Repurposing LNG terminals for Hydrogen Ammonia: Feasibility and Cost Saving
Accident near misses and precursor analysis2
1. Accident Near Misses and
Precursor Analysis
Case Study: Fire in a CSTR System
Presented by:
Mohammed AlShammasi
Ashwin Ariyapadi
Kevin Anglin
November 19, 2012
3. 3
Introduction
November 19, 2012
Definitions
• Near Miss- a severe precursor that is directly
connected to a potential accident in the accident
chain
• Accident Precursor- an event that could escalate
to an accident
A precursor is a more general term while a near
miss can consist of more than one precursor
Accident Near Misses and Precursor Analysis
4. 4
Introduction
November 19, 2012
Precursor Analysis v. Accident Analysis
• Precursor events at the base of the pyramid.
• High consequence accidents at the top of
the pyramid
• Precursor events
▫
▫
▫
▫
simpler to analyze
easier to prevent
smaller in scale
less costly to resolve
Accident Near Misses and Precursor Analysis
5. 5
Introduction
November 19, 2012
Case Study
• Applying QRA concepts
based on precursor
analysis to fire (top
event) in a CSTR system
Accident Near Misses and Precursor Analysis
6. 6
Introduction
November 19, 2012
Case Study, continued
• Hazard barriers:
▫
▫
▫
▫
▫
▫
Coolant system
Temperature controller
Rupture disc
Fire alarm
Manual shutdown
Firefighting emergency
Accident Near Misses and Precursor Analysis
8. 8
Identification of Precursors
November 19, 2012
Identification of Precursors
• Methods to identify precursor events:
▫
▫
▫
▫
▫
▫
Regular surveys
Interviews with safety engineers
Direct observation
Voluntary reporting
Alarm databases
Automated detection systems
Accident Near Misses and Precursor Analysis
9. 9
Identification of Precursors
November 19, 2012
Initiating Events
• Four main initiating events were identified:
Initiating events
Frequency, yr-1
Coolant failure
2.50
Inlet valve malfunction
1.74
External fire
0.82
Agitator failure
4.16
Accident Near Misses and Precursor Analysis
10. 10
Identification of Precursors
November 19, 2012
Basic Failure Events
• An initiating event can be followed by a number
of basic failure events:
Basic Event
Probability of failure
Probability of success
Temperature controller failure
Rupture disk failure
Alarm Fails
0.040
0.120
0.250
0.960
0.880
0.750
No manual shutdown (alarm works)
No manual shutdown (alarm fails)
0.120
0.240
0.880
0.760
Quick emergency response failure (Alarm works)
Quick emergency response failure (Alarm fails)
0.316
0.544
0.684
0.456
Accident Near Misses and Precursor Analysis
11. 11
Identification of Precursors
November 19, 2012
Basic Failure Events
• Contributors to uncertainty in frequencies:
1. Incomplete list of initiation events
2. Incomplete knowledge of hazard scenarios
3. Assumptions about material properties and
technical behavior
4. Common failure causes
5. Unconsidered process upsets
6. Operational errors due to inadequate training or
attitudes
Accident Near Misses and Precursor Analysis
16. 16
Modeling and Analysis
November 19, 2012
Consequences
• Consequences were identified based on
Estimated Monetary Value (EMVs)
Low monetary
value, $ MM
High monetary
value, $ MM
Expected monetary
value, $ MM
No damage
0.00
0.00
0.00
Minor plant damage
0.40
1.00
0.70
Major plant damage without injuries
2.00
3.00
2.50
Major plant damage with injuries
1.00
2.00
1.50
Consequence
Major plant damage with fatalities
7.00
Accident Near Misses and Precursor Analysis
7.00
17. 17
Modeling and Analysis
November 19, 2012
Consequences
• Contributors to uncertainty in consequences:
1. Incomplete knowledge of number of
fatalities/injuries
2. Flammability behavior assumptions
3. Ignition source model
4. Explosion model and efficiency assumptions
5. Changes in market dynamics for acceptable risk
magnitude
6. Neglecting environmental costs of fire
Accident Near Misses and Precursor Analysis
18. 18
Modeling and Analysis
November 19, 2012
Event Tree
Outcome
I
Works
S
Consequence
Scenario
No damage
S1
No damage
S2
S
No damage
S3
S
No damage
S4
S
Minor plant damage
S5
Major plant damage
S6
S
No damage
S7
S
Major plant damage
S8
F
Injuries/fatalities
S9
Fails
Frequency of I
Temp
controller
0.960
Fails
0.040
Works
Rupture
disk
S
0.880
Works
0.750
Fails
Works
Manual
shutdown
0.880
Works
0.120
Alarm
0.120
0.684
Works
Fails
0.316
Fails
0.250
Quick
emergency
response
Manual
shutdown
F
0.760
Works
0.240
Quick
emergency
response
0.456
Works
Fails
0.544
Accident Near Misses and Precursor Analysis
19. 19
Modeling and Analysis
November 19, 2012
Scenario Frequency
• The frequency of each scenario was calculated
for each initiating event as:
Accident Near Misses and Precursor Analysis
20. 20
Modeling and Analysis
November 19, 2012
Scenario Frequency
• Take Scenario 5 for the coolant failure – coolant
fails, temperature controller fails, rupture disc
fails, alarm works, manual shutdown fails, and
emergency response is quick:
F(S) = (2.5 1/yr)(0.04)(0.12)(0.75)(0.12)(0.684)
F(S) = 0.000739 1/yr
Accident Near Misses and Precursor Analysis
22. 22
Risk Management
November 19, 2012
Risk Evaluation
• The risk for each scenario was evaluated by
multiplying scenario frequency by associated
consequence.
