Downloaded from SAE International by University of British Columbia, Monday, July 30, 2018

SAE TECHNICAL
PAPER SERIES 2000-01-1354

Estimation of Crash Injury Severity Reduction

for Intelligent Vehicle Safety Systems
Wassim G. Najm, Marco P. daSilva and Christopher J. Wiacek
Volpe National Transportation Systems Center

Reprinted From: Intelligent Vehicle Systems

(SP–1538)

SAE 2000 World Congress

Detroit, Michigan
March 6-9, 2000

400 Commonwealth Drive, Warrendale, PA 15096-0001 U.S.A. Tel: (724) 776-4841 Fax: (724) 776-5760
 Downloaded from SAE International by University of British Columbia, Monday, July 30, 2018

The appearance of this ISSN code at the bottom of this page indicates SAE’s consent that copies of the
paper may be made for personal or internal use of specific clients. This consent is given on the condition,
however, that the copier pay a $7.00 per article copy fee through the Copyright Clearance Center, Inc.
Operations Center, 222 Rosewood Drive, Danvers, MA 01923 for copying beyond that permitted by Sec-
tions 107 or 108 of the U.S. Copyright Law. This consent does not extend to other kinds of copying such as
copying for general distribution, for advertising or promotional purposes, for creating new collective works,
or for resale.

SAE routinely stocks printed papers for a period of three years following date of publication. Direct your
orders to SAE Customer Sales and Satisfaction Department.

Quantity reprint rates can be obtained from the Customer Sales and Satisfaction Department.

To request permission to reprint a technical paper or permission to use copyrighted SAE publications in
other works, contact the SAE Publications Group.

All SAE papers, standards, and selected

books are abstracted and indexed in the
Global Mobility Database

No part of this publication may be reproduced in any form, in an electronic retrieval system or otherwise, without the prior written
permission of the publisher.

ISSN 0148-7191
Copyright 2000 Society of Automotive Engineers, Inc.

Positions and opinions advanced in this paper are those of the author(s) and not necessarily those of SAE. The author is solely
responsible for the content of the paper. A process is available by which discussions will be printed with the paper if it is published in
SAE Transactions. For permission to publish this paper in full or in part, contact the SAE Publications Group.

Persons wishing to submit papers to be considered for presentation or publication through SAE should send the manuscript or a 300
word abstract of a proposed manuscript to: Secretary, Engineering Meetings Board, SAE.

