a tool for periodic timetabling (for suburban...

A tool for periodic timetabling(for suburban railways)

Soumya Dutta, Narayan Rangaraj, K.N. Singh and Madhu N. Belur

Dept of Electrical Engineering,Industrial Engineering and Operations Research

Indian Institute of Technology Bombay (IITB)

This talk at: http://www.ee.iitb.ac.in/%7Ebelur/railways/

16th Feb, 2017

Soumya, NR, KN, Madhu (EE and IE&OR, IIT Bombay) CST-railways meeting 1 / 15

Outline

Objectives of the tool: ‘IITB-Suburban-Service-Timetabler’Types of constraintsData formats: input and outputMethods/techniquesFurther plans

Soumya, NR, KN, Madhu (EE and IE&OR, IIT Bombay) CST-railways meeting 2 / 15

Objective

Timetabling: Herculean task when done manuallyVarious complex constraintsOnly few experts can modify manuallyCan only ‘tweak’ existing timetablePeriodic constraints: need of the ‘hour’

Soumya, NR, KN, Madhu (EE and IE&OR, IIT Bombay) CST-railways meeting 3 / 15

Periodic timetable

Desired: hourly timetableOnly minute-value to remember (for each service)Need not be exactly evenly spacedRepeats each hour: for example:from CST to Vashi, say 3 services per hour:

8:13, 8:40, 8:52,9:13, 9:40, 9:52,

10:13, 10:40, 10:52Aim: ‘quite’ well-spaced within the hour

Soumya, NR, KN, Madhu (EE and IE&OR, IIT Bombay) CST-railways meeting 4 / 15

Constraints

Hard constraints:Headway constraintsFrequency of serviceMinimum traversal timesTurn-around constraints

Soft constraints:Spacing between consecutive ‘similar’ services

Soumya, NR, KN, Madhu (EE and IE&OR, IIT Bombay) CST-railways meeting 5 / 15

Inputs to the tool

Inputs offline (for now):Different infrastructural parameters:

Stations, trackstypes of services:

Inputs Online: web-version:

Passenger demands, traversal timesturnaround times at terminals

Timetable downloadable from the website in a few hours (or minutes)

Soumya, NR, KN, Madhu (EE and IE&OR, IIT Bombay) CST-railways meeting 6 / 15

Tool output

A feasible periodic timetableGuaranteed to satisfy:

headway, turnaround, traversal constraintsspecified frequency

Incorporated into timetable generationOutput-format: One up-file, and one down-file

Soumya, NR, KN, Madhu (EE and IE&OR, IIT Bombay) CST-railways meeting 7 / 15

More output

Rake linkage chart:

Rake cycleslist of services to be performed by each rakeNumber of rakes also calculated

Soumya, NR, KN, Madhu (EE and IE&OR, IIT Bombay) CST-railways meeting 8 / 15

More output

A terminal occupancy chart:Occupancy of each major stationNumber of trains occupying terminal resources plotted time-wise

Future plan: A graphical way of viewing terminal occupancy

Soumya, NR, KN, Madhu (EE and IE&OR, IIT Bombay) CST-railways meeting 9 / 15

Planned improvements

Line planning can be incorporatedProper platform allocation at all stationsA graphical editing tool for viewing the network and changingany infrastructure

PRESENT PLAN

As of now inputs will be taken from the Central Railways officethrough a web-interface.The inputs asked will be frequency, turnaround time and traversaltimes.The outputs will be displayed on a web page within a day

Soumya, NR, KN, Madhu (EE and IE&OR, IIT Bombay) CST-railways meeting 10 / 15

Sample output timetable

Arrival and departure timings

Each column: arrival or departure at station

Each row: one serviceHour-value not required, minute-values45.5 ≡ 45 minutes, 30 seconds

Soumya, NR, KN, Madhu (EE and IE&OR, IIT Bombay) CST-railways meeting 11 / 15

Sample: rake linkage

Service-type

Servicenum-ber

Source-Dest. Dep-Arr. linkingservice

Soumya, NR, KN, Madhu (EE and IE&OR, IIT Bombay) CST-railways meeting 12 / 15

Sample: platform occupancy

Soumya, NR, KN, Madhu (EE and IE&OR, IIT Bombay) CST-railways meeting 13 / 15

Working time table (taken from Central Railway)

Soumya, NR, KN, Madhu (EE and IE&OR, IIT Bombay) CST-railways meeting 14 / 15

Method/techniques used

Mixed Integer Linear Programming (MILP) formulationDue to soft/hard (and integer) constraintsComputationally difficult to solveGurobi for solving MILPFor pre-post processing: Python and Bash scriptingFinal output: xls-like file (csv file)Web-interface for input/output

Credits for Gurobi/Python implementation: Shashank Dangayach andSoumya Dutta

Thank you

Soumya, NR, KN, Madhu (EE and IE&OR, IIT Bombay) CST-railways meeting 15 / 15

Method/techniques used

Mixed Integer Linear Programming (MILP) formulationDue to soft/hard (and integer) constraintsComputationally difficult to solveGurobi for solving MILPFor pre-post processing: Python and Bash scriptingFinal output: xls-like file (csv file)Web-interface for input/output

Credits for Gurobi/Python implementation: Shashank Dangayach andSoumya Dutta

Thank you

Soumya, NR, KN, Madhu (EE and IE&OR, IIT Bombay) CST-railways meeting 15 / 15

Method/techniques used

Mixed Integer Linear Programming (MILP) formulationDue to soft/hard (and integer) constraintsComputationally difficult to solveGurobi for solving MILPFor pre-post processing: Python and Bash scriptingFinal output: xls-like file (csv file)Web-interface for input/output

Credits for Gurobi/Python implementation: Shashank Dangayach andSoumya Dutta

Thank you

Soumya, NR, KN, Madhu (EE and IE&OR, IIT Bombay) CST-railways meeting 15 / 15