Developing a Fuzzy Rule Based Cognitive Map for Total System Safety Assessment

Total System Performance Assessment, TSPA, for radioactive waste disposal is a multi and interdisciplinary task that is characterized by complex interactions between parameters and processes; lack of data; and ignorance regarding natural processes and conditions. The vagueness in the determination of ranges of values of parameters and identification of interacting processes pose further difficulties to the analysts with regard to the establishment of the relations between processes and parameters. More specifically the vagueness makes uncertainty propagation and sensitivity analysis challenging to analyze. To cope with these difficulties experts often use simplifications and linguistic terms to express their state of knowledge about a certain situation. For example, experts use terms such as “low pH”, “very unlikely”, etc to describe their perception about natural processes or conditions. In this work we propose the use of Fuzzy Cognitive Maps, FCM, for representation of interrelation between processes and parameters as well as to promote a better understanding of the system performance. Fuzzy cognitive maps are suited for the case where the causal relations are not clearly defined and, therefore, can not be represented by crisp values. In other words, instead of representing the quality of the interactions by crisp values, they are assigned degrees of truth. For example, we can assign values to the effect of one process on another such that (+) 1 corresponds to positive, (−) 1 to negative and 0 to neutral effects respectively. In this case the effect of a process A, on a process, B, can be depicted as function of the membership to the fuzzy set “causal effect” of the cause process to the target one. One of the main advantages of this methodology would be that it allows one to aggregate the linguistic expressions as descriptions of processes. For example, a process can be known to have a “very strong” positive effect on another one, or using fuzzy sets terminology the effect is “around (+) 1” with degree of membership μ = 0.9. As another example, a “moderate” negative effect can be represented as “around (−) 1” with degree of membership μ = 0.6 to the set “(−) 1”. Such a methodology can be an important tool for enhancing transparency in the TSPA process by allowing discussions between experts from different fields of research, for example by adding new “what if” analysis, and, therefore, for confidence building.