VTI särtryck
Nr 227 0 1994
Workshop on Assessments
of Driving Safety and of Crashworthiness
Lennart Strandberg
Reprint from the proceedings of 26th ISATA,
International Symposium on Automotive Technology and
Automation, Aachen, Germany, 13 17 September, 1993.
Dedicated Conference on ROAD AND VEHICLE SAFETY,
pp 343 348
Väg- och
VTI särtryck
Nr 227 0 1994
Workshop on Assessments
of Driving Safety and of Crashworthiness
Lennart Strandberg
Reprint from the proceedings of 26th ISATA,
International Symposium on Automotive Technology and
Automation, Aachen, Germany, 13 17 September, 1993.
Dedicated Conference on ROAD AND VEHICLE SAFETY,
pp" 343 348
(db
Väg- och
transport-farskningsinstitutet
,
Author: Lennart Strandberg. Paper 93SF078: Workshop on Assessments of Driving Safety and of Crashworthiness. PJ (6)
Workshop on Assessments of Driving Safety and of Crashworthiness.
by
Lennart Strandberg
Swedish Road and Tra ic Research Institute (VTI)
prepared for
26th ISATA (International Symposium on Automotive Technology and Automation) Dedicated Conference on Road and Vehicle Safety
Aachen, Germany, 13th-17th September 1993
Call for workshop contributions from experienced delegates.
This session of the Safety Conference in the 26th ISATA will be held as a workshop. Each presentation will be followed by connecting discussions with the audience.
In addition to the scheduled papers, delegates are invited to prepare informal contributions, to which a few minutes each will be offered for oral presentation. The informal presentations will be selected from those proposals described to me (the chairman) personally at least three (3) hours before session opening. Due to time restrictions, priority will be given to examples explaining how oversimpli ed interpretations or biased risk assessments have misled (or may mislead) people.
Particularly welcome are cases where shortcomings in accident analysis have caused unjustified damage to the safety reputation of certain companies, products or services. Representatives of pro t making companies seem namely to be met with greater suspiciousness by mass media than what researchers from market independent organizations are, irrespective of their knowledge and experience. Car manufacturers, for instance, are often accused of being evasive, if they question the method in safety rankings where their own models appear dangerous.
Therefore, this workshop aims at an open discussion on facts and principles, without throwing suspicions on certain participants due to their affiliation.
Workshop Program and Objectives
In this Workshop past safety studies with strategic, controversial or unexpected results will be scrutinised. Between the presentations of scheduled papers, related topics will be discussed upon short oral presentations of selected contributions from delegates (not in proceedings, see above).
A major workshop objective is to elucidate the great in uence on safety ratings from the investigators' experience (both practical and scientific) as well as from risk perspective and preferences. Ratings may concern car models, vehicle type and parameters (mass, geometry, de ciencies, etc.), equipment (ABS-brakes, studded tyres, air bags, etc.) or education and supervision (safe-driving courses, vehicle inspection, etc.)
Relevance in safety research requires adequate experience.
To support open-mindedness and methodological rather than personal criticism, I intend to open the Workshop with two of my own research mistakes. These cases illustrate also how vehicle safety assessments may be misleading or irrelevant, if they are carried out without consensus from adequately experienced people.
In the first case I thought that my own extensive driving experience from rallies and racing circuits (during the 60's and 70's) was directly applicable to normal driving in road traffic. Then, I did not consider cars' directional stability important to traffic safety. On the racing circuit, steerability (controllability) is preferred before stability. Oversteering, instability and rear wheel skids are often developed intentionally to turn the car body earlier than its velocity vector. Therefore, I was not recognising stability then, as I do now in my subjective assessments of driving safety. Testing judgements by motor journalists may suffer from similar bias.
Author: Lennart Strandberg. Paper 93SF078: Workshop on Assessments of Driving Safety and of Crashworthiness. P.2 (6)
Today, I have learned from experienced driving school professionals, from analyses of fatal accidents (Strandberg, 1989a) and from observations of average drivers in experiments on ice (Strandberg, 1991b) that instability is a greater and more complex road safety problem than I realised with only my own driving experience.
In the other case, I analysed heavy vehicle safety from the opposite standpoint, i.e. without driving experience. I devoted years of research on lateral dynamic response to pure steering inputs, disregarding braking. See Strandberg et al (1975). I did not recognise the substantial braking problems until the mid 80's, when I completed my driver's licence for heavy truck and trailer, cf. Strandberg (1989b). Now, I do not deny that the variation of braking properties in the vehicle
population may be more decisive of the lateral dynamics in emergencies than the variation of chassis
geometry, which we analysed in detail during the 7O's.
