Closer running: magic capacity potion or poison?

Author: Hongsin Kim
Co-author: Felix Schmid, Philip Dakin, Shuxia Lu
Day: Aspect Day Two
Session: Performance

The low adhesion at the wheel-rail interface is a fundamental characteristic of railway systems, leading inevitably to long stopping distances. As a consequence, traditional fixed block methods of train control prevent railways from using the infrastructure fully. Increases in demand for railway journeys and the difficulty of adding infrastructure has led to a philosophy that enables trains to be separated by less than a full braking distance, similar to motorway driving. This concept is based on the premiss that it is unrealistic to assume that a train stops instantly and, therefore using the full braking distance in a system design is an overly cautious approach. It offers potentially greater capacity but there are concerns that a catastrophic event would occur when the preceding train is stopped instantaneously. Technologies exist for these systems. However, it has not yet been applied to operational railway systems. According to IEC 61508, a safety analysis should be performed early in the system life cycle to reduce systematic risks and reduce cost. In the paper, the authors aim to review critically the research into closer running of trains and conduct a safety analysis of the approach. To investigate the increase in railway capacity, the authors simulated a scenario of two consecutive services diverge from a line and making station stops of 90 s. The model indicated a capacity improvement of 14% over ETCS Level 3.For the safety analysis of the system, the authors compared two approaches: (1) event chain accident causation model based safety analysis methods which include traditional tools, such as HAZOP, FMEA and FTA and (2) a systematic theory accident causation model, which is relatively new. They summarise the advantages and disadvantages of applying this novel approach.For approach (1), Event Tree Analysis and Boolean algebra were applied to quantify the probability of an accident using data from industry statistics and reports covering a 10-year period. The event tree analysis indicates that the probability of a passenger being injured due to a train stopping at an unacceptable rate is 1 in 1.23 million journeys. This falls into the regions of negligible and tolerable risk. Investigations into historical incidents reveal the following as having the potential of decelerating a train rapidly: fallen trees, road vehicles, cattle, engineering work and track plant, most importantly, avalanches, landslides and debris. Thus environmental factors present the greatest hazard. Climate change increases rainfall severity increases the risk of landslides.For approach (2), he authors extended the Unified Modelling Language into a sequence diagram meta-model for STPA, to analyse the safety of the ETCS level 4 system. The authors identified hazards in operational scenarios, built the associated model, and discovered unsafe control actions (UCA) that could lead to hazards. For each unsafe control action, the authors further identified the causal scenarios that lead to the unsafe control action and propose mitigations or solutions. The systematic theory accident causation model based methods are applied to only one hazard analysis. Further research is therefore required to adopt the method for safety analyses.