Enhancing Safety and Minimizing

Corrosion in Refineries through

Process Intelligence Tools

Sridhar Srinivasan and Dr. Russell Kane

Honeywell International, Inc
David Moore
BP North America
Overview

• “Intelligence” in process control

• What is ammonium bisulfide (sour water) corrosion
• Importance of prediction as a means to “Fail-Safe
Control” in Sour Water Systems
• Sour Water JIP (joint industry initiative)
• Predict-SW 2.0 Software System
• Experiences / Case Studies
A Paradigm Shift

• Process Intelligence: Using process information to

generate recommendations on the state of the process.

• Software tools to evaluate process data:

- quantify potential for corrosion damage, and
- generate recommendations for acceptable material solutions or
process changes.

• These tools are being integrated with process modeling

tools and monitoring solutions.
• Predict®-SW 2.0 is one of the Honeywell suite of
corrosion tools to proactively address corrosion in
refinery systems.
Integrating ‘Intelligence’ into Process Design
• Process Intelligence: Using process information to generate recommendations on the state
of the process.
- Honeywell’s UniSim Design is a process modeling tool for development/evaluation of critical process data
- Predict-SW performs corrosion modeling (quantification / prediction) based on process conditions
- Integration (USD R370) provides the ability to design, control and optimize processes to control corrosion

• Vision:
- An integrated USD / Predict-SW provides the ability to:
 “see” what sour water process stream variables contribute to corrosion
 how they impact unit operational boundaries
 how corrosion can be managed as a unit process boundary
 Integration Schematic
Reliability HS&E Compliance

Improve PROCESS Performance Process Intelligence

Unisim Design

Knowledge Capture
Advanced Process Performance Real-Time
Advanced
Control (APC)Process Monitoring
Performance Real-Time (RTO)
Optimization
Control (APC) Monitoring Optimization (RTO) Process
Process Modeling
Process
Optimization
Process Process Modeling
Process Unit Optimization
Productivity

Process
Simulation Process Unit
Monitoring
AndSimulation
Design Data Refinery Planning
Monitoring
And Design Data
Visualization Refinery
And Planning
Scheduling
Laboratory
Process Boundary Visualization And Scheduling
Process Boundary
Information
Laboratory Data Historization
Management Management
Information(LIMS) Data Historization
Management Process
Management (LIMS)
Process
Management
Management

Predict®-SW
Operational Excellence

Corrosion Damage Multiphase Flow

Early Event Remote Perform. Equipment Health Corrosion
Quantification Damage
Prevention Multiphase Flow
Modeling
Early Event
Detection Remote Perform.
Management Equipment Health
Management Quantification Prevention Modeling

Capital Return
Detection Management Management
Material Selection Design Planning
Material Selection Design Planning

Improve ASSET Effectiveness

Risk Management
Importance of Quantified Predictions
• Accurate prediction is critical for
optimum process control –
maximize production goals while
limiting corrosion, operational risk
and asset damage.
• Models provide the ability to
determine critical system points
(i.e. high corrosion rate locations)
- Identify/Extend actual process
boundaries
- Better process control and
improved asset life
- Update Operational Risk
Assessment and Asset
Management
- Identifies locations for real-time
monitoring and feedback to
operators.
• Enhance safety while increasing
plant productivity.
Sour Water Corrosion JIP

• Prior to 2000:
- Several major failures & high operating risk in sour water systems
- Available rules were qualitative and of limited use
• In 2000
- Honeywell* with assistance from Shell Global Solutions organized
a joint industry program (JIP) with 20 leading refinery operators
and service companies
• This initiative developed a definitive, quantitative
engineering basis for dealing with sour water process
corrosion.
• Included chemical & flow modeling and experimental
simulation of actual refinery sour water conditions.

*Acquired InterCorr in 2005

JIP Findings: Important New Insights

• JIP data indicated that • JIP data show:

old rules do not work: - H2S partial pressure has a
significant effect on NH4HS
- Fixed NH4HS concentration corrosion
limit
- Three key variables must be
- Fixed Velocity limit considered together to predict
- pH is not key parameter NH4HS corrosion:
 NH4HS concentration
- KP parameter limited.
 Flow induced wall shear stress
 Kp is the product of mole (Not velocity)
percent Ammonia and mole  H2S partial pressure
percent H2S in the vapor
phase - Consider other relevant
variables
 Temperature, hydrocarbon
phase, inhibition.
Joint Industry Program Summary

• Results
- Corrosion rates obtained for all commonly used alloys in refinery
sour water systems versus wide range of sour water conditions
• Software Application (Predict-SW 2.0)
- Now available for both sponsor & non-sponsor licensing
- Flow modeling calculations for flow-induced wall shear stress
- Corrosion rate algorithms/rule sets to address effects key and
relevant variables.
- A quantifiable basis for selecting materials, performing inspections
and developing asset management programs
- Ability to support process optimization through integration with
process modeling tools such as Unisim Design
- Significant changes to conventional unit design – with substantial
economic benefits
 Predict-SW: Software Interface