Printed in USA
 Downloaded from SAE International by University of British Columbia, Monday, July 30, 2018
Estimation of Crash Injury Severity Reduction
for Intelligent Vehicle Safety Systems
Wassim G. Najm, Marco P. daSilva and Christopher J. Wiacek
Copyright © 2000 Society of Automotive Engineers, Inc.
ABSTRACT
A novel methodology is presented to estimate the safety
benefits of intelligent vehicle safety systems in terms of
reductions in the number of collisions and the number
and severity of crash-related injuries. In addition, mathe-
matical models and statistics are provided to support the
estimation of the crash injury reduction factor in rear-end,
lane change, and single vehicle roadway departure colli-
sions. Simple models based on Newtonian mechanics
are proposed to derive ∆v, the change in speed that a
vehicle undergoes as a consequence of crashing. Statis-
tics on the distribution of vehicle types and weights in the
United States are supplied, which are needed for ∆v esti-
mation. Moreover, mathematical equations are derived to
estimate the average harm per collision. Finally, statistics
on the average harm per occupant are obtained from the
1994 and 1995 Crashworthiness Data System crash
databases.
INTRODUCTION
Intelligent vehicle safety systems (IVSSs), such as rear-
end collision warning systems, have the potential to
increase the safety of motor vehicles. The safety benefits
of IVSSs can be measured by the number of collisions
that might be avoided and by the number and severity of
crash-related injuries that might be alleviated if drivers
were assisted by such systems. An engineering-based
methodology was recently developed to estimate system
effectiveness and predict potential safety benefits in
terms of the number of collisions that might be avoided
with the assistance of an IVSS [1]. This methodology uti-
lizes as input non-crash data collected from driving simu-
lator experiments, test track studies, and field operational
tests. Crash data are not available from field operational
tests since they are usually limited to few instrumented
vehicles operating for a relatively short period of time.
The reduction in the number of collisions is projected
based on driver and vehicle performance in safety-critical
driving conflicts with and without the assistance of an
IVSS. A safety-critical driving conflict leads to a collision if
the driver does not attempt any avoidance action.
1
2000-01-1354
Volpe National Transportation Systems Center
This paper describes a novel methodology to assess the
impact of IVSSs on the number and severity of crash-
related injuries. This methodology estimates an injury
reduction factor in both the number of injured persons
due to collisions that might be avoided and the severity of
injuries due to lower-impact collisions that might not be
avoided if drivers were assisted by an IVSS. In addition,
this paper provides fundamental models and statistics on
crash severity in support of this novel methodology, which
address rear-end, lane change, and single vehicle road-
way departure (SVRD) crash types. Simple mathematical
models are described based on fundamental physics,
which estimate Delta v (∆v) in each of the three crash
types. The parameter ∆v, a preferred measure of the
severity of a collision, denotes the change in speed that a
vehicle undergoes as a consequence of crashing. Crash
statistics are utilized to correlate the values of ∆v to the
injuries suffered in each of the three crash types. The
Crashworthiness Data System (CDS) crash database
provides such statistics [2]. The CDS contains detailed
data on a representative, random sample of thousands of
minor, serious, and fatal crashes. There are 24 field
research teams that study about 5,000 crashes a year
involving passenger cars, light trucks, vans, and utility
vehicles. These researchers perform a site investigation
on the crash scene to provide a detailed description of
the collision, and follow up by interviewing the crash vic-
tims and review medical records to determine the nature
and severity of their injuries.
First, this paper describes the fundamental equations of
a novel methodology that estimates the impact of IVSSs
on the number of collisions and the number and severity
of crash-related injuries. Next, mathematical models and
relevant data are provided in support of estimating the
injury reduction factor. This is followed by mathematical
models that derive ∆v from vehicle speeds at impact in
rear-end, lane change, and SVRD collisions. After, the
distribution of vehicle types and weights in the United
States (US) vehicle fleet is supplied. The 1994 and 1995
CDS crash statistics are then presented to show the rela-
tionship between injury severity and ∆v for a range of
speed bins in each of the three collision types. Finally,
this paper concludes with a summary.
 Downloaded from SAE International by University of British Columbia, Monday, July 30, 2018

SAFETY BENEFITS ESTIMATION pwo(Si): Probability of a collision when encounter-

METHODOLOGY ing a safety-critical driving conflict Si with-
out the assistance of an IVSS.
Intelligent vehicle safety systems (IVSSs) have the poten- M wo (Si ) : Number of encounters with a safety-
tial to reduce the number of motor vehicle collision and to critical driving conflict Si per VMT without
alleviate the number and severity of crash-related inju- the assistance of an IVSS.
ries. The number of collisions that might be avoided with
Instrumented vehicles in field operational tests collect
the assistance of an IVSS is estimated as follows:
data that can be used to estimate the values of M w (Si )
B = Nwo × SE (1)
and M wo (Si ) . An estimate of pwo(Si) is computed as fol-
Nwo: Number of relevant collisions without the
assistance of an IVSS. lows:
SE: Total IVSS effectiveness in mitigating rele- N wo (Si )
vant collisions. p wo (Si ) = (5)
M wo (Si )
Relevant collisions refer to vehicular crashes that the
IVSS is designed to address. Simple queries of crash N wo (Si ) denotes the number of collisions preceded by a
databases, such as the General Estimates System safety-critical driving conflict Si per VMT without the
(GES) database, yield values for Nwo. The parameter SE
assistance of an IVSS. This parameter can be estimated
is defined by the following equation:
using driving exposure statistics and crash data from
n national crash databases. Consequently, the estimation
SE = ∑P
i =1
wo (Si ) × E (S i )
(2) of E(Si) solely relies on estimating pw(Si). Potential esti-
mation methods include Monte Carlo computer simula-
n: Total number of distinct safety-critical driving tion [4] [5], principles of reliability theory [4], application of
conflicts leading to Nwo. "extreme value" theory [6], or other suitable techniques.
Si: A distinct safety-critical driving conflict. The number and severity of crash-related injuries that
Pwo(Si):Ratio of relevant collisions preceded by Si might be alleviated with the assistance of an IVSS is
over all relevant collisions without the assis- measured by an injury reduction factor, R, defined as fol-
tance of an IVSS. lows:
E(Si): IVSS effectiveness in mitigating relevant col-
Hw
lisions preceded by Si. R = 1− (6)
H wo
A set of five pre-crash variables in the GES and CDS
crash databases provides a basis for the identification of H w: Total harm caused by crash injuries with the
distinct safety-critical driving conflicts (pre-crash scenar- assistance of an IVSS.
ios) leading to different crash types. This set of variables Hwo: Total harm caused by crash injuries without
outlines the sequence of events leading from a normal the assistance of an IVSS.
driving scenario into a collision. The following equation
Generally, the “harm” is defined as the sum of injuries of
determines the value of Pwo(Si):
crash victims, with each injury weighted in proportion to
N wo (Si ) the economic cost of the outcome of such injury whether
Pwo (Si ) = (3) fatal or not. The total harm, H, can be formulated as:
N wo
Nwo(Si): Number of relevant collisions preceded by Si H= N×H (7)
without the assistance of an IVSS. N: Number of collisions.
Values for Nwo(Si) can be obtained from the GES [3]. The H: Average harm per collision.
parameter E(Si) is estimated by: Using above equations, the injury reduction factor, R, is
expressed as follows:
p w (S i ) × M w (Si )
E (Si ) = 1 − (4) l
p wo (S i ) × M wo (S i ) n
∑ Pwo (Si ) × {1 − E(Si )} × ∑ Pw (Si , ∆v k ) × H(∆v k )
pw(Si): Probability of a collision when encounter- i=1 k =1
R =1− n l
(8)
ing a safety-critical driving conflict Si with
the assistance of an IVSS. ∑ Pwo (Si ) × ∑ Pwo (Si , ∆v k ) × H(∆v k )
i=1 k =1
M w (Si ) : Number of encounters with a safety-criti- l : Range of ∆v values.
cal driving conflict Si per vehicle-miles
∆vk: Crash severity measure, denoting the
traveled (VMT) with the assistance of an
change in speed bin k that a vehicle
IVSS.
undergoes as a consequence of crash-
ing.
2
 Downloaded from SAE International by University of British Columbia, Monday, July 30, 2018

