RISKTOOLS from Fire Risk Forum

From Toolbox: White Papers Title: Event Tree Fire Loss Scenario Analysis Author: Thomas F. Barry, P. E. Company: TFBarry Publications email: [email protected] web site: www.fireriskforum.com Copyright: You may copy and forward this information for educational purposes only. For other purposes, please contact the author for permission.

ABSTRACT: This paper provides an introduction to developing and evaluating fire loss scenarios using event tree analysis (ETA). Developing credible loss scenarios is an important step in the risk assessment process. This paper introduces the concepts of Source Pathway Target analysis, event tree structuring, incident outcome grouping and also provides some examples. Information in this paper is extracted from the book. Risk-Informed, Performance-Based Industrial Fire Protection. An Alternative to Prescriptive Codes, by Thomas F. Barry, P. E. The information is based on actual projects at industrial facilities.

Become a Fire Risk Forum RISKTOOLS contributor! Share your knowledge and expertise. We need White Papers on risk and performance topics, Excel Spreadsheets for consequence and risk calculations, short reports on risk and performance-based fire safety projects, and fire modeling applications. Please use a similar cover page with contact information and an abstract. Feel free to add some background information about your company or organization. We can accept WORD, WordPerfect, PDF file formats. For further information contact [email protected]

Event Tree Loss Scenario Analysis INTRODUCTION

June 2003

A loss scenario represents the sequence of events that can result in undesirable fire or explosion incidents. The scenario development process must be: Sequentially structured in a time-related manner Credible in terms of realistic incident outcomes Contain sufficient information to allow the risk analysis team to quantify the scenario As presented in Fig. 3.1, loss scenario development is step 3 in the risk-informed, performance-based decision making process which is described in reference [1]. The input from steps 1 and 2 define the parameters for the types of fire scenarios and level of detail that will need to be developed. Fig. 3.1: Risk-Informed, Performance-Based Fire Protection Steps
ANALYSIS 3 Loss Scenario Development 4 2 Risk Tolerance Criteria 5 Initiating Event Likelihood Exposure Profile Modeling 6 Fire Protection System (FPS) Performance Success Probability PERFORMANCE ASSESSMENT 7 Risk Estimation & Comparison With Risk Tolerance 8 Cost/Benefit Analysis of Risk Reduction Alternatives

APPRAISAL

Program Objectives

The term loss scenario development can best be defined as the reasoning methodology that describes the sequential relationships between time dependent fire loss events. In general, scenario modeling is an abstract representation of a real-world system or subsystem. Time limitations, complexity, and costs generally preclude the development of logic models, which represent every aspect and detail of fire loss exposure potential. Therefore, in most cases fire scenario logic models provide us with a simplified approach that attempts to include the significant aspects of fire events relevant to the risk-based decisions under evaluation. Section 3.2 describes event sequencing using the concept of Source Pathway Target. This methodology provides systematic input for structuring event tree logic and focuses on the potential risk at a defined target or targets. Section 3.3 presents an overview of event tree structuring with some examples. Scenario analysis is an important aspect of conducting credible risk-informed, performancebased projects. A comprehensive approach involves: Applying a consistent framework and systematic methodology Identifying contributing factors associated with fire and explosion (F&E) initiating events

www.fireriskforum.com

Page 1

Event Tree Loss Scenario Analysis

June 2003

Evaluating the exposure-versus-time profile following the initiating event. Exposure elements include fire growth, propagation, incident outcomes, and potential consequence levels Assessing the performance of fire protection systems (detection systems, passive and active protection systems, life safety elements, etc.) in terms of modifying the outcome of the initiating event It is important to point out that this paper, introduces the development of a scenario-based framework focusing on event tree structuring. Chapters 4 through 6 in reference [1] provide the additional building blocks towards comprehensive scenario analysis, as illustrated in Fig. 3.2. Fig. 3.2: Building Blocks Towards Comprehensive Scenario Analysis
CHAPTER 4 INITIATING EVENT LIKELIHOOD CHAPTER 5 EXPOSURE PROFILE MODELING CHAPTER 6 FIRE PROTECTION SYSTEM PERFORMANCE Assessment of the performance of risk reduction measures

CHAPTER 3 LOSS SCENARIO DEVELOPMENT Identification of scenario sequences using event trees

