Optimization Models And Algorithms For Multicommodity Logistics Network Design

The logistics network is the collection of related organizations and facilities inlogistics process. The efficiency of the enterprise's logistics systems is decided by itsconfiguration. Also its quality reflects the its ability of competition and integratedmanagement on a certain extent. So the logistics network, which is the main run formof the logistics operation, was get more and more attention. The logistics networkdesign problem(LNDP) was considered as a key problem in logistics system. TheChinese development of the enterprise's logistics is more weakness than it in foreign.So the study of the design and optimization of the logistics network is necessitous andimportant.In the past decades, the optimization design problem of the logistics network wasstudy widely, and obtained abundant outcomes. This dissertation will introduce thecharacteristics, classes and optimization models, algorithms of the LNDP at first. Andthe main content of this dissertation is study for some multi-commodity problems thatbased on the existent literatures.The multi-commodity logistics network design problem(MLDNP) in thedissertation, not only contained the facility location problem or supply chain designproblem, but also study the whole layout optimization problem in logistics nodes,which was been little researched before.Based the common characteristics of LDNP, the dissertation improved thestandard simulation annealing algorithm and proposed the new method ofneighborhood creation, and designed the framework of the combined simulationannealing algorithm(CSA) for the problem. We tested the algorithm framework byenough benchmark problems, and compared its results with other algorithms'. And wefound the CSA can compute the better optimal solutions, and its speed of convergenceis more quickly than other algorithms'. The CSA also has the all-purpose property,that when the constraints of the logistics network optimization model were changed,the CSA only changed a little to adapt the new model.The LNDP has been researched in-depth regarding it as the single-commodity,few levels facility location problem, and there are few literatures discussed the morelevels LNDP. But go with the development of global economic, the enterprise'smarket is more and more wide, and its directly effect is the enterprise's supply chain is more and more longer, the length of its logistics network has been elongated too. Sothe dissertation proposed the optimization model of the capacitated multi-commoditymulti-levels logistics network design problem in chapter 4 and used the CSA to solveand analysed the numerical example. The application denoted the model andalgorithm can convergence in short-time, and compared with the object functions ofthe initial solutions, the optimal solutions that obtained from the CSA relatively savecost 35.45％～81.96％, and the influence size of each kind of factor to the systemtotal expense was studied.Furthermore, the study of the LNDP at present, often disjoin the inventorydecision and the facility location decision and regards both as the unattachedproblems of each other. But both of the two problems are one side of the LNDP inpractice, and they are influenced by each other and impacted on each other. Weproposed the MLNDP which considering the inventory decision and stochasticdemands in chapter 5, and proposed a new optimization model which simultaneitydescribed the inventory decision and the facility location decision. We used the CSAto solve and analysed the numerical example too. The application denoted thealgorithm's convergence speed is quickly, and compared with the initial solutions, theobject cost of the optimal solutions relatively save cost 16.82％～31.72％, and weobtained some rules of this type problem.The layout problem inside logistics nodes is an important problem in the latterphase of the LNDP, but related research is little. The LPILN is that in order tominimize the total cost under the condition of the total usable area in logistics node islimited, the design in the logistics node is required to determine rationally the productallocation to the functional areas as well as the size of each functional area. Based onthe function of the areas, the freight flows in the logistics node were described aseight basic flows. A optimization model was presented and used the CSA to solve it.Application denoted the optimization method operated rapidly and the result wasrational, and the influence of the mainly variable in the model was studied, so itprovides scientific guidance for the design in the logistics node.