H ( ∆v k ) : Average harm per collision resulting in REAR-END COLLISION – In a rear-end collision, the
∆vk for a specific manner of collision (e.g., post-crash speed is expressed mathematically as [7][8]:
rear-end impact).
( m F × v cF + m L × v cL )
N (S , ∆v k ) vf = (11)
Pw (Si , ∆v k ) = w i (9) (m F + m L )
N w (S i )
mF: Mass of following vehicle.
Nw(Si,∆vk): Number of relevant collisions that were
mL: Mass of lead vehicle.
preceded by a safety-critical driving con-
flict Si and resulted in ∆vk with the assis- vcF: Velocity of following vehicle at time of impact.
tance of an IVSS. vcL: Velocity of lead vehicle at time of impact.
Nw(Si): Number of relevant collisions that were vf: Final velocity of the coupled vehicles after
preceded by a safety-critical driving con- impact.
flict Si with the assistance of an IVSS.
The changes in speed as a result of the impact for the fol-
N wo (S i , ∆v k ) lowing vehicle, ∆vF, and for the lead vehicle, ∆vL, are then
Pwo (S i , ∆v k ) = (10) computed by:
N wo (S i )
Nwo(Si,∆vk): Number of relevant collisions that were ∆vF = vcF – vf ∆vL = vcL – vf (12)
preceded by a safety-critical driving con-
flict Si and resulted in ∆vk without the LANE CHANGE COLLISION – Figure 1 illustrates one
assistance of an IVSS. vehicle (Number 1) steering toward another vehicle
(Number 2) that is traveling straight in an adjacent lane,
Nwo(Si): Number of relevant collisions that were
prior to a lane change collision.
preceded by a safety-critical driving con-
flict Si without the assistance of an
IVSS. y

