SAFETY BARRIER PERFORMANCE ASSESSMENT

Transcription

1 SAFETY BARRIER PERFORMACE ASSESSMET I THE FRAMEWORK OF ESCALATIO GABRIELE LADUCCI Dipartimento di Ingegneria Civile e Industriale, Università di Pisa, Largo Lucio Lazzarino 1, Pisa, Italy gabriele.landucci@diccism.unipi.it 1

2 Introduction Domino effect was responsible of severe accidents that took place in the process and chemical industry. Cascade of events in which the consequences of a starting accident (primary event) impacting on target equipment with an impact vector (radiation, overpressure, fragments) are typically amplified (propagation) Confirmed by past accident data analysis Huge destructive potential Significant economic losses High risk perception due to fatalities Priolo Accident, Italy, 1985 Technical standards and legislation (Seveso Directives) concerned with the control of major accident hazard include measures to assess, control and prevent domino effects. 2

3 Domino effect triggered by fire In the case of domino effect caused by blast overpressure or fragment projection, the escalation rapidly occurs after the exposure of the target to the impact vector In the case of fired domino effect a time lapse is present between the start of the primary fire and the failure of the target equipment leading to a loss of containment Pressure build up associated to temperature increase in the equipment needs time time to failure ttf Crucial element to enhance mitigation and domino prevention LPG 3

4 Domino effect and safety barriers The ttf depends on both the features of the primary fire scenarios and shape of the secondary equipment involved in the fire A key point in the assessment of escalation probability in fire scenario is that in most cases both factors may be modified by the installation of mitigation barriers and by appropriate emergency measures OPE ISSUES Quantitative assessment of protection layers relevant to the prevention or mitigation of fired domino effect Comprehensive approach to consider different categories of protection layers PASSIVE ACTIVE PROCEDURAL 4

5 Aims of the work: Quantitative assessment of the action of safety barriers in preventing and mitigating fired domino effect Identification and characterization of relevant safety barriers (passive, active and procedural) Systematic quantitative analysis of safety barrier performance Availability and effectiveness analysis Integration with available probabilistic models for the frequency assessment of domino scenarios Application to case studies 5

6 Methodology 1. Identification and classification of relevant protective measures Information of target equipment and primary fire 2. Quantitative evaluation of safety barriers performances 2.1 Availability evaluation 2.2 Effectiveness definition 2.3 Effectiveness evaluation Gathering of site-specific performance data 3. Probabilistic assessment 3.1 Determination of escalation probability for protected equipment 3.2 Determination of domino scenario frequency 6

7 Step 1: Identification and classification of typical safety barriers Passive protective measures Installation of heat resistant coating (cementitious or vermiculite sprays, intumescent, mineral or ceramic fibers, etc.). Associated with a pressure relief devices (PRD). Combined mitigation action Active protective measures Systems for the delivery of fire-fighting agents (such as water or water-based foam): fixed, semi-fixed, mobile and portable systems. Emergency Shutdown Systems (ESD) and Emergency Depressurization Systems (EDP). Procedural and emergency measures The procedural measures are put in action by the internal personnel during a fire; the external emergency measures represent the coordinated response to a major emergency, involving local authorities, fire brigade, emergency teams etc 7

8 Step 1: Identification and classification of typical safety barriers Protection systems are site-specific. A database of reference equipment (RE) for a number of reference installations (RI) was developed (based on past accident data analysis) Refinery tank farms (RI.1); LPG storage facilities (RI.2); Offshore Oil&Gas (O&G) platforms for hydrocarbon extraction (RI.3) RI RE description RE Active Protection Systems RI.1 RI.2 RI.3 Floating roof diam. > 60 m T.1 Floating roof diam. > 30 m Fixed roof diam. > 20 m Floating roof diam. <30 m Fixed roof diam. < 20 m T.2 T.3 Foam-water sprinkler system Semi-fixed Foam System* + Fixed Water Spray* Foam-water sprinkler system or Semi-fixed Foam System* + Fixed Water Spray* Semi-fixed foam injection system* Rim seal by foam flooding* Abovegroud pressurized vessel V.1 Water Spray system PRD Mounded pressurized vessel V.2 PRD Horizontal separator Condensate treaters S.1 Object specific deluge S.2 Area deluge * not considered in the quantitative assessment ESD&EDP system ESD&EDP system PRD PRD PRD PRD PRD Passive Protection Systems PFP (2h rating) PFP (2h rating) PFP (2h rating) PFP (2h rating) Procedural/ Emergency measures Emergency team Emergency team Emergency team Emergency team Emergency team 8

