QRA Model with Probabilistic Parameters and Its … QRA Model with Uncertainties 2 Contents 2 QRA Model Review 3 QRA with Uncertain Parameters 4 Application for Road Tunnels 5 Conclusion

2010-03-05 QRA Model with Uncertainties 1

QRA Model with Probabilistic Parameters and Its Application for Road Tunnels

Speaker: Xiaobo QuSupervisor: Qiang Meng

Presenter

Presentation Notes

Good afternoon, everyone. I am xiaobo, from national university of Singapore. It is a great honor for me to meet all of you here. Today I will share with you one of my ideas with respect to QRA model. This is a very new research topic and your comments are valuable for this study.


Contents

QRA Model Review2

QRA with Uncertain Parameters3

Application for Road Tunnels4

Conclusion and Future Work5

Background 1

Presenter

Presentation Notes

My presentation will be delivered to you in five parts. First, background. Followed by the QRA model review, which will briefly introduce the QRA model. After that, the QRA model with parameter uncertainty is proposed. Risk assessment for road tunnels is performed as the application of the new model. The last part concludes this presentation.


Background

Background A risk criteria is deemed required for tunnel >240 m in Singapore (the Road Traffic Act, 2005)EU Directive shows that the risk assessment is compulsory for every road tunnel in Europe (EU Directive, 2004)A project quantitative risk assessment of road tunnelscollaborated with LTA

Presenter

Presentation Notes

For any road tunnels longer than 240 meters in singapore, a risk criteria is required. Regarding the case in Europe, risk assessment is compulsory in every road tunnel according to EU directive. This is the motivation of initiating a project….. With lta.


Background

Input parameters Frequency of fire in tunnel (no fire record in CTE since it was officially open in 1989)Probability of fire detection system fails to workProbability of tunnel ventilation system fails to workAir velocity when tunnel ventilation fails to workEvacuate time when fire detection system fails to work ……

Uncertainty is an unavoidable element affecting outputs of the model!

Presenter

Presentation Notes

Actually the project has been completed and LTA is satisfied with our model and software. However, there do have some drawbacks of our model later I will brief the model in Section 2. The software is robust lta uses several very large cases to test it. However, when we are disucssing with lta experts, the input parameters, which are also critical to outputs, is difficult to obtain. For example, regarding the fire in tunnel, for singapore cte road tunnel, no any fire record since it was offically open in 1989, no fire disaster within the past twenty years. So which value should we use? Another very good example is related to the tunnel electrical and mechenical system. The probability of their working condition are random variable which have significant variability.. Therefore, uncertainty is an unavoidable element affecting outputs of the model. This paper will deal with the impacts of those uncertain parameters.


Background

How to represent uncertainties? Probability theory (pdf)Fuzzy theory (membership function)Evidence theory Uncertainty theory ……

Presenter

Presentation Notes

how to represent uncertainties? Literature reviwes shows that probability theory, fuzzy theory, possibility theory, evidence theory can represent various types of uncertainties.


Background

Literature review and objective Huang et al. (2000) initially considered the fuzzy parameters in event tree analysisBaraldi and Zio (2008) took the uncertainty propagation in event tree analysis into considerationThis paper will deal with the uncertainty propagation in quantitative risk assessment

Presenter

Presentation Notes

We went through the literatures with respect to dealing with uncertainties in decision analysis and found that huang et al 2000 initially consider fuzzy parameters in event tree analysis. Baraldi and zio 2008 took the uncertainty propagation in event tree analysis into consideration. His research is just the same idea with ours. However, they just take the frequency analysis into account. Regarding the uncertainty propagation in quantitative risk assessment, problems become different.


Contents

QRA Model Review2




Background 1


QRA Model Review

Introduction Use reliability and statistics to engineering design Proposed by US Atomic Energy Commission for evaluating the safety level of nuclear power plant (US Atomic Energy Commission, 1957)To assess the safety level of hazardous installations such as road tunnel, nuclear power plant, work zone, and etc.


QRA Model Review

Procedure

Identify basic condition of the system

Identify possible hazardous incidents- Analyze possible top events

Consequence estimation- Consequence estimation model

Frequency calculation- Fault tree and event tree analysis

Evaluation of safety level and risk- Calculate individual risk, societal risk and expected number of fatalities

Safety Evaluation - Set risk criteria and limits

EndYes

NoAccept the risks or not ? Risk reduction

measures

Presenter

Presentation Notes

Ever since 1975, this procedure is proposed and today we still use this framework. Firstly, identify the top events such as fire, flood, and then build event trees to get all the possible events. By using fault tree to obtain the frequency of top events, by using consequence estimation to obtain the consequence of each scenario. After that, calculate the risk index to evaluate the risk. Finnally, if necessary, some risk reduction measures are proposed.


QRA Model Review

Input and output of the model Input parameters

Probability of fire detection system fails to workProbability of tunnel ventilation system fails to workAir velocity when tunnel ventilation fails to workEvacuate time when fire detection system fails to work ……

Output Frequency and consequence of each particular scenario Risk index (societal risk and expected number of fatalities)

Presenter

Presentation Notes

For every model, we should pay attention to the input and output. The input of this model is all the parameters related to the tunnel operations. The outputs are freuqncy and conseuqnce and risk index.


