Originally published in Track & Signal January-March 2016. Written by Anjum Naweed, Principal Research Fellow at ACRI.

All images: Identifying the risk factors for SPAD – Images (PDF 574kB).

In the previous issue of Track & Signal, I described how train drivers viewed their “relationship” with signals and illustrated some of the psychological imperatives underlying the eponymous signal passed at danger (SPAD) failure mode.

I presented some research, funded by the Australian Cooperative Research Centre (CRC) for Rail Innovation, that set out to determine certain behaviour-shaping factors that affect SPAD-risk. The research examined this by observing how they engaged with train driving under different conditions and challenging scenarios of their design. In this follow-up article, I will draw on this research to describe some of the error-producing conditions for SPAD-risk. But before I do any of this, a little bit of scene-setting is in order.

So, traditionally, train drivers have always navigated railways by relying on a keen awareness of the route and “likelihood” predictions of future train state. (Fun fact: over the course of my research, drivers have referred to themselves several times as “crystal-ball gazers”.) Obviously, these predictions rely on having a good knowledge of the track. This is called ‘route knowledge’, and in practice comprises both static and dynamic aspects of memory. As you would expect, this knowledge includes a variety of individual and external factors, such as an accurate understanding of train handling characteristics, foundational rules and the position of signals.

Train driving is characterised by a need to sustain attention for tremendously long periods, which increases the drivers’ vulnerability to ‘disturbances’. The key point here is that route knowledge becomes less reliable and information such as the recollection of the previous signal aspect can simply ‘fall behind the fridge’. The error-producing conditions I refer to are those that occur when the driver is fundamentally distracted, not by things like talking on the phone or daydreaming about what they will have for tea but by factors that are clearly task-related.

The plot thickens
Few rail organisations tend to recognise things that are fundamentally task-related as being a contributor to error-producing conditions for SPAD-risk, it alone a cause of distraction from the task in the first place. A good example of such a distraction is the pressure to perform. Under the grip of pressure, attention can easily be allocated to one aspect of the task and diverted away from another task of equal or greater concern. In the industry’s defence, the nature of safety and performance regulation in train driving is actually quite paradoxical – conceptually, keeping time and driving safely is conflicting, and it can be difficult for the train driver to define how they should distribute their attention to maintain these goals.

And so, our research explored the factors, that contributed to SPAD-risk – but originally, we didn’t go out to search on the issue of distraction. We had a very generic approach that asked a number of drivers operating in organisations across Australia and New Zealand to simply invent challenging scenarios in focus groups. We reasoned that their assessment of risk and SPAD likelihood would be grounded in the same sort of cognition they used when “crystal-ball gazing” and so the rationale was reasonably compelling. Each driver created a scenario with felt pens and A3 paper using whatever drawing conventions they liked, and then shared them with the others. The real fun started when we began analysing them.

You see, even though we didn’t set out to look at task-related distractions, almost all drivers were saying the same things: distractions – and from task-related factors no less. The most common were time pressure, which featured in 60 percent of collected scenarios, and station dwells, which were in 50pc of scenarios. The consequence of distraction from these distractors increased risk likelihood, particularly when present with sighting restrictions, which actually featured in 80pc of scenarios. So, let us have a look at some concrete examples.

What did drivers say of time-keeping? “Well, that’s your job. As a drain driver your job is to get the train in on time.” Time keeping was considered a goal-directed activity but time-keeping pressure was described as a distraction.

The drawing in Figure 1 shows a section of railway with a train about to enter a caution zone. The next signal is set at danger and located on a blind corner. While this signal is restricted from view, the driver would know about its location from the route knowledge. They would also know that it would be set to danger, based on the aspect of the immediate signal. In this scenario, the driver experiences what they describe as a “loss of focus” from a radio call as they enter the caution zone, and they have a SPAD. Aside from the train, track and two signals, there is little infrastructure, thought the duration of the conversation is marked, and you will note a meticulously drawn vignette of the driver in the act of losing focus.

On the drawing itself, the driver also notes a “focus on quick turnaround of train due to timetable running late”. Given the propensity for acquiring and maintaining route knowledge, there is no reason why a SPAD would occur from line-of-site restrictions alone. Thus, I would argue that the main risk factor for a SPAD outcome in this scenario is time pressure, which drives the decision error to answer a call under unsafe conditions and as a consequence renders route knowledge far less dependable.

Half of all SPAD-scenarios collected were blind-corner-type scenarios, indicating the importance of route knowledge for overcoming sighting restrictions but also how brittle it was under certain conditions. Some train drivers were very fastidious in their time-keeping: “There are guys that really try to keep up time and … they will do anything to try and make up time. They will, you know, bend the boundaries.”

