crystal. Once Whatever your solidification
for use in safety applications. the gap fills with a liquid the acoustic requirements, choose Rotoform for
The ability to test a float switch is energy is coupled through the liquid reliable, proven performance and a
also suspect. Some manufacturers molecules, reaches the second crys- premium quality end product.
provide a lift arm to physically move tal and completes the circuit, indicat-
the float to make it change state ing that the liquid is present.  High productivity –
from normal to alarm. This test is Ultrasonic gap switches have no on-stream factor of 96%
insufficient to exercise potential fail- moving parts to wear or hang up,  Proven Rotoform technology –
ures, such as leaking floats, and may which is an advantage over mechan- nearly 2000 systems installed
in 30+ years
not identify binding or heavy coat- ical switches, such as floats. How-
 Complete process lines or
ings. Some test arms are fitted with ever, materials that leave coatings retrofit of existing equipment
magnets that will release if the float and materials that have suspended  Global service / spare parts supply
is heavy due to leakage or coating, solids, or are aerated, will block the
but even this precaution is suspect. acoustic energy, causing a failure.
As such, the only true way to test In-situ testing of ultrasonic gap
the float switch is to remove it from switches that validate all potential
the vessel for testing, or to raise the failures is not possible. Some manu-
product level to the high-high switch, facturers provide test buttons that Sandvik Process Systems

which, as previously discussed, is are used to test the switch. This Division of Sandvik Materials Technology Deutschland GmbH
Salierstr. 35, 70736 Fellbach, Germany
not permitted. test operates in one of two ways. In Tel: +49 711 5105-0 · Fax: +49 711 5105-152
info.spsde@sandvik.com
Floats are best suited for simple some products, there is a second set www.processsystems.sandvik.com
non-critical applications. Moving of crystals that are wired together.
Circle 18 on p. 190 or go to adlinks.chemengonline.com/56204-18

Chemical Engineering www.chemengonline.com december 2015 45


SANDVIK_Chemical_ad_55.6x254_MASTER.indd 1 09/02/2015 14:48
These two sets of crystals are used
for the actual measurement, and
when the test button is depressed,
an acoustic frequency travels from
one crystal through a wire to the sec-
ond crystal, indicating a valid test.
The assumption is that if the two test
crystals operate properly, so will the
measurement crystals.
The second approach to this test
is to increase the frequency on the
Figure 4. Vibronic point-level switches (left) are an active technology with constant self-check function
actual measuring crystal, which al- built-in (right) to ensure functional integrity
lows the acoustic energy to travel
through the metal in the gap to the wall or an added ground rod or plate. requirement, a capacitance point-
second crystal completing the cir- As the material in the vessel rises, it level switch may be the best choice
cuit and validating the test. Neither covers the sensing element and the for a safety installation. This is par-
of these tests addresses one of the capacitance increases. The output ticularly true of applications involv-
most common failures in ultrasonic of the electronic unit changes state ing extremely viscous materials that
gap switches — namely, coating or to indicate presence of material once coat sensing elements heavily. It is
plugging of the gap itself. Coatings the capacitance exceeds a preset important to make sure the capaci-
in the gap or material plugging the switch point. tance switch selected provides ac-
gap will prevent the acoustic energy Capacitance point-level switches tive coating-rejection technology to
from crossing the gap and indicating have several advantages over pre- compensate for the coatings.
when the liquid is present. viously discussed technologies. Vibronic (tuning fork) point-level
A second common problem with There are no moving parts to wear switch for liquids. Vibronic point-
gap switches is dis-bonding of the or hang up. Internal diagnostics level switches, also called tuning
crystal. The dual-crystal design will monitor data, such as the base ca- forks, operate by vibrating the fork
not detect this failure. The second pacitance. Reduction of the base at a resonant frequency in the un-
design may detect dis-bonding, capacitance would indicate a wiring covered state. When process mate-
but increasing the frequency could failure or a sensing element that has rial covers the fork it causes the fre-
provide a valid test while the switch lost mass due to damage or corro- quency to shift down, indicating the
may fail to operate at its normal fre- sion. Failures can result in a switch presence of the liquid and changing
quency. Also, neither of these test going into fault mode or activating the output of the switch.
methods will detect potential failures an “alarm” contact. There are a number of advantages
due to liquids with suspended solids One disadvantage to capacitance to vibronic switches (Figure 4). First,
or aeration. switches is that they require calibra- vibronic switches are an active tech-
Performing a valid test on an ultra- tion. Initially, the base capacitance nology. Because they are constantly
sonic gap switch requires removing needs to be balanced, and then an vibrating, additional diagnostics are
it from the vessel and testing it in a additional set-point capacitance possible. The frequency of the fork is
sample of the material from the ves- is added. There are capacitance monitored to determine the covered
sel. For these reasons, ultrasonic gap switches that “calibrate” themselves. or uncovered state. But changes in
switches are best suited for general, These switches follow the same pro- frequency can also indicate damage
non-critical level applications. They cedure as manual calibration with the or corrosion to the fork, heavy coat-
should not be used for high-high safety exception that it is done internally in ings, and objects jammed between
and spill-prevention applications. the electronic unit. If the calibration is the forks. Any of these conditions will
Capacitance point-level switch. not performed correctly it is possible result in a fault output. The electronic
Capacitance point-level switches are that the switch will not respond to in- unit is constantly running self-test
based on a capacitor. A capacitor is creasing material level in the vessel. routines to identify these and other
made of two conductive plates sep- There are products available that potential faults.
arated by a dielectric insulator. The incorporate testing features so in- Some manufacturers have devel-
capacity of the capacitor is based on situ testing can be performed to en- oped additional functions to ensure
the size of the plates, the distance sure the functionality of the switch. the operation of the switch in safety
between the plates, and the dielec- The PTC percentage for a capaci- systems. One such function is to
tric constant of the insulating mate- tance-switch partial proof test tends provide a live signal superimposed
rial between them. In a capacitance to be low. The result is that the in- on the current signal (Figure 5).
point-level switch, one plate of the tervals between a required full proof This live signal is constantly switch-
capacitor is the active center rod test tend to be fairly short, increasing ing from one current to another and
of the sensing element; the second downtime and maintenance. back. This switching current certifies
plate of the capacitor is the vessel Depending on the application that the current signal is not stuck
46 Chemical Engineering www.chemengonline.com december 2015
4 to 20 mA with live wire signal “best practice” and should be fol-
lowed for safety overfill applications.
The reliability of a point-level switch
Safety
PLC with continuous live signal will ex-
ceed other technologies in demand-
ing critical services. As always, the
application will determine the best
technology for safety devices.
Figure 5. Vibronic tuning forks with continuous live signal provide the highest in reliability and maximum
in-situ proof test coverage Summary
There are many point-level switches
and ensures that the current will shift concept is to poll the two transmit- available in the market today. We
when the fork changes from the un- ters and to shut down the process if have reviewed some of the most
covered to covered state. the level reaches a preset high-high common types in this article. As we
Vibronic point-level switches have value, or if the two transmitters differ have discussed, some technologies
no moving parts to hang up or wear by a predetermined percentage. As in are better suited for safety systems
out. Additionally, there is no calibra- the previously mentioned use of con- than others. Technologies with mov-
tion required so you can be sure the tinuous level for overfill prevention, ing parts and those that cannot be
switch is set up properly. the thought is that the continuous easily tested are best suited for non-
Another advantage to vibronic signal provides a measure of security critical level applications.
point-level switches is the ability to that a point-level switch cannot offer. Trends toward using continuous
perform in-situ partial proof tests. However, there are a few concerns level transmitters for overfill preven-
The sophistication in design of these with this approach. tion on the surface have some ap-
switches employs redundant circuitry First, as in the previous example, pealing merits. However, the advan-
along with the diagnostic capability it is possible for a process condi- tages of using a separate point-level
previously discussed. These features tion such as foam, condensation, switch with live signal are clear and
added together result in an extremely or buildup to cause a signal to lock. continue to provide a “best practice”
high proof-test-coverage percent- Using two different technologies in solution to overfill prevention.
age. The high PTC provides the ability redundancy may provide an advan- From an overall standpoint of so-
to test in-situ for an extended period tage in that a process condition that phisticated diagnostics and ease of
of years without having to perform a causes one technology to fail may commissioning with no calibration, vi-
full proof test. Some manufacturers not affect the other. Using two differ- bronic point-level switches are excel-
provide products that will not require ent technologies can introduce other lent choices for safety systems. The
a full proof test for as many as twelve problems. An example would be the extremely high proof-test coverage
years, greatly reducing testing cost accuracy of guided-wave radar may and long intervals between full proof-
and ensuring process availability. be different than a differential pres- test requirements result in the highest
Continuous technologies. Some sure transmitter under the same cost savings and plant availability. From
plants rely on continuous level tech- process conditions. The difference these standpoints, vibronic point-level
nologies, such as free-space radar, in output, while typically negligible, is switches are the clear choice for most
guided radar or ultrasonics, to pro- often hard for operators to overlook. safety-system applications. n
vide for overfill prevention or function There is still the issue with API 2350 Edited by Gerald Ondrey
as a point-level device. Their thought recommended practices.
process is that with a continuous If the overfill prevention transmitter Author
level technology, they would know if needs to be separate from the tank Bill Sholette is the Northeast re-
something was wrong with the trans- gauging transmitter, which continu- gion level product manager at
Endress + Hauser (2350 Endress
mitter, because they have a continu- ous level transmitter is which? This is Place, Greenwood, IN 46143;
ous output. In reality, it is possible certainly a grey area, since they both Phone: 888-363-7377; Fax: 317-
that upset conditions in the process, are outputting continuous level and 535-2171: Email: bill.sholette@
us.endress.com). He has spent
such as foam, condensation and a high-high trip point. Last, it is clear the last 36 years consulting and
buildup, can lock up the signal to a that a high-high point-level switch is specifying on level-measurement
instrumentation. Sholette received
false value. It should also be noted the most reliable and simple device his certification in management and marketing from Vil-
that API 2350 states that instruments for overfill requirements. The avail- lanova University. He previously worked for Ametek,
used to prevent accidental overfilling ability of point-level switches with where he began his career in sales as a regional sales
manager. Later, he moved on to work at Drexelbrook
and spills must be separate from the current output and live signals pro- Engineering as a product manager. In 2012, Sholette
instruments used for tank-gauging vides the same assurance as the came to work for Endress+Hauser as the Level Prod-
the vessel. continuous level transmitter that the ucts business manager for the Northeast region. In this
role, he is responsible for technology application and
A recent trend in overfill preven- switch is operating as required. development of level products. He has also published a
tion has been to use two continuous Using a point-level switch for high- number of white papers and articles pertaining to level
level transmitters in redundancy. The high indication is recognized as a measurement.

Chemical Engineering www.chemengonline.com december 2015 47


Solids Processing

Control Strategies Based On


Realtime Particle Size Analysis
Practical experience
illustrates how Particle size

to achieve better Wireless


TCP/IP network
analyzer

process control Control PC

Jeff DeNigris Wireless PLC


Historian database
Malvern Instruments TCP/IP network

Alberto Ferrari
Figure 1. Integrated system architecture is illustrated here for
Ferrari Granulati an automated mill with an online particle-size analyzer

T
he pursuit of manufacturing This article draws on experience from turnkey integration within a so-
excellence is a strong theme from different industries as they apply phisticated control platform to sensor-
across all of the chemical pro- realtime particle-sizing technology, to only purchase for the inhouse imple-
cess industries (CPI). Improv- demonstrate the various control strat- mentation of simple closed-loop control.
ing operations by eliminating vari- egies that it supports and the benefits This ready availability of realtime mea-
ability and waste, a cornerstone of the that can be derived. surement presents an important oppor-
six sigma approach, is essential for Note: With an online system the tunity to realize a number of important
competitive performance in the global analyzer is typically installed on a practical gains, even before considering
marketplace. In addition, of course, dedicated loop fed from the process the issue of process control.
health, safety and the environment line, while with an inline instrument Full automation — from sample
(HSE) remains the subject of intense the analyzer sits directly in the bulk extraction through the delivery of
scrutiny and concern. Within this cli- process flow. Both approaches provide results to a control system — means
mate, automation of both analysis and continuous realtime data so for the that online analysis completely elimi-
control, can be highly advantageous. purposes of this article the term on- nates the issue of operator-to-operator
Particle size analysis is a measure- line has been used to cover both types variability. In addition, the ability to
ment technique that has already suc- of installation. analyze a much higher proportion of
cessfully completed the transition from the process stream improves the sta-
laboratory to line and become an estab- Moving toward realtime tistical relevance of the data. Equally
lished part of the automation toolkit. It should be stated from the outset important is that the benefits of au-
Reliable, online particle-size analysis that offline analytical capability re- tomation come with a dramatic de-
systems are now commercially proven mains vital. Essential during research, crease in the amount of manual labor
across a number of sectors, on both it is also frequently the norm for final required for analysis, and a significant
wet and dry process streams, with all quality control checks. Furthermore, reduction in the containment and ex-
aspects of project implementation well- certain techniques that yield critical posure risks associated with manual
scoped and understood. With this mat- information have yet to move success- sampling and analysis.
uration has come greater accessibility fully to the process environment. That Investing in realtime measurement
to sectors that previously faced tech- said, for routine process monitoring, technology may be justifiable simply
nical or financial barriers to adoption. an offline regime is far from ideal and on the basis of these practical ben-
Requirements are different in each online measurement, if available may efits, especially where the envisioned
and every case, from a simple single be the preferred option. set-up is relatively simple. The cost
PID (proportional integral derivative) Laser-diffraction particle sizing ex- of installation is offset by savings in
control loop to multivariate statistical emplifies a number of commercially manpower, and the operational team
control or the rigors of operation within proven, online analytical technologies maintains manual control exactly as
a highly regulated environment. Nev- with established credentials for real- before, using the online system solely
ertheless, the building blocks needed to time monitoring. Today it is possible to for particle size monitoring. While this
fashion an optimal solution are there, tailor an online particle-sizing solution approach may deliver improved opera-
for the vast majority of applications. that closely matches user requirements, tions and variable cost gains, it fails
Chemical Engineering www.che.com February 2012 41
Solids Processing

Figure 2. For
Average size and transmission
simple, closed-
2,600 3,600 loop control,
2,800 rpm rpm
rpm 3,200 online analysis
rpm 3,400
rpm 3,800 can be used to
4,000 rpm track particle
rpm
150.00 250.00 size changes that
are induced by
varying mill rotor

Particle diameter, µm
112.50 187.50 speed over time
Transmission, %

Avg{Dv(90)} = 170.54

75.00 125.00
Avg{Dv(50)} = 57.58

37.50 62.50

Avg{Dv(10)} = 9.16
0.00 0.00

11/13/2008 -17:05:00 11/13/2008 -17:15:00 11/13/2008 -17:25:00 11/13/2008 -17:35:00 11/13/2008 -17:45:00

to fully realize the potential that on- Case study: Simple closed loop con- manipulation (Figure 2). A 30-s rolling
line instrumentation offers in opening trol on the basis of realtime data. average Dv50 (median particle size)
up the route to automated control. An An automated mill system recently in- figure from the online particle-size
efficient control-automation strategy stalled at a commercial pharmaceutical analyzer drives this loop.
fully exploits the information stream manufacturing site is shown in Figure To test the response of the system,
provided by realtime measurement 1. It was developed as a widely appli- the setpoint for the loop was reduced
and maximizes return on investment. cable, validated alternative to manual from an initial 58 microns to 50 mi-
mill control using offline particle-size crons, and then back up to the original
Simple closed-loop control measurement. One routine operation value. Despite the absence of compre-
The simplest option when implement- in the company is milling an active hensive loop tuning — proportional
ing automated control on the basis of pharmaceutical ingredient (API), typi- (P) only control was used at this stage
continuous data from an online in- cally recovered through crystallization, — the results were good. Steady op-
strument is usually a single variable to a defined particle size. The particle eration at 50 microns was established
PID control loop. Such an approach size distribution of an API is often a just 30 s after the change was made,
can prove highly productive and be critical quality attribute because of its and the final transition was complete
an efficient way of automating exist- impact on clinical efficacy and drug in under 2 min.
ing manual control strategies. Even product manufacturability. To mill a new batch with this sys-
the implementation of one automated The comminution mill has fast dy- tem, the operator simply selects the
loop changes the process from being namics, making rapid and continuous target particle size and feeds mate-
fixed to becoming responsive. A fixed data acquisition and interpretation es- rial into the mill. Control is then suf-
process translates, or magnifies, up- sential. Since the selected online par- ficiently tight to largely eliminate
stream variations on to the product or ticle-size analyzer has a measurement the production of out-of-specification
downstream process; a responsive one rate of four complete particle-size dis- material. Contrasting this control
either erases or reduces their impact. tributions per second, it can efficiently scheme with offline or manual analy-
For milling, a common approach is track even this swiftly changing pro- sis quickly brings the multiple ben-
to automatically vary either mill speed cess in fine detail. efits into sharp focus:
or downstream separator variables This fully integrated system uses up- • When using offline particle-size
(classifier speed, for example) to meet graded programmable-logic-controller analysis, each new batch required a
the product specification (as in the ex- (PLC) code and proprietary software potentially lengthy iterative process
ample below). An exactly analogous to handle data exchange between the to determine the appropriate rotor
strategy in emulsification processes main hardware units. The operator speed, in order to meet the defined
allows key variables, such as pres- interacts with the central controlling specification. Eliminating this step
sure, to be automatically manipulated PC via the mill HMI (human machine has saved both time and material.
in order to control droplet size. Since interface) and can do the following: • Prior to automation, the rotor
online particle-size analyzers are com- input set points; start and stop the speed was fixed for each batch,
mercially available for both wet and mill or analyzer remotely; perform based on a test sample. However, if
dry process streams, any unit involv- background tests; and receive particle the sample was not representative
ing comminution to a defined particle size results. A closed control loop links of the batch, or if segregation had
size can potentially benefit from this particle size with mill rotor speed, the occurred, the rotor speed would be
very basic type of automation. principal operating parameter for size less than optimal and the product
42 Chemical Engineering www.che.com February 2012
Particle High
size efficiency
product separator
quality
Separator the future this should ease the inte-
speed
To gration of both process and laboratory
storage
silo
systems. Enabling software, based on
this specification, is already avail-
Elevator From able commercially. These advances
power Exit water spray clinker
air flowrate silos will reduce the difficulty and cost of
(separator
feedrate) Elevator implementing customized multivari-
conveyor Feed ate control strategies, bringing the
rate potential rewards within the reach of
Exit
temperature more manufacturers.
Ball mill Case study: Multivariate control
LEGEND
of a heavy commodity milling cir-
Controlled variables
Manipulated variables
cuit. In 2006, Vulcan Materials Com-
pany (Birmingham, Ala.) made the
Florida Rock Industries – Thompson Baker Cement Plant (Vulcan Materials Co.) decision to transform control of its ce-
ment finishing circuit (Figure 3). The
Figure 3. This upgrade to a cement finishing circuit for a milling clinker (and gypsum project involved the three following
and limestone) helps achieve the fineness required to meet final cement specifications significant changes:
• Switching from Blaine measurement
inconsistent. Since the automated ensure that the total feedrate to the to laser-diffraction particle-size anal-
mill responds to, and compensates mill remains constant. In combina- ysis, with the intention of more pre-
for, any variability in the feed tion with the fixed operating strategy cisely targeting cement performance
stock, the product particle size has these loops ensure that exceptional and accessing online technology
become extremely consistent. product quality is achieved at com- • Adopting online, rather than offline
petitive cost. analysis for process control
Multiple control loops As with the previous example, the • Selecting and installing a powerful
The next step beyond a single control architecture of the control loops em- model-predictive-control package to
loop involves multiple, simple control ployed here is relatively simple: one automate process control
loops, which enable the parallel ma- single process variable manipulated Vulcan Materials installed a propri-
nipulation of a number of variables to on the basis of a single measured etary solution for multivariate con-
simultaneously meet product quality, variable. Nevertheless, they are trol and an online laser-diffraction
variable cost and throughput goals. For highly effective with each loop effi- particle-size analyzer. At the heart
example, at its plant close to Verona in ciently targeting one specific aspect of the control package is a multivari-
Italy, Ferrari Granulati mills very fine of process performance. ate process model that is tuned using
marble powders of exemplary quality plant data to accurately predict plant
[1]. Three discrete products are mar- Multivariate process control performance from a range of inputs.
keted with Dv50s in the range of three Just as a seasoned operator bases Automatic manipulation of process
to eight microns. Here, online particle- manual control decisions on every variables, on the basis of these predic-
size analysis has been used extensively piece of relevant information, the tions, achieves plant performance tar-
to develop the design of the mill and a most sophisticated automated control gets, which are as follows:
control strategy for the milling circuit relies on an array of data, rather than • Maintain product quality
(mill and associated classifier). a single input. Multivariate process • Reduce variability and improve op-
The adopted strategy relies largely control systems take in and use data erational stability
on fixing a number of process variables, from a number of sources and, in com- • Maximize fresh feedrate subject to
at values defined through detailed bination with a process model, provide equipment constraints
optimization trials that are based on multiple outputs, simultaneously ma- The process model runs in real time,
realtime measurement. These values nipulating various operational param- employing an integrated steady-state
have been defined for each product. eters. Such systems work within well- optimizer and dynamic controller to
Two independent, automated control defined boundaries to target optimal drive the system toward optimal op-
loops are, however, applied to opti- operation at all times. eration within the above constraints.
mize mill performance and maximize Quite recently, steps have been The impact of changing manipulated
plant throughput on an ongoing basis. taken to reduce one of the barriers to variables is projected into the future;
One loop maintains a prescribed pow- implementing multivariate process predictive control ensures that mul-
der depth on the table of the vertical control: the difficulty of integrating tiple performance targets are met
roller mill to ensure efficient commi- analyzers from different suppliers. simultaneously and that process out-
nution and prevent excessive wear of The new OPC Foundation Analyzer puts are as close as possible to desired
the mill. The other controls the rate of Device Integration (ADI) specifica- reference trajectories. Optimization
fresh feed to the mill with reference to tion [2] provides a common standard procedures are repeated each time
the recycle rate from the classifier, to for instrument manufacturers. Into process values are re-read, following
Chemical Engineering www.che.com February 2012 43
Solids Processing

a change, in order to maintain the fu- The installed solution has pushed operation presents an opportunity for
ture prediction-horizon period. This the circuit into a new operating re- improved control, reduced risk and fi-
is termed “realtime receding horizon gime, uncovering better control strate- nancial gain. For online particle-size
control” and is characteristic of model gies than had been identified through analysis proven automated systems
predictive control. manual operation. Significant benefits have developed to the point of wide-
Here, a number of loops are operat- have accrued, including the following: spread availability, making this op-
ing in tandem (Figure 3). Manipulat- • A 20% reduction in specific energy portunity financially and technically
ing clinker feedrate and separator consumption feasible across a broad range of manu-
speed controls product quality. These • An improvement of 15% in one-day facturing sectors. Options for invest-
same two parameters are also used, strength levels (a primary perfor- ment now range from fully integrated
together with air flow through the mance indicator) turnkey solutions to sensor-only pur-
mill, to drive separator feedrate (mea- • An increase in throughput in excess chase for in-house implementation.
sured as elevator power) toward a de- of 15% Continuous, realtime analysis of a
fined high limit. Air flowrate through These savings are not individually at- critical variable, one that directly influ-
the mill and exit water spray vari- tributable to either the analyzer or the ences product performance, provides a
ables control the temperature of exit- control package but arise from symbi- platform for developing and imple-
ing material. Finally, a stabilizing loop osis between the two, which has un- menting the very best automated con-
minimizes a function defined as “mill locked the full potential of each. The trol strategies for a given application.
condition”, calculated from the rate of payback time for the entire project is Such strategies deliver multiple eco-
change of elevator and mill power. The estimated at just over one year, based nomic benefits in the form of reduced
online particle-size analyzer measures on energy savings alone. waste, increased throughput, enhanced
product quality in real time, providing product quality, reduced manual input
vital information that is used by the Conclusions and lower energy consumption. Auto-
model in combination with an array of The automation of process analysis mating control realizes the full poten-
other process measurements. and its closer integration with plant tial of realtime analysis, extracting
maximum return from an investment
in process analytical technology.  ■
Edited by Rebekkah Marshall

Harness the power of References


1. Ferrari, A. and Pugh, D., Marble fillers made

positive press. to measure, Industrial Minerals, Aug 2008.


2. http://opcfoundation.org/Default.
aspx/02_news/02_news_display.
asp?id=740&MID=News

Custom reprints from Chemical Engineering Authors


could be one of the smartest marketing Jeff DeNigris joined Mal-
vern Instruments in 2005
decisions you make. as national sales manager,
process systems to focus on
supplying online, realtime
particle size analyzers to the
pharmaceutical, fine chemi-
cal, mineral, toner and cement
markets (Address: ; Phone:
XXX-XXX-XXXX; Email: ).
He graduated in 1989 with
a B.S.M.E. degree from Wid-
ener University, College of Engineering in Penn-
sylvania. He has spent most of his career in the
manufacturing sector with top original equip-
ment manufacturers of capital equipment. Par-
ticular focus was given to the plastics industry in
the sales and marketing of recycling and reclaim
systems, and to the design and implementation
of granulation and separation systems.

Contact The YGS Group Alberto Ferrari is the pro-


duction manager of Ferrari
at 717.399.1900 x100 or Granulati S.A.S., a leading
learn more online at producer of marble chips and
www.theYGSgroup.com/reprints powders, situated in Grezzana
(Verona) Italy. He is a special-
ist in the fields of information
technology and industrial
The YGS Group is the authorized automation, working for five
provider of custom reprints from years as R&D manager in
Chemical Engineering. Dellas S.p.A., a producer of
diamond tools, before return-
ing to the family company in 1995. He continues
to work to improve the efficiency of milling and
sieving plants for calcium carbonate minerals.

44 Chemical Engineering www.che.com February 2012


Chem Eng_Third Square.indd 1 3/10/09 3:59:01 PM
Facts At Your Fingertips
Process Hazards Analysis Methods
Department Editor: Scott Jenkins

D
ifferent methodologies are Table: Different PHA methods and Approaches
available for conducting the Method Description
structured reviews known as
Consequence analysis This method quantitatively assesses the consequences of hazardous material
process hazards analyses (PHAs) for releases. Release rates are calculated for the worst case and also for alterna-
new processes. PHAs are often con- tive scenarios. Toxicological endpoints are defined, and possible release dura-
ducted or moderated by specialists, tion is determined
with participation by the design team,
representatives of the facility owner, Hazard identification HAZID is a preliminary study that is performed in early project stages when
and experienced process operators. analysis (HAZID) potentially hazardous materials, general process information, initial flow diagram
and plant location are known. HAZID is also generally used later on to perform
Each different PHA method is bet- other hazard studies and to design the preliminary piping and instrumentation
ter-suited to a specific purpose and diagrams (P&IDs)
should be applied at different stag-
es of the project development. The What-if method The what-if method is a brainstorming technique that uses questions starting
table includes brief descriptions of with “What if...,” such as “What if the pump stops running” or “What if the op-
some of the most widely used PHA erator opens or closes a certain valve?” For best results, these analyses should
be held by experienced staff to be able to foresee possible failures and identify
methods in the chemical process in- design alternatives to avoid them
dustries (CPI).
Hazard and operability The HAZOP technique has been a standard since the 1960s in the chemical,
When to use different methods study (HAZOP) petroleum refining and oil-and-gas industries. It is based on the assumption
Different types of PHA studies have that there will be no hazard if the plant is operated within the design param-
varying impact, depending on the de- eters, and analyzes deviations of the design variables that might lead to unde-
sirable consequences for people, equipment, environment, plant operations or
sign phase in which they are applied. company image.
For example, if a consequence analy- If a deviation is plausible, its consequences and probability of occurrence
sis is not performed in a conceptual are then studied by the HAZOP team. Usually an external company is hired to
interact with the operator company and the engineering company to perform
or pre-FEED (front-end engineering this study. There are at least two methods using matrices to evaluate the risk
and design) phase, important plot- (R): one evaluates consequence level (C) times frequency (F) of occurrence;
plan considerations can be missed, and the other incorporates exposition (E) as a time value and probability (P)
such as the need to own more land ranging from practically impossible to almost sure to happen. In this method,
the risk is found by the following equation: R = E × P × C
to avoid effects on public spaces; or
the fact that the location might have a
Layer-of-protection The LOPA method analyzes the probability of failure of independent protection
different elevation with respect to sea analysis (LOPA). layers (IPLs) in the event of a scenario previously studied in a quantitative hazard
level than surrounding public places evaluation like a HAZOP. LOPA is used when a plant uses instrumentation inde-
impacted by a flare plume. pendent from operation, safety instrumented systems (SIS) to assure a certain
Some other studies, like HAZOP, safety integrity level (SIL). The study uses a fault tree to study the probability of
failure on demand (PFD) and assigns a required SIL to a specific instrumentation
cannot be developed without a con- node. For example, in petroleum refineries, most companies will maintain a SIL
trol philosophy or piping and instru- equal to or less than 2 (average probability of failure on demand ≥10−3 to <10−2),
mentation diagrams (P&IDs), and are and a nuclear plant will tolerate a SIL 4 (average probability of failure on demand
≥10−5 to <10−4)
performed at the end of the FEED
stage or at the end of the detailed
engineering phase (or for improved Fault-tree analysis Fault-tree analysis is a deductive technique that uses Boolean logic symbols (that
is, AND or OR gates) to break down the causes of a top event into basic equip-
results, at the end of both) to define ment failures or human errors. The immediate causes of the top event are called
and validate the location of pressure “fault causes.” The resulting fault-tree model displays the logical relationship
safety valves (PSVs) as well as to between the basic events and the selected top event
validate other process controls and
instrument safety requirements. Quantitative risk assess- QRA is the systematic development of numerical estimates of the expected
QRA or LOPA evaluations (or both) ment (QRA) frequency and consequence of potential accidents based on engineering
evaluation and mathematical techniques. The numerical estimates can vary
are undertaken after the HAZOP study from simple values of probability or frequency of an event occurring based on
to validate siting and define safety in- relevant historical data of the industry or other available data, to very detailed
tegrity levels (SIL), to finally meet the frequency modeling techniques. The events studied are the release of a haz-
ardous or toxic material, explosions or boiling-liquid expanded-vapor explosion
level required by the plant. n (BLEVE). The results of this study are usually shown on top of the plot plan

Editor’s note: The definitions in the table, and associated


comments, were adapted from the following article: Giar- Failure mode and effects This method evaluates the ways in which equipment fails and the system’s re-
dinella, S., Baumeister, A. and Marchetti, M. Engineering for analysis (FMEA) sponse to the failure. The focus of the FMEA is on single equipment failures and
Plant Safety. Chem. Eng., August 2015, pp. 50–58. An ad- system failures
ditional reference is the following article: Wong, A., Guillard,
P. and Hyatt, N. Getting the Most Out of HAZOP Analysis,
Chem. Eng., August 1, 2004, pp. 55–58.
34 Chemical Engineering www.chemengonline.com january 2016
Feature Report

Aging Relief Systems —


Are they Working Properly?
Common problems, Symptoms
Inadvertent Relief lines
relief-valve and/or headers
cures and tips blocking vibrations

to make sure your Equipment


and/or piping
Relief valve
chattering
Other
symptoms
failure
pressure relief valves
Problems
operate properly Unprotected equipment Relief system
when needed or piping during
overpressure scenarios
components malfunction

Level 1
Sebastiano Giardinella causes
Inelectra S.A.C.A.
Improperly

R
Lack of a relief Undersized relief installed relief Miscellaneous
elief systems are the last line of device device device deficiencies
defense for chemical process fa-
cilities. Verifying their capabil-
Level 2
ity to safeguard equipment in- causes
tegrity becomes important as process
plants age, increase their capacities to Undersized Inappropriate
adjust to new market requirements, relief lines or relief line routing
equipment
undergo revamps or face new environ-
mental regulations.
Incorrect relief Block valves
In the past, approximately 30% of device set without involuntary
the chemical process industries’ (CPI) pressure closure prevention
losses could be attributed, at least in
part, to deficient relief systems [1]. Level 3
Furthermore, in an audit performed causes

by an independent firm at more than


250 operating units in the U.S., it was Overpressure Higher relief loads
scenario unforeseen than forseen
determined that more than 40% of the during design during design
pieces of equipment had at least one re-
lief-system-related deficiency [2]. These
indicators underscore the importance Figure 1. A problem tree for relief system shows causes, problems and symptoms
of checking the plant relief systems.
This article presents the most com- ply with “recognized and generally ac- ficiencies can be classified into one of
mon types of relief system problems cepted good engineering practices” [3] three types [2]:
with their possible solutions and offers in relief systems design. The recognized 1. No relief device present on equip-
basic guidelines to maintain problem- and generally accepted good engineer- ment with one or more potential
free relief systems. ing practices are criteria endorsed by overpressure scenarios
widely acknowledged institutes or orga- 2. Undersized relief device present on
Common problems in nizations, such as the Design Institute equipment with one or more poten-
existing relief systems for Emergency Relief Systems (DIERS) tial overpressure scenarios
Problems and their causes or the American Petroleum Institute 3. Improperly installed pressure relief
Relief system problems or deficiencies (API). For instance, in the petroleum device
can be identified, with respect to the refining industry, the accepted good en- The first type of deficiency refers to
U.S. Occupational Safety and Health gineering practices are collected in API the lack of any relief device on a piece
Admin. (OSHA) regulation 29 CFR Standards 520 and 521. of equipment that is subject to poten-
1910.119, as items that do not com- The most common relief system de- tial overpressure. The second type
38 Chemical Engineering www.che.com July 2010
Table 1. Relief-system problem identification during overpressure-scenario modeling
Relief System Deficiency Identified When:
Undersized Insufficient relief- Calculated relief-device area > installed relief device area
relief device device area
Improperly Excessive relief-valve Friction pressure drop in pressure-relief-device inlet line > allowable friction
installed inlet-line pressure drop losses (typically 3% of the set pressure)
relief device
Excessive relief valve Relief valve backpressure > allowable backpressure (typically 10% for conven-
backpressure tional valves, or 50% for balanced bellows valves considering backpressure
capacity-reduction factor)
Incorrect relief-valve Pressure in protected vessel or line > Maximum allowable accumulated pres-
set pressure sure (typically 10, 16 or 21% of the MAWP for pressurized vessels with single
relief valves for non-fire scenario, multiple relief valves for non-fire scenario, and
relief valves for fire-scenario), AND pressure at PSV inlet < PSV set pressure
Miscellaneous Excessive line velocity Line Mach number > allowable Mach number (typically 0.7)
Insufficient knockout Effectively separated droplet size at maximum relief load > allowable droplet
drum liquid separation size (typically 300–600 µm)
Excessive flare radia- Calculated radiation level at a specific point > allowable radiation level (typi-
tion cally 1,500 Btu/h-ft2 where presence of personnel with adequate clothing is
expected for 2–3 min during emergency operations, or 500 Btu/h-ft2, where
continuous presence of personnel is expected, both including sun radiation)

refers to an installed relief device data, “the employer shall document 3. Overpressure scenario identifi-
with insufficient capacity to handle that equipment complies with rec- cation. In this step, the P&IDs are
the required relief load. The third ognized and generally accepted good examined in order to identify credible
type encompasses relief devices with engineering practices” [3] overpressure scenarios for each piece
incorrect set pressures, possibility 2. Process hazards analysis, which of equipment.
of involuntary blocking or hydraulic may include: What-if, hazard and op- 4. Overpressure scenario model-
problems. In addition to these prob- erability (HAZOP) study, failure mode ing. The fourth step is to model each
lems, other less frequent ones can be and effects analysis (FMEA), fault-tree credible overpressure scenario. Each
cataloged as miscellaneous deficien- analysis or equivalent methodologies. model is developed in accordance with
cies. A relief-system problem tree is In order to document that the plant the chosen reference standard (for
shown in Figure 1. equipment complies with recognized instance, API 520 and 521). The fol-
In a previous statistical analysis of and generally accepted good engineer- lowing calculations are typically per-
272 process units in the U.S., it was ing practices, the plant management formed during this step:
observed that [2]: must validate that the facilities are • Required relief load for each over-
• 15.1% of the facilities lacked relief de- protected against potential overpres- pressure scenario
vices on equipment with one or more sure scenarios, in accordance with • Required relief-device orifice area
potential overpressure scenarios accepted codes and standards, such for each overpressure scenario
• 8.6% of the relief devices were un- as API standards 520 and 521. An ef- • Relief line’s hydraulics
dersized fective relief-system-validation study • Knockout drum (KOD) liquid-sepa-
• 22% of the relief devices were im- comprises the following steps: ration verification
properly installed 1. Plant documents and draw- • Flare or vent radiation, dispersion
ings gathering. The first step in- and noise level calculations
Identifying potential problems volves obtaining and classifying the The overpressure scenario modeling
There are work methodologies that existing plant documents and draw- can be done in different ways, be it
allow identifying potential problems ings: process flow diagrams (PFDs), by hand calculations, spreadsheets
in relief systems. OSHA regulation mass and energy balances, product or by the use of steady-state or dy-
29 CFR 1910.119 is based on safety compositions, equipment and instru- namic relief-system simulation soft-
audits that use techniques such as ment datasheets, P&IDs, relief device ware. The results of the models are
process hazard analyses performed at datasheets, relief loads summaries, re- analyzed to identify potential prob-
regular intervals. The work methodol- lief line isometrics, one-line diagrams, lems. Table 1 summarizes the pos-
ogy established by this regulation to unit plot plan, and so on. sible relief system problems and the
identify safety hazards comprises two 2. Plant survey. The second step con- ways to identify them on the calcula-
basic steps [3]: sists of inspecting the installed relief tion results.
1. Process safety data gathering, devices to verify that they are free of
which includes the following: mechanical problems, to update and Available solutions
• Process chemical safety data fill-out missing data in the plant docu- There are various solutions for each
• Process technology data ments and to verify consistency be- type of relief system problem. The
• Process equipment data [materials tween the documents and drawings available solutions can be classified
of construction (MOCs), piping and and the actual as-built plant. During as: (a) modification of existing relief
instrumentation diagrams (P&IDs), plant surveys, other typical indica- system components, (b) replacement
design standards and codes, design tions of relief system problems are the of existing relief system components,
and basis of design of the relief sys- presence of pockets, leaks or freezing (c) installation of new relief system
tems, among others]. As part of these in relief lines and headers. components, or (d) increasing the reli-
Chemical Engineering www.che.com July 2010 39
Table 2. Conditions that increase the probability
and impact of relief system failure
Conditions that increase the probability of Conditions that increase the impact of
Feature Report relief system failure relief system failure
The plant has over 20 years of service The plant handles toxic, hazardous or
flammable fluids
ability of the emergency shut- The plant currently handles different prod- The plant handles gases
ucts to those it was originally designed for
down systems.
The modification of exist- The plant operates at a different load or The plant operates at high pressures
at different conditions to those it was origi-
ing relief-system components nally designed for
includes changes made to in- There have been contingencies that have The plant operates at high temperatures
stalled components, without required the replacement of equipment or
requiring their replacement. lines in the past
Some examples of this type of Rotating equipment (pumps, compressors) The plant has furnaces, or equipment
solution include the following: has been modified (for instance, new im- that adds considerable heat input to the
1. Recalibrating the pressure pellers) or replaced fluids
relief valve by readjusting The relief valves have not been checked or The plant has high-volume equipment
the set pressure (solution validated in the last ten years (such as columns, furnaces)
to incorrect set pressure) or Modifications have been made to existing The plant has exothermic reactors, or
relief valve lines (that is, they have been chemicals that could react exothermi-
the blowdown (solution to rerouted) cally in storage
inlet-line friction losses be- A complete and up-to-date relief valve in- The plant has large relief valves (8T10),
tween 3% and 6% of the set ventory is not available or the relief header has a large diameter
pressure) The relief load summary has not been up- The plant has a high number of opera-
2. Adding locks to relief lines’ dated in the last ten years tions personnel
block valves (to prevent in- A relief header backpressure profile is not The plant is located near populated
voluntary valve closure) available, or the existing model has not areas
The replacement of existing been updated in the last ten years
relief system components in-
volves substituting inadequate relief in which redundant instrumentation Deficiency No. 2
system elements for newer, appropri- and emergency shutdown valves are This type of deficiency involves under-
ate ones. Some examples of this solu- installed in order to cutoff the over- sized relief devices that are present on
tion are the following: pressure sources during a contingency. equipment with one or more potential
1. Replacing the installed pressure The main advantage of this type of overpressure scenarios.
relief valve, either for one with a solution is that it can significantly re- Case 2: Insufficient orifice area
larger orifice area (solution to un- duce the required relief loads, hence after changes in the stream com-
dersized relief device) or for one of a posing an economical alternative to position. In a petroleum refinery, a
different type (solution to excessive the installation of new relief headers, desalter that was originally designed
backpressure) knockout drums or flares. to process heavy crude oil was pro-
2. Replacing relief line sections to tected against a potential blocked
solve hydraulic problems, such as: examples of problems outlet by a relief valve on the crude
excessive relief-valve inlet-line fric- in aging systems outlet. When the refinery started
tion losses, excessive backpressure, What follows are examples of some processing lighter crude, simulations
excessive fluid velocity, pockets, typical relief-system problems that showed partial vaporization in the
among others can be found in aging process facilities relief valve. The vapor reduced the
The installation of new relief system and the recommended remedy. PSV capacity until it was insufficient
components involves the addition of to handle the required relief load. In
relief system elements that were not Deficiency No. 1 this case, the recommendation was to
included in the original design, such The first type of deficiency is when no replace the original PSV for one with
as the following: relief device is present on equipment a larger orifice and appropriate relief
1. New pressure relief valves, either with one or more potential overpres- lines.
on equipment lacking overpressure sure scenarios.
protection, or as supplementary Case 1. New overpressure scenario Deficiency No. 3
valves on equipment with under- after pump replacement. In a pro- The third type of deficiency involves
sized relief valves cess unit, a centrifugal pump was re- improperly installed pressure relief
2. New headers, knockout drums or placed for another one with a higher devices.
flares, when the revised relief loads head, without considering the down- Case 3: Excessive backpressure due
exceed the existing relief system stream system’s maximum-available to discharge line modifications. An
capacity, or when relief system seg- working pressure (MAWP). Since the existing vacuum-distillation column’s
regation (that is, acid flare/sweet downstream system was designed at PSV outlet lines were rerouted from the
flare, high-pressure/low-pressure the previous pump’s shutoff pressure, atmosphere to an existing flare header
flare) is required the installation of a higher shutoff due to new environmental regulations.
Increasing the reliability of the emer- pressure pump created a new blocked The installed PSVs were a conven-
gency shutdown systems is typically outlet scenario. Therefore, the instal- tional type, so with the new outlet-line
done via implementation of high in- lation of a new pressure safety valve routing, the backpressure exceeded
tegrity protection systems (HIPS), (PSV) was recommended. the allowable limit. A recommendation
40 Chemical Engineering www.che.com July 2010
Incorrect Correct

