You are on page 1of 25

Siemens Process Safety Consulting

Do More with Dynamic Simulation


Case study in dynamic simulation for Safety Analysis
Webinar
15June2017
siemens.com

© Siemens AG 2017
Agenda

• Introductions
Topic
Introduction
Safety Moment
Siemens Process Safety Consulting Overview
Dynamic Simulation Overview
Dynamic Simulation Work Process
Case Study - Debutanizer
Q&A

Restricted © Siemens AG 2016


Page 2 DF CS SAF
Zero Harm Culture @ Siemens

Our Vision
“Every Siemens employee should be able to rely on an
intact and safe working environment so that they can
return healthily and safely to their family and friends.”

Our three Principles


1. Zero incidents – it is achievable SAFETY IS
not a program –
2. Health and safety – no compromises It‘s a mindset!
3. We take care of each other!

Joe Kaeser Roland Busch Lisa Davis Klaus Helmrich

Hermann Requardt Siegfried Russwurm Ralf P. Thomas

Restricted © Siemens AG 2016


Page 3 DF CS SAF
Siemens Process Safety Consulting
Operations evergreening with a focus on:
Focused Process Safety Assessments, Studies, Project
Pioneering Software improved asset performance, continuous management of PSI & risk
Services Consulting & Audits, Develop
Solutions based safety,
with breadth of Expertise Guidelines, Best Practice Sharing
identify-track-report performance indicators
To the global Processing
Siemens Process Safety Software Siemens Process Safety Consulting Expertise
Industry
Flare System Pressure Relief Process Hazard Analysis (HAZID, HAZOP, LOPA)
Process Safety Pressure Protection
Analysis Analysis Dynamic Process Simulation
Manager™ PSPPM™

Inspection Risk Assessment


Relief & Disposal
Data Management
System Services
System Services
PS Asset Integrity Manager®
PSAIM™

Management of Process
Process Safety
Safety
Change Services
PS Change Manager® Consulting
Consulting
PSCM™

Safety Lifecycle Technology &


Oil and Gas Manager™ Training Services
Services Safety Advisory
OGM™

Mechanical Integrity System Deficiencies benchmarking


SIF (Safety Instrumented Function) Conceptual Design
Restricted © Siemens AG 2016 Programs for continuous improvement
SIL (Safety Integrity Level) assessment
Page 4 DF CS SAF
Dynamic Simulation - Overview

Restricted © Siemens AG 2015 All rights reserved. siemens.com/answers


Dynamic Simulation Traditional Uses

• Process Optimization
• Startup / Shutdown
• Throughput Changes

• Operator Training
• Emergency Procedures and Response

• Pipeline Design and analysis

• Compressor Systems

Restricted © Siemens AG 2016


Page 6 DF CS SAF
Other Dynamic Simulation uses

• Disposal (Flare) system evaluation

• Control system evaluation

• SIS / Shutdown System Evaluation

• Safety Analysis

• Hydrotreating Reactor loop pressure profile analysis

Restricted © Siemens AG 2016


Page 7 DF CS SAF
Dynamic Simulation Tools

Aspentech
• HYSYS
• Aspen Plus

Honeywell
• Unisim Design (useful for Dynamic engineering studies)
• Unisim Operations (best for OTS)

Schneider Electric (formerly SIMSCI)


• Dynsim
• SimCentral

Restricted © Siemens AG 2016


Page 8 DF CS SAF
Dynamic Simulation for Flare system evaluation

• Can be used to evaluate header pressures and flare tip radiation

• Can be done on entire system, or can take a hybrid approach and focus on largest contributors,
while smaller one rely solely on a quicker/cheaper steady state approach

• Can be used to determine staging of loads to the header (pressure up time, time for relief)

• Computer technology and software advanced enough that expensive servers are no longer a
strict requirement, however performance is impacted by use of standard workstations.

