nr. 2 i den internationale ROADEF-konkurrence
IMADA's group of PhD students finishes second in the 2010 ROADEF Challenge. A large-scale energy management problem with varied constraints July 2010.
An IMADA team of four PhD students made it up to the second place in the category reserved to young scientists in the 2010 ROADEF Challenge.
The ROADEF Challenge is an international competition dedicated to industrial applications. The object of this year's edition was a decision-making problem in the energy sector: planning the maintenance and the production levels of power plants. Nineteen teams from eleven countries worldwide took up the challenge and submitted their algorithms and computer programs. Steffen Godskesen, Niels Kjeldsen, Rune Larsen and Thomas Sejr-Jensen from IMADA ranked second among their peers and seventh overall.
The competition represents an opportunity for the Department to strengthen its position in the field of operations research. The group working on practical applications of combinatorial optimization and effective algorithms has already experience in high-level analyses and has solved industrial problems in a number of areas, including transportation, work scheduling, production management and timetabling.
Organized by the Societe Francaise de Recherche Operationnelle et d'Aide la Decision, the ROADEF Challenge looks at the development of optimization tools for industrial problem solving. The teams are required to develop and submit, in the form of a computer program, a method to solve an industrial problem. These methods are then tested and compared using actual industrial data. The winners of the 2010 competition were announced at the recent EURO conference, held in Lisbon from July 11 to 14.
The problem to be solved this year was suggested by EDF (Electricite de France), the leading company in power generation in France. The varied range of EDF facilities mixes all forms of energy: thermal (nuclear, coal, fuel oil and gas), hydraulic and other renewable energies. The management of thermal power plants is particularly complex, especially for those that have to be repeatedly shut down for refuelling and maintenance, e.g. the nuclear ones. The scheduling of outages has to comply with various constraints, regarding safety, maintenance, logistics and plant operation while it must lead to production programs with minimum costs. The plan extends over at least one year and has to and take into account the forecasts of future electricity demands. EDF was not satisfied with the way they currently handle this decision-making situation and it was only logical to involve the main experts from the field.
The method developed by IMADA team combines, ingeniously, known techniques, such as constraint programming and heuristics. It attains effective solutions and further research will make it more robust against particularly uncommon data.
17.08.2010Tilbage til nyhedsoversigten