International Journal of Crashworthiness

ISSN: 1358-8265 (Print) 1754-2111 (Online) Journal homepage: www.tandfonline.com/journals/tcrs20

Car-car crash compatibility: development of crash

test procedures in the VC-Compat project

Robert Thomson, Mervyn Edwards, Tiphaine Martin, Cor van der Zweep,
Richard Damm & Giancarlo della Valle

To cite this article: Robert Thomson, Mervyn Edwards, Tiphaine Martin, Cor van der Zweep,
Richard Damm & Giancarlo della Valle (2007) Car-car crash compatibility: development of crash
test procedures in the VC-Compat project, International Journal of Crashworthiness, 12:2,
137-151, DOI: 10.1080/13588260701433420

To link to this article: https://doi.org/10.1080/13588260701433420

Published online: 24 Aug 2007.

Submit your article to this journal

Article views: 251

View related articles

Citing articles: 1 View citing articles

Full Terms & Conditions of access and use can be found at

https://www.tandfonline.com/action/journalInformation?journalCode=tcrs20
 Car-car crash compatibility: development of
crash test procedures in the VC-Compat project
doi:10.1080/13588260701433420

Robert Thomsona, Mervyn Edwardsb, Tiphaine Martinc,

Cor van der Zweepd, Richard Damme
and Giancarlo della Vallef
a
Chalmers University of Technology, SE412 96 Gothenburg, Sweden
b
TRL Ltd, Crowthorne House, Wokingham, Berkshire, RG40 3GA, UK
c
UTAC, Autodrome de Linas Montlhery, 91311 Montlhery cedex, France
d
TNO Automotive, Schoemakerstraat, 97, P.O. BOX 6033, 2600 Ja Delft, The Netherlands
e
BASt, Bruederstrasse 53, 51427 Bergisch Gladbach, Germany
f
Fiat Auto S.p.A., Via Fausto Coppi 2, 10043 Orbassano (TO), Italy

Abstract: A major component of the EU Fifth Framework Programme sponsored project “Improvement
of Vehicle Crash Compatibility Through the Development of CrashTest Procedures” (VC-Compat)
focused on car-to-car frontal crash compatibility. The work program was composed of four main
activities, a structural survey, cost-benefit analyses, crash testing, and supporting modeling work. All
these activities focused on the development of two candidate test procedures, namely the full-width
deformable barrier (FWDB) and progressive deformable barrier (PDB), which are capable of assessing
a car’s structural interaction potential. These tests have different approaches; the FWDB assessment
is based on load cell wall force measurements, whereas the PDB assessment is based on deformation
measurements. This work supports the activities of the European Enhanced Vehicle Safety Committee
working group on frontal impact and compatibility, which has the task to propose draft test procedures
to assess a vehicle’s crash compatibility in 2007.

Key words: Crash testing, vehicle safely, compatibility.

INTRODUCTION frontal structures are more effectively utilized in car-to-car

collisions. This should help reduce compartment intrusion
Following the introduction of the European frontal and in severe accidents, and thereby lead to a decrease in the
side impact directives and EuroNCAP, car safety has number of serious and fatal injuries.
taken a major step forward. Even so, there were still more Compatibility is a complex issue that is better analyzed
than 38,000 fatalities and 1.6 million injured people be- by separating it into three subtopics: structural interaction
cause of traffic accidents in the EU15 member states for (SI), frontal force levels, and compartment strength. SI de-
2002 [1]. Passive safety equipment operates well under scribes how well vehicles interact during frontal impacts.
idealized crash test conditions. However, behavior of car If the SI is poor, the energy-absorbing front structures of
structures and safety systems during real-world conditions the vehicle may not function as designed, leading to a risk
is not always directly comparable to crash tested behavior, of compartment intrusion at lower than designed impact
especially in car-to-car crashes. Compatibility aims to fur- severities. Frontal force levels are essentially dependent
ther improve frontal impact protection by ensuring that car on vehicle mass [2] and designed crush zone length. Since
most vehicles exhibit a similar ride down distance (600–
700 mm) in the current offset test [2], small vehicles have
Corresponding Author:
lower force levels than large vehicles. This leads to an over-
R Thomson crushing of small vehicles when they collide with large ve-
Chalmers University of Technology, hicles – they are unable to deform the heavier vehicle with
SE412 96 Gothenburg, Sweden higher force levels. Ideally, matched frontal force levels
Tel: +46 (0)31 772 3645; Fax: +46 (0)31 772 3690 would ensure that both vehicles in an impact absorb their
Email: [email protected]
share of the kinetic energy. This would reduce the risk of

Copyright

C Taylor & Francis Group, LLC 137 TCRS 2007 Vol. 12 No. 2 pp. 137–151
R Thomson, M Edwards, T Martin, C van der Zweep, R Damm and G della Valle

injury for the occupant in the lighter vehicle. Compartment didate test procedures that have been in development prior
strength is closely related to frontal force levels, but is nev- to the VC-Compat project:
ertheless distinguished in compatibility discussions since
• Full-width deformable barrier (FWDB): 100% overlap
it is an important issue for self-protection. In cases where
test at 56 km/h
the vehicle front structures do not absorb the amount of • Progressive deformable barrier (PDB): 50% overlap test
energy as designed – or in cases where the vehicle is ex- at 60 km/h
posed to higher impact severity than it is designed for – the
compartment strength needs to be sufficiently high to resist A compatibility test approach must evaluate SI, vehicle
a compartment collapse. A strong occupant compartment force levels, and occupant compartment strength. The
is necessary to support the front-end structures during a FWDB was primarily designed to assess SI, and the devel-
frontal crash and ensure that the deformation can occur in opers propose it to be used in conjunction with an offset
a stable, predictable manner. deformable barrier (ODB) test at 64 km/h to test force
The work program in “Improvement of Vehicle Crash levels, and potentially a high-speed ODB test at 80 km/h
Compatibility Through the Development of Crash Test to test compartment strength [5]. The PDB is claimed to
Procedures” (VC-Compat) has been designed to sup- address all three aspects of compatibility by its develop-
port the activities of the European Enhanced Vehicle ers [6]. The International Harmonised Research Activity
Safety Committee (EEVC) frontal impact and compati- group for frontal impact has previously recommended the
bility working group (Working Group 15, WG15), which global use of both an offset and a full-width test for assess-
aims to propose draft test procedures to assess a vehicle’s ing a car’s self-protection capability in frontal impact [7].
crash compatibility to the EEVC Steering Committee in This ensures that the car is not optimized to one particular
2007. Recently, the EEVC WG15 has defined a route map crash configuration. Thus, the PDB test should be incor-
to improve frontal impact compatibility. The general ob- porated with a full-width barrier test to assess occupant
jectives of the route map are to: restraint system performance. Further information of the
test approaches is provided in the following sections.
• address partner and self-protection without decreasing The VC-Compat project has focused its research re-
current self-protection levels;
sources on the development and evaluation of the two
• keep number of procedures to a minimum; and
proposed test methods to assess SI. This concept is the
• internationally harmonize procedures;
most challenging issue to address through objective assess-
The short-term objectives are to develop requirements ment criteria. Thus, understanding and assessing SI have
to: been the main activities followed by limited activities to
identify the force levels of current vehicles and propose
• improve SI; new force-level requirements for future test procedures.
• ensure that frontal force mismatch (stiffness) does not in-
No investigation of a test of compartment strength has
crease and compartment strength does not decrease from
been conducted within the project. As a final deliverable,
current levels.
VC-Compat will propose a suite of draft test procedures
The medium term objectives are to develop requirements and associated performance criterion outlines to assess and
to: control car frontal structures for frontal impact compati-
bility. At this stage, the best combination of tests still has
• improve compartment strength, especially for light vehi-
to be determined and it could include both the FWDB and
cles;
PDB tests.
• Take first steps to improve frontal force matching; and
• further improve SI.
FWDB test
The main activities in the VC-Compat project have been The FWDB test has a deformable element and uses mea-
designed to address the short-term objectives of the EEVC surements from a high-resolution load cell wall (LCW) to
WG15. This is achieved through a work program that assess a car’s SI potential and has been described previously
mainly investigates vehicle SI and uses vehicle crash test- [5]. The premise is that cars exhibiting a more homoge-
ing and computer modeling. Cost-benefit analyses are also neous force distribution on the LCW should have a better
included to ensure that the relevance and motivation for SI potential. Metrics that have been proposed to assess
new test procedures are warranted. A description of the SI based on force distribution include the relative homo-
VC-Compat project is available from the project Web site geneity criterion (RHC) and the average height of force
[3] and from an earlier publication [4]. (AHOF). The development of these metrics has been re-
ported previously [2, 5].
CANDIDATE TEST PROCEDURES More recently, a new SI criterion has been developed to
resolve a number of issues with the RHC. The criterion is
The VC-Compat project is evaluating and developing the based on the peak cell loads recorded in the first 40 ms of
compatibility test procedures in line with EEVC WG15 the impact. This time interval should allow the detection of
priorities. The EEVC WG15 has been evaluating two can- structures up to 400 mm into the vehicle, which is adequate

TCRS 2007 Vol. 12 No. 2 138 Copyright

C Taylor & Francis Group, LLC
 Car-to-car crash compatibility

for detection of subframe load paths. In comparison with A proposal for implementing the SI criterion suggests three
using peak cell loads recorded through the duration of the phases of application:
impact (previous RHC criteria), this has the advantage of
assessing SI at the beginning of the impact and minimizing • Phase 1: The vertical and horizontal components of the cri-
the loading applied by structures further back into the ve- terion are applied over assessment Area 1 to ensure that all
hicle such as the engine. In addition, 400 mm aligns with vehicles have adequate structure in a common interaction
the revised NHTSA proposal to assess the AHOF over the zone.
initial 400-mm vehicle displacement. The criterion con- • Phase 2: In addition to the requirement of Phase 1, the ver-
sists of a vertical SI (VSI) component and a horizontal tical component of the criterion is applied over assessment
SI (HSI) component. The following concepts formed the Area 2 to encourage vehicles to spread their load better
basis for the SI criterion: vertically.
• Phase 3: In addition to the requirements of Phase 1 and
Vertical assessment concepts Phase 2 the horizontal component of the criterion is ap-
• Provide a minimum level of support within a common in- plied over assessment Area 2 to encourage better crossbeam
teraction zone. This is to ensure that high vehicles have structures.
sufficient structure in alignment with a common interac-
tion area to interact with cars. PDB test
• Encourage greater vertical load balance and minimum sup-
port within a defined interaction zone, which includes the The PDB test is a 50% overlap offset test that uses mea-
common interaction area. The results of tests performed surements from a PDB to assess a car’s compatibility in
as part of the VC-Compat project have shown that vehicles terms of partner and self-protection [8]. The barrier stiff-
that spread their load well vertically have better SI than ness increases with depth and has upper and lower load
those that do not. levels to represent an actual car structure. The progres-
sive stiffness of the barrier has been designed so that the
Horizontal assessment concepts Equivalent Energy Speed (EES) for the vehicle should be
• Encourage strong crossbeams to adequately distribute the independent of the vehicle’s mass. The use of a PDB bar-
lower rail loading. The aim is to ensure that cars have rier should thus harmonize the test severity among vehicles
adequate bumper crossbeam structures. of different masses by encouraging lighter vehicles to be
• Encourage wider structures to prevent lateral misalign- stronger without increasing the force levels of large vehi-
ment in low overlap impacts. The results of tests analyzed cles. The reader is referred to [8] for more information on
as part of the VC-Compat project have shown dynamic the PDB performance.
lateral misalignment to be a problem. The philosophy behind the PDB test is to assess both
To allow manufacturers to gradually adapt vehicle designs a car’s SI, potential frontal force level, and give informa-
to become more compatible, the criterion could be applied tion on the vehicle’s compartment strength in the same
in three phases using two assessment areas. test [8]. Laser scanning techniques are used to measure the
three-dimensional (3D) barrier deformations. The PDB
• Assessment Area 1 – Rows 3 and 4 of the LCW (330–580 mm): metrics are under development and different parameters
This is the common interaction zone and encompasses are available: homogeneity; the average height of deforma-
the FMVSS Part 581 bumper beam zone (406–508 mm tion (AHOD) – comparable to the AHOF in LCW tests;
from the ground level). The vertical component of the and the average depth of deformation (ADOD) metrics.
criterion (VSI Area 1) assesses whether the vehicle has The AHOD and ADOD are simply weighted averages of
structure capable of generating a minimum load within the barrier deformation in the longitudinal and vertical
the common interaction zone. The requirement is that the directions and are quite straightforward to correlate with
vehicle applies a load to Rows 3 and 4 of the LCW that is compatibility assessment concepts. The calculation of the
greater than the target row load. A mass-dependent target
homogeneity parameters is a new development based on
row load is proposed up to a maximum of 100 kN. The
the gradient method to detect local variations of the defor-
horizontal component of the criterion (HSI Area 1) assesses
mation map of the barrier. For a given investigation area,
the lateral load distribution within the common interaction
zone, specifically the ability of the crossbeam to distribute the total variation of the deformation gradient is defined
the lower rail loads and provide outer support. and normalized by an estimate of the magnitude of the
• Assessment Area 2 – Rows 2 to 5 of the LCW (205–705 mm): deformation map. This allows the resulting rating to be
This is an extension of the common interaction zone to independent of the average depth of the deformation. The
include Rows 2 and 5. The vertical component of the cri- hypothesis used by the gradient method is that large fluc-
terion (VSI Area 2) assesses whether the vehicle has struc- tuations in deformation gradients indicate inhomogeneous
ture capable of generating a minimum row load within loading of the vehicle structure and lower SI potential.
the enlarged assessment area and also assesses load balance The PDB test metrics are based on the longitudinal
within the enlarged assessment area. The horizontal com- and vertical deformation patterns of the barrier face, the
ponent of the criterion (HSI Area 2) assesses the lateral load cells recording the global crash forces, and measure-
load distribution within the enlarged assessment area. ments from the vehicle. In principle, the uniformity of the

