This is the second of two posts published today about offsets. The other post is a useful background into offsets and a broad overview on how we calculate them.
Aviemore was resignalled in March 2019 into the new signalling workstation at Inverness, initially installed as part of the Aberdeen to Inverness resignalling project between Nairn and Elgin. The old signal box at Aviemore was opened in 1898 as part of the station expansion in that year and is a listed building. It was a mechanical lever frame and was the largest surviving operating box of its type. The new signalling system is a digital offering from Siemens Mobility.
We chose Aviemore for this post as it is a relatively rare example of an opportunity to go back to first principles. Train reporting was previously done on a manual basis and therefore we had no previous work to go back on.
The above diagram is a simplified version of Aviemore station and does not include the connection onto the Strathspey Railway or the engineering siding that is adjacent to the through line in the down direction after the station. The identities shown are the berths related to each signal - we will refer to each signal by its berth identity.
Aviemore is a relatively simple layout forming of a single through track and a passing loop. The station area between A313 and A326 on the diagram has a 40mph permanent speed restriction (PSR) and all routes are bi-directional. The diverging routes onto the passing loop are not approach controlled and the system will show the least restrictive aspect possible on any route. Platform 1 is the preferred platform.
There are two arrival routes from each home signal into the station into platforms 1 and 2. There are two departure routes from each platform, one in each direction. The arrival step past the home signal will move to the appropriate departure signal for the direction of travel - this is not always the case across the network. There are 6 distinct steps that require measurement for normal service - the route through platform 2 in the down direction is rare. There are several additional steps that require values, such as those for platform 3 and the Strathspey Railway, but this is not covered on this post.
There is an irregular service that is slightly less than one train per hour in each direction. The passenger traction serving this station when we visited included class 158 and 170s of various lengths, short 2+4 HST sets (ScotRail’s Inter7City service), a daily 2+9 LNER HST, soon to be an IET, and the Caledonian Sleeper then formed of 67 and Mk3s, now 73/9s and Mk5s. There is an infrequent service Royal Scotsman operated with GB Railfreight class 66s.
Aviemore was visited in late June as part of efforts to measure the newly signalled areas of the Highland Main Line as well as other parts of the Scottish Central Belt that required checking. The newly signalled areas were Moy to Kincraig loops and Pitlochry station. Pitlochry is not controlled by Inverness but rather by Stanley Junction - its data is broadcast through the Inverness.
The irregular service pattern meant the most appropriate way to maximise our coverage involved shuttling up and down the line. A Spirit of Scotland Travelpass was used as the easiest way of covering all the routes at a reasonable cost. Our target was to get at least six measurements for all the common steps and four for the less common. We have worked on mathematical analysis to infill the gaps for less common runs based on similar steps across the country.
As an example, on 27 June I did the following journeys:
- 1T98 09:36 Kingussie to Perth (2+4 HST)
- 1H09 11:19 Perth to Pitlochry (170)
- 1H11 12:24 Pitlochry to Aviemore (170)
- 1B31 13:35 Aviemore to Dunkeld & Birnam (170)
- 1H13 15:06 Dunkeld & Birnam to Inverness (170)
- 1T62 17:27 Inverness to Pitlochry (170)
- 1H53 19:50 Pitlochry to Kingussie (2x 170)
Platform 1, Down direction
I’m not going to focus on every platform direction in detail as that will result in a lot of rather boring data, so I will focus in on timings at platform 1 in the down direction.
|Day||Traction||Train ID||A313>A319 time||Wheel stop time||‘Arr’ diff||Wheel start time||A319>C333 time||‘Dep’ diff|
|2806||43026 + 43021||1H05||09:50:53||09:52:06||+73||09:53:44||09:54:26||-42|
|0607||43135 + 43149||1H05||09:53:16||09:54:31||+75||09:57:18||09:58:01||-43|
As is normally the case for a signal slightly further away from the platform, there is a slight range of departure times but most are consistent around 42-43 seconds from wheels start. Arrivals show a broad range and this is largely down to the aspect shown to the driver on approach to A313 signal. On the arrival for 1H13 on 27th, the signal was yellow (A319 was ‘on’) as the train arrived at the station due to passing another. Our data shows that they arrived at virtually the same time.
The 2+4 HST set (Inter7City) differences are slightly higher on arrival and this we consider to be for a number of reasons including a lack of driver familiarity and different braking characteristics from the normal DMU operation. We considered the data as being valid for now, as it is a new area for both us and HSTs, and marked it for revisiting when more HSTs have been introduced on the route. We intend to also measure the class 800 ‘Azuma’ fleet for a value as we believe this may be close to a ‘typical train’ based on experience from the Western region - the characteristics of a multiple unit but of 9 car length being potentially similar to the Inter7City characteristics overall.
Using the above measurements, the calculated offsets for arrival and departure of platform 1 in the down direction were +67 and -43 respectively. Aviemore was marked as a location of interest for further visits and continues to be so - our offsets have been revised since this trip and continue to be so as part of our continual drive for improvement.
Contrasting the data
I thought it would be an interesting contrast to end this blog post with our caluclated offsets from this trip compared those in SMART made available by Network Rail at the time of writing through the Open Data portal.
|Activity||Step||RTT offset||NR offset||Notes|
|Platform 1, down arr||A313>A319||+67||+80|
|Platform 2, down arr||A313>A321||+70||+85||None recorded, forecast using P1 down and P2 up|
|Platform 1, down dep||A319>C333||-43||-30|
|Platform 2, down dep||A321>C333||-45||-30||None recorded, forecast using P1 down and P2 up|
|Platform 1, up arr||A326>A318||+75||+60|
|Platform 2, up arr||A326>A316||+82||+65|
|Platform 1, up dep||A318>K306||-22||-20|
|Platform 2, up dep||A316>K306||-23||-20|
There’s a fairly substantial difference on some of these offsets as you can see. Network Rail’s offsets I believe to be the ones set up at the resignalling stage using mathematical models based on the plans of the scheme. It is unfortunately common that on resignalling the initial offset values are sometimes out by occasionally minutes.
The goal of Realtime Trains is to be the most accurate website for real-time reporting across the entire rail network and I believe that developing our own offsets is the best direction to achieve this within the current data environment.
This is a really interesting topic for myself, and for various people I talk to in the industry (I spent five hours talking about Train Describers and offsets to various people on Friday!), so if you’d like to ask more information please get in touch! You can contact me personally on Twitter, on the Realtime Trains page on Facebook or via our feedback email. If you’re part of the industry and would like to know more or work with us, please see our contact page.