Towards strategic development of maintenance and its effects on production performance : A hybrid simulation-based optimization framework

Abstract: Managing maintenance in manufacturing within an economical short-termism framework and taking the consequential long-term cost effects into account is hard. The increasing complexity of managing maintenance and its impact on the business results calls for more advanced methods to support long-term development through effective activities in the production system environment. This problem-based design science research has evolved into the novel concept of a hybrid simulation-based optimization (SBO) framework which integrates multi-objective optimization (MOO) with system dynamics (SD) and discrete-event simulation (DES) respectively. The objective is to support managers in their decision-making on the strategic and operational levels for prioritizing activities to develop maintenance and production performance.To exemplify the hybrid SBO framework this research presents an SD model for the study of the dynamic behaviors of maintenance performance and costs, which aims to illuminate insights for the support of the long-term strategic development of maintenance practices. The model promotes a system view of maintenance costs that includes the dynamic consequential costs as the combined result of several interacting maintenance levels throughout the constituent feedback structures. These levels range from the applied combination of maintenance methodologies to the resulting proactiveness in production, such as the ratio between planned and unplanned downtime, in continuous change based on the rate of improvements arising from root-cause analyses of breakdowns. The model creation and validation process have been supported by two large maintenance organizations operating in the Swedish automotive industry. Experimental results show that intended changes can have both short-term and longterm consequences, and that the system may show both obvious and hidden dynamic behavioral effects.The application of MOO distinguishes this work from previous research efforts that have mixed SD and DES. It presents a unique methodology to support more quantitative and objective-driven decision making in maintenance management, in which the outcome of an SD+MOO strategy selection process forms the basis for performance improvements on the operations level. This is achieved by framing the potential gains in operations in the DES+MOO study, as a result of the applied strategy in the SD model. All in all, this hybrid SBO framework allows pinpointing maintenance activities based on the analysis of the feedback behavior that generates less reactive load on the maintenance organization.