Return to Archive -by date - by topic.
Network Rail says signalling is too expensive – but the same technology is cheaper in export markets
Like some plaything of the gods in Greek myth, Railtrack was born without those two essential corporate virtues, hindsight and foresight.
Hindsight was deliberately eschewed when Year Zero began at 00.01 on 1 April 1994 . For Chairman Bob Horton, Chief Executive John Edmonds and the other members of the new company's board, anything pre-dating Railtrack's existence was irrelevant – other than as an example of how not to do things.
Foresight was equally unwelcome in an organisation where the corporate memory had been reformatted. Anyone voicing doubts about the amazing opportunities awaiting in the brave new world was dismissed as a wrecker or even worse, hidebound old railway.
And nowhere has the resulting naïve optimism, the open armed welcome for universal panaceas, had such a malign effect as on the signalling industry. Over the past decade we have seen a series of bonfires lit on the ice and yet even now Network Rail doesn't seem to see the link between the steaming patches of exposed water and past decisions. The ability to draw lessons from hindsight and apply them with foresight is still missing.
Just bear with me while I reprise the history of signalling policy, starting in 1995 with the West Coast Main Line Development Company Report. At a cost of £6million in modern money this looked at the way forward for the WCML
The WCML DevCo's signalling McGuffin was a radio transmission based train control system or TBTCS for short. As many of the consultants came from industries outside the railway, they assumed that railways had cost and operational problems only because the technology was archaic and the engineers reactionary.
On this basis they assured me that the digital radio systems, safety critical computers and software needed for TBTCS were available in other industries and ready to be taken down of the shelf and applied to the WCML. The only hard part would be systems integration.
In exchange for embracing this risk, cab signalling would eliminate lineside equipment, saving hundreds of millions on future maintenance. TBTCS would also provide moving block, boosting capacity. These benefits justified the Core Investment Programme to restore the WCML to modern standards.
Naturally, Railtrack was enchanted, even though I never met a senior signalling engineer who thought it would work. When Brian Mellitt joined Railtrack as Director Engineering & Production from LU, where he had sold the board the concept of moving block for the Jubilee Line extension, TBTCS was go.
Oddly, when it went out to tender none of the much vaunted off the shelf suppliers showed any credible interest, and the bidding came down to negotiations under which two offers went forward in parallel. The idea was that after a period of competitive development, the most successful contractor would scoop the pool. Bidders were told that success on the WCML would give them a head start in Europe where the European Rail Traffic Management system (ERTMS) was underway.
Parallel development was junked in 1997 in favour of a single prime contractor. Signalling companies were given six weeks to bid for a system that was in advance of anything currently available and which was to be in service before the hardware and software ERTMS was expected to be ready.
Not in this column‘Who controls the past controls the future. Who controls the present controls the past'. George Orwell 1984 |
Yes, honestly. And a deluded Railtrack director told me that although the expected commissioning date for the initial installation between Euston and Crewe was 2005, if development continued as expected it could be brought forward to 2003.
GEC Alsthom Signalling got the short straw and cut the acronym back to TCS. The Company never received the definitive contact and after the initial nine months feasibility study, subsequent development funding was drip fed a few months at a time.
Eventually, when the timescale had slipped from 2002 tom 2004 to ‘2008 if we're lucky', and Lord Cullen had recommended the installation of ERTMS, TCS was cancelled.
DevCo also proposed a single a single control room for the route to complement TBTCS. This became known as the Network Management Centre NMC).
Now when it came to Network Control BR Research, as was, had followed up the best selling electronic interlocking (SSI) with the Integrated Electronic Control Centre. Just as Airbus introduced the glass cockpit , where VDU screens replace instruments, so IECC was the glass signalbox.
But Research's master stroke was to see that sitting a signalman at a bank of VDUs from which routes could be set with a tracker ball or mouse was to miss the full potential of the IECC.
Enter Automatic Route Setting (ARS) which brought together train location from the train describers, paths available from the interlocking and what the trains were supposed to do from timetable in memory and combined this information to set routes to minimise delay. Thus for much of the time the train service ran itself - often more effectively than a manual signaller - while the signaller did the things the Mk 1 human brain was best at.
As we now know, ARS didn't understand the limitations of the Mk 1 brain and could set routes which, while highly efficient, subjected even experienced drivers to magical mystery tours. After Ladbroke Grove ARS was restricted to a number of 'official' routes.