LO CSO

LO CSO

No measures are taken The block valves on PSV lines are kept
to prevent involuntary open via locks (LO) or car seals (CSO)
PSV blocking

Incorrect Correct

A PSV installed over the mist A PSV installed below the mist
eliminator is ineffective when eliminator is effective even if
the latter gets clogged the latter is clogged

Figure 2. The risk of blocking in a pressure safety valve Figure 3. Non-free-draining lines in installed relief lines, such
(PSV) can sometimes be readily identified on P&IDs as shown in these two constructions, may cause accumulation
of liquids that can hamper relief valve performance

was made to replace the existing PSVs tations, a non-ideal line arrangement only for a fire, additional discharges
for balanced bellows PSVs. is installed, creating pockets on relief were not considered by the designer.
Case 4: Incorrect PSV set pressure lines that may cause liquid accumu- However, the power failure also af-
due to static pressure differential. lation and hamper relief valve perfor- fected these drums. When this case
A liquid-full vessel’s relief valve was mance (Figure 3). was evaluated, the backpressure was
set to the vessel’s MAWP; however, the too high for the installed PSVs, so they
relief valve was installed several feet Deficiency No. 4 had to be replaced by piloted valves.
above the equipment’s top-tangent The fourth category of deficiencies is a
line. The static pressure differential miscellaneous grouping. MAINTAINING PROBLEM-
was such that the pressure inside the Case 8: Problems in an existing FREE RELIEF SYSTEMS
vessel exceeded the maximum-allow- flare network due to additional Some practical guidelines are offered
able accumulated pressure before the discharges. The additional dis- below to help the plant management
PSV would open. The problem was charges of various distillation-column to assess, identify and troubleshoot re-
solved by modifying the existing PSV, relief valves were rerouted to an exist- lief system problems.
recalibrating it to the vessel MAWP ing flare network because of new envi-
minus the static pressure differential. ronmental regulations. The additional Tip No. 1: Assess the risk
Case 5: Incorrect PSV set pressure discharges exceeded the system capac- Some factors tend to increase the prob-
due to higher operating tempera- ity, and the entire flare network and ability and impact of a relief system
ture. The temperature of a stream was emergency shutdown system had to be failure. Table 2 qualitatively shows
increased with the addition of new heat redesigned by selecting the optimum some of them. If several of the condi-
exchangers, and no attention was paid tie-in locations for the discharges, and tions shown on Table 2 apply, then the
to the set pressure of the thermal relief by implementing HIPS in order to re- plant management should consider
valve in the line. By increasing the tem- duce the required relief loads. planning a detailed study, such as a
perature, the pipe MAWP was reduced. Case 9: Sweet and sour flare mix- quantitative risk analysis (QRA), or a
The PSV set pressure was lowered to ing. When revamping a section of a relief-system validation study.
the new MAWP at the new working process unit’s relief headers, some acid
temperature plus a design margin. discharges were temporarily routed Tip No. 2: Maintaining up-to-
Case 6: Risk of blocking the relief to the sweet flare header in order to date relief-valve information
valve. A relief valve can be blocked for maintain operations. Soon afterwards, The plant management should main-
various reasons. Some of the most com- the header backpressure started to in- tain accurate, up-to-date relief-valve
mon include the lack of locked-open crease and scaling was detected upon data for maintenance and future ref-
(LO) or car-seal-open (CSO) indica- inspection. The acid gases could also erence. The following documents are of
tions in the PSV inlet- and outlet-line generate corrosion, as the sweet flare particular interest: (a) relief valve in-
block valves, and installing the PSV header material was inadequate to ventory, (b) relief loads summary and
above the mist eliminator on a sepa- handle them. (c) relief header backpressure profile.
rator. Both deficiencies can be readily Case 10: High- and low-pressure Relief valve inventory. The relief
identified on P&IDs (Figure 2). flare mixing. The discharges of low valve inventory is a list that contains
Case 7: Pockets. Relief lines going pressure PSVs located on drums were basic information and status for each
to closed systems should be self- routed to the closest flare header, which relief valve, which should include the
draining. It is not uncommon during was a high pressure header. Since the following:
construction that, due to space limi- design case for relief of the drums was • Valve tag
Chemical Engineering www.che.com July 2010 41
Feature Report

Table 3. Relief system validation study typical execution phases and deliverables
Phase Deliverable Deliverable description
Survey and Updated relief de- A list containing up-to-date, accurate data for each relief device located in the
data gathering vice inventory plant. The minimum data to be included on the list are as shown in Tip No. 1, and
they should be obtained by combining relief-valve manufacturer documentation
with onsite inspections
Updated P&IDs P&IDs showing the existing installed relief-device information: connection diam-
eters, orifice letter, set pressure, inlet- and outlet-line diameters and block valves
List of pockets A document identifying pockets on relief lines, with the appropriate photographs
Existing relief Updated relief A list containing the required relief loads for each applicable overpressure sce-
system model- loads summary nario of each relief device, the required orifice area and the relieving fluid prop-
ing erties, based on actual process information
Updated relief- A document showing a general arrangement of the relief headers and subhead-
network back- ers, along with updated backpressure profiles for the major plant contingencies
pressure profile
Updated relief de- A document containing the calculations for each relief device under actual op-
vice calculations erating conditions
List of relief sys- A document listing all of the deficiencies found in the existing relief system, cat-
tem deficiencies egorized by type
Relief system Conceptual engi- A document defining the modifications required to solve the relief system defi-
troubleshooting neering ciencies

• Process unit and area for each phase are described in Table be available to provide technical in-
• Location 3. If the plant management has spe- formation and verify the validity of
• Discharge location cific document formats, it should the consulting firm’s calculations. The
• Connection sizes provide them as part of the deliver- typical information that the consult-
• Connection rating able description. ing firm will request in order to com-
• Orifice letter The study may require a number of plete the study includes: relief device
• Manufacturer resources that are not readily avail- inventory, relief loads summary, relief
• Model able in the plant. If the plant man- device datasheets, mass and energy
• Type (conversion, ball, pilot) agement has available resources but balances, PFDs, P&IDs, equipment
• Set pressure lacks specialized software licenses, datasheets and relief line isometrics
• Allowable overpressure then it can assign some of the tasks for each evaluated process unit/area.
• Design case to inner resources, for example, survey The consulting firm may also request
• Installation date and data gathering. Tasks requiring process simulations, if available.
• Last inspection date expertise or software packages above
• Last calculation date the plant’s capabilities, such as com- Tip No. 4: When modeling, go
Relief loads summary. The relief plex distillation column, reactor sys- from simple to complex
loads summary contains all the over- tem or dynamic simulations, should Replacing a relief valve or header
pressure scenarios and relief loads be outsourced. section generates labor, materials, in-
for each relief device at the plant. The A consulting firm should be se- stallation and loss of production costs
data in this document can be used to lected based on its experience in that can only be justified when the
identify the critical overpressure sce- similar projects, technological ca- results of an accurate model identify
narios in the plant. pabilities (specialized software li- the need for it. However, developing
Relief-header backpressure pro- censes) and a reasonable cost esti- an accurate model for every relief de-
file. A backpressure profile of the en- mate. In order for the consulting firm vice in the plant can be impractical
tire relief network is valuable when to deliver an accurate estimate, the and costly, especially if only a small
evaluating the critical contingencies plant management should provide number of relief devices require re-
in the system, as it can be used to the scope definition along with suf- placement at the end.
identify relief valves operating above ficient information to identify each A practical compromise is to verify
their backpressure limits. relief device within the scope of the each system starting from a simple
project, its location and the possible model with conservative assump-
Tip No. 3: Planning and overpressure scenarios. These data tions, and developing a more accu-
executing a relief system study are available in the relief loads sum- rate model for those items that do
The execution of a typical, relief-sys- mary and relief device inventory. not comply with the required param-
tem validation study comprises three One person should be assigned on eters under such assumptions. This
phases: (a) survey and information the plant management side to manage approach minimizes the time and
gathering, (b) existing relief system the project, along with administrative effort dedicated to items, and con-
modeling and (c) relief system trou- personnel, and at least one in charge centrates on those items that could
bleshooting. The typical deliverables of technical issues; the latter should present problems.
42 Chemical Engineering www.che.com July 2010
For instance, for a blocked outlet rally wish to implement the quickest, References
downstream of a centrifugal pump most practical and least costly one. 1. American Institute of Chemical Engineers,
“Emergency Relief System (ERS) Design
and control valve system, the sim- For instance, when a relief valve’s Using DIERs Technology”, New York, 1995.
plest model is to assume a relief load inlet losses are between 3 and 6% of 2. Berwanger, Patrick, others, Pressure-Relief
equal to the pump’s rated capacity. If the set pressure, the valve blowdown Systems: Your Work Isn’t Done Yet, www.hy-
drocarbononline.com, July 7th, 1999.
the relief-valve orifice area is insuffi- can be adjusted instead of replacing 3. Occupational Safety and Health Administra-
cient under the previous assumption, the entire valve inlet line. tion, 29 CFR 1910.119 “Process Safety Man-
agement of Highly Hazardous Chemicals”.
the next step would be to read the
required relief load from the pump Tip No. 6: What to do after
curve with the control valve’s rated validation and troubleshooting Author
discharge coefficient and the valve’s A routine revalidation of the relief Sebastiano Giardinella is
a process engineer at Inelec-
downstream pressure equal to the re- system’s correct operation not only tra S.A.C.A. (Av. Principal con
lief pressure, ignoring piping friction bring that security to the plant man- Av. De La Rotonda. Complejo
Business Park Torre Este
losses. If the orifice area still seems agement over the integrity of its fa- Piso 5, Costa Del Este. Pan-
insufficient, then a rigorous hydraulic cilities, but also to third parties, such amá. Phone: +507-340-4842;
Fax: +507-304-4801; Email:
calculation of the entire circuit should as occupational safety organizations sebastiano.giardinella@
inelectra.com). He has six
be performed to determine the re- and insurance companies. The cost of years’ work experience in
quired relief load. a relief valve study may very well be industrial projects with a
special focus in relief systems design and evalu-
paid with a reduction in the plant in- ation, equipment sizing, process simulation and
Tip No. 5: Evaluate various surance premium. Furthermore, the bids preparation. He has participated in several
relief system evaluation studies, revamps and
solutions to problems image of a company that worries over new designs. Giardinella graduated as Chemical
As was mentioned earlier, there are the safety of its employees and the Engineer, Summa Cum Laude, at Universidad
Simón Bolívar in Venezuela and holds an M.S.
multiple solutions that are possible environment constitutes an impor- degree in project management from Universidad
Latina de Panamá. He has taken part as speaker
for a single relief system problem, and tant intangible benefit. n or coauthor on international conferences and is
the plant management would natu- Edited by Gerald Ondrey affiliated to Colegio de Ingenieros de Venezuela.

SUBSCRIBE SUBSCRIBE SUBSCRIBE SUBSCRIBE SUBSCRIBE SUBSCRIBE SUBSCRIBE SUBSCRIBE SUBSCRIBE


SUBSCRIBE SUBSCRIBE SUBSCRIBE SUBSCRIBE SUBSCRIBE SUBSCRIBE SUBSCRIBE SUBSCRIBE SUBSCRIBE
SUBSCRIBE SUBSCRIBE SUBSCRIBE SUBSCRIBE SUBSCRIBE SUBSCRIBE SUBSCRIBE SUBSCRIBE SUBSCRIBE
SUBSCRIBE SUBSCRIBE SUBSCRIBE SUBSCRIBE SUBSCRIBE SUBSCRIBE SUBSCRIBE SUBSCRIBE SUBSCRIBE
July
2008

SUBSCRIBE SUBSCRIBE SUBSCRIBE SUBSCRIBE


B S C SUBSCRIBE
R I B E SUBSCRIBE
T O SUBSCRIBE SUBSCRIBE SUBSCRIBE
Incorp
oratin

S U
www. So g
Into lids
6
che. Liquid
com s

Focu
s on
Gas
New Detec
Engin tion
eering
Closed Mater
Liquid ials
Dispe
Findin nsing
g the
To Fit Right
the Ap Gloves
Facts plicat
at Yo ion
ur Fin
Vacu gertip
um Pu s:
mps
Flowm
eter
News

Augu
st
2008

Written for engineers, by engineers


www. Sterili
8
che. zation
com PAG
E 34
r

More and more, business in the Chemical Process Industries (CPI) is not
Rever
se
PAG Osmo
E 34 sis
Heat
Transf
er
Fluids
Hydro

local, it’s global. To keep up with this rapidly evolving marketplace, you
car
Prope bon
rties

Focu
Filtra
s on
need a magazine that covers it all, not just one country or region, not just
one vertical market, but the whole CPI.
tion
Facts
at Yo
r Finge ur
Lesso rtips:
ns-Le
Syste arned Valve
ms s
Preven
ting
Cakin
g

With editorial offices in Europe, Asia, and North America, CHEMICAL


ENGINEERING is well-positioned to keep abreast of all the latest innovations
in the equipment, technology, materials, and services used by process
plants worldwide. No other publication even comes close.
The #1 choice
of worldwide To subscribe, please call 1-847-564-9290
CPI organizations or visit clientservices@che.com
www.che.com

Chemical Engineering www.che.com July 2010 43


Solids
Environmental
Processing
Manager

Overpressure Protection: Consider


Low Temperature Effects in Design
Understanding the 0

inherent limitations of
-10
current overpressure-
k=1.1
Temperature drop, °F

protection analyses is -20

key to developing a more -30


robust heuristic k=1.2

-40
Aubry Shackelford
Inglenook Engineering
-50 k=1.3
Brian Pack
BP North America
-60

I
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
n designing and sizing relief- Mach
device and effluent-handling
systems, one commonly over- Figure 1. Temperature drop relative to stagnation as a result of flowing
looked aspect of the perfor-
mance is examining the potential for Current-practice limitations stream piping system using an efflu-
low temperatures that can cause the Common practices for pressure re- ent-handling hydraulic model.
components of the system to reach lief and effluent handling are found Third, the temperature estimates are
temperatures below their respective, in numerous references [1–5]. The typically made for isothermal pressure-
minimum-design metal tempera- processes for estimating a discharge drop equations and do not account for
tures (MDMT), which may result in temperature and performing the out- effects like retrograde condensation.
brittle fracture with subsequent loss let pressure-drop calculations in the Fourth, some simplifications of the
of containment. This article points pressure-relief-device discharge pip- calculations that are used for the pur-
out limitations of the typical over- ing are limited in their ability to ac- poses of estimating the outlet pres-
pressure-protection-analysis philos- curately predict flowing temperatures sure drop do not represent flashing
ophy, discusses common sources of for many situations. effects (for example, highly subcooled
low temperatures for further inves- First, the discharge calculations flashing liquids are often choked at
tigation, and addresses possible de- are quite often only performed for the the bubblepoint; therefore, the sizing
sign remedies for MDMT concerns. controlling contingency for which the of the valve may assume the backpres-
The primary objectives of a process pressure relief device was sized, which sure is at the bubblepoint).
engineering evaluation of an efflu- does not necessarily represent the Finally, the temperature estimates
ent handling system (such as a flare most likely cause of overpressure or tend to be based on either relieving
system) include ensuring that opera- the cause resulting in the lowest dis- temperatures or isenthalpic flashes
tion of the pressure relief devices dis- charge temperatures. from relief conditions, which do not
charging into the collection system Second, the outlet pressure-drop account for kinetic energy effects.
(flare headers, for example) is not ad- calculations for individual pressure These effects can be substantial if the
versely affected; and that the effluent relief valves consider the outlet dis- developed velocity in the outlet piping
handling equipment are properly de- charge piping and potentially exclude is high and can be compounded when
signed to perform safely. The results of the remaining, downstream piping there are multiple relief devices dis-
an overpressure-protection design are system. This practice can result in a charging simultaneously into a collec-
the primary input for this engineering temperature discontinuity between tion system, or when large diameter
evaluation; however, there are several the calculated discharge temperature differences exist between the tail-pipe
potential gaps in the ability of these for the individual relief device and and the main effluent header.
data to identify situations in which that calculated for the same section Temperature drop. Figure 1 shows
the MDMT may be exceeded. of piping considering the entire down- the temperature drop from the stagna-
Chemical Engineering www.che.com July 2012 45
5

NRU
He
6
N2
1 LP
HP Auto-
refrig-
eration

4
Dehydration

2
Refrig-
Sweetening
erant
Demethanizer

Sweeten-
3 ing

NGL
surge
Flash
tank

Deethanizer

Stabilizer
Propane 7
loading

Amine still Depropanizer


Emergency storage

1. Process upsets in the NRU can cause low flowing temperatures to 5. Valve specifications for appropriate MDMT are
propagate downstream to after the gas/gas exchanger, which is typically common issues
designed based on normal operating temperatures 6. Common discharge lines can cause increased velocities, further
2. Relief of unstabilized condensate, a flashing liquid reducing flowing temperatures. Also depressuring is an
3. Breakthrough of high pressure gas or flashing liquid into the flash tank isentropic process that results in low flowing temperatures
4. Tube rupture in the chillers cause relief of expanded gas that starts cold 7. Potential relief of flashing volatile liquids that can cool
and is cooled further by JT effects significantly based on flashing

Figure 2. Typical schematic of NGL processing facility showing common areas of potential MDMT issues

tion temperature (Tstagnation) caused pressure relief valves to leak, or the thus represent a more likely source of
by the kinetic energy developed during effects of other inputs to the effluent fluid input to the collection system.
adiabatic compressible flow of an ideal handling system (such as pressure In addition, these control valves are
gas as a function of the Mach number control valves, depressuring valves, not necessarily sized to handle the full
for ideal gases having different ideal- pump seals and manual purge valves). flow of an overpressure scenario, re-
gas specific-heat ratios (k) (see Ref. 6, A leaking pressure-relief valve is typi- sulting in flow from both the control
Equation 6–128). For the purposes of cally considered an operational and valve and the pressure relief valve,
illustrating the temperature drop, a mechanical issue, not a cause of over- thereby exacerbating velocity effects.
stagnation temperature of 0°F (460R) pressure that needs to be evaluated for Finally, depressuring is a dynamic
was chosen. the sizing of the pressure relief valve process, releasing fluids of different
It is useful to note that while a stag- or for the effects on the downstream pressures and temperatures as a func-
nation temperature of 0°F seems un- collection system; however, many of us tion of time. Considering the most
likely for many cases, this stagnation in the warm Gulf Coast region of the likely behavior of a depressuring sys-
temperature is established after the U.S. recognize an ice-covering as indic- tem to be an isentropic expansion of
fluid has been relieved into the collec- ative of a leaking valve, and the fluids the residing fluid, the inlet fluid tem-
tion system (in other words, after the used in the evaluation of the pressure- peratures can drop significantly as the
isentropic process of flowing through relief-device sizing may not be repre- depressuring progresses.
the pressure-relief-valve nozzle and sentative of the normal process fluid
the subsequent adiabatic process of (for example, the external fire case, Low temperatures
expanding from the nozzle throat to which is a common design basis). While the potential for low flowing
the total backpressure that results in Pressure control valves may also be temperatures falling below the MDMT
Joule-Thompson (JT) cooling, both of called upon to “relieve” fluids, yet are exists in a variety of processing facili-
which can result in significantly lower commonly not accounted for in over- ties, the issue is especially apparent in
stagnation temperatures of the fluid pressure protection studies based on natural-gas processing facilities where
entering into the discharge piping). the desire to not include the positive high pressure, low temperature, low-
Additional limitations. Additional response of control systems in prevent- molecular-weight gases and volatile
gaps in the overpressure protection ing overpressure. In actual situations, liquids are present.
analysis include the common practice the basic process-control systems are Design considerations. Based on
of not considering the potential for expected to function as intended, and recent evaluations of several natural-
46 Chemical Engineering www.che.com July 2012
gas processing facilities with ethane fluid is at elevated temperatures, yet tion of overfilling of vessels), mitiga-
recovery capabilities, the authors during process upsets may experience tion of relieving conditions causing
have identified several common areas significantly lower temperatures. the low temperature excursion via
of concern that may provide a start- Figure 2 provides an overview of safety instrumented systems (SIS),
ing point for other gas processors’ these common areas of concern related performing mechanical stress analy-
investigations into this aspect of col- to low flowing temperatures. NGL ses to establish a better estimate of
lection system design, as well as for and propane processing-and-storage the MDMT per ASME B31.3 (with re-
process piping. These areas include equipment are examples of commonly placement of components not covered
the following: multiple inputs (such overlooked systems that can achieve by stress analysis as needed), add-
as pressure relief devices or control low flowing-discharge temperatures. ing supplemental fluid (such as gas
valves) discharging into subheaders These equipment usually have pres- or methanol) to raise the stagnation
having diameters close in size to the sure relief devices that are sized based temperature, rerouting the discharge
individual discharge piping diameter; on an external fire case, yet also have to a different location (such as to the
flashing liquid relief (unstablized con- the potential for relieving the liquid atmosphere), or conducting Charpy
densate, natural gas liquids [NGL] or either due to blocked discharges, leak- testing on the piping in question to
liquid propane); internal-boundary- ing relief valves or depressuring. establish the actual MDMT.
failure cases (tube rupture, for ex- Alternative solutions. While the For potentially leaking pressure-
ample) in gas chillers; cryogenic drain design issues related to low flowing relief valves, the options also include
operations (such as draining expander temperatures can be dealt with by recognizing the additional conse-
casing for maintenance); pressure- specifying appropriate metallurgy, quences in a risk-based inspection
relief-device MDMT specifications not there are other alternatives for con- protocol, installing rupture disks, or
commensurate with discharge piping sideration. These alternatives can adding skin thermocouples and low
MDMT; and pressure relief devices or include identifying ways to elimi- temperature alarms on the discharge
vents on the outlet of cryogenic cold- nate the cause of overpressure in piping to notify personnel of leakage
box sections where the normal process the first place (for example, preven- before the MDMT is crossed.

BRAND PRODUCTS

MPLIAN
CO T
S
ARM

INT

ITAR Circle 3 on p. 56 or go to adlinks.che.com/40269-03


ERNAT
C IN
FI

ON
I

AL TRAF
Environmental Manager

Final analysis oping a more robust overpressure pro- Authors


In summary, established overpressure- tection analysis heuristic, which more Aubry Shackelford is a
principal engineer for Ingle-
protection-analysis philosophies are fully recognizes the effects of low tem- nook Engineering, Inc., which
not well suited to identify possible ma- perature flashing on material design. provides process engineer-
ing consulting with a specific
terial concerns as a result of process It is the experience of the authors focus on process safety man-
fluid flashing and depressuring. Relief- that modification of the typical over- agement (15306 Amesbury
Lane, Sugar Land, TX 77478;
device and effluent-handling sizing pressure-protection-analysis philoso- Email: aubry@inglenookeng.
conventions and simplified calculation phy to identify and propose alterna- com; Phone: 713-805-8277).
He holds a B.S.Ch.E. from
methodologies limit the ability of the tive solutions for conditions resulting Northeastern University and
is a professional engineer licensed in the state of
designer to recognize potential MDMT in excursions beyond MDMT is pru- Texas and the province of Alberta (Canada).
concerns. Understanding the inherent dent in promotion of enhanced facility Brian Pack is the area engi-
limitations of current overpressure-pro- process-safety management. ■ neering-support team leader
for the Mid-Continent opera-
tection-analysis practice is key to devel- Edited by Dorothy Lozowski tions area in North America
gas region for BP America
Production Co. (501 West-
lake Park Blvd., Houston, TX
77079; Email: brian.pack@
References bp.com; Phone: 281-366-1604).
He holds a B.S.Ch.E. from the
1. API Standard 520, “Sizing, Selection, and 4. Coats, and others, “Guidelines for Pressure University of Oklahoma and
Installation of Pressure-Relieving Devices Relieving Systems and Effluent Handling”, is a professional engineer li-
in Refineries, Part I — Sizing and Selection”, 1st Ed., Center for Chemical Process Safety censed in the states of Texas and Oklahoma.
8th Ed., December 2008. of the American Institute of Chemical Engi-
2. API Standard 520, “Sizing, Selection, and neers, 1998.
Installation of Pressure-Relieving Devices in 5. Gas Processors Suppliers Association, “Engi-
Refineries, Part II — Installation”, 5th Ed., neering Data Book”, 12th Ed., 2004. Note
August 2003. 6. Perry, R.H., D.W. Green, and J.O. Maloney, The views in this paper are entirely
3. ISO 23251:2006 (API Standard 521), “Petro- editors, “Perry’s Chemical Engineering the authors and do not necessarily re-
leum, Petrochemical and Natural Gas Indus- Handbook”, McGraw-Hill, 7th Ed., 1997.
tries — Pressure-relieving and Depressuring 7. ASME B31.3-2008, “Process Piping”, Decem-
flect the views of BP America Produc-
Systems”, 1st Ed., August 2006. ber 31, 2008. tion Co. or its affiliates.

rEcEivE full accEss


to all of chemical Engineering’s
facts at Your fingertips art

Each information packEd pdf articlE includes


❯❯
icles in one convenient location
.

graphs, charts, tables, equations and columns on the full


chemical engineering processes you deal with on a daily basis.
This is the tool you will come to rely on, referring back to the
information again and again with just the click of a mouse.
Facts at Your Fingertips Topics Include:
 Conservation Economics:  Creating Installed Gain Graphs
Carbon Pricing Impacts  Aboveground and
 Distillation Tray Design Underground Storage Tanks
 Burner Operating  Chemical Resistance of
Characteristics Thermoplastics
 Measurement Guide for  Heat Transfer: System
Replacement Seals Design II
 Steam Tracer Lines and Traps  Adsorption
 Positive Displacement Pumps  Flowmeter Selection
 Low-Pressure Measurement  Specialty Metals
for Control Valves  Plus much, much more… 17872

receive full
Receive full access
access today
todaybyby
visiting www.omeda.com/cbm/facts
visiting http://store.che.com/product/

48 Chemical Engineering www.che.com July 2012


Environmental Manager

Things You Need to Know Before Using an


Explosion-Protection Technique
Understanding the different classification methods is necessary to better select the
explosion-protection techniques that will be used

Class I, Division 2 Class I, Division 1


(from 5 ft to 10 ft radius) (5 ft around vent) Zone 1 Vent

Zone 2
Vent
Outdoors Zone 0
Gasoline
storage
10 ft (3m) tank 10 ft (3m) Gasoline
without storage
floating tank
roof without
Circular dike floating
around tank roof

Class 1, division 2

FIGURE 1. Shown here is a typical example of a Class I hazardous area uti- FIGURE 2. The example hazardous area shown in Figure 1 is here classified ac-
lizing division methods of area classification cording to the zones

Robert Schosker stances, where there exists the principles for working or gathering
Pepperl+Fuchs risk of explosion or fire that can be information out of a hazardous area

E
caused by an electrical spark or hot remain the same. It’s just that today
xplosion protection is essen- surface, requires specifically defined we have multiple options. In order to
tial for many companies, and instrumentation located in a hazard- exercise those options, we must first
those companies have deci- ous location. It also requires that determine if the danger of an explo-
sion makers. But before any the interfacing signals coming from sion exists and how severe it may be.
decisions can be made, there are a hazardous location are unable to
some important factors one must create the necessary conditions to What is a hazardous area?
consider. These factors include what ignite and propagate an explosion. Hazardous areas are most frequently
is most efficient and economical, as This risk of explosion or fire has found in places where there is a pos-
well as knowing the basics of explo- been the limiting factor when using sibility of an emission of flammable
sion protection; so the decision mak- electrical instrumentation because gas or dust. A hazardous area can
ers are headed in the right direction. energy levels were such that the occur in normal operation, in the
We will highlight many of the different energy limitation to the hazardous event of a fault, or due to wear and
“things to know,” but first, let’s step location was difficult, if not impossi- tear of seals or other components.
back in time and take a look at the ble, to obtain. For this reason, those Now the risk of an ignition of an
background of explosion protection. parts of the process that were con- air/gas mixture in this hazardous
sidered risky were controlled with area depends on the probability of
Backdrop pneumatic instrumentation. the simultaneous presence of the
After World War II, the increased following two conditions:
use of petroleum and its derivatives Moving forward • Formation of flammable or explo-
brought the construction of a great Now let’s move forward 70 years, sive vapors, liquids or gases, or
number of plants for extraction, refin- where almost everything you can combustible dusts or fibers with
ing and transformation of the chemi- think of can be found at the touch of a atmosphere or accumulation of
cal substances needed for techno- finger. From pneumatics to quad core explosive or flammable material
logical and industrial development. processors, information gathering • Presence of an energy source (elec-
The treatment of dangerous sub- has definitely changed, but the same trical spark, arc or surface temper-
50 Chemical Engineering www.chemengonline.com july 2015
Table 1. Defining areas for Divisions Table 2. The breakdown of Classes into subgroups
Class Type of Material Class Subgroup Atmospheres
Class I Locations containing flammable Class I Group A Atmospheres containing acetylene
gases, flammable liquid-produced va- Group B Atmospheres containing hydrogen and flammable process gases with
pors, or combustible liquid-produced more than 30 vol.% H2, or gases or vapors posing a similar risk level,
vapors such as butadiene and ethylene oxide
Class II Locations containing combustible Group C Atmospheres such as ether, ethylene or gases or vapors posing a similar
dusts risk
Class III Locations containing fibers and Group D Atmospheres such as acetone, ammonia, benzene, butane, cyclopropane,
flyings ethanol, gasoline, hexane, methanol, methane, natural gas, naphtha, pro-
pane or gases or vapors posing a similar threat
ature) that is capable of igniting the Class II Group E Atmospheres containing combustible metal dusts, including aluminum,
explosive atmosphere present magnesium, and their commercial alloys, or other combustible dusts
Determining hazardous areas in whose particle size, abrasiveness and conductivity present similar haz-
ards in the use of electronic equipment
a plant is normally performed by
experts from various disciplines. It Group F Atmospheres containing combustible carbonaceous dusts, including
carbon black, charcoal, coal or coke dusts that have more than 8% total
may be necessary for chemists, pro- entrapped volatiles, or dusts that have been sensitized by other materials
cess technologists, and mechani- so that they present an explosion hazard
cal engineers to cooperate with an Group G Atmospheres containing combustible dusts not included in Group E or
explosion-protection expert in order Group F, including flour, grain, wood, plastic and chemicals
to evaluate all hazards. The possible
presence of a potentially explosive Table 3. The Division Method
atmosphere as well as its properties Division Class I Class II Class III
and the duration of its occurrence
(gases and vapors) (flammable dust or powder) (flammable fibers or sus-
must be established. pended particles)
Also understanding terms such as In accordance with NEC In accordance with NEC
minimum ignition energy (MIE), upper 500.5 and CEC J18-004 500.6 and CEC 18-008 In accordance with NEC
and lower explosive limit (UEL/LEL), 500.5 and CEC 18-010
flash point, and ignition temperature Division 1 Areas containing dangerous Areas containing dangerous Areas containing dangerous
in the evaluation of your hazardous concentrations of flam- concentrations of flammable concentrations of flammable
mable gases, vapors or mist dusts continuously or oc- fibers or suspended particles
area will also provide a clearer direc- continuously or occasionally casionally under normal continuously or occasion-
tion on how severe a hazardous area under normal operating operating conditions ally under normal operating
might be. conditions conditions
In any situation involving an ex- Division 2 Areas probably not contain- Areas probably not contain- Areas probably not contain-
plosive material, the risk of ignition ing dangerous concentra- ing dangerous concentra- ing dangerous concentra-
must be taken into account. In addi- tions of flammable gases, tions of flammable dusts tions of flammable fibers or
tion to the nominal rating of materi- vapors or mist under normal under normal operating suspended particles under
operating conditions conditions normal operating conditions
als under consideration, parameters
related to the process involved are and utilized worldwide. may exist due to flammable gases,
especially important in the evalua- In rather simple terms, we can dif- vapors, or liquids within a plant or
tion. For example, the risk of explo- ferentiate between the International working environment. These are ex-
sion may be caused by the evapora- Electrotechnical Commission (IEC; plained by defining the type of haz-
tion of a liquid or by the presence of Geneva, Switzerland) (zones) and ardous material present, severity of
liquid sprayed under high pressure. the North American (division) proce- the hazard, and probability of the
It is also important to know which dures. The differences lie in the cat- hazard. It may also depend on the
atmospheric conditions are present egorization of hazardous areas, the likelihood of the hazard, risk of an
normally and abnormally. The range design of apparatus, and the installa- explosion, and the boundaries of the
of concentration between the ex- tion technology of electrical systems. hazardous location.
plosion limits generally increases as The categorization of these areas is This is usually determined by a
the pressure and temperature of the carried out in North America in ac- HAZOP (hazard and operability)
mixture increases. cordance with the National Electrical study and documented on a set
Code (NEC) NFPA 70, article 500. of electrical plot plans on record in
Divisions and zones The European Zone practice is de- every plant.
Once it has been determined that a scribed in IEC/EN 60079-10. For divisions, the type of material
hazardous area exists, it now needs So how does each work? First let’s is given by a class designation, as
to be classified. While the physical start at the basics, and then we’ll shown in Table 1. These can be bro-
principles of explosion protection cover each individually. ken down further into sub-groups,
are the same worldwide and are as shown in Table 2.
not differentiated, there are two dif- Defining the area Once we have determined the haz-
ferent and distinct models to define Hazardous location or area clas- ardous material we are working with,
your hazardous area —divisions and sification methods specify the dan- the probability of an explosion and
zones — both of which are accepted ger of fire or explosion hazards that boundaries must also be taken in to

Chemical Engineering www.chemengonline.com july 2015 51


Hazardous atmosphere Hazardous atmosphere
Hazardous
Q atmosphere

Io L R
Interstice

U
C
Length of
S Uo
junction
P

FIGURE 3. Explosion-proof protection is based on FIGURE 4. In purging or pressurization protec- FIGURE 5. Intrinsic safety is based on the prin-
the explosion-containment concept, whereby the tion, a dangerous air/gas mixtures is not allowd ciple of preventing an effective source of ignition
enclosure is built to resist the excess pressure to penetrate the enclosure containing the electri-
created by an internal explosion cal parts that can generate sparks or dangerous
temperatures
ated, T1 to T6. The classes T2, T3
and T4 are divided into further sub-
consideration. The division method better rationalized by looking at the classes, as indicated in Table 7.
is divided into two areas: Division 1 example shown in Figure 2. In Europe, the apparatus are
and Division 2 (Table 3). These were With a slightly different approach, certified on the basis of design
created in 1947 when the NEC first IEC 600079-0 requires apparatus to and construction characteristics.
recognized that different levels of risk be subdivided into two groups, as From a practical point of view, the
exist in hazardous locations. Figure 1 shown in Table 5. two systems are equivalent, even
shows a typical example of a Class The groups indicate the types if there are minor differences, but
I hazardous area utilizing Division of danger for which the apparatus before you run out and choose the
methods of area classification. has been designed. Group I is in- most convenient method for you, it
In comparison to the division- tended for mines. Group II concerns is important that you consult your
based area classification, which is above-ground industries (electrical local authority having jurisdiction to
prevalent throughout North America, apparatus for hazardous areas with learn what method is allowed or, in
the zone-based architecture prevails potentially explosive gas (dust) at- fact, preferred.
in the rest of the world. mosphere except firedamp hazard- The initial steps to determine
Zones are similar in nature to divi- ous mining areas) and is subdivided whether a hazardous area exists and
sions where type of hazardous ma- into II G (gases) and II D (dusts). classify that area may seem rudimen-
terial present, severity of the hazard, Similar to divisions, the zones tary to some, but they are important
and probability of the hazard and offer a sub material classification as they now open up the multiple
boundaries must be determined. as well. Table 6 shows how this methods of protection, which may
Zones are in accordance with IEC/ approach compares to the North or may not be allowed, depending
EN 60079-10, which states that any American equivalent. on whether you classified your area
area in which there is a probability of Finally, when classifying your haz- by divisions or zones.
a flammable gas or dispersed dust ardous area, whether it be division
must be classified into one of the or zones, you must also classify the Protection methods
areas shown in Table 4. maximum surface temperature that There are three basic methods of
Similar to the division method of can go in to the hazardous area. The protection — explosion containment,
area classification, zones can be maximum surface temperature must segregation and prevention.
be below the minimum ignition tem- Explosion containment. This is the
Table 4. Defining areas by Zones
perature of the gas/dust present. only method that allows the explosion
Zone Type of material In North America as in Europe, six to occur, but confines it to a well-de-
Zone 0 An area in which an explosive air/gas temperature classes are differenti- fined area, thus avoiding the propaga-
mixture is continuously present or tion to the surrounding atmosphere.
present for long periods of time
Table 5. Apparatus Groups per IEC Flameproof and explosion-proof en-
Zone 1 An area in which an explosive air/gas 600079-0 closures are based on this method.
mixture is likely to occur in normal
operation Group Apparatus Segregation. This method attempts
Group I Apparatus to be used in mines where to physically separate or isolate the
Zone 2 An area in which an explosive air/gas
mixture is unlikely to occur; but if it the danger is represented by methane electrical parts or hot surfaces from
does, only for short periods of time gas and coal dust the explosive mixture. This method
Zone 20 An area in which a combustible dust Group II Apparatus to be used in surface includes various techniques, such
cloud is part of the air permanently, industries where the danger is rep- as pressurization, encapsulation,
over long periods of time or frequently resented by gas and vapor that has and so on.
been subdivided into three groups:
Zone 21 An area in which a combustible dust A, B and C. These subdivisions are
Prevention. Prevention limits the
cloud in air is likely to occur in normal based on the maximum experimental energy, both electrical and thermal,
operation safe gap (MESG) for an explosion- to safe levels under both normal op-
Zone 22 An area in which a combustible dust proof enclosure or the minimum igni- eration and fault conditions. Intrinsic
cloud in air may occur briefly or dur- tion current (MIC) for intrinsically safe safety is the most representative
ing abnormal operation electrical apparatus
technique of this method.
52 Chemical Engineering www.chemengonline.com july 2015
Table 6. Sub material classification for Zones Table 7. Temperature classes
Material Apparatus clasification Apparatus classification Ignition energy Tmax, °C Tmax, °F T Class in N.A.*
450 842 T1
Europe (*IEC) North America
300 572 T2
Methane Group I (mining) Class I, Group D
280 536 T2A
Acetylene Group IIC Class I, Group A > 20 µJ
260 500 T2B
Hydrogen Group IIC Class I, Group B > 20 µJ
230 446 T2C
Ethylene Group IIB Class I, Group C > 60 µJ
215 419 T2D
Propane Group IIA Class I, Group D > 180 µJ
200 392 T3
Conductive dust (metal) Group IIIC* Class II, Group E
180 356 T3A
Non-conductive dust Group IIIB* Class II, Group F
(carbon) 165 329 T3B
Cereal/flour Group IIIB* Class II, Group G 160 320 T3C
Fibers/suspended Group IIIA* Class III 135 275 T4
particles 120 248 T4A
100 212 T5
My application requirements fication, application, and cost — in
85 185 T6
Now the questions really start racing helping users find the ideal solution
in: Which should I use? Which one to match their needs. *N.A. = North America
offers the best protection? What if In addition to considering the nor-
all of my equipment is not low pow- mal functioning of the apparatus, area classification, there are many
ered? My plant is already using a eventual malfunctioning of the appa- other options when an area is classi-
technique; can I use another pro- ratus due to faulty components must fied using zones, but for now we will
tection method? Can they co-exist? be a consideration. And finally, all concentrate on the above as they
Who makes that decision? Why those conditions that can acciden- are most commonly used.
should I use one method over the tally occur, such as a short circuit, XP. The explosion-proof protection
other? Can I use two methods at open circuit, grounding and errone- method is the only one based on the
the same time? So many questions, ous wiring of the connecting cables, explosion-containment concept. In
all of which are very important, and must be evaluated. The choice of a this case, the energy source is permit-
with a little understanding of your specific protection method depends ted to come in contact with the dan-
process, they will guide you to best on the degree of safety needed for gerous air/gas mixture. Consequently,
method(s) to use. the type of hazardous location con- the explosion is allowed to take place,
Hazardous-area protection method sidered in such a way as to have the but it must remain confined in an en-
selection depends on three impor- lowest probable degree of an even- closure built to resist the excess pres-
tant factors: (1) area classification, (2) tual simultaneous presence of an sure created by an internal explosion,
the application and (3) the cost of the adequate energy source and a dan- thus impeding the propagation to the
protection method solution. gerous concentration level of an air/ surrounding atmosphere.
Area. Area classification depends on gas mixture. The theory supporting this method
the type of hazardous substances None of the protection methods is that the resultant gas jet coming
used, operating temperature, and can provide absolute certainty of from the enclosure is cooled rap-
explosion risk due to how often the preventing an explosion. Statistically, idly through the enclosure’s heat
dangerous substance is present in the probabilities are so low that not conduction and the expansion and
the atmosphere and the boundary even one incident of an explosion dilution of the hot gas in the colder
of the substance from various parts has been verified when a standard- external atmosphere. This is only
of the process. Area classification ized protection method has been possible if the enclosure openings
is determined by either the division properly installed and maintained. or interstices have sufficiently small
method or zone method. The first precaution is to avoid dimensions (Figure 3).
Application. Application character- placing electrical apparatus in haz- In North America, a flameproof en-
istics also affect which protection ardous locations. When designing closure (in accordance with IEC) is, as
method is used. For example, some a plant or factory, this factor needs a rule, equated with the "flameproof"
methods are more appropriate for to be considered. Only when there is designation. In both considerations,
large equipment protection, while no alternative should this application the housing must be designed for
others are more appropriate for high- be allowed. a x1.5 explosion overpressure. The
power applications. North American version “Explosion
Cost. Cost is also an important fac- Choosing the best method proof” (XP) must withstand a maxi-
tor for many engineers. For example, After carefully considering the above, mum explosion overpressure of x4.
if their application requires Division we can look at three more popular Furthermore, in North America,
2 protection, they may not want to methods of protection, XP (explo- the installation regulations (NEC 500)
purchase more expensive equipment sion proof/flameproof), purging and specify the use of metal conduit for
rated for Division 1. For that reason, pressurization, and intrinsic safety. the field wiring installation. It is also
it is important to understand the in- Although these are the most com- assumed here that the air-gas mixture
terplay of all three factors — classi- monly used methods in the division can also be present within the con-

