This study investigates the effects of equipment failure variability on the performance of a maintenance float system. Existing maintenance float policy models assume the rates of equipment failure and repair to follow an exponential(β) distribution. This assumption implies a coefficient of variation of 1 (CV=1). In this paper, the gamma(α,β) and the Weibull(α,β) failure distributions are used to generate a wide range of values for the coefficient of variation. The effect of these values on the average equipment utilization, server utilization, and on the average waiting time to repair are assessed via experimental simulation with subsequent statistical analysis. The results indicate significant departures in the performance measures of the maintenance float system under both a decreasing failure rate (DFR) and an increasing failure rate (IFR). Thus, there is a need to generate new sets of models for these cases. A full factorial experimental design is followed and the simulation output used to generate metamodels for the gamma and Weibull distributions. The limits of these models are also discussed and it is shown how they can be used to reach optimum decisions. The model also showed a significant interaction between the coefficient of variation and the number of servers in determining average equipment utilization (EU). The number of standbys (F) was not found significant in determining EU.
