Experience from designing transport scheduling algorithms Raymond Kwan School of Computing, University of Leeds R.S.Kwan @ leeds.ac.uk Open Issues in Grid Scheduling Workshop, Oct 21-22, 03 1 Outline o Public transport scheduling o Optimisation issues o Discussion 2 Public transport service Depot Operations & management The Public Routes Vehicle & Driver Timetables Operations Fares Transport Operator Payroll Planning & Scheduling 3 Planning and scheduling o Minimise operating costs o Operator: one optimisation problem, all decisions are variables o Solution designer: Sequential tasks Some decisions are fixed by earlier tasks Some decisions are left open for later tasks 4 Planning and scheduling tasks Service and Timetable Planning Vehicle Scheduling Crew Scheduling Crew Rostering 5 Research & Development at Leeds o Span over 40 years (22 years myself) o Algorithmic approaches - hueristics - integer linear programming - rule-based/knowledge-based - evolutionary algorithms - tabu search - constraint – based methods - ant colony o Numerous users in the UK bus and train industries 6 Parties involved in UK train timetabling Strategic Rail Authority Train Operating Companies Office of the Rail Regulator Health and Safety Executive Track Operator UK Train Timetables 7 Train timetables generation o Three key types of decision variable Departure times Scheduled runtimes Resource options at a station 8 Hard Constraints o Headway: time gap between trains on the same track o Junction Margins: time gap between trains at a track crossing point o No train collision! - On a track - At a platform 9 Soft constraints o (TOCs) Commercial Objectives Preferred departure/arrival times Clockface times Passenger connections Even service Efficient train units schedule 10 Bus Vehicle Scheduling o Selection and sequencing of trips to be covered by each bus o Each link may incur idling or deadrun time o Minimise fleet size, idling time, deadrun time o Other objectives: e.g. preferred block size, route mixing 11 Bus Vehicle Scheduling - FIFO, FILO Arrivals Departures FIFO for regular steady service FILO for end of peak 12 Driver Scheduling - Vehicle work to be covered Piece of work Vehicle 38 0600 0742 0935 1110 1304 G S H H S Time Location ( Relief opportunity ) 13 2-spell driver shift example sign on at depot Vehicle 1 meal break sign off at depot Vehicle 2 Vehicle 3 14 More example potential shifts Vehicle 1 Vehicle 2 Vehicle 3 15 Some characteristics of vehicle and driver scheduling o Jobs to be scheduled have precise starting and ending clock times o Scheduling involves trying to get subsets of jobs to fit within their timings to be collectively served by a resource (vehicle or driver) o Not the type of problem where jobs are queued to be served by a designated resource 16 Driver Rostering o To compile work packages for drivers e.g. A one-week rota Mon S46 Tue S46 Wed S46 Thu S07 Fri S14 0512 - 1357 0512 - 1357 1350 - 1815 0512 - 1357 1201 - 1846 Sat REST Sun REST o Rules on weekly rotas o Drivers may take the rotas in rotation o Optimise fairness across the packages subject to rules and standby requirements 17 Multi-objectives – what is optimality? o Operators do not always try equally hard to achieve optimal operational efficiency Union rules Service reliability Problem at hand is not on the “critical path” 18 Global optimisation? o Automatic global optimisation is obviously impractical o Combining two successive tasks for optimisation are sometimes desirable, e.g. Hong Kong: fixed size fleet, fixed peak time requirements, schedule buses & maximise offpeak service Sao Paolo: driver and vehicle tied schedules First (UK bus): “ferry bus” problems 19 Better optimisation through intelligent integration of the scheduling tasks o o Sometimes superior results could be simply obtained where powerful optimisation algorithms fail A more favourable scheduling condition could be achieved from the preceding scheduling task E.g. driver forced to take a break after a short work spell – swap in the vehicle schedule to lengthen the work spell Needs good vision from the human scheduler – rule-based expert system to integrate the scheduling tasks? 20 Scheduling for different service types o Different types of service may pose different levels of difficulty for scheduling (different algorithmic approaches?) Urban commuting: high frequency, many stops Sub-urban and rural: lower frequency, fewer stops Inter-city and provincial: long distance, few stops Some problems have to consider route and vehicle type compatibility 21 Discussion 22