Copyright

C Taylor & Francis Group, LLC 139 TCRS 2007 Vol. 12 No. 2
R Thomson, M Edwards, T Martin, C van der Zweep, R Damm and G della Valle

Figure 1 Progressive deformable barrier test development program [8].

barrier deformation gives a measure of the vehicle’s SI po-

tential and the longitudinal barrier deformation, combined
with force measurements, indicates its frontal force levels.
The development of the PDB test procedure is schemat-
ically presented in Figure 1. The procedure uses tools
(deformable barrier, load cells, and the vehicle itself) to
provide measurements. From these measurements, a set of
assessment parameters is used to identify desirable vehicle
performance.

VC-COMPAT WORK PROGRAM

Figure 2 Vertical positions of significant structural
There are four activities that provide the technical basis components.
for the research:
• A structural survey to create a database of positions and of the main crashworthy structures of the European fleet.
dimensions of the important energy-absorbing structures Information contained in the structural database has been
in vehicles. This will be used to help determine appropriate helpful to understand the results obtained in car-to-car
SI assessment areas for vehicles. and car-to-barrier testing. The database provides the po-
• Accident analysis to estimate the benefit and cost of im- sitions of the main frontal structures that must engage in
proved compatibility.
car-to-car impacts to ensure good SI. A typical analysis is
• Mathematical modeling to support the development of the
shown in Figure 2, where the vertical position of the vehi-
test procedures and the cost benefit analysis.
• A crash testing program of car-to-car and car-to-barrier
cle structures can be described in terms of the maximum,
crash tests to develop and evaluate the crash test procedures minimum, average, and weighted average values. Similar
and appropriate performance criteria. analyses for the lateral position and sectional dimensions
have been conducted. These measurements will be a help
The results of these four activities will be brought together to define an interaction and investigation area for the as-
in another activity to synthesize the crash test procedures. sessment area used for the test criteria. For example, an
In addition, a dissemination activity is communicating the assessment area would encompass a vertical range between
results and findings from this project and soliciting input about 180 and 800 mm to include all the subframe, main
from industry. rail, upper rail, and wheel sill load paths.

Structural survey Cost-benefit study

The structural survey activity was performed by the Research of the benefits and costs of improved compatibil-
UTAC and has been reported previously [4]. The main ity for frontal impact has been led by BASt. The structure
goal of this activity was to identify the geometric position of the benefit study is presented in Figure 3. The UK

TCRS 2007 Vol. 12 No. 2 140 Copyright

C Taylor & Francis Group, LLC
 Car-to-car crash compatibility

Figure 3 Benefit study structure.

activities have utilized data from the Co-operative Crash smaller CCIS dataset to the national level. This provided
Injury Study (CCIS) and the German research was based the reduction in number of annual fatalities and severe
on the German In-Depth Accident Study (GIDAS). injuries for the United Kingdom.
Using the available national and detailed accident data A different analysis approach was employed by BASt
in the United Kingdom and Germany, an estimate of an for the German dataset. Method 2 did not involve the re-
European “target population” for improved compatibil- moval of individual injuries from the injury report for the
ity was made. The target population was defined as those accident cases reviewed in the target population. Instead,
casualties that are likely to experience a reduced risk of in- injury risk curves (probability of injury versus EES) were
jury as a result of the implementation of improved frontal developed using the existing accident data. Probabilities
compatibility measures. Using the European numbers for for no injury, slight injury, severe injury, and fatal injuries
annual road traffic trauma victims and scaling the target were developed and compared with the measured EES dis-
populations identified in German and UK data analyses, tribution in the GIDAS database. The injury risk curves
the following upper (optimistic) and lower (pessimistic) were modified using an assumption that improved compat-
boundary estimates were made [4]: ibility would increase the energy absorption in the vehicle
structures. Using compartment collapse as reference cri-
• Between 3466 (14%) and 7675 (31%) fatally injured car oc-
teria, real-world vehicle performance was compared with
cupants are within the “compatibility target population.”
• Between 50,260 (29%) and 90,122 (52%) seriously injured laboratory test results. From this analysis, improved com-
car occupants are within the “compatibility target popula- patibility should allow vehicles to absorb about 28% more
tion.” energy in a car-to-car crash that would then lead to an
increased EES threshold before serious or fatal injuries
From the target populations, the expected benefit for im- occurred. Similar to the UK data, the benefits estimated
proving vehicle crash compatibility was calculated. Two using the German data were scaled to the national level to
different approaches were employed for the different data obtain the reduction in serious and fatal injuries.
sources available. TRL Ltd analyzed the CCIS dataset Results for the two analyses are summarized in Table 1.
on the basis of the hypothesis that improving compatibil- The United Kingdom and Germany used fundamentally
ity will lead to predictable impact energy absorption in different approaches to the analysis, but the general results
the frontal structures and little compartment intrusion for are the same. These results must be treated as estimates
frontal crashes. The resulting improvements in crash per- knowing the limitations of the data available for the analysis
formance will lead to reduced occupant injuries, which are and the general assumptions used to predict benefits.
predicted using two assumptions:
• Pessimistic (lower) estimate: Eliminate injuries caused by
contact with an intruding front interior structure if ETS Mathematical modeling
<56 km/h. There were three main computer simulation tasks in the
• Optimistic (upper) estimate: Eliminate injuries caused by VC-Compat project that were coordinated by the TNO
contact with the front interior (with or without intrusion) Automotive.
if ETS <56 km/h.
Using these assumptions, relevant cases in the CCIS • Finite element (FE) barrier modeling to support the de-
database were reviewed. Injuries that theoretically would velopment and initial validation of the test procedures.
not have occurred if better compatibility existed are then • Multibody (MB) modeling methodology to develop a fleet
removed from the accident record and any change in model to support the benefit estimation and determine the
the Maximum Abbreviated Injury Score (MAIS) was re- effect of improved compatibility in other crash configura-
ported. The improved MAIS scores were then used to tions.
identify improved injury/fatality for the CCIS dataset. • MB simulation of vehicle force levels to identify strategies
The national statistics for the United Kingdom were then for force matching of vehicles with different masses and
used to scale the proportional benefit obtained in the the consequences for occupant protection.

Copyright

C Taylor & Francis Group, LLC 141 TCRS 2007 Vol. 12 No. 2
R Thomson, M Edwards, T Martin, C van der Zweep, R Damm and G della Valle

Table 1 Results of benefit analysis

Reduction in severe Reduction in fatal
Dataset injures [%] injuries [%]
Co-operative Crash Injury Study (CCIS) 4–9 6–12
German In-Depth Accident Study (GIDAS) 4 8

FE models were developed to support development of the The first major step was to evaluate the overall behavior
FWDB test. A RADIOSS FE model of the FWDB and of the total numerical fleet. This evaluation included
LCW was created by TRL Ltd. The main advantage of this seven different vehicle models in a range of accident
model is that there is a capability to simulate local tearing of configurations. At the end, the NASS-CDS injury
the honeycomb material by a stiff car structure. To achieve criteria and available vehicle data were compared with the
this, the barrier model was constructed from columns of results of the simulations to validate the methodology.
“standard” type honeycomb elements that were joined by Figure 4 gives the flow chart of this process, where the
thin “tear” type honeycomb elements. The “tear” type weight factors (indicating the relevancy of specific crash
elements have strain-based failure criteria that delete the configurations) are depicted as probability p.
element when a prescribed strain is reached. A large set of simulations was performed (>5000 runs)
Parallel to the FWDB barrier modeling, an FE model to simulate the reference fleet performance. A second fleet
of the PDB was developed by the UTAC. The model was was created where the two smallest vehicles (Geo Metro
tuned with the help of three component tests that repre- and Dodge Neon) were modified to improve compatibil-
sent the loading from a crossbeam, a single longitudinal ity. Simulations of the second fleet were performed and
and a half-rigid wall. The model is currently available in compared with the results from the reference fleet.
RADIOSS and soon to be available for Pamcrash software. To create an approximation of real-world collisions, the
A fleet model, the second part of modeling activities in accident variables “impact velocity” and “impact overlap”
VC-Compat, was developed to support the benefit estima- were varied. The initial velocity of each vehicle was within a
tion and determine the effect of improved compatibility range of 20 to 80 km/h and the overlap was varied in a range
in other crash configurations. For this purpose, TNO Au- of 25% to 80% of the smallest vehicle. The distribution
tomotive developed an MB vehicle fleet model based on of the variations was set up with the Latin Hyper Cube
seven vehicle models representative of a real-life car fleet algorithm implemented in ADVISER
C , resulting in 100

[9]. The overall objective of this study was to develop a batches that randomly generated an even distribution over
methodology that can predict the fleet wide improvement a given window for a relatively small number of samples.
of compatibility by changing the front-end design of se- Figure 5 shows the mean overall injury (ISS) values for
lected vehicles. all drivers in all scenarios. Each column represents the in-
The MB models were constructed from existing FE jury outcome for impacts between the vehicles listed on the
models already developed for common US vehicles. Front- axes. Drivers of the small vehicles (GE, Geo Metro; NE,
end structures and passenger compartments were modeled Dodge Neon) had relatively high-injury risks, particularly
in detail to provide realistic deformation modes. Dummies, when involved in collisions with the larger vehicles in the
airbags, belts, and the main interior parts such as the dash- fleet. Improvements to the vehicle compatibility (G1, Geo
board and steering wheel were included to assess occupant Metro; N2, Dodge Neon) led to lower mean overall injuries
injury. Table 2 gives an overview of the available mod- for these particular cases.
els. By simulating impacts between different combinations To show the effect of the design improvement even
of vehicles, a representation of real-life accidents can be more clearly, the following figures show the results when
made. Figure 4 shows how the models fit into the overall only the small passenger cars are included. Figure 6 shows
benefit estimation. only those experiments that included the modified ve-
The stepwise approach to investigate the fleet-wide im- hicles as the target (struck) car (Geo Metro and Dodge
provement of compatibility was based on the fleet systems Neon). The red circles in the graphs represent the original
model initially developed by the NHTSA/Volpe [10]. For numerical fleet and the blue dots represent the updated
the evaluation of the numerical fleet a real-life database is numerical fleet. The lines represent linear regression func-
needed. The NASS-CDS database was used for this study. tions of the data points. The effect of the design improve-
From this database, relevant cases and accident character- ment resulted in an overall reduction of about 35% of the
istics were selected for the design of experiments for the ISS values based on the numerical fleet, as assessed by
simulations series. the linear regression lines. The mean overall ISS score
In this study, vehicle models complemented with decreased from 6.72 to 4.59. Although there is improve-
occupant models were used to research the benefit of ment in the overall ISS score, the updated numerical fleet
improved vehicle front-ends. Real-life crash injury data shows some high ISS scores for
v values between 45 and
were compared with simulation occupant model outputs. 50 km/h.

