Simulation-optimization of inventory control of multiple products on a single machine with sequence-dependent setup times

This work studies the problem of lot sizing and scheduling of multiple products on a single machine, with sto-chastic demand and sequence-dependent setup times, called Stochastic Economic Lot Scheduling Problem (SELSP). The present work differs from others in the literature by considering simple inventory control policies and using the simulation-optimization approach to calibrate their parameters. We consider two inventory control policies: (i) fixed cycling (First in Sequence -FIS) and (ii) dynamic scheduling based on inventory levels (Lowest Days of Supply -LDS), combined with an "order-up-to" lot sizing. The problem is solved using AnyLogic simulation software and the OptQuest search engine to minimize total inventory cost (ordering, holding and shortage costs). The experimental design included the following factors: number of items, coefficient of variation of demand, system workload, and degree of setup increment, allowing the comparison of the two inventory control policies in different scenarios. Experiments show that LDS outperforms FIS in all scenarios, achieving up to 4.6% cost savings for cases of more products, higher workload, and greater demand variance. The developed models proved to solve the problem, effectively generating reasonable solutions. Furthermore, as they are user-friendly, we believe they can be adapted, without great difficulties, to real-life scenarios of the process industry.