Equation (8) assumes that the vehicle crashworthiness 2

(e.g., crash protection offered by vehicles), distribution of x v2c
vehicle weights, and vehicle occupancy remain the same
with and without the assistance of an IVSS. Therefore, v1c
the reduction of injury severity would occur due to lower 1
closing speeds at impact (smaller ∆v) if drivers were θ
assisted by an IVSS. The values of H ( ∆v k ) can be
derived from available crash databases. The ratios Pw(Si, Figure 1. Lane Change Crash Model Schematic
∆vk) and Pwo(Si, ∆vk) are possible to obtain when exercis-
ing Monte Carlo simulation to estimate the probabilities of The longitudinal components of ∆v for vehicle 1, ∆v1x,
a crash in safety-critical driving conflict Si with and with- and vehicle 2, ∆v2x, are computed by:
out the assistance of an IVSS. These simulations yield a ∆v1x = v1c × cos(θ) – vfx ∆v2x = v2c - vfx (13)
number of collisions along with vehicle speeds at impact.
These speeds can then be converted to values of ∆v v1c: Velocity of vehicle 1 at time of impact.
using simple models. An analysis of crash databases v2c: Velocity of vehicle 2 at time of impact.
such as the CDS produces values of P wo(Si, ∆vk). Such θ: Angle of vehicle 1 impacting vehicle 2 mea-
estimates can then be utilized as desired values to cali- sured clockwise relative to road direction (x-
brate the models and techniques that estimate both axis).
Pwo(Si, ∆vk) and Pw(Si, ∆vk). Next, mathematical models vfx: Longitudinal component of final velocity of
and relevant data are presented to enable the estimation coupled vehicles after impact:
of the parameter R as expressed in Equation (8).
(m 1 × v1c × cos( θ) + m 2 × v 2 c )
vf x = (14)
∆v ESTIMATION (m 1 + m 2 )
m1: Mass of vehicle 1.
Simple mathematical models based on Newtonian
mechanics can be utilized to estimate ∆v in rear-end, m2: Mass of vehicle 2.
lane change, and single vehicle roadway departure Similarly, the lateral components ∆v1y and ∆v2y are cal-
(SVRD) collisions. Such models represent inelastic colli- culated using the following equations:
sions among vehicles; i.e., the vehicles remain in contact
as one combined object after the impact resulting in ∆v1y = v1c × sin(θ) – vfy ∆v2y = - vfy (15)
equal post-crash speeds.

3
 Downloaded from SAE International by University of British Columbia, Monday, July 30, 2018

vfy: Lateral component of final velocity of coupled VEHICLE TYPE AND WEIGHT DISTRIBUTION
vehicles after impact:
Distributions of vehicle types and weights on US road-
m × v1c × sin( θ)
vf y = 1 (16) ways are needed to compute vf as indicated in Equations
(m1 + m 2 ) (11), (14), and (16). Figure 2 shows the distribution of
vehicle types based on 1996 US registration statistics
In a lane change collision, total ∆v for vehicle 1, ∆v1, and
that counted approximately 203 million registered vehi-
total ∆v for vehicle 2, ∆v2, are then calculated as follows:
cles [9]. The “truck” type encompasses light trucks (≤
4,500 Kg), medium trucks, and heavy trucks (> 15,000
∆v1 = ∆v12x + ∆v12y ∆v 2 = ∆v 2 2x + ∆v 2 2y (17) Kg). Light trucks accounted for about 93.3 percent of all
trucks in 1996. Figure 3 illustrates the distribution of light
SINGLE VEHICLE ROADWAY DEPARTURE (SVRD) vehicles, autos plus light trucks, which comprised about
COLLISION – The total ∆v in SVRD collision type may be 95.3 percent of all registered vehicles in 1996. Table 2
expressed as: indicates that the average weight of automobiles varied
between 1,139 Kg for subcompact autos and 1,657 Kg
∆v = α × vc (18)
for large autos over 1992-1997 period. The distribution of
α: Impact severity attenuation factor. the truck population by gross vehicle weight, excluding
vc: Speed of vehicle at time of impact. light trucks, is shown in Figure 4 based on 1992 statistics.

This model assumes that the parameter vc is equal to the Auto

61.5%
travel speed of the vehicle when it leaves the roadway.
Moreover, the final speed of the vehicle is assumed to be
zero after the impact. The parameter α is a positive ran-
dom number (≤ 1) that takes into account the effects of
objects collided with in SVRD collisions. The distribution Bus
of α depends on the category of objects collided with. 0.3%