9 Step 2: Quantitative evaluation of the performances of safety barriers Approach Safety barriers reduce the frequency of the unwanted scenario and/or mitigate the consequences. Two relevant parameters are needed availability: estimation of probability of failure on demand (PFD) for each safety barrier; effectiveness: the presence of a protection layer may not guarantee that the mitigation action will be performed successfully barrier specific evaluation mitigation action of the barriers in relationship with the physical modeling of the fire attack on target equipment Comment We analyze Protection Layers (PLs), but not Independent Protection Layers (IPLs), since the safety barriers may not fulfill the criterion of independence [LOPA], as they may be led to failure or malfunction by a common cause (e.g. the fire itself) 9

10 Analysis of passive barriers availability: literature PFD values effectiveness: the presence of the coating delays the ttf of equipment ttf p ttf ttf c Coating category low performance high performance Definition Materials aimed at thermal insulation but not specifically designed for fireproofing applications (glass wool, rock wool, etc.), specifically designed to withstand severe fire conditions (intumescent coatings, vermiculite spray, fibrous mineral wool, etc.), according to standards Value of ttf c (minutes) 0 70 Radiation and emissivity FIRE IMPIGEMET Convection and radiation BULK VAPOUR Convection BULK LIQUID Conduction between coating and tank wall 10

11 Analysis of active barriers Horizontal Pressurized Vessels Water Deluge Systems (WDS) ESD&EBD System Atmospheric tanks (Fixed) Foam Systems Water-foam sprinkler systems; Automatically actuated rim seal fire extinguishing system; Fixed discharge outlets (foam chambers); Availability: due to system complexity a Fault Tree Analysis was performed to evaluate the PFD (Probability of Failure on Demand) of the more common active fire protection schemes 11

12 Analysis of active barriers Availability: evaluation of system Probability of Failure on Demand (PFD) Active Fire Protection Type of actuation Proportioning method Calculated PFD through FTA Literature PFD value In-line educator (a) Pneumatic Metering proportioning (a) Foam-water Bladder tank (a) sprinkler system In-line educator (a) Electric Metering proportioning (a) Bladder tank (a) WDS for LPG Pneumatic ot applicable (b) vessels protection Electric ot applicable (b) WDS for Pneumatic ot applicable (b) horizontal separator Electric ot applicable (b) protection ESD system ot applicable ot applicable a Value for supervised fire sprinkler system b Typical value for water-based fire protection systems 12

13 Analysis of active barriers h Efficiency parameter (%) Efficiency WDS: intensity reduction factor φ in order to estimate the effectiveness. Reduction in the heat load due to heat radiation (Q HL ) obtained due to the presence of the activated deluge system. Q WDS Q HL φ = 0.5 a) Offshore separator Radiation intensity reduction (%) Top Front Rear Area Deluge - Standard flow rate Area Deluge - Double flow rate Object Deluge - Standard flow rate Area Deluge - Double flow rate; Object Deluge Standard flow rate b) LPG storage vessel Target cooling Standard flow rate Pool fire Always effective Jet fire ot effective Enhanced flow rate Pool fire Always effective Jet fire Variable effectiveness from case to case Sprinkler systems: analysis of fire protection systems performance statistics given in several literature studies PFD' PFD h = h1 PFD Modified PFD for the failure branch of event tree FPA [62] Australia FPA [63] Minimum value UF Fire Factory Office Mutual Committee System [65] [64] Maximum value Industrial Risk Insurers [61] U.S. Richardson Deparment (1985) [67] of Energy [66] Finucane et al. (1987) [68] 13

