Transportation modeling and GIS

Transportation modeling and GIS

SHARE

Bharat Salhotra

This paper highlights the use of Geographical Information System for Investment Planning in the Rail transportation sector. 

Indian Railways is one of the largest transport systems in the world with a total of over 60,000 track kilometers spanning the length and breadth of India. About 14 million passengers – about the population of Australia- and over one million tons of freight traffic move over this network every day with a fleet of 6100 locomotives and half a million fleet cars. Indian Railways, is truly the lifeline of India and contributes over 1% of the national GDP, with its annual revenues aggregating to nearly 10 billion dollars.

One of the peculiarities of basic infrastructure – transport, power transmission and telecommunications, – is that it forms and operates as a part of a geographically spread out network industry. As a result, a change in infrastructure / operating practices in one part of the network has a significant impact over other parts of the network. This “ripple effect” translates into complexities for investment planning wherein benefits visualized through additional investment often do not result in achieving the desired objectives. In the absence of a strategic model that ‘sees’ the network as a whole, infrastructure-related investments tend to result in optimizing sub-systems and simultaneously rendering the entire network system sub-optimal. A realization of this ‘ripple effect’ and problems of subsystem optimization has lead to the application of operation research based transportation models.

Growth of transport infrastructure involves a lifecycle of planning, design, construction followed by operation and maintenance (see Adams et al 1992) and spatial data is central to each of these phases. An overwhelming amount of transportation data is location based and geographically referenced and in the light of this, GIS has emerged as the optimal technology to manage this data. There is also a growing realization among transportation experts that design and development of an integrated transport information system requires Geographical Information System. 

Research worldwide has attempted to enhance the power of GIS by adding /attaching ‘TM’ i.e. transportation modeling to the GIS so as to develop an integrated Geographical Information System for transportation. However, looking at the complexities in network modeling in general and peculiarities in transportation modeling in particular, it is not feasible to develop a universal GIS-T for meeting the requirements of all the users. Dedicated GIS-T software for meeting specific user requirements is however, a strong possibility. One such successful case study is the Indian Railways Long-Range Decision Support System (LRDSS)

The long-range Decision Support System (LRDSS) project was started in 1994 as a joint effort between World Bank and a team of Indian Railway Officers. The objective of the LRDSS was the development of a scientific and data based system for drawing up investment plans for track, rolling stock, and locomotives for the Indian Railways network. 

The LRDSS has been developed as a decision support tool for facilitating pre-feasibility level analysis of investment alternatives and for prioritization of projects based on their financial viability. In its present form, LRDSS is being used as a strategic planning tool for drawing up long term investment plans for Indian Railways. It provides an objective, information based, interdisciplinary platform for evaluating different investment alternatives aimed at improving the network wide capability of Indian Railways.

A unique feature of the LRDSS is the Avenue based User Interface, which insulates the planner from complex transportation models and gigabytes of network data as well as transport forecasts, and at the same time, provides him with a very powerful tool for long term planning. The GIS capability at the front end provides top-level decision-makers with a usable state of the art investment analysis tool. The GIS front end facilitates quick evaluation of specific technology, policy, and market related initiatives on a system-wide basis. In addition, the GIS is used for doing some preliminary infrastructure related analysis. 

LRDSS consists of six modules and each of these modules is envisaged to work in a stand-alone mode as well as an integrated whole. All the six modules reside at the back end while the front end and the user interface uses ArcView. A brief description of each module is given below:

Traffic Forecasting Module (TFM): This module is used to obtain the total transport demand between pairs of points for different commodities and for different key years. The inputs used by this module include production and consumption forecasts for ten major commodities for the entire country and for the next fifteen years. The data is based on extensive studies done by the Planning Commission and relevant ministries from time to time. In addition, and as a part of the data collection effort, Rail India Technical and Economic Service (RITES) were retained for conducting road survey to assess the road traffic levels that can be attracted to Rail. 

The TFM uses Linear Programming and “Furness” and factoring techniques for different commodities and generates a commodity wise Origin Destination (O-D) traffic forecast matrix. 

