For many years, I have been working with and for safety on level crossings. The other week, however, I found myself in the local newspaper (front page and all…) arguing against a recently implemented safety measure. Paradoxical, or not? First, some necessary background information…

Level crossings

Level crossings are places on the railroads that generally have a rather high risk. Here is a possibility that common road users (cars, busses, cycles and pedestrians) meet a train. Trains contain a lot of energy (mass and speed!), cannot steer away (being ‘locked’ to the rails) and take a long distance to stop. In many cases, the meeting of train and common road users is fatal, or at least highly damaging, for the latter.

For this reason, often safety measures have been implemented, like automatic barriers, warning lights, signs and the like. Safety measures will vary, depending upon the situation. In Norway, for example, all level crossings with public roads have to be secured, while the smaller infrequently used level crossings in the woods or rural areas will have significantly less safety measures. Locations with the highest risk or locations that provide special problems are preferably removed, by building a fly-over or culvert and making it a crossing in different levels. Separating various modes of traffic in space is better (but more expensive) than building barriers.

The location

When you leave the station of my hometown and travel north, after about a mile you will meet four unprotected level crossings connected to some farms beside the train line. If you try Google Maps and put in the coordinates 59.565506, 11.297634 you get an idea of the situation.

The terrain is relatively flat, but slopes on one side and there are some curves. Vegetation can be another challenge that disturbs the line of sight. Most of these level crossings are used very infrequently and are even closed with a fence that the user can open if necessary (for example, when a farmer has to get to the fields on the other side of the tracks). One of the level crossings is clearly in daily use for farm traffic and people who live in the woods a few hundred meters from the tracks. All level crossings have “Stop, Look, and Listen” signs.

It is safe to say that the situation has been like this for many years, decades probably, and the people who live there and use the level crossings are very familiar with the situation (I actually spoke to one of the farmers). Save for the odd visitor, I dare say that only people with experience and knowledge of the situation use these level crossings. The chance for an unfamiliar random passer-by is remote because the roads are literally going nowhere.

Another safety measure

During a recent upgrade/maintenance on the line, someone noticed that none of the unprotected level crossings on the line had a sign that ordered the train driver to blow the horn to signalise “Train coming”. The infrastructure manager decided to correct this non-compliance. From that moment on, train drivers did as they were ordered and created a decent amount of noise in the wide area that previously had been quiet.

Blowing horns is a safety measure with a long historic tradition. From their early beginning, locomotives have been equipped with loud horns or bells to warn vehicles or pedestrians that they were coming. Steam locomotives had steam whistles, the later diesel and electrical locomotives got specifically designed train air horns. Train air horns are significantly louder than their counterparts in cars and trucks are.

To reduce the impact on people living nearby, train drivers have to blow the horn only between 06:00 and 22:00 - apart from emergencies (like people or animals on the tracks), of course. The problem is that even by restricting the honking to these hours, there still is noise at least 4 times per hour. There are people living within 100 metres of the place where the trains blow the horn - with NO sound-reducing barriers in-between. There is also a nursing home close by.

Additionally, even though rules try to stipulate how the horn signal is to be given (and I assume train drivers are trained about this), not all train drivers do it the same way. Some blow the horn significantly longer, earlier, or louder than others, creating more nuisance than others do.

Some critical questions

I have not been part of, or seen, the assessments that were part of the decision to putting up signs that are causing the noise. When I filed a complaint, the answer I received referred to compliance to some rule. And I presume (but again, I can be mistaken) that this is as simple how the process went: a non-compliance was identified and without much thought a process was started to fix the problem. Undoubtedly with the best of intentions and the assumption that it would improve things, but I doubt that real risk-based thinking was part of the process.

And while I wasn’t part of the process, I can ask a couple of critical questions:

1: Management of Change?

I wonder if someone has assessed the effects of the change. It looks so simple and straightforward, but is it really? One may think that the more safety measures the better, but believing that implementing safety measures is only positive or neutral is a fallacy. Neither do more safety measures automatically mean lower risk/safer.

