Robust Vehicle Routing Solutions to Manage Time Windows in the Case of Uncertain Travel Times

Abstract

The vehicle routing problem requires mostly meta-heuristics to find approximate solutions. Congestion makes it hard for planners to find good routes because travel times are uncertain. The problem is handled by building scenarios with a range of possible travel times to represent the uncertainty. The main goal is to find a robust solution, which performs “well” in bad scenarios. The experiments show that robust results are obtained in a computationally reasonable time, which means that the concept can be used by practitioners. Most applications are commercial, but also social applications exist. An earthquake or a flood might lead to road disruptions. Higher traffic delays appear due to lower speed on flooded roads or due to time spent on finding alternative routes. Routing like evacuation of wounded or delivery of food is hardly studied. The work is inspired by the 2011 flooding in Thailand.