TCRS 2007 Vol. 12 No. 2 142 Copyright

C Taylor & Francis Group, LLC
 Car-to-car crash compatibility

Table 2 Available multibody vehicle models for fleet studies

Model Class Mass [kg] Test mass [kg] Length [m] Width [m]
Geo Metro Subcompact 900 1191 3.85 1.59
Chrysler Neon Compact passenger car 1085 1371 4.36 1.71
Ford Taurus Mid-size passenger car 1488 1728 5.07 1.86
Honda Accord Mid-size passenger car 1396 1636 5.06 1.90
Ford Crown Victoria Large passenger car 1836 2076 5.30 1.96
Ford Explorer SUV 1971 2205 4.81 1.87
Dodge Caravan Full-size MPV 1682 1934 4.70 1.95

deformation zones of larger vehicles and study the new

range of stiffness levels required for small vehicles under
this new traffic condition. The baseline assumption was
that all vehicles had the same deformation zone of roughly
700 mm [2].
The results of this study of vehicle stiffness suggested
that smaller vehicles can be made stiff enough to provide
suitable safety levels in high-mass ratio impacts. The in-
creased stiffness resulted in higher accelerations for the
smaller vehicles, but impacts with mass ratios of 1:1.6 were
survivable with appropriate safety equipment designs. A
similar result was found for the investigation of fleet force
Figure 4 Flow chart of the process for comparing the levels when larger vehicles had a 50-mm longer defor-
accident database and the simulated crash scenarios. mation zone. These cases resulted in similar acceleration
levels in the smaller vehicles, but the force levels of small
vehicles still needed to be increased above current levels.
A study of frontal force levels was the third activity The results indicate that a minimum force of 350 to 400 kN
undertaken in the modeling work package. The objective should be required for all small vehicles. An implementa-
of this research was to investigate the dependency of frontal tion of this requirement for vehicles would be most critical
force level on vehicle mass in current and future tests. In for vehicles under 1400 kg. The relevance of this type of
addition, the influence of the crash pulse on the occupant requirement for small cars is highlighted by the undesir-
response must be identified so that no undesirable side able performance of some vehicles observed in the test
effects of the test procedures arise. program.
This research produced generic vehicle descriptions to
model a range of car-to-car collisions. The goal was to
find how the stiffness of vehicles could be modified so Crash testing
that impacts involving vehicle pairs with reasonable mass The objective of the crash testing activity, led by TRL
ratios could still result in survivable crash environments. Ltd, was to perform full-scale crash tests and associated
This was investigated by first increasing the stiffness of analyses to help develop and validate a suite of test pro-
smaller vehicles (<1500 kg) from their current levels. The cedures to improve car frontal impact compatibility. The
next step for this investigation was to lengthen the existing test program consists of three separate activities. The first

Figure 5 ISS distribution (mean values) for entire subset plotted as a function of target and bullet cars.

Copyright

C Taylor & Francis Group, LLC 143 TCRS 2007 Vol. 12 No. 2
R Thomson, M Edwards, T Martin, C van der Zweep, R Damm and G della Valle

Figure 6 Comparison of original numerical fleet and updated numerical fleet for the updated vehicles: (a) influence of
v;
(b) influence of overlap.

two, car-to-car tests and car-to-barrier testing, have been The tests were performed with a 50% overlap of the nar-
the main focus in the VC-Compat project. A car-to-car test rower vehicle and a closing speed of 112 km/h. The ride
program has been developed to explore the performance height difference for the supermini-to-supermini test was
of different structural designs and identify the critical pa- 60 mm to emphasize any under/override that might have
rameters for improved crash compatibility performance. occurred.
The findings in the car-to-car tests are used to identify In each of the tests, there was initial over/underride
performance features in the test candidates (FWDB and because of unstable activation of the supermini lower rail
PDB). The car-to-barrier tests were used to observe the load path. This was limited by an interaction between the
performance of the FWDB and PDB tests for a range of upper to lower rail vertical connection of the supermini and
different vehicle models. Comparisons between the car- the lower rail/crossbeam structure of the target vehicle.
to-car tests and car-to-barrier tests have been conducted This led to the conclusion that the SI assessment should
routinely by the consortium members to identify and iter- encourage good vertical connections between the upper
atively plan the test series. The third crash testing activity and lower rails. However, the stiffer main rail/crossbeam
has been to collect LCW force data in ongoing EuroNCAP structure of the target vehicle overloaded the weaker upper
tests to identify links between vehicle mass and front-end load path of the supermini, resulting in collapse of the
force levels. This has been a much smaller task than for the occupant compartment (Figure 7). This collapse of the
previously discussed activities. occupant compartment demonstrates the importance of
high compartment strength for light cars.
Car-to-car test program Although there is some evidence that SI with the two-
Car-to-car tests performed as part of the VC-Compat level load path car was better than with the one-level load
project have shown that in impacts against a similar mass path car – the subframe crossbeam of the two-level load
vehicle, the SI performance of a two-level load path small path car engaged the wheel/sill load path of the supermini –
family car was better than a one-level load path small family the collapse of the supermini’s occupant compartment in
car [4]. This was due to the greater vertical load spreading all tests prevented an accurate assessment of the difference
capability of the two-level load path vehicle design. The in the SI between the cars.
two-level load path car has a subframe and lower rails while The objective of the second of the test series was to
the one-level load path vehicle only has lower rails. Fur- investigate the effect of mass ratio and identify how low
ther to this work, two additional series of tests have been structures are required on a SUV to give good interaction
performed to investigate the effect of mass ratio and front with a car. Both SUVs had approximately the same frontal
structure height on SI. The results of these two test series force levels measured in the 64 km/h ODB test. The tests
are reported here. performed as part of this test series were:
The objective of the first of these new test series was to
investigate the effect of mass ratio and whether the per- • Two-level load path small family car to SUV1: mass ratio
formance of lighter vehicle was improved when impacted 1:1.6
against a two-level load path car compared with a one-level • Two-level load path small family car to SUV2: mass ratio
load path car. The tests performed as part of this test series 1:1.8
were:

• Supermini to supermini: mass ratio 1:1 SUV1 employed a secondary energy-absorbing structure
• Supermini to one-level load path small family car: mass (SEAS) below the main rails to promote interaction with
ratio 1:1.3 the small family car front structure. SUV2 (no SEAS)
• Supermini to two-level load path small family car: mass had primary energy-absorbing structures (PEAS) that were
ratio 1:1.3 reasonably aligned with the small family car front structure.