The module as well as databases reside at the back end of the User Interface and the front end Arc View is used to generate and view useful maps that bring out the pattern of traffic – both by Rail and Road- between pairs of points. These maps can be generated for different commodities, different pairs of points and different key years. The user interface also enables the user to generate “Origin Analysis” maps and “Destination Analysis” maps for different commodities/years. The sample map below brings out the outflow of traffic from Satna Area in Madhya Pradesh. While this sample map has been drawn for all the commodities, the user interface allows the user to select a specific commodity, a specific key year and a specific originating area.

Maps of the type shown above help the decision-maker to better appreciate the traffic movement potential as well as traffic leads for specific commodities. These outputs help to initiate corrective action and optimally deploy resources – wagons, locomotives so as to improve the market share. The quantum of traffic is gauged by the thickness of the red lines as shown above. 

In addition to its presentation power, the User Interface can be used for altering the assumptions used during traffic forecasting and for running complex LP based models at the back end for generation of the O-D traffic forecast by commodity. Market Analysis Module (MAM): The objective of this module is to obtain the sensitivity of the railway’s customer to different market parameters such as reliability and price. A Shipper Survey was conducted, as a part of the LRDSS, to find out the weights given to different service parameters by the customers. The results of the survey were used to develop a model to simulate the mode choice behavior of customer with respect to changes in transit time, reliability of service and availability of wagons. The GIS based User Interface provides the User with the facility to modify key parameters in a seamless manner and analyze their impact on mode share for different commodities and over different leads. 

The Market Analysis Module can be used to test the impact of service improvements on the mode share of rail. By running the Market Analysis Module for each key year, each commodity and a set of assumptions, a matrix of modal share percentages is arrived at for pairs of Origin-Destination points. This matrix is then used by TFM to determine the rail demand in total transport as well as corridor wise market share. 

Maps of the type shown above are extremely useful to assess the impact of strategic initiatives on the corridor wise market share of railways. This power of presentation enables the user to gauge how the market share changes with changes in various service parameters such as reliability, tariffs etc.

Facility Performance Module (FPM): This module uses a sophisticated simulation model called RAILS to simulate the train running on a section and evaluates impact of different locomotives and wagon, track conditions like gradients, signaling system etc. on the performance of a section. By using this model, the impact of various investment options can be analyzed at a micro level.

In the LRDSS, FPM is used to generate mathematical congestion functions that relate technology, operating policy and traffic levels with long-term-line-haul-variable costs. These mathematical functions are obtained for different types of trains running on different types of sections over the Indian Railways. Long-term line haul variable cost consists of line haul costs of traction, repairs and maintenance of rolling stock, signaling and track costs and other transportation costs that are congestion dependent. The cost function and the simulated capacity for a section are used by the Traffic Assignment Module to capture the ability of the network to carry traffic.

Simulation modeling is a complex and time consuming exercise and therefore this model forms a cold link with the GIS front end. However, the results of simulation studies reside at the back end so that the user can assess those through the GIS. The User Interface at the front of the FPM Module can be used to generate congestion graphs for different commodities carried over different sections and on different types of trains. Alternatively, the user can click on a specific section of the railways and obtain the cost curve depicting the carrying cost for different commodities at different traffic levels. 

In the map on the following page, the user selected Coal of type “A” as the commodity. The Cost of carrying Coal-A in three different types of sections- Flat (Type 6), Rolling Gradient (Type 10) and mountainous (Type 3) and at different levels of traffic is depicted in the graphical window. This graph has been generated by the user making a selection in the window at the lower left corner of the map.

The red lines in the main map depict movement of Coal-A over the Indian Railways network.

The user can also generate, by clicking on a section and requesting for all-commodity line haul cost, a series of congestion curves for different commodities moving over a single section using the GIS Interface.

Traffic Assignment Module (TAM): This module forms the core of the LRDSS and assigns traffic flows over the Railways network based on the capacity and congestion levels as generated from the FPM, (the supply side) and traffic forecasts available from the TFM (the demand side). 

