Failure record discounting in Bayesian analysis in Probabilistic Risk Assessment (PRA) : a space system application

In estimating a system-specific binomial probability of failure on demand in Probabilistic Risk Assessment (PRA), the corresponding number of observed failures may be not directly applicable due to design or procedure changes that have been implemented in the system as a result of past failures. A methodology has been developed by NASA to account for partial applicability of past failures in Bayesian analysis by discounting the failure records. A series of sensitivity analyses on a specific case study showed that failure record discounting may result in failure distributions that are both optimistic and narrow. An alternative approach, which builds upon NASA's method, is proposed. This method combines an optimistic interpretation of the data, obtained with failure record discounting, with a pessimistic one, obtained with standard Bayesian updating without discounting, in a linear pooling fashion. The interpretation of the results in the proposed approach is done in such way that it displays the epistemic uncertainties that are inherent in the data and provides a better basis for the decision maker to make a decision based on his / her risk attitude. A comparison of the two methods is made based on the case study.

Department dc:contributor.department

System Design and Management Program.

Grantor dc:publisher

Massachusetts Institute of Technology

Year dc:date.issued

2006

Author dc:creator

• Lekkakos, Spyridon-Damianos

Advisor dc:contributor.advisor

• George E. Apostolakis.

• System Design and Management Program.

Statement dc:rights

• M.I.T. theses are protected by copyright. They may be viewed from this source for any purpose, but reproduction or distribution in any format is prohibited without written permission. See provided URL for inquiries about permission.

Licence dc:rights.uri

• http://dspace.mit.edu/handle/1721.1/7582

Language dc:language.iso

eng

Handle dc:identifier.uri

http://hdl.handle.net/1721.1/35113

OAI identifier oai:identifier

oai:dspace.mit.edu:1721.1/35113