Recently, more attention has been directed towards improving and optimising maintenance in manufacturing systems using simulation. This paper aims to report the state of the art in simulation-based optimisation of maintenance by systematically classifying the published literature and outlining main trends in modelling and optimising maintenance systems. The authors investigate application areas and published real case studies as well as researched maintenance strategies and policies. Much of the research in this area is focusing on Preventive Maintenance and optimising Preventive Maintenance frequency that will lead to the minimum cost. Discrete Event Simulation was the most reported technique to model maintenance systems whereas modern optimisation methods such as Genetic Algorithms was the most reported optimisation method in the literature. On this basis, the paper identifies the current gaps and discusses future prospects. Further research can be done to develop a framework that guides the experimenting process with different maintenance strategies and policies. More real case studies can be conducted on multi-objective optimisation and Condition Based Maintenance especially in a production context.

Existing approaches for modelling maintenance rely on oversimplified assumptions which prevent them from reflecting the complexity found in industrial systems. In this paper, we propose a novel approach that enables the modelling of non-identical multi-unit systems without restrictive assumptions on the number of units or their maintenance characteristics. Modelling complex interactions between maintenance strategies and their effects on assets in the system is achieved by accessing event queues in Discrete Event Simulation (DES). The approach utilises the wide success DES has achieved in manufacturing by allowing integration with models that are closely related to maintenance such as production and spare parts systems. Additional advantages of using DES include rapid modelling and visual interactive simulation. The proposed approach is demonstrated in a simulation based optimisation study of a published case. The current research is one of the first to optimise maintenance strategies simultaneously with their parameters while considering production dynamics and spare parts management. The findings of this research provide insights for non-conflicting objectives in maintenance systems. In addition, the proposed approach can be used to facilitate the simulation and optimisation of industrial maintenance systems.

Investigating the optimum blend of maintenance strategies for a given manufacturing system is a continuing concern amongst maintenance academics and professionals. Recent evidence suggests that little research is conducted on the simulation optimisation of maintenance in industrial systems. This study was designed to make an important contribution to the field of simulation-based optimisation of maintenance by presenting two empirical case studies: a tyre re-treading factory and a petro-chemical plant. It is one of the first to optimise various maintenance strategies simultaneously with their parameters in industrial manufacturing systems while considering production dynamics. Stochastic Discrete Event Simulation models were developed and connected to a Multi-Objective Optimisation engine. Various maintenance strategies were investigated including Corrective Maintenance, Preventive Maintenance, Opportunistic Maintenance and Condition-Based Maintenance. The results of this research suggest that over-looking the optimisation of maintenance on the strategic level may lead to sub-optimal solutions. In addition, it appears that traditional trade-offs between maintenance cost and production throughput are not present in some maintenance systems. This is an interesting observation that requires further investigation and experimentation.

Maintenance and spare parts management are interrelated and the literature shows the significance of optimizing them jointly. Simulation is an efficient tool in modeling such a complex and stochastic problem. In this paper, we optimize preventive maintenance and spare provision policy under continuous review in a non-identical multi-component manufacturing system through a combined discrete event and continuous simulation model coupled with an optimization engine. The study shows that production dynamics and labor availability have a significant impact on maintenance performance. Optimization results of Simulated Annealing, Hill Climb and Random solutions are compared. The experiments show that Simulated annealing achieved the best results although the computation time was relatively high. Investigating multi-objective optimization might provide interesting results as well as more flexibility to the decision maker. INTRODUCTIONMaintenance plays a vital role in sustaining and improving asset availability, which in turn affects the total output production of the system. A large and growing body of literature has investigated the optimization of maintenance in manufacturing systems. Dekker (1996) conducted a comprehensive review of maintenance optimization models and applications. It is interesting to note that simulation optimization was not mentioned in his research, rather, the focus was on mathematical modeling only. As the complexity of maintenance systems has increased, the limitation of mathematical modeling became apparent as many maintenance policies are not analytically traceable (Garg and Deshmukh 2006, Nicolai andDekker 2008). In a recent review, Sharma and Yadava (2011) discussed the published maintenance optimization models. They observed that the use of simulation has been an emerging trend which changed the maintenance view. This is mainly because simulation allows experimentation and better understanding of complex systems (Sebastian 2006). Simulation based optimization (Rogers 2002) is an approach that consists of two main elements: a simulation model and an optimization engine. As shown in Figure 1, the optimization engine provides the simulation model with the decision variables. The simulation runs and provides the optimization engine with the objective function. Depending on the optimization algorithm the optimization engine will conduct an analysis and provide the simulation model with a new set of variables seeking to improve the objective function. This cycle will continue until a condition is met. For example a specific number of evaluations without improvement in the objective function. 1109 978-1-4799-3950-3/13/$31.00 ©2013 IEEE

The maintenance function in manufacturing has been gaining growing interest and significance. Simulation based optimisation has a high potential in supporting maintenance managers to make the right decisions in complex maintenance systems. Surveys in maintenance optimisation have repeatedly reported the need of a framework that provides adequate level of details to guide both academics and practitioners in optimising maintenance systems. The purpose of the current study is to address this gap by developing a novel framework that supports decision making for maintenance in manufacturing systems. The framework is developed by synthesising research attempts to optimise maintenance by simulation, examining existing maintenance optimisation frameworks and capturing framework requirements from review papers in the area as well as publications on future maintenance applications. As a result, the framework addresses current issues in maintenance such as complexity, multi-objective optimisation and uncertainty. The framework is represented by a standard flowchart to facilitate its use.