Evaluation of initiating events and contributing factors

Evaluation of incident outcomes, exposure profiles and consequence levels

3.2 SEQUENCE OF EVENTS A scenario represents a set of time-related events (transition states) that can lead to various fire or explosion outcomes. One systematic approach towards structuring fire scenarios is the Source (S) Pathway (P) Target (T) method: Target: The Target is the focus of the risk-based study and is the first thing to define. To be at risk, the target must be vulnerable to loss from the source fire exposure and have defined value to the owner. Source: After defining the targets vulnerability and value, the initiating source fire(s) that could expose the Target to loss must be identified and screened. Pathway: Pathway events include those factors that either propagate or modify the source fire(s). Pathway propagation factors can include fire growth, secondary fuel ignition, uncontrolled flame spread, etc. Pathway modification factors, which will reduce the exposure to the target, include fire protection systems (FPSs) such as detection, emergency control systems (ECSs), suppression systems, fire barriers, manual fire-fighting efforts. Figure 3.3 presents an illustration of the S P T (Source Pathway Target) concept www.fireriskforum.com Page 2

Event Tree Loss Scenario Analysis

June 2003

applied to an example production building. In this example, the scenario boundaries include the

entire production building. The target or risk-study focus is the Control Room. The selected source fire for scenario evaluation is a hydraulic oil system fire. Fig. 3.3: Example of Production Building S P T Concept

An example S P T scenario development worksheet is presented in Table 3.1 for the production building example . The table is set up as follows: A. Evaluation Summary What are the scenario boundaries? What are the specific scenarios under evaluation? What is the risk focus? B. Target What is the target? What is the fire exposure vulnerability to the target? Temperature effects Radiant heat effects Smoke concentration effects Toxic, corrosive gas effects What is the value of the target? Property replacement values Production downtime/business interruption (BI) value Trained operators value C. Source(s) What are the potential initiating fire event occurrences? What are the events that should be selected for further evaluation? D. Pathway What are the pathway propagators? Fire progression potentials Structural failures, domino effects www.fireriskforum.com Page 3

Event Tree Loss Scenario Analysis What are the pathway modifications (i.e., FPSs)? Existing Proposed

June 2003

www.fireriskforum.com

Page 4

Event Tree Loss Scenario Analysis

June 2003

E. Assumptions/Actions What are the limitations and assumptions of the scenario evaluation? Actions How many event trees will need to be structured (i.e., based on credible initiating bounding fire source inputs and FPSs options) Note: Event tree structuring is described in Sect. 3.3. There can be a wide variation in fire hazard evaluations conducted by different people concerning identification and selection of initiating fire event scenarios. For example, one person may speculate a small combustible fluid leak and ignition, while another may imagine a total release of all the combustible fluid inventory. A credible initiating event is one that is both within the realm of possibility and is likely to be severe enough to cause significant damage (i.e., exceed risk tolerance limits). Generally, at this stage in the risk analysis, what constitutes credibility, reasonable likelihood, and significance is mostly based on the experience and qualitative judgement of the risk assessment team. Having a team consensus approach and a consistent methodology to identify and screen scenarios for further risk review is a very important aspect of conducting a credible risk-informed study.

www.fireriskforum.com

Page 5

Event Tree Loss Scenario Analysis Table 3.1: Example Scenario Development Worksheet
Evaluation team date: February 7, 2001 Team members: Stan Smith Safety Tom Jones Engineering Jim Jones Maintenance Hank Edwards Operations Kevin Reed Fire Risk Analyst

June 2003

A: Evaluation summary Production Building ABC an automated production facility. Scenario boundary includes entire production building. Risk focus: fire exposure to critical control room operations B: Target Control Room Vulnerability (1) Electronic Equipment Operators 1. Connective heat Highly Highly from fire plume/ vulnerable vulnerable hot gas layer 2. Smoke, combustion Highly Highly gases (toxic, corrosive) vulnerable vulnerable C: Sources(s) (Initiating fire occurrences) (2) Expanded list of potential initiating fire sources

Target value Control room equipment replacement values: $50M Replacement time: 6 months 100% business interruption for 6 months Highly trained operators staff control room

Selected for further evaluation 1. Hydraulic system oil fire 1A. Torch (jet) fire 1B. Pool fire 2. Production equipment fire 2A. Hydraulic oil 2B. Electrical fire 3. Cable tray fire 3A. Smoldering fire Pathway modification factors (i.e., FPSs) Proposed options to evaluate Detection system above hydraulic systems and cable trays Emergency control system for quick, reliable shutdown of hydraulic oil pumps and electrical power Installation of quick response heads on sprinkler system Reinforcement of control room walls Actions (3) Structure event trees to analyze source fires: 1A 1B 2A 2B 3A

D: Pathway modification factors (i.e., fire protection systems): Existing Ceiling-level, ordinary hazard sprinkler system Note: Existing sprinkler protection would not be effective against a hydraulic oil torch fire or a smoldering electrical fire

E: Assumption List assumptions/limitations of scenario evaluation.