Risk Zooming complementing or substituting Risk Homeostasis?
"To be conscious that you are ignorant is a great step to knowledge" (Disraeli, quoted from Stevenson 1948 by Sampson 1971)
Zooming in on and controlling a few risks, that one is well aware of, may result in ignorance or underestimation of other relevant hazards. Such risk zoom in among road users is likely to decrease the expected safety benefits of various accidents preventive measures.
While these effects are similar to those of Behavioural Adaptation (OECD, 1990) and of Risk Homeostasis (Wilde, 1982), does the risk zooming concept not imply that improvements in roads, vehicles or education lead to compensatory behaviour, which reduces or resets the safety margins to their former width. On the contrary, adequate education and ampli ed risk cues in the traffic system should help the road user to zoom out to a wider risk perspective - and to identify acute hazards requiring particular attention. See row 1 in Table 1.
Safety margins can only be controlled by a road user, who recognises the speci c hazards. Therefore, one might debate Wilde's (1988) emphasis on "motivational interventions that are effective in lowering road user's preferred level of accident risk". The risk homeostasis and adaptation models seem to be based on an oversimpli cation of the risk panorama to an issue of speed level only. (There are evidence of that many fatal head-on accidents have been triggered by sudden attempts to reduce speed.)
It appears more constructive to increase our knowledge on hazards than trying to prevent an assumed natural adaptation to some risk equilibrium state hinted by the term homeostasis. The safety potential of active risk zooming will be illustrated below with examples, common in the antecedents of fatalities.
The owner of a front wheel driven car has recently experienced front wheel hydroplaning with spinning wheels and loss-of-steering. If he buys two new tyres, he will consequently mount them on the front wheels, and keep the old ones at the rear axle. However, this will increase the risk of instability and severe yaw motions on wet roads, even if he drives at the same speed as before. See Figure 1, Strandberg (1989a) and UR (1989).
The driver of a heavy truck-trailer combination has recently experienced brake fading tendencies when driving with fully laden vehicles on hilly but non-icy roads. When arriving at northern Scandinavia he unloads the trailer, believing that the brakeability has improved. However, on slippery winter roads the stable braking distance may be more than twice as long as when fully laden (Strandberg 1987, Table 4).
Quite a few similar examples exist, since stability is impaired by many measures that improve brakeability and steerability. Nevertheless, this is only a fraction of the conceived risk panorama.
In addition, the relevance of different hazards varies between individual road users. Such tragic examples have appeared recently, where children have been killed by the front passenger airbag in crashes at low speeds (ViB, 1993). In the last year's Swedish public media campaigns for air-bags, few (if any) quantitative assessment of their safety promotion did zoom out to this risk perspective.
Author: Lennart Strandberg. Paper 93SFO78: Workshop on Assessments of Driving Safety and of Crashworthiness. P.3 (6)
Can safety be quanti ed in one dimension?
The examples above and in Table 1 show why oversimpli cations and premature generalisations may result in misleading risk assessments. Perhaps should all global 'safety' ratings be abandoned. It may be better to split evaluations into several measures of properties, which are closely related to
speci c risks with varying relevance to different individuals.
For instance, the crashworthiness of a car is often rated without consideration of its aggressive-ness to other road users than its occupants. This may lead to negative effects for society due to vehicle weight escalation through the buyers' demand of heavier cars. See rows 9-10 in Table 1. Also consumers have been unintentionally mislead by mass media reports on studies suffering from bias in data, from neglecting con dence intervals or from mass-signi cance (rows 4&6 in Table 1).
In certain evaluations of safety measures, failures to nd signi cant improvements are inter-preted as if there are no effects. Sometimes such "negative results (expression for non-signi cance by Stöttrup Hansen et al, 1990) are considered 'proofs' for risk homeostasis, thereby supporting a destructive scepticism against ABS and modern vehicle technology (row 3 in Table 1).
Similar results or even risk enhancements have been found in several evaluations of driver training programs, see Struckman-Johnson et al (1989). However, quite a few studies of this kind have been based on self-reported accidents and mileage. If such safety measures increase the drivers' disposition to report accidents, the real safety effects may be severely underestimated. For instance, Maycock et al (1991, p.19) found that one group of drivers forgot twice as many accidents as another (42% annually compared to 20%). One should also question conclusions such as "Mature Driver Course Fails to Lower Crash Involvement" (headline in HHS Status Review referring to Foster, 1991) when the actual investigation was made without data on mileage. See row 5 in Table 1.
