Campbell

Download as pdf or txt
Download as pdf or txt
You are on page 1of 11
At a glance
Powered by AI
The paper reviews different force-crush equations used to estimate vehicle impact velocities from crush data and proposes strategies to select the most appropriate equations. It also provides estimates of error bands given available crash test data.

The paper discusses methods based on a vehicle's crush deformation and energy, as well as equations using conservation of momentum and energy that are implemented in crash reconstruction programs. Empirical relationships between impact speed and crush depth are also used.

The accuracy of impact velocity estimations largely depends on the accuracy of the relationship between vehicle frontal impact force and crush depth, which varies between vehicles. Limited crash test data is available.

REVIEW OF CAR FRONTAL STIFFNESS EQUATIONS FOR ESTIMATING VEHICLE

IMPACT VELOCITIES
Jiang T.1, Grzebieta R.H.1, Rechnitzer G.1,
Richardson S.2 and Zhao X.L.1
1
Department of Civil Engineering,
Monash University, 2DV Experts
Australia
Paper Number 439

where M1, M2 are the mass of the respective


vehicles(kg). Some commonly used crash
reconstruction programs such as CRASH3,
EDCRASH and PCCRASH make use of these type
of equations for predicting pre and post crash
impact speeds [3, 4].

ABSTRACT
This paper reviews different force versus crush
empirical equations for cars used in accident
reconstruction over the past three decades. These
equations are compared to numerous data obtained
from various sources. A strategy for selecting the
most appropriate equations to use for determining
the frontal stiffness characteristics of a car for
accident reconstruction, simulation modeling, and
design purposes is also proposed. Estimates of error
bands for a particular strategy chosen given
available crash test data, are also provided.
INTRODUCTION
Estimates of vehicle speed changes during impacts
are important for assessing the impact speed and
hence crash severity of real world road accidents
for research, insurance claims and litigation
purposes. The term commonly used to define the
speed change during an impact is Delta V. A
method, essentially empirical, based on a vehicles
crush deformation and crush energy has been
widely used to determine Delta V. Generally, for a
frontal collinear car-to-car impact, from the crush
profiles of the two cars involved, the crush energy
can be calculated as

E1 =

w01

C1

F1 (C )dCdw

0 0
w02 C 2

E2 =

F2 (C )dCdw

(1).
(2).

where E is the crush energy (J); F(C) represents the


impact force per unit width of crush (N/m), C is the
residual crush depth (m), w01, w02 are the crush
widths (m) and subscripts 1 and 2 refer to car 1 and
car 2 [1]. Using the conservation of momentum, the
conservation of energy, and assuming zero
restitution, Robinette et al [2] provide the following
equations for calculating Delta V:
Delta V1 =

2M 2 ( E1 + E2 )
;
M 1 (M 1 + M 2 )

Delta V2 =

2M 1 ( E1 + E2 )
M 2 (M 1 + M 2 )

(3).

Obviously, the accuracy of Delta V predicted using


Equation 3 largely depends on the accuracy of the
relationship between the vehicles frontal impact
force per unit crush width F(C) and the crush depth
C. Numerous research and crash tests have been
carried out in regards to the frontal crush resistance
of a car. However, because of the wide diversity of
vehicle frontal structures and their complex crush
behaviour, F(C) must be empirically determined
from full-scale crash tests.
As full-scale crash tests are expensive and need a
lengthy time to prepare, only several car models
have been tested at a limited range of impact
speeds. Nevertheless, different trends of F(C)
versus C were observed from this limited data [5-8].
For most car models, full-scale frontal crash tests
were only conducted at 48.3 km/h (30 mph) for
regulation compliance purposes and/or at 56.3
km/h (35 mph) for New Car Assessment Program
(NCAP) tests. With regards to accident
reconstructions involving cars that have one or two
crash test data points, the common practice is to
empirically assign a linear relationship between
impact speed and residual crush depth, which is
expressed as

V
= b0 + b1C
3.6

(4).