14 Analysis of procedural and emergency measures availability: literature PFD values effectiveness: related to the emergency response time; 3 key parameters Internal emergency response tem 1 = max time required to raise the alarm and start internal emergency procedures tem 2 = max time required to pose in act internal mitigation measures (mobile equipment available on site) ln To calculate Probit coefficients tem lntem2 For conservative estimate: a = lntem1 - lntem2 if ttf = tem 1, EP = 0.90 EP Pr = a b ln(ttf) if ttf = tem 2, EP = b = lntem1 -lntem2 External emergency response tfm= max time required by external emergency teams to provide and keep constant the amount of water necessary to suppress the primary fire or effectively cool the target. Step A Identify suitable fire-fighting strategy Step B Evaluate required water rate G W Step C Localize external water supplies Step D Select type and number of required Water Transport Systems (WTS) Fire Engines (FE) Step E Estimate WTS and FE arrival and deployment time Primary fire extinguishment Target exposure protection G G w w A S EP A fire t arget SF w S Required water application density for 12.2 L min -1 m -2 fire suppression w EP Required water application density for target exposure 10.0 L min -1 m -2 protection SF Safety factor 3 - FPA 15, TO-report 200G4-R0069/B Characteristic time of an effective intervention of the emergency team, to be compared against protected target ttf for final assessment. 14

15 Analysis of procedural and emergency measures availability: literature PFD values effectiveness: related to the emergency response time; 3 key parameters Parameter RI.1 Refinery tank farm RI.2 LPG storage facility tem1 (min) tem2 (min) tfm (min) Depending on primary fire scenario Depending on target geometry RI.3 Offshore installation tfm = tem2 First probit coefficient (a) Second probit coefficient (b)

16 Primary fire fi Y Failure of SB1 Step 3: Probabilistic assessment of mitigated domino events PFD1 1 -PFD1 Y Failure of SB2 Y Failure of SB2 PFD2 1-PFD2 PFD2 1 -PFD2 Intermediate events ttf1 f1 ttf2 f2 ttf3 f3 ttf4 f4 tfm tfm tfm tfm Y Y Y Y PFDem 1 -PFDem PFDem 1 -PFDem PFDem 1 -PFDem PFDem 1 -PFDem tfm ttf1 tfm < ttf1 tfm ttf2 tfm < ttf2 tfm ttf3 tfm < ttf3 tfm ttf4 tfm < ttf4 Unmitigated domino o domino o domino o domino o domino Final events fiia,1 fiib,1 fiia,2 fiib,2 fiia,3 fiib,3 fiia,4 fiib,4 Event tree analysis and LOPA approach for the identification of all possible scenario evolutions in the case of success and/or failure of the installed protection systems and adopted procedural measures. Final events Intermediate events PFD em Unmitigated domino f IIa,1 Y PFD 2 ttf 1 f 1 tfm tfm ttf 1 1 -PFD em f IIb,1 Y PFD 1 Failure of SB2 1-PFD 2 ttf 2 f 2 Y tfm tfm < ttf 1 o domino PFD em tfm ttf 2 1 -PFD em f IIa,2 f IIb,2 Primary fire f I Failure of SB1 PFD 2 ttf 3 Y tfm PFD em tfm < ttf 2 tfm ttf 3 o domino f IIa,3 f IIb,3 16

17 Step 3: Probabilistic assessment of mitigated domino events Characterization of the primary fire: frequency and determining the heat load and the type of exposure. ITERMEDIATE EVETS describing target equipment conditions resulting from the exposure to the primary fire, whose effects have been mitigated to a certain extent by the installed active and passive fire protection systems. FIAL SCEARIOS representing the final outcomes of the propagation of the primary fire after the assessment of emergency response: a) UMITIGATED (escalation) b1) MITIGATED (tfm > ttf i ) b2) O DOMIO (tfm < ttf i ) Overall frequency f D f IIa, i f IIb, i i i fi f j 1 j 0 I, i j1 PFD j j, i j-th barrier available ( 1 j-th barrier unavailable Comparison between the ttf and the time needed for emergency intervention f IIa, i fipd, i f i i IIb,i 1 0 f i if if P D,i PFD tfm tfm em 1 PFD ttf ttf i i em 17 j i )