From IECC BR Research moved on to what was called Control Centre of the Future (CCF). Piloted at Liverpool Street , CCF provides a strategic over-view of train running for Operations Controllers. To avoid contamination from this catalogue of incompetence, CFF is in a separate box.
Anyway, if lack of hindsight and foresight were not bad enough, when combined with residual old railway xenophilia, it multiplied Railtrack's ability to waste money several fold. One of the wonders of the railway world is the Harrington dispatching centre at Omaha . This has a massive wall display straight out of Dr Strangelove and controls trains over thousands of miles of north America .
It certainly made an IECC look pretty low tech, even though a moment's thought suggests that running Liverpool Street on ARS in the peak in real time is a tad more demanding than getting two freight trains to met at a passing loop.
So the creators of Omaha , Union Switch & Signal, got the contract for the first NMC in October 1998. Railtrack said that its software was furthest down the development path and that the ability to extrapolate how an operational situation to develop was a key feature. ‘The ability to look ahead is impressive' gushed a Railtrack executive. Personally, I thought the claims made for the ability to predict conflict in advance and regulate trains through out the route to maintain smooth running extravagant, given the contractor's lack of European railway experience.
Originally the idea was that the new NMC at Saltley would house the TCS kit, the TCS radio block centre and the mega control room and be responsible for the whole of the WCML. But as time went buy, you could see the puddles spreading around another bonfire. Yet for some reason, even after Railtrack has crashed and burned, the NMC contracted was defend by Network Rail to the last.
It was all getting a bit like the Monty Python dead parrot sketch. You could understand Network Rail being twitchy since the £40 million building to house the NMC, including the large circular control room, had been completed back in April 2002.
It was the contractor who broke the code of Omerta first. On 17 February this year Union Switch & Signal announced that it was laying off staff who had been working on the development of control technology for the NMC. US&S explained that Network Rail had ‘suspended the project due to financial and budgetary constraints.'
When Network Rail bit the bullet US&S had a team of 135 working in Pittsburgh on the project of which 37 went immediately. US&S warned that ‘there could be additional workforce reductions based on project close-out activities, and the level of follow-on opportunities with Network Rail.' In the UK , around 50 people were affected.
What really stopped the show was development of the software needed for the US&S kit to interface with the many and various signalling installations used on the WCML. Officially, just under £100million had been spent on the aborted project out of a budget of £150million for Phase 1 of the NMC.
Under Phase 1 the staff at the NMC would have regulated traffic, either by instructing signalmen by phone or by displaying revised timetables in their signalling centres.
But Saltley is not all bad news. It houses the Midland control centre which, naturally, relies on CCF.
Then there was the low cost signalling initiative launched in 1997. Nine organisations put in proposals for replacement of mechanical signalling on lightly used lines, exploiting the new generation of vital safety single processor interlockings. Both the successful schemes were based on North American interlockings.
At the contract announcement, great heaps of combustibles were being heaped on the frozen lake. When I asked about the possibility of delay while single processor interlockings were approved by the Safety Review Panel, Railtrack's Director Engineering & Production told me that the deal would be a partnership contract to move forward technical issues which could have been stumbling blocks. He insisted that Railtrack was not in the business of ignoring safety validation work done elsewhere. ‘That would be a waste of time and money'.
Oh dear. Nuneaton-Peterborough was aborted while Vaughan , with new US owner Harmon Industries, wrestled Norwich-Cromer – nominally a £6 million project, to a protracted and very expensive conclusion.
Currently GE Transportation Systems which took over Vaughan-Harmon is resignalling another low cost line - Bedford-Bletchley as part of an alliance. The contract was signed in June 2001 and the current date for completion of the project is the first week in September – nearly a year late.
Then there was the great CBI boondoggle. This was really summed up by Railtrack's Major Projects Director, another FNB, who during his brief reign called the signalling majors together and told them that SSI was toast and anyone still making it was heading down a dead end.
Since the speaker knew nothing about railways, the fact that SSI was an international best seller and that even French Railways had adopted it was probably overlooked. What we needed was whizzy CBIs – Computer Based Interlockings. SSI is, of course, a CBI.
So Alstom and Westinghouse went on selling SSI to more enlightened railways. And having upped the processing speed of SSI with new hardware, they also got on with their own next generation CBIs, Smartlock and Westlock respectively. Where SSI was a bit too capable, Westinghouse also offered its single processor WestRace interlocking.
Meanwhile, as reported interminably in these columns Railtrack let pilot scheme contracts for CBIs to Antranz (Ebilock/Horsham), Ansaldo (ACC/Manchester South) and Siemens (SIMIS-W/Dorset Coast).
Looking through that list you can see disaster ahead. Bournemouth was a nice quiet place for Siemens to get to grips with our ailway, Horsham was handy for Adtranz, since Ebilok was earmarked for Thameslink 2000. But Manchester South as a nice quiet, no pressure to complete test sight?
Anyway, having been formally opened in shadow mode by Capt Deltic, Horsham was clearly doomed and is now to be resignalled with SSI by Alstom. Ansaldo eventually got a truncated scheme into service – and very nice kit it is too.
Siemens followed with acceptance for Bournemouth Sea Front, the project having been cut back. However what that project does is expose the unsuitability of the Siemens system for long thin railways. Since everything is hard wired and the maximum cable length is 6.5km you need a remote container every 13km, even to drive automatic signals on a lightly used line.
Which brings us up to date, with SSI rehabilitated asthe standard interlocking for new schemes in the short term, CCF the national standard operational control and some hard decision to be taken on future interlockings and ERTMS, or, more properly, ETCS, the European Train Control System.
But dominating all else is the fact that Network Rail, under the lash of the Regulator, says that signalling is too expensive. Which is where we come to the SEU.
How do you measure the cost of signalling schemes? It's a hard question because signalling schemes vary in terms of length and density of equipment. A Westlock CBI that will fit in a briefcase can control the lines out of Euston.
Network Rail's solution is to focus on the hardware the interlocking controls , point ends, signals and level crossings. Each of these is termed a Signalling Equivalent Unit (SEU). Divide the cost of a project by the number of SEUs and you get a tolerably coherent yardstick.
Currently Network Rail has around 1,650 interlockings controlling 61,400 SEU. Table 1 shows the breakdown of signalling assets and Chart 1 their age profile.
|
Signalling centres/boxes |
Signals |
Point mechanisms |
Train detection sections |
Level crossings |
|
Number of assest |
906 |
38017 |
14786 |
60994 |
8612 |
|
Network Rail's current benchmark for planning purposes is £270,000/SEU. The aim is to get down to £100,000/SEU. Some prices are coming in around £300,000/SEU. ETCS is just too expensive.
These costs help explain why, when electro-mechanical signalling still has some life in it, life extension is the preferred policy. Network Rail is also in the throes of developing a National Signalling Plan, which clearly will depend on the costs of various technologies.
Because of the current uncertainty, the Regulator's interim Review funded Network Rail's signalling expenditure for the first two years only. A mini Interim Review will review signalling expenditure from 2006/07 onwards when the plan has been developed.
This is small comfort for the signalling supply industry which has had staff sitting around waiting for projects. And the work situation is not getting any better.
Network Rail has just cancelled the Leicester-Nuneaton resignalling with SSI on the grounds that it was too expensive and, since much of the existing signalling was not life expired the electro mechanical relay interlockings, will be renewed of refurbished.
Cancellation came after just under £2million had been spent on scheme design for the project. Total value of the signalling work, awarded to Westinghouse, was around £12million to buy 53 SEU. The unit cost of £226,400/SEU was well below Network Rail's notional £270,000 in its business plan.
Just out of interest I got a chum with first hand experience of SSI in other countries to cost Leicester-Nuneaton using SSI. The cost came out at just under £100,000/SEU.
Since hardware costs are much the same, it must be a case of ‘it ain't what ya do, it's the way that ya do it'. And as pointed out in this column a while back, if an SSI installation costs £5million for the hardware and software, the associated treacle doubles the price.
According to one very informed source, a footbridge that costs £80,000 to fabricate, will have cost Network Rail £800,000 by the time it is installed, of which £400,000 represents compensation for possessions.
Meanwhile, Network Rail is moving from replacing local interlockings on condition, to a route based renewal strategy, which is progress. One expected benefit is improved efficiency in the supply chain by smoothing the work profile and reducing unit costs.
Amen to that, but on technology, there is still institutional confusion. As this extract from the 2004 Technical Plan shows.
‘We have recently introduced modern CBIs developed by Siemens and Ansaldo which are expected to deliver efficiencies in the medium term. Each of these pilot schemes has exceeded the anticipated budget and timeframe because of the complexity of signaling principles and safety approval processes in the UK . We are therefore working with suppliers to optimize the overall design, , installation and commissioning for these new technologies. A number of existing standards and engineering processes may need to be modified to facilitate this, and a study is being undertaken to identify the required revisions. CBIs will be needed to facilitate the eventual roll-out of ERMS Level 2 once the UK application is mature.'.
Then, almost as an aside, it adds:
‘In addition, both Westinghouse and Alstom are developing second generation SSIs that will be compatible with existing equipment'.
Good grief! Railtrack lives! Hindsight is compounded by ignorance.
Westlock and Smartlock are not some sort of Morris Minor with a Toyota engine and disc brakes compared with the Ford Focus CBIs; they are CBIs running on safety critical processors' But they just happen to use one of the best selling signaling software architectures, namely SSI.
They will be able to interface with ERTMS Level 2, because even Almost is not so thick as to dive down a genuine blind alley. In a sane world, backwards compatibility with SSI and forwards compatibility with ERTMS would be a massive plus point. Not on Planet Network Rail,
And it gets worse. Westinghouse is still waiting for a demonstration application site for Westlock.
Cherwell Valley Phase 1 resignalling, using SSI, was completed in 10 months from a standing start, to the delight of the SRA and Virgin Cross country. Signalling contractor Westinghouse came up with a crafty ploy to simplify training on the new WestCad signalers desk..
They wired up a cubicle with a Westlock interlocking and loaded it with the Leamington SSI data. A computer simulating train operation fed the Westlock which was controlled from the WestCad which also had Leamington loaded. So the signalers could be trained on a layout they knew.
Now Leamington is where the first SSI was piloted in 1985. And the existing SSI is connected to the working railway by a handful of wires. And Westinghouse has an ISRP endorsed safety strategy for a trial installation of Westlock. All it needs is for someone to say where and when.
So, in a sane world, with Westlock already sitting there running a simulated Leamington layout, you might have thought that the opportunity had arrived. All you had to do was fit two way switches in the wires coming from the real world so that either Westlock or SSI could run the junction.
Initially Westlock would have been switched in and tested with the route signaled under possession, but, all being equal, after successful testing, the switchover would become permanent and Network Rail would have a state of the art CBI running, which would not have cost an arm and a leg, didn't require changes to the signaling principles, but was forward compatible with ERTMS where, of course, Westinghouse is a member of the UK and European signaling consortia developing the system.
But, of course, it didn't happen.
But not all is sweetness and light for Ansaldo and Siemens either. At least Ansaldo has got Sandbach-Wilmslow, well it still had the last time I checked. Butt having been promised the Saltley and Glasgow resignalling schemes as a follow-on to BSF, Siemens has heard nicht and is getting a bit tetchy.
Actually they should keep schtumm lest Network Rail discovers that the eponymous Wernher set up his first company in Britain and that Siemens is a major UK employer. Or perhaps the penny has dropped.
AEA Technology Rail's Control Centre of the Future (CCF) uses the same train operating data already produced for the Integrated Electronic Control Centre (IECC) and processes it to give operations controllers a real time display of train running, with the ability to highlight delayed trains identified by headcode. It also records the data, so that an incident can be replayed.
In the privatised railway this facility is invaluable. Since someone has to pay for delays, knowing who was responsible, and for how many delay minutes, is worth serious money. So CCF is now installed at all Network Rail control centres so that incidents can be replayed and blame attributed accurately.
Delay attribution brought CCF back from old-railway oblivion. Now, belatedly Network Rail is starting to appreciate its full potential – doing what the Saltley NMC was supposed to do at a fraction of the cost.
AEA is pushing hard on extending CCF to become a true train path manager. This is well within the state of the art and could have been operational by now had Railtrack had the nous.
Instead of looking backwards for fault attribution, the CCF software looks forward to predict the impact of a developing situation on the timing of trains. This can be presented as a classic a timing graph which the controller can then use to reallocate paths, entering the changes manually into the IECC's timetable.
Say, for example, a freight train is late off the terminal. CCF would take the delay, plot a new graph and the controller could consider the options. This is roughly what Phase 1 of Saltley promised.
But do you think that those wonderful folk who brought you ARS would stop at train graphs for manual intervention? Of course not.
So being demonstrated now in prototype form, having been available for some time, is the 21 st Century version. You give CCF the decision criteria for your late freight train. This could be ‘minimise delay to the freight train', ‘minimise delay to long distance passenger services' or ‘spread the pain equally across all services'.
CCF then predicts the impact, applies the appropriate criteria to produce an amended timetable and downloads it into the ARS software of the IECC. In other words you have automated perturbation management.
I've just had a week's holiday. The red mist will return shortly.