The Traffic Assignment Module uses an Operation Research based non-linear programming algorithm called the Freight Network Equilibrium model. The objective function is the minimization of the carrier cost i.e. the Railways while assigning traffic over the entire network. As is the case with other modules, the TAM too resides in the back end. ArcView is used to manage the databases, edit spatial data (e.g. add a new line) as well as attribute data (e.g. double a railway line from single track to double track, or upgrade the signaling system) and display inputs and outputs of the LRDSS. The TAM Module takes care of 

  • Shortest Path Generator
  • Physical to logical networks Generator
  • Traffic Assignment 

One of the most difficult tasks encountered in the pre-TAM process is the preparation of the right set of paths that connect O-D pairs. Quite often, due to exigencies of operations, the preferred path used by Railways for moving traffic from an Origin to a Destination is not the shortest path. Under these conditions, the shortest paths generated by an algorithm such as “K-Short” do not replicate the actual reality and therefore, the paths have to be manually generated. A typical path may consist of nearly 100 to 200 sections and it is humanly impossible to code these without the assistance of a GIS. 

As a solution to this, a Path Editor was written in Avenue Script. The user simply clicks on a set of stations that must form a part of the preferred route. Once this is done, the preferred path between the Origin and the Destination is automatically appended to the path set generated by the K-Short algorithm. The Path Editor also has the facility to generate maps with the full set of paths that connect an origin-destination. 

The map above depicts the use of the Path Editor in which the user has selected six pairs of Origin -Destination Traffic Analysis Zones and wishes to display the route that connects these three Origin Destinations pairs. Depiction of the path sets in the manner described above enables the planner to better appreciate the currently preferred route apart from making him aware of the other alternative paths available. The Path Editor also helps the planner to understand the implications of routing traffic over preferred paths instead of the shortest paths and throws up interesting possibilities for cost reduction. For example, zooming in on the previous map helped focus on why the preferred path between 148D and 148P has to be longer than the shortest path. Analysis of the shortest path helped to emphasize the need to undertake gauge conversion on the highlighted part of the shortest route. (See map below)

The TAM, while minimizing the cost of carrying freight, assigns the commodity wise traffic over the entire Indian Railways Network. A typical TAM run takes over 8 to 10 hours of preparatory work followed by 4 to 5 hours of computer run and 2 to 3 hours of post proces sing. Broadly, a single run of TAM implies the following inputs:

  • A network comprising of about 1000 nodes and 2000 sections with a cost congestion function attach to each.
  • A Traffic Forecast matrix of the size: 800 originating Traffic Analysis Zones by 800 Destination Traffic Analysis Zones (TAZ) by 10 commodities
  • Approximately 15000 path sets detailing the potential paths that connect the various origin and destination TAZs.

The end result of the TAM run is a commodity wise flow over each section of the Indian Railways. Post processing of the TAM Output provides a list of bottleneck sections on different routes by key years. These bottlenecks can be assessed through the user interface and depicted over the entire network in a bottleneck map. The map above is an output of the TAM and it depicts the total pattern of flow of traffic in the year 2006-’07. The thickness of the lines is a measure of the density of traffic. Red part of the network highlights the sections, which are likely to have capacity problems.

Using this map as the base, the planner can they detail out an investment 

strategy for elimination of these bottlenecks. His strategy may involve one or all of the following:

  1. De-market carriage of some commodities 
  2. Enhance the network through a series of actions such as provision of additional lines, automatic signaling, higher speed locomotives
  3. Set up and strengthen alternative routes
  4. Initiate policy changes

The planner may want to generate a series of scenarios taking these four steps either in sequence or is a simultaneous manner. Looking at the lead times for various projects, he may opt for accelerated, normal or decelerated rate of investments.

Each of these scenarios may involve running of a series of different models in the FPM- e.g. RAILS Simulation Model-, in the TFM (e.g. Furness Matrix Generator) and MAM (e.g. Logistics Cost Model) and generation of a series of databases for input into the TAM. All these tasks can be undertaken through the Avenue User interface and TAM can be rerun to generate a set of scenario specific output tables. The impact of different scenarios or projects can then be studied in detail and annual investment plans made for the Railways as a whole.

