Black Saturday

The popular press loves bad news. So, when Network Rail deployed 11,000 people over Christmas to work on 2,000 sites around the country, the story wasn’t about the 314 projects that were handed back on time, it was about the eight (2.5%) that weren’t. Or, more exactly, it was about two of the eight that weren’t.

That seems like a very small number, but those two delays did shut two of London’s main termini for a day.

Oops!

Unsurprisingly, everyone took this very seriously. Network Rail certainly did, with apologies from Mark Carne, chief executive, who said: “While we have completed a huge amount of work across the country which will improve millions of journeys, the last few days for many passengers have been miserable and again I apologise for the disruption this caused.”

Mr Carne also pledged that an internal report on what went wrong would be prepared – and published. That duly happened. Written by Dr Francis Paonesssa, the managing director of Network Rail Infrastructure Projects, it was published on Monday 12 January.

Mark Carne and Robin Gisby, managing director of Network Operations and the duty director over the Christmas holiday, then met with the Commons Select Committee on Wednesday 14 January. And to cap it all, Rail Engineer was in Francis Paonessa’s office on Thursday 15 January to find out for itself what had happened. It was a tough week for Network Rail’s board.

It must be noted that, although there were two major problems, each of which closed a London terminus (King’s Cross and Paddington), both the projects and the problems were entirely different. There was no connection or even similarity between the two.

Secondly, Dr Paonessa’ published report is remarkably frank. It goes into detail on the problems that occurred, the timescales involved, and the steps that were taken at the time to try and get the projects back on track. There was no attempt at a cover up – everything was laid out for public inspection.

Advance planning

All of the projects that Network Rail and its contractors undertook over Christmas, and there were 322 of them spread across 800 possessions and around 2,000 worksites, were meticulously planned. Every one was also assessed to estimate the likelihood of delays occurring and completion being delayed. Every one had to have at least a 90% chance of being completed on time.

The 59 most critical projects were assessed again, and the most important of those had to pass a 95% threshold. Both of the projects in question were assessed to the 95% level, and both passed.

As it happened, just eight were delayed out of 322 – that’s 97.5%. So, in that respect, Network Rail’s planning worked. It was just Murphy’s Law that two of the ones that overran affected so many people.

King’s Cross / Holloway Junction

Holloway Junction is approximately 1.5 miles north of King’s Cross station on the East Coast main line (ECML). There are four railway lines at Holloway, interconnected by four junctions to allow trains to pass from one line to another.

Two of the junctions and 500 metres of the two railway lines between them were being replaced between Christmas Day and Monday 29 December 2014.

Although this type of renewal is a relatively routine operation on the network, it was large in scale, involving the replacement of 6,000 tonnes of ballast and so was a significant logistics operation. Two other adjacent railway lines were needed for the engineering trains which supported the work.

In parallel, there were a significant number of other engineering works being undertaken between King’s Cross and Peterborough which closed the ECML in multiple locations. As a result, it was not possible to bring the engineering trains straight to site from their depots.

In total, 14 engineering trains were required to support the project at Holloway and they had to be delivered to site ahead of the work and parked up. The large number and size of these trains made the total worksite around nine miles long in order to accommodate them safely and to allow them the space to move in and out of the construction area in a carefully choreographed sequence.

The optimum engineering plan would have required the closure of all four railway lines at Holloway in a seven day blockade and to renew all four junctions and all four stretches of track at one time. This would have been cheaper and would have had inherently more flexibility in the engineering plan, lowering the risk of an overrun. However, because of the significant disruption to passengers resulting from a seven-day blockade, this strategy was not supported.

The next best plan was to have two four-day blockades and to undertake the renewal in two halves, one at Christmas 2014 and one at Christmas 2015. However, the main junction at Watford on the West Coast main line was also being renewed over Christmas 2014. This would have left two of the key North/South routes closed at the same time, which would again cause significant disruption to passengers.

In discussion with the train operators (TOCs), it was agreed to keep one Anglo-Scottish route open to a London terminus and therefore the plan at Holloway was staged in such a way that two of the four lines could be re-opened on Saturday 27 December, enabling a reduced passenger services to run.

To be able to execute this plan, various parts of the project had to be undertaken in parallel rather than in series, reducing flexibility. Whilst the risk of an overrun on Saturday 27 December was now higher, the project was risk assessed using the standard industry tools and still passed with a 95% likelihood of completing on time.

To minimize this risk, various steps were taken. Seven road-rail vehicles (RRVs) were sourced from an approved supplier to ensure that there was minimal risk of mechanical failure. The supplier of the plant also provided eight new log grabs to reduce the risk of key plant failures – one for each machine and a spare – and provided an on-site fitter so that any equipment problems could be fixed quickly.

A contingency allowance was built into the plan so that any unforeseen slight delays would not affect the overall timetable. The plan was therefore as robust as Network Rail could make it.

Cumulative delays

However, delays occurred right from the start of the works. Approximately one hour was lost during the process of isolating the 25 kV overhead lines and then issuing the required permits to work before the site work could start.

“We recognize that taking possessions generates quite a high workload on our electrical control centres, so we had already taken the measure to make sure that Holloway was number two in the queue so that risk was minimised,” Francis Paonessa commented. However, it seemed that didn’t help.

Approximately three hours were then lost during the process of ‘scrapping out’ during which 500 metres of old track and sleepers were flame cut, dismantled and loaded onto engineering trains marshalled on the adjacent track. There were a number of reasons why time was lost, including some due to machine operator inexperience, but the primary reason was that the fittings between the RRVs and the brand-new log grabs kept leaking hydraulic fluid, losing pressure and not working correctly. The on- site fitter was constantly working to fix these problems and delays mounted.

The plant supplier had provided new log grabs to specifically reduce the delivery risk. However, the unintended consequence was to introduce a delivery risk because the grabs had never been operated with these specific RRVs.

On top of that, one of the seven RRVs failed and was beyond the capability of the on-site fitter to mend, so an off-site specialist was called to site. It was 3 hours 35 minutes before that machine returned to service.

Network Rail is getting very hot on plant reliability, so it was unfortunate that this particular machine failed at such a critical time. “We’ve been doing a large piece of work on RRV reliability over the last couple of years,” Francis admitted. “They still fail, but we are in the middle of a scheme of work to drive that reliability up.”

All this meant that, at the end of scrapping out at approximately 09:00 on Christmas Day, the project had lost four hours. Less the 45 minutes of contingency built into the final plan up to this point, and the work was 3 hours 15 minutes behind plan.

The project had now reached the ‘point of no return’. A key decision had to be made on how deep to excavate the ballast. The contingency plan allowed for the project to be up to four hours behind at this point and still deliver the full 300mm deep ballast excavation (around 6,000 tonnes of stone).

Significant time could have been saved at this point if the ballast was purely ‘skimmed’ rather than fully excavated. However, that would have significantly reduced the expected life of the new crossings and track, from a typical 25 years to perhaps only 10 years or less.

“The whole point of doing this job was to fully renew the track and the S&T so the plan was always to maximise the amount of ballast we could take,” Dr Paonessa commented.

“The contingency plan said if you’re up to four hours behind at that point you carry on with the full dig, because we think we’ve got enough time left. So the site team made exactly the right decision based on the contingency plan that they had and they carried on with the full dig. We’d already used up three and a quarter hours of the overall contingency but we had other slots of contingency built in.”

Once that decision had been taken, to go for the full 300mm deep excavation, the die was cast. It had to be completed to that depth along the entire length of the track otherwise it could introduce serious track quality issues.

Shortage of drivers

Shortly after the decision was made to proceed to the excavation phase, two engineering trains, which were loaded with scrap rails and sleepers, were due to leave site. Some of the scrap was not correctly positioned to safely travel on the open railway and had to be adjusted. By the time that this was completed, the drivers of these trains had reached the end of the limit for the length of a shift and were unable to take the trains to New Barnet, approximately seven miles away. This seemingly minor problem proved to be the catalyst for a major issue, though this did not materialise for a further 12 hours.

The drivers of two of the spoil wagon trains (used to transport the old ballast) were cascaded forward to drive the scrap trains away. Moving drivers from one train to another is not unusual; however, the engineering trains were spread out over the nine-mile worksite so time was lost with each driver move.

The site was now two train drivers short and continuing to gradually lose time. By 14:00 on Christmas Day, the project was approximately six hours behind the plan.

The programme delay had now been escalated to senior management, but it was still felt that the Saturday hand- back of two railway lines was possible, although tight, because of the contingencies built into Boxing Day.

However, the project had been cascading train drivers throughout the day and, shortly after midnight, the supply of new drivers to support this cascade ran out. There was one remaining driver and five engineering trains still on-site and this was the point at which the project started to rapidly lose time. Whilst all the drivers involved were cooperative and committed to completing the project, they had reached their maximum shift duration limits, which for safety reasons cannot be exceeded.

