Network Rail is making remarkable progress upgrading its southern territory power supply. Jackie Whitelaw reports.
Andy Mitchell must have nerves of steel. On Wednesday the man in charge of the power supply upgrade for the whole of Network Rail's southern region was due in front of a House of Commons select committee to explain how progress on his scheme was affecting commissioning of a brand new fleet of 'safer' trains - required by legislation to be in service by next year (see box).
If Mitchell had been asked the question just five months ago he would have been forced to say the power project was only 20% complete. This week, however, he could report a remarkable advance and say the job is 55% finished and running faster than the new trains can be put into service.
And he could point out that the cost of the power work has been reduced from £1bn to £700M with the price still falling.
Such progress might look like brinkmanship or last minute panic. But what has got the Network Rail power supply upgrade (PSU) project director and his team to this point is classic project management.
The electrical works themselves are relatively straightforward, Mitchell says (see box). But carrying them out across 5,114km of track, on a live railway when the final scope of the project was only agreed in November, has been a challenge.
Mitchell held his nerve at Christmas when others, looking at virtually no starts on site, were less confident. He insisted that the pre-planning, offsite trials and prefabrication of new sub-stations would all pay off. So far he looks to be right.
The most difficult task when the project was formed in January 2002 was where to begin, given that there had yet to be a decision on the full extent of the works. 'There was a whole lot of discussion over the business case of works to do and not to do. But while that was going on it was obvious to me and the PSU team that the routes running the most trains would have to be upgraded - the Brighton and Portsmouth main lines - so we decided to start there.'
Experience gained on those lines could then be rolled out across the whole project.
'We put a lot of effort into making the work 'plug and play', ' Mitchell says. Substations have been prefabricated in stainless steel for a longer life and to speed installation.
'We've taken as much of the wiring as we can into the factory.
It means we don't hit resource constraints because we can spread the work around the country; it's away from the railway so is safer and cheaper to do; and everything is factory tested so it arrives on site precommissioned.'
When the project parameters were settled it was decided to develop a system to support all existing train routings. The gold tap version of this would have cost an estimated £1bn.
'Apart from the unaffordability of it, we couldn't have built it in the time, ' Mitchell says. 'So we reviewed the design. First of all we had to establish exactly what the existing kit could take.'
The original plan had anticipated the old equipment being able to supply 70% of the new power demand. 'We built ourselves a high voltage test facility, ' Mitchell says, 'by taking bits of existing kit out of service and trying it out. We found we could push it harder than we first thought without shortening its life - 10-15% more in fact.'
Mitchell continues: 'The thing we did which made the most difference was telling the designers that rather than designing to guarantee power in all situations, they were to design a system as lean as they felt prudent.
The understanding was that once the trains were all running, if we'd under-cooked the system anywhere, we'd go back and put that bit right.' The sort of issues they had to debate were whether the chance of losing the air conditioning on the trains for five minutes once every 10 years was worth spending £3M to prevent. The answer was no.
To refine the design Mitchell and his team used Network Rail's Vision software model.
This is a virtual train set, running all the trains in the region to timetable. They then added a piece of software called Oslo which mimicked the power demands, and tested the network under the old power demands and the new to see where the blackspots were. The team then added in the impact of 'train perturbations' - signalling problems, leaves on the line - and various other options for a bad day on the railway. The subjective debate was just how bad a day do you design for?
'We asked the designers to stand as close to the edge as they could and design for a reasonably distressed network, ' Mitchell says.
The underlying assumption was that 2% of new train delay minutes could be down to electrical failure - the same as before.
'All this brought the price down to £709M. By continuing to use a more realistic approach to risk the ambition is to get below £600m and we expect to do that by some way, ' Mitchell claims.
The outcome is 165 pieces of work to be done. These have been bundled into four packages, being dealt with by four different contractors (see box), although that is due to be reduced by the end of the year.
'Of 12 super-critical sites we have completed eight; with 43 critical sub-stations to commission in the next 28 weeks. The whole job is due to finish in June 2005. If we hold to programme, power will never be a constraint on the network, ' he says.