where C is the residual crush depth (m), b0 is the


intercept or zero crush speed (m/s), b1 is the
slope of the speed-crush relationship (ms-1/m) and
V is the impact speed (Note that the term impact
speed is used loosely here. In most reconstruction
handbooks the symbol V is used for velocity and is
expressed in m/s. In this paper V will be used to
symbolize speed expressed in km/h). b0 is usually
set at 2.2 m/s (8 km/h or 5 mph) [5, 6, 9]. Thus, b1
can be calculated from Equation 4 using the data (V
and C) obtained from a single frontal crash test.
Campbell [1] originally proposed that if there is a
linear relationship between impact speed and crush,
the simplest characteristic for a vehicles front
structure that will reproduce the linear V-C
relationship is a linear force-crush relationship.
Hence
(5).
F (C ) = A + BC

Jiang 1

Mb0 b1
w0

(6).

B=

Mb12
w0

(7).

where M is the vehicle mass (kg) and w0 is the


crush width (m) [1].
However, there are some concerns about this
method for determining F(C). One concern is that a
single linear model type is used for all car models
and used over the whole impact speed range.
Vehicles do vary in their structures, within
manufacturing tolerance, from one vehicle to the
next that can result in significant differences in
force versus crush values. Test set-ups can also
vary from one crash to another. Another concern is
that coefficients A and B are derived from only one
crash test. The basic assumption here is that
coefficients A and B hold for a particular vehicle
over all speed ranges.
Hence it is not clear to what level of accuracy Delta
V can be estimated. Variance is inevitable in crash
tests due to various reasons. It is important for
accident reconstructionists to quantify the error
bands in regards to the accuracy of Equation 4.
Over the past three decades, a wealth of crash test
data has been made available from NCAP,
regulatory and laboratory crash tests. This data
provides the possibility to clearly assess the F(C)
equations used for Delta V estimations.
This paper reviews the different force versus crush
equations proposed and various approaches used in
accident reconstruction over the past three decades.
These equations are then compared to numerous
data obtained from various sources. A strategy for
selecting the most appropriate equations to use for
determining the frontal stiffness characteristics of a
car for accident reconstruction, simulation
modelling, and for design purposes is also
proposed. Estimates of error bands for a particular
strategy chosen given available crash test data, are
also provided.

In the late 1960s, Emori [10] suggested that the


impact force a car being subjected to in a head-on
collision is directly proportional to the crush,
somewhat similar to that of a spring force versus
displacement. He also proposed that the frontal
crush should be directly proportional to the impact
velocity.
On the basis of Emoris work as well as a range of
frontal fixed rigid barrier crash tests on 1971~1972
model full size GM cars and 1971~1974 model
Chevrolet Vegas, Campbell [1] found a linear
relationship between residual crush depth (C) and
impact speed (V), as shown in Figures 1 and 2.
60

50
Impact Speed (mph)

A=

Single Linear Equation

40

30

20

10

0
0

10

20

30

40

50

60

Residual Crush (inches)

Figure 1. Impact speed vs residual crush for


1971~1972 model full size GM cars [1].
60

Impact Speed (mph)

where A and B are coefficients. Coefficients A and


B can be determined as

50
40
30
20

REVIEW OF CAR FRONTAL STIFFNESS


MODELS

As mentioned above, because of the high expense


of a crash test, only a limited number of car models
have been crash tested over a range of impact
speeds. Nevertheless, it was possible to estimate
different frontal crush characteristics from this
data.

10
0
0

10

20

30

40

50

Residual Crush (inches)

Figure 2. Impact speed vs residual crush for


1971~1974 model Chevrolet Vegas [1].

Jiang 2

Campbell further proposed that this linear


relationship between V and C could be transformed
into a linear impact force-crush relationship,
expressed as Equation 5. Crash test data indicated
that the linear equation for F(C) was workable for
estimating crush energy in frontal, oblique and offset crashes. Campbell also gave the methods for
determining coefficients A and B in the linear
equation, as expressed in Equations 6 and 7.
Campbells findings have been extensively used for
accident reconstructions and laid the theoretical
foundation for some commonly used reconstruction
programs, such as CRASH3 and EDCRASH [3, 4].
Since Campbells classical work, numerous
research papers have been published and crash tests
have been carried out to investigate a vehicles
crush behaviour in relation to Equations 4 to 7.
Frontal fixed rigid barrier impact tests on several
car models (other than the ones tested by Emori
[10] and Campbell [1]) also showed a linear V-C
relationship, such as those proposed by Navin et al
[6] when he analysed crash test data for 1974 to
1981 model Honda Civic cars (see Figure 3).

Impact Speed (mph)

40
30
20
10
0
10

20

30

40

Residual Crush (inches)

Figure 3. Impact speed vs crush for 1974~1981


model Honda Civic [6].

In order to account for the weight differences


between different test vehicles, Varat et al [7] used
an Energy of Approach Factor (EAF) to analyse
crash test data. If the frontal structure of a vehicle
behaves as a linear dissipator and zero restitution is
assumed, the energy absorbed is

E=

1 2
kx
2

E 1 k 2
= ( )x
w0 2 w0

(9).

Through some algebraic manipulation, we have

2E
=
w0

k
x
w0

(10).

2E
and B = k , and
w0
w0
considering that some initial elastic deformation
energy is required before permanent crush results,
the following equation is arrived at

Assigning

EAF =

EAF = EAF0 + B x

(11).

where EAF0 is the Onset Energy Factor and B


represents the vehicle frontal impact coefficient
which is a constant based on a vehicles crush
properties [7]. Note that for a car crashing into a
rigid barrier, the crush energy E can also be
equated to the cars kinetic energy just prior to
impact such that E = M (V)2 (assuming zero
rebound speed). As can be seen from Equation 11,
the quantity EAF is theoretically linear to the
residual crush depth. If a linear relationship exists
between EAF and the residual crush depth from
crash test data, it can be demonstrated the
assumption that a vehicle behaves as a linear
dissipator is acceptable, and using a linear forcecrush equation to estimate crush energy is
appropriate. Otherwise, if EAF is not linear in
regards to crush depth, a linear force-crush
equation may not be suitable for representing the
vehicle crush behavior.

50

where E is the absorbed energy (J); k is the


dissipators linear force characteristic (N/m) and x
is the crush distance (m). Energy absorbed per unit
crush width is

Varat et al [7] used Equation 11 to analyse a


number of vehicle models where full frontal rigid
barrier crash tests were carried out over a range of
impact speeds. Eleven vehicle models were
analysed. Varat concluded that two vehicle models
showed a linear relationship between EAF and the
residual crush up to 80 km/h (50 mph) (see Figure
4), and four vehicles exhibited a linear relationship
up to 56 km/h (35 mph) (see Figure 5).

(8).

Jiang 3

residual crush plots, found that a bi-linear equation


was more suitable for 1980~1982 model GM
Citation cars (see Figure 7).

300

EAF

250
200

300
R^2 = 0.989

150
100

250

50

200
0

10

15

20

25

30

35

40

EAF

0
Crush (inches)

150

Figure 4. EAF vs crush for 1974 model Ford


Pinto up to 50 mph [7].

100

300

50

EAF

250

200

150

R^2 = 0.990

10

20

30

40

50

Crush (inches)

100

Figure 7. Energy Factor vs crush for 1980~1982


model Citation cars [5].

50
0
5

10

15

20

25

30

35

40

45

Crush (inches)

Figure 5. EAF vs crush for 1981~1985 model


Ford Escort up to 35 mph [7].

Neptune [8] also observed that a linear relationship


existed between EAF and residual crush up to
80km/h (50 mph) based on full overlap and partial
overlap rigid barrier crashes for 1986 to 1991
model Ford Taurus cars (see Figure 6). He further
stated that a review of available crash test data
revealed that the linear relationship between impact
force per unit crush width and crush depth was
valid for full frontal collisions with rigid barriers up
to a speed of 56 km/h.
Occupant
Compartment
Crush

EAF

Engine
Compartment
Crush

Neptune [8] also found that while a linear forcecrush relationship could be demonstrated up to 56
km/h for 1981 to 1985 model Ford Escort cars,
when the impact speed was greater than 56 km/h,
the vehicle no longer displayed a linear
relationship. The Escort dissipated less energy per
unit crush above 56 km/h (see Figure 8). Neptune
further concluded that in high severity collisions,
the crush response characteristic of this vehicle
could be divided into two regions, the engine
compartment crush region and the occupant
compartment crush region. Hence, the vehicle
could be modelled as a bi-linear dissipator where
the second dissipator (occupant compartment) does
not compress until the first dissipator (engine
compartment) bottoms out.

