A Study On The Service Network Design Problem With Warehouse Capacity Constraints And Transportation Outsourcing

This work considers the logistic service network design problem.For the set of location nodes in the logistics service network and for the set of edges related,we consider a discrete optimization problem that aims to minimize logistics costs.In this problem the customer's demand(single goods)and supply of goods come from any number of nodes in the network.At the same time,different types and quantities of vehicles are placed at each node.We strive to meet the needs of all our customers during the planned timeframe and generate a schedule of plans that includes the quantity,source,and destination of the goods,as well as the transportation vehicles.According to the literature review and most common scenarios in the real world,in this problem all the vehicles are also required eventually to return to their original location nodes,so that the problem meets the design balance in a time period.This paper expands the design of the traditional logistics service network as follows.First of all,this paper considers the feature that some nodes("warehouses")in the network have the function of storing goods.Moreover,such warehouse nodes have finite capacity.Secondly,we consider two types of restrictions on the weight of commodities can be carried by vehicles.First one is that different vehicles have different physical load limits,and the second one is maximum load capacity that different roads(bridges)can withstand.These two aspects of the extension constitute two new types of constraints in the optimization model.In order to make the problem always feasible in the case of new constraints,we also introduce outsourced transportation services that are not subject to load restrictions,but have higher costs.This paper analyzes the role of these new types of constraints in different situations.The logistics service network design problem with the above two types of new constraints is the NP-complete problem.In order to solve this problem,this paper proposes a new heuristic algorithm combining linear programming relaxation,column generation,tabu search,constraint programming,and variable fixing.The computational experiments show that for most instances the solution obtained by the heuristic algorithm is very close to the lower bound.By comparing with the commercial MIP solver CPLEX 12.6,we also observed that the heuristic algorithm provides a high-quality solution:it either provides better results after two hours of running,or finds the same result faster in comparing with commercial MIP solver.