Remarks:

(1) Quantification of vulnerability levels is addressed in Chap. 5, Exposure Profile Modeling [1] (2) Initiating source identification and screening is addressed in Sect. 3.5 (3) Event tree structuring is discussed in Sect. 3.3

www.fireriskforum.com

Page 6

Event Tree Loss Scenario Analysis

June 2003

Figure 3.4 presents an example of source fire potentials as listed in sections C and E of the scenario development worksheet, Table 3.1. Note that each selected initiating event source fire has a related event tree. Fig. 3.4: Source Fire Exposures to Target and Related Event Trees

www.fireriskforum.com

Page 7

Event Tree Loss Scenario Analysis 3.2.1 Some Examples

June 2003

The S P T concept provides a method for systematic thinking and reasoning when attempting to develop scenarios. The concept has a wide range of application from outside chemical process structures to multi-compartmented industrial and commercial buildings. Figure 3.5 presents an example of an exterior exposure to a critical control room operation. Applying the concept of S P T and using a scenario development worksheet (such as the in Table 3.1) provides a framework for developing event tree logic. The event tree logic, which will be addressed in Sect. 3.3, expands on pathway modification options. Fig. 3.5: Example of Exterior Exposure

Figure 3.6 applies the concept of S P T to an example multi-floor compartmented building. Again the process is the same: start with qualitative definition of the targets value and vulnerability, identify and screen source fire potentials, identify pathway propagation factors, and identify possibly pathway modification options (i.e., FPSs).

www.fireriskforum.com

Page 8

Event Tree Loss Scenario Analysis Fig. 3.6: Example of Multi-Floor Building

June 2003

Figure 3.7 is an example of the S P T approach applied to off-site risk (i.e., fire or explosion risk beyond the plant site boundaries). For this example, Source 2 may represent the most severe off-site impact from a liquified flammable gas release, flash fire, BLEVE, or vapor cloud explosion. Pathway propagation factors may include topography, wind speed and direction, location of ignition sources, etc. Pathway modification options might include gas and flame detection, an emergency shutdown system (to reduce the release size), and a fixed FPS (e.g., fire proofing, water spray).

www.fireriskforum.com

Page 9

Event Tree Loss Scenario Analysis Fig. 3.7: Example of Off-Site Risk

June 2003

Figure 3.8 provides a general illustration of the relationship between the S P T scenario development concept and event tree structuring. Event tree structuring and analysis is discussed in the next section. Fig. 3.8: Relationship Between S P T to Event Tree Structuring
PATHWAY (P) MODIFIERS Fire Barriers TARGET (T) VULNERABILITY Incident Outcomes

SOURCE(S)

Fire Fire Detection Suppression & Emergency Shutdown

Consequences

Initiating Fire Event Occurrence

Performance of Fire Protection Systems

Exposure at Target

www.fireriskforum.com

Page 10

Event Tree Loss Scenario Analysis 3.3 EVENT TREE STRUCTURING

June 2003

The most widely used risk assessment technique for structuring fire loss scenarios involves the use of event tree analysis (ETA). ETA conveys the initiating source fire event, fire protection performance (FPS), incident outcomes, and consequences. Using ETA provides the major advantage of being able to incorporate time and conditional FPS performance probability into a scenario. ETA is used to analyze both simple and complex processes in which several FPS levels may be in place to respond to specific initiating events. To complete an ETA, a risk analyst must identify initiating source fire events, evaluate the performance of existing or proposed fire protection options and evaluate incident outcomes and consequences. Event trees are similar to failure modes and effects analysis (FMEA) as they start with identifying a failure mode and then assess potential incident outcomes and consequences. However, methods such as FMEA or hazard and operability analysis (HAZOPS) dont usually provide sufficient detail on FPS performance, nor do they provide a vehicle for estimating probabilities of fire protection performance success and quantifying risk. ETA has several advantages in that it: Provides systematic organization of fire loss event stages Orders events in time-related sequences Identifies significant top fire loss events for subsequent FTA and probability modeling Allows evaluation of fire detection and protection system alternatives Demonstrates the relationship between the success of FPSs and the potential fire incident outcomes, consequences, and risk levels To structure event tree modeling logic, the risk evaluation team must have knowledge of potential initiating events (i.e., equipment failures, human errors, external events, or system upsets that could lead to fire or explosions) and intermediate events such as FPS functions and emergency procedures. Event sequences must be individually evaluated to generate F&E loss scenarios. The general event tree modeling logic sequence stages are shown in Fig. 3.9.

www.fireriskforum.com

Page 11

Event Tree Loss Scenario Analysis Fig. 3.9: Event Tree Modeling Logic Sequence Stages Initiating Fire Source(s) Events

