A model for multi-plant coordination : implications for production planning

Firms in several discrete parts manufacturing industries, e.g., electronics equipment, computers, telecommunications equipment etc. operate in a multi-plant environment where products are processed successively at several plants. Prior studies have ignored the interaction between different plants in a multi-plant scenario. The objective of this dissertation is to study the impact of coordination on the cost performance of a two-plant firm.We propose a model that jointly determines production and inventory decisions so that the total cost of holding inventory and overtime, at the two plants is minimized. Our model captures the interaction between the two plants and is preferable to the uncoordinated or the sequential approach which ignores this interaction. We consider the case with limited capacity and explicitly model setup times. Strategies based on Lagrangian relaxation and Lagrangian decomposition methodologies are proposed to solve the model.Two main findings emerge from this research. First, our results indicate that coordination could lead to improved cost performance and enhanced profits for firms. Two parameters, the setup time to processing time ratio and the capacity utilization at the two plants played a significant role in determining the cost improvements. Managerial implications relating to implementation of the coordinated model are discussed. The second important finding of this research is that Lagrangian decomposition consistently outperforms Lagrangian relaxation in terms of achieving better deviation from the optimal solution, for this problem. A Linear Programming based technique for further enhancing the convergence between the upper and lower bounds is presented.In the quest for improved performance, multi-plant coordination represents an important strategy for firms. The contribution of the current research is in modelling some of the salient issues of this problem and exploring promising methodological directions