Here is a paper that discusses how accidents take place and what are the most risk-prone zones in Indian Railway. Scientists from IIT Kharagpur analyse the 2010 Indian Railway Accidents. My report about the same in The Hindu today Click on this to read the article
And if you have a bit more time to read at leisure, here is a longer version...
There are many things going on around us that we
seem so sure about and yet many times it remains a gut feeling only. For
instance the changing of seasons – everyone observes this but we do not seem to
have any scientific study that quantifies this change. Similarly comments about the growing traffic
on railways are rife yet many of these comments remain at a subjective level.
However, there is news on this front. Moved by the spate of railway accidents
that took place in 2010 in India – 19 in all of which 11 were due to
derailments or collisions between trains or some sort of failure of the railway
system itself (as opposed to fire or faults in the train or some other reason), Saptarshi Ghosh,
Avishek Banerjee and Niloy Ganguly of Complex Network research Group (CNeRG),
IIT Kharagpur have taken it upon themselves to analyse this phenomenon.
Published in Physica A, this year, their results are
twofold – first, they identify zones of heavy traffic and the reason for
accidents as the fact that the number of trains has grown over the years
whereas the railroads and routes have not proportionately been scaled up.
Second, they identify serious flaws in the scheduling of trains on some routes.
They find that if on these routes trains were to run as per the Indian Railways
schedule, the system would not be able to handle the traffic. In reality, this
is managed by making trains wait at signals because of which there are long
delays in runtime of trains. The second
is an alarming result because it holds in its heart the possibility of human
errors leading to accidents – what if the signals malfunction or if the driver
fails to respond to the signal or some other such disaster?
The year 2010 was grim in this aspect – there were
19 accidents in all of which 11 fell into the category the authors consider. It
had the feature that eight of the eleven shortlisted accidents took place in a
zone which they call the Indo Gangetic Plain – a north eastern belt of
India. Noting this, the authors identify
statistical parameters to measure this phenomenon and do a statistical analysis
of the existing Indian Railway (IR)
express train routes. They find that the Indo-Gangetic Plain hosts some
of the most traffic intensive segments of rail routes (7 out of the 20 that
they consider high-traffic). Comparing data gathered from 1992 to 2010 from
Trains at a glance, they conclude that this is because the infrastructure such
as railway lines and tracks have not grown over the years, whereas the number
of trains has increased many times. They identify the most risk-prone “trunk
segment” as the Delhi-Tundla-Kanpur one and identify the
Vishakhapatnam-Vijayawada trunk segment from the southern zone as the “safe
standard” owing to the empirical evidence that it has not had any accident so far.
Another parameter is the headway, or time lapsed
between two trains as they cross the same point. As the headway reduces, the
chance of two trains coming dangerously close to one another increases. On
analysis, two segments clearly come out as risk-prone segments The Delhi-Kanpur
segment and the Ahmedabad-Surat segment.
The Vishakhapatnam-Vijayawada segment has a much higher headway and therefore
is safer, relatively speaking. Of the
two segments with low headway, the Ahmedabad-Surat segment has trains with low
headway running throughout the day, whereas in the case of Delhi-Kanpur
segment, they are bunched up in the early hours. This once again points out the
latter as more unbalanced.
Runtime delays of trains on these segments were also
studied and it was found that 20 percent of the trains on the Delhi-Kanpur
segment were delayed by more than one hour, whereas about 3 percent of the
trains on the Vishakhapatnam-Vijayawada segment were delayed to that extent.
The delays reflect the high degree of congestion and frequent waiting of trains
at the signals - which again as pointed
out earlier implies a high probability of collision due to malfunctioning of
signal or driver failing to react to signal.
To analyse the congestion of traffic at a
fine-grained level, a simulation of the traffic flow according to the IR
schedule was done. The authors modelled the “block system” followed by Indian
railways. A railway track is divided into block sections (of about 4 km to 8
km) such that when one train is occupying a block, no other train is allowed to
enter that block on the same track. At
the end of the block, there are signals or stations which control the traffic
entering the block.
From the simulation, it transpires that if all
trains were to keep to the schedule and not stopped by signals etc, then for
trains in the Indo Gangetic plain (IGP) there would be more than two (or three)
trains in one block quite frequently. Now, while some blocks have three tracks,
most of the IR blocks have only two tracks and so can accommodate at most two
trains. So this indicates that the infrastructure is not sufficient to handle
the traffic and this is only being managed by stopping trains and delaying them
beyond the schedule. This is an alarming result because it implies that if the
trains ran as per the schedule, for the IGP trains, the infrastructure would
not be able to accommodate all of them. This is a result that begs for
proportional improvement of infrastructure.
Do they have an alternate method for scheduling
trains which may be safer? Niloy Ganguly replies, “We would like to do a study
on better scheduling of IR trains, but it will only be meaningful if we get to
know the present infrastructure accurately, e.g., the traffic handling
capacities of different routes, etc… we
had requested the IR authorities for these information, but are yet to receive
any favourable reply from them.”
How sound is the rationale behind keeping the south
zone route as a safe standard? Is it not better to keep an absolute value on
safety? The author says, “It will
definitely be better to use an absolute standard, but we do not know of any
such standard for IR…. Note that there have been derailment / collisions even
in south India in 2012. Hence, some segments in south India also seem to be
nearing the risky zone. However, this means that the condition of IGP is even
worse than what we had estimated in our paper (since the safe standard itself
is no longer very safe)”.
