Multiobjective optimization in an unreliable failure‐prone manufacturing system

The purpose of this article is to consider a corrective and preventive maintenance model with a view to both minimizing cost and maximizing system availability.

The proposed experimental multiobjective approach combines a simulation model and a statistical method to determine the best system parameters. The desirability function is used to convert a multiresponse problem into a maximization problem with a single aggregate measure. The model examined is based on a m identical machines system subject to unpredictable breakdown and repair, and the maintenance strategy used is based on the existing block‐replacement policy, which consists in replacing components upon failure or preventively, at scheduled intervals (T). Spare part inventory management is based on the (S, Q) model, whereby an order is placed when the replacement stock level drops below a given safety threshold level (S). At that time, a replacement part quantity (Q) is ordered, and is received after a stochastic lead time (τ).

The proposed model jointly minimizes the overall maintenance cost and maximizes system availability using a multiobjective optimization desirability function.

The multiobjective model can be used in a real manufacturing environment to help business decision makers determine the best compromise system parameters and adjust them to obtain desired response variables (overall production cost and system availability).

The proposed model allows the simultaneous optimization of two response variables, and determines the best system parameter compromise between the system cost minimization and the system availability maximization.