Long blamed for train delays in Europe, autumnal leaves pose a danger on the tracks by forming a slippery coating that makes acceleration and braking more difficult. The leaves also interrupt the electrical conductivity between the train and track needed for the signalling system to work properly.
The London Underground's Central Line is particularly vulnerable, as around 70% of the 46 mile line is above ground, parts of which go through wooded areas.
Software engineers have created a program, in use for the last five years, to help predict when the leaf fall will be the heaviest. The software helps train drivers determine whether they can use a computer controlled system, Automatic Train Operation (ATO), which drives trains harder and faster.
ATO maximises acceleration and braking, which means more Central Line trains can run than if under manual control by drivers, said Gilbert Rowe, project engineer for the Central Line's Leaf Fall Mitigation programme, part of the Metronet Rail Group that runs much of the Tube.
The track must be clean for the trains to run in ATO, Rowe said. The software program, called the Adhesion Controller's Condition Assessment Tool (ACCAT), compiles a variety of data to determine track conditions.
ACCAT downloads regular weather reports, including rainfall and wind speed, along with data from moisture sensors on the tracks. Train drivers can also report conditions. ACCAT also incorporates information from an annual vegetation survey, which pinpoints what species of trees are located next to specific portions of track, Rowe said.
That information is important since different kinds of trees shed their leaves at different times during the fall. For example, oak trees line the tracks between the stops of Epping and Woodford on the north-eastern part of the Central Line, which means heavy leaf fall in early December, Rowe said.
The data and trends are incorporated into the software's logic, which uses algorithms to make a recommendation on risk.
Risk is defined using five colour coded levels displayed on a map of the track shown in the software program, a web application that runs through a browser, said Sean Alty, director and technical architect for SanDS Business Solutions, which designed the program for Metronet. Blue is the lowest risk, followed by green, yellow, pink and red for the highest.
The software's logic is engineered to reflect the trends noticed over years on the tracks. Strong winds tend to stir up leaves, which is a problem. Strong rain is better than light rain, since it clears the track of leafy debris, Rowe said. Light showers and wind mean leaves will stick more easily to the rails.
If conditions are poor, the software can recommended dispatching a special train that coats the track with sandite, a mix of sand and tiny balls of stainless steel. The sand gives trains traction, while the steel restores the conductivity needed for signalling. The software can also recommend that Tube drivers take manual control of their trains rather than using ATO.
Other improvements are on the ACCAT road map. Right now, the program divides the track into segments between stations, but engineers are considering breaking it down further into 250 metre segments, Alty said. Also, reports of track conditions from train drivers are manually entered into the software, and opinions on track conditions can vary according to who is reporting.
Eventually, engineers would like to automate the way those reports are entered into the software and make them more standardised, Alty said. Plans are also under way to upgrade moisture sensors from GSM (Global System for Mobile Communications) modems to 3G technology to transmit data, which would enable near real-time reports, he said.
Overall, ACCAT has enabled Central Lines train to run in ATO for all but two or three weeks of the year, improving the speed of service. "It's a predictive tool," Rowe said. "This leaf fall software is essentially the difference that allows us to continue running in ATO."
Metronet is considering testing the software on other Underground lines, Alty said.