CRITICALITY MULTI-MODELLING AND SIMULATION OF SPARE PARTS INVENTORY CONTROL
ABSTRACT
As the bar for service excellence keeps rising, especially in the request of shorter lead times, higher service levels, lower costs and better customer service support, the conventional models of spare parts inventory control are increasingly becoming inadequate. Therefore, to tackle this challenge, in this study, three novel models of spare parts inventory control have been formulated, developed and packaged into a multi-model and multi-purpose engineering computer software, called U-SPIC. Model 1 used mathematical analysis to integrate 7 spare parts inventory policies together. Model 2 integrated the same inventory policies of Model 1, using stochastic simulation while Model 3 expanded Model 2 by considering bulk demand and supply using stochastic simulation. Chi-square goodness of fit inference statistical technique was employed in the preliminary design to check the reasonableness of using Poisson distribution for the demands and it gave 86% success. Composite stepwise two dimensional graphical representations of the models were formulated, which captured the stochastic demands and stepwise state transitions. The inverse transform algebraic method was applied for the generation of random numbers while next event method was used for the time advancement of the simulation clock. Traces and structured walk through were utilized for debugging the stochastic simulation models. Batch mean method was used in determining the confidence intervals of the simulation models, with 105 days run time and 100 replications each. The developed models results were validated with the case study (ANAMMCO) software package called IDIS which uses standard (r;Q) inventory policy. Beyond that, the models results were compared via an extensive simulation approach. 19 sensitivity parameters were varied in the study, where at each instance of variation, the behaviour of the fill rate of demands, as well as backorders (i.e. with regard to its average number, mean response time and maximum queue length) were analysed. On the average a saving of 18.51% demands in comparison with the conventional models was found, which indeed will result in huge cost savings in absolute terms. Beyond that, the insights from these models will increase the overall efficiency of spare parts inventory control.