Use this URL to cite or link to this record in EThOS: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.575706
Title: Investigation into heuristic methods of solving time variant Vehicle Routing Problems
Author: Harwood, Kieran G.
Awarding Body: Cardiff University
Current Institution: Cardiff University
Date of Award: 2013
Availability of Full Text:
Access through EThOS:
Access through Institution:
Abstract:
Traditionally, Vehicle Routing Problems (VRPs) are modelled with fixed traversal times. The amount of time it takes to drive from one end of a road to the other is unchanged throughout the day. Nearly always, the reality of the situation that is being modelled is very different, with road speeds varying heavily, especially with “rush hour" traffic. Modelling VRPs with time varying congestion means that even slight changes early in a vehicle tour can have major knock-on effects that are hard to predict. Recalculating the total traversal time of vehicles whenever their tours are changed drastically increases metaheuristic calculation times compared to non-time varying models. In this thesis we use a simple technique of calculating the localised change and inferring the global effects resulting from neighbourhood moves. Only if the localised change suggests that the global result is satisfactory do we then calculate the actual global result. Inevitably using these estimates does not give as accurate results as always calculating the changes, but we aim to show that the loss of solution quality is overshadowed by the significant savings in calculation time. We present a series of experiments comparing simple metaheuristics with and without using estimates and show consistent savings in calculation time whenever estimates are used compared to when they are not. These savings shown to increase as the size of the problem (in terms of the number of customers) increases. In addition to synthetic problems, we also present a problem based on real world vehicle traversal times and show that our estimates prove just as accurate, if not more so, at retaining solution quality as the synthetic methods. Lastly, we briefly discuss further methods of solving VRPs that could also benefit from our work here.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID: uk.bl.ethos.575706  DOI: Not available
Keywords: QA75 Electronic computers. Computer science
Share: