Applying Functional Safety Engineering Technique to Optimize Defensive Fire Protection Strategy 66 th Canadian Chemical Engineering Conference October 19 th, 2016 Presented by Sam Sanati, P.Eng, FS Eng. 30 January 2017 1
Fire Protection Design Prescriptive Design Deterministic Design Probabilistic Design Codes Standards Fire Scenario (severity) Specification Fire Scenario (severity) Risk of events Specification Specification Performance based Performance based 30 January 2017 2
Deterministic vs Probabilistic NFPA 551 Definitions: Deterministic Model: In a deterministic model, the quantities being modeled are treated as being completely certain the purpose of the model is to provide an estimate of these quantities. Probabilistic Method: In a probabilistic model, the quantities being modeled are treated as being uncertain the purpose of the model is to quantify the degree of uncertainty in these quantities. 30 January 2017 NFPA 551: Guide for the Evaluation of Fire Risk Assessments 3
Defensive Fire Protection NFPA 600 Definitions: Defensive: The mode of manual fire control in which the only fire suppression activities taken are limited to those required to keep a fire from extending from one area to another. Offensive: The mode of manual fire control in which manual fire suppression activities are concentrated on reducing the size of a fire to accomplish extinguishment. 30 January 2017 4 NFPA 600: Standard on Facility Fire Brigades
Risk Based Fire Protection Design Scenario Selection Consequence & Freq. Analysis Barriers Effectiveness Event Tree Analysis Risk Plotting 30 January 2017 5
Event Tree Analysis Leak Frequency Ignition Probability Immediate to Delay Ignition Event Frequency 3.38% Delayed 50% 5.83x10 5 VCE Loss of Containment 3.45 x10 3 Immediate 50% 5.83x10 5 Jet Fire / BLEVE 96.62% No Ignition 3.33x10 3 Dispersed Gas
ETA with Barriers Leak Frequency Loss of Containment Ignition Probability 3.38% Detection / ESD / De-Press Fail Immediate to Delay Ignition Delayed 50% Immediate 50% Exposure Impinged Exposed Fire Water Pumps Fail Success Deluge System Fail PFP Fail Success Fire Escalation Event Frequency VCE BLEVE Jet Fire BLEVE (Time Dependent) BLEVE (Time Dependent) 3.45 x10 3 Success Fire Escalation Jet Fire Success Fire Controlled 96.62% No Ignition Dispersed Gas
Reliability of Barriers Reliability of the barriers depends on: failure frequency test intervals, mean time to restore (MTTR) and diagnostic coverage. 30 January 2017 8
Failure on Demands Type of Barriers Failures Test Interval (yr) Calculated PFD Selected PFD Remarks (per 10 6 hr) (Based on Industry Practice) Gas Detector (IR) 2.26 1 4.49x10-3 3.07x10-3 Test interval every 3 months. However, standard test interval is every year Fusible Link 1x10-6 Reference: Grosse et al (1996) Deluge System 5.8 1 1.27x10-2 1.5x10-2 Selected PFD is based on OGP Diesel Fire Water Pump 1550.38 0.02 (Per NFPA weekly) 0.133 0.12 Poorly Maintained Diesel pump is 0.12 Passive Fire Protection 0.01 CCPS Emergency Shutdown 6.5 1 2.89x10-3 2.89x10-3 Valve ESD Logic Solver 16 Unknown 3.5x10-2 3.5x10-2 Test Interval is normally 1 yr 30 January 2017 9
Effectiveness of Barriers Effectiveness of barriers and safeguards should be evaluated by considering the following elements: Functionality; Maintenance History; Voting Logic; and Dependability. 30 January 2017 10
Dependability Dependability System Integration Operator Intervention Instrumentation Integrity Mechanical Integrity Activate Emergency System Initiate Suppression 30 January 2017 11
