|
|
Abstracts of the seminar papers
Short bio Planning of busses and gates at Amsterdam Airport Schiphol Abstract All aircraft visiting Amsterdam Airport Schiphol (AAS) need to be assigned to a gate. These gates can be divided into two categories, ordinary gates and remote stands. Aircraft assigned to remote stands must be served by one or more buses to transport the passengers. When looking at the gate assignment problem, one of the stages is the one-day-ahead planning during which a schedule for the next day is generated. The more robust this one-day-ahead schedule is, the less rescheduling will have to be done during the actual day when the real arrival and departure times deviate from the planned times. The bus problem consists of planning for each bus which aircraft it must serve taking into account the legal constraints for the drivers. Here we again aim at finding a robust solution for the one-day-ahead problem to avoid replanning during the day due to deviations from the planned arrival and departure times. For both of these problems, we present a solution approach based on column generation; for both problems we are able to find robust schedules in a matter of minutes.
Short bio Timetable 2007: why and how? Abstract The timetable of 2007 (starting at December 10, 2006) is by far the most significant timetable change since 1970. In the first part of the presentation, we will discuss the main reasons why NS developed a completely, new timetable from scratch. One of the reasons to change the timetable is to improve the robustness of the railway system, which is mostly measured by its punctuality, i.e. the percentage of trains driving on time. Therefore, in the second part of the presentation, we will look at the relation between the timetable and the robustness of the system. Simulation can be used to determine this relation, and therefore, it was frequently used during the construction of the new timetable. Finally, we will discuss some recently developed Operations Research models that can improve the punctuality by slightly modifying the timetable. These models were not used in the timetable of 2007, but will probably be used in the near future. GERARD KINDERVATER (KLM Royal Dutch Airlines) Address KLM Royal Dutch Airlines Department of Pricing and Revenue Management gerard.kindervater@klm.com Short bio Gerard Kindervater has been a research assistant at CWI in Amsterdam in the 1980's. In 1988 he accepted a position as assistant professor at Erasmus University in the Department of Computer Science of the Economic Faculty. Since 1999 he works for KLM Royal Dutch Airlines in the Department of Pricing and Revenue Management. Currently he is involved in the development of a joint revenue management system for Air France and KLM. Title In the airline passenger industry, carriers try to maximize the seat revenue through the overbooking of lights by anticipating on forthcoming cancellations and no-shows, and through the offering of different fares for the same journey thereby limiting the number of seats available to the lower yielding fares. For optimization roughly two models exist: (1) consider all flights independently of each other and treat connecting passengers as local passengers on each flight of their journey, and (2) build a network of connecting flights and evaluate passengers with respect to their complete journey from origin to destination. Whereas the flight based approach was used by all airlines in the past, nowadays it has been replaced by an origin-destination approach by all major network carriers. In this presentation we will present both models and show their pros and cons. We will also discuss the issue of buy-down, which is a major drawback for successfully using present revenue management systems over a longer period, and show a possible way to overcome its negative effect. MARIEKE DE KONING (Pointlogic, Rotterdam) Address Pointlogic P.O. Box 29147 3001 GC, Rotterdam koning@pointlogic.com http://www.pointlogic.com Short bio Marieke de Koning graduated in Econometrics in 2002, with a combined specialisation in Operations Research and Marketing. She has worked as an analytical consultant at Pointlogic since her graduation. Pointlogic is a consulting software company enabling smart decisions for its clients by delivering software based solutions. The company specializes in applying mathematical and econometric techniques in the areas of media, marketing, HRM, health care and security. Title This presentation will focus on a very different application of OR in public transportation, namely on the performance measurement of public transportation ads. Performance of media channels is measured in two different measures: contacts and reach. How many people see an advertisement, and even more important: how many different people see an advertisement? Both measures combined describe indirectly the value of a product. That is why these measures are so important for media owners and media buyers to know. Out of home media are extremely difficult to research. Unlike television, radio, press, etc. there is no editorial environment to provide a context. For these media a questioned individual might remember what they read, seen or listened to. Very few, if any, will have sufficient perception of the outdoor medium to give meaningful responses to question about posters on buses. So, the big question is: how can reach be measured for out of home media and more specifically: public transportation advertising? To answer this question we will briefly explain how reach can be measured for non-moving items like billboards. For this we combine a displacement research with a route choice model and the actual positions of billboards. Then we focus on the aspects for moving billboards on public transport: timing and schedules become an issue, as well as the direction of the respondent compared to the bus or tram. The final result is a database which is integrated in media planning software that can be used to measure the performance of public transportation advertising campaigns. LEO KROON (NS Reizigers, Utrecht; Erasmus University, Rotterdam) Address NS Reizigers Department of Logistics P.O.Box 2025 3500 HA Utrecht lkroon@rsm.nl Short bio A survey of OR models and techniques for the planning of Public Transportation Abstract In this presentation, an overview will be given of quantitative models for public transport systems. Relevant elements to be taken into account are the passengers, the infrastructure, the line system and the timetable, the vehicles and the crew. Traditionally, quantitative models for public transport focused mainly on supporting operational planning processes such as timetabling, rolling stock circulation and crew scheduling and rostering. However, currently also strategic planning and real-time operations control are recognized as fruitful areas for the application of quantitative models. Especially disruption management is an important topic within real-time operations control. RAMON LENTINK (ORTEC, Gouda) Address Department of Algorithmic Developments ORTEC bv Groningenweg 6k 2803 PV Gouda rlentink@ortec.nl Short bio Where do train units stay when they're off duty? Abstract Possibly, you never thought about this. However, off-duty train units pose an interesting planning problem for planners of railway operators. In