Benefit Cost Analysis (BCA) Module: The Benefit Cost Analysis Module analyses the impact of alternative scenarios and projects on system wide costs and benefits. It relates investment costs and investment pattern, project lead times and changes in operating costs with incremental benefits in the form of additional or diverted traffic. The results thus obtained can form the basis of a financial or economic analysis of various investments. The input to the module is the unit investment costs for various types of works, the traffic flows as obtained from the TAM and the unit operating costs (i.e. the long-term line haul variable costs) as generated by the FPM. 

The Avenue User Interface facilitates transfer of TAM output data to Excel. It uses intermediate Visual FoxPro programs and database files to organize the large amount of TAM output data. Sensitivity Analysis as well as “With/Without” Analysis of a series of Investment Projects and their prioritization can be done in the BCAM Excel Sheets 

Financial Forecasting (FF) Module: The Financial Forecasting Module generates cash flows, expenses and revenues for key years based on the forecast traffic levels and investment options selected for implementation. It projects the investment requirements under different plan heads. Additionally, key year wise Contribution and Financial Statements as well as Sources and Uses of Funds over a 15- year financial plan can be generated by this module.

Conclusion
The LRDSS has thus emerged as a world class model using GIS and complex transportation models and data structures, all successfully bundled into a user-friendly decision support system. It is integrated and interdisciplinary investment planning tool for planning of complex transportation systems through a simple and easy to use Geographical Information System User Interface. Its decision support capability is one of the salient features that have been built in. 

LRDSS is being gradually institutionalized into the traditional planning process of the Indian Railway Board. Its use is likely to provide tremendous opportunities for cost reduction and network optimization within the organization. Use of GIS is likely to encourage its application in areas such as Asset Management, Property Management and Customer Response & Prospecting. The LRDSS Model can also be customized to other infrastructure areas such as Roads, Telecom and Power Distribution. 

(Mr. Bharat Salhotra is a Mechanical Engineer and an MBA from IIM, Calcutta. He is currently working as Joint Director in the Planning Directorate of the Indian Railway Board)

Database integrity is a question mark and a lot of work had to be done to ensure that there is no data loss while moving the data from one platform /database to another one.

Indian Railways Network Database: This contains various attributes of the railway network. Important stations and sections of the network have been encoded on a map using a Geographical Information System called Arcview. The network is corrected upto 31.3.97. A future network for 2001-02 has also been prepared on the basis of anticipated completion of works of new line, electrification and gauge conversion etc

Traffic Database: An Origin Destination flow table has been set up for different commodities. This data will be updated annually so that the latest information on annual flows between pairs of points is available by commodity. A commodity wise forecast table has been set up containing the total traffic demand between pairs of points, that is likely to materialize by the year 2001-’02.

Finance & Cost Database: A list of committed works alongwith their expected date of completion and the approximated cost has been entered in a database. Further, a Unit investment cost database for Civil, Mechanical, Electrical and Signaling works has been set up to enable the user to build up total project investment costs for different line capacity projects. A Long-Term Variable Cost database has been created to enable the user to assess costs per net ton kilometer at different congestion levels for different commodities. 

Facility Performance Database: We have aggregated division wise data relating to various aspects and features of operation. This data contains details of interchange points, type and number of trains running on important sections of that division, ruling gradients, commodities carried, type of stock used etc. In addition, detailed data of seventeen important links over Indian Railways such as Sonnagar Mughalsarai, Gomoh Gujhandi, Baroda Surat has been entered for Simulation Modeling. Output analysis related to congestion, failure analysis and impact of different technologies can be obtained for these sections. 

Conclusion
The LRDSS will be able to provide the users with an easy to use information-based tool that can be used for long term planning over the Indian Railways. A user-friendly interface has been developed so that the various types of analysis can be done without much difficulty. 

The basic problem has been to attach large transportation model system to the ArvView. The problem was circumvented by development of a Avenue based Interface. Need to have a more open and more powerful GIS Arcview to facilitate attachment of Transportation Models at the back.