OpenStreetMap

bentrails's Diary

Recent diary entries

Getting railway tunnels (almost) right

Posted by bentrails on 12 November 2011 in English.

I drive commuter trains around Sydney every day... many of the lines I've been GPS tracing recently and this works fine as long as you are above-ground. The challenge now is to get a better alignment of Sydney's underground rail lines.

This week my focus was on the Eastern Suburbs Rail line (ESR) between Erskineville and Bondi Junction. Large parts of it are underground, and the estimated tunnel routes shown on OSM just weren't matching with what I was experiencing as a driver at the front of the train.

So how do we fix this?

My initial thoughts were to make a rough sketch as I drove along. With experience I've become very good at judging distance ahead of me in measures of train-length due to the number of instances where I have to restrict my speed (e.g. going across a set of points - or a turnout) until the entire train has passed the risk. I do this by judging an imaginary point ahead of me as being my train length, and when I reach that point I can resume speed. All the trains I drive are 8 cars long (about 160m in total), so I could say for example, the track continues straight for 2 train lengths, then a gentle left curve for 3 train lengths, immediately followed by a hard right curve for half a train length.

But my sketch idea didn't work. I simply couldn't get everything onto paper that I needed as well as drive the train. It just wasn't practical - so this week I tried the Voice Recorder on my iPhone and verbalized everything I experienced as I drove through the tunnels. This gave me some really useful data which I was able to use to fine-tune the alignment of the underground sections of Sydney's Eastern Suburbs line on OSM.

Information like - "this long sweeping right curve continues for about 2 train lengths, and then we exit the tunnel the same curve continues for another train-length outside".... is really useful as you can already see the ourdoors view of the curve in OSM, and knowing that this same curve extends back into the tunnel for about twice the distance at the same curvature is a great help in approximating the underground alignment.

This isn't perfect. You still must know (through common sense) where the stations and platforms are, and how everything needs to line up... but if we have the platform alignments (and underground platforms are usually aligned with streets or buildings above ground) and the tunnel portals - all we need to do is join them up as best we can.

Also - in some places there are above-ground clues as to where the rail tunnel goes. I found a lot of cases along the ESR where the buildings directly above the underground sections tended to be newer buildings. This rail line was opened in 1979 and there is a distinct strip of newish buildings are above the line in many places. This is especially the case right through Kings Cross, Edgecliff station, and the Bondi Junction end.

So I'm happy to say I've fine-tuned Sydney's underground ESR to a point where as a train driver, I am 95% pleased with it, including the location of the underground sidings and crossovers at the Bondi Junction end. It's even better to then compare my results with other sources to see that I was pretty much spot-on, or even more accurate in some parts.

My next areas to focus on improving will be the Airport Line and Chatswood to Epping link.

Location: Woollahra, Eastern Suburbs, Sydney, Woollahra Municipal Council, New South Wales, 2025, Australia

Here we are

Posted by bentrails on 10 October 2011 in English.

Re-awakening my interest in OSM. On the surface it seems like so much has already been done, but digging deeper there is always more to do. As a train driver I have a lot of knowledge about the CityRail system so that might seem a logical start - then again, do I want to dwell on work during my leisure time?