Chemical Engineering www.chemengonline.com july 2015 53


duit system. Therefore, the resulting starting the electrical equipment. ergy in circuits in the form of capaci-
explosion pressures must be taken The classification of the electrical tance and inductance. These energy
into consideration. The conduit con- apparatus must be based on the storage elements must be limited
nections must be constructed ac- maximum external surface tempera- based on the voltage and current
cording to specification and sealed ture of the enclosure, or the maxi- levels present in a particular circuit
(that is, lead seals) with appropriate mum surface temperature of the in- or make-break component.
casting compound. The housing is ternal circuits that are protected with In normal operation and in the
not constructed gas-tight. Of course, another protection method and that event of a fault, no sparks or thermal
large openings are not permitted on remain powered even when the pro- effects may occur that could lead to
the enclosure, but small ones are in- tective gas supply is interrupted. the ignition of a potentially explosive
evitable at any junction point. Some The purging or pressurization atmosphere. Intrinsically safe circuits
of these gaps may serve as pressure technique is not dependent upon the may therefore be connected and
relief points. Escaping hot gases are classification of the gas. Rather, the disconnected by experts during op-
cooled to the extent that they cannot enclosure is maintained at a pressure eration (even when live), as they are
ignite the potentially explosive atmo- higher than the dangerous external guaranteed to be safe in the event of
sphere outside the housing. Ignition atmosphere, preventing the flamma- a short circuit or disconnection.
is prevented if the minimum temper- ble mixture from coming in contact Intrinsic safety is the only ignition-
ature and minimum ignition energy with the electrical components and protection class that allows connec-
of the surrounding potentially explo- hot surfaces inside. tors to be opened and intrinsically
sive atmosphere is not reached. For In the U.S., the term “pressuriza- safe apparatus to be removed and
this reason, the maximum opening tion” is limited to Class II applica- replaced by an equivalent device in
allowed for a particular type of joint tions. This is the technique of supply- a hazardous area. Because of the
depends on the nature of the explo- ing an enclosure with clean air or an level of freedom this brings, intrinsic
sive mixture and width of the adjoin- inert gas, with or without continuous safety has become one of the most
ing surfaces (joint length). flow, at sufficient pressure to prevent important methods of protection in
The classification of a flameproof en- the entrance of combustible dusts. the industrial automation industry.
closure is based on the gas group and Internationally, the term “pressuriza-
the maximum surface temperature tion” refers to a purging technique Final remarks
which must be lower than the ignition for Zones 1 and 2. Each method offers its own advan-
temperature of the gas present. The divisional model of the purging tages and disadvantages, and in most
Purging or pressurization. Purg- protection method is based on the cases no one method will be or can
ing or pressurization is a protection reduction of the classification inside be the only method used in a process
method based on the segregation the enclosure to a lower level. The plant. Generally, this mixed system
concept. This method does not allow following three types of protection (X, does not present installation difficulty
the dangerous air/gas mixture to pen- Y, and Z) are identified in relation to if each of the protection methods is
etrate the enclosure containing electri- the hazardous-location classification appropriately used and is in compli-
cal parts that can generate sparks or and the nature of the apparatus. ance with the respective standards.
dangerous temperatures. A protective • Type X: reduces the inside of the No matter how you classify your
gas — air or inert gas — is contained enclosure from Division 1 to a non- plant or which method of protection
inside the enclosure with a pressure hazardous state that requires an you chose, it is always important
slightly greater than the one of the ex- automatic shutdown of the system to remember that the method you
ternal atmosphere (Figure 4). in case of pressure loss choose today may not necessarily
The internal overpressure remains • Type Y: reduces the inside of the be the appropriate choice tomorrow.
constant with or without a continu- enclosure from Division 1 to Divi- Evaluate, choose and protect not
ous flow of the protective gas. The sion 2 only to keep your plant safe, but to
enclosure must have a certain de- • Type Z: reduces the inside of the keep your personnel safer. n
gree of tightness; however, there enclosure from Division 1 to a non- Edited by Gerald Ondrey
are no particular mechanical require- hazardous state, requiring alarm
ments because the pressure sup- signals only Author
ported is not very high. Intrinsic safety. Finally, intrinsic Robert Schosker is the product
To avoid pressure loss, the pro- safety is based on the principle of manager/team lead for intrinsic
safety (IS), remote I/O, HART, sig-
tective gas supply must be able to preventing an effective source of ig- nal conditioners, power supplies
compensate during operation for nition. The electrical energy is kept and surge protection at
enclosure leakage and access by below the minimum ignition energy Pepperl+Fuchs Inc. (1600 Enter-
personnel where allowed (the use of required for each hazardous area prise Parkway, Twinsburg, OH
44087; Phone: 330-425-3555;
two interlocked doors is the classical (Figure 5). Fax: 330-425-4607; email:
solution). Because it is possible for The intrinsic safety level of an elec- rschosker@us.pepperl-fuchs.
the explosive atmosphere to remain trical circuit is achieved by limiting com). Since joining the company in 1995, Schosker has
been focused on technology and product-related sup-
inside the enclosure after the pres- current, voltage, power and tem- port, and is involved in a wide range of activities and
surization system has been turned perature; therefore, intrinsic safety is roles including certifications, sales, and marketing. He
off, it is necessary to expel the re- limited to circuits that have relatively has been the key lead in many IS and HART projects
resulting in the development of new products for intrin-
maining gas by circulating a certain low levels of power. Of critical impor- sic safety and HART infrastructure. Schosker holds a
quantity of protective gas before re- tance are the stored amounts of en- B.S.E.E. from the University of Akron.

54 Chemical Engineering www.chemengonline.com july 2015


Cybersecurity Defense
for Industrial Process-
Control Systems
Security techniques widely used in information technology (IT) require special considerations to
be useful in operational settings. Here are several that should get closer attention

Honeywell
Mike Baldi
Figure 1. Expansion of the
Honeywell Process Industrial Internet of Things
Solutions (IIoT) and cloud storage offers
benefits, but raises security
concerns
In Brief
cyber threats and
consequences

Defense in Depth

Adapting to the
needs of Operational
Technology

Risk-analysis
solutions

Next-generation
Firewalls

Endpoint protection

looking to the
Future

I
ndustrial cybersecurity risks are widely has brought tremendous benefits, along
appreciated. In April, the deputy direc- with more complex security concerns. The
tor of the U.S. National Security Agency, same convergence, however, has allowed
Rich Ledgett, warned that industrial the industrial world to adopt cyberdefense
control systems (ICS) and other critical in- techniques that have been widely used in
frastructure assets remain vulnerable to IT. This article discusses several key cy-
attack (Figure 1). Robust cyberdefense of bersecurity IT tools that can help industrial
industrial facilities remains an ongoing chal- facilities establish a layered cybersecurity
lenge for the chemical process industries system for its operations.
(CPI). The convergence between the world
of information technology (IT) and the world Cyber threats and consequences
of operational technology, in which con- The Stuxnet worm, a computer virus that in-
trol systems for industrial facilities reside, famously affected Iran’s nuclear centrifuges,
36 Chemical Engineering www.chemengonline.com july 2016
Honeywell
and the damage due to a cyberattack of a
German steel mill reported in 2014 are evi-
dence that cyberattacks can have physical,
real-world impacts. But it is not necessary
to prompt an explosion to cause significant
disruption. A cyber attack on Ukraine’s elec-
tric power grid, and subsequent widespread
power failure last December, was evidence
of that.
As NSA’s Ledgett put it, “You don’t need
to cause physical harm to affect critical infra-
structure assets.”
Cybersecurity risks are not easily ad-
dressed, however. One challenge is the
increasing sophistication of attacks. The
German government report on the steel
mill incident, for example, noted that the at-
tackers demonstrated not only expertise in
conventional IT security, “but also detailed
technical knowledge of the industrial control opportunity to store and process the large Figure 2. A layered approach
systems and production processes used in volumes of data resulting from the prolifera- to cybersecurity, with several
types of different cyber-
the plant.” tion of connected devices, such as with the defenses should be the ob-
Moreover, once the tools and knowledge IIoT. Again, however, it introduces new con- jective of industrial control
to enable such attacks are developed, they nection and communication channels that systems
are often quickly commoditized and shared, would-be cyberattackers will try to exploit.
allowing others with fewer technical skills to
use them. Defense in depth
Another challenge, however, is simply the In fact, the security issues related to the IIoT
increasing vulnerabilities introduced by the and Cloud storage result from the longer-
growth of intelligent, connected devices in in- term challenges surrounding the conver-
dustrial control systems. As Chris Hankin, di- gence between the IT and operational tech-
rector of the Institute for Security Science and nology (OT) worlds. Open platforms and the
Technology (ISST) at Imperial College, Lon- proliferation of third-party and open-source
don (www.imperial.ac.uk/security-institute), software in industrial control systems has
remarked recently: “Almost every component long brought the power and efficiencies from
of such systems now has fully functional com- the enterprise side of the business to the
puting capability and most of the connections process side. But along with those benefits,
will now be Ethernet, Wi-Fi or will be using the convergence also brings associated se-
Internet protocol.” curity concerns.
The growth of the Internet of Things — To complicate matters, while the vulner-
and, more specifically the Industrial Internet abilities on both sides — enterprise and op-
of Things (IIoT), in particular — is adding to erations — may be similar, the solutions are
both the number of devices and their con- often not directly transferable. The priorities
nectivity. Today, the IT research and advisory of each are necessarily different: while confi-
company Gartner Inc. (Stamford, Conn.; dentiality can be prioritized in the enterprise;
www.gartner.com) estimates 6.4 billion con- availability and integrity must, for the most
nected devices are in use worldwide. By part, take priority on the OT side. In prac-
2020, it forecasts, that total will reach 20.8 tice, a security solution cannot be allowed
billion. Moreover, heavy industries such as to shutdown operator access to data or de-
utilities, oil and gas, and manufacturing are vices that are essential to the safe running of
among the leading users. Each device and the plant, even if the security of those data is
connection expands the possible attack sur- at risk of being compromised.
face for cyberattacks. ISST’s Hankin acknowledged this reality in
Closely connected to the increasing num- his speech: “While there has been a conver-
ber of connected devices is the growth of gence between the two worlds [IT and OT],
the network of remote computer servers particularly in the past five years, there are
casually known as the “Cloud,” which pro- major differences, such as the fact the in-
vides access to infinitely scalable computing dustrial control systems (ICS) tend to have
power and storage. The Cloud provides an to operate in a time-critical way; they have to

Chemical Engineering www.chemengonline.com july 2016 37


Honeywell
• Application whitelisting and device control
to protect individual end nodes
• Advanced analytics, focused on using
“big data” to detect and predict
cyberattacks
The first three are already seeing signifi-
cant uptake, and accompanying security
benefits, among industrial users. The last of-
fers a glimpse at how industrial cybersecurity
is likely to continue to develop in the future,
based on IT trends. It also demonstrates
how the increasing connectivity and elastic
computing power embodied by the IIoT and
Figure 3. Risk analysis operate around the clock; and edge clients, the Cloud can contribute to the security chal-
enables the prioritization of such as sensors and actuators, are becom- lenges they have done so much to highlight.
cybersecurity risks so that
limited resources can be ap- ing much more important” (Figure 2).
plied intelligently In essence, the options for ensuring secu- Risk analysis solutions
rity are more limited in the OT world. This is A key value of risk analysis is that it recog-
partly why the concept of “defense in depth” nizes that resources are finite. Plant owners
is so important to industrial security: without face numerous choices about where and
the option of configuring protection mecha- how to apply security controls and solutions.
nisms to potentially inhibit system availabil- Risk analysis techniques provide a way to
ity, it is even clearer in an OT setting that no quantify, and therefore prioritize, cybersecu-
single security solution can provide complete rity risks, to ensure that limited resources are
protection. A layered approach that employs applied effectively and efficiently to mitigate
several different defenses is the better goal. those that are most severe.
Such an approach means that if (or rather, That quantification is aided by the exis-
when) one layer fails or is bypassed, another tence of standard definitions of risk from
may block the attack. Defense in depth bodies such as the International Organiza-
makes it more difficult to virtually break into tion for Standardization (ISO; Geneva, Swit-
a system, and, if it includes active monitoring zerland; www.iso.org) and the National In-
and a good incidence-response plan, pro- stitute of Standards and Technology (NIST;
motes quicker detection and responses that Gaithersburg, Md.; www.nist.gov). The for-
minimize the impact where an attack does mer defines risk as “the potential that a given
breach security. threat will exploit vulnerabilities of an asset
This also means that — perhaps even or group of assets, and thereby cause harm
more so than in the IT world — security in an to the organization.” The latter characterizes
operational setting cannot rely solely on soft- risk as “a function of the likelihood of a given
ware. As in all operations, success is only threat source’s exercising a particular poten-
achieved through a combination of people, tial vulnerability, and the resulting impact of
processes and technology. that adverse event on the organization.”
Cybersecurity risk is therefore a function
Adapting to the needs of OT of vulnerabilities, threats and potential con-
Notwithstanding these points, though, secu- sequences of a successful compromise. By
rity developments in the IT world do prove accepting this as a definition, risk can be
valuable to operations. Provided the priori- quantified and prioritized.
ties of OT users are accommodated, and the In practice, vulnerabilities will always exist
solutions are implemented in an appropriate — whether in the form of a software bug
framework, recent IT developments offer sig- or due to weak passwords or poor system
nificant potential to boost security in the OT configuration. They cannot be entirely elimi-
world of industrial facilities. nated. Threats, meanwhile, constantly vary,
Four recent technologies, in particular, are and will be driven not just by the availability
worth looking at in more detail: of malicious software or technical knowl-
• Risk-analysis technologies that enable edge, but also by the motivation and means
plants to prioritize investments in cyber- of potential attackers. The consequences of
security exploiting a specific threat have to be calcu-
• Next-generation firewalls, which can bring lated into a relative risk score for each vul-
about radical improvements in network nerability (Figure 3). Owner-operators of in-
protection dustrial control systems can then determine
38 Chemical Engineering www.chemengonline.com july 2016
Honeywell

what level of risk to mitigate, and which


risks they are willing to accept — their
risk appetite.
Since vulnerabilities and threats con-
tinually evolve and expand (with 200,000
new variants of malware identified every
day, for example), the process must be
continuous. Automating the risk-analy-
sis process brings significant benefits to
the security of a plant.
Risk-analysis software does so, and
enables users to monitor networks and
system devices in realtime (Figure 4).
By consolidating complex site-wide
data, risk-analysis software significantly • Non-standard network traffic into Figure 4. By compiling complex, site-
improves the ability to detect threats poorly managed ports can cause un- wide data, risk-management software
can improve the ability of plants to detect
and identify vulnerabilities. Perhaps intended consequences — including threats and identify vulnerabilties
more importantly, by calculating the locking up a communications port,
risk for each device in realtime, it en- tying up resources on the end node,
ables prioritization of risks by their po- or even hanging up an entire end
tential impact to the plant or business. node. This type of probing can reveal
It also provides a realtime update when weaknesses in the configuration or
the risks change due to new threats programming of applications that
or vulnerabilities to the system. Com- results in unintended consequences
bined with well-configured alerts, users • Network scanning can increase the
can assign resources more efficiently, load on an end node to an unman-
and respond more effectively and more ageable level, resulting in a denial
quickly to risks. of service (with the node unable to
In the IT world, risk-analysis and complete normal operation), or even
risk-management solutions have seen a node crash. To avoid this vulner-
widespread uptake, but there are dif- ability, scanners must be “throttled”
ficulties in simply transposing these to properly to protect both the end
an industrial setting. First, the require- nodes as well as the network latency
ments and competencies of the users and bandwidth
— control engineers and operators, An IT VA tool may therefore introduce
as opposed to IT staff — are different. risks to the safe operation of an ICS, as
An OT risk-analysis tool must present much as it may identify them.
results that are meaningful to non-se- Essentially, realtime risk analysis in an
curity specialists who operate the ICS OT environment must be tailored to en-
around the clock. sure that it never interferes with normal
Second, allowance has to be made plant operation or control. It must also
for the OT environment. Many traditional provide realtime, actionable information
vulnerability assessment (VA) tools used that can be used by operators, security
in enterprise systems may be unsuitable administrators and business leaders.
(and possibly unsafe) when applied to VA tools tailored to the ICS environ-
network activity in an ICS. ment are now becoming available, and
This is because they probe aggres- are seeing good uptake. With the scale
sively to test for vulnerabilities, launch- of the cybersecurity challenge continu-
ing a variety of network packets di- ally growing, they are likely to become
rected at every possible port on an an increasingly important tool in helping
end node. The responses are used to operators focus and tailor their cyberse-
determine the state of each port, and curity strategies.
whether the protocols are actively sup-
ported. A database of known vulner- Next-generation firewalls
abilities is then used to match the re- In IT systems, firewalls are among the
sponses, and then further scanning of most widely used cybersecurity mea-
the device is attempted. sures. While antivirus software protects
There are two key problem areas with the end nodes, the firewall monitors and
this technique. controls network traffic based on con-

Chemical Engineering www.chemengonline.com july 2016 39


figured security rules to detect and prevent desirable traffic through as detecting and
network-based cyberattacks. For most busi- blocking threats.
ness, they are the first line of defense in their More highly sophisticated control gives
cybersecurity strategy. plant operators not only increased protection,
Next-generation firewalls (NGFWs) signifi- but also the confidence to allow connections
cantly enhance the protection capabilities they would otherwise feel forced to block: to
of these systems. In addition to traditional enable and control access for an increasing
network protection which restricts access range of applications; to facilitate authorized
to a particular port or address, NGFWs in- personnel using mobile devices; and to pro-
clude deep packet inspection of network mote collaboration across the enterprise with
traffic in realtime. controlled access to realtime data.
Increased analysis of the content of net-
work traffic (not just the source and desti- End-point protection
nation addresses) facilitates a range of ad- Application whitelisting (AWL) is another sta-
ditional defenses: ple in traditional cybersecurity approaches. It
• Application profiling — tracking applica- protects individual end nodes by restricting
tion behavior to raise alerts or interrupt the files that can be executed to only those
communications displaying abnormal specifically authorized to run.
behavior, or patterns associated with Its value is well recognized. Whitelisting
known malware is listed first among the top four strategies
• Protocol support — including, in industrial listed by the Australian government intelli-
NGFWs and most industrial control sys- gence agency, the Signals Directorate, and
tem protocols, such as Modbus, DNP3, last October, NIST published a guide to
OPC and HART. This allows the NGFW to whitelisting for businesses.
be configured to restrict protocols to only As the NIST guide notes, the power of
specific functions, such as restricting the application whitelisting comes from its pre-
ability of applications using Modbus to scriptiveness: “Unlike security technologies,
write to certain registers, or restricting all such as antivirus software, which block
write commands coming into the ICS known bad activity and permit all other, ap-
• Potential to interface with the ICS domain plication whitelisting technologies are de-
controller to identify the user associated signed to permit known good activity and
with specific application traffic on the block all other.”
plant control network and to block unau- Added to this, whitelisting avoids some of
thorized users the maintenance required for technologies
• Advanced threat detection (on high-end like antivirus software or intrusion preven-
NGFW), based on network traffic pat- tion/detection systems (IPS or IDS). Such
terns, and signatures of known malware “blacklisting” technologies require frequent
The potential benefits of NGFWs may even updates to the “known bad” signatures; DAT
be greater in an OT than IT setting. Network files (binary data files with .dat filenames) for
traffic in the OT environment is typically antivirus solutions are updated daily with
more “predictable,” with most communica- new “known malware” signatures. More so-
tion channels clearly defined. That makes it phisticated malware, meanwhile, is being
possible in many cases to more tightly lock designed to evade detection by signature-
down communications traffic on an ICS — based security protections.
and easier to determine deviations from nor- Application whitelisting therefore repre-
mal network traffic patterns. sents a strong additional line of defense
Again, there are significant challenges, against malware that is designed to add new
though: an NGFW can decode some, but software or modify existing software on an
not all, encrypted traffic, for example. ICS end-node. It can also offer some protection
owners also need to coordinate the NGFW for obsolete operating systems no longer
selection with their process control vendors supported by new security patches (such as
to ensure the correct configuration and to Windows Server 2003 and Windows XP op-
ensure that network performance is not af- erating systems).
fected when on critical operations and net- There are challenges for an ICS, how-
work traffic latency. ever. Whitelisting takes time to set up and
However, the potential rewards make this configure in all systems. The difficulty lies in
worthwhile. An NGFW not only provides ensuring that all applications that need to
tighter control of network traffic, but more be run on a particular node are enbled (or
intelligent control: it is as much about letting not blocked). In an ICS, the risks of blocking
40 Chemical Engineering www.chemengonline.com july 2016
or impacting normal operations are often While the value of risk analysis is that it
greater, however. If improperly configured, a recognizes resources for cybersecurity are
whitelisting solution can prevent normal op- finite, the value of advanced analytics is that
erations, causing operators to lose visibility it accepts that complete security is unach-
or control of the plant. It must therefore be ievable. With the threat landscape constantly
tightly integrated into the control system evolving, it is impossible to completely miti-
operation, because it is active before every gate all threats to the ICS.
file execution on the system. Those that have the potential to do the
To minimize the risk, the AWL solution most harm will be those threats of which
should be fully qualified by the ICS vendor organizations remain unaware. The faster
or end user before use. Most solutions also plants can detect malicious actors on the
offer various operation modes: monitoring system or network, the faster they can ad-
or observation, in which users can monitor dress them and minimize the damage.
unauthorized file execution without blocking Advanced analytics uses big data tools
any operations; “self approval” — in which to monitor and analyze a whole range of
message pop-ups enable users to override information sources, from email and social
any blocked executable; and full implemen- media, to network flows and third-party
tation in which whitelisting policies are fully threat feeds. With this information, it can
executed and enforced. The last should identify abnormal patterns that indicate at-
only be used after the site has validated the tacks or intrusions. Not only can advanced
whitelisting configuration against all normal analytic techniques detect recognized
plant usage scenarios. threats, but they can also allow the ability to
Where this is done, however, whitelisting predict new, emerging dangers. Such sys-
has proven an effective and safe solution in tems, for example, can automatically notify
industrial settings, bringing similar benefits users of a cyberattack occurring on a re-
for cybersecurity that have been realized in lated system elsewhere in the world — in
the IT world. In addition to managing execut- realtime — enabling them to take precau-
able files, whitelisting solutions increasingly tions to protect their own sites.
offer a wide range of functionality: While advanced analytics are increasingly
• Managing USB (universal serial bus) and important in cybersecurity, there is little up-
removable storage devices, allowing take to date in the OT world. That, however,
users to restrict USB device usage by is likely to change — as it has with other key
vendor, serial number or function (restrict- technologies in the IT realm. Convergence
ing to read-only, for example) between IT and OT means the challenges
• Extending device management capabil- facing the two are often similar. As long as
ity to control wireless, Bluetooth and all industrial users pay due regard to the dis-
plug-and-play devices on the system tinctive requirements of process control sys-
• Protecting access to the local registry tems, there is no reason the solutions for OT
• Managing access to non-executable files cannot draw on the lessons that have been
• Protecting against malicious behavior learned. In time, it may have insights to share
of programs in memory (such as buffer with IT as well. n
overflows) Edited by Scott Jenkins
• Controlling execution of scripts or activeX
controls Author
• Executing files with reputation-based Mike Baldi is a cybersecurity solutions ar-
decisions chitect at Honeywell Process Solutions (1860
West Rose Garden Lane, Phoenix, AZ 85027;
• Tracking processes changing files on the Email: mike.baldi@honeywell.com); Phone:
system 602-293-1549). Baldi has worked for Honey-
Like NGFWs, application whitelisting is a well for over 36 years. He led a team providing
technical support for Industrial Process Con-
mature technology and integral part of most trol Systems and advanced applications, and
IT cybersecurity strategies. Increasingly, the was the lead systems engineer for HPS sys-
tem test. Baldi joined the HPS Global Architect
same is becoming true in the OT space. team in 2009, and became the chief cybersecurity architect for
HPS, and the lead architect for the HPS Cyber Security Center-of-
Looking to the future Excellence. He lead the design for security initiative — integrating
security into HPS products and the HPS culture. He was also the Editor’s note: For more informa-
Advanced analytics, by contrast, remains primary focal point for HPS product and customer security issues, tion on cybersecurity in the CPI,
resolutely immature in the industrial environ- and for HPS product security certifications and compliance. Baldi visit our website (www.chemen-
ment. It is, however, an important emerging recently moved to the Honeywell Industrial Cyber Security organiza- gonline.com) and see articles by
tion as a cybersecurity solutions architect. Baldi holds a B.S. degree Andrew Ginter (Chem. Eng., July
technology that once again offers significant in computer science, an MBA degree in technology management, 2013) and Eric C. Cosman (Chem.
potential for OT systems. and is CISSP certified. Eng., June 2014).

Chemical Engineering www.chemengonline.com july 2016 41


Plant Functional Safety
Requires IT Security
Cybersecurity is critical for plant safety. Principles developed for plant safety can be applied to
the security of IT systems
Peter Sieber
HIMA Paul Hildebrandt
GmbH

In Brief
Safety and security
standards

What requires
Protection?

Applying Safety
principles to
security

Integrating BPCS and


SIS

IT Security and Safety


recommendations

W
hen the Stuxnet computer username/password authentication for re-
worm attacked programmable mote access. The attack was discovered
logic controllers (PLCs) at Ira- only after the data transmission volume ex-
nian nuclear facilities running ceeded the company’s data plan.
an integrated system, centrifuges were Cyber-related safety risks do not neces-
commanded to literally rip themselves sarily result from criminal activity. During the
apart. This clear demonstration of the link commissioning of one plant, for example,
between cybersecurity and safe industrial the failure of engineering software during
operations was a worldwide wakeup call for the recompiling of the memory mapped
plant managers, IT and automation manag- input (MMI) following a plant shutdown led
ers, safety engineers and many others. to a situation in which an incorrect modifi-
Of course, smaller-scale attacks are much cation was loaded into an integrated safety
more likely, and they are happening. At one controller, and then activated.
plant, where system maintenance was car- These incidents demonstrate the need for
ried out remotely, a cyber attack from abroad specific IT security improvements, and at the
revealed the vulnerability of using simple same time, raise broader questions about
42 Chemical Engineering www.chemengonline.com July 2016
HIMA Americas

tem itself and the distributed control system


(DCS), of which the safety system may be a
part. This is one reason why so many safety
experts call not only for the physical separa-
tion of safety instrumented system (SIS) and
DCS components, but also for different en-
gineering staffs or vendors to be responsible
for each.
To answer the other questions, we need
to highlight two other standards. One is the
international standard IEC 61511 for SIS
in the process industries. Whether inde-
pendent or integrated into an overall basic
process control system (BPCS), the SIS is a
fundamental component of every industrial
process facility.
In this model, the industrial process is sur-
rounded by different risk-reduction layers,
which collectively lower the risk to an ac-
ceptable level (Figure 1). The risk reduction
Figure 1. Under a model the relationship between cybersecurity and claim for the safety layer is set by the safety
put forth under IEC standard plant safety: integrity level (SIL).
61511, an industrial process
is surrounded by a series of
1. Can the “insecurity” of integrated control The first line of protection for any plant is
risk-reduction layers that act systems influence the functional safety of the control and monitoring layer, which in-
together to lower risk a plant? cludes the BPCS. By successfully carrying
2. What needs to be protected? out its dedicated function, the BPCS reduces
3. Can the principles developed for functional the risk of an unwanted event occurring.
safety be applied to security? Typically, IEC 61511 stipulates that the risk
This article considers these questions and reduction claim of a BPCS must be larger
includes operational examples and specific than 1 and smaller than 10. A risk-reduction
recommendations for improving security and capability of 10 corresponds to SIL 1.
safety at industrial facilities. The cyberattack and IT vulnerability pre-
vention layer includes the SIS. The hardware
Safety and security standards and software in this level perform individual
The International Electrotechnical Commis- safety instrumented functions (SIFs). During
sion (IEC; Geneva, Switzerland; www.iec.ch) the risk and hazard analyses carried out as
standard IEC 61508 is the international stan- part of the basic design process of every
dard of rules for functional safety of electri- plant, the risk-reduction factor to be achieved
cal, electronic and programmable electronic by the protection layer is determined.
safety-related systems. According to IEC In most critical industrial processes, the
61508, functional safety is “part of the overall SIS must be rated SIL 3, indicating a risk-
safety that depends on functional and physi- reduction factor of 1,000, to bring the overall
cal units operating correctly in response to risk to an acceptable level.
their inputs.” At the mitigation layer, technical systems
By this definition, the answer to the first are allocated, allowing mitigation of dam-
question posed earlier — Can the “insecu- ages in case the inner layers of protection
rity” of integrated control systems influence fail. In many cases, mitigation systems are
the functional safety of a plant? — has to be not encountered as being part of the safety
“yes.” In the examples cited above, vulner- system, as they are only activated after an
abilities to people and facilities were intro- event (that should have been prevented)
duced. Clearly, functional safety was com- happens. However, in cases where the miti-
promised, and while security breaches may gation system is credited as part of defining
not have led to deaths or injuries, there is additional measures, it may be covered by
no evidence to suggest that such a situation the safety evaluation as well.
could not occur in the future. Now consider the IEC standard for cyber-
Even ruling out malicious threats, the fact security. IEC 62443 covers the safe security
remains that IT security-based vulnerabilities techniques necessary to stop cyber attacks
can be found in all kinds of automation sys- involving networks and systems at indus-
tems. This includes the safety-related sys- trial facilities.
44 Chemical Engineering www.chemengonline.com July 2016
What requires protection? system is only as strong as its weak- the overall risk reduction is reached
According to the most recent ver- est link. When using embedded when two different protection layers
sion of IEC 61511, the answer to the safety systems, all hardware and are in place.
question of what needs to be pro- software that could impair the safety
tected is that both norms and physi- function negatively should be treated Integrating BPCS and SIS
cal structures need to be protected. as being part of the safety function. As an illustrative example, assume
As for norms, the standard calls for IEC 61511 requires different, in- that a risk analysis of a given pro-
the following: dependent layers of protection. cess has led to the conclusion that a
• SIS security risk assessment Unifying two layers of protection will SIL-3-compliant SIS is required. The
• Making the SIS sufficiently resilient require the new risk-reduction evalu- traditional approach implies that a
against identified security risks ation to prove that compliance with risk reduction of greater than 1,000
• Securing the performance of the
SIS system, as well as diagnostic
and fault handling, protection from
unwanted program alterations,
data for troubleshooting the SIF,
and bypass restrictions so that
alarms and manual shutdown are
not disabled
• Enabling/disabling of read/write
access via a sufficiently secure
method Where can you find all your CPI solutions in one spot?
• Segregation of the SIS and BPCS
networks
As for the structural requirements, The Chemical Processing Industry covers a broad range of products such
IEC 61511 instructs operators to as petrochemical and inorganic chemicals, plastics, detergents, paints, pulp
conduct an assessment of their SIS
& paper, food & beverage, rubber and many more. Chemical Engineering
related to the following:
• Independence between protection magazine is uniquely suited to cover this worldwide market.
layers
• Diversity of protection layers
• Physical separation between dif-
ferent protection layers
• Identification of common-cause
failures between protection layers
One other IEC 61511 note has
particular bearing on the issue of
cybersecurity and plant safety. The
standard states: “Wherever practi-
cable, the SIF should be physically
separated from the non-SIF.” Also,
the standard demands that coun-
termeasures be taken for foresee-
able threats.

Applying safety principles Written for engineers by engineers, Chemical Engineering delivers solid
The IEC 61511 (safety) and IEC
engineering essentials and developing industry trends to keep its readers
62443 (security) standards coincide
on the demand for independent lay- abreast of everything they need to keep their facilities running smoothly.
ers of protection. Together, these
standards prescribe:
• Independence between control Missing archived issues or what to share
systems and safety systems
• Reduction of systematic errors Chemical Engineering with your colleagues?
• Separation of technical and man-
agement responsibility
Visit www.chemengonline.com/chemical-engineering-magazine
• Reducing common-cause errors for more information.
The standards also reinforce that 27584

anything and everything within the

Chemical Engineering www.chemengonline.com July 2016 45


and less than 10,000 will be achieved. engaging the various parties to make
The risk reduction is partly covered by sure that potential deficiencies in each
the BPCS (up to 10, as per IEC 61511) task are identified and corrected. While
and by the SIS (1,000 in a SIL-3-com- integrated tools can support the effec-
pliant solution). tiveness of engineering processes, ad-
In the integrated solution, there will be dressing aspects like common-cause
common components for the BPCS and failures requires first narrowing integra-
SIS. Depending on the individual setup, tion to a sustainable level. This helps
this will be either the central processing maintain both efficient engineering pro-
unit (CPU), input-output (I/O) buses or cesses and functional safety at the re-
(parts) of the solution software (for ex- quired level.
ample, the operating system), and sym- The previous comments about BPCS
bol libraries. and SIS independence and diversity
The argument could be made that dif- also apply to engineering tools. A po-
ferent components (of the same make) tential hidden failure of the engineering
may be used for the SIS and BPCS. tool may impair the desired reduction in
However, if common elements (such as overall risk.
operating systems and buses) are used, There are two types of integrated so-
the systematic capabilities of such com- lutions that have either a common con-
ponents may need to comply with the figuration database for SIS and BPCS,
requirements mentioned above. or have independent databases for SIS
It should also be kept in mind that and BPCS, but use the same data ac-
using components such as CPUs with cess mechanisms. Both solutions have
freely configurable software on board – the disadvantage of having a common

The quality of engineering processes, tools and associated services may be even more
important to overall safety results than BCPS and SIS hardware.

and using these same components for cause for potential failures, which would
different tasks – may not be considered infect both the BPCS and SIS.
sufficient leveraging of the integrity level The engineering tool for safety sys-
of the solution. tems should overcome these issues by
These commonly used components, remaining independent (to the greatest
in order to comply with the initial risk extent reasonably possible) from the
reduction requirements, will need to hardware and software environment.
maintain a risk reduction of greater than This is accomplished by having the com-
1,000 by less than 10,000. Practically, plete functionality of the safety engineer-
this means SIL 4, which is currently an ing tool, running in a Windows software
unachievable level. environment, implemented in a way that
allows it to be independent from Win-
Engineering’s key role in security dows functions. This concept allows
The quality of engineering processes, maximum protection from errors and
tools and associated services may be creates a trusted set of engineering data
even more important to overall safety that can be used to program the SIS.
results than BCPS and SIS hardware. Nevertheless, the engineering tool
Proper engineering includes the fol- should allow integrated engineering by
lowing aspects: maintaining interfaces that permit au-
• Reducing complexity by splitting tasks tomated transfer of configuration data
into independent modules (tag-oriented information as well as log-
• Properly defining and verifying ic-oriented data) from third-party sys-
interfaces tems into the trusted set of engineering
• Testing each module intensively data used for programming the SIS.
• Maintain the “four-eyes” principle Furthermore, having the same engi-
when reviewing engineering docu- neers in charge of programming the DCS
ments and results of implementation and safety system ignores the proven
tasks, according to IEC 60158-1, benefits of the checks and balances of
paragraph 8.2.18 independent thinking. For this reason,
Application of this strategy requires IEC 61508 is setting recommendations
46 Chemical Engineering www.chemengonline.com July 2016
for the degree of independence of par- products. Additionally, it should feature
ties involved in design, implementation easy-to-use engineering tools with fully
and verification of the SIS. integrated configuration and program-
ming and diagnostic capabilities.
IT security recommendations Applying these recommendations and
Cybersecurity and plant safety are so adhering to international standards for
intertwined in the connected world of in- separate BPCS and SIS systems help
dustrial processes that an equal commit- plant operators meet their obligation to
ment to both is required to achieve the protect people, communities, the envi-
needed protection. Following the recom- ronment and their own financial security.
mended international standards for func- The good news is that hardware, soft-
tional safety for PLCs (IEC 61508), safety ware and expertise are available today
instrumented systems (IEC 61511) and to help operators meet their obligations
cybersecurity (IEC 62443) provides a for the full lifecycle of their plants. n
path to a safe, secure facility. Edited by Scott Jenkins
For the most robust security and
reduced safety risks, the author ad- Author
vocates the traditional approach of Peter Sieber is vice president for
standalone SIS and BPCS units — ide- global sales and regional development
for HIMA Paul Hildebrandt GmbH (Al-
ally from different vendors — versus an bert-Bassermann-Strasse 28, 68782
integrated BPCS/safety system from Bruehl, Germany, Phone +49-6202
the same vendor. 709-0, p.sieber@hima.com), a leading
specialist in safety automation systems.
For valid security and safety reasons, Sieber is participating in the ongoing
it is also good practice for companies to effort by the steering committees work-
ing on functional safety and IT security
consider an independent safety system standards, IEC 61508 and IEC 62443, respectively. He has
built on a proprietary operating system. been actively involved in the development of the definition of
Of course, such a system can and should both functional safety guidelines and IT security guidelines
for process automation applications.
be completely compatible with DCS

Chemical Engineering www.chemengonline.com July 2016 47


Solids Processing Diverter valves

Dilute-phase Pneumatic Conveying:

Instrumentation and
High
level

Conveying Velocity Feed


Low
level
Follow these guidelines bin

to design a well-instrumented
and controlled system, and to Rotary
Receiving bins with bin
vent filters
Air valve
optimize its conveying velocity inlet
Run or position light
Blower
Amrit Agarwal FIGURE 1. This figure is a schematic flow diagram of the
Consulting Engineer conveying system with run and position lights to show the
operating condition of each component of the system

D
ilute-phase pneumatic con- Process logic description not stop the blower and the solids
veying systems must be Feeding solids into a conveying line feed at the same time.
operated in a certain se- that does not have an airflow with When a conveying cycle has
quence and have sufficient sufficiently high conveying velocity been completed and the solids flow
instrumentation and operating con- will result in plugging of the line. into the conveying line has been
trols to assure reliable operation To prevent this, solids must be fed stopped, the blower motor must
and prevent problems. This article into the conveying line only after continue to run for at least a few
discusses two subjects that are im- the required airflow has been fully more minutes to ensure that all of
portant for successful dilute-phase established. This requirement is the solids that are still inside the
conveying. Discussed below are de- met by allowing the solids feeder conveying line have been conveyed
sign guidelines for instrumenta- to start only after the blower has to the destination bin. If these sol-
tion and controls that can prevent been running for at least five ids are allowed to remain in the
operating problems, such as pipe- minutes. To do this, the rotary- conveying line, they may plug the
line plugging, downtime, equipment valve motor should be interlocked line when the system is restarted.
failure, high power consumption, with the blower motor so that the These solids may also cause con-
product contamination and more. blower motor has run for five min- tamination if a different solid is
The article also provides a simple utes before the rotary-valve motor conveyed in the next cycle.
methodology for finding out if the can start. Solids feed must stop immediately
presently used conveying veloc- When the conveying system is if the normal operating pressure of
ity is too low or two high and for running, the rotary-valve motor the blower increases by 10% and
making the required changes in must stop immediately in the event continues to rise. This is because
this velocity. that the blower motor stops for any the pressure increase is most likely
The required instrumentation reason. If the rotary valve is not due to the conveying line starting to
depends on the degree of automa- stopped, solids feed will continue plug. If the ongoing feed stream is
tion that is necessary, and whether and will plug the pipeline below not stopped, the pressure will keep
the system is to be controlled lo- the feeder. To remove this plug, the increasing, making the plugging
cally or remotely. When manual pipeline will need to be opened. This situation worse.
control of the conveying system required control option is imple- After stopping the feed, the
is used, problems can arise, espe- mented by interlocking the rotary- blower is allowed to run for about
cially if the operators do not have a valve motor with the blower motor five minutes in an effort to flush the
thorough understanding of the de- so that the rotary-valve motor stops plug. If the plug does not flush out
sign and of the required operating when the blower motor stops. and the blower pressure remains
method of the conveying system, or Should the conveying system need high, the blower motor should be
if they do not pay close attention to to be stopped, certain steps must be stopped. The plug is then removed
day-to-day operation of the system. followed: The first step is to stop the by tapping the pipeline to find the
For conveying systems — where solids feed, after which the blower plug location and opening up the
even a single error can result in a is allowed to run until the convey- plugged section of the pipeline.
large financial loss — a well-instru- ing line is empty and the blower Solids feed must also be stopped
mented and automated control sys- discharge pressure has come down if the receiving bin or silo becomes
tem is highly recommended. to the empty-line pressure drop. Do full, as indicated by its high-level
54 Chemical Engineering www.che.com March 2014
light and alarm. If the feed is con- ing status of the blower motor and is located far away from the rotary
tinued, the bin will overfill and the the rotary-valve motor valve, a second pressure indicator
solids will back up into the convey- • Position lights to indicate the di- should be provided just upstream of
ing line, causing pluggage. vert or through position of the di- the rotary valve.
If a conveying line has diverter verter valves These two measurements will
valves, the position of the diverter • Position lights to indicate the low show the overall pressure being
valves must be set up in a “through” and high levels in the receiving provided by the blower, and the
mode or in a “divert” mode before bin or silos pressure drop in the conveying line.
starting the blower and the solids • Run lights to show the operating In addition to local pressure indica-
feed. If the destination bin or silo status of the bin vent filters/dust tors, these pressure measurements
is changed for the next conveying collectors should also be provided in the
cycle, the diverter valves position Figure 1 shows in one glance how control room using pressure
must be changed before the con- the conveying system has been set transmitters.
veying blower and the rotary valve up, and the operating status of all Digital pressure indicators are
are started. components of the system. better than the analog type, because
Graphic control panel. In the Monitoring conveying air pres- they can show the pressure much
central control room, a graphic sure. Conveying pressure is a key more accurately, up to two decimal
panel (Figure 1) should be pro- parameter in pneumatic convey- points. These pressure measure-
vided to show a schematic diagram ing systems. It must be regularly ments should be archived on the
of the conveying system, start- monitored from the control room computer so that historical data are
ing from the air supply blower as well as locally at the blower. available if needed in the future.
to the receiving bins or silos. For measurement of the conveying An alarm for high blower-discharge
This panel should have the follow- pressure, a locally mounted pres- pressure should also be provided in
ing lights: sure indicator should be provided at the control room.
• Run lights to indicate the opera- the blower discharge. If the blower Monitoring blower discharge air

MONITOR VISCOSITY SIMPLY


SENSE MIXER MOTOR HORSEPOWER
WITH UNIVERSAL POWER CELL
EASY INSTALLATION PROFILING A PROCESS
• No holes in tanks or pipes • Power changes reflect viscosity changes
• Away from sensitive processes • Good batches will fit the normal “profile” for
that product
VERSATILE 24
POWER DECREASE
• One size adjusts to motors, from 22
SHOWS BATCH
small up to 150hp 20
IS DONE

• Works on 3 phase, fixed or variable 18

frequency, DC and single phase power POWER


16
SENSOR
SENSITIVE 14

• 10 times more sensitive than 12

just sensing amps 10 DRY MIX ADD LIQUID


HIGH SPEED LOW SPEED MIXER
CONVENIENT OUTPUTS 8 MOTOR
BEGIN HIGH
• For meters, controllers, computers 6
SPEED MIX
4-20 milliamps 0-10 volts 4

0
BATCH 1 BATCH 2 BATCH 3

CALL NOW FOR YOUR FREE 30-DAY TRIAL 888-600-3247


WWW.LOADCONTROLS.COM

Circle 22 on p. 68 or go to adlinks.che.com/50974-22
Chemical Engineering www.che.com March 2014 55
Solids Processing

temperature. A locally mounted veying characteristics of solids rotary valve if the conveying pres-
temperature indicator should be being conveyed, it must be moni- sure starts to increase (indicating
provided at the blower discharge, tored closely. start of formation of a line plug); 2)
and also at the blower after-cooler Rotary-valve motor interlocks Allow the blower motor to continue
discharge if an air cooler is used. with the blower motor. A manu- to run for the selected time, such
This temperature is needed to ally adjustable timer with a selec- as 10 to 15 minutes (in an effort
carry out calculations for the “as- tor switch should be provided in to clear the line plug); and 3) Re-
built” conveying system. If this air the control room to provide three start the rotary-valve motor if the
temperature can affect the con- functions: 1) Automatically stop the conveying pressure falls to the nor-
mal pressure.
Diverter valves. Position lights are
provided in the control room graphic
panel to indicate if the valves are in
the “through” or “divert” position.
Receiving bins. Low- and high-
level lights are provided in the
graphic panel for the receiving bins.
An alarm should be provided in the
control room to indicate high level
in the bins. At the high level, the ro-
tary valve motor should be stopped
automatically.
Bin vent filters/dust collectors.
The bin vent filters or the dust col-
lectors on the bin vents must be
running before the conveying sys-
tem is started. A “run” light for the
Understanding Protective Coatings filter should be provided in the
graphic panel.
in Hot Environments Pressure drop indicators should
be installed locally to show the pres-
sure drop across the filter elements.
The use of fireproofing, high-temperature, and other coatings to Their locations should be easily ac-
protect infrastructure in high-temperature facilities is becoming more cessible to the operating staff. For
widespread around the world. The 2014 Bring on the Heat Conference conveying materials that have high
is focused on providing an informative look into how these coatings are dust loading, alarms for low- and
used and how they benefit different industries. This event will provide high-pressure drops should be pro-
presentations, case studies, and forum discussions on the following vided in the control room. The low-
topics: pressure drop alarm would indi-
cate a ruptured filter element, and
• Corrosion under insulation • Passive fire protection the high pressure drop alarm
• Thermal insulation coatings • Coatings needs for owners would indicate a completely clogged
• Thermal spray aluminum filter element.

