Robust scheduling of the vehicle routing problem with time windows
Summary
In this thesis we extend the Vehicle Routing Problem with Time Windows by adding time-dependent and stochastic travel times.
By allowing any probability distribution to represent our stochastic travel times at any point of time, we allow great flexibility in defining these travel times.
Feasibility of a route is defined using reliabilities; the probability that the vehicle will arrive on time at the customer.
This problem is solved using a column generation heuristic.
To solve the pricing problem, we define four methods; two mixed integer programs, a local search heuristic and a dynamic programming heuristic.
We will compare these methods to find out which performs best on our problem.
Taking the new feasibility criterion for routes into account, we will calculate the latest possible departure time from the depot for a given route.
This can be done using different methods, which we will explore.