及時化(JIT)政策曾被成功地應用在製造產業，並在降低存貨水準與快速反應顧客需求方面獲得顯著的成效。事實上，JIT觀念亦可應用在供應鏈體系中，以時間為規劃基礎，精準地將採購、生產與運送活動緊密連結一起，達成同步化物流的運籌系統。因此，JIT供應鏈的建立將是國內企業邁向全球化市場重要的關鍵，其中整合批量政策在供應鏈管理中是一個重要的研究課題。供應鏈管理者必須規劃出具有協調機制的採購、生產與運送批量決策，使得整體供應鏈總聯合成本最小化。本論文提出四種JIT供應鏈整合批量模式，旨在有系統地探討不同JIT供應鏈結構的整合批量決策問題，作為管理者規劃最佳批量政策之參考。

本研究首先針對單一買方與單一賣方系統發展一種兩階層序列JIT供應鏈批量模式，用來協助買賣雙方共同擬定採購、運送與生產批量決策。本模式將一些強烈性假設予以釋放，例如：生產批量與採購批量呈倍數關係，並考慮有限計畫期與次數政策。這些作法使得模式更具合理性，而適切地應用在JIT環境。此外，為了發展一般化模式，運用複製方法將兩階層序列模式延伸至多階層序列供應鏈批量模式討論，並且考慮不確定運送時間對於JIT供應鏈的影響。多階層序列供應鏈批量模式同時採用緩衝時間與緊急借調政策以降低運送延誤與缺貨風險。

本研究進一步分析單一賣方與多名買方之生產配送體系。考慮單一賣方生產多項產品供應給多名買方，進而提出多物項兩階層樹狀供應鏈批量模式。本模式主要在解決經濟採購、生產、運送與排程問題，並同時決定計畫期內賣方的最佳共同週期時間以及多名買方的最佳採購與運送次數政策。另外，在多階層樹狀供應鏈方面，本研究建構一種多階層樹狀供應鏈批量模式，用以探討製造商、批發商與零售商在電子商務與JIT供應鏈的環境下進行資訊溝通與業務整合，使得易腐壞商品得以在高新鮮度狀況下，迅速及時供應給市場顧客。因此，本模式將可規劃易腐壞商品供應鏈成員的最佳整合採購、生產與運送批量決策。

本研究考慮有限計畫期，並將批量決策轉換成次數決策進行規劃，以符合實務營運需要。然而，最佳化次數決策乃相當於求解非線性整數規劃問題。本研究分別提出各模式特性，並發展有效率的分枝界限演算法來搜尋最佳次數政策。

The just-in-time (JIT) policy has been successfully implemented by manufacturing industries to eliminate unnecessary inventories and to obtain efficient customer responses. Recently, the JIT concept has focused on improving supply chain management to integrate the purchasing, manufacturing, and delivery activities. The problem of making integrated lot-size decisions is one of the most important research problems for supply chain management, since logistics managers must determine the coordinative lot-size policy that effectively links retailers, distributors, and manufactures at a minimum total joint cost. Consequently, the purpose of this dissertation is to systematically study the integrated lot-size decisions for a JIT supply chain. This study also provides four JIT supply chain models with various structures to help logistics managers in making the integrated lot-size decisions.

To begin with, this study addresses the buyer-vendor coordination problem in which a single buyer and a single vendor jointly make mutually beneficial lot-size decisions, including those related to JIT purchasing requirements, economic production lot sizes, and multiple small-lot deliveries. Therefore, an integrated lot-size model for a two-echelon JIT supply chain is developed. Some strong assumptions associated with that the vendor’s production lot size is an integer multiple of the buyer’s purchase lot size are relaxed and a finite time horizon is considered, so the proposed model is a more reasonable model that can be more practically applied in the JIT environment. To develop a generalized model, the two-echelon serial supply chain model is extended to the multi-echelon serial supply chain model by using a duplication methodology. In this model, since the problem of delivery uncertainty is involved, a time buffer policy and an emergency borrowing policy are applied simultaneously to deal effectively with random delivery lead times and to protect a JIT system against shortages of goods.

Furthermore, this study presents an integrated multi-item lot-size model of a two-echelon JIT supply chain in which one vendor produces several different items on a single facility for a number of buyers and the vendor and buyers exchange information to jointly make mutually beneficial decisions over a finite planning horizon. The model aims to find the optimal frequency policy that simultaneously determines the optimal numbers of purchase orders, production runs, and deliveries for the economic purchasing, production, and delivery scheduling problem. In multi-echelon arborescent supply chain, this study analyzes the problem that companies efficiently utilize collaborative logistics activities and the coordinating lot-size policy to achieve rapid flows of a perishable product through a JIT supply chain. Accordingly, an integrated lot-size model is constructed for the production-distribution-retail supply chain of a perishable product with fixed lifetime. An effective branch-and-bound algorithm is derived to determine the optimal frequency solution with integer decision variables, including the optimal numbers of purchasing orders, production runs, and deliveries during the finite planning horizon.

This study converts the lot-size policy into the frequency policy to ensure that the numbers of purchase orders, production runs, and deliveries during the finite planning horizon are integers. Optimizing the frequency policy is equivalent to solving a nonlinear integer programming (NIP) problem to which an exact solution can not easily be found. The branch-and-bound algorithms based on the model’s properties are developed to search for the optimal frequency solution.

none