Tonight’s British Computer Society (BCS) event in the Davidson Building opposite the Savoy in London was the first one I’ve attended for years. The subject was the Network Rail “ORBIS” programme which focuses on data asset management.
The attraction for me was three-fold: First, I’m developing a keen interest in the railways generally thanks to our work with ATOC RSP and second, because of Smart421’s focus on BigData challenges. Thirdly, the agenda mentioned something about specialist mobile apps, which is my main interest in leading our Enterprise Mobile capability in Smart421 which is backed with IBM Mobility solutions.
The slick presentation from the impressive Davin Crowley-Sweet and his team brought out the approaches to data as an asset and highlighted the challenges of modernising an organisation that still collects much of its information, such as survey data, on paper. The slideshow was “Prezi” style zooming in and out like a game of Angry Birds and contained a LOT of detailed analysis and roadmaps which made me glad I sat in the front row to try and take it all in.
My rough notes are summarised here but I believe the interested reader can learn much more about ORBIS (which has been “reverse-acronymed” with the name “Offering Rail Better Information Services”) from the Network Rail website. The first point to note is that this is not another BigData experiment but an extremely far-reaching business transformation programme, focussing on business processes and aiming to be business-led. There are very serious business benefits already being gained and (according to the paper linked to above)
the £327 million total cost of ORBIS is expected to yield £270 million in asset benefits plus £100 million per year in maintenance improvements.
An iPhone for every maintenance engineer
The company took the far-sighted step of giving every engineer their own ruggedized iPhone along with the request to use the built-in features as best they could to suggest what can be used. Not surprisingly, one of the most useful features was GPS, accurately pinpointing problems with rails. Then when engineering work comes along on the Sunday consigning us travellers to the replacement bus service the disruption should be minimised as engineers should be guided directly to the problem location. Taking photos with the iPhone of problems with rails was another obvious use and now there are Apps being developed (one per month on average) that can automate some of the data capture along with the photos. I asked the presenter Edward afterwards about the apps and he said there are now developments across other platforms and O/S to use other (perhaps cheaper Android-based) handsets. Having around 12,000 engineers with a ~£500 iPhone is significant cost although the positive aspect of the workers being given the desirable devices should not be overlooked.
These unstructured data inputs were combined with significant mapping data, collected using helicopters and cameras on trains similar to “Google Streetview”. I think they mentioned collecting a few Terabytes of data from each train journey that was recorded in this way. That is some BigData problem to deal with the analysis and Network Rail to develop some sophisticated decision support rules around which stretches of railway need maintenance work most urgently based on the data analysed around curvature of the line, weather conditions, etc.
Some 20,000 miles of track, 40,000 bridges and tunnels and huge electrical, telecoms and physical networks make for a highly complex set of problems to manage. The seven year mission focuses on 5 asset types: Fixed assets, Fleet assets, Topology, Topography, Unstructured data (schematics, drawings, etc). It has several defined stages, in stage 1 it asked “what?” and “where?” and continues in stage 2 to try joining up and optimising the collected asset data.
The Contribution of Data Analysis to saving lives
There is a strong emphasis on improving safety through improvements to data management; recommendations following rail disasters like the terrible Lambrigg crash in Cumbria, for which Network Rail are still apologising, said that the points failure could have been detected and fixed earlier with better data analysis. Literally a case of life and death software.
Davin answered questions after the presentation and made the very relevant observation that enterprises should manage their operational data in the same way as they manage any other (physical) asset: know what it is and where it is, monitor its quality, use it while it is relevant and when it reaches end of life get rid of it.