June 2003

7
Intermediate Conditional Events Fire Protection System Performance

7
Fire or Explosion Incident Outcomes

7
Consequences at the Target This section introduces event tree structuring and provides an overview of ETA steps. Quantification of initiating event likelihoods, FPS performance success probabilities, and exposure profile modeling at the target are covered in other chapters of reference [1]. 3.3.1 ETA Steps Figure 3.10 present an example fire risk event tree framework. Event tree development steps include:

Identifying the initiating fire source event. Identifying the pathway factors to be evaluated within the event tree analysis. For example, FPSs, existing or proposed, which will modify the propagation of the initiating fire source event. Structuring the event tree branch logic and time line. Assessing the incident outcomes. Identifying and quantifying exposure and consequences of concern at the target. Quantifying branch line probabilities. This involves quantifying the initiating event likelihood and conditional probabilities of pathway factors (i.e., FPS performance success). Quantifying the risk.

www.fireriskforum.com

Page 12

Fig. 3.10:

Example Fire Risk Event Tree Framework

1 2 5
PATHWAY FACTORS TARGET
FIRE PROTECTION SYSTEM PERFORMANCE SUCCESS PROBABILITY

SOURCE

www.fireriskforum.com

INITIATING EVENT LIKELIHOOD

3
EXPOSURE PROFILE/CONSEQUENCE LEVELS
[G] [J] [H] [I] [E] [F]

RISK ESTIMATIONS

[A]
Emergency Control System Successful Automatic Suppression System Successful Manual Fire Department Suppression Successful

[B]

[C]

[D]

[K]

[L] [L] = [F] X [K]

Flammable Liquid Release and Ignition

Detection System Successful

PROPERTY DAMAGE AGGREGATE LIFE BRANCH EQUIVALENT ANNUALIZED F&E RISK STOCK PRODUCTION SAFETY MONETARY LINE INCIDENT BUILDING EQUIPMENT BRANCH $ / YEAR DAMAGE DOWNTIME EXPOSURE VALUE, $ ID PROBABILITY OUTCOME DAMAGE DAMAGE

Event Tree Loss Scenario Analysis

6
.60 .70 [C1] .40 [D1] [E1] .20 .30 [C1] .80 [D2] .10 [E2] .60 .20 [C2] .40 [D3] [E3] .20 .80 [C2] .80 [D4] [E4] Check 2-5 5 - 10 10 - 30 12 .90 [E4] .10 11 [D4] 10 .017 .15 0.33 minutes .04 9 .017 .80 [E3] 8 .20 .004 G2 G3 G1 G2 G4 [D3] 7 .03 G1 6 1 2 3 1 2 4 .0016 G3 3 [E2] .90 5 .014 G2 2 [D2] 4 2 4 2 2 3 2 2 5 .004 G1 1 2 3 .0009 G3 3 3 .05 [E1] 2 3 1 2 3 1 2 3 2 2 4 .95 .018 G2 1 2 2 [D1] 1 .028 G1 1 1 1

1 1 3 1 2 3 1 2 3 1 2 4

0 0 0 0 0 1 0 0 3 0 0 4

5k 10k 1M 7k 25k 2M 7k 25k 2.5M 10k 25k 6M

140 180 900 28 350 3,200 210 100 42,500 400 425 900,000

.20

[B1]

0.33

YES

Fires/Year

Success ?

Initiating Event Likelihood

NO

[B1]

.80

TIME LINE

1-3

Total Annualized Risk; $/Year

948,433

June 2003

Page 13

Event Tree Loss Scenario Analysis

June 2003

Identifying initiating fire source events

The initiating event is the first event that must be identified in constructing fire or explosion event tree scenarios. This event may be a system or equipment failure; a human error; a process upset; an external exposure such as an earthquake, a vehicle accident, arson, or a security failure, which when coupled with ignition, creates an initiating fire source event. As depicted in Fig. 3.11, we want to use a systematic process for identifying, screening, and grouping initiating source fire events to keep the number of event trees and probability calculations manageable. Fig. 3.11: Example of Input Approach for Initiating Fire Events
Event Tree Logic

Initiating Fire Events Selection & Quantification Input

Screening & Classification

Design Basis Events

High Challenge Events

Worst Case Events

Review & Grouping

$ Representative Grouping $ Compilation & Review

Initiating Event Identification

Scenario Identification Worksheet

Fault Tree Analysis

Historical Incident Data

Plant Data Past History

Experience / Judgments

www.fireriskforum.com

Page 14

Event Tree Loss Scenario Analysis

June 2003

As illustrated in Fig. 3.11, identification sources for primary initiating events include: C Scenario identification worksheets C FTA C Historical incident data C Plant data, plant history C Hazard surveys, experience, engineering judgements The initiating events are grouped in such a way that the failure modes of each group would impose the same or very similar performance demands on the FPSs. Groups of initiating events are then screened further in terms of the likelihood and the consequences of possible outcomes. Initiating event groupings are then placed into a scenario classification scheme such as designbasis events, high challenge events, or worst case events which are defined in Sect. 3.5. The risk evaluation team selects the initiating event cases of interest to the risk-based study. This process will also determine how many ETAs will have to be conducted. Initiating event analysis using qualitative FTA and quantification is addressed in detail in Chap. 4, Initiating Event Likelihood, in reference [1] and therefore will not be expanded on in this section.

Identifying pathway factors

Pathway factors are those events that follow the initiating event in sequence. In constructing event tree logic, the analyst must identify significant and relevant pathway factors that will affect fire growth and propagation or mitigation of the initiating event. Intermediate pathway factor events represent both a conditional state and a function of time, which is very important to recognize when addressing conditional probabilities. Both propagating and FPS factors need to be identified. Figure 3.12 provides an example breakdown of intermediate pathway events in terms of: C Fire propagation factors C FPSs (i.e., fire mitigation) Some fire growth or propagation factors of interest in scenario development are: C Fuel properties (heat release rate) C Flame spread and secondary ignition C Ventilation effects C Structural failures C Operator emergency response

www.fireriskforum.com

Page 15

Event Tree Loss Scenario Analysis

June 2003

FPSs of interest are: C Detection systems C ECSs C Automatic suppression systems C Propagation limiting features (i.e., fire barriers) C Manual suppression systems C Confinement features (construction or separation features that confine a fire to the area of origin) Fire growth and propagation factors are discussed in detail in Chap. 5, Exposure Profile Modeling and FPS performance is covered in Chap. 6 in reference [1]. Fig. 3.12 : Example of Some Pathway Factors
PATHWAY MODIFICATION FACTORS

Fire Propagation Factors

Fire Protection Systems

Structural Failures Ventilation / Weather Conditions Human Error Improper Response

Fuel Properties / Heat Release, Flame Spread

Domino Effect Secondary Fire or Explosions/ Secondary Fuel Releases / Equipment Failures

Detection Systems

Emergency Control Systems

Automatic Suppression Systems

Propagation Limiting Features

Manual Suppression Systems

Confinement Features (Barriers)

www.fireriskforum.com

Page 16

Event Tree Loss Scenario Analysis

June 2003

Structuring the event tree branch logic

The event tree displays the development of accident sequences, beginning with the initiating event and proceeding to the responses of the FPSs. The results are defined fire incident outcomes that can result from the initiating event. The FPS functions should be depicted chronologically, although many times the events may occur almost simultaneously. Both existing and proposed FPSs can be evaluated. The first step in constructing the event tree is to enter the initiating event and FPSs that apply to the analysis. The initiating event is listed on the left side of the page (i.e., first event), and the FPSs are listed across the top of the page from left to right. An example event tree is shown in Fig. 3.13 depicting the headings: C Initiating Event Likelihood [A] C Fire Protection System Events [B] [E] C Exposure Profile Consequence Levels [G] [J] The initiating source event is a frequency or likelihood (i.e., events/year). The pathway factor (i.e., FPS performance) are conditional probabilities (i.e., a number between 0 and 1). Concerning the event tree structure: C Event tree logic is from left to right C This event tree indicates system success in the upward (YES) branch segments Probability of Success = 1 - Probability of Failure C Branch line probabilities are calculated by multiplying the initiating event likelihood [A] and the conditional branch event segment conditional probabilities [B] [E], for example: Branch lines (Branch I.D.) 6 and 12 could represent an uncontrolled fire scenario. If we wanted to estimate the annualized probability of an uncontrolled fire situation: Branch Line 6 Probability = [A] x [B1] x [C1] x [D2] x [E2]

Branch Line 12 Probability = [A] x [B1] x [C2] x [D4] x [E4]

The line over the letter, (for example; ) indicates that this is a system failure branch. NOTE: Chapter 7 on Risk Estimation in reference [1] provides detail on ETA probability calculations. C Many fire or explosion incident outcomes are possible and are developed through different branches of the event tree. The event tree outcomes are then assessed according to potential exposure versus consequence level or category. C The time line establishes a frame of reference for estimating FPS success probabilities and estimating the magnitude of the consequences at the time intervals. For example, a 10-min limit could be established as the point where structural steel damage could start to occur from a flammable liquid fire exposure. www.fireriskforum.com Page 17

[C1]

Event Tree Loss Scenario Analysis

June 2003

Therefore, to be successful in minimizing damage, we would want to detect, shutdown and suppress a fire before10-min elapsed. NOTE: Chapter 5, Exposure Profile Modeling in reference [1], addresses time line development. Fig. 3.13:
SOURCE
INITIATING EVENT LIKELIHOOD [A]
Flammable Liquid Release and Ignition

