INTRO: The Railtrack board will shortly receive the business case for concentrating operational control of Britain’s 16600 route-km in about nine Network Management Centres. Richard Hope asked Andrew McNaughton to explain the strategy
CSX TRANSPORTATION was the first major railway to exploit the almost limitless capacity of modern communications technology by bringing all aspects of train control into a single operations room. The Dufford Centre at Jacksonville went live in April 1988, and today controls some 1400 daily train movements on 29000route-km.
Union Pacific was not far behind, opening its Harriman Dispatching Centre at Omaha in June 1989. Subsequent mergers, notably with Southern Pacific in 1996, have expanded UP’s network to 58000 km carrying some 2000 daily trains.
Burlington Northern had its Fort Worth control centre up and running by 1995, just as the merger with Santa Fe came along. Fort Worth is currently the biggest rail operations room in the US with around 100 workstations, but CSXT is looking to upgrade or replace Dufford as soon as the division of Conrail routes with Norfolk Southern is confirmed.
Impressive as these US control centres are in geographical reach, the dense national networks of Europe are far busier in terms of train movements. In Britain, for example, 25000 trains are scheduled daily on 16600route-km compared to around 2000 for each of the big five US freight railroads.
Although many British passenger trains complete a one-way trip in under an hour, whereas US freights take many hours and even days from origin to destination, it is the frequency of interaction of trains at junctions and interlockings which determines the flow of data to be handled at the control centre.
Moreover, away from the big cities CTC often does not extend through terminal areas, such as UP’s yards in El Paso where through trains simply slow down and drive on-sight over hand-thrown switches innocent of any interlocking. In Europe, virtually all running lines are signalled, vastly increasing the complexity of software needed.
Both DBAG in Germany and Railtrack in Britain have announced plans to bring train control over the whole of their respective networks into fewer than 10 regional centres, although neither has taken the final decision to go ahead. The ultimate objective is to eliminate all existing signalboxes and manned level crossings, though this will take many years and is obviously dependent upon the replacement of manually worked points, signals and crossings on secondary routes.
First approval later this year
Andrew McNaughton, Railtrack’s Engineering & Operations Development Director, is charged with preparing the business case for building the first of about nine Network Management Centres covering the 16600route-km national network. He expects to place this before the Railtrack board in midsummer. If it is accepted, construction of the first pilot NMC should be authorised by September; this is likely to be at Ashburys in Manchester.
Though by no means the only factor, the order in which NMCs are built and equipped will be governed primarily by the need for resignalling. The North West is favoured for the pilot NMC because the Manchester South area will need a new command system anyway by the time that the first release of software will be available. Immediate resignalling mainly affects lines out of Manchester’s Piccadilly and Victoria stations, but the North West Zone extends from Crewe to Carlisle and Holyhead.
McNaughton points out that ’the alternatives to creating an NMC are to install obsolete systems such as panel processing or early software-based Integrated Electronic Control Centres, or to accept for each scheme whatever is offered and end up with a hotch-potch of systems. So the more we resignal, the greater the need for an NMC in that zone.’
The position at the end of May was that four groups had submitted bids to develop the software and hardware needed for the first stage of the North West NMC, and indicative bids for the remainder. A contract will be awarded when the Railtrack board accepts the business case for going ahead with NMCs as a network-wide strategy.
Whilst Railtrack Chief Executive Gerald Corbett is concerned that the ’headline prices’ are too high (below left), the NMC strategy will be implemented over two decades or more, and the suppliers have had to price their future risk. McNaughton feels the indicative bids have given Railtrack a clearer understanding of the costs and difficulties of the various elements. The NMC project team ’can now see the way to an initial system and future upgrades’ within the overall business case, although there will inevitably be ’plenty of hard negotiation ahead of us to get there.’
Year of preparation
The team has spent the last year trying to define what operational processes should be in 5 to 10 years time. This has been matched not only against what the technology will be capable of, but also what the staff will be capable of doing. NMCs are seen as an operational rather than technical development - ’fundamentally, it is about how we manage the network.’
State of the art in Railtrack’s current resignalling projects - driven primarily by the need to replace relay interlockings which are becoming unreliable or potentially unsafe because of wiring insulation failure - is the solid-state interlocking commanded from an IECC. These were developed in the early and late 1980s repectively.
According to McNaughton, ’there will soon be more computing power in the average Game Boy than in our 10-year-old IECCs. As for the older SSIs, they still run at 1MHz. We urgently need to introduce faster computer-based interlockings which are now available’. So ’it is no longer a case of us specifying the technology we want.’ The appropriate question is: ’what around the world is capable of being introduced into our environment?’
He sums up the year’s work as ’understanding what we need to do; determining its value; understanding how fast we should move; and finally, developing a philosophy as to how we should do it.’
Role of the NMC
McNaughton says his view of the functions that an NMC should perform was strongly influenced by a visit to UP’s Harriman Dispatching Centre in early 1995. ’The Americans see them as a means of delivering services, not just a big signalbox to keep trains apart ... that’s why we are calling them management and not signalling centres.’
At the core of the NMC is everything to do with operating and monitoring the network. The plan is to encompass all activities within a 12 week time horizon, because that is when the timetable is frozen so far as disruptive engineering work is concerned. Those engaged in longer-term planning may be in an adjacent room, if convenient, but will not be on the operating floor. In some cases it may be appropriate to co-locate the NMC and zonal headquarters buildings.
In addition to signalling and train regulation, the NMC will assume control of traction power switching and isolations, passenger information at stations, and engineering work affecting the running of trains. It will also monitor the condition and status of key equipment such as points, or often-bashed bridges, and perhaps some station plant as well.
Another important function to be handled at NMCs will be the analysis and attribution of train delays and incidents, both for the performance regime payment process and to decide and action solutions to avoid repetition.