1988 Offset Frontal

Engine
Compartment
Crush

Occupant
Compartment
Crush

EAF

1986 Full Frontal


(Repeat Series)

56 km/h
(35 mph)

(Repeat Series)

1986-91 Full

Frontal (Single Tests)

Crush (inche s)

Figure 6. EAF vs crush for 1986~1991 model


Ford Taurus [8].

8 to 64 km/h

63 to 64 km/h
80 km/h
Crush (inche s)

Bi-Linear Equation

Figure 8. Energy factor vs crush for 1981~1985


model Ford Escort [10].

Some vehicles crash test data also showed a nonlinear trend. Strother et al [5], using EAF versus

Varat [7] also drew similar conclusions in his


research. For the six vehicle models that had a

Jiang 4

EAF

range of impact speeds up to 80 km/h, four vehicles


displayed a bi-linear trend as shown in Figure 9.
400
350
300
250
200
150
100
50
0

48 km/h (30 mph)


data point

10

15

20

25

30

35

40

45

Crush (inches)

Figure 9. Comparison of a linear (dash) and a


bi linear (solid) model [7].
Constant Force Equation

Some crash test data also displayed a quadratic


trend, e.g. full frontal barrier tests of 1974 model
Plymouth Satellites as shown in Figure 10.
Accordingly, Strother [5] proposed that a constant
force value, expressed as F/w0 = D, might be more
suitable where F is the crush force.
Some researchers used sophisticated finite element
models in conjunction with test results to develop a
two-stage constant force relationship as shown in
Figure 11. Wood et al [11] cited Sakurias work
and stated that a two-stage constant force-crush
relationship with a transition as the deformation
reaches the engine, could be used to represent
vehicles frontal crush characteristics. Futamata [12]
and Toyama [13] confirmed Sakurias two-stage
force model by examining the pattern of energy
absorbed by the various elements, such as side
rails, suspension members and so on, in the course
of crushing.
250

Figure 11. A two-stage constant force-crush


relationship [11].
CRASH DATA

Most of the findings in the previous section were


based on a limited number of crashes of vehicle
models dating from 1960 through 1986. To extend
these findings, as well as investigate the accuracy
of equations adopted that describe a cars crush
characteristics, data from over 1000 crash tests
were collected. Vehicle models ranged over a large
number of vehicle types and manufacturers and
over a period starting from 1960 through to 2002.
Data Collection

200

Over the past three decades, a wealth of crash test


data has been made available from NCAP,
regulatory and laboratory crash tests. A literature
and web search was carried out to collect as many
frontal rigid barrier crash test results as possible.
As a result, 1368 crash test points were collected
from the database of the National Highway Traffic
Safety Administration (NHTSA) [14], 50 crash
tests were found from Australian NCAP [15], and
38 crash tests were obtained from some
publications [6, 7, 9].

EAF

150

100

50

0
0

10

15

20

25

30

Crush (inches)

Figure 10. Energy Factor vs crush for 1974


model Plymouth Satellites [5].

Data Analysis

As rebound velocities are not available for all tests,


the EAF method was used where

Jiang 5

2E
M (V / 3.6) 2
=
w0
w0

(12).

where the crush width w0 was taken as the vehicles


overall width obtained from vehicle specification
data sheets. Average crush depth was used in this
analysis and was calculated based on a minimum of
at least three crush measurements. However, in
most cases, average crush was calculated from six
measurements (C1~C6) [16, 17].
The data was first grouped according to the
following body styles: passenger cars, vans, pickup
trucks and four-wheel-drive vehicles (4WDs).
Figures 12 through 15 show the plots of EAF
versus average crush for these vehicles.
700
600
EAF

500
400
300
200
100
0
0

300

600

900

1200

Average Crush (mm)

Figure 12. EAF vs crush for passenger cars.


700

EAF

EAF =

700
600
500
400
300
200
100
0
0

300

600

900

1200

Average Crush (m m)

Figure 15. EAF vs crush for 4WDs.

