Optimisation for transportation & energy markets
Dr Tibor Illés, who also works at the department of Differential Equations, Budapest University of Technology and Economics, has been actively involved in a number of industry projects, developing practically efficient solution methods.
Optimisation models for railway freight transportation
Although freight and passenger railway transportation have many common features there's at least one major difference, there's no timetables for freight trains, as are produced for passenger trains. Instead there's timetables of planned freight trains with quite flexible rules of changing those in case of available capacities for railway transportations.
In collaboration with MÁV Trakció Zrt. (part of the MÁV Group - Hungarian State Railway Company), we've aimed to address the issues of planning of the train and duty assignments for a month. We've also developed a re-optimisation model for train assignment due to the fact that the changes in regular trains have similar patterns. We've developed single and multi-commodity network flow models, as well as integer linear programming models with single and multi-objective functions.
In in a project supported by the Hungarian Government (EITKIC_12-1-2012-0001), we wanted to find paths in the Budapest urban area for any pair of origin-destination, considering private, public, and pedestrian transportation modalities, with the objective of minimising the overall travel time, number of transfers, or walking time.
Some similar but limited applications are available. For example, Google Maps can determine the shortest path and some alternatives based on distance or expected time. An optimisation-based multimodal transportation application could have several different extensions and still many open questions remain. As is usually the case in multi-objective optimisation, the objectives are in conflict. Therefore, in general, there's no single optimal solution, but rather a set of non-dominated solutions, called Pareto optimal solutions, from which the user should select the most preferred one.
Day-ahead power markets with coupled regions
The coupling of day-ahead power markets is a cornerstone of the wider goal of integrating the European power market. By combining the implicit auctions of regional markets and the previously explicit auction of cross-border capacities, this system facilitates the optimal utilisation of resources. This induces increased trading, narrower price spreads between the regions, and an overall more stable supply and pricing.
As part of the GOP-1.1.1-11-2012-0583 project supported by the Hungarian Government, the goal of our research in collaboration with electrical engineers is extending the currently used models, and developing robust and efficient algorithms to solve these variants. In the first stage of this research we implemented different formulations of the currently used model, which is a mixed integer program with linear constraints and convex quadratic objective functions. We then tested these implementations on a real dataset using multiple solvers.