• Can help analyze control system actions/reactions to take credit for additional layers of
protection

Restricted © Siemens AG 2016


Page 9 DF CS SAF
Dynamic Simulation for Disposal (Flare) system evaluation - Performance

• Studies have shown some promising results for certain global scenarios
• Boil-up scenarios (loss of feed and reflux, heat input continues) in particular have shown significant
reductions compared to traditional steady state analysis, upwards of 90% in some cases
• Lower load reduces relief device backpressures and can also reduce predicted radiation levels
• Often, the flare header design scenario is a total power failure which results in boil up at many locations
in the refinery
• May also demonstrate that system is not capable of reaching a relief condition
• Other scenarios, such as loss of cooling with continued reflux, do not have the same magnitude of reduction
• Some systems as low as 5-15% reduction
• Feed boiling point range can serve to predict effectiveness of approach.

Restricted © Siemens AG 2016


Page 10 DF CS SAF
Dynamic Simulation for Disposal (Flare) system evaluation - Challenges

• Simulations require a good deal of effort to construct properly

• Higher cost compared steady state analysis

• High level of competency required to build and maintain models

• Some simulations can be very complex


• Complexity can significantly impact simulation performance

• Simulation Time

Restricted © Siemens AG 2016


Page 11 DF CS SAF
Dynamic Simulation – Work Process

Restricted © Siemens AG 2015 All rights reserved. siemens.com/answers


Building a Distillation Column System

1. Data Collection
a) Equipment Mechanical details – dimensions, elevations, piping information
b) Equipment internal details – Column tray/packing (type, spacing, HETP), reboiler sump,
exchanger setp, accumulator vessel internals
c) Process data – Steady state H&MB, PI data
d) Control system information – Valve details, tuning parameters, control system logic, ESD logic,
emergency procedures
e) Utility system information - may include piping isometrics depending on complexity requirements
f) Relief system information
g) Piping isometrics

Restricted © Siemens AG 2016 All rights reserved. siemens.com/answers


Page 13
Building a Distillation Column System

2. Data Entry
a) Certain equipment may be represented via multiple unit operations in simulation software, i.e.
column reboiler sump is built separately to model thermosiphon effect
b) Piping/hydraulics have several options depending on tool selected
c) ‘Dummy’ control valves and controllers required to help set flows and pressures to unit operations
d) Component setup
e) Integrator details – step size, calculation priority, acceleration
f) Strip Charts – setup to view key process variables, controller actions
g) Weeping factor
h) OTS Framework (AspenTech only)

Restricted © Siemens AG 2016 All rights reserved. siemens.com/answers


Page 14
Running the simulation

Several key variables have been identified that can greatly affect performance

• Hydraulic calculations take a toll on the pressure/flow solvers


• detailed hydraulics should be limited to specific areas
• May require several iterations to find a configuration that provides good balance between
resolution and simulation time
• Relief system piping on outlet side must be truncated, backpressure is another key variable
• Time step can have a slight effect on results (sensitivity), and can impact simulation times
• More components = more simulation time
• Must run model for some time to confirm steady state is achieved and tuned to match base case
• Changes to model require a re-initiation
• No more than 5-10% delta compared to PI data/base simulation
• Once steady state is achieved, events are triggered and model is allowed to run for some time (60-90
minutes)

Restricted © Siemens AG 2016 All rights reserved. siemens.com/answers


Page 15
Running the simulation - Tips

• PF solver failures typically can be mitigated by reductions in complexity


• Reduce hydraulic complexity by using valves with holdup volume and resistances instead of pipe
operations
• Reduce number of pseudo components as much as possible
• Increase time step where high resolution is not necessary (i.e. around condenser/accumulator).
AspenTech software can be set up with multiple time steps via OTS framework
• System hardware and background processes can impact simulation time
• Ideally should use a computer or server that can be dedicated to running the simulation
• Avoid virtual machines where resources are shared
• Current iterations of software tools do not utilize multi-threading
• Reboiler modeling is another key variable
• Can use simple or complex models
• Can simulate thermosiphon effect if desired

