Construction of a Demand-Response Two-Phase Time-Dependent Logistics Distribution System

• Main Authors: I-Hsiang Lin, 林翊翔
• Other Authors: I-Hwa Chou
• Format: Others
• Language: zh-TW
• Published: 2004
• Online Access: http://ndltd.ncl.edu.tw/handle/08789254939596402920

碩士 === 國立臺灣大學 === 土木工程學研究所 === 92 === Logistics distribution systems and marketing are closely integrated nowadays. As a result, to satisfy customer needs, distribution suppliers classify their customers and provide specific services for each category of customers. This research presents a highly flexible two-phase distribution model, which allows delivery companies to not only reduce running costs but also boost customer satisfaction, by providing specific services to different categories of customers. The model is divided into customer grouping – phase one – and production distribution – phase two. In the first phase, customers are classified into groups based on customers’ attributes. In the second phase, products are delivered to demand points, i.e. customers as grouped in phase one. Customer grouping reinforces the specific attributes of customers and thus helps delivery suppliers to work out appropriate distribution strategies. Production distribution is then divided into two sub-questions: vehicle assignment, and routing. The objective functions of these two sub-questions include terms reflect different requests from supply and demand ends. The vehicle assignment sub-question indicates how to best serve the orders of customer groups, based on the delivery company’s strategy. The vehicle routing sub-question indicates the most effective routing path for each vehicle assignment, as defined in the vehicle assignment sub-question. Vehicle routing paths are determined based on the preferences of the delivery companies: control of total vehicle routing time, early arrival and delay time. As the traveling costs of the vehicle routing links are time-dependent, results obtained are particularly practical. To solve the vehicle routing sub-question, a dynamic programming method is presented, which reduces both the number of computing iterations and the solving time. Applied to delivery company S, this two-phase distribution model suggests distribution plans with lower costs than either the customer location grouping model or the delivery company’s original distribution plans. Terms of the objective function in the model accurately reflect requests from the demand and supply ends of the distribution operations. That is, if weights of the supply terms are raised, vehicle routing time and lead time decrease, while customer waiting time increases. On the other hand, if weights of the demand terms are raised, vehicle routing time and lead time increase, while customer waiting time decreases. Furthermore, if the cost per unit of vehicle operation time is raised, vehicle routing time decreases and customer waiting time increases. Conversely, if penalty costs for disobeying the customer time window are raised, vehicle routing time increases and customer waiting time decreases. In addition, if the fleet size is enlarged, the lead time decreases. Finally, if goods are delivered by batch, all costs go up. When planning delivery routes, distribution supplier should be aware of the conflicts among all these costs to control both relative cost and customer satisfaction.