Study On Milkrun Planning And Vehicle Routing Problem

Milkrun is a new supply model in enterprise supply logistics, which has been popularly run in foreign countries and gained huge economical benefits. In fact, this unique model is only been implemented in inbound logistics of several large-sized auto manufacturing factories. Related theories analysis and practical implementation and operation on milkrun are very weak. Comparing with traditional supply models, milkrun has its own particular requirement in operation procedures and implement conditions. In this background, on the basis of learning from successful operation experience of milkrun in domestics and outsides, and site investigation in relevant logistics company, this paper gave insight research of milkrun.Firstly, this paper dwelled on research and elaboration of theoretical conceptions and operation processes. Secondely, to solve the problem of high operating cost in implementation of milkrun basing on zero store or low store, we constructed a new mathematical model which concerned the influence of collection frequency on stcok cost and transportation cost in together. We also gave two algorithms to deal with the model, which were heuristic approach and genetic alogtithm. For the first one, we gave improvements to batch collection and routes arrangement in order to reduce operating cost on the basis of initial routing plan with“zero store”. For the second one, we designed genetic algorithm and used its high global searching ability to solve the model. We also gave special treatments to chrosome encoding and crossover mutation.In the process of implementation of milkrun, the critical factor is vehicle routing arrangement, which influents the cost of transportation. In the next charter we constructed mathematical model of split delivery vehicle routing problem (SDVRP) for milkrun, in which the constraint that one point can only be served by one vehicle was eliminated, and developed tabu search algorithm to solve the model. In this paper, we designed the crucial factors on the basis of character of SDVRP, and referred to the concreted instances in relevant literatures to demonstrate high performance of the algorithm.In the end, we used two algorithms to solve a concreted instance of milkrun. The results given by two algorithms illustrated the rationality of our mathematical model, which also demonstrated that we should fully concern the relations between stcok cost and transportation cost in the operation of milkrun. Rational collection planning can efficiently reduce the operating cost.