BRITISH train operator First Capital Connect (FCC) plays a vital role in getting London's commuters to and from work, carrying over 150,000 passengers each weekday on two separate routes, the Great Northern from King's Lynn, Cambridge and Peterborough to London King's Cross and Moorgate, and Thameslink from Bedford and Luton through the heart of London to Wimbledon, Gatwick Airport, and Brighton.

FCC operates more than 200 emus of six different types, ranging from the inner-suburban class 313 sets built in 1976-77 to the dual-voltage class 377s supplied by Bombardier in 2008-09 for Thameslink services. The Thameslink fleet is maintained at Bedford, while Hornsey in north London is responsible for Great Northern trains and carries out some heavier maintenance on the Thameslink fleet.

Keeping this busy network on the move demands high levels of availability and rapid response when a train failure occurs on the main line. A number of high-profile failures between 2009 and 2011 demonstrated that split-second decisions can cost hours, and FCC has been working to improve its response times to incidents by providing better information to drivers, allowing faults to be isolated and dealt with more quickly.

At the heart of this strategy is the Train Assisted Faults and Failures System (Taffs), a web-based diagnostics tool developed in-house by FCC and named after its creator, rolling stock inspector Mr Keith Taffs. The system guides controllers or engineers in contact with the stricken train through the most logical steps to fix the fault as quickly as possible. It is also used by depot staff to speed up fault-finding on out-of-service trains.

"Before Taffs there was no centrally-based system staff could access out on site or at the control centres," says FCC's head of engineering production Mr Jason Long. "So we wanted a one-stop shop that staff could look at regardless of their knowledge or experience that would give them a correct diagnosis of the fault in the shortest space of time."

Faster-fault-fixingThe Taffs home page links to three main options. The faults and failures menu provides access to the fault trees. These are step-by-step top-down deductive failure analysis tools that allow the controller to establish the cause of a particular problem by relaying questions to the driver over the phone. Taffs contains around 2000 fault trees, all of them written by Taffs himself, covering almost any possible fault on all six train types in the FCC fleet.

"With six train types, we knew early on that having a general fault tree would not be enough for ease of use and correct information to be relayed; it had to be location-specific," explains Taffs. "One fault tree can link into another and you can look back through the steps you've visited. The system has a report button so the user can contact the administrator if they feel a step is missing that needs to be included."

Each fault tree starts with a straightforward 'yes' or 'no' question. Photographs are included to show the position and illumination of indicators on the driver's desk, and these can be enlarged by the controller to give a clearer idea of what the driver sees. The controller can bypass certain steps in the fault tree if the driver has already taken them prior to contacting control.

Having identified the fault, the controller enters the Safety Management System (SMS) menu, which includes information on how to deal with the fault, including all the rules and regulations that might be relevant to the incident. All of the fault trees link to the SMS at appropriate steps and this section also links directly to the latest version of the Railway Safety and Standards Board (RSSB) rulebook.

If the train cannot be moved under its own power, the controller can access the assistance matrix which provides movement procedures for the driver, detailing coupling compatibility between the failed and assisting trains, driving position, maximum speed, and the required braking configuration. Instructions for the movement procedure can also be configured according to the direction of travel, telling the driver which cab should be used to move the train.

"We built in the assistance matrix because in the past we've had incidents where trains have been coupled but the train crew has been unable to establish whether the formation can run with passengers, what needs to be isolated, and what type of coupling is needed," says Long.

Naturally, developing such a comprehensive system of fault trees is a laborious task - in the case of Taffs it has taken 10 months of intensive work to write and input all the required information. But this is already being rewarded with operational benefits.

In full operating mode Taffs operates a shadow system which records every step the controller goes through while responding to an incident. Over time, this will allow FCC to review which steps in the process are statistically more effective in solving defects, allowing them to be brought forward to speed up the process. The logging system also allows the performance of the controller to be evaluated. Because all incident data is recorded, depot staff can filter information on defects according a variety of parameters including train, route, and date, showing the steps taken and decisions made by controllers.

"Taffs was conceived as a self-learning diagnostics tool that allows us to look at all the faults that have been fixed, the step at which they have been fixed, and optimise the fault tree accordingly," says Long. "The more it's used the more efficient it becomes, which means what is currently step 10 in a particular fault tree could become step 6 or step 1 in the future."

As soon as the controller enters Taffs, a clock appears on the screen which shows the operator they have five minutes to identify the defect before making a decision on whether the train should be removed from service. The use of this function depends on the traffic situation, and the controller may allow more time for fault-finding during quieter periods.

Help from staff

FCC was keen to include fleet management staff in the development of Taffs. "As an engineering controller based in the depot you do not see the bigger picture of what is occurring on the main line, so we wanted the ability for a service delivery centre controller to remotely monitor the situation in the control centre away from the depot," says Taffs. This means service delivery centre controllers with access to Taffs receive an alert when engineering controllers open a new fault tree and a further message when that fault has been dealt with.

"The system will really take off when we start to measure ourselves on delay-per-incident performance, which it was really designed for," says Long. "We estimate Taffs will reduce total delay minutes by about 5%, or 4900 minutes per year, and during 2012 it helped us to reduce delay minutes attributable to train faults by 46%."

Long adds that the data coming from Taffs reports could ultimately drive reliability modifications in the fleet, and the technology could have other asset management applications beyond fleet support in the longer-term.

Taffs has been optimised for portable devices such as smartphones and tablet computers, to allow use anywhere on the FCC route with a 3G connection to obtain required information on site. Currently the system is being enhanced so that anyone without a personal login to Taffs can have procedures sent to their portable devices by the controllers.

Long says the system offers numerous benefits to depot staff. "Anything documented in our SMS could run on Taffs and it means you don't have to remember anything, it's all there," says Long. "The biggest issue we have is convincing people to use it. If you have engineers who've been on the railway for a long time, often they know exactly what to do to fix a fault. What we want to get across is that we've put logical steps that tell us statistically where the fault is most likely to be, so we can reach the fault more quickly. We are working to embed Taffs into our engineering processes, but it takes time."