INDIA RAILWAYS GIS-BASED DECISION-SUPPORT SYSTEM

This paper describes the implementation of Phase I of a Long Range Decision-Support System for the Indian Railways (IR) which has been under development during the last two years. The system incorporates GIS (ArcView) as a user interface, as means of storing and retrieving system inventory and facility management data, as a link to transportation models, and as an interface to a set of evaluation tools for investment and marketing decisions. This set of functions provides an user- oriented system that has greatly improved information available to IR managers over the previous system, which depended on manual, hard copy reports. The system added flexibility and analytic power as well as spatial data that was not previously available in a timely fashion to managers. The system has the potential to identify major costs savings in achieving rail line capacity expansion and contribute to the process of changing IR priorities to achieve more cost-effective investments and marketing strategies. The paper will describe data base and modelling issues as well as GIS and analysis issues.

INDIA RAILWAYS GIS-BASED DECISION-SUPPORT SYSTEM

Background

The new economic environment in India has given rise to a 
growth rate of GDP averaging 5.6 % per year and growth of 7 to 
9 % annually in surface transport demand.  Export growth is 
averaging 18-21%.  This growth is creating pressures on both 
the road and rail modes and it is anticipated that the railway 
will be called on to handle 125-150 million tons of additional 
freight (a 34-41 percent increase or average annual growth of 
more than 6%) and an even faster increase in passenger flows 
by the year 2000.  Therefore, identification and evaluation of the 
most cost-efficient means for achieving capacity expansion is a 
top priority for IR.

With this heavy pressure for growth and a shortage of 
investment funds, effective planning tools are needed by IR to 
provide a comprehensive basis for the screening and 
evaluation of proposed improvement projects and to provide 
better forecasts of traffic for each corridor, taking into account 
the competition from road haulage. Since the capacity 
expansion alternatives for Indian Railways lie in a 
complicated combination of investments, there is a need for 
relatively sophisticated but cost-effective methods to determine 
the relative priorities of system-wide investments. The IR also 
identified a need for access to more market information to help 
it compete for the most profitable business in the future freight 
market.

Brief Description of Indian Railways

The Indian Railway is a large system by international 
standards with 62,000 route kilometers and transportation 
revenues of Rs 758 billion (U. S. $21.6 billion) in 1994-1995.  It 
carried 365 million tons of freight in that year, an increase of 3 
% over the preceding year, which reflects the continuing 
growth of railway traffic over the last few years.  The railway is 
a unique mixture of long haul bulk freight operations and high 
volume inter-city passenger operations.  It is a continent-wide 
railway with average freight hauls of about 705 km compared 
to the European average freight hauls of 200-400 km.  It is both 
a short-haul suburban mass transit railway and a long-haul 
inter-city passenger railway.

Despite the growth in rail services and the successful 
introduction of modern container operations, there has been a 
relative decline of rail output to road output over the last 15 
years.  This reflects the specialization of IR in carrying low-
valued, long-haul bulk cargoes, such as coal and the very low 
unit revenues from inter-city passenger operations. Road 
competition for freight has also been very effective, although it 
has lessened in the last two years due to congested road 
conditions.  The planned construction of better intercity roads 
and expressways will increase the challenge for the Railway to 
become more competitive.

The Long Range Decision-Support System

In 1994, IR management decided to undertake the development 
of a decision-support system which would provide the tools to 
carry out evaluations of all the critical factors and help set 
investment priorities that had been missing in the past.  The 
IR called this a Long Range Decision Support System (LRDSS) 
and enlisted assistance from the World Bank for its design and 
implementation.  It was conducted as an institution-building 
exercise with a substantial training component, involving the 
creation of a well-trained multi-disciplinary LRDSS team 
supervised by a Steering Committee in the Railway Board. 

