Vehicle Rollovers Analysis

Finite Element Model Based Analysis 

Fadi Tahan
Kennerly Digges
Pradeep Mohan

 
 

Abstract – The Automotive Safety Research Institute (ASRI) has commissioned a series of
dynamic rollover tests using the Jordan Rollover System (JRS). The Jordan Rollover System
(JRS) has the flexibility of closely controlling the initial conditions of the test vehicle being
subjected to a rollover. The controlled conditions include drop height, roll velocity, roadway
velocity and angular orientation at drop. A previous series of dynamic rollover tests sponsored
by the Santos Family Foundation were conducted under a constant set of initial conditions. The
ASRI proposed this study in order to compare these test conditions with other alternatives.

The baseline conditions for the Santos tests were: Roll angle of 145° at impact; Pitch angle of 5°;
Yaw angle of 10°; Roadbed speed of 24kph; Vertical drop of 10cm and roll velocity of
190deg/sec. This test procedure produced severe roof damage on some, but not all of the
vehicles tested. However, the new static test standard (FMVSS216) doubles the minimum
allowable roof strength as measured by a static test. This change in the standard should permit
vehicles to withstand more severe rollover conditions. A purpose of this research is to investigate
how changes in the initial rollover conditions on the JRS would affect the roof loading and
dynamics of a vehicle with a strong roof for both the driver and passenger sides of the vehicle.

The approach used in this study makes use of Finite Element Analysis. A FMVSS No 216
validated Finite Element model of a 2003 Ford Explorer. In order to exclude crushing factors, a
strong roof structure was adopted and made of high strength steel without failure and plasticity.

In this initial study, the angle of impact was held constant at the baseline conditions. The initial
rollover parameters of road speed, roll rate and drop height were changed one at a time in order
to study their effects on vehicle rollover behavior and loading of the roadbed.

The following observations were made based on simulation results:

1- Roadbed speeds of 0, 9, 18, 24, 30 and 36kph were investigated. A minimum roadbed speed
of 24kph was necessary to produce roof contact on the near and the far sides. For roadbed speeds
lower than 24kph, the far side impact was not significant.

2- Roll rates of 190, 360 and 540deg/sec were investigated. As the roll rate increases, the
roadbed force measured at the near side impact decreases. The roadbed force effect at the far side
impact sustains for a longer time in order to manage the additional energy in the system. After
the near side hit is completed, the roadbed speed controls the roll rate.

3- Three drop heights of 10, 20 and 30cm were investigated. The increase in potential energy
increases the force that is managed by both sides of the roof. The drop heights have shown no
effect on the roll velocity.

 
 

Introduction 

FMVSS No 216 specifies a quasi-static test procedure that measures the force required to push a
metal plate into the roof at a constant rate [1]. It requires a reaction force equal to 1.5 times the
weight of the vehicle be reached within 5 inches of plate displacement. In 1991, the standard was
extended to apply to light trucks and vans with Gross Vehicle Weight Ratings (GVWR) less than
6,000 lbs [2]. In May 2009, the rule doubled the amount of force the vehicle’s roof structure
must withstand in the specified test, for vehicles of a Gross Vehicle Weight Rating (GVWR) of
2,722kg (6,000 lbs) or less, to 3.0 times the vehicle’s unloaded weight. The rule requires, as well,
that all vehicles must meet the specified force requirements in a two-sided test, instead of a
single-sided test, i.e., the same vehicle must meet the force requirements when tested first on one
side and then on the other side of the vehicle. NHTSA adopted this upgraded quasi-static
requirement for now and is conducting research for a dynamic test standard [3].

Figure 1. FMVSS No 216 Test setup [1]

The Insurance Institute of Highway Safety (IIHS) has developed its own rollover
crashworthiness rating. The IIHS indicated that the boundary for a good rating in their program
will be a Strength to Weight Ratio (SWR) of 4.0 in a one-sided platen test condition similar to
the existing FMVSS No 216 test procedure. For an acceptable rating, the minimum SWR is 3.25.
A marginal rating value is 2.5. Anything lower than that is rated as poor. This rating system is
based on the institute research showing that occupants in rollover crashes benefit from stronger
roofs [4].

 
 

In developing the upgraded FMVSS 216 NHTSA considered, several dynamic rollover tests.

The FMVSS No. 208 dolly rollover test was originally developed only as an occupant
containment test and not to evaluate the loads on specified vehicle roof components [3]. The
vehicle is rolled laterally off the inclined ramp of a dolly moving at 30mph. The vehicle typically
rolls one to four times and, to pass the test, no part of an unrestrained dummy in the vehicle may
be ejected. The National Highway Traffic Safety Administration (NHTSA) believed that this test
lacks sufficient repeatability to serve as a structural component compliance requirement [3].

