以資料包絡分析法評選連續盤點存貨策略(s, Q)

Abstract

本研究考慮某公司使用連續盤點(s, Q)之存貨系統，其中該公司的每天需求量與訂單到貨前置天數時間根據歷史經驗得知皆服從Poisson分配。在允許該公司缺貨待補，且不考慮上游會有銷售損失的情境下，發展出一個程序去找出最適的存貨策略。首先我們設計各種存貨策略組合，藉由模擬其各種存貨策略組合的真實情況，得到五個指標：訂購次數、存貨件數、缺貨次數、每次發生缺貨時，缺貨件數與補足缺貨所需天數之乘積、出貨件數。我們發展一個新的模組能針對各個存貨策略做綜合性的績效評選，此模組以CCR產出導向模式加入虛擬權重限制以符合指標之特性與管理意涵。

We consider a company controlled by continuous review (s, Q) inventory model. The company experiences Poisson demand. The lead time of orders follow Poisson distributions. We assume unsatisfied orders are backordered in the company. Headstream lost sales is neglected. The purpose of this research is to develop a procedure to select the continuous review (s, Q) inventory model. Given the levels of s and Q, the combinations of them are the alternatives for selection. Then for each alternative, we used simulation to collect the values of the five indices: times of ordering, stock level, times of shortage, the product of backorder and lead time, ordering quantity. These five values reveal the property of the alternative in managerial tasks. The aim of this research is to assess those alternatives with the five indices. A data envelopment analysis (DEA) model is developed. The model is a CCR output-based measure with some side constraints for the virtual weights restrictions of the indices.