• This type of risk evaluation is based on expected
monetary values.
Accident Near Misses and Precursor Analysis
23. 23
Risk Management
November 19, 2012
Total Risk
• The total risk was estimated by summing the
risks for each initiating event over all scenarios
Accident Near Misses and Precursor Analysis
24. 24
Risk Management
November 19, 2012
Scenario Importance
• The importance percentage of each scenario was
evaluated as follows:
• The importance % is a measure of scenario
contribution to the total risk
Accident Near Misses and Precursor Analysis
25. 25
Risk Management
November 19, 2012
Total Fire Risk
• The total risk of fire was calculated by summing
the risks of all initiating events:
Accident Near Misses and Precursor Analysis
29. 29
Recommendations
November 19, 2012
Risk Control
• An example of risk control is installation of
automatic sprinkler system as an emergency
response system in case of fire
Cost of installation, $/yr
Associated costs, $/yr
No sprinkler system
0
Sprinkler system 1
14,000
0.99
2,100
Sprinkler system 2
10,000
0.82
5,400
Sprinkler system 3
7,000
0.72
8,700
Accident Near Misses and Precursor Analysis
30. 30
Recommendations
November 19, 2012
Risk Control Continued
• The expected value of control was calculated
assuming a useful life period of 10 years for each
system
Accident Near Misses and Precursor Analysis
31. 31
Recommendations
November 19, 2012
EMVs for Control
• The EMVs for various control systems were as
follows:
Expected monetary values of control, $/yr
No sprinkler system
Sprinkler system 1
Sprinkler system 2
Sprinkler system 3
Accident Near Misses and Precursor Analysis
32. 32
Recommendations
November 19, 2012
Identification
•
•
•
•
•
•
Other potential initiating events
Other weak points/critical basic events
Examining of mechanical integrity of CSTR
Considering common failure modes
Performing HAZOP
Implementing reliability testing on hazard
barriers (TC, alarm, …)
Accident Near Misses and Precursor Analysis
33. 33
Recommendations
November 19, 2012
Analysis & Modeling
• Using Bayesian Network software
• Using probability distribution functions instead
of single-point probabilities
• Analyzing effects of fire or other initiating events
on other equipment in the facility
• Developing empirical probability functions
based on test data
• Creating a risk-level matrix
Accident Near Misses and Precursor Analysis
34. 34
Recommendations
November 19, 2012
Risk Management
• Adding redundancy in hazard barriers for
▫ temperature controller
▫ overpressure alarm system
• Using risk distribution functions
• Using utility function to account for risk
attitudes
• Accounting for societal risk
• Developing frequent maintenance and testing
procedure
• Adapting an inherently safer reactor design
Accident Near Misses and Precursor Analysis
35. 35
Recommendations
November 19, 2012
Inherently Safer Design
• The current cooling jacket has a failure
frequency of 2.5 times per year
• A jacket with a failure rate of only 2 times per
year is safer:
Accident Near Misses and Precursor Analysis
36. 36
November 19, 2012
Conclusion
• A brief risk assessment of a CSTR system based
on precursor analysis was presented using fire as
top event.
• Four initiating events were identified followed
by a number of possible basic failures.
• A number of recommendations were provided
for precursor identification, analysis and
modeling, and risk management.
Accident Near Misses and Precursor Analysis