整合IC封裝與IC最終測試之生產規劃系統之構建

Abstract

隨著晶圓製造廠產出量的增加，IC封裝與IC最終測試的產能需求亦跟著水漲船高。各企業因而透過增設廠房以獲得更大的產能，來滿足訂單的需求，而形成了雙階多廠之生產型態，進而造成了各廠產能負荷分配的困難。因此本文考量IC封裝與IC最終測試之生產特性，發展一生產規劃系統，以規劃出各階段各廠所應生產之最適產品種類與數量。 本系統主要分為三個規劃階段，其中「IC封裝廠之生產規劃階段」是針對規劃週期內預定的需求量，考量各封裝廠內瓶頸工作站之產能水準及製程規格能力的限制，建構一整數規劃模式，將需求分配至各個封裝廠。而後依據各廠之產能水準與需求數量進行各廠之產線配置。並考量各產線瓶頸工作站之產能水準與必須生產之產品數量，估算投料時點、生產週期時間與各工單完成封裝作業的時間，以完成各工單完工日彙整表。 「IC最終測試廠之生產規劃階段」則利用「IC封裝廠之生產規劃階段」所規劃出來之完工日彙整表，考量各最終測試廠的製程規格能力與產能水準，建構一混合整數規劃模式，以求得各封裝廠分派至各最終測試廠之產品種類及數量，以平衡各機群的利用率，促使同產品在各廠之生產週期時間能更為一致。接著考量各廠之產能水準與負荷，估算各工單之投料點，並利用黃氏[20]所提出的最終測試廠之生產週期時間估算法求得生產週期時間，以推算各工單之可交貨日，並完成日產出總表。 「訂單管理階段」則將各訂單轉換成企業內部所使用的工單，並對照日產出總表的內容，使生管人員能夠快速的反應顧客，該臨時訂單的允收與否，最後再進行針對接收後的訂單進行指派工作。 實例驗證顯示，本文所提出之生產規劃模式可有效的估算出封裝階段之生產週期時間與產出量，且在面對產品組合不同與各廠產能供給不同時，本文所提出之工單移轉分派模組可使同產品別的生產週期時間變異降低。

Following the throughput increasing in IC foundries, the demand of IC packaging and final testing is also increasing. Each enterprise thus establishes new factories to satisfy the customers’ need. The enterprise of IC package & final testing thus faces a difficulty to deal with demand allocation that makes the production smoothly between the up-stream and down-stream factories. This study will focus on IC packaging and final testing processes and will propose a production planning system for planning the product type and quantity that each factory in each stage shall produce. The thesis includes three planning stages. In “the planning stage for IC packaging process”, according to the scheduled throughput in the planning horizon, we propose an integer programming model to allocate the product mix to each IC packaging factory with consideration to the capacity supply of bottleneck machines and capability constraints. Then we plan the product lines for each IC packaging factory with consideration to the capacity supply and product quantity assigned for each factory. Finally, we estimate the release time, cycle time and predict the due date of each job order at the packaging stage. In “the planning stage for final testing process”, after collecting the delivery date of each job order from each packaging factory, we propose a mixed integer programming model to derive the product type and quantity that each IC packaging factory has to transport to any specific final testing factory. Leveling the utilization rates of all testes is the major concern so as to make the cycle time of the same product type in each factory being consistent. Then, according to the capacity loading of each factory, we can estimate the release time of each job order and cycle time for each product type using Huang’s cycle time estimation algorithm for a final testing factory. Finally, we predict the due date of each job order at the final testing stage. In “the order management stage”, we transform a customer order into job orders in the production system. And according to the table of deliverable date, the production planner can quickly reply the customer whether we can accept the customer order or not. Finally, we collect all of the orders accepted and distribute the orders to each factory. Experimental studies reveal that the proposed planning system can efficiently estimate the cycle time and throughput for packaging process and can minimize the variation of cycle time given that the production target and capacity supply of each factory are different.