Example: General Event Tree Structure

PATHWAY FACTORS
FIRE PROTECTION SYSTEMS PERFORMANCE [B] [C]
Emergency Shutdown System Successful

TARGET
EXPOSURE PROFILE CONSEQUENCE

[D]
Automatic Suppression System Successful

[E]
Manual Fire Department Suppression Successful

[F]

[G]

[H]

[I]

[J]

Fire Detection Successful

BRANCH ID

BRANCH INCIDENT PROBABILITY OUTCOME PEOPLE PROPERTY OTHER

[D1] [C1] [D1] [B1] [D2] [A] Fires/Year

Initiating Event Frequency Flammable Liquid Fire

1 2 3 4 5 6 7 8 9 10 11 12

[E1] [E1]

YES

[C1] [D2]

[E2] [E2]

NO [C2]

[D3] [E3] [D3] [E3] [D4] [C2] [D4] [E4] [E4]

The Event Tree indicates fire protection systems success in the upward (YES) branch segments

[B1]
FTA-A1

Event Tree Logic is from left to right

TIME LINE, minutes

Incident Outcome Assessment

As presented in Fig. 3.14, potential incidents of primary interest in fire and explosion (F&E) risk-based evaluations include the following: C Heat from a fire: pool fire, torch fire, flash fire, BLEVE C Explosion overpressures: vapor cloud explosions, dust explosions, tank or equipment rupture or fragmentation C Corrosive smoke or combustion products concentration, toxic gas concentrations

www.fireriskforum.com

Page 18

Event Tree Loss Scenario Analysis Fig. 3.14: Example of Some Fire and Explosion Incident Outcomes
INCIDENT OUTCOMES

June 2003

Heat Effects Exposure

Toxic Combustion Product(s) Exposure

Explosion Overpressure Exposure

Pool Fires

3-D Fire Situations Solids, Storage Fires

Torch Fires (flame jet)

Flash Fires

Fire Balls (i.e., from BLEVES)

Vapor Cloud Explosion

Missles/ Fragments from BLEVES

Confined Explosions

Dust Explosions

NOTE: Incident outcome exposure and consequence assessment is addressed in detail in Chap. 5, Exposure Profile Modeling in reference [1]. Incident Outcome Grouping Of special interest in the fire and explosion (F&E) risk assessment process are those event tree scenario branch lines that represent: C The best-case scenario C The worst-possible case scenario C Other likely case scenarios that may be of interest Traditionally, in the fire protection industry, scenarios have been related to the following definitions, which were initially established by the fire insurance industry:

C C C

Normal loss expectancy (NLE) best case Probably maximum loss (PML) other likely scenarios Maximum foreseeable loss (MFL) worst case

Although there are different variations used in the definition of these loss expectancy levels, in general terms they can be defined as follows: C NLE This is the loss scenario assuming all detection and protection features are in service and operating as designed. C PML This is the loss scenario assuming the primary automatic protection system (i.e., such as an automatic water spray system) is out of service. C MFL This is the worst-case loss scenario, which usually assumes that all detection and automatic protection systems are out of service or

www.fireriskforum.com

Page 19

Event Tree Loss Scenario Analysis

June 2003

ineffective and in many cases the loss severity is only limited by passive mitigation features (fire walls, separation, etc.). Figure 3.15 provides a general event tree depiction of how these scenarios can be related to the branches of the event tree. Fig. 3.15:
SOURCE
INITIATING EVENT LIKELIHOOD [A]
Flammable Liquid Release and Ignition

Example of ETA Incident Outcome Grouping

PATHWAY FACTORS
FIRE PROTECTION SYSTEM PERFORMANCE [B] [C]
Emergency Shutdown System Successful

TARGET
EXPOSURE PROFILE CONSEQUENCE

[D]
Automatic Suppression System Successful

[E]
Manual Fire Department Suppression Successful

[F]

[G]

[H]

[I]

[J]

Fire Detection Successful

BRANCH INCIDENT BRANCH PROBABILITY OUTCOME PEOPLE PROPERTY OTHER ID

[D1] [C1] [D1] [B1] [D2] [A] Fires / Year

Initiating Event Frequency Flammable Liquid Fire

1 2 3 4 5 6 7 8 9 10 11 12

G1 G2 G3 G1 G2 G3 G1 G2 G3 G1 G2 G4

[E1] [E1]

YES

[C1] [D2]

[E2] [E2]

NO [C2]

[D3] [E3] [D3] [E3] [D4] [C2] [D4] [E4] [E4]