Just some points for consideration:

  • More often than not, new measures create new problems. In this case among other things noise.
  • The frequent use of sound signals may cause desensitisation. After a while, people may get used to it and no longer react to it. The warning signal thereby loses its function because it is overused. Kind of the “Cry wolf” effect. Experience shows that people do not react (or only badly) when an alarm goes off in a shopping centre and offices (“Oh, it’s probably another false alarm”), and the “Don’t leave your luggage unattended” messages on airports have become part of the background noise, nothing more.
  • It can lead to changed behaviour, for example that people start trusting on the horn signals and stop looking out for themselves, leading to risky situations when a train driver forgets to give the signal, or outside the period of 06:00 - 22:00.

2: Suitable and effective measure?

While the measure has a long tradition and is implemented as a standard solution, I wonder if it is a solution that suits our modern time at all?

These days, people walk around with their iPods or mobile phones and headphones all the time, or have loud music in cars. On the locations we are talking about here, there are probably tractors that are pretty noisy themselves and some farmers use hearing protection while operating them. So it's doubtful that the warning signal reaches its intended 'audience'.

We should also consider how honking is perceived in normal life. While we cannot fully compare road and rail traffic (for example the meaning of a ‘red signal’ does not have the same impact), the function of using a horn is generally as a warning signal in extraordinary situations (although I know that drivers in some ‘southern’ or ‘eastern’ cities appear to use it as a means to signalize that they participate in traffic). So shouldn’t be using the train horn also be reserved for extraordinary situations, like when there are people on the rails when the train approaches, or possibly when there is fog and visibility is bad?

Using a danger/warning signal as a constant signal means that its effect wears off very quickly, leading to the above mentioned desensitisation.

3: Gain or pain?

One may wonder what the cost/benefit looks like in this case.

Typically these cost/benefit assessments only look at safety and money. I presume that the process (if there was any, because usually they are done very implicitly, especially when triggered by non-compliances) went somehow like this: There is an additional safety measure, so we assume safety improves. The expenses are relatively limited for the infrastructure manager who has to put up a couple of signs (although you may be surprised about the total costs if you factor in the design, planning, updating of maps, etc.) while the cost for the railway undertaking is nil (train drivers have to do their job, and that they do already).

However, this look at cost/benefit is too limited. Besides the elements safety and money, also other factors should be included in the assessment, like efficacy (the assumed safety gain may not be there at all) and the effect on other areas than safety (like environment and welfare in this case). One will also notice then that we are dealing with various, non-overlapping groups of stakeholders that all have different costs and benefits. Safety gain (or loss) concerns passengers and personnel in the train and road traffic using the level crossing (gains for these two groups are different, mind you!), cost and compliance concern the infrastructure manager and railway undertaking while environmental and welfare aspects concern the people living in the area.

Dealing with risk means making trade-offs between various aspects. Even if safety would improve thanks to the signalling with the horn, someone else pays the price. What is more important? Quite importantly also, who gets to decide? Were other stakeholders, like the neighbours or the municipality, even involved in the decision? Or, were they only confronted with the results? (As far as I know the answers the answers to the last two questions are No and Yes, respectively).

I am not necessarily a fanatic adept of utilitarianism, but I am rather sure that Bentham would not approve of this. To me it does not sound like the solution with the greatest gain for the greatest number.

What can we learn

Let me first say very clearly that I do not know if the implementation of this ‘safety measure’ was wise. Answering the questions raised above might help determining that. And there are surely other important factors not mentioned here that should be considered too. For now, I am very sceptical.

Some may wonder what the fuss is all about. Why make safety so complicated. Simple answer: because it already is. I do not make it so. In fact, I think that this case illustrates very well that safety is often way too complex to dumb it down into compliance with a rule and implementing a simplistic standard measure. Other lessons might be:

  • One size usually fits no one really well.
  • More is not always better. It is not always a matter of the old Dutch saying of “baat het niet dan schaadt het niet” (“if it doesn’t help, it doesn’t do harm”), because it just might do harm.
  • Implementing safety measures will change the system that can (will) lead to changed behaviour, which may actually lead to a reduced level of safety.
  • Compliance is not the same as safety.
  • Talking to various stakeholders will give you a richer picture of the problem, and lead to better (more robust) decisions.


Also published on Linkedin.