TCRS 2007 Vol. 12 No. 2 144 Copyright

C Taylor & Francis Group, LLC
 Car-to-car crash compatibility

Figure 7 Deformation of the supermini occupant compartment.

The tests were performed with a 50% overlap of the nar- plied loads into the stiffer parts of the opposing vehicle
rower vehicle and a closing speed of 112 km/h. structure.
For the test with SUV1, there was initial over/underride
of the opposing lower rail structures. This was countered FWDB test program
by interaction between the subframe crossbeam (SEAS) of FWDB tests have been performed with a range of vehicles
SUV1 and the lower rail to the subframe hanger of the car, covering a test mass range of 900 to 2300 kg. For the
as well as by the engagement of SUV1 wheel/sill load path vehicles used in the car-to-car test program, the assessment
with the subframe crossbeam of the car. The result was that of the one-level load path and two-level load path family
the loads applied by SUV1 were well distributed into the cars used in the car-to-car test program has been reported
occupant compartment of the car, which made the most of previously [4]. The FWDB test has been shown to be
its compartment strength. This limited the intrusion into capable of distinguishing the presence of an additional load
the occupant compartment (Figure 9). path. The assessment of the supermini and the two SUVs
For the test with the SUV2, there was dynamic lateral used in the car-to-car test program are reported here.
misalignment of the lower rails in both vehicles. The width SI potential of the vehicle is assessed on the basis of the
of front structure of SUV2 in this test was less than for the LCW peak cell force distribution. The force distribution
SUV1 in the previous test. The lower rail of SUV2 directly for the supermini and the two SUVs are shown (Figures 10
loaded the footwell of the car, resulting in penetration of and 11). Note that for the SI assessment criterion, an event
the footwell, while the crossbeam of the SUV2 loaded the time of 40 ms is used when selecting the peak cell forces. As
A-Pillar. The strong crossbeam of SUV2 in this test limited discussed earlier, there are different assessment areas for
the maximum extent of the footwell penetration by direct- the implementation of the SI criterion. The VSI and HSI
ing the lower rail loading into a stiffer part of the opposing assessments for the common interaction area (Rows 3 and
vehicle structure. Figure 8 shows the limited deformation 4, Area 1) as well as the VSI for the extended assessment
of the lower rail and crossbeam structures of SUV2. The area (Rows 2–5, Area 2) are reported here.
result of this test was higher occupant compartment intru- For the supermini, the stiff lower rails and weak bumper
sion than in the test with SUV1 (Figure 9). crossbeam result in a poor horizontal force distribution
The performance of the car in the impact against SUV1 (Figure 10). The absence of a subframe load path and the
(with the SEAS) demonstrates the importance of load paths low load applied by the upper rail load path result in a poor
below the lower rail. The SI assessment in a compatibility vertical force distribution. For the SUVs, the high loads
test procedure should encourage load paths below the lower applied by the lower rail load paths are more widely spaced
rail. The assessment should also encourage wider front for SUV1 than for SUV2 (Figure 11). This indicates that
structures to prevent lateral misalignment in low overlap SUV1 has the wider front structure. The bumper cross-
impacts and also strong crossbeams for directing the ap- beam of SUV2 gives a more uniform force distribution

Copyright

C Taylor & Francis Group, LLC 145 TCRS 2007 Vol. 12 No. 2
R Thomson, M Edwards, T Martin, C van der Zweep, R Damm and G della Valle

Figure 8 Deformation of the impact side lower rail and bumper crossbeam for SUV2.

Figure 9 Deformation of the small family car occupant compartment.

laterally, with higher loads at its center point. This indi- in Rows 3 and 4. Each of the vehicles tested has structure
cates that SUV2 has the stronger bumper crossbeam. The in alignment with the FMVSS Part 581 zone and each
vertical force distribution is similar for the two SUVs, with scored zero. This indicates that the VSI Area 1 part of the
the lower rail load paths applying significantly higher loads criterion works. Note that a load balance is not applied in
than the structures above and below the lower rails. Note Area 1.
that SUV1 was tested with the lower edge of the LCW at 50 The VSI Area 2 scores are shown in Figure 12 for
mm from ground level, compared with 80 mm for SUV2. the range of vehicles tested in VC-Compat, plotted in
For VSI Area 1, a score of zero ensures that the vehicle order of increasing vehicle mass. The score is a com-
applies a minimum row load equal to the target row load bination of minimum load and load balance measures; a

Figure 10 Supermini load cell wall force contour plot (up to 40 ms) and front structure deformation.

TCRS 2007 Vol. 12 No. 2 146 Copyright

C Taylor & Francis Group, LLC
 Car-to-car crash compatibility

Figure 11 SUV load cell wall force contour plots (upto 40 ms) and front structure comparison.

lower score indicates better vertical SI potential. The su- The HSI component of the criterion is based on the
permini tested in the car-to-car program (Supermini 2) concepts of encouraging strong crossbeams to adequately
has a relatively high score in comparison with the other distribute lower rail loading and wider front structures to
small passenger cars. This high score is primarily due to prevent dynamic lateral misalignment in low overlap im-
the minimum support component. This was expected as pacts. A lower score indicates better horizontal SI poten-
the absence of a subframe load path limited the load mea- tial. The HSI Area 1 scores are shown in Figure 13 for the
sured in Row 2, while the low load applied by the up- range of vehicles tested, plotted in order of increasing ve-
per rails limited the load measured in Row 5. The VSI hicle mass. For the Supermini 2, the outer support part of
Area 2 score for the two SUVs was also high in compar- the HSI score is low, while the crossbeam assessment part
ison to other vehicles tested. This was expected as the of the HSI score is high. This was expected, as the front
higher front structure meant that they were not able to structure width represents a greater proportion of the ve-
meet minimum load requirement for Row 2, while the hicle width for narrower vehicles, while the weak bumper
high loads applied by the lower rails resulted in poor crossbeam for this vehicle failed to adequately distribute
vertical load balance. The VSI Area 2 part of the crite- the lower rail loading.
rion correctly assessed the vertical SI potential of the test In the HSI assessment for the SUVs, the outer support
vehicles. component is higher for SUV2 than for SUV1. This was

Figure 12 Vertical structural interaction (VSI) Area 2 assessment plotted against increasing vehicle mass.