Data Output

Process Interface

0
 Predict-SW: BP Utilization Objectives

• BP was part of the Predict-SW JIP development

group.
• They selected Predict-SW to help enhance plant
safety and minimize potential for failure
- Methodology for rapid coverage of asset base for corrosion risk
identification
- Minimize staffing requirements to enhance productivity
- Quickly identify critical areas for appropriate action (material
enhancement, inspection, process change)

1
 Case Study A: BP Refinery

• Iso-cracker stage I air cooler tubes were seeing corrosion

problems
- Up to 6% NH4HS and 300°F at inlet.
- Predicted corrosion rate of carbon steel with Predict-SW = 12 mpy
- Measured average corrosion (metal loss) rate in the first four years
of unit operation: 13 mpy
- Predict-SW recommended material for low (<1 mpy) corrosion:
Alloy 825 (21Cr-3Mo-20Fe-Ni)

2
 Case Study B: BP REAC Multipoint Data Analyses
Superficial
Liquid PP
Case ID Shear Stress Velocity Superficial Gas Velocity Lab flow velocity H2S Flow Regime

Tag Pa m/s m/s m/s Kpa(a)

IN-100%WR 1.9 0.1 4.4 0.8 130.1 Horizontal - Slug Flow

IN-125%WR 2.0 0.1 4.4 0.8 129.41 Horizontal - Slug Flow

IN-150%WR 2.1 0.1 4.4 0.8 128.66 Horizontal - Slug Flow

IN-200%WR 2.2 0.1 4.3 0.8 126.59 Horizontal - Slug Flow

IN-75%WR 1.9 0.1 4.4 0.8 130.79 Horizontal - Stratified Flow

IN-50%WR 1.8 0.1 4.4 0.7 132.24 Horizontal - Stratified Flow

IN-100%WR-Weld 6.8 0.1 4.4 1.6 130.1 Horizontal - Slug Flow

IN-125%WR-Weld 7.0 0.1 4.4 1.6 129.41 Horizontal - Slug Flow

IN-150%WR-Weld 7.2 0.1 4.4 1.6 128.66 Horizontal - Slug Flow

IN-200%WR-Weld 7.7 0.1 4.3 1.7 126.59 Horizontal - Slug Flow

IN-75%WR-Weld 6.6 0.1 4.4 1.5 130.79 Horizontal - Stratified Flow

IN-50%WR-Weld 6.4 0.1 4.4 1.5 132.24 Horizontal - Stratified Flow

IN-75%WR 1.9 0.1 4.4 0.8 130.79 Horizontal - Stratified Flow

IN-50%WR 1.8 0.1 4.4 0.7 132.24 Horizontal - Stratified Flow

IN-75%WR-Weld 6.6 0.1 4.4 1.5 130.79 Horizontal - Stratified Flow

IN-50%WR-Weld 6.4 0.1 4.4 1.5 132.24 Horizontal - Stratified Flow

3
 Case Study B: BP REAC Multipoint Data Analyses
Carbon Alloy 20Cb-
Case ID Steel Alloy 400 12 Cr Steel 304 SS 316 SS Alloy 2205 Alloy 800 Alloy 600 3 Alloy 825

Tag mmpy mmpy mmpy mmpy mmpy mmpy mmpy mmpy mmpy mmpy

IN-100%WR < 0.025 < 0.025 < 0.025 < 0.025 < 0.025 < 0.025 < 0.025 < 0.025 < 0.025 < 0.025

IN-125%WR < 0.025 < 0.025 < 0.025 < 0.025 < 0.025 < 0.025 < 0.025 < 0.025 < 0.025 < 0.025

IN-150%WR < 0.025 < 0.025 < 0.025 < 0.025 < 0.025 < 0.025 < 0.025 < 0.025 < 0.025 < 0.025

IN-200%WR < 0.025 < 0.025 < 0.025 < 0.025 < 0.025 < 0.025 < 0.025 < 0.025 < 0.025 < 0.025

IN-75%WR 0.11 < 0.025 < 0.025 < 0.025 < 0.025 < 0.025 < 0.025 < 0.025 < 0.025 < 0.025

IN-50%WR 0.11 < 0.025 < 0.025 < 0.025 < 0.025 < 0.025 < 0.025 < 0.025 < 0.025 < 0.025

IN-100%WR-Weld < 0.025 < 0.025 < 0.025 < 0.025 < 0.025 < 0.025 < 0.025 < 0.025 < 0.025 < 0.025

IN-125%WR-Weld < 0.025 < 0.025 < 0.025 < 0.025 < 0.025 < 0.025 < 0.025 < 0.025 < 0.025 < 0.025

IN-150%WR-Weld < 0.025 < 0.025 < 0.025 < 0.025 < 0.025 < 0.025 < 0.025 < 0.025 < 0.025 < 0.025

IN-200%WR-Weld < 0.025 < 0.025 < 0.025 < 0.025 < 0.025 < 0.025 < 0.025 < 0.025 < 0.025 < 0.025

IN-75%WR-Weld 0.18 < 0.025 0.03 < 0.025 < 0.025 < 0.025 < 0.025 < 0.025 < 0.025 < 0.025

