Post-processing techniques for the joint CEC/USNRC uncertainty analysis of accident consequence codes

A summary description of the Joint CEC/USNRC uncertainty analysis of accident consequence models for nuclear power plants is given. The problem of post processing is introduced as the problem of projecting a joint distribution over observable variables onto the parameter space of a mathematical model. A conditional sampling algorithm is presented which works in simple cases. For more complex cases, two approaches are described and results are compared.