Depreciation from such risk assessments has lead to opposition from experienced people in the car industry and in the safety education branch. The Workshop will aim at mutual understanding and effective measures, such as reinforcing driver education on the speci c limitations of ABS, skid-pad training, etc.
If conventional exposure measurements are used, cars with studs in their tyres may be consid-ered more dangerous than those without, since studded tyres have a greater percentage of their mileage on risky winter roads. Methods and perspectives from epidemiology may be more adequate here than those used in road safety assessments. American and Swedish case-control studies with police recordings from accident sites indicate that the relative risks induced by driver and vehicle properties may differ from common belief. See row 4 in Table 1.
Ignoring the safety-relevance of vehicle properties may lead to similar human losses as the neglecting of Semmelweis' request for better hospital hygiene during the 19th century. The association between lung cancer, asbest exposure and smoking is now well-known and accepted. Similar correlation and synergetic effects were found between accident risk and driver-vehicle characteristics by Jones and Stein at HHS (1987). See also row2 in Tablel and Strandberg (1991c)
Decision-makers' ranking of safety measures should be based on repeated results from several independent investigations and consensus in the scienti c community rather than on negotiations and agreements within some exclusive group. Recently, a report by an expert commission recom-mended "a lower level of ambition" for the periodic car testing programme. This was based on an appendix, which has been heavily criticised for its anecdotal perspective on accident causes and for considering a number of vehicle de ciencies (e.g. rust damage in the car body, poor shock absorbers, etc.) unimportant to traffic safety. See row 8 in Table 1. No reference is given here to avoid embarrassing this particular group of experts and to restrict the Workshop discussion to principles.
Author: Lennart Strandberg. Paper 93SF078: Workshop on Assessments of Driving Safety and of Crashworthiness. P.4 (6)
References
Aschenbrenner K M, Biehl B, Wurm G W (1992). Mehr Vehrkehrssicherheit durch bessere Technik? Feldtuntersuchungen zur Risikokompensation am Beispiel des Antiblockiersystem
(ABS). Forschungsberichte der Bundesanstalt ir Strassenwesen, no. 246.
Biehl B, Aschenbrenner M, Wurm G (1987). Einfluss der Risikokompensation auf die Wirkung von Verkehrssicherheitsmassnahmen am Beispiel ABS. Unfall und Sicherheitsforschung Strassen-verkehr, Heft 63. Bundesanstalt fiir Strassenwesen.
DN: Dagens Nyheter (1992). Renault most dangerous car. Saab 9000 safest in DN's investigation of accidents in 1991. (In Swedish: Renault farligaste bilen. Saab 9000 säkrast i DN:s undersökning om 1991 års olyckor.) Front cover headline and article in Swedish daily Newspaper, December 30, 1992. Dagens Nyheter, S-105 15 Stockholm.
DN: Dagens Nyheter (1993). Renault first to declare safety. (In Swedish: Renault först att deklarera säkerhet.) Front cover paragraph and article in Swedish daily Newspaper, June 12,
1993. Dagens Nyheter, S-105 15 Stockholm.
Fosser S (1991). The effect of periodic motor vehicle inspection on accident risk. Results of an experiment with 205 000 passenger cars and vans in Norway (in Norwegian with English summary). Report 0070/1991, Transportökonomisk Institutt, N 0602 Oslo 6.
Foster K K (1991). Annual tabulations of mature driver program driving record comparisons -1991. Report CAL-DMV-RSS-91-130, Dept. of Motor Vehicles, California.
IIHS (1987). Double-Trailer Trucks More Than Double Danger on Highways. Status Report, Vol.22, No. 1, Insurance Institute for Highway Safety, 1005 N Glebe Road, Arlington, Virginia
22201, USA.
IonesI S & Stein H S, (1987). Defective Equipment and Tractor-Trailer Crash Involvement. Insurance Institute for Highway Safety, 1005 N Glebe Road, Arlington, Virginia 22201, USA. Maycock G, Lockwood C R, Lester Julia F (1991). The Accident Liability of Car Drivers. Research
Report 315, Transport and Road Research Laboratory, United Kingdom.
OECD (1990). Behavioural adaptations to changes in the road transport system. Organisation for Economic Co-operation and Development, Road Transport Research, Paris. IRRD No 824028. Priez A, Petit C, Guezard B, Boulommier L, Dittmar A, Delhomme A, Vernet-Maury E,
Pailhous E, Foret-Bruno J Y, Tarriere C (1991). How About the Average Driver in a Critical Situation? Can He Really Be Helped by Primary Safety Improvements? To be published in proceedings of the 13th ESV Conference, Paris, November, 1991. Paper no.91-S7-O-07.
Sampson E E (1971). Social Psychology and Contemporary Society. John Wiley & Sons, Inc. Schlesselman I I (1982). Case-Control Studies. Oxford University Press.
SOU (1991). Safer Drivers (in Swedish: Säkrare förare. Summary in English). Statens Offentliga Utredningar 1991:39, Department of Transportation, Stockholm. ISBN 91-38-10785-6.
StevensonB (Ed. 1948). The home book of proverbs, maxims, and familiar phrases. Macmillan, New York. Reference from Sampson, 1971.
Strandberg L, Nordström O, Nordmark S (1975). Safety Problems in Commercial Vehicle
Handling. Proc. of a Symposium on Commercial Vehicle Braking and Handling, pp. 463-528,
May 5 7, 1975, Ann Arbor, Michigan. Also in VTI Report 82A, Swedish Road and Traffic Research Institute (VTI), S-581 95 LINKOEPING.
Strandberg L (1987). On the Braking Safety of Articulated Heavy Freight Vehicles. Proc. of OECD Symposium on the Role ofHeavy Freight Vehicles in Tra ic Accidents, pp.3 23 -3-40. In VTI Reprint 134 also.
Strandberg L (l989a). Skidding Accidents and their Avoidance with Different Cars. Proceedings on the 12th ESI/Conference, pp.825 828, Göteborg, Sweden, May 29 June l, 1989. In VTI Reprint 158 also.
Author: Lennart Strandberg. Paper 93SFO78: Workshop on Assessments of Driving Safety and of Crashworthiness. P.S (6)
Strandberg L (1989b). Braking Characteristics of 400 Heavy Trailer Combinations from Denmark, Finland, Norway and Sweden. Proceedings on the IZth ESVConference, pp.139-159, Göteborg, Sweden, May 29 - June 1, 1989. In VTI Reprint 135 also.
StrandbergL (199la). Case-Control Studies for Assessment of Accident Risks in Drivers and Vehicles (in Swedish: Bestämning av olycksrisker hos trafikant och fordon). VTI Note TF 50-20 (to appear as VTI Report 367).
StrandbergL (1991b). Crash Avoidance Capability of 50 Drivers in Different Cars on Ice. Thirteenth ESV Conference, Paris, November, 1991. VTI Reprint 179.
StrandbergL (1991c). Experience and Science Preventing Childbed Fever, Cancer and Traffic Accidents. (In Swedish: Praktik och vetenskap mot barnsängsfeber, cancer och trafikolyckor).
Nordic Accident Researcher Seminar, Gotland, Sweden, August, 1991. Proceedings in Arbete
och Hälsa, National Institute of Occupational Health, Solna, Sweden.
Struckman-JohnsonD L, Lund AK, Williams AF, Osborne D W (1989). Comparative Effects of Driver Improvement Programs on Crashes and Violations. Accid. Anal. & Prev. Vol.21, No.3, pp.203-215.
Stottrup-Hansen E, Ahlbom A, Axelson O, Hogstedt C, Juul Jensen U, Olsen] (1990). Negative Results - no effect or information? Arbete och Hälsa l990:l7, National Institute of Occupat1onal Health Solna, Sweden
UR: UtbildningsRadion (1989). 'Drive Alright' (speaking1n Swedish: 'Ratta Rätt'). Educational TV programmes Nos 89432/tvl and 89432/tv3, 30 minutes each with about 15 minutes demonstration of car stability problems. Swedish Television Broadcasting Company, Stockholm.
ViB: Vi Bilägare (1993). Small girl killed by Volvo's Air-Bag (in Swedish: Liten flicka dödad av Volvos Air-Bag). Article in Swedish weekly motor journal, no.13-14. Vi Bilägare, S-104 35 Stockholm.
Wilde GJ S (1982) The Theory of Risk Homeostasis: Implications for safety and Health. Risk Analysis, Vol 2, No 4.
Wilde GI S (1988). Risk homeostasis theory and traffic accidents: propositions, deduction and discussion of dissension111 recent reactions. Ergonomics, vol.31 (4), pp.441--468.
Temporary sideslip 51». (angle exaggerated) &
[_rista) e11 3,3XX x_x __ Countersteering x required t ___. » / x \-\ "A a. ,» - 1 _ . _ , ** ~ ,x' 1 **. 1 ' Stable .:. *F-» _ ,. 54,3: L \ . x \\\\\\\\\\\\\\'s>'
Spontaneous sideslip reduction
No steering required
Figure 1. Side forces (S), their resultant (F)and the position (P) of its application point in relation to the mass centre (T). Stability threatenedm upper car, since front tyres are better than the rear ones. One single example of many hazards that are unknown to quite a few road users.
Author: Lennart Strandberg. Paper 93SF078: Workshop on Assessments of Driving Safety and of Crashworthiness.
P.6 (6)
Table 1. Some examples of general problems and oversimpli cations in connection with risk assessments (C: Crashworthiness, D: Driving Safety).
General problem Consequences in specific cases Suggested remedies Risk homeostasis and
behavioural adaptation
considered unavoidable.
Safety potential of many measures is underestimated. Hazard education and risk cue amplification poorly developed.
Research for better knowledge on
risk complexity and development of active 'risk zooming', see text.
Accidents are blamed on the human factor or on
negligence by the directly
involved individuals.
Important hazards in vehicles or their
behaviour (and synergy interactions with
driver experience) remain unknown, while public safety campaigns are
burdened by self-evident information.
See discussion by Strandberg (19910).
Recognise the safety-relevance of vehicle properties quicker than what
was needed during the 50's and 60's to accept the association between
lung cancer and asbest (in synergy
with smoking).
Since one or a few studies
fail to find significant
improvements from safety measures, it is concluded that there are no effects.
The safety potential of ABS brakes has been questioned in several reviews (e.g. by OECD, 1990) mainly referring to a
conscientious but single study of German taxi drivers (Biehl et al, 1987 or Aschenbrenner et al, 1992).
Mis interpretations of 'negative results' from scientific research are elaborated
on by Stottrup Hansen et al (1990).
Repeat investigations until
consen-sus has been arrived at.
Consult experienced people to
iden-tify relevant but neglected
confound-ing factors and to improve safety by
combined measures. See Priez et al
(1991) for driver education in the
use of ABS. Exposure and accident
» measures are incompat
ible or biased due to
sampling method, con-founding or interaction.
See applications to
epi-demiology by Schlesselman (1982).
Common rental car type appeared to be the most risky one according to crash-worthiness ratings in a leading daily
newspaper (DN, 1992). All police repor-ted accidents were relarepor-ted to private only mileage from separate interviews.
Studded tyres may be considered more risky than summer tyres, since they are
over-represented on slippery roads with much greater accident risk.
Neutralise interaction from road en vironment hazards by better match-ing (e.g. selection of both case and
control vehicles from the accident
site, see Jones & Stein, HS 1987 or Strandberg 1991a). Use knowledge
from such studies to support use of
safer vehicles or to make super
vision and driver education more safety-relevant.
Definition of accident may vary between subjects and their ability to remember may be correlated to treat ment or experience
(Maycock et al, 1991).
Safety education has been considered worthless or even impairing safety on the basis of evaluations with self-reported accidents and mileage (SOU, 1991)
Carry out multiple investigations on
the correlation between safety
edu-cation treatments and the tendency
to under- or overestimate one's own
accidents and mileage.
Ignorance of confidence
intervals in mass media reports on test results or on risk assessments.
Small differences between cars, tyres, etc. are reported as ifthey were
signifi-cant both statistically and practically,
though they may be only a result of ran
dom variation and without practical im-portance.
Offer education on relevant statistics and stimulate journalists to
communicate the most important concepts to interested readers in the public.
Fatal and serious injury cases are not
distin-guished from cases with only property damage in accident counts.
Possible reductions in fatal accidents from car inspections may have been disregarded, when no significant effect was found on the total number of acci dents in a study by Fosser (1991).
Specify conditions and events
distin-guishing between fatal and
non-injury accidents (Strandberg, 1989). Expert groups base risk
conclusions on decisions rather than on scientific studies and consensus.
Public traffic safety development will not be optimal, since the knowledge and
risk preferences from other branches of
society are ignored.
Discuss major issues in seminars with industrial and university profes-sionals. Publish controversial results
awaiting consensus.
Poor or neglected
know-ledge on which car weight
will be optimal to society and minimise injuries among all road users.
Certain mass media recommend people to avoid light car models.
Car weight less than 1600 kg (arbitrary limit?) required for approval in a 9 point safety declaration published in a leading
daily newspaper (DN, 1993).
Assess optimal weight and frontal stiffness.
Adjust tax and insurance rates accordingly.
10
Crashworthiness rating considers occupants only (and disregards
aggres-siveness to other road users.
People and authorities neglect the safety potential in heavy vehicles of front underrun protection and de-formation zones adapted for cars, see Jones (1987) or Strandberg (1987).
Tax and insurance discount for
trucks and busses with non
agg-ressive fronts. File ISAT93TRDOC last saved 93-07-01 23:49, printed 93-07-02 11:26. Page 6 (6)