IN-50%WR-Weld 0.17 0.03 0.03 < 0.025 < 0.025 < 0.025 < 0.025 < 0.025 < 0.025 < 0.025

IN-75%WR 0.51 0.36 0.12 0.06 0.06 < 0.025 0.06 0.06 0.06 0.1

IN-50%WR 0.71 0.46 0.15 0.07 0.07 < 0.025 0.09 0.09 0.09 0.1

IN-75%WR-Weld 0.58 0.38 0.13 0.07 0.07 < 0.025 0.09 0.09 0.09 0.1

IN-50%WR-Weld 0.74 0.48 0.18 0.09 0.09 0.03 0.13 0.13 0.12 0.1

4
 Case Study C: Problem

• Sour water overhead condensers with high alloy material were

experiencing corrosion problems
• Important need to identify the source of the problem and solution
for effective mitigation
• Predict-SW utilized for process analyses and corrosion
quantification
5
 Case Study C: Analysis
• Corrosion problems were indicated on the Shell side of the overhead
condensers.
- This was of concern since the material of construction was a highly alloyed
stainless steel
• Process and flow modeling was coupled with Predict-SW corrosion
modeling
- Goal: To determine effect of flow induced wall shear stress on corrosion rates
• Corrosion rates predicted for base case and sensitivity studies were
conducted for worst case scenarios.
• No corrosion on stainless alloy was indicated from ionic modeling and
corrosion analyses
• However, inspection data was indicating corrosion propensity for 904L
stainless steel
- So, what could contribute to corrosion here?
- Clearly, a factor was not considered in the analyses

• One possibility was the presence of cyanides in the process stream, but
was not reported.
6
 Case Study C: Base Case Corrosion Rates

NH4HS H2S Partial Pressure Temperature Wall Shear Stress CS Corrosion Rate 904L Corrosion Rate

wt% psia F Pa mpy mpy

11.41 12.53 195 5 14 <1

11.41 12.53 195 10 16 <1

11.41 12.53 195 50 33 <1

11.41 12.53 195 100 51 <1

11.41 12.53 195 150 58 <1

11.41 12.53 195 200 62 <1

11.41 12.53 195 250 65 <1

11.41 12.53 195 500 68 <1

11.41 12.53 195 1000 72 <1

           

18 12.53 195 5 33 <1

18 12.53 195 10 38 <1

18 12.53 195 50 67 <1

18 12.53 195 100 78 <1

18 12.53 195 150 83 <1

18 12.53 195 200 88 <1

18 12.53 195 250 92 <1

18 12.53 195 500 96 <1

18 12.53 195 1000 99 <1

7
 Corrosion Prediction for System with Cyanide

NH4HS NH3 PP H2S PP Temperature CN WSS 2205 CR 904L CR CS CR

wt% psia psia F ppm Pa mpy mpy mpy

11.41 15.9 12.53 190 40 500 9.2 7.4 55

11.41 15.9 12.53 190 40 1000 13.3 11.2 74.7

11.41 15.9 12.53 190 40 1400 16 13.7 83.3

11.41 20 12.53 190 40 500 11.8 10.2 62.4

11.41 20 12.53 190 40 1000 16.1 14.3 84.4

11.41 20 12.53 190 40 1400 19 17.1 94.4

11.41 25 12.53 190 40 500 15.4 14.3 70.5

11.41 25 12.53 190 40 1000 20 18.8 94.9

11.41 25 12.53 190 40 1400 23.1 21.7 106.3

11.41 30 12.53 190 40 500 19.5 19.1 77.5

11.41 30 12.53 190 40 1000 24.5 23.9 103.7

11.41 30 12.53 190 40 1400 27.6 26.9 116.5

Analyses identified the problem and the need for

8 A high nickel alloys with <1 mpy corrosion rate)
 Corrosion Analyses: Summary
• Each refinery application has different process criteria.
- Tools such as Predict-SW facilitate identification of critical operating
parameters
• Multiphase streams across multiple tube rows resulted in non-
uniform flow distribution
- This means that different rows may see different flow patterns and wall
shear stress levels.
• Predict-SW used to perform multipoint sensitivity analyses to
characterize worst case locations and conditions
• Pro-active application of Predict-SW resulted in the following
benefits:
- Give plant engineers a tool to “see” corrosion vs. process.
- The ability to make changes to avoid potential costly failures.
- Significantly reduced operational risk…. That actually enhanced
productivity and increase plant safety.
- Predict-SW is now being applied to BP refinery sour water handling
systems enterprise wide with similar expected benefits.

9
 Conclusions
• Predict-SW can be used by refinery operators:
- Integrates with process data for better process optimization and
asset protection
- Quickly identifies critical locations for risk reduction
- Supports better inspection and maintenance planning, and
allocation of resources towards area of high corrosion risk
- Determine the right alloy or process conditions for your unit.
- Mitigate and minimize risk of failure (as a component in any RBI
program).
- Potential cost of failure in REAC system has been reported to be
up to $60 million per incident!
• Utilization of Predict-SW allowed BP to enhance plant
safety, minimize corrosion risk and ensure more reliable
unit operations.
• JIPs extending technology to amine, alkylation, and
naphthenic acid applications.