Restricted © Siemens AG 2016 All rights reserved. siemens.com/answers


Page 16
Case Study - Debutanizer

Restricted © Siemens AG 2015 All rights reserved. siemens.com/answers


Case Study - Debutanizer

Restricted © Siemens AG 2016 All rights reserved. siemens.com/answers


Page 18
Case Study - Debutanizer

System Details
• Two section trayed distillation column
• One steam thermosiphon reboiler shell
• Two cooling water condensers run in series
• Reflux drum equipped with control valve vent to flare header
• Control system is fairly simple, no emergency shutdown systems or HIPPS.
• Two cascaded control loops, for steam flow and overhead product/reflux
• System has three available relief devices
• Two located on column overhead prior to condenser inlet, and venting to flare header
• One more located on reboiler return that also vent to flare header
• Scenario evaluated is a total power failure, which results in a loss of feed, reflux and overhead cooling.
Steam continues to provide heat input
• Sensitivity analysis was performed on several key variables

Restricted © Siemens AG 2016 All rights reserved. siemens.com/answers


Page 19
Case Study - Assumptions

Some simplifying assumptions are necessary to build a model that will run in a reasonable amount of
time

• System is at a pre-defined steady state based on maximum expected throughput prior to initiating an
event.
• Instrument air system has enough capacitance to keep certain instruments active
• Control loops that provide benefits to the system are assumed not to act/remain in position
• Control loops which provide adverse affects to the system in response to the initiating event are
allowed to continue functioning
• Example: Overhead vent from accumulator to flare is set to act below relief pressure, and allowed
to open at it’s set point to provide flow to flare header
• Steam turbine pump availability is determined based on whether keeping them online would benefit
the system or not (pumps that provide benefit are usually assumed to fail, unless automated pump
starts are being analyzed)
• Reboiler is allowed to pinch as LMTD changes throughout the scenario

Restricted © Siemens AG 2016 All rights reserved. siemens.com/answers


Page 20
Case Study

Peak Total Relief Load


Case Steady State Analysis Dynamic Analysis (lb/hr) Difference / Reduction
(lb/hr)
Loss of Condenser, feed 260,000 93,700 64%
and reflux
Loss of condenser and 186,000 170,000 9%
reflux, feed continues
Loss of condenser, 184,000 173,000 6%
while feed and reflux
continue
Loss of condenser and Not Evaluated 108,000 N/A
feed, reflux continues

Restricted © Siemens AG 2016 All rights reserved. siemens.com/answers


Page 21
Sensitivity analysis

Accumulator Vent Valve


Vent active Vent inactive
Peak relief load (lb/hr) 78,400 93,700
Time to peak load (minutes from 3.4 3.3
event initiation)
Observed Reboiler Duty at peak 15.1 15.3
(MMBTU/hr)

Integrator time step size


Time step 0.025 0.050 0.100
Peak relief load (lb/hr) 94,050 93,700 127,500
Time to peak load (minutes
from event initiation) 3.5 3.3 3.9
Observed Reboiler Duty at 14.9 15.3 13.5
peak (MMBTU/hr)

Restricted © Siemens AG 2016 All rights reserved. siemens.com/answers


Page 22
Sensitivity analysis

Weeping Factor
Weeping Factor 1 0.3
Peak relief load (lb/hr) 93,700 86,300
Time to peak load (minutes from 3.3 3.1
event initiation)
Observed Reboiler Duty at peak 15.3 14.5
(MMBTU/hr)

Reboiler Setup
Reboiler Setup 1 2 3 4
Peak relief load 93,700 94,600 104,000 191,000
(lb/hr)
Time to peak load 3.3 3.4 3.1 5
(minutes from event
initiation)
Observed Reboiler 15.3 15.5 17 32
Duty at© Siemens
Restricted peak AG 2016 All rights reserved. siemens.com/answers
(MMBTU/hr)
Page 23
Conclusions

• Dynamic simulation can significantly reduce predicted relief rates vs traditional steady state methods
for certain scenarios

• Not as effective at reducing rate when reflux continues

• Models can be very sensitive to key variables, so these must be chosen wisely

• More data is needed to validate these models for widespread use

• Maintenance of models due to changes at the plant need to be handled carefully

Restricted © Siemens AG 2016 All rights reserved. siemens.com/answers


Page 24
Thank you for your time and attention!
……………………………………………………Any Questions????

Dan Pulis

Technical Advisor
Office: +1 713 570 2979
Mobile: +1 8323504031

Daniel.Pulis@Siemens.com

Restricted © Siemens AG 2016


Page 25 DF CS SAF

You might also like