[B1]
FTA-A1

TIME LINE, minutes

[G] INCIDENT OUTCOME GROUPINGS

INCIDENT OUTCOME GROUPINGS

G1 - Automatic Suppression Successful G2 - Fire Department Successful G3 - Automatic Suppression/Fire Department Not Successful (assumes delayed Fire Department Suppression at 60 minutes) G4 - Uncontrolled Fire (assumes 2-hour duration)

Table 3.2 provides an example of grouping scenarios based on the definition of loss expectancy. www.fireriskforum.com Page 20

Event Tree Loss Scenario Analysis Table 3.2: Loss Expectancy Grouping
GROUPED SCENARIOS G1 Automatic suppression system successful G2 Automatic suppression system not successful/fire department successful G3 Automatic suppression system/fire department not successful assumes delayed fire department suppression at 60 min G4 Uncontrolled fire assumes a 2-hr fire duration

June 2003

LOSS EXPECTANCY DEFINITION Best case situation generally related to an NLE normal loss expectancy analysis Selected probable case generally related to an PML probable maximum loss analysis Selected probable case generally related to an PML probable maximum loss analysis Worst case situation generally related to an MFL maximum foreseeable loss analysis

Chapter 7, Risk Estimation and Comparison, in reference [1] discusses how these scenario groupings are used in the development of graphical risk profiles.

Identification and quantification of exposure and consequence levels

Figure 3.16 provides a summary of some consequence concerns in F&E risk-based studies. The important thing to recognize here is that fire and explosion incidents represent a multiconsequence exposure, including property damage (PD), business interruption (BI), people exposure, environmental impacts, regulatory fines, and company image. Combined, these consequences can lead to very high monetary loss. Fig. 3.16: Example of Some Consequence Concerns
CONSEQUENCES

Property Damage $ Buildings $ Equipment $ Contents

Production Downtime (Business Interruption)

People Exposure $ On-Site $ Off-Site

Environmental Exposure $ Air, Soil Contamination

Regulatory Fines Legal Lawsuits

Company Image Loss of Customers

www.fireriskforum.com

Page 21

Event Tree Loss Scenario Analysis

June 2003

Quantification of branch line probabilities

Chapter 4, Initiating Event Likelihood, and Chap. 6, Fire Protection System Performance Success Probability in reference [1], provide information on developing probability and likelihood estimates.

Quantify the risk

The likelihood of each incident outcome is determined by multiplying the initiating event frequency with the conditional probabilities of the FPS performance success along each path leading to that outcome. As illustrated in Fig. 3.10, risk is the product of multiplying the branch line likelihood with the consequence value. Risk estimation is described in Chap. 7, Risk Estimation and Comparison with Risk Tolerance in reference [1]. 3.3.2 Example Industrial Building ETA Structure An example Production Building is shown in Fig. 3.17. Pathway Modifiers, which in this case are FPS options, are listed. Figure 3.18 presents an example event tree that could be used to evaluate risk levels versus fire prevention and fire protection options. Note how the sequence of events is set up: Release and Ignition (Initiating Fire Event)

Automatic Detection Performance

Emergency
Control Performance

Automatic Sprinkler System Performance

Control Room Barrier Performance

www.fireriskforum.com

Page 22

Event Tree Loss Scenario Analysis Fig. 3.17: Example of Potential Pathway Modifiers for a Production Building

June 2003

www.fireriskforum.com

Page 23

Event Tree Loss Scenario Analysis Fig. 3.18:

SOURCE
INITIATING EVENT LIKELIHOOD [A]
Hydraulic Oil System Release and Ignition

June 2003

Example Production Building Fire Risk Event Tree Structure

PATHWAY FACTORS
PERFORMANCE OF FIRE PROTECTION SYSTEMS [B] [C]
Emergency Control System Successful

TARGET
EXPOSURE PROFILE CONSEQUENCE LEVELS

[D]
Automatic Sprinkler System Successful

[E]
Control Room Fire Barrier Successful

[F]

[G]

[H]

[I]

[J]

Automatic Detection System Successful

BRANCH ID

CONTROL ROOM BRANCH INCIDENT PROBABILITY OUTCOME PEOPLE EQUIPMENT OTHER

[D1] [C1] [D1] [B1] [D2] [A] Fires/Year

Initiating Event Likelihood

1 2 3 4 5 6 7 8 9 10 11 12

[E1] [E1]

YES

[C1] [D2]

[E2] [E2]

NO [C2]

[D3] [E3] [D3] [E3] [D4] [C2] [D4] [E4] [E4]

[B1]

TIME LINE, minutes

www.fireriskforum.com

Page 24

Event Tree Loss Scenario Analysis 3.3.3 Example Exterior Exposure ETA Structure

June 2003

Figure 3.19 shows an exterior transformer fire exposure. For example purposes, some pathway modifiers (FPS options) have been listed. Fig. 3.19: Example of Potential Pathway Modifiers for an Exterior Exposure

