考量整備時間與不確定性非等效平行機台之排程探討－以半導體產業封裝晶粒黏貼製程為例

Abstract

工廠在實務運作時，為求生產線產能擴充、產量大幅增加，常會添購新的機台，新機台的生產速度與所能製造的產品往往較原有機台來的更快、更多，但也致使訂單排程更加複雜，就此，本研究定位於探討非等效平行機台之排程問題。為使排程符合實際情況，本研究以國內一半導體公司封裝廠晶粒黏貼製程為實際案例，考量換線時所造成的整備時間與其他可能造成停機之不確定性因素，探討現有訂單排程之改善方案。由於半導體產業的訂單無法由單一機台生產，而需將其切割至其他機台，再由各機台同時加工，本研究以最小完工時間為目標函數，在整備時間、機台數量與訂單數量等資源限制條件下，建立一線性規劃模型，再考慮如換線整備時間是否相同與機台是否尚有未生產完成之訂單等不同情況，應用LINGO求解各情況下之最佳解。本研究發現經由模型求解後，機台將以其相對優勢的生產速度進行更有效率的排程，使整體完工時間下降21.1%，除此之外，藉由考量不確定性之因素，可以得知接近實際的平均完工時間與最長完工時間，作為該製程生產週期時間之依據。

To expand production-line capacity and increase factory output, a manufacturing company usually purchases new machines which possess higher production speed and manufacturing capacity than the original ones. It is also very often to distribute production to several machines in semiconductor industry since one order cannot be produced by one single machine. The above situations result in more complex scheduling problems. A compromised resolution is therefore needed. To that end, this research was carried out. Special attention was paid to a semiconductor company’s local assembly plant. In particular, the setup-time caused by changeover and the downtimes of machine used in die-mounting process were paid attention to. A mathematical programming model considering constraints of setup-time, machine quantity and volume of orders was constructed in which the objective function was selected to be the minimized completion time. Various possible conditions including different setup-times and unfinished production of orders were taken into consideration. LINGO was adopted to seek for the best solution. The result indicates that the completion time could be decreased by 21.6% when the advantage of production speed of the machine was taken. If uncertainties were considered, the near actual completion time and the longest completion time could be obtained for use in assessing the production cycle time.