As can be seen from Figure 12, an approximate bilinear trend can be observed for all passenger cars.
With regards to vans, pickup trucks and 4WDs,
crash data over 56 km/h is not available, and there
are only several crash tests available at low impact
speeds. Nevertheless, a linear trend up to 56 km/h
is evident as shown in Figures 13, 14 and 15.
Crash data of passenger cars was further grouped
according to engine configurations. Data was
segregated and graphed for cars with the same
engine placement (transverse or inline), the same
number of cylinders (4 cylinders, V6 cylinders,
Straight 6 cylinders or V8 cylinders) and with
similar engine capacity. Only the data for cars that
were crash tested over a range of impact speeds are
shown here. Where cars were only tested at two
speeds (48 km/h and 56 km/h) data was omitted as
these were effectively single point tests.

600
EAF

500
400
300
200
100
0
0

300

600

900

1200

Figure 16 shows the data plotted for cars that have


a 4-cylinder transverse engine with engine
capacities ranging from 1.0 L to 1.9 L. All data is
clearly located in the same narrow band that has a
bi-linear trend. Similarly, data for cars with a 4cylinder inline engine and for cars with a straight 6cylinder transverse engine exhibit a bi-linear trend,
as shown in Figures 17 and 18 respectively.

Average Crush (mm)

Figure 13. EAF vs crush for vans.

EAF

EAF

700
600
500
400
300
200

700
600
500
400
300
200
100
0

1.6L
1.7L
1.8L
1.9L
1-1.5L

100
0
0

300

600

900

1200

Average Crush (m m)

Figure 14.

300
600
900
Average Crush (m m )

1200

Figure 16. EAF vs crush for cars with a 4cylinder transverse engine (1.0L~1.9L).

EAF vs crush for pickup trucks.

Jiang 6

600
500
EAF

EAF

700

700
600
500
400
300
200
100
0

400

2.5L

300
200
100
0

300
600
900
Average Crush (m m )

1200

Figure 17. EAF vs crush for cars with a 4cylinder inline engine (1.3L~2.6L).

300
600
900
Average Crush (mm)

1200

Figure 20. EAF vs crush for cars with a 4cylinder transverse engine (2.5L).

DETERMINING F(C) MODELS


700
600
EAF

500
400
300

S6

On the basis of the literature review and the data


presented in Figures 12 to 20, a strategy for
determining a vehicles frontal crush force F(C)
and the coefficients A and B in Equation 5 is
proposed here.

200

When A Range Of Crash Test Points Are


Available

100
0
0

300
600
900
Average Crush (mm)

1200

Figure 18. EAF vs crush for cars with an S6cylinder transverse engine (2.5L~3.8L).

Figure 19 shows data for cars that have a V8cylinder inline engine with engine capacities
ranging from 4.9 L to 6.9 L. The maximum impact
speed in this case is 57 km/h. Figure 20 shows data
for cars that have a 2.5L 4-cylinder transverse
engine where the maximum impact speed is 56.6
km/h. Both data plots indicate that a linear
relationship exists between EAF and average crush.

ECF =

2E
=
w0

M (V 2 VR2 )
w0

(13).

where VR is the rebound speed (km/h) and ECF is


the Energy of Crush Factor ( N ).

700
600
5.0L

500
EAF

The most accurate way to determine the F(C) of a


given vehicle model is to conduct a series of full
frontal rigid barrier crash tests at different speeds.
Obviously, from a financial perspective, this is the
most expensive method. If such crash test data is
available, plot EAF versus residual crush, or plot
the Energy of Crush Factor (ECF) versus residual
crush, as outlined by Kerkhoff et al [18], if the
rebound speed is available. ECF is expressed as

5.2L

400

5.7L

300

If the data plot is linear, Equation 5 can be used


where the stiffness coefficients are

5.8L

200

5.9-6.9

100

4.9L

0
0

300
600
900
Average Crush (mm)

B = ( slope) 2
A = EAF0 B

(14).
(15).

1200

Figure 19. EAF vs crush for cars with a V8cylinder inline engine (4.9L~6.9L).

where slope is the slope of the graphed line and


EAF0 is the intercept with the vertical axis [5].
An alternative is to use a V-C plot (e.g. Figures
1 and 2). The least squares method is
recommended to obtain the best-fit line of the
plotted data. b1 and b0 in Equation 4 is the slope
of the straight line and the intercept of the line
with y-axis respectively. The coefficients A and
B can then be determined from Equations 6and7.

