Pipeline Risk Assessment using a Fuzzy Systems Network

The Pipeline Risk Assessment using a Fuzzy Systems Network (PRAFSYN) presented in this paper was built using a set of Fuzzy Logic Systems (FLS) arranged in a network to behave as a Virtual Filter and to produce final risk evaluations. The FLS network arrangement contains seventeen FLSs that produce all the risk components detected through the knowledge engineering process, namely: Corrosion, Third Party, Construction, Design, Operation, Sabotage Hazard and Environmental Hazards. The Rule Base of each FLS contains the component of that knowledge describing the cause-effect relationships between the inputs and the corresponding risk contribution or risk factor. This knowledge is stored inside the FLS´s Rule Base by means of a linguistic model describing the expert´s approximate reasoning strategy to estimate the risks involved. The inputs to every one of the components are the expert identified variables affecting the corresponding risk factor. The risk contributions produced are finally combined within a last FLS containing al linguistic model describing the qualitative relationships between the contributing risk factors and the total risk estimation. All the knowledge was gathered from Houston pipeline experts and every FLS was designed and tested using the UNFUZZY [3] development tool, which allows proper adjustment of design parameters to obtain desired system behavior. C++ source code generated by UNFUZZY was finally assembled to produce PRAFSYN.