A risk-based maintenance planning in process industry using a bi-objective robust optimization model

Abstract

We have developed an innovative risk-based maintenance planning methodology using a bi-objective scenario-based robust optimization model. This approach determines robust, optimal maintenance intervals for process industries. Our methodology comprises two main phases: risk assessment and maintenance planning. In the initial phase, we identified critical items using a Bow-tie diagram, which is subsequently mapped into a Bayesian network to estimate the overall risk based on historical data and process knowledge. In the second phase, we developed a bi-objective scenario-based robust optimization model to Pareto-optimize both risk and cost under operational risks. This results in a robust maintenance plan capable of withstanding time, costs, and failure rate uncertainties inherent in process industries with considering decision-makers' attitudes to risk (risk-averse, risk-neutral, or hybrid attitude). The computational results demonstrate the significant impact of considering uncertainty of critical data, and robustness on the selected maintenance plan and system performance.