Jiang 7

If the data plot shows a quadratic trend, the


constant F(C) equation can be used. In such
cases, construct a plot of energy (E/w0) versus
residual crush. A true constant force model
would appear as a straight line and the slope of
the line is the constant crush force value [5].
When Only One Crash Test Data Point Is
Available

Most cars only have a 48 km/h impact test and/or a


56 km/h NCAP crash test. In such cases where the
relationship between the EAF and crush is not
known, a bi-linear F(C) model can be used.
The first stage of the bi-linear F(C) model can be
used up to 56 km/h. b0 is usually set at 2.2 m/s (8 km/h
or 5 mph) on the basis of some crash test data, and
this has been commonly accepted in accident
reconstruction literature [5,6,9]. If either a 48 km/h
test or a 56 km/h test is available and the vehicles
crush profile is uniform, b1 from Equation 4 is

(V / 3.6) b0
b1 =
C

ERROR BANDS

It should be noted that errors are inevitable using


the strategy outlined in the previous section to
determine F(C) and its coefficients. This is
particularly so when using only one crash test data
point. Figure 21 shows a plot of impact speed
versus crush for 1971~1974 model Chevrolet
Vegas vehicles [1]. A linear V-C relationship can
be derived from the range of test data as shown in
Figure 21. Assuming that only one 48km/h (30
mph) crash test point is available and it happens to
be the test with minimum crush depth, a linear V-C
line for a 2.2 m/s (8 km/h or 5 mph) intercept (b0)
shown as Line 1 in Figure 21, can be plotted.
Alternatively, another V-C line (Line 2) can be
obtained if the crash test point happens to be the
test with maximum crush depth. Obviously, if Line 1
or Line 2 is used to derive A and B and to predict
Delta V, errors can be expected.
60
Impact Speed (mph)

If the data plot displays a bi-linear relationship,


then the linear F(C) equation (Equation 5) can
be used in two stages. The same method as
presented above can be used to determine the
coefficients A1 and B1 for phase one and A2 and
B2 for phase two.

50
Line 2

40
30

Line 2

20

(16).

10

If the crush profile is not uniform and is measured


via 5 equal crush width zones (C1~C6) [16, 17],
Neptune et al [19] propose that b1 can be calculated
such that

0
0

10

20

30

40

50

Residual Crush (inches)

2 / 3

(17).

where: = C1 + 2(C 2 + C3 + C 4 + C5 ) + C 6

= C12 + 2(C22 + C32 + C42 + C52 ) + C62 +


C1C2 + C2C3 + C3C4 + C4C5 + C5C6

Coefficients A and B for the first linear phase can


then be determined using Equations 6 and 7.
For the second stage of the bi-linear model, where
impact speeds are over 56 km/h and up to 80 km/h,
Varat [7] recommends setting a new intercept with
'

the y-axis ( b0 ) at 6.7 m/s (24 km/h or 15 mph)


according to crash test data he analysed. Using

b0' =6.7 m/s and the 48 km/h crash test point, b1'
can be obtained via Equation 16 or 17. Similarly,
coefficients for the second phase linear equation
can be determined using Equations 6 and 7.

Figure 21. Impact speed versus crush for 1971~


1974 model Chevrolet Vegas [1].

Crash test data also indicate that even for the same
model car tested at the same impact speed, the test
results differ, as shown in Figure 22. To illustrate
the effect of test data variation (scatter) on the
estimation of Delta V, an example is given here.
2001 Toyota Echo 1.5L
1999 Hyundai Accent1.5L
1999 GM Saturn 1.9L
1999 Audi A8 3.7L

Impact Speed (km/h)

(b0 ) 2 20 (b02 (V / 3.6) 2 ) / 3

b1 = b0 +

70
60
50
40
30
20
10
0
0

300
600
900
Average Crush (mm)

1200

Figure 22. Comparison of average crush at


the same impact speed.