Table 1 presents the distribution of objects collided with in

SVRD collisions based on 1994 and 1995 CDS crash
statistics. Fixed objects include fences, walls, ground, Motorcycle
1.9%
curbs, fire hydrants, and other. Breakaway objects Truck
36.3%
encompass trees, bushes, embankments, and posts
designed to break away upon impact. Non-breakaway Figure 2. Distribution of Vehicle Types in the US Based
objects consist of poles or posts that are not designed to on 1996 Statistics
break away upon impact. Fixed traffic barriers comprise
concrete barriers, impact attenuators, and guardrails.
24.1
Non-fixed objects include passenger cars, trucks, pedes-
trians, animals, trailers, and other. Detailed analyses of 20.1
SVRD collision cases are needed to statistically describe 18.6

the distributions of α for the different object categories. 14.9

11.4
Table 1. Distribution of “Objects Collided With” in SVRD
8.9
Crashes

Object %*
Fixed 30.1 1.2 0.8

Breakaway 24.3 Truck <= 2700 Kg Compact Midsize Subcomp Truck<= 4500 Kg Large 2-Seaters Minicompact

Non-breakaway 22.6 Figure 3. Distribution of Light Vehicles Based on 1996

Statistics
Fixed traffic barrier 19.4
Non-fixed 3.6
Total 100.0
*: Based on 1994 & 1995 CDS

4
 Downloaded from SAE International by University of British Columbia, Monday, July 30, 2018

Table 3 defines the injury scale of MAIS and presents the

Table 2. Weight of Automobiles by Size “fatal equivalent” values that correspond to each MAIS
severity level based on its crash economic cost [10] [11].
Auto Type Weight (Kg)*
MAIS values are converted to “fatal equivalent” values so
Minicompact 1,227 that injuries of different severity can be measured on a
Subcompact 1,139 single ratio scale [12]. Table 4 presents statistics on the
Compact 1,211 average number of occupants per vehicle and the aver-
age number of vehicles involved in three collision types
Midsize 1,442
for both light vehicles and trucks.
Large 1,657
2 Seaters 1,328 Table 3. MAIS levels and their “Fatal Equivalent” values
*: Average weight of new autos from 1992 through 1997.
I MAIS w*
0 Uninjured 0.0014
% 1 Minor 0.0087
40.3 40.5 2 Moderate 0.0417
3 Serious 0.1250
4 Severe 0.2765
5 Critical 0.8483
6 Fatal 1
*: Based on 1994 economic cost [10]

7.6 The 1994 and 1995 CDS crash databases were queried
4.5 4.3
2.8
to estimate the values of H o ( ∆v k ) for rear-end, lane
"4.5 – 6.4" "6.4 - 7.3" "7.3 – 8.8" "8.8 – 11.8" "11.8 - 15" ">15" change, and SVRD collisions. A total of 1,235 crash
Thousand (Kg)
cases or about 28 percent of all examined cases con-
tained valid ∆v information. Each crash case was
Figure 4. Distribution of Medium/Heavy Trucks by
weighted using its Ratio Inflation Factor in the CDS so as
Weight Based on 1992 Statistics
to approximate the national profile. Moreover, injuries are
scored in the CDS database using the police-reported
AVERAGE HARM PER COLLISION
injury score KABCO which is converted to the MAIS
scale by a translator derived from the 1982-1986 National
The average harm per collision resulting in ∆vk, H ( ∆v k ) Accident Sampling System (NASS) data.
in Equation (8), is obtained for rear-end, lane change,
and single vehicle roadway departure (SVRD) collisions Table 4. Average Values of Vehicles Involved and
in 10 Km/h speed bins as follows: Vehicle Occupancy in Collisions^

H ( ∆v k ) = H o ( ∆v k ) × {Average number of Crash # Vehicles # Occupants

occupants per vehicle} × {Average Type Vehicle Type per Crash per Vehicle
number of vehicles involved Rear- Light Vehicles 2.2 1.3
in a specific collision} (19) End Trucks 2.2 1.2
H o ( ∆v k ) : Average harm per occupant as a result
Lane Light Vehicles 2.0 1.5
of ∆vk in a specific manner of collision,
estimated by: Change Trucks 2.0 1.3
6 Light Vehicles 1 1.4
∑
1
H o ( ∆v k ) = × w (i) × I(i) (20) SVRD
N o ( ∆v k ) i =0 Trucks 1 1.1

N0(∆vk):Number of occupants of a specific crash type ^: Based on annual statistics from 1989-93 GES Crashes [11].
resulting in ∆vk.
w(i): Weighting coefficient corresponding to maxi- AVERAGE HARM PER OCCUPANT IN REAR-END
mum injury severity i based on the Maximum COLLISIONS – A total of 455 rear-end crash cases or
Abbreviated Injury Scale (MAIS). about 38 percent of all rear-end crash cases in the 1994
I(i): Total number of persons involved with a max- and 1995 CDS databases had valid total ∆v information
imum injury severity i based on MAIS result- for both the lead and following vehicles. These cases
ing from ∆vk collisions.
were examined to estimate H o ( ∆v k ) based on the high-
est total ∆v experienced in each collision. Figure 5