general, the idle rolling stock is parked at a shunt yard, keeping the main railway infrastructure available for other train services. Shunt plans are created to coordinate the temporary parking of off-duty train units. Shunt planning is one of the last elements of the planning process of a passenger railway operator. Therefore, every change in a previous step of the planning process is likely to require changes in the shunt plans at one or more stations, a total of 130 out of 380 planners at NSR are currently involved in shunt planning. Advanced planning support for shunt planners enables an increase of efficiency of the shunt planners. Moreover, it provides opportunities for reducing the throughput time of the overall planning process. In turn, a reduced throughput time enables a later start of the planning process, which results in more accurate input data for the planning process and therefore better plans. In this presentation, we will introduce the most important processes that take place at a shunt yard during the night. Moreover, we will discuss the parking of train units at shunt yards in detail and describe a solution approach for this problem. The solution approach is based on the well-known technique column generation, enabling the explicit generation of only a fraction of all possible columns, while guaranteeing the quality of the solution. Computational tests on real-life data show that high-quality solutions are typically found within minutes of computation time. In addition, the developed algorithms are designed to interact with shunt planners. They provide a firm basis for an advanced planning system to support shunt planners in their everyday work. MICHIEL ODIJK (ORTEC, Gouda) Address Department of Algorithmic Developments ORTEC bv Groningenweg 6k 2803 PV Gouda modijk@ortec.nl Short bio In June 2001, Dutch Railways faced an internal conflict regarding a new crew assignment model. This led to a period of much turmoil, including nationwide strikes that lasted several days. The model assigned drivers and guards to train series that started or ended in their home base exclusively, yielding duties with little variation. To address the conflict, the board of Dutch Railways and its Works Council agreed that the Council could develop an alternative crew assignment model. This agreement was spectacular in many ways, as rarely could the Works Council of a large company directly influence such a crucial part of the business process. In October 2001, ORTEC and Basis & Beleid were asked to assist the Works Council with the development of an alternative crew assignment model. The project started with a thorough survey of the planning process at Dutch Railways. Furthermore, the team investigated the crew assignment model challenges by visiting every crew base and discussing the situation with over 600 active personnel. This process led to the development of five alternative crew assignment models by ORTEC / Basis & Beleid. In the two months thereafter, a team of six experts from ORTEC analysed these models in close cooperation with the Logistics department of Dutch Railways. After five interactive sessions, each with over 70 personnel, the final alternative was selected by the personnel and then adopted by the Works Council. On May 22, 2002, the project reached a milestone with the presentation of the alternative model to the board of Dutch Railways. The board accepted the alternative model on June 13 ending the period of turmoil. The alternative model is called "Sharing Pleasures & Burdens" (Dutch: delen lusten & lasten). The philosophy behind the model is that all personnel should experience an equal amount of pleasure and an equal amount of burden in their work. This innovative model sets boundaries on * the amount of time a driver or guard works on fast trains and on trains that are labeled as socially unsafe, * the number of different tracks and rolling stock types a certain crew base will be assigned, and * the variation in the duties In the presentation we will further set out this model, the road that lead to it, and the role Operations Research played in this process. NIELS VAN OORT (HTM Personenvervoer N.V., Den Haag; TU Delft) Address HTM Personenvervoer N.V. Afdeling Vervoersontwikkeling Postbus 28503 2502 KM Den Haag http://www.htm.net Short bio Niels van Oort (28) finished his Master study Civil Engineering at the Delft University of Technology in 2003. His thesis dealt with the relation between network planning of public transport and reliability of operations. After his study, he started working at the public transport company of The Hague, HTM. He works as a consultant/researcher at the department of research and development. He is involved in the RandstadRail project, the new light rail connection between the cities of The Hague and Zoetermeer. In this project his main focus is on the new operation control system: a system which shows the driver his punctuality and the dispatchers can monitor the position as well as the punctuality of all the vehicles. This system helps to achieve high reliability of RandstadRail. Besides this project Niels participates in different infrastructure projects and rider ship studies of HTM. At the beginning of 2006, Niels started a PhD study at the Delft University of Technology, section Transport and Planning. Niels is a member of the board of "Jonge Veranderaars", an organization of young rail professionals and he is a board member of "KIVI NIRIA, verkeerskunde en vervoerstechniek", The Dutch Engineers Association, department of transport and traffic. Title The moment of execution: real-life monitoring and control of public transport operations in major urban areas Abstract RandstadRail (RR) will be a new regional public transport system between The Hague, Zoetermeer and Rotterdam. RR will be a high level public transport system: short travel times and high frequencies. During peak hours the intensity on some trajectories will be about 24 vehicles an hour. RR will share some tracks with other tramlines as well. Dealing with this high frequencies and deviations, a system for controlling operations is necessary: because of offering travellers a high quality product, according to waiting times and chance of getting a seat, as well as making optimal use of the restricted capacity. To prevent congestion the operation of RR should be as much as possible according to the timetable. To reach a high level of reliability, HTM designed a controlling philosophy. This exists of three steps. The most important aspect is distribution in deviation of the timetable. First step is preventing this deviation to occur: the infrastructure will be as much as possible exclusive right of way and at intersections RR will get priority over the other traffic. RR will stop at every stop and will never leave before schedule time. Second step in the philosophy is dealing with the deviation by planning extra time in the schedule at stops, trajectories and terminals. Small deviations can get solved in this way. Final step to get vehicles back on schedule will be done by the traffic control: they have a total overview of all vehicles and their deviations. They can take measures like slowing down vehicles nearby a delayed vehicle to optimize headways. In case of big disturbances derouting and shortening of lines are possible. |