The huge amount of construction activity around the country over Christmas required the support of over 200 engineering trains. This demand exhausted the national supply of freight train drivers and some planned work had already been cancelled in the run up to Christmas because it could not be supported by train crew. The train plan for each worksite was carefully constructed but, due to the constraints of driver hours and shift patterns, could only cope with a finite amount of project delay. After this point, the work activities and the engineering trains to support them become out of sync and this is what happened at Holloway during the early hours of Boxing Day.

“When it got to about 1am, we hadn’t finished the heavy dig and we only had one driver left on-site to move the five trains around. Then we were really in difficulties,” Francis Paonessa admitted.

“We’d made assumptions that we would be digging out at a certain rate, but we were not achieving it. The train wasn’t where it’s supposed to be and you can imagine the difference between just scooping up a bucket of material to tip into the train and having to trundle backwards and forwards to where the train was. It was in that period overnight and into the morning, when that logistics plan and the work on the ground was completely out of sync, that we made virtually no progress.”

This was compounded by a mechanical failure of one of the ballast wagons which failed in such a way that it could not be moved for a number of hours.

On the morning of Boxing Day, senior staff sought to identify a solution but, by 11:00, the project was around 15 hours behind the original plan. Network Rail IP Tactical Control and LNE Route Control were made aware that an overrun would occur on Saturday 27 December. The overrun was also escalated to Network Rail’s managing director Infrastructure Projects and chief executive.

By 13:00 on Boxing Day, a revised plan had been put together which showed that the railway through Holloway to King’s Cross could not be safely handed back until Saturday night. This was after the time when any passenger trains were planned to be operating so a 24-hour overrun was therefore declared. This revised plan had two through lines being opened for passenger service on Sunday 28 December at 05:30 and all four lines as originally planned at 05:30 on Monday morning.

Don’t forget the passengers

So what to do about the 36,714 tickets that had been booked in advance for 85 trains planned to run into and out of King’s Cross on Saturday?

“We have a number of standing plans that sit there in any case of major problems,” Francis Paonessa said. “As you can imagine, our teams are dealing with service incidents on a daily basis across the country. All sorts of things are happening from animals on the track to a tree coming down to a points failure or signal failure; so the route teams are very familiar with generating service recovery plans quickly.

“Alpha 1 plan is the one for King’s Cross so, if for any reason King’s Cross is out of action, whether it be a fire or a bomb alert or something, the standing plan that we have, and the one that we adapted on Boxing Day, utilises Finsbury Park.”

Finsbury Park is two tube stops north of King’s Cross. London-bound trains would arrive at Platform 4 to allow passengers to disembark onto an empty platform, and then be shunted across to Platform 5 to pick up passengers before departing for destinations north of London.

The decision to run trains had to balance the disruption that would be caused by running services terminating at Finsbury Park against the even-greater disruption that would have been caused by cancelling services altogether or directing them to other destinations more remote from King’s Cross.

At the start of services on 27 December, local trains began to arrive (southbound) and depart (northbound) with some using Platform 4 in both directions. Around 100 had arrived, departed or passed through by 10.00 with average delays of around 10 minutes.

At this stage, it was agreed locally between the station staff and the King’s Cross signal box that long-distance services would also arrive and depart from Platform 4, whereas the plan agreed by all parties the previous night was that they would arrive at Platform 4 but depart from Platform 5. As a result, passengers were unable to get off London-bound trains onto Platform 4 due to the platform already being occupied by northbound passengers.

This error contributed to overcrowding in the rest of the station and understandable passenger confusion.

This was corrected after the third London-bound train arrived at Finsbury Park around midday following the implementation of a passenger flow system at the station.

A further 106 trains were operated between 10:00 and 17:00 with average delays of around 37 minutes, although a number of trains experienced delays of up to two hours. While the vast majority of passengers were able to board a train to their destination, a reduced service meant that trains were cramped with many passengers forced to stand for all or part of their journey.

Passenger services to and from King’s Cross were resumed on Sunday 28 December on two lines and all four lines were opened, as originally planned, for services on Monday 29 December.

A number of immediate lessons have been learned. These include a requirement for plant operators to test all new equipment before it will be allowed to be used on the railway and that engineering train crew and contingency at times of peak work will be treated with the same level of nationwide cross-project scrutiny and planning as other resources in short supply, such as signal testers and overhead line engineers. No doubt more changes will be implemented after a thorough examination of everything that occurred.