Jiang 8

Two data points were selected from the NHTSA


database (Test Numbers 3127 and 3113) [14]. The
crash data points represented results from full
frontal crashes of 1999 model GM Saturn sedans
into a rigid concrete barrier. This model car has a
1.9L 4-cylider transverse engine. The mass, crush
width and average crush depth for the first crash
(Test 3127) were assumed as known (M1=1242 kg,
w01=1.685m and C=0.468 m), whereas the impact
speed had to be determined. The second data point
(Test 3113) was used to estimate the coefficients
b1, A and B in order to estimate the impact speed
for the first unknown crash.

However, the actual Delta V for the first data point


was in fact 48.1 km/h. The predicted Delta V
overestimates the actual Delta V by nearly 26%.
Therefore, when using one crash test data point to
predict Delta V, care must be exercised to evaluate
the accuracy of the test data.
Figure 16 shows the data band for cars with a 4cylinder transverse engine (1.0L~1.9L). A low
F(C) line, an average F(C) line and a high F(C)
line is graphed in Figure 23. These F(C) lines were
also used respectively to estimate Delta V of the
above 1999 model GM Saturn example. Figure 24
shows the comparison of the actual Delta V and
predicted Delta V using the different F(C) lines.

The mass of the car representing the second


known data point was 1181 kg, its width was
1.682 m, its impact speed was 48.3 km/h and the
average crush depth was 0.350 m. Using Equation
16 and b0=2.2 m/s (8 km/h),
b1 =

As can be seen, for a car with a 4-cylinder 1.0L to


1.9L transverse engine crashing into a rigid barrier,
when the frontal crush is 0.468 m, the possible
Delta V ranges from 41.2 km/h to 72.1 km/h and
the average Delta V is 56.7 km/h. In other words, if
a high F(C) line was used from a single data point,
the error could be as much as 50%. This data band
as well as all other data plots for cars with different
engine configurations can be used as a guidance for
determining the possible Delta V range, particularly
when the scatter of the crash test is uncertain or no
crash test is available.

-1
(V / 3.6) b0 (48.3 / 3.6) 2.2
=
= 32.0 (ms /m)
C
0.350

From Equations 6 and 7, the coefficients A and B


are

Mb0 b1 1181 2.2 32.0


=
= 49431 (N/m)
1.682
w0
Mb12 1181 32.0 2
2
=
= 718992 (N/m )
B=
w0
1.682
A=

CONCLUSION

Using Equations 1 and 5, and assuming zero


restitution, we have
E=

w0 C
1
M ( Delta V ) 2 = F (C )dCdw
0
0
2

w0

( A + BC )dCdw

w0

= ( AC +
0

1
1
BC 2 + G )dw = w0 ( AC + BC 2 + G )
2
2

where G = A2/2B [1, 5]. Hence, Delta V can be


determined such that

Delta V =

2w0 ( AC +

1
A2
BC 2 +
)
2
2B
M

(18).

Thus, the predicted Delta V of the car crashed


representing the unknown first data point is
Delta V =

2 1 .685 49431 0 .468 +

1
49431 49431
718992 0 .468 0 .468 +
2 718992
2
1242

= 16.8 (m/s) = 60.5 km/h

Theoretically, there is no unique frontal stiffness


equation that can represent all vehicle models
because of the wide diversity of vehicle frontal
structures and their complex crush behaviour. In
practice, unless the stiffness equation for a
particular vehicle can be determined via a range of
crash test data points, a linear stiffness equation can
be used for impact speeds of up to 56 km/h and a
bi-linear model can be adopted for high severity
collisions for impact speeds ranging from 56 km/h
to 80 km/h. In most cases, when using only one
crash test point to determine the coefficients of the
linear F(C) equation to estimate Delta V, extreme
care must be exercised when the crush versus speed
scatter is uncertain. Determining the possible Delta
V range is a useful guidance in Delta V estimations
for crash reconstructions. This conclusion would be
of particular interest to research centres relying on
this methodology to estimate Delta V during the
collection of real world data.
What is of particular importance is the need for
frontal crash tests of common vehicles over a large
range of speeds. National authorities should
provide funding for tests because such data is being
used in Civil and Criminal legal cases every day
around the world. It is also essential for safety
research and design of roadside barrier systems and
impacts into structures.

