Workload Balancing for Production Planning With Lot Streaming and Multilevel BOM

摘要

This article addresses a real-world tactical production planning problem in which a series of real-world constraints need to be considered, such as no backorder, products with multilevel bills of material (BOMs), and lot streaming production. Under the premise of no backorder, the objective of this plan is to make the workload as balanced as possible throughout the planning horizon. An integer quadratic programming model is first proposed to formulate this problem. Then, based on the analysis of the optimal solution for a common BOM structure, this problem is reformulated to a simplified problem where only one item in BOM needs to be considered. Some optimality properties are further derived to help solve this problem. An enhanced variable neighborhood search algorithm is developed to solve this problem, and a lower bound is put forward to measure the performance of the algorithm. Experimental results show that this algorithm can obtain high-quality solutions in a short time. Note to Practitioners-Effective and intelligent decisions about production planning could have a significant influence on the performance of the manufacturer. This article is motivated by a tactical production planning problem that we encounter in a factory producing discrete equipment. A production plan needs to be made with the consideration of various realistic characteristics in this factory. Without delaying the orders, the objective of the plan is to balance the workload throughout the planning horizon. Currently, the planners make the production plan manually based on their experience in Microsoft Excel. Due to the complexity of this problem and the increasing number of orders, it is difficult and burdensome for planners to make a high-quality production plan manually. In this article, a mathematical model is proposed to formulate this problem, and some theoretical properties are derived to help solve this problem. An enhanced variable neighborhood search-based algorithm is proposed to solve large-scale problems, which could provide a more detailed plan than the current plan. This study could ease the planner from burdensome planning work, improve the quality of the production plan significantly, and be adapted to solve planning problems with similar production characteristics and objectives. In the future, more production characteristics will be considered to meet the diversified demands of production planning.