5
 Downloaded from SAE International by University of British Columbia, Monday, July 30, 2018

illustrates the relationship between ∆v and average harm Fatal Equivalent Value
0.035 0.034 0.034 0.034
per occupant in rear-end collisions, expressed in terms of 0.032
fatal equivalent values. The average harm per occupant 0.03
should increase with increasing values of ∆v. However, 0.024 0.025
0.025
there appears to be an anomaly in the distribution due to
the low number of cases available in some ∆v bins, espe- 0.02 0.017
cially in the “15 – 25” and “> 65” Km/h ∆v bins. It is rec- 0.015 0.012
ommended that a fatal equivalent value of 0.028 be
0.01 0.008
assigned to H o ( ∆v k ) at ∆v above 55 Km/h (using the 0.005
0.005
weighted average of fatal equivalent values at the last two
0
∆v speed bins). Moreover, a value of 0.006 is suggested "5 - 15" "15 - 25" "25 - 35" "35 - 45" "> 45"
for the “15 – 25” Km/h ∆v bin based on the increasing ∆v (Km/h)
rate of H o ( ∆v k ) between “25 –35” and “45 – 55” Km/h CDS Modified
∆v bins. Figure 6. Distribution of Average Harm Per Occupant in
Lane Change Collisions by Highest ∆v
AVERAGE HARM PER OCCUPANT IN LANE CHANGE
COLLISIONS – The 1994 and 1995 CDS crash data- AVERAGE HARM PER OCCUPANT IN SVRD
bases contained 67 cases or about 22 percent of all lane COLLISIONS – A total of 713 cases or about 24 percent
change crash cases with valid total ∆v information for of all SVRD crash cases with valid total ∆v information
both vehicles involved in the collision. The values of were extracted from the 1994 and 1995 CDS databases.
H o ( ∆v k ) are estimated using the highest total ∆v expe- Figure 7 illustrates the distribution of the average harm
rienced by either vehicle in each collision. Figure 6 illus- per occupant in SVRD collisions by 10 Km/h ∆v bins. The
trates the relationship between ∆v and average harm per average harm per occupant is expected to increase with
occupant in lane change collisions as estimated from the ∆v as evident in Figure 7. This harm increased dramati-
CDS database and displays a modified distribution to cally between “45 – 55” and “55 – 65” Km/h ∆v bins, and
adjust the observed anomaly. As seen in Figure 6, an then was reduced slightly at ∆v over 65 Km/h. This may
anomaly exists in the distribution of observed injury be attributed to the small number of valid cases available
severity data due mainly to the very low number of lane for the last two speed bins. It is recommended that a fatal
change crash cases available in the two-year period equivalent value of 0.376 be assigned to H o ( ∆v k ) at ∆v
under consideration.
above 55 Km/h (using the weighted average of H o ( ∆v k )
Fatal Equivalent Value at the last two ∆v bins).
0.041
0.04 Fatal Equivalent Value
0.4 0.389
0.359
0.03
0.024
0.3
0.02 0.017
0.015 0.014

0.01 0.009 0.2

0.004
0.127
0
"5 - 15" "15 - 25" "25 - 35" "35 - 45" "45 - 55" "55 - 65" ">65" 0.1 0.072
∆v (Km/h) 0.021
0.008 0.017
Figure 5. Distribution of Average Harm Per Occupant in
0
Rear-End Collisions by Highest ∆v "5 - 15" "15 - 25" "25 - 35" "35 - 45" "45 - 55" "55 - 65" ">65"
∆ v (Km/h)
Figure 7. Distribution of Average Harm Per Occupant in
SVRD Collisions by ∆v