Dependable Equipment Impairment 30 January 2017 12
Maintenance Condition & History 30 January 2017 13
Functionality Deluge for BLEVE impingement (PFD=1) Using Improper type of fireproofing or jacketing (PFD=1) 30 January 2017 14
Voting Logic ESD System G D G D PFD = 4.98 x 10-2 Sensor 2oo2 Logic Solver 1oo1 Final Element 3oo3 Detectors/Operators/ ESD Logic / F&G ESDV / Alarm 30 January 2017 etc. Logic 15
Voting Logic No epressurization Sensor 1oo1 Logic Solver 1oo1 Final Element 2oo2 PFD = 4.70 x 10-2 Detectors, etc. ESD Logic / F&G Logic ESDV / Alarm Operator Intervention Operator Intervention Logic Solver 1oo1 Final Element 3oo3 PFD = 1.41 x 10-1 ESD Logic / F&G Logic ESDV / Alarm 30 January 2017 16
Case Study: NGL Bullets Fire Protection System: No PFP No F&G No De-Pressurization Poor Firewater system 30 January 2017 17
Consequence Analysis Results Event Threshold Impact Distance BLEVE Fireball Distances (m) at 1 m above Ground 1000 tdu 2000 tdu 3200 tdu (%100 Fatality) Jet Fire 35 kw/m 2 (%100 Fatality) Overpressure Overpressure 5 psig (%100 Fatality indoor) 7.5 psig (%50 Fatality indoor) 196 m 123 m # Fatality 72 m 5* 46 m 5* 64 m 5* 51 m 2.5* * Assume 10 workers/operators are working outdoor and 5 operators inside the building. 30 January 2017 18
Original ETA Leak Frequency Loss of Containment Ignition Probability 3.38% Detection / ESD / De-Press Fail Immediate to Delay Ignition Delayed 50% Immediate 50% Exposure Impinged Exposed Fire Water Pumps Fail Success Deluge System Fail PFP Fail Success Event Frequency 5.83 x10 5 2.92x10 5 3.5 x10 6 3.08 x10 7 VCE BLEVE Jet Fire BLEVE (Time Dependent) BLEVE (Time Dependent) 3.45 x10 3 Success 2.53 x10 5 Jet Fire Success Fire Controlled 96.62% No Ignition 3.33 x10 3 Dispersed Gas
Risk Plotting <10-6 10-6 < to <10-4 10-4 < to <10-3 10-3 < to <10-2 10-2 < avg. 10-5 avg. 10-4 avg. 10-3 avg. 10-2 N= 50+ N= 10 to 50 BLEVE N= 3 to 10 Jet Fire N= 1 to 2 Medical Aid 30 January 2017 20
Defensive Fire Protection Design Options 1. Emergency Systems (ESD, F&G, Depressurization) 2. Passive Fire Protection 3. Emergency systems + Active Fire Protection 4. Emergency system + Passive Fire Protection 30 January 2017 21
Improvements Improvement Improvement Existing PFD Improved PFD Firewater Diesel Pump Deluge Valve tests Interval Detection system Valve tests Interval From 3yrs to 6 months From 1 yr to 6 months ESDV test interval From 1 yr to 6 months 0.12 0.065 0.015 0.0063 4.95 x 10-3 2.5 x 10-3 2.9 x 10-3 1.5 x 10-3 30 January 2017 22
Improve Risk Frequency Event 1 Detection/ ESD/ Depress. 2 Passive Fire Protection 3 Improved F&G/ESD/ Depress. & AFP 4 Improved F&G/ESD/ Depress. & PFP VCE 2.9 x 10-6 5.83 x 10-5 1.36 x 10-6 1.34 x 10-6 BLEVE 2.9 x 10-6 5.83 x 10-8 7.15 x 10-7 1.34 x 10-9 Jet Fire N/A 5.82 x 10-5 6.2 x 10-7 1.33 x 10-6 Controlled Fire 1.11 x 10-4 N/A 1.14 x 10-4 1.14 x 10-4 BLEVE Event Frequency 2.9 x 10-6 5.83 x 10-8 7.15 x 10-7 1.34 x 10-9 30 January 2017 23
Improved Risk Plotting <10-6 10-6 < to <10-4 10-4 < to <10-3 10-3 < to <10-2 10-2 < avg. 10-5 avg. 10-4 avg. 10-3 avg. 10-2 N= 50+ N= 10 to 50 Option 2,3 &4 Option 1 N= 3 to 10 N= 1 to 2 Medical Aid 30 January 2017 24
What would change the results Risk matrix structure Type of Fire Population Distribution Spacing, siting, building type Equipment arrangement Mechanical Integrity Program 30 January 2017 25
Thank you If you d like to find out more contact: Sam.Sanati@atkinsglobal.com Or visit: www.atkinsglobal.com Atkins Limited except where stated otherwise. 30 January 2017 26