Instrumentation checklist
Register Today A summary of the instrumentation
requirements, as described above, is
provided below:
Register by May 16 For the blower:
• Local and control room mounted
to SAVE! running lights for the blower
motor
• Local pressure indicator at the
blower discharge
To register or for more information go to • Local temperature indicator at
www.nace.org/both2014 the blower discharge
• Local temperature indicator at
the blower after-cooler discharge,
Circle 27 on p. 68 or go to adlinks.che.com/50974-27 for applications using a cooler
56 Chemical Engineering www.che.com March 2014
Saltation line, Dilute
joining pressure phase:
M Unstable minima and suspension
flow saltation velocity flow

d
H

be
∆P

L
Log (pressure drop per unit length), L

W2
ed
=
ck
d ing • Alarms for low- and high-pres-
oa
Pa

sl sure drop across filter elements

e
lid C

im
K So (optional)

reg
W 1
G g= Graphic control panel:
din

ble
loa • Graphic panel showing the con-
sta
G
l ids B
Dense Un So e veying system route with run
F ip
phase
ptyp lights for the blower motor and
regime E Em rotary valve motor, position lights
Dilute for the diverter valves, low- and
phase:
strand D high-level lights for the receiving
flow bins, and run lights for the bin-
A vent filters
Saltation velocity/minimum pressure

Finding the conveying velocity


Log (gas velocity), V
Along with conveying pressure, con-
Dilute phase: Dilute phase: veying velocity is perhaps the most
Packed bed Dense phase strand flow suspension flow
important variable in pneumatic
conveying. After a conveying sys-
tem has been installed and is going
through startup, its conveying veloc-
FIGURE 2. This figure shows the relationship of conveying velocity with conveying ity should be checked to make sure
pressure at different solids-loading rates W1 and W2. The solids-loading rate is the it is not too low or too high, and is
solids-conveying rate divided by the internaal cross-sectional area of the convey-
ing pipeline. For these two loading rates, the figure also shows the transition points about equal to the conveying veloc-
(Points D and G) at which the conveying system migrates from dilute to dense phase. ity that is required. If the convey-
For solids-loading rate W1, as the conveying velocity is reduced, the conveying sys- ing velocity is too low, it may cause
tem's operating point moves from Point C to Point D in dilute phase; and then in the line plugging problems; if it is too
dense phase, from Point D to Points E, F and G. Similarly, for the solids-loading rate
W2, the operating point moves from Point H to Point G in the dilute phase; and then
high, it will result in higher particle
in the dense phase from Point G to Points K, L and M attrition, pipeline wear, and higher
energy usage.
The conveying velocity used in
the conveying system’s design cal-
Feed
bin culations may be too low or too high
because it is difficult to find a reli-
Vent
Pressure Flow control able method to determine its cor-
indicator PI valve and
flow indicator rect value. This value depends upon
many variables, such as solids par-
FLC
Air
Rotary ticle size, bulk-solids density, solids-
valve
inlet to-air ratio, air density, pipeline di-
Blower
ameter and others. Presently, there
are two methods to find the convey-
FIGURE 3. This figure shows the design of the vent air system for venting out a ing velocity. The first method is to
portion of the blower airflow to determine saltation velocity use equations to calculate saltation
velocity (the gas velocity at which
• Pressure transmitter at the • Interlocks with the blower motor particles will fall out of the gas
blower discharge with a pressure For the diverter valves: stream). These equations have been
indicator in the control-room con- • Position lights to indicate developed by researchers to find
trol panel. Computer storage of “through” and “divert” positions the impact of the above-mentioned
pressure data • Hand switches for control room variables on saltation velocity. As
• Control room alarm for high operation of valve positions they are based on research work
blower discharge pressure Receiving bin: that is carried out in small-scale
• Blower motor interlocks with the • Low-level and high-level switches test equipment in a laboratory, they
rotary-valve motor with indicating lights for the re- do not cover the entire range of sol-
For the rotary valve: ceiving bins ids and all of their properties. These
• Local and control-room-mounted • Control room alarm to indicate equations can be found in published
running lights for rotary valve high level in the bin books and literature.
motor Bin vent filters/dust collectors: Th second method is to use con-
• Control-room-located, manually • Running lights for the bin vent fil- veying velocity values that are avail-
adjustable timer for starting and ters or dust collectors able in published literature such as
stopping the rotary valve motor • Local pressure-drop indicator those given in Table 1. It should be
Chemical Engineering www.che.com March 2014 57
TABLE 1. COMMONLY USED CONVEYING VELOCITIES
Conveying Conveying
Solids Processing Material velocity, ft/ Material velocity,
min ft/min
Alum 5,100 Malt, barley 3,300
noted that these published values Alumina 3,600 Oats, whole 4,200
are applicable to only those pneu- Bentonite 3,600 Nylon, flake 4,200
matic conveying systems from which
Bran 4,200 Paper, chopped 4,500
they were derived, but may or may
not be applicable for new conveying Calcium carbonate 3,900 Polyethylene pellets 4,200
systems. This is because the convey- Clay 3,600 Polyvinylchloride, powder 3,600
ing velocity for a particular convey- Coffee beans 3,000 Rice 4,800
ing system depends on the values of
Coke, petroleum 4,500 Rubber pellets 5,900
various factors and variables such
as solids particle size, particle size Corn grits 4,200 Salt cake 5,000
distribution, particle density, air Corn, shelled 3,300 Salt, table 5,400
density, solids conveying rate, pipe- Diatomaceous earth 3,600 Sand 6,000
line diameter and more. As shown
Dolomite 5,100 Soda ash, light 3,900
in Table 1, the published values
may not be applicable because they Feldspar 5,100 Starch 3,300
do not give any information on the Fluor (wheat) 3,600 Sugar, granulated 3,600
values of the variables on which Flourspar 5,100 Trisodium phosphate 4,500
they are based.
Lime,hydrate 2,400 Wheat 3,300
A proposed method Lime, pebble 4,200 Wood flour 4,000
This third method is based on run-
ning a test on the as-designed and is then subtracted from the air sup- decreases when solids flow through
built conveying system to determine plied by the blower to determine a bend. This decrease can be 5 to
the true value of the solids saltation the airflow going to the conveying 20% depending on the properties of
velocity. The value of the saltation line. The conveying velocity is then the solid being conveyed. Unless the
velocity obtained by the test will be calculated based on this airflow and conveying velocity is high enough,
accurate because it is based on the pipeline diameter. such a decrease can result in sal-
properties of the solids being con- To run this test, the conveying tation of the solids and plugging of
veyed and on the as-designed and system is started and run at full the bend or its downstream convey-
built conveying system. This value capacity for a few minutes to bring ing line.
is then used to determine the value it to steady-state conditions. Keep- This test-derived optimum con-
of the conveying velocity. ing the solids flowrate constant, the veying velocity is compared with the
This test requires gradually re- vent valve is manually and gradu- velocity that is actually being used.
ducing the airflow that goes into the ally opened to start ventinga few If the actual velocity currently in
conveying line so that the conveying cubic feet per minute of the convey- use is lower, then the blower speed
velocity continues to decrease until ing air, reducing the conveying air- is increased to match the optimum
it reaches saltation conditions. The flow and the conveying velocity. conveying velocity; if it is higher,
Zenz diagram (Figure 2) shows both A close watch is kept on the dis- then the blower speed is decreased.
the dilute- and dense-phase con- charge-pressure indicator installed The change in speed is determined
veying regimes, and the saltation at the blower outlet. This pressure from the blower performance curve.
velocity interface between them. As will keep falling with the decrease The speed change is implemented
shown, the conveying pressure is at in airflow, but as shown in Figure by changing the belts and sheaves
a minimum at the saltation velocity. 2, its value will eventually reach a of the blower. ■
In the test, the airflow and hence point after which it will start to in- Edited by Suzanne Shelley
the conveying velocity is reduced crease. The objective of the test is to
until this minimum pressure point find the airflow at that point. The Author
is reached, after which the pressure vent airflow is gradually increased Amrit Agarwal is a con-
sulting engineer with Pneu-
starts to increase. until this point is reached and the matic Conveying Consulting
The equipment required for this pressure, instead of falling, starts to (7 Carriage Rd., Charleston,
WV 25314; Email: polypcc@
test is shown in Figure 3. A vent increase. This is the minimum pres- aol.com). He retired from The
Dow Chemical Co. in 2002,
line is installed in the air-supply sure point beyond which the convey- where he worked as a resident
line at the discharge of the blower. ing system migrates to dense-phase pneumatic-conveying and sol-
ids-handling specialist. Agar-
Its purpose is to vent off to the at- conveying. At this point, the solids wal has more than 40 years of
mosphere some of the conveying air reach their saltation velocity. design, construction, operat-
ing and troubleshooting experience in pneumatic
that is being supplied by the blower. The saltation velocity value ob- conveying and bulk-solids-handling processes. He
In this vent line, a flow-control valve tained by the test is increased by a holds an M.S. in mechanical engineering from the
University of Wisconsin, Madison, and an MBA
with a flow indicator is used to con- safety factor of about 30% to select from Marshall University (Huntington, W. Va.).
He has written a large number of articles and
trol the airflow that is to be vented an appropriate value for the convey- given classes on pneumatic conveying and bulk
out. The airflow that is vented out ing velocity. Solids velocity always solids handling.

58 Chemical Engineering www.che.com March 2014


Alarm Management By
the Numbers
Deeper understanding of common alarm-system metrics can improve remedial actions and
result in a safer plant

Kim VanCamp
Emerson Process
Management

In Brief
Alarm management
Performance Metrics

Alarm system
example metrics

average alarm rates

Peak Alarm rate

Alarm priority
distribution

Alarm source
contribution

stale alarms

Closing remarks

Figure 1. A better understanding of alarm system metrics can lead to more focused remedial actions and help to make the
plant safer

D
o you routinely receive “alarm man- defining a valid alarm), alarm management
agement performance” reports, methods (for instance, rationalization tech-
or are you expected to monitor niques), justification (such as the benefits of
a managerial dashboard equiva- investing in alarm management) and tools
lent? What do you look for and what does it (including dynamic alarming enablers). This
mean? We all know that fewer alarms mean article provides a different perspective. Writ-
fewer operator interruptions and presumably ten for process plant operation managers or
fewer abnormal process or equipment con- others that routinely receive alarm manage-
ditions. But a deeper understanding of the ment performance reports, this article aims
more common alarm-management metrics to explain the most common metrics, with-
can yield greater insight, leading to more fo- out requiring an understanding of the alarm-
cused remedial actions and ultimately to a management discipline in depth.
safer, better performing plant (Figure 1).
This article reviews the now well estab- Alarm-management KPIs
lished benchmark metrics associated with The first widely circulated benchmark met-
the alarm-management discipline. Most arti- rics, or key performance indicators (KPIs), for
cles previously published on alarm manage- alarm management relevant to the chemical
ments cover alarm concepts (for example, process industries (CPI) were published in the
50 Chemical Engineering www.chemengonline.com march 2016
Table 1. Example of typical alarm performance metrics, targets and action limits
Metric Target Action limit
Average alarm rate per operator (alarms per day) < 288 > 432
Average alarm rate per operator (alarms per hour) < 12 > 18
Average alarm rate per operator (alarms per 10 minutes) 1–2 >3
Percent of 10-minute periods containing > 10 alarms < 1% > 5%
Maximum number of alarms in a 10 minute period ≤10 > 10
Percent of time the system is in flood < 1% > 5%
Annunciated priority distribution (low priority) ~80% < 50%
Annunciated priority distribution (medium priority) ~15% > 25%
Annunciated priority distribution (high priority) ~5% >15%
Percent contribution of top 10 most frequent alarms < 1% to ~5% > 20%
Quantity of chattering and fleeting alarms 0 >5
Stale alarms (number of alarms active for more than >24 hours) < 5 on any day >5

1999 edition of the Engineering Equipment publications state that an average long-term
and Materials Users Association publication rate of new alarms occurring at a frequency
EEMUA-191 Alarm Systems – A Guide to of up to 12 alarms per hour is the maximum
Design, Management and Procurement [1]. manageable for an operator. Suggested
Later works from standards organizations, performance levels such as this can pro-
such as the 2009 publication International So- vide a reasonable starting point if you are
ciety of Automation (ISA) 18.2 Management just beginning an alarm-management pro-
of Alarm Systems for the Process Industries gram. But before deciding what constitutes
[2] and the 2014 publication IEC62682 Man- a reasonable set of targets for your site, you
agement of alarms systems for the process should also consider other firsthand inputs,
industries [3], built upon EEMUA-191 and like surveying your operators and reviewing
have furthered alarm-management thought in-house studies of significant process dis-
and discipline. For example, they provide a turbances and alarm floods. Note that more
lifecycle framework for effectively manag- research into the human factors that affect
ing alarms and establish precise definitions operator performance is needed to validate
for core concepts and terminology. Yet fif- and potentially improve on the current pub-
teen years later, little has changed regarding lished performance targets. Important work
the metrics used to measure alarm-system in this area is ongoing at the Center for Op-
performance. This consistency in measure- erator Performance (Dayton, Ohio; www.
ment has been positive in many respects, operatorperformance.org).
leading to the wide availability of generally
consistent commercial alarm analytic re- Alarm system example metrics
porting products, from both control-system A typical alarm-performance report contains
vendors and from companies that specialize a table similar to Table 1, where the metrics
in alarm management. Consequently, selec- and targets are based upon, and in many
tion of an alarm-analysis product may be cases, copied directly from, the EEMUA-
based on factors such as ease of use, inte- 191, ISA-18.2 and IEC62682 publications. It
gration and migration, reporting capabilities, is also common to see locally specified ac-
price, support availability and so forth; with tion limits based on a site’s alarm philosophy.
reasonable certainty that the KPIs derived When a target or action limit is exceeded, it
from the chosen product can be interpreted is important to ask: what problems are likely
consistently and compared across sites and contributing to the need for action, and what
across differing process control, safety and are the actions? These questions are the
other open platform communications (OPC)- focus of the following discussion.
capable alarm-generating sources.
In addition to defining the KPI measure- Average alarm rate
ments, the EEMUA-191, ISA-18.2 and The average alarm rate is a straightforward
IEC62682 publications also suggest perfor- measure of the frequency with which new
mance targets, based in large part on the alarms are presented to the operator, ex-
practical experience of the companies par- pressed as an average count per day, hour
ticipating in the committees that contributed or per 10-minute interval. As alarm frequency
to each publication. As an example, these increases, an operator’s ability to respond

Chemical Engineering www.chemengonline.com march 2016 51


Figure 2. Timeline views of 1400 Alarm rates for Figure 2 on a per-hour
the data can reveal periods
where alarm performance is 1200 basis
not acceptable • Overall: 16.5
1000

Average alarm rate


• During alarm floods: 100.7
800 • Excluding alarm floods: 7.9

600

400 n Critical
n Warning
n Advisory
200

5/6/2009
5/7/2009
5/8/2009
5/9/2009
5/10/2009
5/11/2009
5/12/2009
5/13/2009
5/14/2009
5/15/2009
5/16/2009
5/17/2009
5/18/2009
5/19/2009
5/20/2009
5/21/2009
5/22/2009
5/23/2009
5/24/2009
5/25/2009
5/26/2009
5/27/2009
5/28/2009
5/29/2009
5/30/2009
5/31/2009
Date

correctly and in time to avoid the ultimate mal condition requiring a timely response
consequence of inaction decreases. If the is occurring
rate is excessively high, it is probable that • Chattering or other frequently occurring
some alarms will be missed altogether or the nuisance alarms are present. These often
operators will ignore them, thus eroding their originate from non-process alarm sources
overall sense of concern and urgency. So of marginal interest to the operator, such
clearly it is an important metric. as field devices or system hardware diag-
Averages can be misleading, however, be- nostics. Chattering alarms can also indi-
cause they provide no sense of the peaks in cate an incorrect alarm limit or deadband
the alarm rate, making it difficult to distinguish • Redundant alarms, where multiple alarms
“alarm floods” from steady-state “normal” are presented when a single abnormal situ-
operation. Consequently, most alarm per- ation occurs. An example is when a pump
formance reports supplement this basic KPI is shut down unexpectedly, generating a
value with a timeline view or separate calcu- pump fail alarm in addition to alarms for low
lation of alarm rates for both the times when outlet flow and low discharge pressure
operation is normal and for times of an alarm • A problem with the metric calculation is oc-
flood. Figure 2 presents a typical example. curring. A correct calculation only counts
The average alarm rate of 16.5 alarms per new alarms presented to the particular
hour exceeds the target KPI value of 12 from operator or operating position for which
Table 1, but is slightly less than the action the metric is intended, taking into consid-
limit of 18 per hour, and so might not raise eration any by-design threshold settings or
concern, while the timeline view shows that other authorized filtering mechanisms that
there are significant periods of time where cause fewer alarms to be presented to the
the performance is unacceptable. operator than may be recorded in system
Common contributors to an excessively event logs
high alarm rate include the following:
• The alarm system is being used to notify Peak alarm rate
the operator of events that do not consti- The two metrics — the percentage of
tute actual alarms, such as communicat- 10-minute periods with more than 10
ing informational “for your information” alarms, and the percent of time spent in
messages, prompts, reminders or alerts. an “alarm flood” state — are calculated dif-
According to ISA-18.2, an “alarm” is an in- ferently, but are highly similar in that they
dication to the operator that an equipment quantify how much of the operator’s time
malfunction, process deviation or abnor- is spent within the highly stressful circum-
stance of receiving more alarms than can
Figure 3. Pie charts can sup- New alarm activation rate distribution
plement alarm performance be managed effectively.
reports and give information 6.6%
n Acceptable EEMUA-191 defines the start of an alarm
on how much time is spent in 10.1% (0–1 per 10 min.) flood as a 10-minute period with more than
the acceptable range
n Manageable 10 new alarms, continuing through subse-
(2–4 per 10 min.) quent 10-minute intervals until reaching
n Demanding a 10-minute interval with fewer than five
20.0% 63.4% (5–9 per 10 min.) new alarms. Equally acceptable is to de-
n Unacceptable fine a flood simply as a 10-minute period
(≥10 per10 min.) with more than 10 new alarms. Often, an
alarm-performance report will supplement
these two metrics with a pie chart (Figure
52 Chemical Engineering www.chemengonline.com march 2016
3) that segments the report period into Alarm priority distribution Figure 4. When the number of high-pri-
10-minute periods that are catego- ority alarms exceeds that of low-priority
alarms, the methodology of how alarms
rized into named alarm-rate ranges, 8.7%
are assigned priority should be evaluated
such as acceptable, manageable, de-
manding and unacceptable. n Medium
Another commonly included metric in 51.8% n High
39.4%
the alarm-performance report, the peak n Low
number of alarms within a 10-minute
period, is a straightforward measure
of the degree of difficulty of the worst-
case alarm flood for the operator. In
poorly performing alarm systems, it is
common to see peak alarm counts in ties: low, medium, high and very-high.
a 10-minute period that exceed 250, There are a number of well accepted
a total that would overwhelm even the methods for assigning priority, the most
most highly skilled operator. common being a systematic guided
Common contributors to high peak- (selection-based) consideration of the
alarm-rate frequency and severity in- severity of the consequence of inaction
clude the following items: combined with the time available for the
• Multiple redundant alarms for the operator to take the required action.
same abnormal condition. The op- Conventional wisdom says that the an-
timum situation is of course that any nunciated alarm-priority distribution ex-
single abnormal event will produce perienced by the operator for low-, me-
just one alarm, representing the best dium- and high-priority alarms should
choice in terms of operator compre- be in an approximate ratio of 80, 15 and
hension and the quickest path to take 5%. Ultimately however, the goal should
remedial action. This requires study of be to guide the operator’s determination
alarm causes and often leads to the of the relative importance of one alarm
design of conditional, first-out or other compared to another, based on their
form of advanced alarming logic importance to the business.
• Cascading alarms. The sudden Figure 4 illustrates a situation where
shutdown of equipment often trig- the number of high-priority (critical)
gers automated actions of the control alarms being presented to the operator
system, which in turn, triggers more far exceeds the low-priority (advisory)
alarms alarms, suggesting the need to review
• False indications. When routine the consistency and methodology of
transitions between process states the priority assignment.
occur, the alarm system is not usually Common contributors to out-of-bal-
designed to “follow the process,” so ance alarm-priority distributions include
it can therefore produce a multitude the following:
of false indications of an abnormal • Alarm prioritization (a step in the ratio-
condition. Likewise, logic is typically nalization process) has not been per-
required to detect state changes and formed and alarm priorities have been
suppress or modify alarms accord- left at their default values
ingly • Misuse of the priority-setting scheme
Some systems provide specialized to classify alarms for reasons other
alarm views that present alarms in a than providing the operator with a tie-
graphical pattern to aid an operator’s breaker during alarm peaks. For ex-
comprehension of peak alarm events ample, using priority to classify alarms
and their associated causality, supple- by impact categories, such as envi-
menting the classic alarm list to help pro- ronmental, product quality, safety/
vide a built-in layer of defense against the health, or economic loss
overwhelming effects of an alarm flood. • Lack of discipline in setting prior-
ity based on consideration of direct
Alarm priority distribution (proximate) consequences rather than
When faced with multiple alarms, the ultimate (unmitigated) consequences.
operator must decide which to address While it may be the case that a de-
first. This is — or should be — the basis signed operator action could fail, fol-
for assigning priority to an alarm. Most lowed by a protective system failure,
systems will employ three or four priori- followed by a subsequent incorrect

Chemical Engineering www.chemengonline.com march 2016 53


450 100.0% — alarms that operators have come to ex-
400 pect, and in most cases, ignore or con-
350 80.0% sider to be informational
300
• The alarm system is being misused to (fre-
Number of alarms

60.0% quently) generate operator prompts based


250
on routine changes in process conditions
200
40.0% or operating states that may or may not
150 require action
100 20.0% Eliminating chattering alarms is generally
50 straightforward, using signal-conditioning
0 0.0% features found in most control systems,
such as on-delay, off-delay and hysteresis
FITST111

IIUP16P1

FICUP1516

IIPX15P1

OPC_FI-N2-051

TIFH42106

FIC-1252

PICFP2043

TIFG41106

FIFC1054
n Alarms
— Cumulative % (deadband).

Stale alarms
Alarm source
A stale alarm is one that remains annunci-
ated for an extended period of time, most
Figure 5. A small number human response, such what-if consider- often specified as 24 hours. Stale alarms
of alarm sources can often ations are likely to lead to a vast skewing are surprisingly challenging to quantify.
account for the majority of
alarms of alarm priorities toward critical Metrics based on event histories require the
presence of both the start and ending alarm
Alarm source contribution event in order to compute an alarm’s an-
The percent of alarms coming from the top- nunciated duration. There is no event rep-
ten most frequent alarm sources relative to resenting the attainment of a certain age of
the total alarm count is a highly useful met- an annunciated alarm. Thus, it is common
ric for quantifying, identifying and ultimately to miss counting stale alarms if their acti-
weeding out nuisance alarms and alarm- vation event or all-clear event falls outside
system misuse. This is especially true if the the range of dates and times covered in the
alarm performance report covers a range of event history. Consequently, there are alter-
time where operations were routine and with- nate methods for quantifying stale alarms,
out significant process upsets or equipment such as periodic sampling of the active
failures. The top-ten alarm sources often alarm lists at each operator workstation, or
provide “low-hanging” fruit for alarm-man- simply counting the number of alarms that
agement performance improvement. They attained an age greater than the threshold
are a handful of alarms, which if addressed, age. Given this variation in methods, it is im-
will create a noticeable positive change for portant to exercise caution when compar-
the operator. ing stale-alarm metrics across different sites
Figure 5 shows a pattern observed in that may be using different alarm-analytic
many control systems, where as few as applications.
ten alarm sources (like a control module or In addition to being hard to quantify, stale
transmitter) out of the many thousands of alarms can also be some of the most difficult
defined alarm sources, collectively account nuisance alarms to eliminate. Thus in some
for about 80% of all of the alarms presented respects the upward or downward trend in
to the operator. In this example, the first stale alarm counts provides an informal in-
alarm source (FIST111) alone was respon- dication of the overall ongoing health of the
sible for 15% of all of the alarms presented alarm management program.
to the operator. Common contributors to stale alarm
Another related metric is the count of counts include the following:
chattering alarms — alarms that repeatedly • Routine transitions between process states
transition between the alarm state and the where the alarm system is not designed to
normal state in a short period of time. The adapt and therefore provides false indica-
specific criteria for identifying chattering tions of an abnormal condition
alarms vary. The most common method is • Alarms associated with standby or idle
to count alarms that activate three or more equipment
times within one minute. • Alarms configured to monitor conditions no
When the top-ten alarm sources generate longer relevant or available, an indicator of
over 20% of all the alarms presented to the poor management-of-change processes
operator, it is a strong indicator that one or • Alarms that are essentially latched due to
both of the following is the case: excessive application of hysteresis
• Some of those alarms are nuisance alarms • Alarms that persist beyond the called-for
54 Chemical Engineering www.chemengonline.com march 2016
operator action, waiting for maintenance methods for corrective action, one outstand-
action. This likely constitutes an incorrect ing resource is Ref. 4. n
use of the alarm system, using it as a re- Edited by Scott Jenkins
cording method for outstanding mainte-
nance actions References
In conjunction with reviewing the number 1. EEMUA Publication 191 — Alarm Systems: A Guide to Design,
of stale alarms or the list of stale alarms, it is Management and Procurement – Third edition, published by the
Engineering Equipment and Materials Users Association in 2013.
also important to review what alarms have
2. ANSI/ISA–18.2–2009 — Management of Alarm Systems for
been manually suppressed (thus removing the Process Industries – approved June 23, 2009. ISBN: 978-1-
them from the view of the operator). Sup- 936007-19-6.
pressing the alarm will remove a stale alarm 3. ANSI/ISA–18.2–2009 — Management of Alarm Systems for
the Process Industries – approved June 23, 2009. ISBN: 978-1-
from the alarm list (effectively reducing the 936007-19-6.
number of stale alarms), but will not address 4. International Society of Automation. Technical Report ISA-TRI
the underlying condition. 18.2.5, Alarm System Monitoring Assessment and Auditing, ISA.
2012.
Closing remarks
This article touches on just some of the key Author
alarm-system performance metrics and Kim VanCamp is the DeltaV marketing prod-
what the numbers represent, in terms of the uct manager for alarm management at Emer-
son Process Management (8000 Norman
issues that lay behind them and possible ac- Center Drive, Bloomington, MN 55437; Phone:
tions to address them. With this understand- 1-952-828-3500; Email: Kim.VanCamp@
ing, periodic reviews of alarm-performance emerson.com). He joined Emerson in 1976
and has held senior assignments in manufac-
reports should lead to more focused actions turing, technology, field service, customer
that can improve operator effectiveness and service, service marketing and product mar-
keting. VanCamp is a voting member of the
thereby reduce the risks for economic loss, ISA-18.2 committee on Management of Alarm Systems for the Pro-
environmental damage or unsafe situations. cess Industries and has published multiple papers on alarm man-
For further reading on these and other alarm agement. He holds a bachelor’s degree in electrical engineering
from the University of Nebraska.
performance metrics, including suggested

Oil, Gas and Chemicals Filtration & Separations Conference – Expo

American Filtration and Separations Society • May 9-11, 2016 • Houston Marriott Westchase, Houston, TX
The American Filtration & Separations Society invites you to
Oil, Gas and Chemicals Filtration & Separations Conference – Expo
Conference website: http://spring.afssociety.org/, Contact Conference Chair for questions, David Engel +1 (832) 296-6624
Conference Features Technical Conference Topics Participating Companies
• 3 Plenary sessions • Filtration • Berry Plastics (formerly AVINTIV)
• 3 Concurrent tracks • Coalescing • Delta Pure Filtration
• 72 Technical papers • Adsorption & Absorption • Dorstener Wire Tech
• Student Poster Competition • Air/Gas Purification • GKD – USA Metal Fabrics
• Short Courses on Monday, May 9 • Bulk Separations and Cyclonics • IFTS Filter Testing
• Vendor Expo • Chemical Assisted Separations • Industrial Netting
Plenary Speakers • Equipment & Systems • Nexo Solutions
• Larry Ryan – Dow Chemical Company • Media Technology • Onyx Specialty Paper
• Michael Spearman – Otronix • Produced Water • Sefar Filtration
• Scott Northrop – Exxon Mobil • R & D Innovation • Spifil Filtration

Don’t miss the premier filtration & separations Expo in the USA.
Expo only passes are complimentary but registration in advance is required.
Circle 2 on p. 90 or go to adlinks.chemengonline.com/61493-02

Chemical Engineering www.chemengonline.com march 2016 55


Part 2

Understand and Cure


High Alarm Rates
Alarm rates that exceed an operator’s ability to manage them are common. This article explains
the causes for high alarm rates and how to address them

Bill Hollifield
PAS Inc.

In Brief
alarm rates

Averages can be
misleading

bad actor alarm


reduction

Alarm rationalization

Alarm management
work processes

Concluding remarks

M
odern distributed control sys- Figure 1. Alarm rates on the order of thousands per day are
tems (DCS) and supervisory con- not uncommon in some CPI facilities
trol and data acquisition (SCADA) to create an alarm for any imaginable con-
systems are highly capable at dition. A simple analog sensor, such as one
controlling chemical processes. However, for temperature, will likely have a dozen alarm
when incorrectly configured, as is often the types available by simply clicking on check
case, they also excel at another task — gen- boxes in the device’s configuration. Without
erating alarms. It is common to find alarm following sound alarm-management prin-
rates that exceed thousands per day or per ciples, the typical results are over-alarming,
shift at some chemical process industries nuisance alarms, high alarm rates and an
(CPI) facilities (Figure 1). This is a far greater alarm system that acts as a nuisance distrac-
number than any human can possibly handle tion to the operator rather than a useful tool.
successfully. This article examines the nature Whenever the operators’ alarm-handling
of the problem and its cure. capacity is exceeded, then operators are
The alarm system acts as an intentional in- forced to ignore alarms, not because they
terruption to the operator. It must be reserved want to do so, but because they are not able
for items of importance and significance. An to handle the number of alarms. If this is the
alarm should be an indication of an abnormal case, the average, mean, median, standard
condition or a malfunction that requires op- deviation, or other key performance indica-
erator action to avoid a consequence. Most tors (KPIs; see Part 1, p. 50) for alarms do
alarm systems include interruptions that not matter, because plant managers have no
meet this definition, but also many miscel- assurance that operators are correctly ignor-
laneous status indications that do not. ing inconsequential alarms or are paying at-
A major reason for this situation is that con- tention to the ones that matter. This situation
trol system manufacturers make it very easy contributes to many major accidents.
56 Chemical Engineering www.chemengonline.com march 2016
Alarm rates 3,000 Annunciated alarms per day Peak rates
— Annunciated alarms
The International Society of Automation (ISA; — Annunciated alarms without the 10 most frequent 9,195
Research Triangle Park, N.C.; www.isa.org) 2,500 14,899
Standard 18.2 on alarm management identi-
fies the nature of the problem and offers a 2,000
variety of assessment measurements. An
important measurement is the rate of alarms
1,500
annunciated to a single operator.
Figure 2 shows an overloaded alarm sys-
1,000
tem. The difference between the two lines is
the effect of including or removing only 10
individual high-rate nuisance alarms. This is 500
a common problem that is discussed later in
the article. 0
Acceptable range: 150 to 300 58 days
To respond to an alarm, an operator must
detect the alarm, investigate the conditions with the proper determination of the cause Figure 2. Removing a small
causing the alarm, decide on an action, take or causes. The diagnosis time varies based number of high-rate alarms
can have a large effect on the
the action and finally, monitor the process upon the operator’s experience and involve- alarm system’s overall profile
to ensure that the action taken resolves the ment in previous similar situations.
alarmed condition. These steps take time Process control graphics (human-ma-
and some must necessarily be executed se- chine interfaces; HMIs) play a major role in
quentially. Others can be performed in paral- effective detection of abnormal situations
lel as part of a response to several alarms and responses to them. Using effective
occurring simultaneously. HMIs, an operator can quickly and prop-
Given these steps, handling one alarm in erly ascertain the cause and corrective ac-
10 minutes (that is, approximately 150 over a tion for an abnormal situation. However, the
24-h period) can generally be accomplished quality of the HMI varies widely throughout
without the significant sacrifice of other op- the industry. Most HMI implementations
erational duties, and is considered likely to be are little more than a collection of numbers
acceptable. A rate greater than 150 per day sprinkled on a screen while showing a pip-
begins to become problematic. Up to two ing and instrumentation diagram (P&ID),
alarms per 10-minute period (~300 alarms/ making diagnosis much more difficult. For
day) are termed the “maximum manageable.” more discussion on this topic, search the
More than that may be unmanageable. Internet for the term “High-Performance
The acceptable alarm rates for small peri- HMI,” or see the comprehensive white
ods of time (such as 10 minutes or one hour) paper cited in Refs. 1 and 2.
depend on the specific nature of the alarm, As a result, the diagnosis and response
rather than the raw count. The nature of the to a simple high-tank-level alarm becomes
response varies greatly in terms of the de- quite complicated. Given the tasks involved,
mand upon the operator’s time. The duration it might only be possible to handle a few
of time required for an operator to handle an such alarms in an hour.
alarm depends upon the particular alarm. Other alarms are simpler, such as, “Pump
As an example, consider a simple tank with 412 should be running but has stopped.”
three inputs and three outputs. The tank’s The needed action is very direct: “Restart the
high-level alarm occurs. Consider all of the pump, or if it won’t restart, start the spare.”
possible factors causing the alarm and what Operators can handle several such alarms
the operator has to determine: as these in 10 minutes. It takes less time to
• Too much flow on inlet stream A, or B or C assess and work through the situation.
• Too much combined flow on streams Response to alarm rates of 10 alarms per
A-B, A-C, B-C or A-B-C 10 minutes (the threshold of a “flood”) can
• Not enough flow on outlet stream D, E or F possibly be achieved for short periods of time
• Not enough combined flow on streams — but only if the alarms are simple ones. And
D-E, D-F, E-F or D-E-F this does not mean such a rate can be sus-
• Several more additional combinations of tained for many 10-minute periods in a row.
the above inlet and outlet possibilities. During flood periods (Figure 3), operators are
The above situation takes quite a while likely to miss important alarms. Alarm rates
to diagnose, and involves observing trends per 10 minutes into the hundreds or more,
of all of these flows and comparing them to lasting for hours, are common. What are the
the proper numbers for the current process odds that the operator will detect the most
situation. The correct action varies highly important alarms in such a flood? Alarm

Chemical Engineering www.chemengonline.com march 2016 57


Figure 3. During alarm flood Annunciated alarms per 10 minutes Alarm floods – alarm count
periods, it is very likely that
operators will miss important 160 Highest 10-min. rate 1,000
alarms = 144 820 separate
900 floods
140 Alarm flood = 10 or
more in 10 min. Several Highest count
800 peaks in an alarm
120 above flood = 2,771
700 1,000 Longest
duration of
100 600 flood = 19 h
500
80
400
60
300
40 200