6
 Downloaded from SAE International by University of British Columbia, Monday, July 30, 2018
SUMMARY AND CONCLUDING REMARKS
This paper delineated a novel methodology and funda-
mental supporting data to assess the impact of intelligent
vehicle safety systems on the number of collisions and
the number and severity of crash-related injuries using
computer simulations. Simple mathematical models with
representative parameters were described which derive
the parameter ∆v from speeds at impact and weights of
vehicles involved in rear-end, lane change, and SVRD
collisions. The SVRD collision model requires data on the
impact severity attenuation factor, α, which are not readily
available. Data are needed to better describe α for differ-
ent object categories. The 1994 and 1995 crash data-
bases were queried to identify the relative frequency of
striking five object categories in SVRD collisions. In sup-
port of ∆v estimation, statistics were presented on vehicle
types and weights based on 1996 vehicle registration in
the U.S.
Statistical data on the number of vehicles involved per
crash and the number of occupants per vehicle were pro-
vided in order to estimate the average harm per collision.
The 1994 and 1995 CDS crash databases were exam-
ined to obtain relationships between ∆v and average
harm per occupant in each of the three collision types.
Unfortunately, a small number of crash cases in the CDS
database contained valid information on ∆v. Conse-
quently, the results obtained had anomalies in the distri-
bution of the average harm per occupant with respect to
∆v. To remedy this problem, at least 5 years of CDS data
might be needed to generate the appropriate distribu-
tions.
ACKNOWLEDGMENT
Mr. John Smith of the Volpe Center’s Accident Prevention
Division is recognized for his contribution to this research
work by providing data from the 1994 and 1995 CDS
crash databases.
REFERENCES
1. Najm, W.G. and daSilva, M.P., “Benefits Estimation
Methodology for Intelligent Vehicle Safety Systems
Based on Encounters with Safety-Critical Driving
Conflicts”. Submitted for Publication at Intelligent
Transportation Society of America’s Tenth Annual
Meeting & Exposition, Boston, MA, May 2000.
2. National Center for Statistics and Analysis, “National
Automotive Sampling System, 1995 Crashworthiness
Data System, Data Collection, Coding and Editing
Manual”. United States Department of Transporta-
tion, National Highway Traffic Safety Administration,
Washington, D.C., 20590.
3. Najm, W.G., Wiacek, C.J., and Burgett, A.L., “Identifi-
cation of Precrash Scenarios for Estimating the
Safety Benefits of Rear-End Collision Avoidance Sys-
tems”. Fifth World Congress on Intelligent Transport
Systems, Seoul, Korea, October 1998.
4. NHTSA Benefits Working Group, “Preliminary
Assessment of Crash Avoidance Systems Benefits”.
Version II, United States Department of Transporta-
tion, National Highway Traffic Safety Administration,
December 1996.
5. Najm, W.G. and Burgett, A.L., “Benefits Estimation
for Selected Collision Avoidance Systems”. Fourth
World Congress on Intelligent Transport Systems,
Berlin, Germany, October 1997.
6. Campbell, K.L., Joksch, H.C., and Green, P.E., “A
Bridging Analysis for Estimating the Benefits of
Active Safety Technologies”. NHTSA Contract No.
DTNH22-93-D-07000, UMTRI-96-18, April 1996.
7. Evans, L., “Driver Injury and Fatality Risk in Two-Car
Crashes Versus Mass Ratio Inferred Using Newto-
nian Mechanics”. GMR-7928, April 1993.
8. Khadilkar, A.V., Redmond, D., and Ausherman, V.K.,
“Collision Avoidance System Cost-Benefit Analysis”.
Volume I, United States Department of Transporta-
tion, National Highway Traffic Safety Administration,
DOT HS 806 242, September 1981.
9. Davis, S.C., “Transportation Energy Data Book: Edi-
tion 18”. ORNL-6941, U.S. Department of Energy,
DE-AC05-96OR22464, September1998.
10. Blincoe, L.J., “The Economic Cost of Motor Vehicle
Crashes, 1994”. United States Department of Trans-
portation, National Highway Safety Administration,
DOT HS 808 425, July 1996.
11. Wang, J.-S., Knipling, R.R., and Blincoe, L.J., “Motor
Vehicle Crash Involvements: A Multi-Dimentional
Problem Size Assessment”. ITS America Sixth
Annual Meeting, Houston, TX, April 1996.
12. Knipling, R.R., Mironer, R., Hendricks, D.L., Tijerina,
L., Everson, J., Allen, J.C., and Wilson, C., “Assess-
ment of IVHS Countermeasures for Collision Avoid-
ance: Rear-End Crashes”. Appendix B, United
States Department of Transportation, National High-
way Traffic Safety Administration, DOT HS 807 995,
May 1993.