A parametric analysis of heuristics for the vehicle routing problem with side-constraints

Abstract

The behavior of the parameters of 10 heuristics for the vehicle routing problem (VRP) with side-constraints is analysed. Statistical analysis techniques are used to determine the significant effect of the heuristic parameters on the solution value. All analyses are performed on a huge test set containing 420 problems with specific geographical characteristics and including a variety of vehicle-related, customer-related and time-related constraints. A classification of the heuristic parameters is presented as a result of the analyses. The findings can be considered as a basis for the conception of automated systems capable of providing support in selecting an appropriate heuristic and its parameter value combinations for a VRP. (C) 2002 Elsevier Science B.V. All rights reserved.