While Railtrack’s NMCs will be bigger than the existing US ones in terms of workload and processing power, they will not be much bigger. BNSF’s dispatching centre at Fort Worth has over 100 workstations, whereas the largest Railtrack is looking at is about 50. However, BNSF is an integrated railway whereas Railtrack is only responsible for infrastructure maintenance and train control.
US control centres also handle locomotive and crew rostering, and issue instructions for the delivery and collection of wagons from industrial sidings. Britain’s 30-odd train operating companies will be able to locate staff in the NMCs if they wish, but English Welsh & Scottish Railway, the principal freight operator, is already building its own national control centre at Doncaster to manage its business.
To establish the practical limits on the size of an NMC, Railtrack undertook a study of the main networks and hubs, and their interaction. This suggested that nine was a logical number, reflecting the practicality of previous management structures for the British network. The favoured locations for six of the NMCs fall one each in six of Railtrack’s seven zones - which may lead to some adjustment of boundaries in due course.
For the Southern Zone, which covers the complex 750 V third-rail commuter network south of London, the plan is to build three NMCs on a ’common campus’. While this is essentially dictated by the volume of data, the risk to performance of a major failure suggests that it would be unacceptable to run the whole third-rail network from one centre.
Flexible staffing
Another lesson learned from the Americans is the need to migrate people on and off the operating floor continuously as the workload varies - a good reason for locating all of the production function in the NMC and not leaving part at zonal headquarters.
McNaughton cites the Long Island Rail Road’s ’low-tech but very effective control centre where, when it snows heavily, the whole finance team stops paying bills and assists with recording train performance, freeing the specialist staff to concentrate on running the service.’ In the last analysis, though, ’the location of an NMC will be determined by the need to keep our skilled and competent people.’
’We want a system where, when there is a major problem, an individual workstation, which might have been supervising 100 km of railway, can zoom in to deal with a particular junction, while other operators reconfigure their control areas to help out. We estimate that we can reduce total delays by 25% to 30% with this flexibility and the faster communication and decision making from having everyone together.’
Asked how far multi-skilling is likely to go, allowing operators to move from one function to another as workload dictates, McNaughton is more cautious. ’It is far too early to say - we will have to see whether it is sensible for train operators to work in all areas. We want an orchestra, not a collective of buskers.’
The level crossing problem
One particular difficulty which Railtrack faces is level crossings. At March 31 1997 there were still 581 manned crossings on the network with hinged gates or lifting barriers, and a further 321 supervised remotely by CCTV. Every such crossing must be inspected visually once the gates or barriers have closed, before protecting signals are cleared, to ensure that no road vehicle is trapped.
CCTV crossings create a major stream of work. If staff have to be retained solely to operate level crossings, as happened when the East Coast main line was resignalled because 200 km/h operation precludes the use of automatic crossings under current rules, the financial benefits of controlling large areas from one location can be undermined.
To get around this problem, Railtrack is starting some fundamental research to see whether there are cost-effective ways of achieving even greater safety without viewing the crossing manually. Programs that scan video pictures to detect the presence of a vehicle are a possibility; ultrasonic detectors show promising results in Japan (RG 6.98 p423).
Another reason why automatic crossings are not acceptable in specific locations is excessive warning time before a slow train arrives. Speed measurement to permit constant warning time could resolve this difficulty.
Control migration strategy
Putting up a building containing an operating floor big enough to supervise a large route network is easy. Developing the software and installing the hardware will take much longer, and the cost is substantial. More complicated is the transfer of control from up to 30 power signalboxes and two or three electrical control centres.
McNaughton sees the key question about migrating control as the ’rich heritage of technologies and conditions’ to be found around the network. The suppliers’ indicative bids showed relatively high cost for developing interfaces with older relay-based interlockings, where the diversity of equipment precludes a standard ’plug-and-play’ interface. This is likely to have a significant impact on the business case for migration from existing signalboxes.
Asked when the last power signalbox will close down, McNaughton says some ’may migrate fairly quickly, certainly by 2005, whereas those in good condition may live on well beyond 2010. There are many factors including synergy of whole route control which we are still analysing.’
Oddly enough, and assuming all goes according to plan, he believes ’it is likely to be the remaining mechanical signalboxes which survive longest, though many may disappear through local schemes that offer adequate return on the investment.’ And closure of the very last signalbox on Railtrack’s network? ’Not in my lifetime.’
The basic rule governing the pace at which NMCs will be set up and expanded is that each step ’has to demonstrate best value.’ This means that the system upon which NMCs rest has to be inherently flexible, and ’must be capable of several generations of upgrading as the network and business environment develop.’
NMCs must not be technology driven, though. ’They must be business driven through operational need, with people who will work in them at the heart of the process, not add-ons. So we will do it at a pace that makes business sense - and on the least busy 10% of the network, NMCs will have virtually no impact at all.’
Above all, McNaughton emphasises, ’NMCs are about integrating all the service delivery functions. There would be no value in simply building ever-larger signalboxes.’ o
CAPTION: As Railtrack’s Engineering & Operations Development Director, Andrew McNaughton CEng MICE is responsible for integrating command and control developments including signalling and the DART digital train radio project
CAPTION: Computer technology has advanced so far and so fast that the major problem for Railtrack has been deciding exactly what the scope of a Network Management Centre should bePhoto: AEA Technology
CAPTION: Workstations offer the flexibility to be configured to monitor almost any aspect of railway operationPhoto: AEA Technology
Bids higher than expected
Four firm bids have been received for equipping the first Network Management Centre in Manchester, along with indicative bids for completing the programme. Our newsletter Rail Business Intelligence reported on May 14 that the bids were not low enough to support the business case. The four bidders are:
