[SystemSafety] Royal College of Paediatrics weighs in

Peter Bernard Ladkin ladkin at rvs.uni-bielefeld.de
Wed Nov 19 07:19:03 CET 2014


Continuing our critically acclaimed series concerning Guardian outtakes on UK child road safety:

http://www.theguardian.com/uk-news/2014/nov/18/children-doctors-20mph-speed-limit

Some numbers jump out. 25 children died as pedestrians in urban areas in 2011, compared with 311
adults. So child-urban-pedestrian deaths are (were in 2011) less than 1% of road deaths overall in
the UK. Some observations:

* The RCP is actually asking for a whole package of measures. There are 2000 child deaths per year
in the UK and they want to reduce that number. Urban road unsafety accounts for just 1.2% of that.

* The general "risk pyramid" for road unsafety in the UK has factors of around 10. That suggests
round 250 severely injured and 2500 lightly injured children, if the uniformity implied by the risk
pyramid carries through to specific categories (for which I have no evidence one way or the other).

* If we are talking around three thousand injured children per year in urban road accidents, that
does suggest to me that measures might be worth implementing, on the basis that simply exhorting the
kids to be more careful is unlikely to affect behavior or outcomes much. I would have thought
adducing those overall figures would have had more impact than just adducing the small number of deaths.

* Does anyone know what physical basis the "collisions at 20mph" / "collisions at 30mph" etc figures
have? Being hit by a car *initially travelling at 30mph* is of course different from a collision at
a closing speed of 30mph. Also, many collisions with an initial closing speed of 30mph would simply
not take place if the initial closing speed had been 20mph. The dynamics of the interaction must be
considered.

The dynamics of a typical collision are
1. Perception by one or more participants of a space conflict, an imminent collision (stages PA,
Perception and Attention, of a PARDIA sequence. "Perception" refers to circumstances being presented
to a sensory field; "Attention" to them being cognitively registered. Informal use of the word
"perception" commonly refers to both stages, but we found it appropriate to separate them for
accident analysis);
2. Latency between perception and action (representing RD: Reason and Decision; the cognitive
formulation of a course of action). In an unanticipated situation, this commonly takes between 1 and
2 seconds. (German traffic law says "1 second". In keeping with Napoleanic law, it is normalised,
and as so often the normalisation of the phenomenon is not bound well to the reality.)
3. The decision made on the action. For a more-than-two wheeled-vehicle operator, usually some
combination of swerve; brake; do nothing. For a two-wheeler, including bicycles, add lay-down.
4. The executed action (here the IA parts together: Intention and Action. They are separated in
PARDIA primarily to account for the phenomena in GLOC incidents, which don't occur in road traffic).

Relatively few collisions occur without a progression through these stages. The stage 4 events
change the dynamics of the collision: the incidence of the trajectories and the relative velocity.
And of course it is the dynamics of the collision which largely determine the damage, along with the
3D spatial configuration of the participant objects. These causal factors are constrained by the
initial conditions (trajectories before the PA of PARDIA; 3D spatial configuration of participants)
but I would propose that the initial constraints are not as determinant of outcome as traditional
approaches to road safety suggest.

In 2011, I considered motorway Auffuhr accidents, which have been occurring regularly (but
thankfully rarely) on motorways/freeways/autobahnen in Western Europe and the USA for at least 60
years, and are usually put down somehow to irresponsible driving behavior. They are in fact system
phenomena and this is easy to show: they result from rational decision making on the part of
multiple operator/participants with limited mutual communication - and a relatively rare property of
the environment. Indeed, I imagine similar conditions, decision making and behavior occur far more
regularly than the accidents. The crucial difference is the relatively rare environmental property.

Nevertheless, we may expect the accidents to carry on happening and the police and judicial
authorities in some countries continuing to seek out participants to blame. Yet systems analysis
shows this is a fundamentally mistaken approach. Systems analysis is a foreign discipline to
professionals in road safety. We could change that.

In 2012, I performed a qualitative systems analysis of same-direction spatial conflicts. It took me
at least a week just to derive an appropriate ontology. For example, it turns out that one doesn't
need 2D-continual trajectories. It suffices for the decision-theoretic analysis to consider
trajectories akin to specific configurations of railway tracks. The analysis gets quite complicated
(I didn't spend enough time to make it simpler) and remains incomplete. I should really get around
to finishing it.

Unlike some here, I regard road safety as the most pressing system safety issue most of us
encounter. There are system-safety issues with living in buildings, but these at least in Germany
are a tenth of the size - about 600 people die in building fires in Germany each year; 5000 or so
die on the roads.

(Apropos indirect-electrical-safety: about a third of fatal building fires are solely caused by
electrical faults. Let's say that's 600/3= 200 deaths. Whereas only 15 or so people die each year
directly from electrocution. So the real electrical-safety issue lies in implementing effective
arc-fault detection in building circuits. The detectors do exist, thanks to aerospace initiatives
which I looked at a decade ago, but an arc-fault-detection/protection device costs upwards of ten
times the price of the usual Class A residual-current + short-circuit + overcurrent protection,
which costs a tenth of the price. Nevertheless, I am thinking of spending the thousand or couple
thousand euros it would cost to equip all my building circuits with arc-fault detection/protection.
I do have pervasive smoke detection, and a climbing rope and good anchor on the top floor with the
bedrooms. I did check I can still rappel half a decade ago. But I don't insist that visitors
can....... The social cost-benefit analysis doesn't yet work for regulation: VSL is way lower. But
*my* life is worth way more to me than VSL.)

I should note, apropos some recent comments here, that many road safety issues are country-specific.
For example, the single most prominent factor in rural bike-car collisions in Germany is a specific
roadway configuration which doesn't exist in the UK. Also, there are many urban bicycle-pedestrian
conflicts in Germany from the provision of cycle paths on the sidewalks. That doesn't occur in the
UK either. Or in the Netherlands, where city cycle paths are separated from pedestrians, and often
from the motorised-traffic roadway. Many pedestrians, even those who have grown up with the system,
wander all over the German urban cycleways without looking. Some cities, such as Bielefeld, are
coming slowly to realise the consequences of requiring cycle paths be used (namely, the city assumes
some liability) and are relaxing the requirements.

Notice that those are all system-safety issues stemming from road design.

And none of any of that occurs anywhere in the USA as far as I am aware. A couple decades ago, I
once walked from a colleague's house in Palo Alto to the California-Street downtown area, a couple
miles of suburbs. I encountered *no one else* on the sidewalks the entire journey, until I was in
the downtown area. And the walking was uncomfortable because of the configuration of the driveway
ramps. Whereas here, people are continually using the sidewalk outside my house to get to and from
the bus stop and their houses nearly a kilometer up the road. If I come out just after 6 in the
morning to clear overnight snow, there are often already sets of footprints from people walking to
the first bus. And I live in a semi-rural village!

PBL

Prof. Peter Bernard Ladkin, Faculty of Technology, University of Bielefeld, 33594 Bielefeld, Germany
Tel+msg +49 (0)521 880 7319  www.rvs.uni-bielefeld.de






More information about the systemsafety mailing list