Study On Milk-Run Routing Models And Optimization Algorithms For Parts Inbound Logistics In Automobile Manufacturing

This paper studies Milk-run routing models and optimization algorithms in parts pickup system for automobile manufacturers. Milk-run is a kind of optimized logistics network, which is characterized by multi-frequency, low-volume, and timing. Trucks leave from manufacturing plant or RDC on time, picking up parts at supplier and going to the next supplier along designed routes. Milk-run trucks return to RDC or plant after having visited all the suppliers. Such pickup mode can not only make up for shortcomings of traditional transportation, reducing transportation cost and providing a more efficient inventory control, but also reduce the number of containers and control quality of suppliers better. This is very important for the implement of JIT production in an automobile manufacturing plant which has a large number of suppliers.By analyzing conflicts between inventory cost and transportation cost while implementing JIT supply in an auto assembly plant, Milk-run pickup system is proposed and introduced. Combined with the application and research of Milk-run, the principle, advantages and disadvantages, operation methods are studied. Routing for a Milk-run system is a very important job, for the routing has a major impact on costs, but research is limited at present. An optimization-based approach is developed for routing a Milk-run supply system, which requires for high-frequency, small quantity deliveries. The mathematical model formulated is called common frequency routing problem (CFR), defining routes among suppliers and frequencies that they are run, under the restriction of maximum space for parts storage. Common frequency routing provides parts supply system with picking up large quantities of products in small batches while keeping a reasonable transportation cost. In contrast to the traditional routing problem, CFR introduces inventory constraints and equal-sized split loads, which are more appropriate for real problems.In solving the model, a tabu search heuristic has been developed, which is suitable for realistic-sized problems. The utility of the approach is demonstrated through a computational experiment. At the same time, computational experiments are designed to test the relationship between logistics costs and performance of JIT supply based on inventory level which is a key impact factor. Finally, a Milk-run routing system has been studied in detail and used as a case study for the model. Compared to solving the problem manually, this model gives a better solution.This paper aims to provide decision support and suggestions for real Milk-run operation in automobile manufactures. The mathematical models and algorithms this paper proposed can also be used to provide theory support for the vehicle routing in Milk-run scheduling.