University of Twente Student Theses


Optimizing transportation in the network of food banks in the region of Twente-Salland based on the Vehicle Routing Problem.

Schotman, K (2022) Optimizing transportation in the network of food banks in the region of Twente-Salland based on the Vehicle Routing Problem.

[img] PDF
Abstract:This research is conducted at the food bank in Almelo. The food bank is a non-profit organization, to provide people with little financial means food aid. This means that they aim to form food packages that contain different products, to fulfill the nutritional needs a human has. They do this by collaborating with companies, the township and volunteers. Like the food bank in Almelo, there are another 170 food banks in the Netherlands doing the same work. Currently, a sixth of the annual budget of the food bank in Almelo is spend on transport of products between their location and the Regional Distribution Centre (RDC) in Deventer. This RDC supplies food to another ten food banks in the region of Twente-Salland. Each of them travels to the RDC multiple times a week to pick up food, and occasionally to deliver food in case of a donation larger than their needs. This means that half of the trips they make are empty. Due to the large influence this has on the annual budget of the food banks, it is important that a more efficient method of transportation is found. This transportation problem of the food banks can be formulated as the Vehicle Routing Problem (VRP), which can extensively be found in literature. Specifically, a multi-commodity VRP with time windows and heterogeneous fleet of vehicles forms the basis of the transportation problem faced by the food banks. However, the biggest difference between the VRP and the problem the food banks are facing, is the location of the vehicles. In VRP literature, the vehicles are located at the depot (comparable to the RDC) and routed to visit the customers (the food banks) in the most efficient manner. Here the customers (food banks) are the ones with the vehicles, and visiting the depot (RDC). Furthermore, not all products can be delivered with the use of the same vehicle due to food safety regulations. The goal of this research is to optimize the transport between the RDC and the food banks. In doing so, a Mixed Integer Linear Program (MILP) has been developed taking the several constraints into consideration. Unfortunately, a MILP is not able to solve large instances in an acceptable time. Therefore a metaheuristic, the Variable Neighborhood Search (VNS) algorithm to solve this problem is developed. Unlike the MILP, the VNS is able to take the supply towards the RDC into consideration, but it is also less strict on multiple usages of the same vehicle. Both the MILP and VNS have been fine tuned with the use of six test data instances. For the MILP, it was concluded that a maximum running time of 600 seconds results in decent solutions while have an acceptable running time. The VNS is consisting of an initialization, shaking phase and local search phase, for which the parameters had to be determined. Thus, experiments for the parameter settings were conducted and it was concluded to use a randomized initialization, have an extensive shaking phase leading to solutions further away, and ordered the operators used for the local search.
Item Type:Essay (Master)
Voedselbanken Nederland, Enschede
Faculty:BMS: Behavioural, Management and Social Sciences
Subject:01 general works
Programme:Industrial Engineering and Management MSc (60029)
Link to this item:
Export this item as:BibTeX
HTML Citation
Reference Manager


Repository Staff Only: item control page