The primary objective of LRDSS Phase I is to develop a decision 
support system (DSS) for the management of the Indian 
Railways  that would allow it to evaluate decisions that affect 
railway capacity in a comprehensive, system-wide, multi-
modal context and to evaluate potentially profitable markets for 
railway service. The system must be capable of evaluating a 
complicated and inter-related set of investments  (e.g.,  gauge 
changing, improved signaling systems, lengthened sidings, 
urban bypasses, high horsepower locomotives, low tare weight 
high axle load freight wagons, and train operations policies 
designed to maximize the capacity of the track). 

The analysis objectives for the DSS are to: (a) forecast traffic 
flows in total and by major category for freight traffic and 
passenger traffic; (b) eliminate bottlenecks by taking cost-
effective measures to improve utilization of existing track and 
rolling stock assets; (c) evolve a least-cost operating strategy for 
movement of traffic between pairs of points served by more 
than one route; (d) determine priorities among already-
sanctioned transport capacity augmentation projects; (e) select 
new, high priority investments for increasing total system 
throughput within prevailing budget constraints; and (f) 
achieve a revenue maximization strategy within broad policies 
of transportation on Indian Railways.

To meet these objectives the decision-support system requires 
of a set of analytic tools and models integrated into a user-
friendly  system with access to a wide range of data.  The most 
appropriate and cost-effective framework for the DSS was 
found to be a combination of commercial simulation software, 
specially-designed system optimization software, specially-
designed database structure and a Geographic Information 
Systems (GIS) which provided both customized dialog with the 
users and the ability to use a mixture of map and table-based 
information to identify and evaluate the best alternatives. 

The DSS selected for Indian Railways was an advanced version 
of a similar DSS developed for China Railways in the early 
1990s (Cook, 1993).  The main differences from the earlier 
system, were a much closer integration of the GIS with the 
rest of the DSS (e.g. input screens for the investment options 
database, broader simulation capability, customized menus for 
evaluation and closer linkage of model inputs and outputs with 
GIS displays of the transportation network).  The structure of 
the DSS is shown in Figure 1.

Implementation of the DSS

The analytical tools for the DSS are structured around six key 
strategic modules, which are capable of analysis of 
investments over a 20-year planning period with detailed 
analysis for the first five years and for every fifth year 
thereafter.  The six modules are:

	a)	Traffic Forecasting Module: This module forecasts 
goods and passenger traffic demand between major 
origin-destination pairs for various commodities 
under different assumed demand scenarios;

	b)	Facility Performance Module:  This module 
estimates capacity, cost and transit time for existing 
and proposed, converted and new rail lines, yards, 
transhipment points and other congested facilities.  
For rail line costs and delay functions, it uses the 
results of a detailed rail line simulation model 
(RAILS);

	c)	Traffic Assignment Module: This module ssigns the 
forecasted traffic and compute financial costs for the 
major railway network under different assumed 
scenarios of investment and demand.  It includes a 
network-wide model based on non-linear 
programming;

	d)	Cost-Benefit Analysis Module:  This module provides 
an economic and financial cost-benefit analysis in 
summary form for each proposed investment 
alternative over a 20-year period under a selected 
demand scenario;

	e)	Financial Forecasting Module: This module 
translates the results of the above modules into a 
summary of the costs and revenues of IR for each 
major commodity group and for passengers.

	f)	Market Analysis Module:  This module stores the 
results of the shipper survey and analyzes 
information on cost and traffic relevant to shipper 
decisions on the choice of road or rail for goods 
movements.  It includes a basic mode choice model, 
calibrated from the shipper survey results.

These modules are all linked together with a GIS-based user 
interface, which provides graphics, dialog boxes, spatial 
analysis tools and other decision-support features (see Figures 
2 and 3 for examples).

As the reader can see, the LRDSS is a relatively ambitious 
undertaking, especially for an organization of the size of 
Indian Railways, with limited data processing facilities. The 
implementation of the system, therefore, went through some 
time-consuming data collection and model calibration steps.  
There were several significant issues that slowed down the 
implementation. The most important of these was the 
significant unanticipated data entry and data cleanup 
problems (traffic and railway line operations data) and the 
discovery of major data gaps (road traffic and shipper 
information) that needed to be filled by formal surveys.  

