Folie 1

1
Side-effects on safety by making cars lighter
to reduce carbon dioxide emissions Ulrike
Trommer supervisor and examiner Lennart
Strandberg
2
Structure of the written thesis

• PART I
• - motivation - climate policy (CO2 reduction
  goals) vs. road safety policy (Vision Zero) -
  literature review on papers investigating
  weight-safety relationship
• PART II
• - formulate hypotheses based on the results of
  Part I
• - data basis methodology some important
  definitions
• - performed a pre-study investigating the data
  material and extracting confounding variables
• - analysis of single vehicle accidents (focus on
  individual safety)
• - analysis of two vehicle accidents (focus on
  net safety)
• Conclusion
• Does downweighing of Swedens passenger car fleet
  decrease road safety?

3
Agenda for todays presentation

• Summary of Part I as an introduction
• Hypotheses as basis for my quantitative analyses
• Data basis lack of data
• Methodology of the quantitative analyses
• Pre-study aim, data, results
• Analysis of single-vehicle accidents
• Analysis of two-vehicle accidents
• Conclusion

4
Motivation

• transport sector was important for economic
  development
• concern about the impact of especially road
  traffic on human health and the environment
• - CO2 abatement policy
• - Road safety
• focus on weight reduction as a measure to reduce
  CO2 emissions
• downweighing, is there a road safety
  side-effect?
• results of the literature review
• - general problem to distinguish between weight
  effects and safety and size effects
• - definition of safety important for method of
  analysis
• - different approaches and no consensus on
  safety-weight relation

two societal goals ? conflict?
5
Motivation
Passenger transport in EU15 Modal split
source of data Eurostat (2000)

?dominant role of the car in passenger transport
6
Motivation
Main emissions from road traffic in the EU 1,000
tonnes source of data Eurostat (2000)


?emissions mainly from road transport
technical developments promise improvements
7
Motivation
CO2 emissions from transport in the EU mio
tonnes CO2 source of data Eurostat (2000)

• CO2 emissions still increasing in the road
  transportation sector!

8
Motivation
Causes of disease or injury worldwide source
WHO (2001)

• Road traffic accidents represent a rapidly
  growing problem worldwide
• 2020 road traffic estimated moved to the 3rd
  leading cause of
• disability-adjusted life years lost
• gt40,000 fatalities and 1,700,000 seriously
  injured every year (EU)
• - road traffic accidents cost the EU yearly 160
  billion (2 GNP)
• ?economies and people suffer from road traffic
  accidents

9
Motivation
Cars in use 1/1 2003 by year model (Sweden)
source of data BIL Sweden (2003)


?almost 50 of Swedens passenger cars 10 years
or older (could have a negative effect on safety
as well)

10
Motivation
Injured and killed persons in road traffic ,
Sweden 1970-2002 source of data SIKA



?high safety level in Sweden, but lately no
further decrease but increase in fatalities and
severely injured

11
Motivation
Injured and killed persons in road traffic,
Sweden 1970-2002source of data SIKA

• Vision Zero no one gets killed or seriously
  injured
• a new focus on the whole system, how can it be
  operated safely

12
Hypotheses
1. in order to reduce CO2 emissions the Swedish
passenger car fleet changed in mass distribution
? process of downweighing 2. downweighing of
the car fleet in a country will increase its
safety problem the outcome of road traffic
accidents will get more severe. Subdividing the
second hypothesis leads to the following
hypotheses a) heavier passenger cars provide
better protection in single vehicle accidents b)
in a collision with a car of approximately the
same weight heavier passenger cars provide better
protection than lighter ones in a collision with
a car of approximately the same weight c) in a
collision with a car of a different weight class
(heavier or lighter) is the driver of the
heavier car better protected than the one of the
lighter car d) the introduction of lighter
passenger cars resulted in an increased safety
problem on Swedish roads and counteracted Vision
Zero
13
Hypotheses
Trends in average CO2 emissions, power, weight
and engine capacity for all new cars in EU (base
100 1990) source ECMT (2000), p. 8

• Has hypothesis 1 to be rejected already?
• At least no downweighing effect can be seen
• (compare figure on next slide)

14
Motivation
Newly registered car in Sweden 1991-2002 by
service weight source of data BIL Sweden
(2003)



?Swedens car fleet gets heavier, is already one
of the heaviest
15
Data basis

• Accident data
• - police-reported accidents occurred on Swedish
  roads (1994-99)
• - data basis containing info about vehicle,
  driver, accident
• - limitations variables (impact point) or
  values (air bag) missing
• Definition of injury severity
• - by the police at accident site
• - scale killed (1) uninjured (4)
• Exposure data, possible sources
• - estimates based on fuel consumption and road
  counts
• - detailed road counts on a sample of roads
• - information from each vehicles road distance
  meter from annual inspections (Bilprovningen)
• ? no such exposure data, only estimated based on
  vehicle age (for passenger cars only)

16
Data basis
Main accident databases
Annual travelled vehicle kilometres as function
of vehicle age according to the EMV-model
version 2.0 taken from Johansson (2002)

17
Methodology

• pre-study
• - detect confounding factors
• - accidents shall be analysed which are
  independent of driver skills
• - animal and pedestrian accidents included in
  order to represent the whole fleet
• Single vehicle accidents (passenger cars)
• - individual safety is analysed
• - simple graphs to reveal basic trends ( of
  accidents, proportion of injury levels by
  weight class)
• - comparing vehicles of comparable/not
  comparable weight
• - regression injury rate as a function of
  weight
• Two-vehicle accidents (passenger cars)
• - frontal collisions
• - safety of the driver in the OTHER car is taken
  into account
• - same procedure as for single vehicle accidents