18 Step 3: Probabilistic assessment of mitigated domino events Estimation of time to failure: use of simplified correlations ttf d exp cv e ln( Q ) HL Item Fire exposure conditions Correlation coefficients f e d c Pressurized vessel Flame engulfment Pressurized vessel Distant source radiation Atmospheric vessel Any f 18

19 Application to a case-study Escalation of a jet fire from the full bore rupture of an hydrogen pipeline in a LPG storage facility PV101-PV110 PL a) 20m b) 20m Heat load Flame engulfment 50 kw/m kw/m kw/m 2 PV101 PV102 PV103 PV104 a) Simplified layout for the analysis b) Heat load caused by the jet fire Target equipment features Horizontal pressurized LPG vessel Water Deluge Systems Pressure Relief Device PFP with intumescent coating External emergency team intervention Characteristics of equipment items under investigation Item ID Type Substance Capacity (m 3 ) Diameter (m) Length (m) PL Pressurized pipeline Hydrogen PV101-PV110 Pressurized vessels LPG Pressure (barg) Release and primary fire characterization Description Full bore rupture of hydrogen pipeline Release diameter (mm) Unit length frequency (y -1 m -1 ) Evaluated frequency (y -1 ) LOC Associated primary fire scenario Primary fire frequency (y -1 ) Jet fire

20 Determination of the complete event tree for the system Final event 1a Failure of coating Intermediate event 1 tfm =53 min Final event 1b Final event 2a Intermediate event 2 tfm =53 min Final event 2b Failure of PRD Final event 3a Failure of coating Intermediate event 3 tfm =53 min Final event 3b Final event 4a Intermediate event 4 tfm =53 min Final event 4b Jet-fire impinging on PV Failure of WDS Failure of coating Intermediate event 5 tfm =53 min Final event 5a Final event 5b Final event 6a Failure of PRD Intermediate event 6 tfm =53 min Final event 6b Final event 7a Failure of coating Intermediate event 7 tfm =53 min Final event 7b Final event 8a Intermediate event 8 tfm =53 min Final event 8b 20

21 LOPA integrated approach for the Probabilistic Assessment of Domino Effect Escalation: Methodology application (3) Intermediate Event ID Heat Load (kw/m 2 ) Time to failure (s) Escalation Probit Associated Escalation Probability Final Event ID Final Event Evaluated Probability Final Event Frequency (y -1 ) Type of resulting scenario 1a Unmitigated Domino 1b Mitigated Domino 2a Mitigated Domino 2b o Domino 3a Mitigated Domino 3b Mitigated Domino 4a Mitigated Domino 4b o Domino 5a Mitigated Domino 5b Mitigated Domino 6a Mitigated Domino 6b o Domino 7a Mitigated Domino 7b Mitigated Domino 8a Mitigated Domino 8b o Domino Overall Probability Overall frequency (y -1 ) Unmitigated Domino scenario scenario Escalation chain interrupted In absence of safety barriers EP = EF = y -1 21

22 Conclusions A methodology for the probabilistic assessment of fired domino effect was developed taking into account the role of safety barriers Site specific approach supported by a dataset of availability and effectiveness data Specific analytical functions (probit functions) were adopted to describe target equipment vulnerability The developed methodology was applied to a sample case study, in order to estimate the frequency of mitigated domino scenarios triggered by fire and the correspondent values of escalation probability. The analysis allowed quantifying the contribution of mitigation barriers in reducing the frequency of residual unmitigated domino events. Hence, if a proper fire protection strategy and adequate safety barriers are effectively designed and maintained, a significant reduction in credibility of fired domino escalation may be achieved. 22