SIMULATION OF DECISION-MAKING POLICIES FOR THE STOCHASTIC INVENTORY ROUTING PROBLEM

This document addresses the replenishment of perishable products in a dairy industry network composed of one plant and multiple distribution centers, considering capacity constraints, quarantine times, transportation lead times, and control of item age in inventory. The objective is to evaluate the performance of different replenishment policies in a stochastic environment. For this purpose, a modular simulator was developed in Python, capable of representing perishable inventories under a FEFO logic, quarantine stock, in-transit shipments, and demand fulfillment within a trace-driven approach. Four decision policies were compared through 10,000 paired replications per policy. The evaluation was based on total cost, its components, and service level. The results indicated significant differences among the policies, with the replenishment policy standing out by presenting the lowest average total cost and the highest average service level in the scenario analyzed.