18
Methodology
Weight of passenger cars involved in animal and
pedestrian accidents 1999 (left) and vehicle
weight of registered passenger cars in Sweden
(right)
?comparable distribution of cars involved and
registered
19
Methodology
Age distribution of passenger cars involved
in animal or pedestrian accidents and age
distribution of registered passenger cars in
Sweden
?comparable age distributions
20
Methodology

• Classification methods
• 1. subdivision into 28 class intervals of
  weight intervals are bounded at the top
  by the following percentiles of weight ? 1st,
  2nd, 4th, 6th, 8th, 10th, 15th, 20th, 25th, 30th,
  35th, 40th, 45th, 50th, 55th, 60th, 65th,
  70th, 75th, 80th, 85th, 90th, 92nd, 94th,
  96th, 98th, 99th, max
• average values of weight and variables
  (weighted by vkm)
• 2. subdivision as in 1. but into less class
  intervals of weight
• 3. subdivision into 3 groups according to the
  weight
• ? light 20 lightest cars
• ? medium lighter than 40 but also heavier
  than 40
• ? heavy heaviest 20 of all cars considered
• boundaries in kg differ and depend on the
  set of cars analysed

heavy cars
medium cars
all cars (min weight ? max weight
light cars
21
Methodology

• Regression analysis
• - explanatory variables (x-values)
• vehicle weight (size) and age, driver age and
  gender, speed limit
• - injury level (and injury rate, proportions) as
  response (y-value)
• - linear regression
• - logistic model for injury proportions as
  y-values
• - least squares fit
• - two different data sets 1. set of cars
  classified by weight
• 2. set of cars classified by length

22
Pre-study Animal and Pedestrian Accidents
Results of the analysis of animal accidents
(1999)

• statistically significant correlation found
  (animal accidents)
• vehicle age, driver gender, speed limit

23
Single vehicle accidents
Trends of uncontrolled analysis of driver
injuries (I)
- proportions of severe and fatal injuries remain
rather unchanged - proportion of minor injuries
decreases - proportion of no injuries increases
24
Single vehicle accidents
Trends of uncontrolled analysis of driver
injuries (II)
- absolute number of KSI diminishes with
increasing weight - KSI per annual travel
distance decreases with weight
25
Single vehicle accidents
Trends of uncontrolled analysis of driver
injuries (III)

• ? drivers in heavier cars seem to be better
  protected
• trends so far are a coincidence or not? ?
  further analysis

26
Single vehicle accidents
Trends of uncontrolled analysis of driver
injuries (IV)
?additional use of an exposure variable confirms
the trend
27
Single vehicle accidents
Controlled analysis of driver injuries (I) 407
records considered
requirements for driver age, gender, vehicle age
and speed limit are met
?heavier cars account for only few KSI, but
medium show high values as well as light cars

28
Single vehicle accidents
Controlled analysis of driver injuries (II) 242
records considered
requirements for driver age, gender, vehicle age
and speed limit are met

• conclusion drawn so far supported
• from individual point of view the fleet should
  get heavier

29
Single vehicle accidents - Regression
Classification by weight
30
Single vehicle accidents - Regression
Classification by length

• increases in either weight or length reduce the
  risk
• of driver fatalities and injuries per distance
  and
• proportion of killed or injured drivers

31
Two-vehicle accidents
Trends of uncontrolled analysis of driver
injuries (I)
- proportions of minor, severe and fatal injuries
decrease - proportion of no injuries increases
32
Two-vehicle accidents
Trends of uncontrolled analysis of driver
injuries (II)
?heavy vs. heavy car accidents produce the lowest
KSI or INJ rate (medium vs. medium car rates
rather high)
33
Two-vehicle accidents
Trends of uncontrolled analysis of driver
injuries (III)
?heavy vs. medium car accidents produce the
highest KSI rate
34
Two-vehicle accidents
Controlled analysis of driver injuries (I) 107
records considered
requirements for driver age, gender, vehicle age
and speed limit are met
?proportion of uninjured drivers decreases with
the weight of the OTHER car!
35
Two-vehicle accidents - Regression
36
Two-vehicle accidents - Regression
Classification by weight
37
Two-vehicle accidents
Classification by length

• results depend on the classification chosen
• reduced safety due to heavier car fleet
  possible see table
• increase in length of the other reduces injuries
  in case car

38
Conclusion

• Lack of data
• - exposure
• - measure of crash severity
• Single vehicle accidents (passenger cars)
• - individual safety is analysed
• - drivers in heavier cars seem to be better
  protected
• Two-vehicle accidents (passenger cars)
• - frontal collisions
• - safety of the driver in the OTHER car is taken
  into account
• - the results are not so easy to interpret, but
  there are tendencies
• that the heavier the OTHER car is the more
  severe are the
• injuries in the car it collided with
• ?other accidents types (side-accidents) and
  collision partners (truck)
• should be considered as well

39
Discussion
40
Motivation
Performance by mode of passenger transport EU15
1,000 mio pkm source of data Eurostat (2000)

? transport by passenger cars experienced
enormous growth
41
Motivation
Worldwide per-capita movement of people and
freight, 1850-1990 source Pastowski, Gilbert
(2003), p. 2

42
Motivation
Traffic injury volume model describing the
traffic safety problem source ETSC (1999), p.
22



? countermeasures have to act on at least one of
the axes

43
Methodology
Vehicles weight in pedestrian and animal
accidents happened in 1999
? similarities to the weight distribution of
Swedens car fleet?