[Pycon] [new paper] "Chetan Chauhan" - Using PuLP for Optimizing Train Scheduling for Indian Railways

[info a pycon.it]

Title: Using PuLP for Optimizing Train Scheduling for Indian Railways
Duration: 60 (includes Q&A)
Q&A Session: 15
Language: en
Type: Talk

Abstract: ndian Railways is the 4th largest train network in the world in terms of size and largest in terms of number of passengers it carries. On a daily basis almost 20,000 passengers trains and about 10,000 freight trains run on its 65,000 KM long network. As is with any network this complex, it faces issues with scheduling of trains. There are multiple constraints like speed restrictions, train prioritization, track maintenance, limits on traffic on one section and of course limitations around infrastructure & personnel. All of these combined cause suboptimal utilization of the network and results in delays for existing scheduled trains. All of this results in loss of revenue for the Indian Railways along with indirect impacts on the economy resulting from delayed train operations.
In order to improve the current situation, Indian Railways approached some companies for analytically driven solutions around this problem specifically for the freight trains. They simplified the problem into a POC (Proof-Of-Concept) solution with an objective to schedule more freight trains without affecting existing (Scheduled) passenger trains
The solution which we will present is broken down into two parts. One part focusses on a solution presented for a Double Line Track from New Delhi to Kanpur (about 30 stations). Solution includes the algorithm and a visual train diagram to help interpret the solution. The second part focusses on some even more complicated, i.e. a Single Line Track where trains run in both directions. The approach is similar with additional safety constraints

Tags: [u'Railways', u'Transportation Analytics', u'Rail Traffic Optimization', u'Research', u'PuLP', u'Operations Research', u'Python', u'logistics', u'optimization', u'Operations', u'Rail Traffic O', u'Linear Programming']