Application of a fuzzy inference system for functional failure risk rank estimation: RBM of rotating equipment and instrumentation

• NORSOK standard Z-008 and risk based maintenance (RBM) planning.
• Functional failure risk (FFR) estimation.
• Risk matrix and risk based maintenance prioritization with fuzzy inference system.
• Minimizing suboptimal classification at the boundary of the input ranges.
• FFR estimation using quantitative data and, vague or imprecise information.

An increased focus on risk based maintenance (RBM) optimization has been prompted in the offshore oil and gas (O&G) production and process industry due to the currently existing regulations and guidelines on preventing the functional failure risk (FFR) of rotating equipment and instrumentation. The RBM optimization approach prioritizes functions based on the potential risk of a given functional failure. Then, the equipment and instrumentation connected to the different functions are assigned with appropriate maintenance routines according to the potential risk of corresponding functional failure. This manuscript focuses on mitigating the suboptimal prioritizations made in implementing petroleum safety authority (PSA) specified regulations and NORSOK standard Z008 specified guidelines. It is mandatory to follow these to prevent functional failures on offshore O&G production and process facilities (P&PFs) operational on the Norwegian Continental Shelf (NCS). This manuscript suggests a fuzzy inference system (FIS) to minimize the suboptimal prioritizations of functions in the FFR analysis using an illustrative tailor-made risk matrix. This risk matrix has been developed to be similar to currently existing tailor-made risk matrices that have derived from Z008 guidelines and which have been used for FFR in different P&PF owner organizations operational on the NCS. Membership functions and the rule base were developed utilizing Z008 guidelines, data, experience, intuition and intentions of maintenance engineers. A risk rank calculation has been performed using the suggested FIS. The results indicate how the proposed approach helps in simplifying and making a more reliable and uniform FFR estimation.