Research On Inventory Routing Problem Of Short Sea Fuel Oil Distribution Under Demand Uncertainty

Ship oil distribution of short sea is an important support work for short sea shipping and on-water producing,fuel oil is saved in the onshore ports for the demand of ships of call,most fuel oil of small port is from the central depot in this area which is under the charge of the fuel oil company.During the operation,oil companies replenish ship fuel for ports and quays that they are responsible for at a certain frequency to meet the need in a period.Company of ship fuel distribution arranges vessels to replenish fuel for one or more ports or quays from oil base,which is also stored in the oil warehouse for the future use of the local ships.Because of the uncertainty of the demand,the amount of the fuel to be transferred can only be got from the experience of the daily operation,and just a range.For the fuel If stockout,repeated replenishment will be a problem which will cause tremendous shipping cost,and if overflow,the inventory cost will rise.So,the supply companies are supposed to make appropriate plans to make a balance and try to reduce the cost.As for the uncertain parameters,Robust optimization method is applied where the demands are constrained in a space of ellipse ball to find the optimal solution.During the distribution companies should deploy ships,decide the amount and design a route to fulfill the distribution task,what worthy a mention is that the route design is similar with a multi-allocation VRP problem.So,the problem in this paper can be summarized as a VRP problem under uncertainty,besides the Quadratic ellipsoidal constraints make it equal the SOCP(Second Order Cone Programming)problem.Comparing with the linear VRP problem,this problem is harder in terms of solution.To solve this problem a two-stage method was designed,where the original problem is cut into two problems,one is a pure SOCP about demand the other is a linear VRP,which substantially raised the solving rate.In term of the conservation of robust model,a parameter between 0?(?)2 was given to adjust the robustness of the model,which make the model more flexible with the orientation against risks.