PATHWAY MODIFIERS Fire Protection System Options: Automatic Water Spray Install as exposure protection between the transformer and wall Detection and Fire Brigade Response Upgrade Wall Integrity Upgrade existing wall to a 2 hr rated fire barrier wall Assumptions Cost prohibitive to relocate transformer

Figure 3.20 provides an example event tree for what we will call risk reduction option 1, installation of an automatic water spray system to provide exposure protection between the transformer and wall.

www.fireriskforum.com

Page 25

Event Tree Loss Scenario Analysis Fig. 3.20:

[A]
Initiating Fire Occurs

June 2003

Example: Event Tree for Risk Reduction Option 1 Automatic Water Spray
[B]
Fire Detection Successful

[C]
Automatic Water Spray Protection Successful

[D]
Fire Brigade Successful

[E]
Wall Barrier Integrity Maintained CONSEQUENCE LEVELS BRANCH ID CONTROL BRANCH INCIDENT PROBABILITY OUTCOME PEOPLE EQUIPMENT

[C1]

[B1]

YES [D1]

2 3 4

[A] Fires / Year

Initiating Event Frequency Transformer Oil Fire

[C1] [D1]

[E1] [E1]

[C2] NO [B1] [D2] [C2] [D2] [E2] [E2]

6 7 8

TIME LINE, minutes

1-2

1-3

5-7

10

20 - 30 Suppression Limit

Figure 3.21 provides an example of a risk reduction option 2, upgrading the existing wall to a 2-hr rated fire barrier wall and relying on the fire brigade for fire suppression.

www.fireriskforum.com

Page 26

Event Tree Loss Scenario Analysis Fig. 3.21:

[A]
Initiating Fire Occurs

June 2003

Example Event Tree for Risk Reduction Option 2 Fire Barrier Wall
[B]
Fire Detection Successful

[C]
Automatic Water Spray Protection Successful

[D]
Fire Brigade Successful

[E]
Wall Barrier Integrity Maintained CONSEQUENCE LEVELS BRANCH ID CONTROL BRANCH INCIDENT PROBABILITY OUTCOME PEOPLE EQUIPMENT

[C1]

N1

[B1]

YES [D1]

1 2 3

[A] Fires / Year

Initiating Event Frequency Transformer Oil Fire

[C1] [D1]

[E1] [E1]

[C2] NO [B1]

N1

[D2] [C2] [D2] [E2] [E2]

4 5 6

TIME LINE, minutes

0 Note 1 (N1)

1-2

1 - 20

10

20 - 30 Suppression Limit

No automatic water spray protection included in Risk Reduction Option 2.

3.3.4 Example Off-Site Risk ETA Structure Figure 3.22 presents an example for off-site risk. For example purposes, some pathway modifiers (FPS options) have been added to Source 2, Liquified Flammable Gas (LFG) Tanks.

www.fireriskforum.com

Page 27

Event Tree Loss Scenario Analysis Fig. 3.22: Example of Some Potential Pathway Modifiers for an Off-Site Risk

June 2003

PATHWAY MODIFIERS Fire Protection System Options: Automatic Detection Systems Gas releases Flame Emergency Shutdown Systems (ESS) Automatic, Fail-safe isolation valves to minimize release size Automatic Water Spray Cooling on Tanks Fireproofing on Tanks Fire Department Response

Figure 3.23 provides an example event tree for off-site risk from a LFG tank. Following accidental release from containment, we need to examine the potential for immediate ignition creating a torch fire, emergency shutdown and isolation to reduce the release size or fire duration, and if there is not immediate ignition, the chances of delayed ignition resulting in a flash fire or vapor cloud explosion. The modeled vapor cloud or BLEVE exposure distance, which could create major injuries and fatalities would be quantified as part of the consequence assessment. This is addressed in Chap. 5 in reference [1].

www.fireriskforum.com

Page 28

Event Tree Loss Scenario Analysis

June 2003

www.fireriskforum.com

Fig. 3.23:

Example Event Tree for Off-Site Risk from LFG Tanks

Page 29

Event Tree Loss Scenario Analysis REFERENCES 1

June 2003

Barry, Thomas, F., Risk-Informed, Performance-Based Industrial Fire Protection, An Alternative to Prescriptive Codes. TFBarry Publications, Knoxville, TN. November 2002. Available at www.fireriskforum.com. Guidelines for Hazard Evaluation Procedures, Second Edition with Worked Examples, Center for Chemical Process Safety (CCPS), American Institute of Chemical Engineers (AIChE), New York, NY, 1992.

www.fireriskforum.com

Page 30