A discrete-event simulation algorithm for the optimization of multi-scenario maintenance policies

Abstract

Maintenance policies that may incorporate alternative work packages with a different intention of implementation have an explicit effect on the availabilities and the operating costs of the production systems. These work packages can be classified under corrective and preventive actions, which offer run-to-failure and failure defeating approaches, respectively. Sub-classes of these actions show a variation by implementation interval, technology and materials used, and the point of application. Corrective maintenance, preventive maintenance by inspections, and opportunistic maintenance are the most frequently observed application types in the production industries. Their financial and functional effectiveness in a maintenance policy may vary depending on the operational substantiality and complexity of the system on target. Therefore, the system operability needs to be ensured with a proper maintenance policy where the redundant contents in the policy should be removed to mitigate the production loss. On this basis, the current study introduces a discrete-event simulation algorithm that is capable of simultaneous evaluation and comparison of multiple maintenance scenarios for a production system. The optimization criterion in the algorithm is interchangeable that can be achieved in two ways as maximization of system availability and minimization of total maintenance cost, based on the available maintenance and cost datasets. For a clear illustration of the algorithm, two case studies with different optimization criteria were provided for multi- and single-system earthmoving operations. The systems were selected from the mining industry, where the high uncertainties in production and machine performance are observed. As well as production and cost outputs, the sensitivity of the failure profiles of subsystems and components to the introduced maintenance scenarios were also discussed in the case studies to reveal the vulnerability zones in the systems.