Aggregate production for supply network - a case of electronics manufacturing services industry

• Main Authors: Huang, Chen-Hsiang, 黃貞翔
• Other Authors: Huang, Chin-Yin
• Format: Others
• Language: zh-TW
• Published: 2012
• Online Access: http://ndltd.ncl.edu.tw/handle/19839386218294232656

碩士 === 東海大學 === 工業工程與經營資訊學系 === 100 === Owing to the characteristics of low technique threshold, high labor demand and huge fluctuation of market demand, the low efficiency of human resources allocation for the Electronics Manufacturing Service (EMS) industry is often induced with changing economic climate. Previous research rarely addressed the policy of human resources, and mostly developed production planning for only a single period. This causes the weakness of a low reaction to the variation of the market. Thus, our study proposes an aggregate production planning model as a better solution for EMS. Under satisfying the following conditions: material supply constraints, raw material prices, capacity limitation of each plant and the lead time of production and transportation into consideration, the model proposed contains many issues regarding human resource issues which includes labor hiring, discharge, training and cross-department labor transfer. We also develop a method to solve the model in order to figure out the accurate HR demand of every production period and the needed amount of backup HR. Because the model proposed is the NP-Complete problem in this study, the distributed and parallel computing method is exploited. To speed up the convergence of the solving procedure, both the upper and lower bounds for each solution scheme are found. There are two parts of Design of Experiments (DOE) as well as results analysis: For each part of the experiments, the levels of demand are arranged in the order of “rush- off” or “off- rush” to simulate multi- periods order variation. First, a small scale model is built to clarify the accuracy of our model and the feasibility of our solving procedure (distributed and parallel computing). According to the results, both distributed and parallel computing and single-machine computing generate the same optimal solution, but the former takes only one-tenth of the time to obtain the solution compared to the latter. In the final part, we design a large scale model in order to illustrate the advantage and the feasibility of distributed and parallel computing. The results show that distributed and parallel computing can obtain the approximate optimal solution in 11 hours, while it takes more than 80 hours for single machine computing. As a conclusion of this research, aggregate production planning not only effectively responds to t high variation of demand for the EMS industry, but also decreases the production cost. In addition, compared to other methods in the literature, distributed and parallel computing can approximate the optimal HR and order allocation much faster.