Capacitated production planning by parallel genetic algorithm for a multi-echelon and multi-site TFT-LCD panel manufacturing supply chain

Abstract

Capacitated production planning of thin-film transistor liquid crystal display (TFTLCD) manufacturing is challenging to optimize owing to its complex sequential operations of array, cell and module plants, which constitute a multi-echelon supply chain. To address this issue, this study proposes a capacitated production planning model for such multi-echelon and multi-site supply chain in the TFTLCD manufacturing industry. Major domain characteristics, such as simultaneous allocation for capacity and transportation, multiple simultaneous resources, glass substrate slicing economics, capacity transformation, and the trade-off between inventories and backorders, are incorporated to maximize the profit of supply chain. To deal the high problem complexity, a parallel programming-based genetic algorithm accompanied with a repair operator is developed to gain the result rapidly. Experiment results indicate that the proposed algorithm outperforms the other benchmark algorithms. The impact of demand uncertainty is also investigated to quantify the information value of the demand. The contribution of the present study lies on the extension of existing literature by proposing a generalized mathematical model for the capacitated production planning problem in the multi-echelon TFTLCD-PM SC, with consideration of key characteristics in such industry. Further, this study improved a regular GA by designing a repair operator and employing a parallel computation architecture to solve a large-scale problem in finite time.