Selected Distribution Network Optimisation Problems and Solving Methods

The paper presents a set of methods and tools for solving several problems related to the broadly understood optimisation of a logistics network. The described results are the outcome of a research project, the results of which are currently being implemented in logistics companies. We define three basic problems that are part of the distribution process: (1) determining the routes and order of visited recipients (modified VRPTW problem in time-dependent stochastic networks), (2) determining the location and parameters of distribution points (branches, sorting plants) and warehouses and grouping the collection points (the so-called problem of districting or hub location problem), (3) determining the parameters of transport networks. For each problem, the solution algorithm is described, and selected research results are presented. A modified simulated annealing algorithm was developed to solve the modified VRPTW problem. We have extended the algorithm by including several new elements: heterogeneous fleet, modification of searching initial solution, different methods of cost calculations, time-dependent network of travel, parallelization, and others. The algorithms have been tested on Solomon and Gehring-Homberger benchmarks, as well as based on real data from courier companies. To solve the hub location problem, we proposed a k-mean based algorithm. A regression analysis and clustering approach was used to solve the problem of determining the parameters of transport networks.