Inventory segmentation and production planning for chemical manufacturing

Abstract

Developing a cyclical schedule for producing multiple items on a single processor under stochastic demand that minimizes total setup and inventory holding costs is an important problem. This problem is faced in many continuous production environments such as chemical or petrochemical production, where the costs of switching production from one item over to another are very high. Due to high changeover costs, each item is produced in large batches and its demand is fulfilled from the finished goods inventory. Holding this inventory incurs inventory carrying costs. Therefore, good production planning policies are required to determine when and how much of each item should be produced so that the total cost of setup changeovers and holding inventory is minimized, while ensuring that sufficient inventory is available to meet customer demand which varies over time. In this thesis, I present the work done to develop production planning policies for a large chemical manufacturing company that operates in the environment described above. The problem described above is called the Economic Lot Scheduling Problem (ELSP) and is known to be NP-hard. So, optimal solutions are hard to find and one has to rely on heuristic procedures to find good solutions.

(cont.) In this thesis, I first present four fundamental inventory planning models relevant to the ELSP and discuss research works that specifically address the ELSP. I then describe the characteristics of the production and the planning processes at the chemical manufacturer where this work was carried out and present a heuristic procedure to solve the ELSP. This is followed by a demonstration of how the procedure can be applied at the manufacturing company and presentation of the results of a simulation experiment conducted to test the effectiveness of the solution. Finally, I will discuss two important issues related to the implementation of the solution at the company.