The position is meant to conduct research on multi-level vehicle routing problems. The work, which involves the implementation and testing of heuristic optimization strategies, will be carried out in close collaboration with Kenneth Sörensen and Peter Goos. The candidate is expected to enroll the university’s Ph.D. program and to obtain a Ph.D. at theend of the project. A short project description can be found below.
Short project description
The main aim of this project is to develop efficient solution methods for several reallife multi-level vehicle routing problems. In multi-level vehicle routing problems, decisions are taken on different levels. For
example:
• In multi-depot vehicle routing problems, decisions need to be taken on which customer to service from which depot, besides the typical routing decisions (determining the sequence of the customers in the routes and the assignment of customers to vehicles).
• In school bus routing problems, decisions to be taken include the assignment of students to bus stops, the selection of bus stops to use, as well as the typical routing decisions.
• In location-routing problems, the algorithm decides the optimal location of distribution facilities, as well as the typical routing decisions.
A second research topic of this project is the search for the most appropriate type of solution method for multi-level routing problems: iterative or integrated. We will examine specifically whether integrated optimization yields better results than sequential or iterative optimization, and what the main influencing factors are for the differences in performance.
A third important research topic that we intend to study is the interaction between different neighborhoods that operate on different levels of a multi-level problem. This will lead to a better understanding of the working of these neighborhoods and their interaction, and make the design of more powerful metaheuristics possible.
Finally, as a fourth research topic we intend to study the integration of exact methods into the developed heuristics. More and more, the study of so-called matheuristics, heuristics integrating mathematical programming methods, are being viewed as an interesting research avenue. Progress in the development of general integer programming solvers (such ILOG’s Cplex) during the last decade has been impressive, and has become such that using these solvers for real-life (sub)problems has become an attractive option.
In fact, multi-level problems offer an interesting playground for this type of methods, as some of the sub-problems can potentially be solved using exact methods, whereas others can be solved using neighborhood search heuristics.
We offer a four-year position (with annual evaluation), a net monthly grant of about 1900 euro, and a stimulating working environment in a lively cosmopolitan city. For more information, please contact Kenneth Sörensen with email address: kenneth.sorensen@ua.ac.be
