Bayesian analysis for parameter estimation for use in PSA - a case study

One of the approaches existing for parameter estimation is Bayesian method. In this methodology, based on prior plant experience or industry experience prior distribution is assigned to the parameter to be estimated. Based on the evidence during the period of observation, the analyst´s prior belief about the parameter is updated using Baye´s theorem. In this paper general methodology for carrying out the Bayesian updation has been discussed. Both conjugate prior and non conjugate prior have been considered in the analysis. Kalmogorov-Smirnov hypothesis test has been performed for checking the goodness of fit of the distributions. A case study has been discussed.