100
20
0
8 weeks
0
8 weeks
ing), fleeting (occurring and clearing in very
floods can make a difficult process situation short intervals), stale, duplicate and so forth.
much worse, and are often the precursors to Alarms with such behaviors are called “bad
major upsets or accidents. actors.” The most common cause of high
alarm rates is the misconfiguration of spe-
Averages can be misleading cific alarms, resulting in unnecessarily high
Alarm performance should generally be alarm occurrence rates. Commonly, 60–80%
viewed graphically rather than as a set of of the total alarm occurrences on a system
averages. Imagine that during one week, come from only 10–30 specific alarms. Chat-
your alarm system averaged 138 alarms per tering alarms and fleeting alarms are both
day and an average 10-minute alarm rate of common. Simply ranking the frequency of
0.96. That would seem to be well within the alarms will identify the culprits. Finding and
bounds of acceptability. But the data pro- correcting these rate-related nuisance be-
ducing those average numbers could look haviors will significantly reduce alarm rates
like that shown in Figure 4. with minimal effort.
The first flood lasted 40 minutes with 118 In the example data shown in Figure 6,
alarms. The second flood lasted 30 minutes 76% of all alarm occurrences came from
with 134 alarms. How many of those alarms only 10 individual configured alarms. In fact,
were likely to be missed? A simplistic answer the top two alarms make up 50% of the
(but good enough for this illustrative purpose) total load, with about 48,000 instances in
is to count the alarms that exceed 10 within 30 days. Alarms are never intentionally de-
any 10-minute period for the duration of each signed to annunciate so frequently, but they
flood, which, for the current example, would do. In this configuration, they would not per-
be a total of 182. In other words, despite form a useful function; rather, they would be
these seemingly great averages (many plant annoying distractions.
managers would consider these averages to Many of these were chattering alarms. In
be strong alarm-system performance and that summarizing 15 alarm-improvement projects
they would be happy to achieve), the alarm at power plants, the author’s employer found
pattern still puts the operators in the position that 52% of all alarm occurrences were as-
of likely missing almost 200 alarms. Missing sociated with chattering alarms. Proper ap-
so many alarms can result in improper opera- plication of alarm deadband and alarm on-
tor actions and undesirable consequences delay/off-delay time settings usually corrects
— perhaps quite significant ones. the chattering behavior. The calculations for
It is easy to plot such data, as in Figure 5. determining those settings are straightfor-
During an eight-week period, almost 21,000 ward (but beyond the scope of this article).
alarms were likely to be missed. A weekly Much more detailed information for solving
view of such data in this way will likely gain the all types of nuisance alarm problems can be
attention of management, whereas viewing found in Ref. 3.
the overall averages alone would indicate that
things are satisfactory when they are not. Alarm rationalization
The other cause of high alarm rates requires
Bad actor alarm reduction more effort to address. Most alarm systems
Many types of nuisance alarm behaviors are initially configured without the benefit of
exist, including chattering (rapidly repeat- a comprehensive “alarm philosophy” docu-
58 Chemical Engineering www.chemengonline.com march 2016
ment. This document sets out the rules Annunciated alarms per 10 minutes Figure 4. Different alarm data can gen-
70 erate similar average alarm rates, and the
for determining what kinds of situa- average rate may not tell the full story
tions qualify for alarm implementation. 60
Flood of
118 alarms Flood of
It specifies methods for consistently de- over 40 134 alarms
min. over 30
termining alarm priority, controlling alarm 50
min.
suppression, ongoing performance 40
analysis, management of change, and
dozens of other essential alarm-related 30
topics.
20
Systems created without such a doc-
ument are usually inconsistent collec- 10
tions of both “true alarms,” along with
many other items, such as normal sta- 0
7 days
tus notifications that should not use the
alarm system. Such non-alarms dimin- neously as an initial improvement effort
ish the overall effectiveness of the sys- with fast, high-impact results.
tem and diminish the operator’s trust in Step 1: Develop, adopt and maintain
it. They must be purged. While it may an alarm philosophy. A comprehensive
be easy to spot things that clearly have guideline for the development, imple-
no justification for being alarms by look- mentation and modification of alarms,
ing at the list of most frequent alarms, an alarm philosophy establishes basic
a comprehensive alarm rationalization is principles for a properly functioning
needed to ensure the consistency of the alarm system. It provides an optimum
overall alarm system. basis for alarm selection, priority set-
With alarm rationalization, every exist- ting, configuration, response, handling
ing alarm is compared to the principles methods, system monitoring and many
in the alarm philosophy document and other topics.
is either kept, modified or deleted. Set- Step 2: Collect data and benchmark
points or logical conditions are verified. the alarm system. Measuring the exist-
Priority is assigned consistently. New ing system against known, best-practice
alarms will be added, but the usual performance indicators identifies spe-
outcome of rationalization is a reduc- cific deficiencies, such as various types
tion in configured alarms by 50–75%. of nuisance alarms, uncontrolled sup-
Since the alarm-management prob- pression, and management-of-change
lem was identified in the early 1990s, issues. A baseline is established for im-
thousands of alarm systems have un- provements measurement.
dergone this process and achieved the Step 3: Perform “bad actor” alarm
desired performance. resolution. Addressing a few specific
After the bad actor reduction and the alarms can substantially improve an
rationalization steps, alarm rates are alarm system. Bad actor alarms, which
usually within the target limits. A typical can render an alarm system ineffective,
result is shown in Figure 7. Significant are identified and corrected to be con-
process upsets, particularly equipment sistent with the alarm philosophy. An
trips, may still produce some alarm ongoing program to identify and resolve
floods, which can be addressed in Step nuisance alarms is necessary.
6 listed below. Step 4: Perform alarm rationalization.
The 2009 publication of the ISA-18.2 Alarm rationalization is a comprehensive
Alarm Management Standard includes review of the alarm system to ensure it
both having an alarm philosophy docu- complies with the principles in the alarm
ment and performing alarm rationaliza- philosophy. This team-based effort re-
tion as mandatory items. For a compre- examines existing and potential alarms
hensive white paper on understanding configured on a system. Alarms to be
and applying ISA-18.2, see Ref. 4. added, deleted and reconfigured are
identified, prioritized and documented.
Alarm management work process The resulting alarm system has fewer
There is an efficient seven-step plan for configured alarms and is consistent and
improving an alarm system, proven in documented with meaningful priority
more than 1,000 improvement projects and setpoint values.
in plants throughout the world. Steps Step 5: Implement alarm audit and
1–3 are simple, and often done simulta- enforcement technology. Once an

Chemical Engineering www.chemengonline.com march 2016 59


Alarms per day likely to have been missed Alarms per day before and after rationalization
1200 8000
Week 1: 3,885
Week 2: 2,281
Week 3: 2,728 7000
1000 Average before: 2,417
Week 4: 1,903 Average after: 249
Week 5: 2,173 6000 An 89% reducation
Week 6: 1,443
800 Week 7: 2,253
Week 8: 4,260 5000
Total: 20,926
600 4000

3000
400

2000
200
1000

0 0
8 weeks
31 days

Figure 5. Despite sound averages for alarm rates, it can still Figure 7. Alarm rates can usually be brought into target lim-
be the case that many alarms could be missed during alarm its by alarm rationalization and bad-actor reduction steps
flood periods

alarm system is rationalized, its configura- program of system analyses that may in-
tion must not change without authoriza- clude KPI monitoring and the correction of
tion. Because DCS systems can be easily problems as they occur.
changed by a variety of sources, they often
require mechanisms that frequently audit Concluding remarks
(and enforce) the approved configuration. The various problems with alarm systems are
Step 6: Implement advanced alarm man- well recognized and there are proven solutions
agement. Certain advanced alarm capa- to these problems. The principles from these
bilities may be needed on some systems to solutions have been successfully applied to
address specific issues. For example, state- thousands of alarm systems worldwide. The
based alarming monitors the current process alarm management body of knowledge is ma-
state, and alarm settings are dynamically al- ture. Solving alarm-system problems simply
tered in predetermined ways to match the requires the will and effort to do so. n
alarming requirements of that process state. Edited by Scott Jenkins
Alarm flood suppression temporarily elimi-
nates the expected and distracting alarms References
from a unit trip, leaving the relevant alarms 1. Hollifield, B. and Perez, H. Maximize Operator Effectiveness: High
that assist the operator in managing that Performance HMI Principles and Best Practices, Part 1 of 2. PAS
Inc., Houston, 2015.
post-trip situation. Such advanced methods
2. Hollifield, B. and Perez, H. Maximize Operator Effectiveness: High
can ensure that the alarm system is effective Performance HMI Case Studies, Recommendations, and Standards,
even in abnormal situations. Part 2 of 2. PAS Inc., Houston 2015.
Step 7: Control and maintain the im- 3. Hollifield, B. and Habibi, E. The Alarm Management Handbook, 2nd
proved system. An effective alarm system Ed., PAS Inc., Houston 2010.
requires an ongoing and typically automated 4. Hollifield, B. Understanding and Applying the ANSI/ISA 18.2 Alarm
Management Standard. PAS Inc., Houston 2010.
Alarm
count Most frequent annunciated alarms Author
35,000 100 Bill Hollifield is the principal consultant at
30 days PAS Inc. (16055 Space Center Blvd., Suite
90
30,000 600, Houston, TX 77062; Phone: 281-286-
80
6565; Email: bhollifield@pas.com). He is re-
25,000 70 sponsible for alarm management and high-
Cumulative %
20,000
60 performance HMI. He is a member of the
50 ISA-18 Alarm Management committee, the
Ten alarms make
15,000 up 76% of the total ISA-101 HMI committee, and is a co-author of
40
alarm loading the Electric Power Research Institute’s (EPRI)
10,000 30 Alarm Management Guidelines. Hollifield is
20 also coauthor of the Alarm Management Handbook and The High
5,000
10
Performance HMI Handbook, along with many articles on these top-
ics. Hollifield has a dozen years of international, multi-company ex-
0 0 perience in all aspects of alarm management and effective HMI
Tag1.Alarm

Tag2Low1

Tag3.Alarm

Tag4.Alarm

Tag5.Low1

Tag6.Low1

Tag7.Alarm

Tag8.Low1

Tag9.High1

Tag10.Alarm

design consulting for PAS, coupled with 40 years overall of industry


experience focusing on project management, chemical production
and control systems. Hollifield holds a B.S.M.E. from Louisiana Tech
University and an MBA from the University of Houston. He’s a pilot
Figure 6. In many cases, the most frequently occurring and has built his own plane (with a high-performance HMI).
alarms make up the bulk of the total alarm load

60 Chemical Engineering www.chemengonline.com march 2016


Feature Report

Wireless
Communication
in Hazardous Areas
Stephan Schultz
R. Stahl Consider these criteria in deciding where
wireless fits in today’s CPI plants and the

W
ireless communications have
great potential in the chemi-
cal process industries (CPI)
explosive atmospheres that permeate them
because they do away with
complex and costly cable installations In existing plants, power sources are application for wireless technology. As
and enable completely new applica- around nearly every corner, so the cost of yet, though, most reading devices
tions. And while a recent wave of suc- of wiring for power is not nearly as used for this purpose are handheld
cessful demonstrations has begun to significant as the cost of the wiring for terminals with a cable that curtails
emerge in the CPI (for more, see CE, the control signals themselves. their operation. Portable radio devices
Nov. 2009, p. 17–23), a number of hur- A look at typical routines in process capable of both acquiring data and
dles stand in the way of a completely plants will identify the potential ancil- passing it on via wireless link to MES
wireless Utopia. In most cases, the lary application areas with a view to (manufacturing execution system)
totally reliable, uncompromised avail- how and how much they may benefit. and ERP (enterprise resource plan-
ability of a production plant remains a Once a case is made for wireless tech- ning) servers save time and costs, and
paramount objective, and it will there- nology in general for these purposes, increase data reliability due to exact
fore likely take some more time before users are faced with various solutions to and nearly instant data acquisition.
radio transmissions of critical signals choose from for actual implementations. RFID tags can be expected to increase
in control loops take root. And last but not least, there are addi- their foothold in the CPI due to reli-
One impediment often cited as a limit tional safety considerations for applica- ability and safety benefits, since one
for wireless solutions is power. In fact, tions in hazardous areas. All of these key RFID advantage over barcodes is
many process applications basically aspects will be discussed in order to en- that even smudged and stained labels
rule out wireless field devices with- able users to make informed choices, or are still legible. Also, there are other
out an independent, onboard source to at least prime themselves for further convenient features that previous so-
of power. Granted, there have been a consultations with specialist manufac- lutions could not provide; for instance,
number of promising approaches in turers or systems solution providers. data can be written to the tags more
this regard, which are based on con- than once and it is possible to acquire
sumption-optimized electronic circuits application Areas several tags at the same time.
and alternative sources of power using Logistics and supply chain
accumulators or solar cells, or on so- State-of-the-art logistics solutions Maintenance and monitoring
called energy harvesting, where energy depend on systems that acquire data Anyone in the field who is servicing
is recovered from vibration, tempera- on the flows of goods with the highest a plant is likely to benefit from using
ture fluctuations, and so on. possible degree of precision, and pref- portable devices with a connection to
At the same time, there are a range erably at the very instant when stock a central management system, since
of ancillary functions in almost any items are taken out or replenished. doing so enables optimization of typi-
plant today for which wireless com- In the CPI, many raw materials and cal routines and measurements. For
munications truly are already a boon. products are transported in containers example, maintenance instructions
In these cases, power is not an insur- such as drums, tanks, intermediated can be automatically dispatched be-
mountable hurdle because the power bulk containers (IBCs), and so on. Most cause all relevant information can be
requirements are low enough to main- containers are marked with either provided via radio to a portable hand-
tain battery life of five or more years. barcodes or RFID (radio-frequency held device that service engineers can
Meanwhile, the use of wired power identification) tags. Acquiring RFID carry with them in the field. Staff are
should not be ruled out automatically. tag information is an obvious model then able to inspect equipment as
Chemical Engineering www.che.com may 2010 39
Feature Report

needed and, upon completion or even Networking Options


while they are taking care of a main- Whatever functions are to be enabled,
tenance task, enter the results of the all wireless network installations re-
inspection, or repairs made, directly quire thorough planning, which starts
into the portable device. Those data with the definition of the requirements
are then instantly available in a cen- for the wireless network. A range of
tral database and can be utilized, for aspects has to be considered, includ-
instance, for documentation purposes ing bandwidth, mobility, hardware re-
or even to speed up billing. Similar quirements in terms of realtime signal
advantages apply to operating and transmission, the encryption system,
monitoring tasks in industrial plants. information-technology (IT) depart-
Portable devices make it possible to ment demands and so on.
Figure 1.
read realtime measured values and 1. Using a floor plan, it is possible, in State-of-the-
therefore keep an eye on the actual principle, to assess the radio frequency art Wireless
state of the production plant onsite. (RF) coverage in the area with the aid LAN systems
At the same time, operators in the of planning programs. However, prac- ensure secure data
handling, unlike
field have access to ancillary informa- tical experience shows that the effort earlier versions, which
tion such as maintenance schedules, to emulate the complete structure of were easier to crack
operating instructions, ATEX or other a CPI plant is too high. Experience is
hazardous area certificates and much the key to success. Meanwhile, users Most users will therefore have to ex-
more. As a result, routine procedures deploying a new network also have to amine at least some of the following
can be modified to become consider- know exactly which wireless systems options to assess whether they fit the
ably more efficient. are already in use in the same place application at hand.
and in neighboring areas. The location
Security and asset management and selection of the antennas can then Wireless LAN
Using radio transmission, camera sys- be established. All radio technologies currently avail-
tems or sensors at distant measuring 2. In the next step, the deployment able on the market have specific ad-
points — for instance, within pump plan should be verified against a so- vantages and disadvantages. It is
stations — can be integrated into the called onsite survey. This is a live on- worth noting, however, that the most
site’s human-machine-interface (HMI) site inspection of the area to check widely used solutions have originated
concept at a low cost and can be read- the values previously determined on in the office IT sector and were not
ily displayed where needed. While pro- the computer in the real environment, genuine developments for industrial
cess signals in the narrow sense are using a portable access point. In this applications. Wireless LAN (local area
absolutely needed to ensure proper confirmation process, some additional network) is the most prominent case
control of a plant, a host of other mea- information can be gathered that in point. In an industrial environment,
sured values may be useful only for cannot be anticipated in a floor plan, Wireless LAN (WLAN) is quite suit-
operative improvements or preven- such as the effects of vehicles passing able for use with portable equipment,
tive maintenance. Radio transmission through, or of mobile containers that such as barcode scanners or handheld
is a good alternative for such signals may have appeared in the way in un- HMI devices. It provides the great-
not only if they are particularly hard expected places. The survey will also est bandwidth (for IEEE 802.11b, 11
to acquire any other way, but, more allow users to realistically determine Mbit/s, or IEEE 802.11g, 54 Mbit/s
broadly, for all kinds of non-critical the effective bandwidth in the central gross data throughput) and is de-
asset management data. and the outer areas of RF coverage. signed for the transmission of Ether-
For the time being, HART communi- 3. Finally, the RF system can be in- net-based protocols. It is important to
cation is most commonly implemented stalled, commissioned, and put through keep in mind, though, that most CPI
to transmit signals that are only used a final test under real operating condi- applications only require bandwidth
for process optimization and similar tions to avoid unpleasant surprises. in the 100–500 kbit/s range.
purposes. Wireless solutions are well While the many steps of this proce- In a WLAN network, an access cli-
suited to satisfy the growing demand dure might appear to drive expenses ent, such as a PDA (personal digital
for higher-level asset management. up, they have indeed proven to be by assistant), can also roam from one
There is good cause for its increasing far the most reliable way to ensure access point to another without any
importance in the process industry: that a new wireless system really interruption in transmission. This
live information about the current works as expected and brings about means users carrying portable devices
state of production equipment in a the desired process improvements. can move freely around the site with-
plant in as much detail as possible Obviously, wireless communications out losing their connections to the net-
gives staff a better means to antici- can be implemented using a variety work. State-of-the-art WLAN systems
pate imminent plant failures and to of different radio technologies. As is also ensure secure data handling, un-
adjust maintenance intervals to ac- so often the case, there is no one stan- like earlier versions, which only relied
tual needs. dard that meets all requirements. on Wired Equipment Privacy (WEP),
40 Chemical Engineering www.che.com may 2010
bandwidth of only 50 kbit/s, GPRS is gether to find a way to merge both
also considerably slower than WLAN standards or enable interoperability.
and other radio protocols. This would be important to tearing
down the last obstacle for the success
Figure 2. This WirelessHART gateway Bluetooth of wireless technology in the CPI.
is suitable for Zone 2 areas Bluetooth does not provide a band-
width that can match WLAN network Coexistence
WLAN’s original, out-dated encryp- performance, but recent systems do While the industrial-scientific-medical
tion method that was very easy to achieve transmission rates of up to (ISM) frequencies (the radio bands at
break by brute force. 2 Mbit/s. In addition, due to its syn- 2.4 GHz used by most common wire-
chronous communication modes, Blu- less solutions) are licence-free and
GPRS on public GSM networks etooth provides a very good basis for therefore help to reduce operating
The General Packet Radio Service realtime applications. One key Blu- costs, they do have the disadvantage
(GPRS) enables the transfer of data etooth feature is the frequency hop- that they must be shared by differ-
packets on the public radio networks ping spread spectrum (FHSS) scheme, ent applications. The standardization
that were originally built for cell- which makes this technology signifi- forums are aware of this fact and
phone voice communications, and cantly less susceptible to interference have come forth with some adequate
have since been enhanced for other than WLAN. FHSS also provides some approaches for resolving potentially
data transfer. GPRS is, for instance, additional protection against eaves- problematic side effects. For instance,
the base for the popular Blackberry droppers. Bluetooth works well for Bluetooth’s adaptive frequency hop-
technology. In CPI applications, the networks with up to eight users, while ping scheme enables an operation of
service can be used for remote main- greater numbers will require increased WLAN and Bluetooth networks at the
tenance and remote monitoring func- technical efforts. Bluetooth radio con- same time in the same environment.
tions in pumping stations, remote sumes less power in operation than WirelessHART and ISA 100 enable
tank farms, centrifuges, compressors WLAN. Due to its characteristics, it a so-called blacklisting of channels,
and other machines. Unlike WLAN, is particularly suitable for integrating which are occupied by other wireless
GPRS operates on a licensed frequency fixed devices such as HMI stations or applications. Given thorough and sen-
range, which means that less interfer- sensors. Like WLAN, Bluetooth boasts sible wireless network planning and
ence occurs in the radio connection specifications that have been interna- deployment, interference can be prac-
than in the frequency bands used by tionally agreed upon, which ensures tically eliminated in most scenarios.
most other established wireless-data- that devices from different manufac- Besides the more or less established
communications standards. Since it is turers are fully, or at least to a great standards just discussed, there are
based on the existing, fully developed extent, compatible with each other. numerous other proprietary protocols.
Global System for Mobile Communi- However, users will more often than
cations (GSM) mobile networks, GPRS WirelessHART and ISA 100.11a not be inconvenienced by them due to
requires no extra investments for a WirelessHART and ISA 100.11a are incompatibilities between devices from
purpose-built, self-run radio system. standards that are dedicated for sen- different vendors. Based on the exist-
On the contrary, GPRS connections sor networks in the CPI. Both stan- ing standards for WLAN, Bluetooth
constitute communication routes that dards promise to connect field devices WirelessHART and ISA 100 technol-
are totally independent from a compa- of various vendors onto one network. ogy, various committees and organiza-
ny’s own, existing IT infrastructure. The network structure could be point- tions in several countries have been
Also, the technology can be used for to-point, star or — the most inter- trying and keep trying to improve
additional services, such as alerting esting way — meshed. The meshed standardization and provide users and
responsible staff via text message or structure offers two advantages. First manufacturers with implementation
Email in case of a malfunction. of all, if a field device is installed out guidelines. Major protagonists include
Some restrictions and weak points of the range for a direct connection to the German VDI/VDE GMA work-
must be taken into account, however. the gateway, it may use a neighbor- ing committee 5.21, the ZVEI, and a
GPRS is, for instance, not yet univer- ing field device as a repeater. This Namur subcommittee. Contributions
sally available worldwide. In some method extends the communication in this field also come from organiza-
countries, such as Japan and Korea, range. Secondly, the meshed structure tions such as the ISA’s (Instrumenta-
GPRS coverage will remain unavail- enables a self-healing of the network tion, Systems and Automation Society
able, since these countries’ mobile in case of interruptions, which could of America) SP100 committee and the
radio networks do not use the GSM happen, for instance due to delivery or HCF’s (HART Communication Foun-
standard. A more extensive coverage service trucks parking in front of a de- dation) WirelessHART.
and better bandwidth will only be vice. The first field trials have proven
achieved when the followup technology that the technology and components Hazardous areas
to GPRS, the so-called UMTS service, are ripe for industrial use. Radio devices emit electromagnetic ra-
becomes well established around the Both WirelessHART and ISA diation that is clearly a possible source
world. Last but not least, with a net 100.11a committees are working to- of ignition in an explosive atmosphere.
Chemical Engineering www.che.com may 2010 41
Feature Report

The main risk lies in the induction of less relevant. IEEE studies on elec-
electrical currents in metallic objects tromagnetic radiation in hazardous
or electronic circuits that are inade- areas have shown that even RF with
quately protected from electromagnetic power of 6 W can become a potential
interference (EMI). These currents hazard in terms of induction in metal
can result in excessively high tem- objects. Because of this danger, the IEC
peratures and the formation of sparks. 60079-0 2008 and the upcoming EN
Other dangers, such as direct ignition 60079-0 for continuous high frequency Figure 3. An external antenna in wire-
less units, such as this access point, is cur-
of an explosive atmosphere, are much sources limit the maximum permitted rently required to attain an individual ATEX
certification for use in hazardous areas

transmitting power in wireless net-

Mission: Immersion. works that are operated in hazardous


areas. The location of a wireless node
in Zone 0, 1 or 2 can be disregarded and
Immersion Engineering™™ has no relevance for the limit, since an
goes deep to solve your RF signal will obviously not stop at the
boundary between two zones.
heat transfer problems.
Safe emission levels
The threshold is set to a value be-
tween 6 and 2 W emitted power, with
the lower end applying to atmospheres
with group IIC explosive gases, such
as hydrogen or acetylene. WLAN, Blu-
etooth, WirelessHART and ISA 100
all predominantly use the aforemen-
tioned ISM bands at 2.4 GHz, which
are restricted to low-power radio
transmissions anyway. More specifi-
cally, WLAN access points using this
band are limited by RF regulations to
no more than 100 mW. Fortunately,
Bluetooth, WirelessHART and ISA
100 transmissions typically require
only about 10 mW of energy in the
first place. At face value, all of these
technologies therefore need or can do
with significantly less energy than the
Even though you may call us on the performance and applications of heat maximum allowed by the standard.
phone miles away, we're so deep into transfer fluids than we do. However, the so-called antenna gain
your stuff--your fluid, your equipment, So pick a service and call one of our must also be factored into the calcula-
your system--we can virtually touch it, technical specialists. Or, check out our tion, as the ignition risk is also defined
see it. web site for case histories, data sheets, by the magnitude of the field strength.
Immersion Engineering is a bundle of comparisons, user’s guide, tip sheets Antenna gain is a parameter that de-
very specialized services that you can and technical reports. It’s all there, it’s scribes the concentration of radio en-
cherry pick. Some are free, some you deep, it’s Immersion Engineering.
ergy emitted in a specific direction.
pay for. We’re the only company offering
them all.
Such directional gain increases as radio
One thing is for sure; when you need emissions in other directions decrease
HTF help you need it now. Nobody because the total energy emitted re-
knows more about the chemistry,
HEAT TRANSFER FLUIDS
mains the same. Antenna gain is mea-
4 Portland Road sured in relation to a specific reference.
West Conshohocken PA 19428 USA
Eyeball this selection of services. If the gain value is stated in dBi units,
� Fluid Analysis � Troubleshooting
800-222-3611 ®
then it refers to an isotropic radiator,
610-941-4900 • Fax: 610-941-9191
� Fluid Maintenance � Consulting or omnidirectional radiator (the theo-
info@paratherm.com
� Training retical model of an antenna that evenly
www.paratherm.com
distributes energy in all directions
Stocking & Sales Locations: USA • Canada • Mexico • Brazil • Argentina • Guatemala • Netherlands • Belgium • Denmark
• United Kingdom • Australia • China • Japan • Thailand from a point source). Typical values for
rod antennas and directional antennas
Circle 34 on p. 70 or go to adlinks.che.com/29250-34
42 Chemical Engineering www.che.com may 2010
P2008A
1/2 Page
are between 5 and 9 dBi. Users have to implement Ex ib (intrinsically safe) Author
take antenna gain into account when type protection, which would allow for Stephan Schultz is senior
product manager automa-
they refer to the values given in the communication via an intrinsically tion, isolator and wireless
tables in IEC 60079-0. safe HF signal. Such solutions are at R. Stahl (Am Bahnhof 30,
74638 Waldenburg, Germany;
currently in development. Once they Phone: +49 7942-943-4300;
Suitable Zone 1 device designs actually become available, users will Fax: +49 7942-943-404300;
Email: stephan.schultz@stahl.
With few exceptions, automation com- finally have access to the full range of de; Website: www.stahl.de).
ponents and devices currently avail- standard antennas. ■
able on the market must not be used Edited by Rebekkah Marshall
in Zone 1* right out of the box. This re-
striction is largely a consequence of the 5297-Convey 7.562x10.5 10/3/07 3:55 PM Page 1

rapid pace of development for new de-


vices, which are released in very short
intervals and are therefore often af-

The Smartest Distance


fected by incomplete standardization.
One possible solution to the problem
is an installation of such RF equip-
ment without Zone 1 approval in hous-
ings featuring a flameproof enclosure.
This includes Ex d type of protection,
or another suitable type. The major-
ity of all Ex d enclosures are made of
Between Two Points.
metal, which shields electromagnetic
Pneumatic Conveying Systems
radiation from the antenna as a side
effect. Obviously, not just any antenna from VAC-U-MAX.
can be installed inside a housing of VAC-U-MAX is a premier manufacturer of custom pneumatic
this type without additional measures. systems and support equipment for conveying, batching,
In some cases, a housing with a glass and weighing materials. With a VAC-U-MAX system on
site, your company’s product can move gently and
pane can be used in combination with quickly from point to point, with nothing in the way to
a directional antenna installed within. impede the efficiency of its movement.
However, tests have shown that only
Count on us for:
antennas specially matched to a par-
ticular type of flameproof enclosure • Decades of engineering and conveying expertise.

will actually work well, since the sig- • Customized solutions that
nal loss is otherwise excessive. meet your specific needs.
Because our systems are
Another possible option is the use of not “off the shelf,” they
external antennas. However, hazard- are always on the mark.
ous area requirements demand that • Reliable equipment that’s
special explosion-protected anten- proudly made in America.
nas have to be installed in this case. • Our Airtight Performance Guarantee™. We stand
They usually have to be designed for behind every system we engineer. And we say it
increased safety (Ex e) protection, be- in writing.
cause, in the event of a short circuit For more information about our custom-engineered
between the power supply and the pneumatic systems and solutions, call:
output or input stage in the RF device,
no excessively high currents or volt- 1-866-239-6936
ages are allowed to coincide with the or visit online at:

explosive atmosphere without protec- www.vac-u-max.com/convey


tion. Zone 1 GPRS modems typically
have GSM antennas connected via
an Ex i interface, and also feature an Air-driven solutions.
intrinsically safe Ex i SIM card slot.
One way to do away with most limita-
tions concerning the choice of antenna Belleville, New Jersey
would be antenna breakouts for de- convey@ vac-u-max.com

vices in encapsulated housings that

* For a one-page reference card on hazardous


area classifications, see http://www.che.com/
download/facts/CHE_0507_Facts.pdf
Circle 41 on p. 70 or go to adlinks.che.com/29250-41
Chemical Engineering www.che.com may 2010 43
Feature
Cover Story
Report

Piping-System Leak Detection


and Monitoring for the CPI
Eliminating the potential for leaks is an integral part ACRONYMS

of the design process that takes place at the very



AVO = Audio/visual/olfactory
CAA = Clean Air Act
onset of facility design

HAP = Hazardous air pollutants
HON = Hazardous organic NESHAP
W. M. (Bill) Huitt When discussing anything to do LDAR = Leak detection and repair
W.M. Huitt Co. with government regulations, the LUST = Leaking underground
terminology quickly turns into an storage tank

L
eaks in a chemical process in- “alphabet soup” of acronyms. The NEIC = National Enforcement
dustries (CPI) facility can run box on the right lists, for easy refer- Investigations Center
the gamut from creating a ence, the titles and acronyms that
NESHAP = National Emission
costly waste to prefacing a cat- will be used in this discussion.
Standard for Hazardous Air
astrophic failure.. They can be an Pollutants
annoyance, by creating pools of liq- Leak mechanisms
uid on concrete that can become a Eliminating the potential for leaks NSPS = New Source Performance
Standards
possible slipping hazard and house- is an integral part of the design
keeping problem, or a leak that can process that takes place at the very RCRA = Resource Conservation and
emit toxic vapors, causing various onset of facility design. It is woven Recovery Act
degrees of harm to personnel. In into the basic precept of the piping SOCMI = Synthetic organic chemical
some cases a leak may be a simple codes because it is such an elemen- manufacturing industry
housekeeping issue that goes into tal and essential component in the TSDF = Treatment, storage and
the books as a footnote indicating process of designing a safe and de- disposal facilities
that a repair should be made when pendable piping system. UST = Underground storage tank
resources are available. In other Piping systems, as referred to
cases it can become a violation of here, include pipe, valves and other VOC = Volatile organic
compounds
regulatory compliance with statu- inline components, as well as the
tory consequences, not to mention equipment needed to hold, move and
a risk to personnel safety and the process chemicals. Why then, if we (CA) is used as an applied factor
possible loss of capital assets. comply with codes and standards, in calculating, among other things,
Understanding the mechanisms and adhere to recommended indus- wall thickness in pipe and pressure
by which leaks can occur and priori- try practices, do we have to concern vessels. The CA value assigned to
tizing piping systems to be checked ourselves with leaks? Quite point- a material is theoretical and predi-
at specific intervals based on a few edly it is because much of what we cated on four essential variables:
simple factors is not only a prag- do in design is theoretical, such as material compatibility with the
matic approach to the preventive material selection for compatibility, fluid, containment pressure, tem-
maintenance of piping systems, but and because in reality, in-process perature of the fluid and velocity
is part of a CPI’s regulatory com- conditions and circumstances do of the fluid. What the determina-
pliance. This includes compliance not always perform as expected. tion of a CA provides, given those
under both the U.S. Environmen- Whether due to human error or variables, is a reasonable guess at
tal Protection Agency (EPA) Clean mechanical deficiencies, leaks are a uniform rate of corrosion. And
Air Act (CAA; 40CFR Parts 50 to a mechanism by which a contained given that, an anticipated loss of
52) and the Resource Conservation fluid finds a point of least resistance material can be assumed over the
and Recovery Act (RCRA; 40CFR and, given time and circumstances, theoretical lifecycle of a pipeline
Parts 260 to 299). We will get into breaches its containment. What we or vessel. It allows a reasonable
more detail with these regulations, look into, somewhat briefly, are two amount of material to be added into
as well as the leak detection and general means by which leaks can the equation, along with mechani-
repair (LDAR) requirement within occur; namely corrosion and me- cal allowances and a mill tolerance
the above mentioned regulations, as chanical joint deficiencies. in performing wall thickness cal-
we move through this discussion. Corrosion. Corrosion allowance culations. The problem is that be-
44 Chemical Engineering www.che.com May 2014
Table 1. Elements of a Model LDAR Program Pump, compressor and agitator
seals can develop leaks where shaft
Written LDAR compliance First attempt at repair
misalignment plays a part. If the
Training Delay of repair compliance assurance shaft is not installed within recom-
LDAR audits Electronic monitoring and storage of data mended tolerances or if it becomes
misaligned over time there is a
Contractor accountability QA/QC of LDAR data good possibility the seal will begin
Internal leak definitions Calibration/calibration drift assessment to fail.
Less frequent monitoring Records maintenance The LDAR program
Promulgated in 1970 and amended
yond the design, engineering, and areas for abnormal loss of wall in 1977 and 1990, the Clean Air
construction phase of building a thickness, hydrogen stress-corro- Act requires that manufactur-
facility, the in-service reality of cor- sion cracking (HSCC), and others. ers producing or handling VOCs
rosion can be very different. The LDAR program does not develop and maintain an LDAR
Corrosion, in the majority of specify the need to check anything program in accordance with the
cases, does not occur in a uniform other than mechanical joints for po- requirements set forth under the
manner. It will most frequently tential leaks. Monitoring pipe and Clean Air Act. This program moni-
occur in localized areas in the form vessel walls, particularly at welds tors and documents leaks of VOCs
of pits, as erosion at high-impinge- that come in contact with corrosive in accordance with Method 21 —
ment areas, as corrosion under chemicals, is a safety consideration Determination of Volatile Organic
insulation, at heat-affected zones and practical economics. Perform- Compound Leaks.
(HAZ) where welding was improp- ing cursory examinations for such Table 1 provides a listing of key
erly performed, causing a localized points of corrosion where the po- elements that should be contained
change to the mechanical or chemi- tential exists should be made part in an LDAR program. Those ele-
cal properties of the material, and of any quality assurance or quality ments are described as follows:
in many other instances in which control (QA/QC) and preventive Written LDAR compliance. Com-
unforeseen circumstances create maintenance program. pile a written procedure declaring
the potential for corrosion and the Mechanical joints and open- and defining regulatory require-
opportunity for leaks in the pipe ended pipe. Mechanical joints ments that pertain to your specific
itself or in a vessel wall. Because can include such joining methods facility. This should include record-
of that incongruity, corrosion is an as flanges, unions, threaded joints, keeping certifications; monitoring
anomaly that, in reality, cannot valve bonnets, stem seals and clamp and repair procedures; name, title,
wholly be predicted. assemblies. It can also include and work description of each person-
Corrosion-rate values found in pump, compressor and agitator nel assignment on the LDAR team;
various published resources on the seals. Other potential points of tran- required procedures for compiling
topic of material compatibility are sient emissions include open-ended test data; and a listing of all process
based on static testing in which a piping, such as drains, vents, and units subject to federal, state and
material coupon is typically set in the discharge pipe from a pressure- local LDAR regulations.
a vile containing a corrosive chemi- relief device. Any of these joints or Training. Assigned members of
cal. This can be done at varying interfaces can be considered poten- the LDAR team should have some
temperatures and in varying con- tial leak points and require both experience base that includes work
centrations. After a period of time, monitoring and record-keeping doc- performed in or around the types of
the coupon is pulled and the rate umentation in compliance with the piping systems they will be testing
of corrosion is assessed. That is a EPA’s LDAR program. and monitoring under the LDAR
simplification of the process, but Mechanical joints can leak due to program. Their training should in-
you get the point. When a material improper assembly, insufficient or clude familiarization with Method
of construction (MOC) and a po- unequal load on all bolts, improp- 21 and also training as to the cor-
tentially corrosive chemical come erly selected gasket type, sufficient rect procedure for how to examine
together in operational conditions, pressure or temperature swings the various interface connections
the theoretical foundation upon that can cause bolts to exceed their they will be testing. They should
which the material selection was elastic range (diminishing their also receive training on the test
based becomes an ongoing realtime compressive load on the joint), and instrument they will be using and
assessment. This means that due an improperly performed “hot-bolt- how to enter the test data in the
diligence needs to be paid to exam- ing” procedure in which in-service proper manner. All of this needs to
ining areas of particular concern, bolts are replaced while the pipeline be described in the procedure.
depending on operating conditions, remains in service. “Hot bolting” is LDAR audits. An internal audit
such as circumferential pipe welds not a recommended procedure, but team should be established to en-
for cracking, high-impingement is nonetheless done on occasion. sure that the program is being car-
Chemical Engineering www.che.com May 2014 45
Cover Story

ried out on a routine basis in an ef- ity has consistently demonstrated monitoring equipment should be
ficient and comprehensive manner good performance under monthly made at the end of each monitor-
in accordance with the written pro- testing, then the frequency of test- ing work shift using approximately
cedures. A third-party audit team is ing could be adjusted to a quarterly 500 ppm of calibration gas. If, after
brought in every few years to con- test frequency. the initial calibration, drift assess-
firm that internal audits are being First attempt at repair. Upon de- ment shows a negative drift of more
carried out in the proper manner tection of a leak, most rules will re- than 10% from the previous cali-
and that all equipment that should quire that a first attempt be made bration, all components that were
be included in the monitoring is to repair the leak within five days tested since the last calibration
listed as such. It also ensures that of detection; if unsuccessful, any fol- with a reading greater than 100
the tests are being carried out prop- low-up attempts need to be finalized ppm should be re-tested. Re-test all
erly and that the test results are within 15 days. Should the repair pumps that were tested since the
entered properly. remain unsuccessful within the 15- last calibration having a reading of
Contractor accountability. day time period, the leak must be greater than 500 ppm.
When selecting an outside con- placed on a “delay of repair” list and Records maintenance. Internal
tractor to perform internal LDAR a notation must be made for repair electronic record-keeping and re-
audits for a facility or when bring- or component replacement during porting is an essential component to
ing in an outside contractor to in- the next shutdown of which the a well-implemented LDAR program.
spect the work of the internal audit leaking component is a part. It is an indication to the NEIC that
team, it is recommended that the Delay of repair compliance as- every effort is being made to comply
contract be written in a manner surance. Placing a repair item on with the regulations pertinent to a
that places appropriate responsi- the “delay of repair” list gives assur- facility. It provides ready access to
bility on that contractor. In doing ances that the item justifiably be- the personnel associated with the
so there should be penalties de- longs on the list, that a plan exists program, the test data, leak repair
scribed and assessed as a result to repair the item, and that parts reports and so on.
of insufficient performance or in- are on hand to rectify the problem.
accurate documentation of pre- It is suggested that any item being Testing for leaks
scribed testing and documentation listed in the “delay of repair” list au- Results, when using a leak detec-
procedures. Expectations should tomatically generate a work order tion monitor, are only as accurate
be well defined and any deviation to perform the repair. as its calibration and the manner in
from those prescribed norms by a Electronic monitoring and stor- which it is used. Calibration is dis-
third-party contractor should con- age of data. Entering leak-test cussed in the next section, “Method
stitute a breach of contract. In all data into an electronic database 21.” To use the monitor correctly, the
fairness, both parties must under- system will help in retrieving such auditor will need to place the nozzle
stand exactley what those expecta- data and in utilizing them in ways or end of the probe as close as pos-
tions are. that help provide reports highlight- sible to the flange, threaded joint, or
Internal leak definitions. Inter- ing areas of greater concern to areas seal interface as follows:
nal leak definitions are the maxi- of lesser concern. Such information • In the case of a flange joint test:
mum parts per million, by volume can help direct attention and re- 180 deg around perimeter of the
(ppmv) limits acceptable for valves, sources away from areas of least flange joint at their interface
connectors and seals, as defined by concern, while mobilizing resources • In the case of a threaded joint test:
the CAA regulation governing a fa- to areas of greater concern. This en- 180 deg around perimeter of inter-
cility. For example, a facility may be ables a much more efficient use of face of the male/female fit-up
required to set an internal leak-def- information and resources. • If it is a coupling threaded at both
inition limit of 500 ppm for valves QA/QC of LDAR data. A well ends, check both ends 180 deg
and connectors in light liquid or gas/ written LDAR program will include around the perimeter
vapor fluid service and 2,000 ppm a QA/QC procedure defining the • If it is a threaded union, then
internal leak definition for pumps process by which it is assured that check both ends and the body nut
in light liquid or gas/vapor fluid Method 21 is being adhered to, and 180 deg around the perimeter
service. “Light liquid” is defined that testing is being carried out in • In the case of a valve test:
as a fluid whose vapor pressure is the proper manner and includes the 180 deg around perimeter of
greater than 0.044 psia at 68°F. proper equipment and components. all end connections if anything
Less frequent monitoring. Under This also includes the maintenance other than welded
some regulations it is allowed that of proper documentation. 180 deg around perimeter of
a longer period between testing is Calibration/calibration-drift body flange
acceptable if a facility has consis- assessment. LDAR monitoring 180 deg around perimeter of
tently demonstrated good perfor- equipment should be calibrated in body/bonnet interface
mance (as defined in the applicable accordance with Method 21. Cali- 180 deg around perimeter of
regulation). For example, if a facil- bration-drift assessment of LDAR stem packing at stem
46 Chemical Engineering www.che.com May 2014
National Cleanup Backlog
(confirmed releases, cleanups completed) 160,000

140,000

120,000 by ionizing a sample and then mea-


National backlog

suring the charge (that is, number


100,000
of ions) produced.
80,000 Two methods of ionization cur-
rently used are flame ionization
60,000 and photoionization. The flame ion-
129,828
142,709

