Strategies to cope with disruptions in urban public transportation networks Evelien van der Hurk Department of Decision and Information Sciences Complexity in Public Transport:
AN INTRODUCTION From Rotterdam, The Netherlands
AN INTRODUCTION From Rotterdam, The Netherlands Collaboration with Netherlands Railways
AN INTRODUCTION From Rotterdam, The Netherlands Collaboration with Netherlands Railways Thesis focus on –analysing passenger flows/behavior –Disruption Management –Interaction between passenger and logistic system
AN INTRODUCTION From Rotterdam, The Netherlands Collaboration with Netherlands Railways Thesis focus on –analysing passenger flows/behavior –Disruption Management –Interaction between passenger and logistic system 3 months at MIT, Prof Larson, Prof Sussman, Prof Wilson
RESEARCH QUESTION Is it possible to use the interaction between passenger route choice, the operations, and operation's control to increase service level by dynamically changing the network structure?
AN EXAMPLE CLOSE TO HOME – MBTA NETWORK
LONGFELLOW BRIDGE CLOSURE - MBTA’S PLAN
PLANNING SHUTTLES
THE LINE PLANNING PROBLEM – EXAMPLE NETWORK Station Red Line Broadway Downtown crossing Kendall/MIT Back Bay Community College Orange Line
THE LINE PLANNING PROBLEM – EXAMPLE NETWORK Station Red Line Orange Line Entrance Exit Enter, exit and transfer arcs Choose line with operating frequency and capacity Broadway Downtown crossing Kendall/MIT Back Bay Community College
THE LINE PLANNING PROBLEM – EXAMPLE NETWORK Station Red Line Orange Line Entrance Exit Enter, exit and transfer arcs Broadway Downtown crossing Kendall/MIT Back Bay Community College shuttle 1 shuttle 2 Choose lines and shuttles with operating frequency and capacity
LINE PLANNING MODEL
SUMMARY Is it possible to use the interaction between passenger route choice, the operations, and operation's control to increase service level by dynamically changing the network structure? Planned Disruptions Network effects Both Passengers and Logistics Practical examples (but theoretical model) –MBTA – longfellow bridge –TfL – to be decided Outcome: plan for logistics & plan for detour of passengers
CASE STUDY OF LONGFELLOWBRIDGE
DISCUSSION
MOTIVATION – DEDUCTION OF PASSENGER’S ROUTE CHOICE Knowledge on passenger route choice provides Estimate demand for capacity Test assumptions on passenger behavior and route choice Hind-sight analysis of passenger service (delays) Forecasting of future denand and effects in network So far: Surveys and panel data to deduce route choice Models for route choice: maximum utilitym regret minimization,… Now: Automated Fare Collection (AFC) Systems generetae detailed data on journeys Question: Can we deduce route choice from the Automated Fare Collection Systems data?
PROBLEM OVERVIEW ROUTE DEDUCTION FROM AFC Which route (time, space, trains) did a passenger take? Station AStation B Platform iPlatform k ci co timeci co trains Time +Station Conductor check
DATA Smart card data –Origin station, destination station, start time, end time, card id Realized timetable –Departure time station, arrival time station, train number Conductor checks –Card id, time, train number General: 5 days Over 500,000 journeys, about 1/3 with conductor check full Dutch Railway network of Netherlands Railways trains Comparison between disrupted and non disrupted days
MODEL Generate Paths Based on Realized Timetable Link a route to a path: –Find the set of routes leading from O to D that fit within the time interval of check-in, check out –If multiple routes fit, select one based on: 1) First Departure (FD) 2) Last Arrival (LA) 3) Least Transfers (LT) 4) Selected Least Transfers Last Arrival (STA) Check accuracy of matching based on conductor checks: –does assigned route have train?
MODEL - SCHEMATIC
EXAMPLE Journey: FromToDepatureArrivalCard ID AB8:008:46xxyy
DepartureArrivalTransfersTrain Numbers 7:558: :028:432100,200,300 8:058: :058:431200,300 8:208: :328:571400,500 EXAMPLE – STEP 1 ROUTE GENERATION (PREPROCESSING) FromToDepatureArrivalCard ID AB8:008:46xxyy Journey: Preprocessing – Route generation. Results for A-B:
DepartureArrivalTransfersTrain Numbers 7:558: :028:432100,200,300 8:058: :058:431200,300 8:208: :328:571400,500 EXAMPLE – STEP 2 ROUTE SELECTION FromToDepatureArrivalCard ID AB8:008:46xxyy Journey: Select Routes within check-in and check-out
EXAMPLE – STEP 2 ROUTE SELECTION FromToDepatureArrivalCard ID RotterdamAmsterdamxxyy Journey: Select based on Decision rule. 4 scenarios for decision rules: FD: First Departure LA: Last Arrival LT: Least Transfers STA: Selected least Transfers last Arrival
EXAMPLE – STEP 2 ROUTE SELECTION FromToDepatureArrivalCard ID RotterdamAmsterdamxxyy Journey: Select based on Decision rule (tested 4 decision rules) DepartureArrivalTransfersTrain Numbers 7:558: :028:432100,200,300 8:058: :058:431200,300 8:208: :328:571400,500 FD LT + LA STA
EXAMPLE – STEP 3 VALIDATION FromToDepatureArrivalCard ID RotterdamAmsterdamxxyy Journey: Check selection with MCL data : STA is correct choice, Other decision rules or wrong (in example) FromToTimeTrainNumberCard ID CD8:20400xxyy DepartureArrivalTransfersTrain Numbers 8:028:432100,200,300 8:058: :208: FD LT + LA STA
RESULTS Results for 5 days with extended list of journeys, realized timetable Results for 1 day with different settings Extended List: using conductor checks to find addtional routes
CONCLUSIONS / FUTURE WORK Conclusions Method for linking routes up to an accuracy of over 85% Passengers do not travel only on shortest paths Increasing path side based on conductor checks improves linking Based on linking insight into behavior in disruptions can be obtained, e.g. change in arrival at platform when timetable changes Future work Include learning of routes based on historic conductor data Research individual choice rules instead of one global behavioral rule Formulate general rules for route choice of passengers
QUESTIONS? Questions? Suggestions? Thanks for your attention!
DIFFERENCE IN TRAVEL BEHAVIOR Compare in-vehicle travel time differences with departure-arrival travel time differences between normal days and disrupted days: