Chapter 4 Source

Models
Most accidents in chemical plants result in spills of
toxic, flammable, and explosive materials.

Accidents begin with an incident, which usually

results in the loss of containment of material from
the process. The material has hazardous properties,
which might include toxic properties and energy
content.
Once the incident is known, source models are
selected to describe how materials are discharged
from the process.
The source model provides a description of the rate of
discharge, the total quantity discharged (or total time
of discharge), and the state of the discharge (that is,
solid, liquid, vapor, or a combination). A dispersion
model is subsequently used to describe how the
material is transported downwind and dispersed to
some concentration levels.
For flammable releases fire and explosion models
convert the source model information on the
release into energy hazard potentials, such as
thermal radiation and explosion over-pressures.

Effect models convert these incident-specific

results into effects on people (injury or death) and
structures.
 4-1 Introduction to Source Models
Source models are constructed from fundamental or
empirical equations representing the physicochemical
processes occurring during the release of materials. For
a reasonably complex plant many source models are
needed to describe the release. Some development
and modification of the original models is normally
required to fit the specific situation. Frequently the
results are only estimates because the physical
properties of the materials are not adequately
characterized or because the physical processes
themselves are not completely understood. If
uncertainty exists, the parameters should be selected
to maximize the release rate and quantity. This ensures
Release mechanisms are classified according
to aperture releases:
wide
limited

In the wide aperture case a large hole

develops in the process unit, releasing a
substantial amount of material in a short
time. An excellent example is the over
pressuring and explosion of a storage tank.

For the limited aperture case material is

released at a slow enough rate that
upstream conditions are not immediately
Limited aperture
releases. For these
releases material is
ejected from holes
and cracks in tanks
and pipes, leaks in
flanges, valves, and
pumps, and severed
or ruptured pipes.
Relief systems,
designed to prevent
the over pressuring
of tanks and process
vessels, are also
potential sources of
released material.
 Figure 4-3 shows how the physical
state of the material affects the
release mechanism. For gases or
vapors stored in a tank, a leak
results in a jet of gas or vapor. For
liquids a leak below the liquid level
in the tank results in a stream of
escaping liquid. If the liquid is
stored under pressure above its
atmospheric boiling point, a leak
below the liquid level will result in
a stream of liquid flashing partially
into vapor. Small liquid droplets or
aerosols might also form from the
flashing stream, with the possibility
of transport away from the leak by
wind currents. A leak in the vapor Figure 4-3 Vapor and liquid are ejected from
process units in either single- or two-phase states
space above the liquid can result in
There are several basic source models that are
used repeatedly and will be developed in detail
here. These source models are
flow of liquid through a hole,
flow of liquid through a hole in a tank,
flow of liquids through pipes,
flow of vapor through holes,
flow of gases through pipes, flashing liquids,
and
liquid pool evaporation or boiling.
4-2 Flow of Liquid through a Hole
A mechanical energy balance describes the various energy forms associated
with flowing fluids:
P is the pressure (force/area),
is the fluid density (mass/volume),
u is the average instantaneous velocity of the
fluid (length/time),
gc is the gravitational constant (length
mass/force time2),
is the unitless velocity profile correction
factor with the following values:
= 0.5 for laminar flow,
= 1.0 for plug flow, and
1.0 for turbulent flow,
g is the acceleration due to gravity
(length/time2),
z is the height above datum (length),
F is the net frictional loss term (length
force/mass),
Ws the shaft work (force length), and
M is the mass flow rate (mass/time).
Figure 4-4 Liquid
escaping through a
hole in a process unit.
The energy of the
liquid resulting from its
pressure in the vessel
is converted to kinetic
energy, with some
frictional flow losses in
the hole.
The discharge coefficient C0, is a complicated function of
the Reynolds number of the fluid escaping through the
leak and the diameter of the hole. The following guidelines
are suggested:
For sharp-edged orifices and for Reynolds numbers
greater than 30,000, C0 approaches the value 0.61. For
these conditions the exit velocity of the fluid is
independent of the size of the hole.
For a well-rounded nozzle, the discharge coefficient
approaches 1.
For short sections of pipe attached to a vessel (with a
length-diameter ratio not less than 3), the discharge
coefficient is approximately 0.81.
When the discharge coefficient is unknown or uncertain,
Example 4-1
At 1 P.M. the plant operator notices a drop in pressure in a
pipeline transporting benzene. The pressure is
immediately restored to 100 psig. At 2:30 P.M. a 1/4-in-
diameter leak is found in the pipeline and immediately
repaired. Estimate the total amount of benzene spilled.
The specific gravity of benzene is 0.8794.
Solution
The drop in pressure observed at 1 P.M. is indicative of a
leak in the pipeline. The leak is assumed to be active
between 1 P.M. and 2:30 P.M., a total of 90 minutes. The
area of the hole is
The density of the benzene is