136,265

102,798
191,242

113,919

108,766

100,165

93,123

87,983
40,000 ization detector (FID) theoretically
measures the total carbon content
20,000 of the organic vapor sampled. The
photoionization detector (PID)
0
2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 uses ultraviolet light to ionize the
organic vapors. With both detec-
Fiscal year
tors, the response will vary with
Figure 1. Progress is slowly being made to clean up leaking underground storage the functional group in the organic
tanks under the RCRA program compounds. PIDs have been used to
detect equipment leaks in process
• In the case of a rotating equipment calibration. One is referred to as a units in SOCMI facilities, particu-
shaft seal test: 180 deg around the “zero gas,” defined as air with less larly for compounds such as form-
perimeter of the interface of the than 10 ppmv (parts per million aldehyde, aldehydes and other oxy-
seal and the shaft by volume) VOC. The other cali- genated chemicals that typically do
bration gas, referred to as a “refer- not provide a satisfactory response
Method 21 ence gas,” uses a specified reference on a FID-type unit.
Method 21, under 40 CFR Part 60, compound in an air mixture. The Operation of the non-dispersive
Appendix A, provides rules with concentration of the reference com- infrared (NDIR) detector is based
respect to how VOCs are moni- pound must approximately equal on the principle that light absorp-
tored and measured at potential the leak definition specified in the tion characteristics vary depending
leak points in a facility. Those po- regulation. The leak definition, as on the type of gas. Because of this,
tential leak points include, but are mentioned above, is the threshold NDIR detection can be subject to
not limited to: valves, flanges and standard pertinent to the govern- interference due in large measure
other connections; pumps and com- ing regulation. to such constituents as water vapor
pressors; pressure-relief devices; and CO2, which may absorb light
process drains; open-ended valves; Monitoring devices at the same wavelength as the tar-
pump and compressor seals; de- A portable VOC-monitoring device geted compound. This type of detec-
gassing vents; accumulator vessel will typically be equipped with a tor is typically confined to the de-
vents; agitator seals and access door rigid or flexible probe. The end of tection and measurement of single
seals. It also describes the required probe is placed at the leak inter- components. Because of that pro-
calibration process in setting up the face of a joint, such as a flange, clivity, good or bad, the wavelength
monitoring device. Essentially any threaded connection or coupling, at which a certain targeted com-
monitoring device may be used as or at the interface of a pump, com- pound absorbs infrared radiation,
long as it meets the requirements pressor, or agitator seal where it having a predetermined value, is
set forth in Method 21. interfaces with the shaft. With its preset for that specific wavelength
Cylinder gases used for calibrat- integral pump, the device, when through the use of optical filters. As
ing a monitoring device need to be switched on, will draw in a contin- an example, if the instrument was
certified to be within an accuracy uous sample of gas from the leak- set to a wavelength of 3.4 microm-
of 2% of their stated mixtures. It is interface area into the monitoring eters, the device could detect and
recommended that any certification device. The instrument’s response measure petroleum fractions, such
of this type be filed in either digital or screening value is a relative as gasoline and naphtha.
form or at the very least as a hard measure of the sample’s concentra- The combustion-type analyzer is
copy. There should also be a speci- tion level. The screening value is designed to measure either thermal
fied shelf life of the contents of the detected and displayed in parts per conductivity of a gas or the heat pro-
cylinder. If the shelf life is exceeded, million by volume, or if the instru- duced as a result of combustion of the
the contents must be either re-ana- ment is capable and the degree of gas. Referred to as hot-wire detectors
lyzed or replaced. accuracy needed, in parts per bil- or catalytic oxidizers, combustion-
Method 21 goes on to define how lion by volume (ppbv). type monitors are nonspecific for
to test flanges and other joints, as The detection devices operate on gas mixtures. If a gas is not readily
well as pump and compressor seals a variety of detection principles. combustible, similar in composition
and various other joints and inter- The most common are ionization, to formaldehyde and carbon tetra-
faces with the potential for leaks. infrared absorption and combus- chloride, there may be a reduced re-
There are two gases required for tion. Ionization detectors operate sponse or no response at all.
Chemical Engineering www.che.com May 2014 47
Table 2 – Federal Regulations That Require a Formal LDAR
Cover Story Program With Method 21
40 CFR Regulation Title
Part Subpart
Due to the variability in the sen- 60 VV SOCMI VOC Equipment Leaks NSPS
sitivity of the different monitoring
60 DDD Volatile Organic Compound (VOC) Emissions from the Poly-
devices, the screening value does mer Manufacturing Industry
not necessarily indicate the actual
60 GGG Petroleum Refinery VOC Equipment Leaks NSPS
total concentration at the leak in-
terface of the compound(s) being 60 KKK Onshore Natural Gas Processing Plant VOC Equipment
Leaks NSPS
detected. The leak interface is the
immediate vicinity of the joint 61 J National Emission Standard for Equipment Leaks (Fugitive
being tested — the point at which Emission Sources) of Benzene
the end of the probe is placed. Re- 61 V Equipment Leaks NESHAP
sponse factors (RFs), determined 63 H Organic HAP Equipment Leak NESHAP (HON)
for each compound by testing or
63 I Organic HAP Equipment Leak NESHAP for Certain Processes
taken from reference sources, then
correlate the actual concentration 63 J Polyvinyl Chloride and Copolymers Production NESHAP
of a compound to that of the con- 63 R Gasoline Distribution Facilities (Bulk Gasoline Terminals and
centration detected by the moni- Pipeline Breakout Stations)
toring device. As mentioned previ- 63 CC Hazardous Air Pollutants from Petroleum Refineries
ously, the monitoring device must 63 DD Hazardous Air Pollutants from Off-Site Waste and Recovery
first be calibrated using a certified Operations
reference gas containing a known
63 SS Closed Vent Systems, Control Devices, Recovery Devices
compound at a known concentra- and Routing to a Fuel Gas System or a Process
tion, such as that of methane and
63 TT Equipment Leaks – Control Level 1
isobutylene. RFs at an actual con-
centration of 10,000 ppmv have 63 UU Equipment Leaks – Control Level 2
been published by the EPA in a 63 YY Hazardous Air Pollutants for Source Categories: Generic
document entitled “Response Fac- Maximum Achievable Control Technology Standards
tors of VOC Analyzers Calibrated 63 GGG Pharmaceuticals Production
with Methane for Selected Organic 63 III Hazardous Air Pollutants from Flexible Polyurethane Foam
Chemicals.” Production
Method 21 requires that any se-
63 MMM Hazardous Air Pollutants for Pesticide Active Ingredient
lected detector meet the following Production
specifications:
63 FFFF Hazardous Air Pollutants: Miscellaneous Organic Chemical
• The VOC detector should respond Manufacturing
to those organic compounds being
processed (determined by the RF) 63 GGGGG Hazardous Air Pollutants: Site Remediation
• Both the linear response range 63 HHHHH Hazardous Air Pollutants: Miscellaneous Coating Manufac-
and the measurable range of the turing
instrument for the VOC to be 65 F Consolidated Federal Air Rule — Equipment Leaks
measured and the calibration gas 264 BB Equipment Leaks for Hazardous Waste TSDFs
must encompass the leak defini-
265 BB Equipment Leaks for Interim Status Hazardous Waste TSDFs
tion concentration specified in the
regulation
• The scale of the analyzer meter Federal regulations program and help establish a com-
must be readable to ±2.5% of There are federal regulations that prehensive and detailed procedure.
the specified leak definition pertain to monitoring for VOCs
concentration and require the implementation of RCRA
• The analyzer must be equipped a formal LDAR program in concert The Solid Waste Disposal Act of
with an electrically driven pump with the rules of Method 21. There 1965 was amended in 1976 to in-
so that a continuous sample is are other federal regulations that clude the Resource Conservation
provided at a nominal flowrate of require the rules of Method 21, but and Recovery Act (RCRA), which
between 0.1 and 3.0 L/min do not require a formal LDAR pro- encompassed the management of
• The analyzer must be intrinsi- gram. Tables 2 and 3 list those vari- both hazardous waste and solid
cally safe for operation in explo- ous regulations. waste. Prompted further by an ever
sive atmospheres It is the manufacturer’s responsi- increasing concern of underground
• The analyzer must be equipped bility to make the proper determina- water contamination, this act was
with a probe or probe extension tion as to what regulations it needs again amended in 1984 to address
not to exceed 0.25 in. outside di- to comply with. Those specific regu- underground storage tanks (USTs)
ameter with a single end opening lations, coupled with the Method 21 and associated underground piping
for sampling requirements, will define the LDAR under Subtitle I. This Amendment
48 Chemical Engineering www.che.com May 2014
Table 3 – Federal Regulations that Require the Use of Method 21
But Not a Formal LDAR Program
40 CFR Regulation Title
Part Subpart
60 XX Bulk Gasoline Terminals • Oversee cleanups of petroleum re-
60 QQQ VOC Emissions from Petroleum Refinery Wastewater Systems leases by responsible parties
60 WWW Municipal Solid Waste Landfills • Enforce cleanups by recalcitrant
parties
61 F Vinyl Chloride
• Pay for cleanups at sites where
61 L Benzene from Coke By-Products the owner or operator is unknown,
61 BB Benzene Transfer unwilling, or unable to respond,
61 FF Benzene Waste Operations or those that require emergency
action
63 G Organic Hazardous Air Pollutants from SOCMI for Process
Vents, Storage Vessels, Transfer Operations, and Wastewater
• Conduct inspections and other re-
lease prevention activities
63 M Perchloroethylene Standards for Dry Cleaning In Figure 1 the progress being
63 S Hazardous Air Pollutants from the Pulp and Paper Industry made by the program can readily
63 Y Marine Unloading Operations be seen. In 2002, RCRA was looking
at 142,709 LUST sites — sites that
63 EE Magnetic Tape Manufacturing Operations
were flagged for cleanup. Through-
63 GG Aerospace Manufacturing and Rework Facilities out the following nine years, 2002
63 HH Hazardous Air Pollutants from Oil and Gas Production through 2011, 54,726 of those sites
Facilities were cleaned, leaving 87,983 still
63 OO Tanks ­— Level 1 targeted for cleanup.
63 PP Containers Within the RCRA program there
are requirements that impact de-
63 QQ Surface Impoundments
sign, fabrication, construction, loca-
63 VV Oil/Water, Organic/Water Separators tion, monitoring and operation of
63 HHH Hazardous Air Pollutants from Natural Gas Transmission and USTs and associated underground
Storage piping. The EPA has provided a
63 JJJ Hazardous Air Pollutant Emissions: Group IV Polymers and number of sites on the internet that
Resins provide a great deal of information
63 VVV Hazardous Air Pollutants: Publicly Owned Treatment Works on the various CFR Parts. 40 CFR
Part 260 contains all of the RCRA
65 G CFAR ­— Closed Vent Systems
regulations governing hazardous
264 AA Owners and Operators of Hazardous Waste Treatment, Stor- waste identification, classification,
age, and Disposal Facilities — Process Vents generation, management and dis-
264 CC Owners and Operators of Hazardous Waste Treatment, posal.
Storage and Disposal Facilities — Tanks, Surface Impound- Listed wastes are divided into the
ments, Containers
following group designations:
265 AA Interim Standards for Owners and Operators of Hazardous • The F group — non-specific source
Waste Treatment, Storage, and Disposal Facilities — Process wastes found under 40 CFR
Vents
261.31
265 CC Interim Standards for Owners and Operators of Hazardous • The K group — source-specific
Waste Treatment, Storage, and Disposal Facilities — Tanks,
Surface Impoundments, Containers wastes found under 40 CFR
261.32
270 B Hazardous Waste Permit Program — Permit Application
• The P and U group — discarded
270 J Hazardous Waste Permit Program — RCRA Standardized Per- commercial chemical products
mits for Storage Tanks and Treatment Units found under 40 CFR 261.33
Characteristic wastes, which exhibit
regulates the construction, moni- by current UST regulations to pre- one or more of four characteristics
toring, operating, reporting, record- vent and detect releases and were defined in 40 CFR Part 261 Subpart
keeping, and financial respon- performing the necessary UST sys- C are as follows:
sibility for USTs and associated tem operation and maintenance. • Ignitability, as described in 40
underground piping that handle In 1986, the Leaking Under- CFR 261.21
petroleum and hazardous fluids. ground Storage Tank (LUST) Trust • Corrosivity, as described in 40
As of 2011, there were 590,104 Fund was added to the RCRA pro- CFR 261.22
active tanks and 1,768,193 closed gram. The trust financing comes • Reactivity, as described in 40 CFR
tanks in existence in the U.S. Of the from a 0.1¢ tax on each gallon of 261.23
still active tanks, 70.9% were under motor fuel (gasoline, diesel or bio- • Toxicity, as described in 40 CFR
significant operational compliance. fuel blend) sold nationwide. The 261.24
This means that they were using LUST Trust Fund provides capital Table 4 provides a listing of ad-
the necessary equipment required to do the following: ditional CFR parts that further
Chemical Engineering www.che.com May 2014 49
Table 4 – Resource Conservation and Recovery Act (RCRA)
Feature
Cover Story
Report Information
40 CFR Part Regulation Title
260 Hazardous Waste Management System: General
define the regulations under the 261 Identification and Listing of Hazardous Waste
Resource Conservation and Recov-
ery Act. 262 Standards Applicable to Generators of Hazardous Waste
264 Standards for Owners and Operators of Hazardous Waste Treat-
Final remarks ment, Storage and Disposal Facilities
I am fervently against overregula-
265 Interim Status Standards for Owners and Operators of Hazardous
tion and watch with keen interest Waste Treatment, Storage and Disposal Facilities
the unfolding debate occurring on
Capitol Hill over the amendment 266 Standards for the Management of Specific Hazardous Wastes
and Specific Types of Hazardous Waste Management Facilities
to the Toxic Substances Control
Act (TSCA) for example. But the 267 Standards for Owners and Operators of Hazardous Waste Facili-
improved safety, clean air, clean ties Operating Under a Standardized Permit
water, and cost savings realized 270 EPA Administered Permit Programs: The Hazardous Waste Permit
from the CAA and RCRA programs Program
are four major returns on invest-
272 Approved State Hazardous Waste Management Programs
ment that come back to a manufac-
turer from the investment in a good 273 Standards for Universal Waste Management
leak-detection program. Whether
279 Standards for the Management of Used Oil
monitoring and repairing leaks
above ground, in accordance with 280 Technical Standards and Corrective Action Requirements for
the CAA, or below ground, in accor- Owners and Operators of Underground Storage Tanks (UST)
dance with the RCRA, it is, simply 281 Approval of State Underground Storage Tank Programs
put, just good business. As alluded
to at the outset of this article, leaks 282 Approved Underground Storage Tank Programs
RG LeaseFleet Ad_4.5625 x 4.875.pdf 1 4/7/14 11:47 AM
in hazardous-fluid-service pip-
ing systems have served, in many
cases, as an early-warning indicator
of something much worse to come.
At the very least, such leaks can
contribute to air pollution, ground-
water contamination, lost product
revenue, housekeeping costs, and a
risk to personnel — a few things we
can all live without. ■
Edited by Gerald Ondrey INTRODUCING THE WORLD’S LARGEST FLEET OF
SUCCESSFULLY TESTED BLAST-RESISTANT BUILDINGS.
Author
W. M. (Bill) Huitt has been No matter how many you need, how big you need them to be or when
involved in industrial pip-
ing design, engineering and you need them, get proven protection from the safety authority.
construction since 1965. Posi-
tions have included design en-
gineer, piping design instruc-
tor, project engineer, project
supervisor, piping depart-
ment supervisor, engineering
manager and president of W. • F O R M E R LY A B O X 4 U •
M. Huitt Co. (P.O. Box 31154,
St. Louis, MO 63131-0154;
Phone: 314-966-8919; Email: wmhuitt@aol.
com), a piping consulting firm founded in 1987.
His experience covers both the engineering and
construction fields and crosses industry lines
to include petroleum refining, chemical, petro-
chemical, pharmaceutical, pulp & paper, nuclear
power, biofuel and coal gasification. He has writ-
ten numerous specifications, guidelines, papers,
and magazine articles on the topic of pipe design
and engineering. Huitt is a member of the In-
ternational Society of Pharmaceutical Engineers
(ISPE), the Construction Specifications Institute
(CSI) and the American Society of Mechani-
cal Engineers (ASME). He is a member of the
B31.3 committee, a member of three ASME-BPE
subcommittees and several task groups, ASME
Board on Conformity Assessment for BPE Certi-
fication where he serves as vice chair, a member
of the American Petroleum Institute (API) Task
Group for RP-2611, serves on two corporate spec-
ification review boards, and was on the Advisory 855.REDGUARD
Board for ChemInnovations 2010 and 2011 a
multi-industry conference & exposition.
Circle 1 on p. 76 or go to adlinks.che.com/50976-01
Chemical Engineering www.che.com May 2014 51
Environmental Manager

Monitoring Flame Hazards In Chemical Plants


The numerous flame sources in CPI facilities necessitate the installation of
advanced flame-detection technologies
Ardem Antabian
MSA — The Safety Company

F
ire is a primary and very real
threat to people, equipment
and facilities in the chemical
process industries (CPI), es-
pecially in the refining and storage of
petrochemicals. The consequences
of failing to detect flames, combusti-
ble gas leaks or flammable chemical
spills can have dire consequences,
including loss of life and catastrophic
plant damage.
The monitoring of flame hazards is
mandated by the U.S. Occupational
Safety and Health Administration
(OSHA; Washington, D.C.; www.
osha.gov) through its comprehen-
sive Process Safety Management
(PSM) federal regulation. Interna-
tionally, the European Union (E.U.)
splits gas and flame safety respon-
sibilities between E.U. directives and
European standards organizations, FIGURE 2. Flame detectors, such as those shown here, implement ultraviolet and infrared detection
including the European Committee technologies
for Electrotechnical Standardization
(Cenelec; Brussels, Belgium; www. to train employees to follow related including ultraviolet (UV) and infra-
cenelec.eu), the International Elec- safety procedures consistently. In red (IR) spectroscopy and visual
trotechnical Commission (IEC; Ge- either case, it is important to under- flame imaging. The source of flames
neva, Switzerland; www.iec.ch) and stand the many different sources of in CPI plants is typically fueled by
several other bodies. flame hazards, the detection sensor hydrocarbons, which when sup-
Many accidents are the result of ei- technologies that can warn of immi- plied with oxygen and an ignition
ther failing to implement these stan- nent danger and the proper location source, produce heat, carbon diox-
dards properly with suitable flame- of flame detectors in today’s com- ide and other products of combus-
detection equipment or the failure plex chemical plants. tion. Intense flames emit visible, UV,
In the petrochemical plant environ- and IR radiation (Figure 1). Flame
Ultraviolet Visible Infrared
ment, the range of potential flam- detectors are designed to detect
mable hazards is expansive and the emission of light at specific
growing as materials and processes wavelengths, allowing them to dis-
become more complex. These haz- criminate between flames and false
Relative energy

Sun's
energy ards have led to the development of alarm sources.
reaching
the more sophisticated combustible-gas
earth and flame-sensing technologies with Flame-sensing technologies
embedded intelligence that can bet- The flame safety industry has de-
ter detect the most common indus- veloped four primary optical flame-
300 nm 400 nm 800 nm 4-5 m
trial fire sources, some of which are sensing technologies: UV, UV/IR,
Wavelength listed in Table 1. multi-spectrum infrared (MSIR),
and visual flame imaging (Figure 2).
FIGURE 1. Flame detectors can detect light emis-
sions at specific wavelengths across the UV, vis-
Principles of flame detection These sensing technologies are all
ible and IR spectrum to distinguish between actual Industrial process flame detectors based on line-of-sight detection of
flames and false alarm sources detect flames by optical methods, radiation emitted by flames in the
70 Chemical Engineering www.chemengonline.com May 2016
Table 1. Common Industrial Input layer Hidden layer Output layer
Fire Sources
Alcohols Liquefied natural gas (LNG)
Diesel fuels Liquefied petroleum gas (LPG) Sensor 1
Gasoline Paper
Kerosene Textiles
Jet fuels Solvents Sensor 2
Output
Ethylene Sulfur
Hydrogen Wood
Sensor 3

UV, visible and IR spectral bands.


Process, safety and plant engi- Sensor 4
neers must choose from among
these technologies to find the de-
vice that is best suited to their in- FIGURE 3. Many flame detectors employ technology based on artificial neural networks (ANNs) to more
dividual plant’s requirements for accurately analyze flames
flame monitoring by deciding upon
the importance of the detection events observed in the past. These technology alternatives, there are
range, field of view, response time detectors have been trained through many performance criteria that must
and immunity against certain false NNT intelligence to recognize flames be considered. The following sec-
alarm sources. based upon those data, and deter- tions outline some of these impor-
Ultraviolet/infrared (UV/IR). By in- mine if they are real events or poten- tant detector criteria.
tegrating a UV optical sensor with tial false alarm sources. False alarm immunity. False alarm
an IR sensor, a dual-band flame de- Visual flame-imaging flame rejection is one of the most impor-
tector is created that is sensitive to detectors. The design of visual tant considerations for the selection
the UV and IR radiation emitted by flame detectors relies on stan- of flame detectors. False alarms are
a flame. The resulting UV/IR flame dard charge-couple-device (CCD) more than a nuisance — they are
detector offers increased immunity image sensors, commonly used in both productivity and cost issues. It
over a UV-only detector, operates at closed-circuit television cameras, is therefore essential that flame de-
moderate speeds of response, and and flame-detection algorithms to tectors discriminate between actual
is suited for both indoor and out- establish the presence of fires. The flames and benign radiation sources,
door use. imaging algorithms process the live such as sunlight, lighting fixtures, arc
Multispectral infrared (MSIR). Ad- video image from the CCD array welding, hot objects and other non-
vanced MSIR flame detectors com- and analyze the shape and pro- flame sources.
bine multiple IR detector arrays with gression of would-be fires to dis- Detection range and response
neural network intelligence (NNT). criminate between flame and non- time. A flame detector’s most
They provide pattern-recognition ca- flame sources. basic performance criteria are de-
pabilities that are based on training Visual flame detectors with CCD tection range and response time.
to differentiate between real threats arrays do not depend on emis- Depending on a specific plant-ap-
and normal events, thus reducing sions from carbon dioxide, water plication environment, each of the
false alarms. MSIR technology allows and other products of combustion alternative flame-detection tech-
area coverage up to six times greater to detect fires, nor are they influ- nologies recognizes a flame within
than that of more conventional UV/IR enced by fire’s radiant intensity. As a certain distance and a distribu-
flame detectors. a result, they are commonly found tion of response times. Typically,
NNT is based on the concept of ar- in installations where flame detec- the greater the distance and the
tificial neural networks (ANNs), which tors are required to discriminate shorter the time that a given flame-
are mathematical models based on between process fires and fires re- sensing technology requires to de-
the study of biological neural net- sulting from an accidental release tect a flame, the more effective it is
works. A group of artificial neurons of combustible material. at supplying early warning against
in an ANN process information and Visual flame detectors, despite fires and detonations.
actually change structure during a their many advantages, cannot Field of view (FOV). Detection
learning phase. This learning phase detect flames that are invisible to range and FOV define area cover-
allows ANNs to model complex re- the naked eye, such as hydrogen age per device. Like a wide-angle
lationships in the data delivered by flames. Heavy smoke also impairs lens, a flame detector with a large
sensors in a quick search for pat- the detector’s capacity to detect field of view can take in a broader
terns that results in pattern recogni- fire, since visible radiation from the scene, which may help reduce the
tion (Figure 3). fire is one of the technology’s fun- number of flame detectors required
Flame detectors with NNT oper- damental parameters. for certain installations. Most of to-
ate similarly to the human brain; day’s flame detector models offer
they have thousands of pieces of Flame detection requirements fields of view of about 90 to 120
data stored in their memories from When configuring a flame-detection deg (Figure 4).
hundreds of flame and non-flame system and evaluating the available Self diagnostics. To meet the high-

Chemical Engineering www.chemengonline.com May 2016 71


FIGURE 5. Three-dimensional mapping of a facil-
ity is useful in determining the most appropriate
installation locations for flame detectors
FIGURE 4. Field of view is an important factor to consider in the installation of flame-detection equip-
ment. This diagram shows the distance a flame can be detected at various angles. For example, at 0 deg, to protect large and complex areas
a flame can be detected at 230 ft, and at a 50-deg angle, it can be detected at 50 ft (in this figure, the filled with process equipment from
degree symbol ° is used for angle degrees, and the prime symbol ’ is used for feet) the risks of flames, explosions and
toxic gases. For these systems to be
est reliability standards, continuous that address safety and reliability effective, their location and installa-
optical-path monitoring (COPM) di- based on optimizing processes for tion are important so that they offer
agnostics are often built into opti- risk. The IEC 61508 standard is a a high likelihood of detecting flame
cal flame detectors. The self-check risk-based approach for determin- and gas hazards within monitored
procedure is designed to ensure ing the SIL of safety-instrumented process areas.
that the optical path is clear, the functions. Unlike other international Three-dimensional mapping. De-
detectors are functioning, and addi- standards, IEC 61508 takes a ho- termining the optimal quantity and
tionally, that the electronic circuitry listic approach when quantifying location of flame and gas detec-
is operational. the safety performance of electri- tors is therefore critical to ensure
Self-check routines are pro- cal control systems — the design the detection system’s effective-
grammed into the flame detector’s concept, the management of the ness. Flame and gas three-dimen-
control circuitry to activate about design process and the operations sional mapping is a solution that
once every minute. If the same fault and maintenance of the system assists in the evaluation of flame
occurs twice in a row, then a fault is throughout its lifecycle are within and gas risks within a process fa-
indicated via a 0–20-mA output or the scope. cility and also reduces these risks
a digital communications protocol, toward an acceptable risk profile.
such as HART or Modbus. Location and installation Flame and gas mapping includes
SIL/SIS standards. When plant A variety of processes and sources the placement of detectors in ap-
safety engineers choose detectors within the plant environment can propriate locations to achieve the
certified to safety integrity levels lead to flame and fire incidents, best possible detection coverage
(SIL) and properly integrate them including leaking tanks, pipes, (Figure 5).
into safety-instrumented systems valves, pumps and so on. Accu- The use of three-dimensional
(SIS), they have again added an- rate detection while avoiding false flame and gas mapping helps
other layer of safety. Certification alarms is also important because plant, process and safety engi-
to these standards plays a valuable false alarms result in unnecessary neers in a number of ways. First,
role in effective industrial gas and process or plant shutdowns, slow- mapping helps to increase plant
flame detection. ing production and requiring time- safety by improving the likelihood
Normative standards establish consuming reviews, paperwork of detecting flame and gas haz-
minimum requirements for the de- and reporting. ards. Also, it allows facilities to
sign, fabrication and performance of False alarms can, over time, pro- quantify their risk of a fire or a gas
flame detectors and other safety de- vide a false sense of security, be- leak, and then assess the overall
vices as necessary to maintain pro- cause employees can become com- effectiveness of their flame- and
tection of personnel and property. placent if alarms go off frequently for gas-detection coverage. For new
The ANSI/ISA S84.00.01 standard no apparent reason and are continu- installations, mapping can help im-
was enacted to drive the classifica- ally ignored. The problem is that per- prove the design of new fire and
tion of SIS for the process industries sonnel alone cannot really determine gas systems to mitigate risks from
within the U.S., as well as the norms the difference between a false alarm accidental gas releases or fires.
introduced by the IEC (IEC 61508 and a serious accident that is about For existing installations, mapping
and IEC 61511). to happen. provides a method for assessing
Together, these standards have Fixed flame- and gas-detector the risk-reduction performance
introduced several specifications systems are designed and installed of existing fire- and gas-detector
72 Chemical Engineering www.chemengonline.com May 2016
systems and recommends ways to the flame detectors, then installa- makes them experts in helping to
improve coverage. tion depends on the type of flame identify potential hazards and the
Mapping assists facilities in un- detector chosen. Most optical-type best way to prevent accidents.
derstanding their risk of a fire or flame detectors are placed high and Remember, too, that no single
a gas leak, and then allows them are pointed downward either inside flame-detection sensing technol-
to optimize their flame- and gas- or outside buildings or structures to ogy is right for every potential plant
detection protection layout by monitor tanks and pipelines running layout and hazard. For this reason,
recommending the appropriate throughout the plant. adding multiple layers of flame- and
detector technologies, detector gas-detection technology provides
locations and quantities. Mapping Wrapping up a multi-sensory approach that in-
also equips the engineer with the In order to protect chemical pro- creases detection reliability and also
means to measure detection im- cesses and plants from flame haz- can prevent false alarms. ■
provements when small incremen- ards, it is important to understand Edited by M. Bailey and D. Lozowski
tal design changes are made. Map- the basic detection sensor technol-
ping can therefore help to minimize ogies and their limitations. Defining Author
overall system costs. the type of potential hazard fuels, Ardem Antabian is currently the
With mapping, determining detec- the minimum fire size to be detected OGP (Oil & Gas Products) seg-
ment manager at MSA — The
tor layouts becomes much simpler, and the configuration of the space Safety Company (26776 Simpat-
because mapping provides a me- to be monitored through three-di- ica Circle, Lake Forest, CA
thodical and systematic approach mensional hazard mapping can in- 92630; Email: Ardem.Antabian@
for determining the areas with the fluence the choice of instrument. msasafety.com; Phone: 949-
268-9523. Website: www.
highest likelihood of flame and gas When reviewing a plant’s flame- msasafety.com). Antabian joined
risks. Understanding the locations safety protection level, be sure to the company in 1999, and has
and likelihood of risks will help re- ask for assistance from any of the held various positions, including global assignments in
Dubai, U.A.E. and Berlin, Germany. He also helped de-
move guesswork and uncertainties flame detection equipment manu- velop the company’s advanced-point and open-path
from engineering. facturers. They have seen hun- infrared gas detectors, as well as its multi-spectral
Once the optimal locations are dreds, if not thousands, of plants infrared flame detector. Antabian holds dual B.S. de-
CIC-10307 halfp grees in chemical engineering and chemistry from
determined forpage
the ad.qxd 3/25/07
placement of 6:19their
and PM Page 1 layouts, which
unique California State University, Long Beach.

PLASTIC CONTROL VALVES FOR


ALL YOUR CORROSIVE APPLICATIONS
Collins plastic control valves are
highly responsive control valves
designed for use with corrosive
media and/or corrosive atmos-
pheres.
Collins valves feature all-plastic
construction with bodies in PVDF,
PP, PVC and Halar in various body
styles from 1/2" - 2" with Globe,
Angle or Corner configurations and
many trim sizes and materials.
Valves may be furnished without
positioner for ON-OFF applications.
Call for more information on our
plastic control valves.

P.O. Box 938 • Angleton, TX 77516


Tel. (979) 849-8266 • www.collinsinst.com

Circle 07 on p. 94 or go to adlinks.chemengonline.com/61495-07

Chemical Engineering www.chemengonline.com May 2016 73


Feature Report Part 2

Integrated
Risk-Management Matrices
An overview of the tools available to reliability professionals for making their
organization the best-in-class
In Brief
Reliability,
historically

Reliability, today

Risk-mitigation
approaches

How do we measure
risk?

KPIs and risk

Nathanael Ince competitive environment.


PinnacleART With this in mind, a best-in-class reliability

S
organization combines state-of-the-art theory,
ince the 1960s, process facility op- software and condition-monitoring techniques
erators have made concerted ef- with a strong collaboration of departments
forts to improve the overall reliability and associated personnel. An independent
and availability of their plants. From risk-based inspection (RBI) program or reli-
reliability theory to practical advancements ability-centered maintenance (RCM) program
in non-destructive examination and condi- no longer suffices as cutting-edge. Rather, the
tion-monitoring techniques, the industry has inspection department (power users of RBI)
significantly evolved and left key operations and maintenance department (power users
personnel with more tools at their disposal of RCM) are integrating with process, opera-
than ever before. However, this deeper ar- tions, capital projects and other teams to form
senal of tools, coupled with more stringent an overall reliability work process for the suc-
regulatory scrutiny and internal business cess of the plant.
pressure, introduces a heightened expecta- To highlight reliability’s growing prominence
tion of performance. Now, more than ever, within process facilities, this article addresses
companies recognize that best-in-class reli- the following:
ability programs not only save lives but in- • A brief history of reliability practices in the
crease the bottom line. These programs are 20th and 21st centuries
also one of the foremost “levers” for C-level • Examples of current reliability program tools
personnel to pull when trying to contend in a • A characterization of three different

Chemical Engineering www.chemengonline.com may 2016 65


Table 1. Example Mechanical-Integrity and Maintenance-Program them, or to show negligence to these
Improvements steps in the event of a release, im-
Mechanical integrity improvements Maintenance/reliability improvements plied the possibility of criminal activity.
In other words, if those responsible
Assessments and audits Assessments and audits
in the event of a release were found
Damage/corrosion modeling Preventive and predictive maintenance to be negligent in these items, they
Risk-based inspection Equipment hierarchies and data cleanup could go to jail. The other business
Inspection data management and trending Operator-driven reliability (rounds)
implication of this standard was that it
meant that other individuals, and de-
Piping circuitization Mobile platforms
partments, now had a part to play in
Integrity operating windows Reliability operating windows reliability and overall process safety.
Corrosion monitoring locations (CML) and thickness Maintenance data/order management (computer- While reliability was confined to
management locations (TML) optimization ized maintenance-management system; CMMS) designing equipment that could last
Asset retirement calculation Spare parts optimization a certain time and coupling it with a
Corrosion under insulation (CUI) program Reliability-centered maintenance
non-certified inspector to make gen-
eral observations in the 1950s, by the
Utilizing advanced non-destructive evaluation Reliability-centered design mid-1990s, reliability had become a
Continuous condition monitoring Repair procedures much more complex, integrated and
accelerated science.

risk-mitigation applications that cycle. As a result, several different de- Reliability today
are currently applied in process partments and individuals cooperated With the greater expectation on to-
facilities to ensure they attained reliability. day’s programs, department manag-
• The case for ensuring these risk The concept of RCM pushed ers (including reliability, mechanical-
mitigation frameworks are working through some industries quicker integrity or maintenance managers)
together than others. While it started with the face a powerful, but often intimidat-
• The value of key performance airlines, it flowed quickly into power ing array of tools available to them for
indicators (KPIs) in providing generation, petrochemical and petro- improving their reliability programs.
transparency and accountability leum-refining operations thereafter. Examples are listed in Table 1.
to the effectiveness of these risk Fast-forward to 1992, and another While this only represents a sub-
mitigation frameworks facet, called process-safety manage- set of the options available to the
ment (PSM), was introduced into the re- manager, all of these activities aim at
Reliability, historically liability picture. In response to a growing doing the following:
When one thinks about process reli- perception of risk related to hazardous 1. Reducing the risk of unplanned
ability, a variety of definitions come processes, the Occupational Safety downtime.
to mind. However, it has come a and Health Administration (OSHA) is- 2. Limiting safety and environmental
long way since the early 20th cen- sued the Process Safety Standard, risk.
tury. From the 1920s to the 1950s, OSHA 1910.119, which includes the 3. Ensuring compliance with regula-
reliability went from being classified following 14 required elements: tory standards.
as “repeatability” (how many times • Process-safety information 4. Doing steps one through three for
could the same results repeat) to de- • Process hazard analysis the least cost possible.
pendability (hours of flight time for an • Operating procedures To summarize, the goal of these
engine), to a specific, repeatable re- • Training managers is to put a plan in place
sult expected for a duration of time. • Contractors and execute a plan that identifies and
Through the 1950’s age of industri- • Mechanical integrity mitigates risks as efficiently as pos-
alization, reliability’s evolving definition • Hot work sible. To do that, one has to system-
was still very much focused on de- • Management of change atically identify those risks in addition
sign and not as much on operations • Incident investigation to the level to which those risks must
or maintenance. Then in the 1960s, • Compliance audits be mitigated. If this is done correctly,
the airline industry introduced the • Trade secrets the design, inspections, preventative
concept of reliability centered mainte- • Employee participation maintenance, operational strategies,
nance (RCM), pushing the idea that • Pre-startup safety review and other program facets should
the overall reliability of a system in- • Emergency planning & response all be aligned in attaining steps one
cluded not only the design, but also The intent of the regulation was to through four.
the operations and maintenance of limit the overall risk related to danger-
that system. In other words, reliability ous processes, and “raise the bar” Risk-mitigation approaches
engineering was now stretching into for compliance expectation for facili- Since the 1960s, there have been
other departments, mandating that ties with these “covered” processes. substantial efforts on figuring out
the overall risk of failure was tied to At that point, it became law to fulfill how to best characterize both
multiple aspects of the asset’s life- these 14 elements, and to ignore downtime and loss-of-containment
66 Chemical Engineering www.chemengonline.com may 2016
team on the identification of process
PHA/HAZOP/QRA deviations that can lead to undesir-
(quantitative risk assessment)
able consequences, the risk ranking
of those deviations, and the assign-
ment of actions to either lower the
Consequence of failure

RBI Extreme
probability of those failures or the
High consequence if the failures do occur.
While a PHA would not identify main-
Med high
RCM tenance strategies or detailed corro-
Med sion mitigation or identification strate-
Med low gies, it focuses on safety and not unit
reliability. In the end, the major deliv-
Low
erable is a set of actions that have to
Negligible be closed out to ensure compliance
Likelihood of failure (failure rate) with the PSM standard. Typically, this
Figure 1. This graphical “consequence-of-failure” risk matrix shows the areas covered by process haz- process is owned and facilitated by
ard analysis (PHA), risk-based inspection (RBI) and reliability centered maintenance (RCM) the PSM manager or department.
RBI. RBI arose from an industry
risk in a facility so that appropriate Let’s briefly characterize each. study in the 1990s that produced
and targeted mitigation actions can PHA. The PHA came out of OSHA’s API (American Petroleum Institute)
be taken at the right time. That being PSM standard and is one of the 14 580 and 581, which describe a sys-
said, there are three common risk elements listed above. Every five tematic risk identification and mitiga-
identification and mitigation frame- years, subject matter experts come tion framework that focuses only on
works that are currently being used together for a couple of weeks and loss of containment. For this reason,
in process facilities today. These in- identify the major events that could when an equipment item or pip-
clude process hazard analysis (PHA), happen at different “nodes” in a unit. ing segment (typically called “piping
risk-based inspection (RBI), and reli- The general idea is to use guide- circuit”) is evaluated, the only failure
ability-centered maintenance (RCM). words to systematically focus the that is of concern to the facility is the

Quest Integrity’s Furnace Tube


INSPECT Inspection System (FTIS™) is
the proven technology providing
100% 100% inspection coverage of your
serpentine coils.
OF yOur
HEATEr The FTIS inspection results are
processed with our LifeQuest™
• Pitting (interior or
exterior of pipe)

COILS Heater engineering software,


• Corrosion (interior
or exterior of pipe)
providing a comprehensive
• Erosion and flow
fitness-for-service and remaining assisted wear
life assessment compliant with • Denting and
the API-579 Standard. Quest ovality
Integrity delivers a complete • Bulging and
swelling
solution that helps transfer your
• Coke and scale
integrity and maintenance risk
build-up
into reliability.

QuestIntegrity.com
CHALLENGE CONVENTION

Circle 30 on p. 94 or go to adlinks.chemengonline.com/61495-30
CE-half-page-FTIS-Feb-2016.indd 1 1/15/2016 2:10:50 PM

Chemical Engineering www.chemengonline.com may 2016 67


of failure is on the horizontal axis, and
the consequence of failure is shown
on the vertical axis.
As shown, the frameworks reveal
the following characterization for
each of the three covered risk mitiga-
tion frameworks:
• PHA — High consequence of
failure events but lower probability
that they will happen (an example
would be an overpressure on a
column with insufficient relief-
systems capacity)
• RBI — Medium consequence of
failure events (loss of containment)
breach of the pressure boundary. probabilities of those failure modes and a medium probability that they
As an example, the only failure occurring at any given time. Once will happen (an example would be
mode evaluated on a pump would the potential failure causes are iden- a two-inch diameter leak of a flam-
typically be a leak in the casing or tified, strategies are recommended mable fluid from a drum)
the seal. The consequence of those that mitigate the failure mode to ac- • RCM — Low consequence of
losses can be business, safety or ceptable levels. Unlike RBI, RCM failure events (loss of function) but
environmental, and while a variety accounts for all failure modes relat- a higher probability that they will
of software packages and spread- ing to loss of function, including loss happen (an example would be a
sheets can be used to accomplish of containment (although it typically rotor failure on a pump)
the exercise, the deliverable is an RBI outsources this exercise to the RBI While each of these frameworks gen-
plan targeting the mitigation of loss- study), and the end deliverable is a erally operate in different areas on the
of-containment events. set of predictive maintenance, pre- matrix, they are still standardized to a
In addition, a best-in-class RBI ventative maintenance, and operator consistent amount of risk. The need to
program will not just be a systematic activities that lower loss-of-function include all three risk-management tools
re-evaluation of that plan every five risks to acceptable levels. Typically, into one standard matrix is twofold:
or ten years, but an ongoing man- this process is owned and facilitated 1. Making sure the data, calcula-
agement strategy that updates the by the maintenance or reliability man- tions and actions coming from one
framework whenever, the risk fac- ager or department. study are properly informing the
tors change. Therefore, if an equip- other studies.
ment’s material of construction was How do we measure risk? 2. Insuring that the actions being
changed, insulation was added to While it’s not uncommon for a single produced by each framework are
an asset, or a piece of equipment facility to run PHA, RBI and RCM at being prioritized appropriately, as
was moved to a different location, once, it begs the question, which one determined by their risk.
a re-evaluation of the asset loss-of- is right? To find the answer, let’s briefly Making sure each of the three
containment risk and an associated discuss risk matrices. A risk matrix is a frameworks is communicating with
update of the RBI plan would be ap- tool that allows one to associate indi- one another is a common omis-
propriate. Typically, this process is vidual assets, failure modes or situa- sion in facilities and programs. Many
owned and facilitated by the inspec- tions with specific levels of risk. There is times, facilities spend millions of dol-
tion or mechanical integrity manager both a probability of an asset failing and lars building out and managing these
or department. a consequence of an asset failing, and frameworks, but there is often overlap
RCM. As mentioned earlier, RCM each is represented by one axis on the between them and data gathered for
was spawned out of the aviation in- matrix. The multiplication of both prob- one framework could be utilized for
dustry, but the focus was to identify ability of failure and consequence of another framework. As an example,
a proactive maintenance strategy failure (represented by the actual loca- an inspection department representa-
that would ensure reliability and that tion of the asset on the matrix) equals tive should be present to ensure the
performance goals were met. While risk. What’s interesting is that many RBI study is aiding the PHA effort.
this has been loosely codified in SAE facilities that are utilizing multiple-risk In addition, prioritizing risk between
(Society of Automobile Engineers) frameworks in their facility are utilizing each framework is another chal-
JA1011, there are a variety of meth- multiple-risk matrices. This again begs lenge. A plant manager is not wholly
ods and approaches and therefore the question, which one is right? concerned about each individual risk
RCM isn’t as controlled as RBI. Figure 1 is a risk matrix that is much framework but rather a prioritized list
However, much like RBI, the RCM larger than the typical 4 × 4 or 5 × 5 of actions with those action’s pro-
study itself aims at identifying the dif- risk matrix, but it shows each of the jected return-on-investment (whether
ferent failure modes of an asset, the previously discussed risk frameworks it is reduction of risk, a reduction of
effects of those failure modes, and the on one larger matrix. The probability cost, or a reduction of compliance
68 Chemical Engineering www.chemengonline.com may 2016
fines). The objective of the integrated reduction, cost reduction or com- ever at their disposal to increase
and organization-wide risk mitigation pliance satisfaction). overall plant availability and process
system should be that all possible For example, at any point in time, safety. To drive systematic improve-
failures must be identified, assessed, the plant (or group of plants) could ments in plant reliability with all these
properly mitigated (whether through see the total risk of loss-of-con- different tools, it is essential for fa-
design, maintenance, inspection, or tainment or loss-of-function events cilities to get the data-management
operations) and monitored in order across their units and assets, the strategy right, to synthesize one’s
of priority with an expected amount total risk of loss of function events approach to measuring, reporting
of return. If a consistent risk matrix is across its units and assets, the total and mitigating risk, and to roll it up in
used effectively, this can inform single planned and unplanned downtime a KPI framework that combines risk,
asset or system decisions and con- across the plant with associated cost and compliance reports. n
tinue to ensure reliability value is being causes, and the total cost associ- Edited by Gerald Ondrey
driven consistently across the facility. ated with running those programs
broken out by activity, area and other Author
KPIs and risk helpful specifics. When one or many Nathanael Ince is client solutions
A good set of key performance in- of those rollout KPIs reveal concerns, director, supporting the Solutions
Department of Pinnacle Advanced
dicators (KPIs) is needed as well sub KPIs should be accessible to ex- Reliability Technologies (One Pin-
to help identify root causes and plore the root cause of those risks, nacle Way, Pasadena, TX 77504;
guide programmatic decisions. downtime or costs. It’s from this KPI Phone: +1-281-598-1330; Email:
nathanael.ince@pinnacleart.com).
Once systematic risk management, drill-down, empowered by synthe- In this capacity, he works closely
production loss, and enterprise- sized risk frameworks, that targeted with his team of solutions engi-
neers to ensure the department is
resource-planning (ERP) systems initiatives and actions can be driven. building and implementing the best asset integrity and
are properly setup, roll-up KPIs reliability programs for PinnacleART’s clients. With more
can be reported regularly that re- Summary than eight years on the PinnacleART team, Ince is an
expert source on mechanical integrity, including proper
veal the overall trending of the re- Reliability programs have come a assessment and implementation of risk-based mechan-
liability program and drive specific long way in 100 years, and reliability ical-integrity programs. Ince has a B.S.M.E. degree from
initiatives with targeted results (risk professionals have more tools than Texas A&M University.

