Multi-objective Logistics Network Integration Planning Under Stochastic Demand

As one of the pillar industry of the national economy, the logistics industry plays an important role in production and consumption category. During the last twenty years, Chinese logistics industry which is supported by national policy and promoted by the vigorous development of economy has experienced a rapid development stage. And the scale of logistics industry has significantly improved. However, the total cost of social logistics is still higer than that of those developed countries. In the last two years, the reduction of operating profit in the background of macroeconomic slowdown urges companies seeking reformation from the perspective of logistics.Facility location problem studied in this paper is an improtant part of enterprise logistics planning. Most of the existing discussions about facility location problem only take cost into consideration during optimization process. And also decision-making of facility location and inventory control are separated during logistics planning which will lead to a solution that is not systematic optimal.This paper starts from two different kinds of inventory control tactics, namely expectation order cycle constraint and order size constraint, to form two different mathematic models which both involve two objectives. And one of the objectives is the total logistics cost, and another is weighted distribution distance. Then an solving algorithm based on the non dominated sorting genetic algorithm (NSGAⅡ)is invented for the models. By studying and comparing the pareto frontier solutions of two models mentioned above we can illustrate the rationality and the necessity of the two selected objectives. Still we can find that inventory control level will affect the distribution range of the pareto frontier. Also we find that cost reduction due to inventory centralization is mainly embodied in the reduction of fixed cost of warehouse when the discussion is limited to commodities in this article.Then, this paper studies a location inventory model that involves adjustment of inventory control level during the planning period. The pareto optimal solution of this model is compared with the actual decision-making. Result verifies the validity of the mathematical model. Also we analyze the reasons of difference between the two solutions.Due to the limitations of time and ability, some deficiencies exist in the study which need further research. First of all, the mathematic models in this paper remains to be imperfect which require further improvement so as to moving closer to actual enterprise operation. Secondly, a new distribution way that demands from a same demand zone can be allocated to different warehouses can be took into consideration when we construct mathematic models. Finally, dynamic location research can be done as the inventory control level adjustment exists during the planning period.