Jiang 9

Energy Factor vs Crush of Cars with Transverse Engines


1.6L

1.7L

1.8L

1.9L

1-1.5L

700
600

High F(C) =80473 + 976144 C

EAF

500
400
300

Average F(C) = 48100 + 611524 C

200
100

Low F(C) =23926 +330625 C

0
0

200

400

600

800

1000

1200

Crush (mm)

Figure 23. EAF vs crush of cars with a 4-cylinder transverse engine (1.0L~1.9L) and its data band.

90

72.1
(50 %)

80

60.5
(26 %)

Delta V (km/h)

70
60

56.7
(18 %)

48.1

41.2
(14 %)

50
40
30
20
10
0
Actual result

From one crash


test data point

Using average
F(C) line

Using high F(C)


line

Using low F(C)


line

Figure 24. Comparison of the actual Delta V and predicted Delta V using different F(C) lines.

Jiang 10

REFERENCES

[1] Campbell, K.L., "Energy Basis for Collision


Severity", SAE Technical Paper 740565, 1974.
[2] Robinette, R.D., R.J. Fay, and R.E. Paulsen,
"Delta-V: Basic Concepts, Computational
Methods, and Misunderstandings", SAE
Technical Paper 940915, 1994.
[3] NHTSA, CRASH 3 User's Guide and Technical
Manual, U.S. Department of Transportation,
National
Highway
Traffic
Safety
Administration, National Center for Statistics
and Analysis, Accident Investigation Division,
Washington, D.C. 20590, 1982.
[4] EDCRASH Program Manual (Version 2),
Engineering Dynamics Corporation, Lake
Oswego, Oregon, USA, 1989.
[5] Strother, C.E., et al., "Crush Energy in Accident
Reconstruction", SAE Technical Paper 860371,
1986.
[6] Navin, F. and M. Macnabb, "Crash 3 and
Canadian Test Data", SAE Technical Paper
870499, 1987.
[7] Varat, M.S., S.E. Husher, and J.F. Kerkhoff,
"An Analysis of Trends of Vehicle Frontal
Impact Stiffness", SAE Technical Paper
940914, 1994.
[8] Neptune, J.A., "A Comparison of Crush
Stiffness Characteristics from Partial-Overlap
and Full-Overlap Frontal Crash Tests", SAE
Technical Paper 1999-01-0105, 1999.
[9] Bellion, P., "Frontal Stiffness Coefficients of
Australia Passenger Cars for Use with
EDCRASH", in ICrash Conference 2002,
Melbourne, Australia, 2002.
[10] Emori, R., "Analytical Approach to
Automobile Collisions", SAE Technical Paper
680016, 1968.
[11] Wood, D.P., M. Doody, and S. Mooney,
"Application of a Generalised Frontal Crush
Model of the Car Population to Pole and
Narrow Object Impacts", SAE Technical Paper
930894, 1993.
[12] Futamata, T., "Crash Simulation Methods for
Vehicle Development", in 12th E.S.V.
Conference, Coteborg, 1989.
[13] Toyama, A., "Numerical Analysis of Vehicle
Frontal Crash Phenomena", SAE Technical
Paper 920357, 1992.
[14] NHTSA, Vehicle Crash Test Database. 2002,
National
Highway
Traffic
Safety
Administration
(http://wwwnrd.nhtsa.dot.gov/database).
[15] Crash Rating Report (Volume 1), New Car
Assessment Program, Australia, 1993~1994.
[16] Tumbas, N.S. and R.A. Smith, "Measuring
Protocol for Quantifying Vehicle Damage from
an Energy Basis Point of View", SAE
Technical Paper 880072, 1988.

[17] Collision Deformation Classification - SAE


J224 MAR80, Society of Automobile
Engineers, Warrendale, PA, USA, 1980.
[18] Kerkhoff, J.F., et al., "An Investigation into
Vehicle Frontal impact Stiffness, BEV and
Repeated Testing for Reconstruction", SAE
Technical Paper 930899, 1993.
[19] Neptune, J.A., G.Y. Blair, and J.E. Flynn, "A
Method for Quantifying Vehicle Crush
Stiffness Coefficients", SAE Technical Paper
920607, 1992.

Jiang 11

You might also like