HEXOLOY
®

The Sustainable
Solution in the Harshest Environments
■ Universal corrosion resistance in extremely high
temperatures; small to large complex shape & tight
tolerance capabilities.
■ Our customers minimize maintenance, energy
& downtime costs, reduce emissions & maximize
productivity through Saint-Gobain Engineered
Ceramics' innovation & relentless sustainability focus.
Leverage Saint-Gobain's
Corrosion & Heat Problems that

global footprint, technical


resources & manufacturing

SHUT YOU DOWN


operations to solve your
most difficult problems.

For more information, technical data and case studies contact:


SCD.sales@saint-gobain.com
716-278-6233

Visit our NEW website: refractories.saint-gobain.com

Circle 34 on p. 94 or go to adlinks.chemengonline.com/61495-34

Chemical Engineering www.chemengonline.com may 2016 69


Environmental Manager

Process Safety and Functional Safety in


Support of Asset Productivity and Integrity
Approaches to plant safety continue to evolve based on lessons learned, as well as
new automation standards and technology
Luis Durán
ABB Process Automation Control
Technologies

I
n the chemical process industries
(CPI), one incident can have a tre-
mendous impact on the people
in the plant, the communities
around it, the environment and the
production asset.
This article outlines how learn-
ing from past incidents continues
to drive the development of both
newer standards, as well as new ap-
proaches to process automation as
it relates to plant safety and security.

Learning from incidents


Today, there is a lot of informa-
tion available about process inci-
dents and industrial accidents from
sources such as the Chemical Safety
Board (www.csb.gov), Industrial
Safety and Security Source (www.
isssource.com) or Anatomy of an In- Figure 1. Safety should always a priority at a process plant
cident (www.anatomyofanincident.
com). Regardless of the source, and (1992–2011). (OSHA; www.osha.gov) — the
considering the amount of public On the positive side, the Marsh re- agency tasked with safety of per-
discussion that takes place, particu- port shows that there has been im- sonnel — issued the Process Safety
larly following the very large and visi- provement in the sector after 1992. Management of Highly Hazard-
ble industrial incidents, it’s important This improved safety can be attrib- ous Chemicals standard (29 CFR
to take the opportunity to learn and uted, in part, to the introduction of 1910.199). This regulation set a re-
seek opportunities to improve and the process safety management quirement for hazard management
prevent these incidents from hap- (PSM) standards. and established a comprehensive
pening again (Figure 1). Taking a closer look, it is evi- PSM program that integrates tech-
The impact of incidents and acci- dent that the significant loss for the nology, procedures and manage-
dents on people, the environment and 1987 –1991 period was dominated ment practices. The introduction
plant assets is significant. According by three explosion events, two of of this standard may be credited
to a Marsh LLC (www.marsh.com) which were vapor-cloud explosions with improving process safety per-
publication [1], there is evidence that and account for 70% of the total formance in U.S. hydrocarbon pro-
the petrochemical sector suffered a losses for this timeframe. The key cessing facilities.
terrible period in terms of accidents takeaway from this is that a single
between 1987 and 1991 (Figure 2). incident can have a tremendous Defining safety
The losses (property damage of the impact on the people in the plant, In industry, safety is defined as a re-
production assets, liabilities and so the communities around it, the envi- duction of existing risk to a tolerable
on), recorded in that period were ronment and, last but not least, the or manageable level; understanding
about ten times worse than previous production asset. risk as the probability of occurrence
periods (1976–1986) and about 3.5 In 1992, the U.S. Occupational for that harmful incident and the
times worse than following periods Safety and Health Administration magnitude of the potential harm. In
66 Chemical Engineering www.chemengonline.com november 2015
Figure 2. The 1987–1992 period was exceptionally bad for the petrochemical Figure 3. Operators at a modern control room monitor both the operation
sector due to a few major accidents (Source: Marsh LLC [1]) of the process as well as the safety and security of the plant

many cases, safety is not the elimi- Connecting PSM and FS safety instrumented systems (SIS)
nation of risk, which could be im- Organizations, such as OSHA, rec- that were designed following a pre-
practical or unfeasible. ognize Functional Safety Standard vious RAGAGEP, and to effectively
Although the CPI must accept ISA 84 as a Recognized and Gen- keep its older equipment as long
some degree of risk, that risk needs erally Accepted Good Engineering as the company has determined
to be managed to an acceptable Practice (RAGAGEP) and one way that the equipment is designed,
level; which in turn makes safety a to meet the PSM requirements de- maintained, inspected, tested and
societal term as well as an engineer- fined in 29 CFR 1910.199. Apply- operated in a safe manner. As in-
ing term. Society establishes what ing ISA 84 is more than purchasing dicated by Klein [3], that does not
is commonly accepted as safe and a technology with a given certifica- mean that the existing system can
engineers have to manage risk by tion or using a particular technology be grandfathered and ignored from
introducing risk-reduction methods scheme or architecture. Industry best that point forward.
including human elements, such practices such as ISA 84 consider a The intent of the clause is for
as company culture and work pro- great deal of applied learning. ISA the user to determine if the PSM-
cesses and technologies that make 84 is a performance-based standard covered equipment, which was de-
the production facilities an accept- and describes multiple steps before signed and constructed to comply
able place to work and a responsible and after selecting and implementing with codes, standards or practices
neighbor in our communities. safety system technologies. These no longer in general use, can con-
The CPI has applied learnings steps — commonly referred to as tinue to operate in a safe manner,
from numerous events over the last the safety lifecycle — are also the re- and to document the findings.
40 years. These incidents and ac- sult of applying lessons learned from Therefore, the emphasis should be
cidents have resulted in changes to incidents and events. on the second part of the clause,
regulations and legislation and have Research (as documented in the which states that “the owner/op-
driven the adoption of best practices book “Out of Control” [2]) has shown erator shall determine that the
that address the known factors at that many industrial accidents have equipment is designed, main-
the root of those events. their root cause in poor specifica- tained, inspected, tested and op-
A lot of the best practices are re- tion or inadequate design (about erated in a safe manner.” And that
lated to understanding and evalu- 58%). Additionally, users should determination is a continuous effort
ating hazards and defining the ap- consider that installing a system is that should be periodically revised
propriate risk reduction, including not the “end of the road,” but rather until said equipment is removed
measuring the effectiveness of the another step in the lifecycle of the from operation and replaced with a
methodologies or technologies used facility. Approximately 21% of inci- system that is designed in line with
in reducing the risk. dents are associated with changes current best practices.
Risk-reduction methods using after the process is running, and Another consideration is that the
technology — including digital sys- about 15% occur during operation clause would cover not only hardware
tems — have received extensive cov- and maintenance. and software, but also management
erage in trade publications over time and documentation, including mainte-
as they are important contributors to ISA 84’s grandfather clause nance, all of which should follow cur-
process safety and plant productiv- It is well-known that Functional rent standards — that is, the most re-
ity. However, it is critical to recognize Safety Standard ISA 84.01-2004 cent version of ISA 84 or IEC 61511.
human factors and their impact on contains a grandfather clause
process safety in the design, selec- based on OSHA regulation Emerging technologies
tion, implementation and operation 1910.119. This clause allows users The last few decades have seen
of technology. to continue the use of pre-existing technology changing in all aspects

Chemical Engineering www.chemengonline.com november 2015 67


Potential common-cause failures
There has been a long-standing re-
quirement that a safety system must
be different (or diverse) technology
from its process-automation coun-
terpart to avoid common-cause
failures. But most safety systems
rely on component redundancy
(hardware fault tolerance [HFT]) to
meet reliability and availability re-
quirements, introducing a degree
of common-cause failure directly
into the safety system. Rather than
redundancy, modern systems now
provide a diversity of technologies
designed into logic solvers and
input/output (I/O) modules, along
with a high degree of diagnostics,
to allow a simplex hardware config-
uration to meet safety integrity level
Figure 4. This diagram illustrates the concept of functionally independent protection layers
(SIL) 3 requirements.
of humankind’s daily activities. Pro- provided by known technologies, Product-implementation diver-
cess automation and safety au- such as general-purpose program- sity is also key. Even though most
tomation have not escaped from mable logic controlers (PLCs), it is safety systems are manufactured
such changes (Figure 3). Neverthe- important to keep in mind that those by process-automation vendors,
less, technology-selection criteria platforms might not satisfy the cur- organizational diversity between the
should respond to the risk-reduc- rent risk-reduction requirements in two product teams is only the first
tion needs in the manufacturing the facility and a significant invest- level of separation. Within the safety
facility and consider the improve- ment to maintain the risk-reduction product team, leading suppliers will
ments that some of these technol- performance over the lifecycle of the also be separating the design group
ogies offer, such as enabling bet- plant asset micht might be required. from product-development group
ter visualization of the health of the Also, users will need to develop new and then again from the product-
production asset. competencies in order to under- testing group.
The new breed of systems not stand new risk-reduction require-
only addresses the need to protect ments and apply the next genera- Systematic capabilities
plant assets, but allows users to tion of technology accordingly. Systematic capabilities address how
bring safety to the center stage, side Performance-based safety stan- much protection against human fac-
by side with the productivity of the dards (IEC 61508 and IEC 61511/ tors is built into the safety system.
plant, in many cases by eliminating ISA 84) have changed the way Users should look for the following:
technology gateways and interfaces safety systems should be selected. • Certified software libraries that
that were common a few years ago. The days of simply choosing a certi- offer functions according to the SIL
There are also new developments, fied product, or selecting a preferred requirements of the application
particularly in software, that help technology architecture should be • Compiler restrictions to enforce
prevent human errors in the design, behind us; today’s system selection implementations according to the
and that guide users to fulfill industry is driven by performance require- SIL requirements
best practices using standard off- ments and the risk-reduction needs • User-security management to
the-shelf functionality. Off-the-shelf of the plant. separate approved from non-ap-
products avoid the introduction of proved users for overrides, bypass
error by complex manual program- Understand the hazards and other key functions
ming and configuration. Although this has nothing to do with • Audit-trail capability to record
Although productivity and profit- the safety system technology, it is and document changes to aid in
ability of many manufacturing pro- critical in the selection process to compliance with functional safety
cesses limit the rate of change in the understand the scope of the process standards
process sector, whenever there is an hazards and to determine the nec-
opportunity, facilities should explore essary risk reduction required. This Separate, interfaced or integrated
modern technologies and determine should be done to create the safety Typically based on the SRS and other
if they are a good fit. One should not requirements specification (SRS) business needs, it is important to de-
assume the system shouldn’t be necessary to start a system selec- fine one of these three integration
touched behind the shield of “grand- tion. Even when replacing an exist- philosophies. Integrated systems
father clauses” that are believed to ing system, this is critical because offer many key benefits, drawing on
justify maintaining the system “as- the risk profile of the plant may have common capabilities of the process
is.” Once again, despite the comfort changed since installation. automation system not related to
68 Chemical Engineering www.chemengonline.com november 2015
sified. This reduces the risk that the
system is compromised if one secu-
rity measure fails or is circumvented.
Defense-in-depth tactics can be
found throughout the SD3 + C se-
curity framework (secure by design,
secure by default, secure in deploy-
ment, and communications).
Examples of defense-in-depth
tactics include the following:
• Establishing defenses for the pe-
rimeter, network, host, application
and data
• Security policies and procedures
• Intrusion detection
• Firewalls and malware protection
• User authentication & authorization
• Physical security
The key message is that, like in
the case of safety, security is not re-
Figure 5. Integrated control and safety is a modern alternative to traditional point solutions solved only by certification and it’s not
an isolated activity after the product
the safety functions directly. But only maintenance, both in compliance to development is completed. Security
being interfaced or even kept com- functional safety standards and at a is part of the design considerations
pletely separate are also options, and lower cost over the lifecycle. early in the process and must be
need to be thoroughly considered. supported over the site lifecycle.
Network security
Protection layers The extended use of networked Summary
The use of multiple protection lay- systems is also territory for potential Although following the functional
ers, or functionally independent vulnerabilities. A lot of ground has safety standards is not a silver bullet,
protection layers (Figure 4) to be been covered in this area over the it’s a good place to start the journey
precise, is common in industry. last five years and industry has ex- to improve safety in the process sec-
These include technology elements perienced the emergence of stan- tor. If your industry requires compli-
such as the process control system dards to address new threats and ance to OSHA regulation 1910.119,
and alarms. Safety instrumented has the accelerated development of for the automation portion of any
systems are a last resort to prevent a strong relationship between safety project, complying with the require-
a hazard from escalating. and security. To satisfy the security ments of ISA 84 is a way to address
There are additional layers that requirements of a system network, PSM requirements.
mitigate the impact of a hazard or the user should do the following: Adopting ISA 84 is more than se-
contain it. Once more, there are • Perform a full vulnerability assess- lecting a certified or SIL-capable logic
other layers of protection that are ment/threat modeling and testing solver or having a given redundancy
not based on technology, but on of the different subsystems scheme on the instrumentation. It
work processes or procedures that • Define the best security mecha- requires a lifecycle approach that
might be even more critical than the nism for each of those subsystems starts with the hazards analysis and
technology in use. to cover any identified gaps defines the required risk reduction. It
Most times, system interfaces • Perform a full vulnerability assess- also involves evaluating technologies
are not designed, implemented or ment/threat modeling and testing that better address the hazards and
tested in accordance to industry of the entire interfaced architecture reduce the risk, as well as consid-
best practices or current functional For users of an interfaced system, ering the technical requirements to
safety standards, and therefore they which could be “secured” using “air- mitigate risk to an acceptable level.
have an impact on the performance gaps,” the key is establishing a se- Although existing systems can be
of the system. It has been common curity management system (SMS) of grandfathered, they can’t be ignored
to ignore safety requirements on the interface architecture and sup- from that point forward. Rather, it is
these interfaces. Failure of these in- porting it over the system lifecycle. a continuous effort that should be
terfaces should not compromise the periodically revised until the equip-
safety system. Defense-in-depth in security ment is removed from operation and
Integrated control and safety (Fig- The principle of “defense in depth” replaced with a system designed fol-
ure 5) is a modern alternative to pre- (Figure 6) means creating multiple lowing current best practices.
vious point solutions that takes into independent and redundant preven- When it’s time for selecting a new
consideration the best practices tion and detection measures. The risk-reduction technology, consider
and solves issues related to inter- security measures should be lay- that choosing a given technology
face design, implementation and ered, in multiple places, and diver- scheme is not enough to address the

Chemical Engineering www.chemengonline.com november 2015 69


Industry, 22nd ed., New York, N.Y., 2012.
ing technology gateways and inter-
2. Health and Safety Executive (HSE), "Out of Control:
faces that were common a few years Why Control Systems Go Wrong and How to Prevent
ago, therefore also reducing lifecycle Failure," HSE, London, 2003; available for download
cost and maintenance efforts. at www.hse.gov.uk.
The selection criteria should begin 3. Klein, Kevin L., Grandfathering, It’s Not About Being
Old, It’s About Being Safe, ISA, Research Triangle
with a proper understanding of the Park, N.C., 2005; Presented at ISA Expo 2005, Chi-
hazards and a technology assess- cago, Ill., October 25–27, 2015.
ment to address human factors, 4. Durán, Luis, Safety does not come out of a box, Con-
avoidance of common factors that trol Engineering, February 2014.
could disable the safety instru- 5. Durán, Luis, Five things to consider when selecting a
mented system, and the integration safety system, Control Engineering, October 2013.
of process safety information to the 6. Durán, Luis, The rocky relationship between safety
and security, Control Engineering, June 2011.
process automation systems; this
FIGURE 6. The concept of “defense in depth in
security” is illustrated here
integration is possible and must be Author
done right. Luis Durán is the global prod-
Like in the case of safety, security uct manager, Safety Systems at
ABB Inc. (3700 W. Sam Houston
functional safety requirements. As- (or network security) is not resolved Parkway South, Houston, TX
suming that your existing technology only by certification and it’s not an 77042; Phone: 713 587 8089;
or a “replacement in kind” still com- isolated activity after the product Email: luis.m.duran@us.abb.
plies with the safety requirements of development is completed but part com). He has 25 years of expe-
rience in multiple areas of pro-
your process might lead to a “false of the design considerations early in cess automation and over 20
sense of safety.” Consider the new the process and that must be sup- years in safety instrumented
breed of systems that not only ad- ported over the site lifecycle. n systems. For the last 12 years, he had concentrated
on technical product management and product mar-
dresses the need of protecting the Edited by Gerald Ondrey keting management of safety automation products,
plant assets, but allows users to publishing several papers in safety and critical con-
bring safety to the center stage side References trol systems. Durán has B.S.E.E. and M.B.A. degrees
from Universidad Simon Bolívar in Caracas, Venezu-
to side with the productivity of the 1. Marsh LLC, The 100 Largest Losses 1972-2011: ela and is a certified functional safety engineer (FSE)
plant — in many cases by eliminat- Large Property Damage Losses in the Hydrocarbon by TÜV Rheinland.

Statement of Ownership, Management, and Circulation (Requester Publications Only)


1. Publication Title: Chemical Engineering 2. Publication Number: 0009-2460 3. Filing Date: 9/30/15 4. Issue Frequency: Monthly 5. Number of Issues Published
Annually: 12 6. Annual Subscription Price $149.97. Complete Mailing Address of Known Office of Publication: Access Intelligence, 4 Choke Cherry Road, 2nd
Floor, Rockville, MD 20850-4024 Contact: George Severine Telephone: 301-354-1706 8. Complete Mailing Address of Headquarters or General Business Office
Publisher: Access Intelligence, LLC, 4 Choke Cherry Road, 2nd Floor, Rockville, MD 20850-4024 9. Full Names and Complete Mailing Addresses of Publisher,
Editor, and Maging Editor: Publisher: Michael Grossman, 4 Choke Cherry Road, 2nd Floor, Rockville, MD 20850-4024 Editor: Dorothy Lozowski, 4 Choke Cherry
Road, 2nd Floor, Rockville, MD 20850-4024 10. Owner if the publication is owned by a corporation, give the name and address of the corporation immediately
followed by the names and addresses of all stockholders owning or holding 1 percent or more of the total amount of stock: Veronis Suhler Stevenson, 55 East 52nd
Street, 33rd Floor, New York, NY 10055 11. Known Bondholders, Mortgagees, and Other Security Holders Owning or Holding 1 Percent or More of Total Amount
of Bonds, Mortgages, or other Securities: None 12. Non-profit organization: not applicable. 13. Publication: Chemical Engineering 14. Issue Date for Circulation
Data: September 2015.
Average No. of No. Copies of
15. Extent and Nature of Circulation: Copies Each Issue Single issue
During Preceding Nearest to
12 Months Filing Date
a. Total Number of Copies (Net press run) 35,015 34,489
b. Legitimate Paid and/or Requested Distribution
(1) Outside County Paid/Requested Mail Subscriptions 29,370 29.921
(2) Inside County Paid/Requested Mail Subscriptions 0 0
(3) Sales Through Dealers and Carriers, Street Vendors, Counter Sales or Other Paid
or Requested Distribution Outside USPS 2.615 1.915
(4) Requested Copies Distributed by Other Mail Classes 118 66
c. Total Paid and/or Requested Circulation 32,103 31,902
d. Nonrequested Distribution (By Mail and Outside the Mail)
(1) Outside County Nonrequested Copies 758 752
(2) Inside-County Nonrequested Copies 0 0
(3) Nonrequested Copies Distributed Through the USP by Other Classes of Mail 0 0
(4) Nonrequested Copies Distributed Outside the Mail
(Include Pickup Stands, Trade Shows, Showrooms and Other Sources) 772 902
e. Total Norequested Distribution 1,530 1,654
f. Total Distribution (Sum of 15c and 15e) 33,633 33,556
g. Copies not Distributed (Office, Returns, Spoilage, Unused) 1,382 933
h. Total (Sum of 15f and g) 35,015 34,489
i. Percent Paid and/or Requested Circulation 95.45% 95.07%
16. Electronic Copy Distribution: None Reported
17. Publication of Statement of Ownership for a Requester Publication is required and will be printed in the November 2015 issue of this publication
18. Signature of Fulfillment Manager: George Severine Date: 9/29/15
PS Form 3526-R, July 2014

70 Chemical Engineering www.chemengonline.com november 2015


Engineering Practice

Improving the Operability of Process Plants


Turndown and rangeability have a big impact on the flexibility and efficiency of
chemical process operations
Mohammad Toghraei reliably across a wide range of flow- Equipment
Consultant rates without sacrificing the overall

D
quantity or quality of product(s).
uring the design of a chemi- From a process standpoint, a Utility
cal process plant, the main chemical process plant is a combi-
focus is on which process nation of equipment, utility networks
units or unit operations and control systems. To design a Control valve, instruments

must be integrated to convert the plant with sufficient flexibility, each FIGURE 1. Different elements of a plant need dif-
feed streams into product stream(s). of these three elements needs to ferent levels of operating flexibility. Since the utility
Design engineers work to achieve allow flexibility. Generally speaking, network provides support duty to the equipment,
this goal; however, in terms of mak- the control system (including con- it needs a higher turndown ratio. Control valves
ing sure the plant operates smoothly, trol valves and sensors) and util- and other instruments have a duty to take care of
equipment across a wider operating range; thus
which is equally important for opera- ity network should offer the largest they require an even higher rangeability
tion engineers and operators, there amount of operating flexibility, while
are less well-know parameters fac- the equipment itself could offer the or even to support the shutdown of
ing the design engineers. lowest amount of flexibility (Figure 1). downstream equipment. But it may
There are five primary process This requirement for larger flexibility also occur accidentally due to, for
parameters in each plant — flow, for control items and utility network example, a drop in feed flowrate.
(liquid) level, pressure, temperature, considerations is important because But process plant operators like
and composition. Composition can of the supporting role of the util- to know by how much the flowrate
be considered a collective term that ity system and the controlling role of the equipment (and in the larger
reflects all parameters (chemical and played by instruments in a plant. sense, the entire plant capacity) can
physical), and provides an indicator Two important concepts are used be decreased without compromis-
of the quality of the stream. Com- to quantify flexibility: turndown (TD) ing the process goal or generating
position can be used to describe ratio and rangeability. These are off-specification product. Thus, TD
the moisture of a gas stream or the discussed below, and illustrated in ratio can be defined as the ratio of
octane number of a gasoline stream, Figure 2. high flow to normal flow, as shown in
or even the electric conductivity of a Equation 1.
water stream. Turndown ratio
During operation, equipment pro- The flexibility of equipment or a plant QHigh
cess parameters generally deviate can be defined using the TD ratio. TD ratio =
from the design values (normal level) The most common definition for TD
QLow (1)
over time. Five levels can be defined ratio is “ratio of the normal maximum
for each process parameter: nor- parameter (numerator) to the normal QHigh = the flowrate of the system at
mal level, high level, high-high level, minimum parameter (denominator).” high level
low level and low-low level. In es- However, the meaning of “normal QLow = the flowrate at low level
sence, the operational parameters maximum parameter” and “normal The numerical value of the TD ratio
of a plant relate to the behavior of minimum parameter” is not always is typically reported as a ratio, such
the plant between the low level and clear and the interpretation may vary as 2:1.
high level of each parameter of the in different companies and plants It is important to note that the
individual equipment components, (This is discussed below). denominator term is flowrate in low
individual units or the entire plant. For an individual equipment com- level, and not low-low level. This is
In most cases, the operability of a ponent, or multi-component equip- important as it is the differentiator
plant can be defined using at least ment systems, low-flow or low-ca- between the concept of TD ratio
three key parameters: flexibility in pacity operation happens frequently and rangeability, which is discussed
operation, resistance against surge over the lifetime of a plant. The re- later. Generally, flowrate in low level
(or upset) and the speed of recovery duced-capacity operation may be (as shown in Figure 2) is considered
from upset. intentional or accidental. to be the minimum level of flow at
For instance, reduced-capacity which the process goals can still
Maintaining operating flexibility operation could be planned for the be reached.
Flexibility of operation in this context purpose of off-loading the equipment However, there is another interpre-
means the ability of a plant to operate for inspection, testing, monitoring, tation of TD ratio that is often used
72 CHEMICAL ENGINEERING WWW.CHEMENGONLINE.COM SEPTEMBER 2015
TABLE 1. TURNDOWN RATIO OF SELECT EQUIPMENT 2:1 for a plant. The required TD
Item Turndown ratio ratio could be as high as 3:1 or 4:1
Pipe Large, but depends on the definition of maximum and minimum flow
for a plant.
Storage containers Very large; The maximum value is the total volume of the container, but the
(Tank or vessels) minimum value could be dictated by a downstream component. For example,
Equipment flexibility
a centrifugal pump may dictate a minimum volume to provide required NPSH The TD ratio can also be determined
Centrifugal pump Typically: 3:1 to 5:1
for a given piece of equipment, using
other values that are stated for the
Positive-displ. pump Theoretically infinite
component. For example, even when
Heat exchanger Small, depends on the type; for instance, less than 1.5:1 a TD ratio is not explicitly stated for a
Burner [1] Depends on the type; for example: centrifugal pump, when the pump is
Pressure jet type: ≈ 2:1 said to have a capacity of 100 m3/h
Twin fluid-atomizing type: >8:1
and a minimum flow of 30 m3/h, this
TABLE 2. UTILITY SURGE CONTAINER TO PROVIDE TD RATIO means that the centrifugal pump has
a TD ratio of 3:1.
Surge container Residence time
The TD ratio of a reciprocating
Instrument air (IA) Air receiver 5–10 min. or higher depending on pump could theoretically be defined
whether it is connected to UA or not
as infinite because it can work over a
Utility water (UW) Water tank Several hours very wide range of flows. However, in
Utility steam (US) Utility steam cannot be stored for a long time without condensing; the options practice, such a pump cannot handle
for storing steam are the steam drum of a boiler, or if a conventional boiler is any flowrate that fails to fill the cylin-
not available, a vessel as an “external steam drum” could do the same task
der of the pump in one stroke. Partial
Utility air (UA) No dedicated container; could “float” with IA filling of the cylinder may cause some
Cooling water (CW) Cooling tower basin Depends on the size of the network damage to mechanical components
Cooling/heating glycol Expansion drum Depends on the size of the network of the pump over the long term. Thus
the minimum required flow is a func-
TABLE 3. TURNDOWN RATIO OF SELECT INSTRUMENTS tion of cylinder volume and stroke
Item Turndown ratio speed of a specific pump.
Flowmeter: orifice-type 3:1 [2 ] The TD ratio for pipelines presents
Flowmeter: vortex-type 10:1 to 50:1 [2 ] a more complicated situation. With
piping systems, there are several dif-
Flowmeter: Coriolis-type 5:1 to 25:1 [2 ]
ferent ways to define the minimum
Control valve Depends on type and characteristics; generally 50:1, and less than 100:1 flow. For instance, it could be defined
TABLE 4. ARBITRARY VALUES OF FLEXIBILITY PARAMETERS as the minimum flow that does not
Low flexibility Medium flexibility High flexibility
fall into the laminar flow regime. Or,
it could be considered as the mini-
Equipment (TD ratio) < 1.2:1 to 2:1 2:1 to 3:1 5:1 to 8:1
mum flow that keeps a check valve
Instrument, control valves (rangeability) ≈ 4:1 10:1 to 30:1 20:1 to 100:1 open (if a check valve is used).
For liquid flows in pipes, the mini-
by operations staff. During opera- number. In general, the academic mum flow is more commonly inter-
tion, people expect the TD ratio to definition of TD ratio generally uses preted as the minimum flow that
answer the question in this scenario: a high-to-low values set up, while in makes the pipe full, or the sealing
“My plant is running normally and the field, operators often define TD flowrate (that is, no partial flow), or a
all parameters are normal. How- ratio using normal-to-low values. flow threshold below which the fluid
ever, occasionally, because of differ- The TD ratio can be defined for will freeze in an outdoor pipe. If the
ent reasons (including shortage of parameters other than flowrate, but flow bears suspended solids, the
feed, reduced plant or unit capac- it generally refers to flowrate. One minimum flowrate could be defined
ity), the flowrate falls. What is the reason for this is because flowrate as that at which sedimentation of
minimum value I can withstand can be the most important param- suspended solids may occur.
without compromising the quality of eter of a plant, helping to define the Table 1 provides examples of
the product?” economy of the system. The other typical values and rules of thumb
They basically interpret the TD ratio reason is because the flowrate might regarding the TD ratio for various
so that the numerator is the “normal be influenced by constraints out- types of process equipment. Note
level parameter” (and not the “high side of the plant (for instance, a lack that in Table 1, the TD ratio of stor-
level parameter”). However, the dif- of stored feed), which the control age containers is relatively large. This
ference in the interpretation does not system cannot necessarily adjust high TD helps to explain why large
generate a big difference in numerical (thus making a reduction in flowrate containers are used for surge damp-
value of TD ratio, as the normal and unavoidable). ening as part of a typical plant-wide
high level of parameters are often While the TD ratio is not always a control system.
not very far from each other. Due to requested parameter, and is often In some cases deciding on a re-
this potential confusion, the TD ratio not mentioned in project documents quired TD ratio needs good judg-
should be considered an approxi- for design purposes, operators are ment. One example is chemical-
mate parameter and not a precise usually looking for a TD ratio of least injection packages. The TD ratio

CHEMICAL ENGINEERING WWW.CHEMENGONLINE.COM SEPTEMBER 2015 73


Trip
High-high flow Thus, the rangeability cannot be de-
Alarm fined only as a function of flowrate
High flow
— pressure drop also needs to be
incorporated. The rangeability of

Turndown ratio
Rangeability

Process control valves is a function of the


goal Normal flow
achieved control-valve flow coefficient (Cv).
Rangeability can also be defined for
Alarm other parameters, such as tempera-
Low flow ture, but generally defining rangeabil-
Trip ity with regard to flowrate is the most
Low-low flow
important parameter. Table 3 shows
FIGURE 2. Process plants typically define different threshold values for flowrate levels, and set appropri- some typical rangeability values for
ate alarms and trips when the threshold values of this important parameter are reached. The concept of commonly used instruments.
turndown ratio and rangeability are shown, in relation to these key threshold flowrate values
It should be stressed that TD ratio
is important for chemical-injection port this concept. and rangeability are two separate pa-
packages to protect against chemi- Different segments of a utility net- rameters, for two separate systems.
cal overdosing or underdosing. work experience different levels of They cannot be used interchange-
Chemical-injection packages typi- turndown, and consequently each ably and attempts to relate or con-
cally provide a TD ratio of about 100:1 segment may need a different TD vert them to each other do not have
or lower. In some cases, 10:1 can ratio. For instance, as shown in Figure much meaning.
be provided by stroke adjustment, 3, the main header could need the
and another 10:1 through the use minimum TD ratio, while sub-headers Providing required flexibility
of a variable frequency drive (VFD) may need a higher TD ratio. There are three main ways that one
to control the motor. But the ques- The good news is that achieving a can provide a specific TD ratio for
tion that arises is why such a large high TD ratio for the utility network process equipment, and each is dis-
TD ratio is necessary if the host flow and related instruments is not dif- cussed below:
experiences, for example, only a 2:1 ficult. The overall utility network is • Using equipment with an inher-
TD ratio. This high TD ratio is gener- mainly a series of pipe circuits that ently high TD ratio
ally desired because of uncertainty inherently show a large TD ratio. If • Replacing equipment with multiple
in the required chemical dosage and instruments are included in the util- similar, smaller-capacity equip-
the variety of available chemicals. ity network, this poses no problem. ment in a parallel arrangement
The required dosage of a chemical Many instruments (including control • Providing a recirculation route
depends on the type of chemical valves and sensors) have an intrin- Using equipment with an inher-
and the host stream properties. sically large TD ratio — generally ently higher TD ratio. Some pro-
Thus, during the design phase of a greater than 20:1. cess elements have an inherently
project, the designer doesn’t exactly higher TD ratio. Two of them, tanks
know what the optimum dosage Instrument rangeability and pipes, were mentioned above.
would be, even though a chemical Instruments typically need to oper- It is not always easy to recognize
provider recommends a specific dos- ate over a wider range of process if a piece of equipment has an inher-
age. Often, he or she prefers to con- conditions than other equipment or ently high or low TD ratio. However,
servatively have a chemical-injection utilities. This is because their duty is the following rules of thumb can be
system with a high TD ratio. not limited to normal operation, or a used as guidelines:
There is generally less uncertainty band defined by low and high val- • Smaller-volume equipment tends
when using chemicals of known ues. Rather, they have to be opera- to have a smaller TD ratio than
composition, rather than proprietary tional across the entire, wider band larger-volume equipment
mixtures. If the dosage is fairly firm from low-low to high-high threshold • Equipment with internal baffles
and the chemical used is a non-pro- values. Therefore, rangeability, R, tends to have a lower TD ratio
prietary type, the TD ratio could be can be defined as: (a good example is some gravity
decreased, to reduce the overall cost separators, such as baffled skim
of the chemical-injection system. QHigh high tanks)
R= • Equipment in gas service may
Utility network flexibility
QLow low (2) show a higher TD ratio than equip-
The flexibility of a utility network is ment used in liquid service
also defined by the TD ratio. As men- Where: • Equipment with an internal weir
tioned above, when a plant requires QHigh-high = the flowrate of the in- (especially fixed ones) may have a
a TD ratio of, say, 2:1, the TD ratio of strument or control valve at the high- very low TD ratio
the utility network should be higher. high level threshold value • Equipment that uses some prop-
To accommodate a larger TD ratio, QLow-low = the flowrate at the low- erties of the inlet stream for their
the utility network generally requires low level threshold value functioning, may have a lower TD
containers to absorb fluctuations For control valves, the formula is a ratio. For example, in cyclones or
that may be caused by utility usage bit different because a control valve hydrocyclones, the energy of the
changes in process areas. Table 2 is a device that passes flow and inlet stream (“energy” as a prop-
provides additional details to sup- also drops the pressure of the flow. erty of the inlet stream) is used to
74 CHEMICAL ENGINEERING WWW.CHEMENGONLINE.COM SEPTEMBER 2015
control system are designed to han-
dle a maximum flowrate of 30 m3/h, it
Utility consumer branch means the TD ratio of the pump can
theoretically be increased to infinite,
Sub-header by zeroing the minimum flow.
Another example is a vertical
Utility
Main header
Low turndown ratio
falling-film evaporator. This type of
generation unit evaporator has a vertical tube bundle
that is similar to the ones found in a
shell-and-tube heat exchanger. The
tube-side flow is two-phase flow. The
liquid flows down by gravity, and the
High turndown ratio
vapor (of the same liquid) is pushed
down by liquid drag. The flow inside
the tubes is an “annular regime,”
FIGURE 3. Shown here is a map of turndown ratio for a typical utility network.The pipes closer to the util- meaning the liquid covers the inter-
ity generation system (main header) need less turndown ratio compared to sub-headers and branches
nal perimeter of tubes and the vapor
generate centrifugal force, so any The TD ratio may actually be higher is in the center of the tubes.
reduced flow will reduce the cen- because each shell-and-tube heat In the case of low flow, there is a
trifugal force, which may reduce exchanger has an inherent TD ratio chance of “dry patches” forming on
the effectiveness of the system too, even though it is very small. This the tube’s internal surface. Because
• Equipment containing loose, po- technique has additional benefits. of this, vertical, falling-film evapora-
rous media may show a lower TD The parallel arrangement provides tors are typically equipped with recir-
ratio in liquid service, and the TD higher availability for the system, culation pipes to provide a minimum
ratio may be lower when the po- because the failure of two or three practical TD ratio (Figure 5).
rous media is comprised of larger parallel equipment components is However, this method cannot be
solid particle sizes. Examples less likely than the potential for failure applied for all equipment. For ex-
include sand filtration systems, when the system relies on a single ample it is not a good technique to
catalyst contactors and related equipment component. increase the TD ratio of a furnace or
systems Using two control valves in paral- fired heater, because recirculation
• Despite a common misconcep- lel in a single control loop (through a of fluid around a furnace may in-
tion, perforated-pipe flow distribu- “split range” control) is also another crease the furnace coil temperature
tors do not necessarily have lim- example of this technique in the area and cause burning out if the firing
ited TD ratios [3] of instrumentation. system doesn’t have sufficient TD
As noted, the utility network should However, there are some disad- ratio. Table 4 provides some rules of
have a relatively large TD ratio. Fortu- vantages associated with this tech- thumb to gauge the flexibiliy of differ-
nately, utility networks consist mainly nique. In particular, capital cost and ent elements of a process plant.
of pipes in different sizes, which have operating cost considerations may
inherently large TD ratios. If control rule against it. Resistance against surge
valves are needed on the network, Providing recirculation pipe. Im- While TD ratio refers to the static be-
their lower TD ratios may generate plementing a recirculation pipe from havior of a plant, there are two addi-
bottlenecks. In such situations, it the equipment outlet to its inlet is tional parameters (resistance against
may be necessary to install parallel a widely used method to increase surge, and speed of recovery from
control valves with split control, be- the TD ratio of the system. In many upset) that refer to its dynamic
cause of the required large TD ratio. cases, a pump and definitely a con- behavior. However, there is less
Using parallel equipment. Instead trol system, are needed to imple- emphasis on dynamic theories,
of using a component with a capac- ment this technique. As long as you and only practical aspects of
ity of 100 m3/h, this technique is es- can afford an extra pump and control dynamic behavior.
sential to use an arrangement that system on the recirculation pipe, this A process upset could result from
employs two parallel components, technique can be used. The recircu- a surge. Surge can arbitrarily be de-
each with the capacity of 50 m3/h. lation pipe needs a control system, fined as the deviation of a parameter
By doing so, a TD ratio of at least otherwise all flow goes through the (such as flowrate) beyond its normal
2:1 can often be provided. It should recirculation pipe back to the inlet of level. The final value of the param-
be noted that the equipment by itself the unit of interest (Figure 4). eter may or may not be in a band
may have some inherent TD-ratio ca- One example of this technique is between high level and low level and
pability, which may have to be added using a minimum-flow line for a cen- the change often occurs quickly.
to the provided 2:1 TD ratio. trifugal pump. A centrifugal pump with When a parameter moves quickly,
For example, instead of using one a capacity of 100 m3/h and a mini- an upset could happen. The surge/
shell-and-tube heat exchanger with mum-flow line of 30 m3/h (thus, with a upset could be defined for each pa-
the capacity of 100 m3/h, three heat TD ratio of 1:3) can be equipped with rameter including flowrate, tempera-
exchangers — each with the ca- a minimum-flow line with an appropri- ture, pressure and even composition.
pacity of 33 m3/h —can be used ate control system to increase its TD A surge in the composition is often
to achieve a TD ratio of at least 3:1. ratio. If the minimum-flow line and the called a slug. Level surge is generally

CHEMICAL ENGINEERING WWW.CHEMENGONLINE.COM SEPTEMBER 2015 75


Unit Feed

Vs.
Recirculation pump
Compressor

Recirculation pipe
Unit
Feed
FIGURE 4. By providing a recirculation pipe, the
turndown ratio of a piece of equipment can be Distillate
increased. If the fluid pressure is not enough, a
pump (or compressor) may be needed, and a con-
trol system is definitely needed

a consequence of other surges and


it can be dampened in surge-equal-
ization tanks or drums.
Surge can also be defined by its
Brine
shape (in a diagram of parameter
change versus time), and by its mag- FIGURE 5. Shown here is a system for brine recirculation in a vertical falling-film evaporator. The brine-
nitude. The magnitude of surge can recirculation line in the vaporizer plays an important control role. Without the recirculation line, the
be stated as a relative number or vaporizer has a very narrow turndown ratio, which is not generally acceptable for optimal operation
an absolute number. For example, tial surge remains less than 2°C per Some systems show different be-
a flow surge of 2% per minute is a minute (an absolute value). havior against surge, when it is a
relative number and means if a surge A 2%-per-minute surge means positive surge (an increase in the pa-
occurs in every minute, the flowrate that the flowrate could start at 100 rameter value), or a negative surge
is increased or decreased by 2%. In m3/h and then increase to 102 m3/h, (a decrease in the parameter value).
another example, a system can be then to 104 m3/h and so on. Or the Therefore, it is good idea to clarify it.
said to be resistant to temperature surge may start at 100 m3/h and For example, an API separator could
surge (thus no upset conditions will then decrease to 98 m3/h, then 96 be more resistant to the impact of
be generated) as long as any poten- m3/h and so on. decreasing inlet stream compared to