Over half of the scenarios also featured SPAD error-producing conditions on station platforms. The vast majority of these occurred on departure and also involved time pressure: “Drivers accelerate away from platforms, trying to maintain a schedule.”

The drawing in Figure 2 illustrates a train waiting to depart the platform. In this scenario the driver closes the doors and departs when station staff announces that the train is ready to leave, instead of departing at the signal’s authority on a proceed aspect. The gravity of this error has been emphasised by a near-collision with an express passenger train on a parallel line and the proximity of a rail level crossing. Additionally, the train stop mechanism, which would detect the false start and automatically trigger the brakes of the departing train, has been located near the crossover, allowing the train to attain a faster speed before it is arrested. This SPAD is also unintentional, but did more than time pressure contribute?

I would argue that a key risk factor for the SPAD in this scenario is station-dwelling – not in the sense of the slack built into timetables but the experience of staying longer at a station than deemed to be necessary. A station dwell gives rise to feelings, anxieties and perceptions of workload that transcend the confines of usual scenario pathway. Ordinarily, the driver would complete platform work (assisted by the train guard if present) and then depart when the signal is clear to proceed. In the scenario given, the driver misses the last step altogether and departs without signal consultation. The inattentiveness is rooted in the disengagement from driving and distraction from the station dwell.

Train drivers also indicate that time pressure, or, alternatively, the motivation to avoid time pressure plays a part in premature departure. Station-dwelling is described as an anxious station – on the one hand, participants are relieved not to be driving but they also feel ill at ease from a compelling need to keep moving.

It would be remiss of me not to finish this article with something more on sighting restrictions. Many scenarios converged sighting restrictions with time pressure and station-dwelling the drawing Figure 3 depicts the scenario of a train waiting to depart a station.

In this scenario, the driver arrives on the caution signal, which means their next signal (identified in the drawing as ‘B’ and located on a blind corner) is set to danger. The driver misreads signal ‘D’ as their own, which is set to a proceed aspect for the other bit of line, departs and – lo and behold! – has a SPAD. Originally, the driver who invented this scenario highlighted the tree as the main contributor to the SPAD and proposed a mitigation strategy to “cut it down”.

However, the other train drivers in the focus group suggested the tree was not the issue and the driver should know from route knowledge that the signal visible from that side of the platform is not their own. As a consequence, the driver added the notation that they were “distracted at station”. The temptation to misread the signal and power away from the station instead of pulling away at caution was also attributed to time pressure.

It is important to note that errors reading across to other signals do occur in real-world situations, thus the physical contribution (in this case the tree) should not be disregarded. However, the scenario given in Figure 3 exemplifies more instances where route knowledge was less reliable and attention was diverted. In this scenario, the experience of distraction projecting from both time pressure and station-dwelling separates the driver from their route knowledge and the safe working requirement in the task. Thus, the dynamics of time-keeping and station-dwelling affects the ability to navigate the railway, particularly under the conditions where sighting is restricted.

In the brief error-producing conditions presented, what we have learned is that distraction during train driving can happen from a number of task-related sources. I’m hoping we’ve also recognised that for these error-producing conditions, SPAD-likelihood converges from a number of risk factors contemporaneously. In my preliminary work, I modelled this conceptually in three-dimensional form to convey the dynamism and temporal depth between the factors. I’ve been good enough to show you this in Figure 4.

The point I’ve been trying to make with this is that some SPADs escape single-factor (and ostensibly judgemental) accounts of failure. There’s more complexity in there than is immediately apparent, and it is important for organisations to be mindful of these issues, particularly when some of them may be of their own design.

References (for more information on this research)

Naweed, A, & Rainbird, S (2015). Recovering Time or Chasing Rainbows? Exploring Time Perception, Conceptualization of Time Recovery, and Time Pressure Mitigation in Train Driving. IIE Transactions on Occupational Ergonomics and Human Factors, 3 (2), 91-104.

Naweed, A, Rainbird, S, & Dance, C (2015). Are you fit to continue? Approaching rail systems thinking at the cusp of safety and the apex of performance. Safety Science, 76, 101-110.

Naweed, A, Rainbird, S, & Chapman, J (2015). Investigating the Formal Countermeasures and Informal Strategies used to Mitigate SPAD Risk in Train Driving. Ergonomics, 58 (6), 883-896. (Q1, 1.608, NP.)

Acknowledgements
The author is extremely grateful to the train drivers and railway and operator staff who helped with the research project.

Dr Anjum Naweed is a senior researcher at Appleton Institute for Behavioural Science, CQUniversity Australia, and a Principal Research Fellow at the Australasian Centre for Rail Innovation. His contact details are anjum.naweed@cqu.edu.au, +61 (8) 8378 4520.