Copyright

C Taylor & Francis Group, LLC 147 TCRS 2007 Vol. 12 No. 2
R Thomson, M Edwards, T Martin, C van der Zweep, R Damm and G della Valle

Figure 13 Horizontal structural interaction (HSI) Area 1 assessment plotted against increasing vehicle width.

and previous research [6, 8], the first observations of the

PDB test performance show that:
• the barrier tool and deformation measurements are able to
detect front-end force design in terms of load path force,
position, and geometry;
• the LCW data associated with impacts without bottom-
ing out of the barrier is able to measure front end force
structural forces with more accuracy than the EEVC ODB
barrier since it reduces the effects of the engine loading
directly to the LCW; and
• the dummies and vehicle give self-protection information
similar to current procedures.

To use the barrier deformation to evaluate the SI, an in-

vestigation area must be defined. The area of the barrier
has been identified using the structural survey information
from VC-Compat. The vertical assessment range is from
Figure 14 Progressive deformable barrier assessment area.
180 to 650 mm above ground level and the horizontal area
is half the vehicle width but excludes the first 150 mm adja-
expected as the front structure width represented a lower cent to the barrier edge (Figure 14). The PDB deformation
proportion of the vehicle width for SUV2. The crossbeam is used to calculate the partner protection parameters listed
assessment component was the same for SUV1 and SUV2 previously: Homogeneity, ADOD, and AHOD. Further
despite the apparent stronger crossbeam of SUV2. The development of the assessments must be correlated to the
reason for this was that the crossbeam of SUV2 applied car-to-car test performance and the respective “level of
load to a single row, while the lower rails applied load to aggressivity” for the vehicles.
Rows 3 and 4. Therefore the load distribution was good for A preliminary indication of the performance of the PDB
one row and poor for the other row. To improve the HSI homogeneity assessment is shown in Figure 15. A range
score for SUV2 would require a crossbeam that applies of test vehicles has been arranged by ascending mass. The
load to both Rows 3 and 4. objective is to have low homogeneity scores to produce
The test results confirm that the SI criterion is appropri- improved SI. The PDB results for the same vehicles dis-
ately ranking the VSI and HSI potential of the test vehicles cussed in the FWDB evaluation and car-car test discussion
in the car-to-car test series of VC-Compat. are also identified in Figure 15 (results for small family car
1 are not available). As desired, the PDB indicates that the
PDB test program supermini has a poorer homogeneity than a similar mass
PDB assessment is based on barrier deformation for partner vehicle with better load distribution. In addition, the SUVs
protection and on vehicle and dummies performances for were ranked in the order that was to be predicted from the
self-protection. Based on the testing within VC-Compat car-car test results. The SUV with the SEAS (SUV1) had

TCRS 2007 Vol. 12 No. 2 148 Copyright

C Taylor & Francis Group, LLC
 Car-to-car crash compatibility

Figure 15 Progressive deformable barrier assessment of homogeneity.

better homogeneity than SUV2, with only one load path. influence in the ADOD because they were localized (see
This better homogeneity can be interpreted in Figure 16. Figure 16) to the crossbeam.
The deformation contours for SUV1 are, in general, more The final parameter in the PDB SI assessment is the
widely spaced than in SUV2, where there is a very wide AHOD and is also shown in Figure 17. We can see that the
pattern left by the crossbeam that creates a large gradient supermini had a very low AHOD compared with similar
field across the assessment area. vehicles, recording the lowest AHOD for all test vehicles.
The ADOD and AHOD results for the same vehicles are The SUV results indicated that the SEAS in SUV1 was
shown in Figure 17. As expected, the ADOD for heavier detected in the PDB barrier tool and produced a lower
vehicles (to the right of the graph) are greater than lighter value than the SUV with only a PEAS (SUV2).
vehicles. The PDB indicates that the supermini was pene- In summary, car-to-car testing performed as part of the
trating the barrier as much as similar vehicles in its class. VC-Compat project has shown:
The SUVs had slightly reversed results since SUV1 had a
greater ADOD than SUV2. It must be remembered that
• additional load paths improve vehicle SI performance;
the ADOD is based on a weighted average so the large • compartment strength of light cars is important in high
deformations of crossbeam of SUV2 did not have as much mass ratio impacts;

Figure 16 Progressive deformable barrier deformation contours for SUV1 and SUV2 (contour depth scale [mm] on far right).

Copyright

C Taylor & Francis Group, LLC 149 TCRS 2007 Vol. 12 No. 2
R Thomson, M Edwards, T Martin, C van der Zweep, R Damm and G della Valle