MONITOR VISCOSITY SIMPLY


SENSE MIXER MOTOR HORSEPOWER
WITH UNIVERSAL POWER CELL
EASY INSTALLATION PROFILING A PROCESS
• No holes in tanks or pipes • Power changes reflect viscosity changes
• Away from sensitive processes • Good batches will fit the normal “profile” for
that product
VERSATILE 24
POWER DECREASE
• One size adjusts to motors, from 22
SHOWS BATCH
small up to 150hp 20
IS DONE

• Works on 3 phase, fixed or variable 18

frequency, DC and single phase power POWER


16
SENSOR
SENSITIVE 14

• 10 times more sensitive than 12

just sensing amps 10 DRY MIX ADD LIQUID


HIGH SPEED LOW SPEED MIXER
CONVENIENT OUTPUTS 8 MOTOR
BEGIN HIGH
• For meters, controllers, computers 6
SPEED MIX
4-20 milliamps 0-10 volts 4

0
BATCH 1 BATCH 2 BATCH 3

CALL NOW FOR YOUR FREE 30-DAY TRIAL 888-600-3247


WWW.LOADCONTROLS.COM

Circle 24 on p. 94 or go to adlinks.chemengonline.com/56201-24

76 CHEMICAL ENGINEERING WWW.CHEMENGONLINE.COM SEPTEMBER 2015


the impact of increasing the inlet stream.
The first line of defense against a surge is provided by
the control system or control valves. However, control
valves alone cannot totally eliminate a surge, but will
only stop a surge from impacting a downstream sys-
tem. Ultimately, the surge needs to be handled, but by
other methods.
There are basically two surge-management methods
that can be implemented for each piece of equipment
or group of equipment in a plant:
• Boxing-in a surge in a specific equipment component
or series of equipmen
• Transferring the surge to an external or auxiliary
system
Understanding the applicability of each of these tech-
niques requires some knowledge about the inherent
dynamic characteristics of the systems from a process
control viewpoint. The three dynamic features of each
equipment or unit are resistance, capacitance and in-
ertia (dead time) [4]. A brief qualitative explanation of
these three features is presented next.
If a system is more dominantly a “resistance” type,
this system will be able to prevent the surge from trans-
ferring to downstream equipment. A piece of pipe is
one example of a resistance-type element. A pipe
could inherently stop the surge if it is narrow enough.
However, because a pipe's main function is to transfer
fluid, the designer generally sizes the pipe based on SEE US AT CHEMSHOW - BOOTH #543
its duty (transferring fluid) and then, if needed, a control
valve is placed on the pipe to stop a potential surge. Circle 13 on p. 94 or go to adlinks.chemengonline.com/56201-13
The capability of a system to dampen the surge de-
pends on the “capacitance characteristics” of the sys-
tem. The higher the capacitance characteristic, the
more it is able to dampen a surge.
It’s more than a check valve...
Here, a capacitance-type element refers to whatever
element that can be used to temporarily store excess IT’S A CHECK-ALL®
mass (such as liquid volume or gas pressure) or energy
(such as thermal or chemical energy). Our spring loaded check
For instance, large-volume equipment generally have valves are assembled to
a higher capacitance feature. Implementing a surge your exact needs, ensuring
tank, equalization tank, surge drum (or even pond) is absolute precision and
one means of providing a system with sufficient capac- reliability. They work as
ity to dampen the surge. they should in any
Another example of using a high-capacitance system orientation. Most lead
is when transferring a surge to heat-exchange media. times are less than one
Utility heat exchangers use streams such as cooling week. That’s what makes
water, steam, and other media, to transfer the heat
Check-All® the only choice.
to or from process streams. These utility streams are
also able to absorb a temperature surge in the system.
The capacitance feature of a utility network can be pro-
vided in part by pipes in the network (the pipes func-
tion mainly as resistance elements but they have some
capacitance too), and also their surge tank, as
discussed above.
A system is called robust against upset when it can
tolerate a large surge (as defined for each process pa- SINCE 1958

rameter) and no upset occurs, thereby allowing the


process to proceed smoothly. GET ME A CHECK-ALL®!
If an upset cannot be tolerated, one solution is to
implement a rate-of-change control loop in the system. Manufactured in
West Des Moines, Iowa, USA
The following llist provides some general rules of thumb 515-224-2301 • sales@checkall.com • www.checkall.com
on the capability of a system to handle surges:
1. Generally speaking, equipment with larger volume
Circle 7 on p. 94 or go to adlinks.chemengonline.com/56201-07

CHEMICAL ENGINEERING WWW.CHEMENGONLINE.COM SEPTEMBER 2015 77


and fewer internals is better able means the system requires a longer cal may kill a large portion of the bio-
to dampen upsets. time to recover from an upset. material growing there. n
2. Containers with plug-flow regime However, in addition to this inher- Edited by Suzanne Shelley
are more susceptible to upset ent characteristic of a system, other
from surge compared to mixed- features can also impact (and de- References
flow-regime containers. crease) the speed of recovery from 1. Mullinger, P., and Jenkins, B., “Industrial and Process
3. The equipment that exerts cen- an upset. Sometimes these features Furnaces,” 1st Ed., Amsterdam: Butterworth-Heinemann,
2008, p. 171.
trifugal effect on the process fluid (rather than the dynamic behavior of
2. Upp, E., and LaNasa, P., Fluid flow measurement,” 2nd
is more sensitive toward the upset the system) govern the behavior of Ed., Gulf Professional Publishing, Boston, 2002, pp.
(Examples include centrifuges and the system. For example, a hot lime 157–158.
centrifugal pumps). softener within a water-treatment 3. Perry, R., Green, D. and Maloney, J., “Perry's Chemical
4. Containers that hold loose media system has an established sludge Engineers' Handbook,” 7th Ed., McGraw-Hill, New York,
are less robust against upsets. blanket. It takes time to “heal” 1997, pp. 6–32.
5. Non-flooded containers can han- a broken sludge blanket if an upset 4. Liptak, B., “Instrument Engineers Handbook — Vol 2.
Process Measurements and Analysis,” 4th Ed., CRC
dle and dampen a surge better creates “breaks” in it. Press, Boca Raton, 2003. Chapter 2.
than flooded containers. Another example is “vessel-
media” systems. These are systems Author
Speed of recovery from upset that are used in operations such as Mohammad Toghraei, is an in-
structor and consultant with En-
The speed of recovery from an ion exchangers, loose-media filtra- growth Training (Email: mohtogh@
upset situation primarily depends on tion systems, packing-type absorp- gmail.com; Phone: 403-808-
the dynamic characteristics of the tion towers, catalyst beds and so 8264; Website: engedu.ca), based
in Calgary, Alberta, Canada. He
system, and more specifically, the on. A big surge in flow may displace has more than 20 years of experi-
“process dead time” and “pro- the media in a way that leads to flow ence in the chemical process in-
cess time constant” of a sys- channeling. Putting the displaced dustries. Toghraei has published
articles on different aspects of
tem. The dead time is a result media back into a homogenous form chemical process operations. His main expertise is in
of inertia characteristics of the takes time. the treatment of produced water and wastewater from
system, while the process time con- Similarly, a surge to a biological petroleum industries. He holds a B.S.Ch.E. in from Isfa-
stant is a function of capacitance and system will generally require a long han University of Technology (Iran), and an M.Sc. in en-
vironmental engineering from Tehran University (Iran).
resistance features of the system. recovery system, because a surge in He is a professional engineer (PEng) in the province of
A larger dead time or time constant temperature or slug of a toxic chemi- Alberta, Canada.

“Your #1 replacement for C.T.F.E, Silicone & FluoroSilicone Lubricants”


Inert Light Oil & Grease
Designed For: Focus Industries:
• Chemical Metering Pumps • Chemical Toll Processing Content Licensing for
• Hydrocarbon Metering Pumps • Fuel Refineries Every Marketing Strategy
• Diaphragm Pumps • Cryogenic Plants
• Mechanical Seals • High Pressure Gases
• Valves • Water Treatment Marketing solutions fit for:
• O-rings • Oxygen • Outdoor
• Direct Mail
• Print Advertising
• Tradeshow/POP Displays

m • Social Media
TM

s • Radio & Television

Logo Licensing | Reprints


Eprints | Plaques
m
TM

s Leverage branded content from Chemical Engineering


to create a more powerful and sophisticated statement
MS-1010 MS-2010
about your product, service, or company in your
next marketing campaign. Contact Wright’s Media
m
TM

to find out more about how we can customize your


s acknowledgements and recognitions to enhance your
marketing strategies.
Connecticut • Illinois • California • Canada For more information, call Wright’s Media at 877.652.5295
For technical information: 800.992.2424 or 203.743.4447 or visit our website at www.wrightsmedia.com
supportCE@mschem.com • miller-stephenson.com

Circle 27 on p. 94 or go to adlinks.chemengonline.com/56201-27
78 CHEMICAL ENGINEERING WWW.CHEMENGONLINE.COM SEPTEMBER 2015
Solids Processing

Solids Discharge: Characterizing


Powder and Bulk Solids Behavior
How shear-cell testing provides a basis for predicting flow behavior
Core flow
A B C

Flowing core

Static
material

Hopper half angle

Mass-flow Drain down


angle of repose Critical rat-hole
diameter

Robert McGregor FIGURE 1. Three common types of flow behavior for powder in a bin are mass flow (1a), core flow or fun-
Brookfield Engineering Laboratories nel flow (1b) and rathole formation (1c)

P
have been used in the minerals in- from a container. Figure 1a shows
owder jams are the once- dustry for decades. Recent improve- how particles move uniformly down-
in-a-month catastrophe that ments in the design of this equipment ward in lockstep with one another as
can bring processing opera- and the processing power available the fill level in the bin reduces. The
tions to a standstill. Whether in today’s personal computers (PCs) fundamental principle is referred to
it’s erratic flow behavior or complete make them more affordable and user as “first in, first out.” One obvious
stoppage of powder discharge, the friendly. The bottom line is that shear advantage is that blends of powders
consequence is the same. Shutdown cells can predict powder flow behav- retain their component ratio without
may be necessary before startup can ior using a proven scientific principle segregation. This is one of the most
take place. Why? Formulations often that measures inter-particle sliding important considerations for formu-
involve multiple component powders friction. Mathematical calculations lators who must ensure that final
blended together. If the flow becomes embedded in the software used product has the intended makeup
disrupted, one of the possible conse- with shear cells provide estimates as designed in research and devel-
quences is segregation of compo- for “arching dimension” in mass flow opment (R&D).
nents. Smooth and continuous flow and “rathole diameter” in core flow. More typical of powder process-
of powder from start to finish is the These values become design limits ing in most plant operations is “core
operating goal to minimize the onset for hopper openings and half angle. flow” or “funnel flow” as shown in
of other problems like segregation. This article addresses the rheol- Figure 1b. Particles at the top of the
Traditional testing techniques ogy of powder-flow behavior and container move toward the center
used to predict flow performance, explains how the shear cell is used and then downward through the
such as flow cup, angle-of-repose to make these types of powder mea- middle, discharging out the hopper
measurement and tap test, actually surements and calculations for stor- well before the powder that had been
have limited relevance to whether a age equipment design (see also, “A further down in the vessel. Larger
powder will flow. They are relatively pragmatic Approach to Powder Pro- particles have a tendency to move
affordable in terms of equipment cessing,” Chem. Eng., August 2015, more readily than smaller particles,
purchase and easy for operators to pp. 59–62). potentially resulting in segregation.
use. The data, however, do not pre- This type of behavior is called “last in,
dict whether reliable discharge will Types of powder flow first out.” One possible unfortunate
take place from the storage vessels In a perfect world for powder proces- consequence is that powder around
containing the powder. sors, “mass flow” would take place the outer wall of the vessel becomes
Shear cells for testing powder flow all the time when powder discharges stagnant, consolidates over time,
62 CHEMICAL ENGINEERING WWW.CHEMENGONLINE.COM SEPTEMBER 2015
is needed to allow the powder to dis- sample. The “tap density,” ρtapped,
charge from the cup. In a practical is calculated by dividing the reduced
sense, this instrument is used as a volume of powder at the end of the
“go” or “no-go” indicator for powder test into the sample weight. The two
processing on a regular basis. density values are compared to one
Angle of repose. This is a simple another, giving an indicator for the
test method that observes powder in consolidation that can take place
a pile and measures the angle of the over time when the powder settles.
pile relative to horizontal. Note that Two standard calculations that are
both the angle-of-repose method typically used by industry to evaluate
and the flow-cup test work with pow- tap test data are called Carr Index
ders that are loosely consolidated. (Carr%) and Hausner Ratio (HR), as
They do not attempt to evaluate the defined in Equations (1) and (2):
powder as it settles, which is what
happens when powder is placed in a
containment vessel of any kind. This
phenomenon, called “consolidation,”
is an important distinction to keep in (1)
mind because it has direct impact on
FIGURE 2. The flow cup test is relatively easy to how flow behavior can change.
setup and perform, and the data are used to cal-
culate the Carr index, Equations (1), and Hausner
Tap test. The tap test takes a cylin- (2)
ration, Equation (2) der of powder and shakes it to de-
termine how much settling will occur.
and then becomes lodged in place. The change in volume of the powder Shear cell test for flowability
This type of structure is referred to as from start to finish is a measurement Shear cells measure the inter-parti-
a “rathole” shown in Figure 1c. The of the powder’s tendency to consoli- cle friction of powder materials. This
rathole may extend from top to bot- date. The “loose fill” density, ρpoured, type of test has direct application
tom of the bin and may change in of the powder at the start of the test to predicting flow behavior in grav-
diameter of opening as a function of is calculated by dividing the cylin- ity discharge for powders stored in
powder depth. der volume into the weight of the vessels of any kind. Shear cells were
Processors prefer mass flow for
obvious reasons. Cohesive materi-
als will generally exhibit core flow
in plant equipment as originally de-
signed. The hopper wall angle and
its material of construction have a di- ENSURE
rect impact on flow behavior. There-
fore the challenge is to manage the YOUR PIPING
problem with the equipment that
exists, which means modifying the INTEGRITY
formulation, or redesigning the bin
equipment, if practical.

Traditional tests for flowability


As mentioned earlier, there are three In today’s operating environment, it’s more
common methods for predicting • 100% inspection of
important than ever that the piping within
flow: flow cup, angle of repose and internal and external
your Mechanical Integrity Program complies pipe surfaces
the tap test. with standards such as API-570 and API-574.
Flow cup. The most popular test- • Inspection results
ing method is the flow cup, which is Quest Integrity offers a comprehensive solution tailored to comply
quick and easy to use. The cup is for piping circuits using our proprietary, with API-570 and
API-574
basically an open cylinder with a re- ultrasonic-based intelligent pigging technology
movable disc that is inserted into the combined with LifeQuest™ Fitness-for-Service • LifeQuest Fitness-
bottom (Figure 2). A family of discs, software. for-Service results
each with a different hole diameter in tailored to comply
Ensure your piping integrity by identifying with API-579
the middle, is provided with the cup.
Once the disc is in place, the cup is degradation before loss of containment occurs.
filled with powder and the operator
observes whether the material dis-
charges through the hole. Proces- QuestIntegrity.com
sors may know from experience CHALLENGE CONVENTION

what difficulties they are likely to face


depending on the hole diameter that
Circle 34 on p. 94 or go to adlinks.chemengonline.com/56201-34

CHEMICAL ENGINEERING WWW.CHEMENGONLINE.COM SEPTEMBER 2015 63


A 1. The vane lid, which is attached
to a torsional spring, rotates with
the trough as long as the frictional
force between powder particles
is greater than the torsion in the
spring. When the lid stops moving
with the trough, the torsion in the
spring exceeds the inter-particle
friction. The moment when this
stoppage in lid movement takes
place defines the yield stress be-
tween powder particles and is a
measure of what is referred to as
“failure strength” of the powder.
2. The wall-friction lid behaves in
a similar fashion to the vane lid
while measuring the sliding fric-
tion between the powder particles
and the surface material of the lid.
When rotation of the wall lid stops
C during the test, the yield stress for
powder flow on this particular sur-
face is established.
Movements of trough and lid during
the shear-cell test are very small and
almost unobservable to the naked
eye. Increasing consolidating pres-
sures are applied to the powder sam-
B ple to construct a picture of how the
powder’s failure strength will change.
D
This equates with vessels that have
FIGURE 3. For shear-cell testing, powder is placed into a ring-shaped trough for annular shear cell (3a), increasing fill levels of material.
which is placed into a commercial powder flow tester (3b), which uses either a vane lid (3c) or a wall-
friction lid (3d) Three key graphs
first applied to powders and bulk Basic operation of the instrument Basic tests run with the shear cell
solids in the minerals industry over during the test procedure is to bring address flow behavior of powder
50 years ago. More recent advance- the lid down onto the powder sample in gravity discharge from a storage
ments in the use of computers to and compress the material to a speci- vessel. The following summarizes
automate testing and improvements fied pressure. This action consolidates the three primary graphs used to
in shear cell design have allowed this the powder, forcing the particles to characterize flow characteristics.
type of instrument to become more move closer to one another. With the Flow function. The flow-function test
commonplace throughout the pow- powder in this compressed state, the evaluates the ability of the powder to
der-processing industries. trough rotates at a low speed — per- form a cohesive arch in the hopper
The current popular design is haps 1 rpm. The following is observed, that could restrict or prevent flow out
the annular shear cell. Powder is depending on the lid in use: the opening in the bottom. Result-
placed into a ring-shaped cell called Flow function graph
the “trough,” shown in Figure 3a, 8.0
weighed in order to calculate the 7.0 Very cohesive File
“loose fill” density, and then placed Very cohesive
Unconfined failure strength, kPa

Cohesive
onto a test instrument such as that 6.0
Easy flowing
shown in Figure 3b. The lid, which Free flowing
5.0 Non flowing Data set #1
will fit on top of the cell, is attached Data set #2
4.0 Cohesive Data set #3
to the upper plate on the instrument Data set #4
and can be one of two types: 3.0
1. The vane lid (Figure 3c) has individ- Easy flowing
2.0
ual pockets separated by vanes.
2. The wall-friction lid (Figure 3d) is a flat 1.0
surface and is made of material sim- 0.0
Free flowing
ilar to the hopper wall in the powder
0 1 2 3 4 5 6 7 8 9 10 11 12 13
storage vessel on the production Major principal consolidating stress, kPa
floor. Examples might include mild FIGURE 4. The flow-function graph shows how the failure strength for the powder changes as a function
steel, stainless steel or Tivar. of increasing consolidating stress

64 CHEMICAL ENGINEERING WWW.CHEMENGONLINE.COM SEPTEMBER 2015


Friction angle graph
90
File the dimension of the opening, flow
Data set #1
80 Data set #2 restrictions may result.
70
Data set #3 2. The rathole diameter is the poten-
Data set #4
tial diameter of a hole in the center
60 of the vessel through which pow-
Friction angle, deg

50 der will move when the type of be-


havior is “core flow.” The rathole
40
diameter may change in value as
30 a function of the powder depth in
20
the vessel. Powder particles that
are located radially outside of this
10 diameter dimension may become
0 lodged in place over time and po-
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 tentially not flow at all.
Normal stress, kPa 3. The hopper half angle is the required
FIGURE 5. Data from the wall-friction test show how the effective friction angle for the hopper wall to angle — relative to vertical in the
allow gravity-driven powder flow on its surface changes as a function of consolidating stress hopper section — that is needed to
achieve mass flow behavior.
Density curve graph
These three values can be used for
1200
File
Data set #1
the design of powder storage equip-
Data set #2 ment or to characterize reference
Data set #3
1000 Data set #4 powders that constitute benchmarks
for future production batches.
Bulk density, kg/m3

800
Concluding remarks
600
Shear cells provide a scientific basis
400
for analytically predicting flowability
of powder in gravity discharge. Their
200 use is becoming more accepted
because improved designs for the
0 instrument make them affordable,
0 1 2 3 4 5 6 7 8 9 10 11 12 13
Major principal consolidating stress, kPa user friendly, and automatic in op-
FIGURE 6. The density of a powder in a vessel will vary depending on the consolidating stress, which in eration under control of a computer.
turn is a function of the fill level The most notable change in the past
year is the reduction in time needed
ing data form the flow-function graph cern. Data from this test may also to run a standard flow-function test
(Figure 4), which shows how the fail- have some correlation with findings from 45 min to 15 min. Productivity
ure strength for the powder changes obtained in the angle-of-repose test gains with the current generation of
as a function of increasing consolidat- described earlier in this article. instrumentation certainly give rise to
ing stress (height of powder-fill level Density. Density of powder in a ves- their potential use in quality control
in the vessel). Industry has agreed sel will vary depending on the con- as well as R&D. The chemical pro-
to classify regions of flow behavior solidating stress, which in turn is cess industries on the whole view
as shown in the figure, ranging from a function of the fill level. Figure 6 the shear cell as an important tool
“free flowing” to “non-flowing.” As shows an example. If the change in for improving rapid scaleup of new
might be expected, many powders density increases by more than 50% formulations into full production. n
exhibit “cohesive” or “very cohesive” relative to the “loose fill” condition, Edited by Gerald Ondrey
flow and are likely to be problematical then there is an expectation that
in terms of processability. flow problems may exist. Note that Acknowledgement
Wall friction. The wall-friction test the density test will very likely have All photos courtesy of Brookfield En-
measures the flowability of the pow- a point on the curve that correlates gineering Laboratories, Inc.
der on the material comprising the with the findings in the tap test de-
hopper wall. Data from the wall- scribed earlier in this article. Author
friction test (Figure 5) show how the Robert McGregor is the general
effective friction angle for the hopper Data analysis manager, global marketing and
sales for High-End Laboratory In-
wall to allow gravity-driven powder Parameters of interest that can be struments at Brookfield Engineer-
flow on its surface changes as a func- calculated from the data in the above ing Laboratories, Inc. (11 Com-
tion of consolidating stress (height of tests include the following: merce Blvd., Middleboro, MA
powder-fill level in the vessel). Expe- 1. The arching dimension is the 02346; Phone: 508-946-6200
ext 7143; Email: r_mcgregor@
rience indicates that friction angles length of a bridge section that the brookfieldengineering.com; Web:
below 15 deg will have relatively easy powder has sufficient strength to www.brookfieldengineering.com).
flow behavior whereas friction angles create in the hopper section of a He holds M.S. and B.S. degrees in mechanical engi-
neering from MIT (Cambridge, Mass.; www.mit.edu).
above 30 deg will be cause for con- vessel. If the bridge is longer than

CHEMICAL ENGINEERING WWW.CHEMENGONLINE.COM SEPTEMBER 2015 65


Advantages Gained in
Automating Industrial
Wastewater Treatment Plants
Process monitoring and automation can improve efficiencies in wastewater
treatment systems. A number of parameters well worth monitoring, as well as tips for
implementation are described

JP Pasterczyk Pretreatment Primary Secondary Disinfection


GE Water & Process 2
Grit Hypochlorite
Technologies removal Sedimentation Aeration Clarifier

Dechlorination
IN BRIEF Pump Bar screen
Chlorine
contact

IMPLEMENTING Collection system


PROCESS CONTROL Advanced tertiary/recycled Open water
Tertiary filter
PROCESS PARAMETERS

IMPLEMENTING Reuse Storage 4 3


PROCESS ANALYTICS UV

T
here is growing interest in auto- FIGURE 1. Most wastewater treatment systems use a com-
mating wastewater treatment pro- mon sequence of steps to treat influent wastewater and then
discharge, store or reuse it in line with local regulations.
cesses across a broad range of Automating this approach helps an operator more effectively
industries. In particular, a paradigm manage and treat wastewater, saving time and money in the
shift is starting in automating industrial process
wastewater treatment in various sectors of
the chemical process industries (CPI), such undergo tertiary treatments to be further oxi-
as foods (especially grain processing, sug- dized or disinfected, or to undergo additional
ars, sweeteners and edible oils), beverages purification, including by granular activated
(mainly soft drink bottlers and breweries), carbon (GAC) or membrane separation, be-
and hydrocarbon and chemical processing fore reuse or discharge to a public sewer or
(particularly petroleum and petrochemical open body of water.
plants). The driving forces behind this evo- A fully optimized, industrial wastewater-
lution are economic. Wastewater process treatment plant will operate at a lower total
optimization most often leads to a more cost of materials, labor and energy to do the
efficient use of chemicals, reduced energy following:
consumption and less solid waste. • Remove or reduce large solids and par-
Most wastewater-treatment systems use a ticulate matter (primary)
common sequence of steps (Figure 1), with • Remove or reduce fats, free oil (and
the purpose of first removing solids materi- grease), dispersed oil and emulsions
als in the influent wastewater, recovering • Remove organic materials efficiently
lost product, removing solids, fats, oils and (secondary) and withstand higher vari-
greases (FAG), treating the water biologi- able loading, with enhanced, biological
cally and chemically enhancing flocculation, activated sludge systems through:
coagulation and physical removal of the bio- ❍❍ Control of dissolved oxygen levels,
logical solids and sludge. The clarified and minimizing energy required for
decanted wastewater is the effluent that may aeration
44 CHEMICAL ENGINEERING WWW.CHEMENGONLINE.COM SEPTEMBER 2016
Circle 22 on p. 98 or go to adlinks.chemengonline.com/61499-22
❍❍Maintaining food-to-mass ra- • Disinfect pathogens and produce pending upon the reuse application
tio, pH and nutrient balance, effluent water quality for reuse or and corresponding water quality re-
minimizing chemical usage below discharge limits to open quirements, tertiary disinfection for
and system upsets body, waterway or public waste- pathogens and final polishing with
• Produce a readily settleable bio- water treatment plant GAC or reverse osmosis (or both)
logical floc (small microbial mass); More advanced integration of may be needed.
less energy to coagulate and technologies can be applied to meet
separate (Figure 2) requirements for reuse, whether Implementing process control
• Generate minimal volume of within the facility (for example, wash In general industry, process auto-
sludge and biosolids to dewater, water), for irrigation and agricultural mation is ubiquitous and integral to
minimizing energy, chemical us- purposes or higher purity applica- upstream control mechanisms and
age and disposal costs tions, like clean water utilities. De- production yield. Statistical process
control (SPC) can use process ana-
lytical technology to generate high-
value data in real- and near-time,
and is critical to closely control pro-
cesses, quality and maximum pro-
duction yield. There is a prevailing
interest across industries to identify
opportunities to gain process knowl-
edge by understanding process ef-
fluent streams. These waste streams
Your Valve and combine to become the wastewater
treatment influent. Companies are
Instrumentation investing in multiple tools, devices,
analyzers and sensors, and integrat-
Partner ing these measurements into pro-
cess automation and control sys-
tems for the wastewater treatment
GEMÜ Valves features plant (WWTP). They are looking at
collecting useful data with the right
diaphragm valves, parameters, and applying SPC tools,
previously reserved for production
butterfly valves, angle purposes, to continually analyze and
seat and globe valves, optimize their wastewater treatment
processes. The proper design and
lined metal valves, execution of experiments can help
show the pertinent relationships be-
flow measurement, tween multiple parameters that yield
and multiport the best process performance. The
application of this empirical process
block valves. knowledge can translate into sig-
nificant performance improvements
and efficiencies.

Process parameters
Depending upon the physical and
chemical characteristics of waste
Visit us at streams, a number of treatment
modules are employed to remove,
Booth #134 reduce and change sample stream
constituents including, but not lim-
ited to, the following:
3800 Camp Creek Parkway • Building 2600 • Suite 120
• Bar screens and strainers for grit
Atlanta, GA 30331 • 678-553-3400 • info@gemu.com
and particles
• API (American Petroleum Institute)
www.gemu.com separators and corrugated plate
separation for free oil and grease
• Chemicals and dissolved or in-
duced gas (or air) flotation for oily
Circle 17 on p. 98 or go to adlinks.chemengonline.com/61499-17
46
chemEng201609_gemu_isld.indd 1
CHEMICAL ENGINEERING
8/1/2016 3:12:07 PM
WWW.CHEMENGONLINE.COM SEPTEMBER 2016
solids and emulsified oils Organic carbon: The influent, or-
• Biological activated sludge and ganic carbon loading is a key pro-
advanced membranes for organ- cess parameter for a WWTP, and
ics, nitrogen and heavy metals has historically been quantified using
• Physical and chemical clarification chemical oxygen demand (2 hours)
and advanced membranes for or biochemical oxygen demand (5
microbial flocs days; BOD5). With the availability of
• Chlorine (gas, hypochlorite and online, process instrumentation for
chlorine dioxide solution) and total organic carbon (TOC) analysis,
ozone for trace organics and a direct measurement of the organic
pathogens
• Granular activated carbon (GAC)
concentration can be used to im-
prove downstream performance.
PROVEN
for organics
• Chemical disinfection for patho-
Specifically, by knowing the exact
values of TOC, the plant can be op-
PERFORMANCE
gens (typically chlorination) erated to accommodate variation in
• UV (ultraviolet) for pathogens, the amount of organics, and remove
trace organics and residual them efficiently. For instance, there
ozone destruction is often an introduction of chemicals
• Chemical pH neutralization (such as potassium permanganate,
• Reverse osmosis for inorganics hydrogen peroxide or chlorine) after ROTOFORM
and minerals primary solids removal to reduce the GRANULATION
By employing a combination of total oxygen demand, often referred
discrete (grab) and online mea- to as pre-oxidation. This step can FOR PETROCHEMICALS
surements before, after and at in- be eliminated with lower influent or- AND OLEOCHEMICALS
termediate process points, each ganic concentrations, or minimized
module’s performance can be by using it only when the load is
monitored and improved over time. above a threshold limit based on the
Some of the parameters measured plant’s treatment capacity.
by the available probes, meters, Dissolved oxygen: In a biological or
sensors and analyzers include: activated sludge system, there is an
flow, pH/ORP (oxidation-reduction opportunity to adjust the amount of
potential), conductivity, dissolved dissolved oxygen generated by the
oxygen (DO), suspended solids, aeration system to a level commen-
specific ions [for example: nitrogen surate with the organic load, while High productivity solidification of
(ammonia, nitrates, nitrites), phos- avoiding excessive aeration that can products as different as resins, hot
phorus (phosphates), chlorine], shear or tear the biological flocs, melts, waxes, fat chemicals and
total organic carbon, sludge den- which in turn reduces the overall ef- caprolactam has made Rotoform® the
sity index and turbidity. fectiveness of organics and biosol- granulation system of choice for
chemical processors the world over.
Free oil and grease: Before intro- ids removal. Continuous monitoring
Whatever your solidification
duction of the waste stream to the of influent organic loading and dis-
requirements, choose Rotoform for
biological or activated sludge sys- solved oxygen levels in select zones reliable, proven performance and a
tem, free oil and grease should be of the activated sludge basin provide premium quality end product.
removed or reduced to below a an opportunity to optimize the aera-
maximum threshold of 50 mg/L, and tion system, the largest energy ex-  High productivity –
ideally below 25 mg/L, to avoid in- pense in the operation of a WWTP. on-stream factor of 96%
terfering with the microbial activity. Food-to-mass ratio: Industrial  Proven Rotoform technology –
Some of the negative repercussions wastewater-treatment systems are nearly 2000 systems installed
of allowing excess levels of free oil looking at the ratio of organic load in 30+ years
 Complete process lines or
to come into contact with the bio- or “food,” to the total biomass pres-
retrofit of existing equipment
mass are rapid oxygen depletion, ent in the biological system. The
 Global service / spare parts supply
encapsulation of the bacteria, and biomass of the mixed liquor can be
foaming. Depending upon the levels estimated by measuring mixed li-
of free oil, and geometry of the oil quor suspended solids and sludge
droplets, one can use API separa- density. This F:M or food-to-mass
tors or corrugated-plate separation. ratio, is a critical process control
Dispersed and emulsified oils are re- parameter that can indicate system Sandvik Process Systems
moved and reduced through a com- overload or when there are insuffi- Division of Sandvik Materials Technology Deutschland GmbH
Salierstr. 35, 70736 Fellbach, Germany
bination of chemicals, for lowering cient organics to “feed” the micro- Tel: +49 711 5105-0 · Fax: +49 711 5105-152
info.spsde@sandvik.com
pH and enhancing the dissolved or bial population. The plant operation
www.processsystems.sandvik.com
induced gas flotation unit(s). can use near realtime information
Circle 38 on p. 98 or go to adlinks.chemengonline.com/61499-38

CHEMICAL ENGINEERING WWW.CHEMENGONLINE.COM SEPTEMBER 2016 47


SANDVIK_Chemical_ad_55.6x254_MASTER.indd 1 09/02/2015 14:48
FIGURE 2. Wastewater treat- and take actions to address and improve rus, commonly referred to as the CNP ratio,
ment often involves settling process conditions before they become a conventionally follows 100:10:1 (using BOD5
of solids in a tank such as
this one stress to the biological system. instead of carbon). The amount of nitrogen
Nutrient addition: The organic or carbon or phosphorus present in a system depends
loading can be used to assure the most upon the upstream processes and can be
appropriate levels of nutrients, specifically optimized using chemical addition, often
nitrogen and phosphorus, and improve the through pH control. For example, if there is a
efficiency of the biological system. The pro- deficient amount of phosphorus, and a basic
portion of carbon to nitrogen to phospho- pH, phosphoric acid can be used to reduce

MONITOR VISCOSITY SIMPLY


SENSE MIXER MOTOR HORSEPOWER
WITH UNIVERSAL POWER CELL
EASY INSTALLATION PROFILING A PROCESS
• No holes in tanks or pipes • Power changes reflect viscosity changes
• Away from sensitive processes • Good batches will fit the normal “profile” for
that product
VERSATILE 24
POWER DECREASE
• One size adjusts to motors, from 22
SHOWS BATCH
small up to 150hp 20
IS DONE

• Works on 3 phase, fixed or variable 18

frequency, DC and single phase power POWER


16
SENSOR
SENSITIVE 14

• 10 times more sensitive than 12

just sensing amps 10 DRY MIX ADD LIQUID


HIGH SPEED LOW SPEED MIXER
CONVENIENT OUTPUTS 8 MOTOR
BEGIN HIGH
• For meters, controllers, computers 6
SPEED MIX
4-20 milliamps 0-10 volts 4

0
BATCH 1 BATCH 2 BATCH 3

CALL NOW FOR YOUR FREE 30-DAY TRIAL 888-600-3247


WWW.LOADCONTROLS.COM

Circle 23 on p. 98 or go to adlinks.chemengonline.com/61499-23

48 CHEMICAL ENGINEERING WWW.CHEMENGONLINE.COM SEPTEMBER 2016


the pH while supplementing the phosphorus These processes may require a combina- FIGURE 3. The flocculation
concentration. Supplemental nitrogen can tion of pretreatment, pH control and physi- and coagulation steps, which
allow small microbial flocs
be added using nitric acid, urea or anhy- cal treatment steps. to form and join together for
drous ammonia. Final polishing and purification: Tertiary removal by clarification is
Clarification: The flocculation and coagula- treatment typically refers to final polishing, achieved through a combina-
tion of chemical addition and
tion steps, which allow small microbial flocs but can be interpreted differently by indus- physical separation
to form and join together for removal by try and is dependent upon the composition
clarification (Figure 3), is achieved through of the water and the next purpose, whether
a combination of chemical addition and some form of reuse or discharge. Disinfec-
physical separation. The chemical
feedrates are typically flow-paced,
metered in direct proportion to the Bungartz
system flowrates. By utilizing online Masterpieces
organic measurements, the chemi-
cal addition can be “trimmed” for
better performance at a lower chem-
ical cost.
Nitrogen removal: Systems with
excess nitrogen can employ a bio-
logical or membrane-enhanced the Power
nitrification/denitrification process of regulation
after the aerobic, activated sludge Representing the groundbreaking special
system. Nitrifying bacteria can con- centrifugal pumps from the Bungartz collection,
series production
vert ammonia nitrogen to nitrite,
then nitrate, which can then be de-
A top-class factory expertly picks
nitrified to nitrogen gas. These bac- up on the current movement towards
teria are more sensitive to process pure autonomy. The dry-running
changes, particularly temperature, centrifugal pump Van regulates
and may require an alternate food itself and requires no interference.
source, such as methanol and mo- It works free of cavitation, which
lasses, to supplement when nitro- makes it even more fascinating.
gen levels are low. Online nitrogen When working with toxic media it
and organic measurements can be unfolds its full potential. An asset
used to regulate the amount of or- for many companies!
ganic food sources used in these
applications.
Heavy metals: Some residual
heavy metals, such as arsenic and
selenium, can be removed through More under +49 (0) 211 57 79 05 - 0
and: www.bungartz.de/masterpieces5
chemical, physical, biological and/
or membrane-enhanced processes.
Circle 30 on p. 98 or go to adlinks.chemengonline.com/61499-30

CHEMICAL ENGINEERING WWW.CHEMENGONLINE.COM SEPTEMBER 2016 49


Continuous monitoring of influent organic loading and dissolved oxygen levels in select
zones of the activated sludge basin provide an opportunity to optimize the aeration
system, the largest energy expense in the operation of a WWTP

tion can be accomplished by several differ- tion can also be used to prevent or mitigate
ent chemical and physical methods, such as the negative impact of process upsets. In
chlorine gas, sodium or calcium hypochlo- the case of an unexpected event or exces-
rite solution, chlorine dioxide, ozone, and sive “shock” load to the system, the influent,
UV light (254 nanometer wavelength). After online TOC measurement can be used to
disinfection, the end-of-pipe purpose will de- automatically divert to an equalization basin
termine if additional treatment is necessary. or temporary storage vessel, sometimes re-
Some industrial utilities have reused waste- ferred to as a calamity tank.
water with a GAC step to absorb organics Effluent discharge monitoring: Meet-
and excess chlorine, and reverse-osmosis ing regulatory requirements for effluent dis-
membrane separation to remove inorganics charge levels is critical to any business op-
and trace organics, achieving higher purity. eration. There are continuous monitors for
Managing process upsets: Upsets in the many of the common effluent-wastewater-
wastewater process can affect removal effi- quality characteristics, including pH, dis-
ciencies at each treatment step. More severe solved oxygen, total dissolved solids, total
upsets can overload a system, even lead- suspended solids, and total organic carbon
ing to the loss of an entire, activated sludge (often used to trend chemical and biochemi-
biomass. The cost and time to reseed and cal oxygen demand). Finally, effluent pH for
restore lost biomass are significant, often discharge should almost always be neutral,
upwards of tens of thousands of dollars and ideally pH 6.8 –7.2.
several months. Real- and near-time detec- Solids disposal: The biosolids produced

Circle 36 on p. 98 or go to adlinks.chemengonline.com/61499-36
50 CHEMICAL ENGINEERING WWW.CHEMENGONLINE.COM SEPTEMBER 2016
from management of the activated can be impacted by changes in energy consumption, and solid waste
sludge volume in the aeration basins the upstream processes, as well as disposal, by maintaining the dynamic
and during clarification, are typi- seasonal variations in environmen- treatment system in an optimal opera-
cally dewatered using a belt press tal conditions such as ambient tem- tional state. n
or centrifuge, before being used as perature and rainfall. Edited by Dorothy Lozowski
fertilizer or disposed of as waste. By employing continuous process
The cost of sludge handling and de- monitoring tools and integration to Author
watering, in energy, chemical usage automation and process control sys- J.P. Pasterczyk is the corporate
and disposal, is often the second tems, more industries are finding bet- key accounts manager ­— analyti-
cal instruments for GE Water &
highest expense in a wastewater ter ways to effectively manage and Process Technologies (6060 Spine
treatment facility, after aeration. The treat their process and wastewater Road, Boulder, CO 80301-3687;
ability to use the dewatered sludge effluents. This automation provides Email: john.pasterczyk@ge.com;
Phone: 720-622-0166). He has
as fertilizer is dependent upon the more predictable and controllable pro- 25 years of international experi-
content of undesirable constituents, cesses, reducing the frequency of up- ence in water and wastewater
treatment, from water quality
such as heavy metals or residual sets and assuring a more consistent monitoring to pretreatment, biological treatment pro-
pathogens, including fecal coliforms effluent that meets discharge require- cesses and disinfection. Pasterczyk has spent the last
such as E. coli (Escherichia coli). ments. The efficiency of the biological 17 years with GE’s Analytical Instruments, primarily fo-
cused on total organic carbon analysis and integration
Instead of land application for ag- system to remove organics depends of water quality monitoring with process automation in
ricultural purposes, the solid waste upon the quality of the upstream pro- petroleum refining and petrochemicals, chemical, mu-
can be compacted or incinerated cesses — oil and grease and solids nicipal water, pharmaceutical and semiconductor indus-
tries. He is an expert in industrial wastewater treatment,
(or both) to reduce volume for dis- removal, and the controllable, ambi- applied statistics, statistical process control and optimi-
posal. A more sustainable approach ent conditions, such as dissolved zation, Lean Six Sigma methods and advanced quality
is sending the sludge to anaerobic oxygen, food-to-mass ratio and nutri- management systems. Pasterczyk received a B.S. de-
gree in physics from Drexel University and a Master of
digesters to produce methane gas, ent balance (CNP ratio). Utilization of Engineering degree from the Lockheed Martin Engi-
which can be fed to gas-fired tur- process analytical instrumentation and neering Management Program at the University of Colo-
bines to generate electricity. automation controls enables these rado, specializing in business performance excellence
and applied statistics/Six Sigma.
facilities to reduce total chemical and
Implementing process analytics
The data for each measured pa-
rameter can be tracked through
a data collection and visualization
system. A wide range of commer-
cially available software, as well as
discrete supervisory control and
data acquisition (SCADA) systems,
are employed by treatment facilities
to monitor critical and complemen-
tary water-quality characteristics.
With these tools, each treatment
module indicates the measured pa-
rameters before, during and after
treatment, while steady-state con-
ditions can be established to bet-
ter detect and anticipate upset and
sub-optimal conditions. Many pa-
rameters integrate into a feedback
or feed-forward loop for chemical
feed, becoming statistical process
control applications. New, multi-
variate relationships can be tested
and inferred through sound experi-
mental design and intrinsically valid,
statistical analyses. Good process
data leads to process understand-
ing and SPC brings and maintains
processes in control. Empirical evi-
dence can support or modify pre-
liminary assumptions and control
schemes. This acquired learning
Circle 06 on p. 98 or go to adlinks.chemengonline.com/61499-06

CHEMICAL ENGINEERING WWW.CHEMENGONLINE.COM SEPTEMBER 2016 51

You might also like