1. Erasmus Center for Optimization in Public Transport 1 Shunting passenger train units: Practical planning aspects Ramon Lentink, Pieter-Jan Fioole, Dennis Huisman, Leo Kroon Erasmus University Rotterdam, The Netherlands http://go.to/ecopt AMORE, 3 rd Research Seminar October 1 – 4, 2002, Oegstgeest, The Netherlands

2. Erasmus Center for Optimization in Public Transport 2 Contents Introduction to the problem Solution approach Demonstration of the prototype Summary

3. Erasmus Center for Optimization in Public Transport 3 Examples of train units MAT64 - 2 MAT64 - 4 ICM3 DD-IRM

4. Erasmus Center for Optimization in Public Transport 4 Layout of a station

5. Erasmus Center for Optimization in Public Transport 5 Problem aspects Railway station and surroundings Arriving and departing trains Coupling and uncoupling of units Mixed arrivals and departures within planning period Different types of units Different types of shunt tracks

6. Erasmus Center for Optimization in Public Transport 6 Shunting overview Routing train units Scheduling personnel Parking train units Cleaning and maintenance SHUNTING

7. Erasmus Center for Optimization in Public Transport 7 Problem definition Given –arriving shunt units, departing shunt units, shunt tracks and station infrastructure Match the arriving units and the departing units and assign each unit to a shunt track between arrival and departure if necessary, such that: –costs are minimized, the capacity of the shunt tracks is never exceeded and there are no conflicts between arriving and departing shunt units and through trains

8. Erasmus Center for Optimization in Public Transport 8 Solution approach 4 step approach: 1) Match supply and demand of train units 2) Generate efficient routes 3) Assign units to shunt tracks 4) Determine all final routes

9. Erasmus Center for Optimization in Public Transport 9 What´s new? Increased support for planner More accurate implementation (especially w.r.t. free tracks) Prototype evolved (e.g. input generation, integration with personnel planning, different views on the results) Application to a different station

10. Erasmus Center for Optimization in Public Transport 10 Computational results I Railway station Zwolle in the Netherlands 8 am - 8 am next morning in March 2000 19 shunt tracks, 100 - 400 meters, 16 platforms 84 train units to be shunted during the day, 45 - 190 meters Runtime third step ±8 minutes 731 shunt movements

11. Erasmus Center for Optimization in Public Transport 11 Computational results II Railway station Enschede in the Netherlands 8 am - 8 am next morning in March 2000 13 shunt tracks, 50 - 685 meters, 4 platforms 28 train units to be shunted during the day, 50 - 190 meters Runtime third step ±5 seconds 199 shunt movements

12. Erasmus Center for Optimization in Public Transport 12 The DEMO …

13. Erasmus Center for Optimization in Public Transport 13 Summary Short description of shunting problem and implemented solution approach Application of solution approach to Zwolle and Enschede Demo of the prototype decision support system