199411 – Milroy & Guglielmucci – Operational Effects of ATP Braking Algorithms

This paper is in two related parts.

In Sections 2 and 3 we briefly review developments in Automatic Train Protection (ATP) technology since 1988.

We then draw attention to benefits (not only those related to safety) which become available with the introduction of ATP technology.

These include increased capacity, an up-grade path towards full ATO, the use with manually driven trains of regulatory modules such as Automatic Train Supervision (ATS) which are normally associated with full ATO, and eventual elimination of the capacity restrictions and other costs associated with fixed trackside signalling installations.

In sections 4 and 5 we discuss specific research results. We analyse the conflicts that can occur when ATP braking profiles (trajectories in the velocity/distance plane that must not be violated on safety grounds) tend to be more conservative than actual train performance would allow. These tendencies are unavoidable because simple models of train performance, with worst-case parameters such as braking performance, are needed for safety verification.

We evaluate the target errors that occur as a result of small variations in relevant train braking parameters, and then consider the effect of the distributed mass of a train on variable gradients.

In Section 6 we conclude that while target errors can be quite large after ATP interventions are allowed to occur, use of the regulatory tactics discussed previously should avoid such problems except when there is a genuine safety hazard. Use of driver training simulators can help to avoid such errors.

We also conclude that the cost-effectiveness of ATP should be evaluated not only for its contribution to safety, but also for its ability to facilitate the regulation of traffic, so as to increase capacity in congested sections of the network, keep trains in their planned time-slots, allow for effective recovery from operating disturbances, and conserve energy.

The authors’ experience with ATP and. related systems has been primarily in the context of research and development, with the Scheduling and Control Group at the University of South Australia.

We have also been involved in some aspects of ATP algorithm development on a consultancy basis, and in recent months one of us (IPM) has been involved with the specification and selection of multiple-train simulation packages for modelling the operational performance of trains running under ATP.

We welcome the opportunity to discuss these matters with railway signalling engineers, many of whom are involved with railways already operating under ATP, or in the practical implementation of ATP development projects.