ABSTRACT

In the field of nuclear safety, the value of many variables are tainted with uncertainty. This uncertainty can be due to a lack of knowledge, of experimental value or simply because a variable cannot be directly observed and must be evaluated by some mathematical models. Two common problems encountered in such situations are the following:

1. The difficulty to build synthetic representations of our knowledge of a variable;

2. The need to compare, analyse and synthesize results coming from different mathematical models modeling a common physical phenomenon.