The leak mass flow

rate is given by
Equation 4-7. A
discharge coefficient
of 0.61 is assumed for
this orifice-type leak:

The total quantity of benzene spilled is

4-3 Flow of Liquid through a Hole in a Tank

Figure 4-5 An orifice-

type leak in a process
vessel. The energy
resulting from the
pressure of the fluid
height above the leak is
converted to kinetic
energy as the fluid exits
through the hole. Some
energy is lost because
of frictional fluid flow.
Mass discharge rate at any time t:

The first term on the right-hand side of

Equation 4-19 is the initial mass discharge
rate at hL = hL0.
The time tc for the vessel to empty to the level of the
leak is found by solving Equation4-18 for t after setting
hL = 0:

If the vessel is at atmospheric pressure, Pg = 0 and

Equation 4-20 reduces to
Example 4-2
A cylindrical tank 20 ft high and 8 ft in diameter is
used to store benzene. The tank is padded with
nitrogen to a constant regulated pressure of 1 atm
gauge to prevent explosion. The liquid level within
the tank is presently at 17 ft. A 1-in puncture occurs
in the tank 5 ft off the ground because of the
careless driving of a forklift truck.
Estimate
(a)the gallons of benzene spilled,
(b)the time required for the benzene to leak out,
(c) the maximum mass flow rate of benzene through
the leak.

The specific gravity of benzene at these conditions

 Solution
The density of the
benzene is

The area of the tank is

The area of the leak is

The gauge pressure is

a. The volume of benzene above the leak is

This is the total

benzene that
will leak out.

b. The length of
time for the
benzene to leak
out is given by
Equation 4-20:
c. The maximum discharge occurs at t = 0 at a liquid level
of 17.0 ft. Equation 4-19 is used to compute the mass flow
rate:
4-4 Flow of Liquids through Pipes
Figure 4-6 Liquid flowing
through a pipe. The frictional
flow losses between the fluid
and the pipe wall result in a
pressure drop across the pipe
length. Kinetic energy
changes are frequently
negligible.
The frictional loss term F in Equation 4-28 represents the
loss of mechanical energy resulting from friction and
includes losses resulting from flow through lengths of pipe;
fittings such as valves, elbows, orifices; and pipe entrances
and exits. For each frictional device, a loss term of the
following form is used:

where
Kf is the excess head loss due to the pipe or pipe fitting
(dimensionless)
u is the fluid velocity (length/time).
Table 4-1 Roughness Factor for Clean Pipes
Pipe (mm)
material

Riveted 1-10
steel
Concrete 0.3-3
Cast iron 0.26
Galvanized 0.15
iron
Commercial 0.046
steel
Wrought 0.046
For fluids flowing through pipes
the excess head loss term Kf is
given by

where
f is the Fanning friction factor
(unitless),
L is the flow path length
(length),
d is the flow path diameter
(length).
The Fanning friction factor f is a function of the Reynolds
number Re and the roughness of the pipe . Table 4-1
provides values of for various types of clean pipe.
Figure 4-7 is a plot of the Fanning friction factor versus
Reynolds number with the pipe roughness, /d, as a
parameter.
For laminar flow the Fanning friction factor is given by

For turbulent flow the data shown in Figure 4-7 are

represented by the Colebrook equation:
Figure 4-7 Plot of
Fanning friction
factor f versus
Reynolds number
For fully developed turbulent flow in
rough pipes, f is independent of the
Reynolds number,
For smooth pipes, = 0