QRA Model Review

Top events Depends on expert judgment and historical recordTrigger the event tree

Fault treeEstimate the frequency of top eventPresent all possible causes rendering a single event

Presenter

Presentation Notes

In the following I will briefly introduce various components of the model. Top events stands for the key hazards it depends on expert judgment and hisrot.


QRA Model Review

Event treeAll the possible scenarios Different frequencies and consequences Example road tunnel


QRA Model Review

Event Tree for road tunnels

Ventilation

Success (0.95)

Failure (0.05)

Off peak (0.7)

Frequencies Deaths & Injuries

f1

f2

f3

f4Fire in tunnelW = 9.095e-5

Peak hour(0.3)

Fire detection

f5

f6

f7

f8

Period of Day

Success (0.95)

Failure (0.05)

Success (0.99)

Failure (0.01)

Success (0.99)

Failure (0.01)

Success (0.99)

Failure (0.01)

Success (0.99)

Failure (0.01)

1 0.1n =

2 1.0n =

3 0.1n =

4 2.0n =

5 0.2n =

6 1.1n =

7 0.2n =

8 1.1n =

Event tree

Fault tree Consequence

Model


QRA Model Review

Risk IndexExpected number of fatalities Societal risk

Relationship between the frequency and number of people suffering from a specified level of harm from a top eventRepresented by F/N curve

F: likelihood of N or more fatalities (frequency)N: number of fatalities

Widely used in risk assessment for hazardous installations


QRA Model Review

Societal risk


Contents

QRA Model Review2




Background 1


QRA with Uncertain Parameters

Probabilistic Parameters Event tree parameters Fault tree parameters Consequence parameters



Example

Ventilation

Success (1-pd)

Failure (pd)

Off peak (0.7)

Frequencies Deaths & Injuries

f1

f2

f3

f4Fire in tunnelW = 9.095e-5

Peak hour(0.3)

Fire detection

f5

f6

f7

f8

Period of Day

Success (1-pd)

Failure (pd)

Success (1-pv)

Failure (pv)

Success (1-pv)

Failure (pv)

Success (1-pv)

Failure (pv)

Success (1-pv)

Failure (pv)

1 0.1n =

2 1.0n =

3 0.1n =

4 2.0n =

5 0.2n =

6 1.1n =

7 0.2n =

8 1.1n =



Methodology (Monte Carlo Simulation)Step 1: sample the ith realization of the probabilistic variable vector Step 2: compute the frequencies and fatalities for the ithrealizationStep 3: return to step 1 and repeat Step 4: stop when all the realizations are computed

The number of fatalities and frequencies of each scenario are uncertain variables which has n realizations



Risk indices Expected number of fatalities (Uncertain variables)Societal risk

Each realization has a societal risk


Contents

QRA Model Review2




Background 1


Application for Road Tunnels

Event tree structure



Probabilistic parameters Probability of fire detection working normally Probability of ventilation system working normallyEvacuate time when fire detection works normallyEvacuate time when fire detection fails to workAir velocity when ventilation system fails to workAir velocity when ventilation system works normally……



Probabilistic parameters

Detection failure rate pdf Ventilation failure rate pdf



Probabilistic parameters

Air velocity (ventilation normal ) Air velocity (ventilation failure)



Frequency estimation model

Consequence estimation model ( )∏

==

K

kjk SEP

ijf

1

)(

( )5.1849 0.0273( , ) / TDH DHF F t T t e −= =( )1.036

COCO

K X tF

D=

22

8.13 0.54(20.9 )oo XtF

e− −=

22

6.1523 0.5189* coco XtF

e−=



Frequencies of Scenario 1Scenario 1

0

0.2

0.4

0.6

0.8

1

1.2

1.00E+00 1.00E+01 1.00E+02 1.00E+03

Frequency

Pro

babi

lity

Scenario 1



Consequences of Scenario 1Consequence

0

0.2

0.4

0.6

0.8

1

1.2

0 10 20 30 40 50

Consequence

Prob

abili

ty

Consequence



Expected Number of FatalitiesExpected Value

0

0.2

0.4

0.6

0.8

1

1.2

0.00E+00

1.00E-01 2.00E-01 3.00E-01 4.00E-01 5.00E-01 6.00E-01 7.00E-01

Expected Value



Societal risk


Contents

QRA Model Review2




Background 1


Conclusion and Future work

Conclusion Proposed an approach to take the variability of parameters into account Apply the new model to road tunnel in Singapore Propose the risk index for new model


Conclusion and Future work

Future Work Degree of uncertainty Fuzzy or uncertainty Risk deduction Uncertainty deduction

QRA Model with Probabilistic Parameters and Its … QRA Model with Uncertainties 2 Contents 2 QRA Model Review 3 QRA with Uncertain Parameters 4 Application for Road Tunnels 5 Conclusion

Documents