Paddington / Old Oak Common

In contrast to the catalogue of problems at Holloway, the overrun that closed Paddington was much simpler. Between 24 December 2014 and 2 January 2015, Network Rail undertook major engineering works across the 25-mile stretch of railway between Paddington and Maidenhead to replace ageing infrastructure and to upgrade the railway for future services.

The work to deliver this huge level of activity peaked on Christmas Day and Boxing Day when, as planned, all of the railway lines into Paddington were closed with over 1,200 construction and test staff, split into two 12-hour shifts, working on the infrastructure. Over the following days, between 27 and 30 December, around 350 to 400 people were employed per shift.

The work required the support of 18 engineering trains and over 200 construction machines such as 500-tonne mobile cranes, RRVs and piling machines. 10,000 tonnes of track ballast was removed and a similar amount of new ballast brought in. During this period around 2km of track work was completed and a further 2km of new track brought into use. 23 sets of points were commissioned and multiple new signalling systems were brought into service.

There were detailed contingency plans for all elements of the planned construction and signalling testing activities and these were enacted a number of times to bring the construction and signal testing activities back on schedule. This was achieved by a combination of providing additional resources or de-scoping non-critical activities and these works finished on time.

The contractor responsible for signal testing, one of the last stages of the project to be completed before the lines were returned to service, reported that its signal testing work was complete on the main lines at 03:30. At this point, the project appeared to be on time to hand back the planned railway lines at 07:00 to allow passenger services to commence. As it turned out, this report was inaccurate.

Incomplete validation

“The signalling testing and evaluation process is a huge job and it happens all the way through the commissioning,” Dr Paonessa explained. “It’s not something that happens right at the very end, there’s a massive programme of work all the way through that is the only thing that absolutely guarantees that the railway is safe to hand back on time. That process is led by the tester-in-charge of our framework contractor and it’s a difficult job, there’s no doubt about that.

“At that point, when that validation process was coming to an end, they believed that they had a couple of hours work to do to finish it off. In the process of doing that validation, they found that they had some additional work to do, some additional checks that they needed to redo. So we took back possession of the railway to go out and do those activities and they took a lot longer than we had originally thought and longer than they originally thought.”

At approximately 08:30, the Network Rail project team asked the tester-in-charge for a candid view of the issues and confirmation of what the revised timescales would be. The response was that the document for Old Oak Common would be issued at 11:15. Route control and the operations teams therefore planned on that basis.

Subsequently, the tester-in-charge found that the amount of work that still needed to be done was greater than initially expected. This came to light as safety validation checks on the paperwork were completed. The cause of this was a combination of physical testing work needing to be redone or rechecked and inconsistencies in the paperwork needing to be resolved.

Ultimately, the documentation was finally issued to enable the signalling and thus the railway lines to be handed back into use at 13:14.

An investigation is underway to ascertain exactly what went wrong with the testing process. Consideration will also be given to providing additional contingency time for the validation process where major signalling works or multi-disciplined works are being undertaken.

The effects on services

As soon as it was recognised that the overrunning engineering would affect train services, action was taken to implement the agreed passenger contingency plans. Reading station was running with only three through platforms on 27 December so many London-bound trains were turned at Swindon and Didcot to keep them away from Reading and avoid congestion. Passengers had to change trains to continue their journeys but, overall, this delivered more punctual/reliable journeys.

Passengers to/from London Paddington were generally diverted from Reading to London and vice versa using services to/from London Waterloo. However, a rugby match at Twickenham meant that neither additional services nor longer trains were available, which led to overcrowding.

Oxford to London passengers were diverted via Banbury and Chiltern services and this generally worked well, although the services were very busy with displaced passengers from the West Coast main line in addition to the normal levels of Chiltern passengers.

Because of these two problematic projects, passengers had been badly affected both at King’s Cross and Paddington, and that reflected badly on Network Rail.

“At the end of the day, the railway is here for passengers. It’s not here to be a railway,” Francis Paonessa affirmed. “It’s here to transport people around and we must put their needs and requirements first. It shouldn’t matter to the passenger how we’re making sure that they’re not disrupted, whether it’s through guaranteed delivery plans or excellent service, all that matters is that the trains run on time. We need to do what we say we will do which is to deliver an effective railway and make sure that it is there when the passengers need it. That’s our mission, that’s the mission we set out to do. It’s one that Mark Carne has said quite clearly that we’re not doing well enough.

“We need to do better and we will do better.”