• dynamic lateral misalignment of vehicle structures in The accident and benefit work have identified the target
frontal impacts is a problem; and population and expected benefit for improved compatibil-
• strong crossbeams help SI performance. ity for Europe by extrapolating data from Great Britain and
Germany. The target population is defined as those casu-
The FWDB and PDB test tools and assessment criteria alties that are likely to experience a reduced risk of injury
have been shown to be capable of distinguishing the dif- from improved compatibility measures. The number of
ference in vertical and horizontal frontal force distribution casualties prevented, that is, the benefit, will be a subset of
of the test vehicles. These differences include the number the target population. It was estimated that between 14%
of load path levels, the width of the front structure, and (3466) and 31% (7675) of fatally injured car occupants and
the ability of crossbeam structures to adequately distribute between 29% (50,260) and 52% (90,122) of seriously in-
lower rail loads. jured car occupants lie within the target population for Eu-
rope. Following this analysis, compatibility improvements
to vehicles should yield reductions in the order of 4% to
9% for severe injuries in vehicle impacts in Europe while
CONCLUSIONS
fatalities could be reduced by 6% to 12%. These benefit
VC-Compat is an ambitious project attempting to develop calculations can be considered conservative since any ben-
a complex set of test procedures that will assess vehicle efits for other impact configurations were not considered.
crash compatibility. The project is building on previous Car-to-car testing has shown that there is a compatibil-
development work conducted at various European research ity problem for European vehicles that is different from
institutes that has been previously reported to the EEVC the “SUV versus passenger car” issue identified in North
WG15. From the history of the two candidate test proce- America. In two of the test series, impacts between two
dures, new information is being compiled in VC-Compat identical small cars caused much more occupant compart-
so that the FWDB and PDB can be compared with each ment deformation than would be expected. This indicates
other, and in particular, evaluate the tests’ ability to assess that current self-protection tests do not ensure that a ve-
SI of cars. hicle can impact itself and maintain the same crash per-
The structural survey and mathematical modeling ac- formance. Occupant compartment strengths were demon-
tivities provide support to the main crash testing activity strated to be a foundation for compatibility since the front-
in the program. An inventory (structural survey) of ve- end performance requires a stable support (firewall and
hicle structures has provided important geometrical data occupant compartment).
that can be used in the analysis of test data. In particu- The early results of the VC-Compat program indicated
lar, the geometry of the vehicles is needed to interpret the that SI performance of a two-level load path car was bet-
influence of different structures on the SI of vehicles. Com- ter than a one-level load path vehicle. Subsequent tests
puter models have provided general information about the confirmed this finding. In tests between small family cars
recommended force levels of small vehicles. Fleet models and SUVs, the presence of lower load paths (subframes)
have been used to demonstrate the benefits gained by in- in the car was beneficial when impacting an SUV, which
troducing new vehicle designs that improve compatibility. also had a lower load path. This property is deemed impor-
Changes to the smaller vehicles appear to be an important tant for compatibility and should be detected by the test
first step. procedures.

Figure 17 Progressive deformable barrier assessment of average depth of deformation (ADOD) and average height of
deformation (AHOD).

TCRS 2007 Vol. 12 No. 2 150 Copyright

C Taylor & Francis Group, LLC
 Car-to-car crash compatibility

The dynamic misalignment observed in the SUV crash Dutch Ministry of Transport, and Fiat Auto S.p.A. The
test series demonstrated serious consequences for the VC-Compat Car-to-Car research team: TRL: Mervyn
smaller vehicles. It is important that this characteristic Edwards, Huw Davies; UTAC: Tiphaine Martin, Pascal
is identified in a new compatibility crash test procedure. Delannoy; BASt: Eberhard Faerber, Claus Pastor, Richard
Crossbeam strength and outer support must be sufficient Damm; Chalmers: Robert Thomson, Fredrik Jenefeldt;
so that SI between vehicles can occur for lower overlap FIAT: Giancarlo Della Valle, Fredriqo Pasqui, Danilo
values. Barberis; and TNO Automotive: Cor van der Zweep, Floris
Crash testing work to date has focused on the develop- Leneman, Ton Versmissen.
ment and validation of the FWDB and PDB test proce-
dures. Both tests encourage car designs with good vertical REFERENCES
load spreading capability. The FWDB and PDB test tools
1. Community Road Accident Database (CARE). Available at:
have been shown to be capable of distinguishing the pres-
http://europa.eu.int/comm/transport/care/. Accessed
ence of subframe load paths and different bumper cross-
September 2004.
beam behavior. The proposed FWDB and PDB assessment 2. M Edwards, A Benedetto, P Castaing, H Davies, E Faerber, A
criteria have been calculated and compared for a range of Fails, T Martin, R Schaefer, and A Thompson, ‘A study to
vehicles. The results to date are encouraging and both test improve the crash compatibility between cars in frontal
methods appear to discriminate between the vehicles as impact’, Final report, Directorate-General for Energy and
expected by the investigators. Transport, Contract Reference: E3-3 B2702/SI2.318663/2001
The EEVC WG15 route map requires a test procedure TRL, July 2002.
to assess a vehicle’s SI potential in the short term. The 3. VC-Compat Web page. Available at: www http://
VC-Compat project will continue to focus on the develop- vc-compat.rtdproject.net/. Accessed April 2006.
ment of the FWDB and PDB test procedures, as both these 4. R Thomson and M Edwards, ‘Passenger vehicle crash test
tests have the potential to achieve this goal. In the limited procedure developments in the VC-Compat project’, 19th ESV
Conference, Washington D.C., USA, Paper 05-0008, 2005.
tests that have been analyzed to date, both the PDB and
5. M Edwards, H Davies and C A Hobbs, ‘Development of test
FWDB can discriminate between different vehicle struc- procedures and performance criteria to improve compatibility
ture designs. Thus, the testing tools – the LCW or PDB – in car frontal collisions’, Proc 18th ESV Conference, Nagoya,
can assess SI. An important step is still needed to deter- Japan, Paper 86, 2003.
mine the threshold values that separate compatible vehicles 6. P Delannoy, T Martin and P Castaing, ‘Comparative
from incompatible vehicles. The test procedures must also evaluation of frontal offset tests to control self and partner
demonstrate sufficient sensitivity and robustness so they protection’, Proc 19th ESV Conference, Washington D.C.,
can be confidently applied for rating vehicles. These chal- USA, Paper 05-0010, 2005.
lenges must be overcome before the final application of a 7. P O’Reilly, ‘Status report of IHRA compatibility and Frontal
compatibility test procedure for governmental or consumer Impact Working Group’, Proc 18th ESV Conference, Nagoya,
rating test programs. Japan, Paper 402, 2003.
8. P Delannoy and J Faure, ‘Compatibility assessment proposal
close from real life accident’, Proc 18th ESV Conference,
ACKNOWLEDGEMENTS Nagoya, Japan, Paper 94, 2003.
9. C van der Zweep, G Kellendonk and P Lemmen, ‘Evaluation
The consortium is grateful for the financial sponsorship of: of fleet systems model for vehicle compatibility’, Int J
the European Commission DG-TREN, DSCR (French Crashworth 2005 10 (No. 5) 483–494.
Department of Transport), Swedish Road Administra- 10. A C Kuchar, R Greif and G W Neat, ‘A systems methodology
tion, UK Department for Transport, Federal Ministry of for evaluation of vehicle aggressivity in the automotive
Transport, Building and Urban Development (Germany), accident environment’, SAE, Paper 2001-01-1172, 2001.

Copyright

C Taylor & Francis Group, LLC 151 TCRS 2007 Vol. 12 No. 2
TCRS 2007 Vol. 12 No. 2 152 Copyright

C Taylor & Francis Group, LLC