For smooth pipe with a Reynolds

number less than 100,000

A single equation has been

proposed by Chen to
provide the friction factor f
over the entire range of
Reynolds numbers
4-9 Realistic and Worst-Case Releases
The realistic releases represent the incident outcomes with a high probability
of occurring. Thus, rather than assuming that an entire storage vessel fails
catastrophically, it is more realistic to assume that a high probability exists
that the release will occur from the disconnection of the largest pipe
connected to the tank.
The worst-case releases are those that assume almost catastrophic failure of
the process, resulting in near instantaneous release of the entire process
inventory or release over a short period of time.
The selection of the release case depends on the requirements of the
consequence study. If an internal company study is being completed to
determine the actual consequences of plant releases, then the realistic cases
4-10 Conservative Analysis
All models, including consequence models, have uncertainties. These
uncertainties arise because of:
(1) an incomplete understanding of the geometry of the release (that
is, the hole size),
(2) unknown or poorly characterized physical properties,
(3) a poor understanding of the chemical or release process,
(4) unknown or poorly understood mixture behavior, to name a few.

Uncertainties that arise during the consequence modeling procedure

are treated by assigning conservative values to some of these
unknowns. By doing so, a conservative estimate of the consequence is
obtained, defining the limits of the design envelope. This ensures that
the resulting engineering design to mitigate or remove the hazard is
overdesigned. Every effort, how-ever, should be made to achieve a
result consistent with the demands of the problem.
For any particular modeling study several receptors might
be present that require different decisions for
conservative design. For example, dispersion modeling
based on a ground-level release will maximize the
consequence for the surrounding community but will not
maximize the consequence for plant workers at the top of
a process structure.
To illustrate conservative modeling, consider a problem
requiring an estimate of the gas discharge rate from a hole in
a storage tank. This discharge rate is used to estimate the
downwind concentrations of the gas, with the intent of
estimating the toxicological impact. The discharge rate
depends on a number of parameters, including
(1) the hole area,
(2) the pressure within and outside the tank,
(3) the physical properties of the gas,
(4) the temperature of the gas,
The reality of the situation is that the maximum discharge
rate of gas occurs when the leak first occurs, with the
discharge rate decreasing as a function of time as the
pressure within the tank decreases.
The complete dynamic solution to this problem is difficult,
requiring a mass discharge model cross-coupled to a material
balance on the contents of the tank. An equation of state
(perhaps non ideal) is required to determine the tank
pressure given the total mass.
Complicated temperature effects are also possible. A
A much simpler procedure is to calculate the mass discharge
rate at the instant the leak occurs, assuming a fixed
temperature and pressure within the tank equal to the initial
temperature and pressure. The actual discharge rate at later
times will always be less, and the down-wind concentrations
will always be less. In this fashion a conservative result is
ensured.
For the hole area a possible decision is to consider the area
of the largest pipe connected to the tank, because pipe
disconnections are a frequent source of tank leaks. Again,
Unfortunately, this procedure can result in a
consequence that is many times larger than the
actual, leading to a potential overdesign of the
mitigation procedures or safety systems.
This occurs, in particular, if several decisions are
made during the analysis, with each decision
producing a maximum result.
For this reason, consequence analysis should be
approached with intelligence, tempered with a
Table 4-5 Guidelines for Selection of Process Incidents
Incident Guideline
characteristic
Realistic release
incidents
Process pipes Rupture of the largest diameter process pipe as follows:
For diameters smaller than 2 in, assume a full bore rupture.
For diameters 2-4 in, assume rupture equal to that of a 2-inch-
diameter pipe.
For diameters greater than 4 in, assume rupture area equal
to20% of the pipe cross-sectional area.

Hoses Assume full bore rupture.

Pressure relief devices Use calculated total release rate at set pressure. Refer to
relieving directly to the pressure relief calculation. All material released is assumed to
atmosphere be airborne.

Vessels Assume a rupture based on the largest diameter process pipe

attached to the vessel. Use the pipe criteria.
Worst-case
incidents
Quantity Assume release of the largest quantity of substance handled
onsite in a single process vessel at any time. To estimate the
release rate, assume the entire quantity is released within 10 min.

Wind Assume F stability, 1.5 m/s wind speed, unless meteorological data
speed/Stability indicate otherwise.

Ambient Assume the highest daily maximum temperature and average

temperature/hu humidity.
midity

Height of Assume that the release occurs at ground level.

release
Topography Assume urban or rural topography, as appropriate.

Temperature of Consider liquids to be released at the highest daily maximum

release temperature, based on data for the previous 3 years, or at process
substance temperature, whichever is highest. Assume that gases liquefied by
refrigeration at atmospheric pressure are released at their boiling