Secondary issues were software bugs and the linkage of rail 
network data between the GIS graphics and the model data 
bases.  It took approximately eighteen months to implement 
Phase I, including the first three modules of the DSS (with 
initial calibrations and four months of team training), but 
without the formal survey data and the broader road-rail 
aspects of the DSS which are needed for the remaining 
modules.  Also resources were diverted from the 
implementation of the three remaining modules to deal with 
the data cleanup problems in Phase I.  A second phase has 
been approved by IR to carry out the surveys to fill in the data 
gaps and complete and calibrate the remaining modules. 

Although the full objectives of the DSS were only partly 
achieved in Phase I, the foundation has been laid for a system 
that meets the strategic capacity planning needs of IR.  A 
substantial database has been built up, an initial set of 
hardware and software is now in place, both line capacity and 
strategic models have been preliminarily calibrated and basic 
forecasts and traffic assignment have been carried out for a 
key forecast year (2000).  Also a multi-disciplinary team has 
been created in the Railway Board to utilize the LRDSS for 
investment and policy analysis.

Initial Results of LRDSS

The initial results of the LRDSS Phase I are of four types: (a) 
provision of data access to planners and managers that did not 
exist before; (b) creation of  traffic forecasts by origin-
destination that did did not exist before; (c) analysis of system 
bottlenecks, both with existing train routing procedures and 
with alternative routings and (d) identification of alternative 
locations for system improvements that could be more cost-
effective than the previously identified alternatives.  The 
anticipated outputs of Phase II will identify and evaluate even 
more cost-effective options for IR management, and give 
access by managers to an even wider range of data that they 
could not access before the LRDSS.

The data on the IR rail system available to planners and 
managers in the Railway Board has been very limited before 
the LRDSS.  It has been primarily in the form of reports and 
with many hidden assumptions and biases.  The statistics 
available were of an aggregated nature that did not allow 
much analysis and more detailed data took a long time to 
acquire.  With the LRDSS much of the basic system data (track, 
traffic, facilities, etc.) is now available in both map and table 
form to anyone in the Board or on the support staff with access 
to a computer with LRDSS software and databases.  (It will be 
even more available through local area networks in the 
future.)  This is a major increase in information flow in the 
Railway.

Traffic data is now available by origin and destination to 
managers.  This data was previously stored on tapes but in an 
unreadable format and not consistent between the zonal 
railways, and therefore not available to anyone who wanted to 
use it.  The availablity of this data and the capability to display 
it in both map and table form is a significant tool for planners 
and managers.

The ability to identify system bottlenecks is the first step in 
identifying cost-effective solutions to system problems.  This 
ability of the LRDSS is also tied to scenario analysis or "what 
if" types of analyses, that combine the power of a systems 
model (based on detailed simulation of operations on selected 
links) with the power of a GIS to display the results in multiple 
views and tables.

Finally, the identification of potentially cost-effective 
alternatives, even without the cost-benefit framework to be 
provided in Phase II, is a major step forward for IR managers.  
This gives them the scope to investigate a specific range of 
options that is broader than the high-cost construction 
alternatives, that have been the focus of previous alternatives 
for system capacity improvements and that relates specifically 
to the forecast traffic flows that they will have to carry on 
different routes through the rail system.  The savings from 
identifying better investment alternatives can run into the 
billions of dollars in investments as previously found by China 
Railways.

With the LRDSS, the preliminary evidence is that the Railway 
Board of the IR will have an improved ability to make better 
decisions and get more productivity out of their railway 
system.  Phase II will expand that capability even more.

REFERENCES

Cook, Peter, "The Use of GIS in Improving the Cost-
Effectiveness in Transport Investment Decisions: The China 
Railway Example," Pacific Rim TransTech Conference 
Proceedings, Volume II, July 25-28, 1993.