Decision models for single-period and lateral stock-transfer inventory policies in disribution systems

Abstract

Firms are facing intense competition in today’s global business environment. They have to produce quality products at competitive prices with unpredictable demand. To be competitive, a firm has to quickly respond to demand fluctuations, competition, and new market opportunities. All these require an efficient supply chain and a well-designed distribution system. Information Technology is becoming a critical enabler of an efficient supply chain management system, which can help in sharing and utilizing the information in decision-making related to distribution systems.

We have studied the impact of electronic linkage between supply chain partners on the performance of distribution systems through case studies of two manufacturing firms. The two firms studied were AMUL, a fast moving consumer goods manufacturing company, and IPCL, an industrial product manufacturing company. These firms are different from each other in terms of supply chain structure and operations. These firms have implemented Web-applications for performing some business operations and are in the process to automate some more operations. These case studies have helped us to formulate two models for optimal replenishment decisions in distribution systems.

The first model is concerned with a manufacturing company/central warehouse serving multiple retailers dealing with multiple products. A fleet of vehicles is used between the central supply point and the retailers for daily replenishment to meet the random demand. A shortage or excess of replenishment quantity is a cost to the company. The objective is to determine the vehicle routing and the replenishment quantities to minimize the total under-stocking, over-stocking and the transportation cost. The replenishment quantities are constrained by storage at retailers, available supply quantities and vehicle routing norms.

A multi-product, multi-constraint single-period distribution (MMSD) model has been formulated to solve the problem. This problem being NP-hard, a multi-stage heuristic method has been developed to solve the model. Each stage is concerned with one of the following five sub-problems into which the original problem is decomposed.

1. Inventory Allocation: The objective is to minimize the sum of total expected cost of under stocking and over stocking for all products at all retail outlets based on only retail outlet storage capacity and supply constraints. 2. Expected Benefit: The objective is to determine the expected benefit of supplying a specific quantity over zero supply. 3. Maximum Vehicles: The objective is to determine an upper bound on the number of vehicles to be used in the system. 4. Minimum Vehicles: The objective is to determine the lower bound on the number of vehicles to be used in the system. 5. Vehicle-Inventory Allocation: The objective is to determine the number of vehicles to be used for delivery purposes, and vehicle – retailer – inventory allocation by a heuristic method. Various experiments were conducted to test the performance of the heuristic solutions of the sub-problems and the MMSD model. The results show that the heuristic solution of the MMSD model deviates less than 15% from the lower bound solution. Further, the heuristic for vehicle allocation gives better results than the existing bin packing heuristics. The above model was extended to incorporate the initial stocks for each product. The situations with and without initial stock information have been compared. The results show that the availability of information on actual stock has a potential to save approximately 10% on the total cost. The second model is concerned with a system that allows stock transfer between warehouses. A simulation model involving a manufacturing unit and two warehouses has been developed to analyze the stock transfer policies within the same echelon. The results of the experiments on this model have identified scenarios where the stock transfer can reduce the total cost, number of stock-out days and inventory costs by approximately 3-4 %. We have also studied the effect of information availability on the quality of the decisions in both models. Further, we have tried to obtain the value of information sharing between the partners, and through experimental results showed that the online information sharing has the potential to reduce the expected cost by approximately 5%, with respect to no information sharing, or information sharing with some delay. The major contributions of the research are as follows: • We have identified the current issues in the distribution systems of two manufacturing firms pertaining to the information sharing for better decisions. • We have formulated and solved through decomposition heuristic technique a multi-product, multi-constraint, single-period distribution model. The above model is obtained by extending the classical single-product newsboy problem to take into account multiple products, and multiple. • We have developed a hierarchical procedure to solve the multi-product, multi-constraint single-period inventory problem efficiently. • We have developed a heuristic to solve the bi-packing problem with additional constraints and applied it to vehicle allocation problem. • We have developed a simulation model to study the effect of stock-transfer policy, which extends the emergency stock-transfer models.