The Controlled Rollover Impact System (CRIS) test is achieved by mounting a vehicle on a spit
at the back of a moving tractor-trailer. It is rolled as the truck drives along a flat surface at a
given speed and is dropped so that it lands on its roof while rolling. After release, the vehicle
continues to roll to rest [5]. CRIS is considered to produce repeatable vehicle and occupant
kinematics for the initial vehicle-to-ground contact but NHTSA have no indication that this
procedure is repeatable after the initial ground contact [3].

The Jordan Rollover System (JRS) test is attained by mounting a vehicle on an axis that permits
it to roll and be dropped. As the vehicle is rotated, a roadway segment is run underneath and the
vehicle is dropped so that its roof strikes the road as it would in an actual rollover. After both
sides of the roof have impacted the roadway, the vehicle is then lifted by the spit so that it will
sustain no further damage to the vehicle sides or undercarriage. Subsequent rolls can be
conducted by resetting and running the JRS test a second time [5, 6]. NHTSA believes there is
lack of real-world data in order to determine the JRS test parameters for different vehicles [3].

In this research, the JRS initial conditions were the base parameters. Only 3 quarter turns were
simulated since we are interested in loading the near and far sides structure of the roof and the
vehicle dynamics during the impacts.

Research Approach 

Finite Element (FE) modeling was used in this analysis. FE modeling has been indispensable in
the development of component design, and vehicle crashworthiness evaluations. This study
utilizes commercial FE code, LS-DYNA to simulate roof strength for multiple loading
conditions [10].

FE Model 

The full vehicle FE model used in this study was developed at the National Crash Analysis
Centre (NCAC) under a co-operative agreement between Federal Highway Administration and
National Highway Traffic Safety Administration and The George Washington University. The
FE model of a 2003 Ford Explorer has been validated to several sub-system tests and to a full
frontal rigid barrier test conducted by NHTSA.

 
 
FMVSS 216 Validation 

Further model validation was carried using FMVSS No 216 test set up with two different pitch
angles. The FE model was verified against two FMVSS No 216 tests conducted by NHTSA to
establish the baseline performance for this vehicle. The first test (NHTSA test number C0139)
was conducted as per the FMVSS No 216 protocol platen angles of 5 degree pitch and 25 degree
roll [7]. The second test (NHTSA test number C0140) was conducted at platen angles of 10
degree pitch and 45 degree roll to investigate the change in roof crush resistance [8]. The platen
reaction force versus roof deformation from the NHTSA tests and the corresponding FE
simulations are shown in Figure2. The reaction force is presented as a percentage of the unloaded
vehicle weight. The FE model shows good correlation in both tests.

Figure2. Force-Crush Comparison Tests C039 & C040 and FE Simulations [8]

FE model set up 

Once completed, the model was given initial conditions similar to the JRS common values [6].
The FE model had some differences from the JRS test. The FE model was free of any outside
structure constraints, the roadbed was considered moving at a constant speed and the roof
structure was strengthened by eliminating the plastic deformation in steel. These assumptions
were made in order to investigate solely the sensitivity effects on future strong roofs that would
meet the new regulations and to exclude any interactions with the vehicle’s surroundings (i.e.
spit interactions). The front and side views are shown in Figure3.

 
 

Figure3. FE model front and side views setup

Rollover simulations usually take long computational times. Therefore, in order to reduce the
model size and simulation run time, some components of the vehicle were removed. The
suspensions, drivetrain and engine were selected because they have minimal influence on the
roof strength. The weight and inertia of these parts were replaced by adding concentrated mass
and inertia to six points at the ladder frame. LS-DYNA provides *ELEMENT_INERTIA [10] for
this kind of purpose. The inertial properties of the removed parts were determined and divided
into the engine, the transmission, the front suspension, and the rear suspension sections. These
sections were replaced and applied to nodes attached to the ladder frame at the mounting points
of the removed components. Vehicle weight was dropped from 2237.8 kg to 719.7 kg. Several
iterations were conducted in order to calibrate the model and the element inertias in order to
closely match the original model. These components reduced the number of elements from
619161 to 420517, which had subsequent effects on run time. The values are shown in Table1
and the full and the reduced models are compared in Figure4 [9].

Model Mass I11 I22 I33

Description
Full Vehicle 2237.8 kg 725.97 kg×m2 3805.4 kg×m2 4063.8 kg×m2
Model
Reduced Vehicle
2254.7 kg 604.9 kg×m2 3884.0 kg×m2 4073.7 kg×m2
Model

Table1. Inertial properties of the original and the reduced models [9]

The roadbed was represented by *RIGIDWALL [10] in order to exclude any variations
occurring from the ground. The roadbed speed was maintained fixed during this representation
using the MOTION [10] option since it excludes any variability. The roadbed was 2.6m wide
and 2.8m long.

 
 

Figure4. Full and Reduced models. Yellow parts are the removed components; Red elements are
the added mass with inertias [9]

To represent future vehicles that are going to meet the new FMVSS No 216 requirements, all
roof components were switched to pure elastic properties. This includes the A- B- C- D-pillars,
roof rails, roof cross members and roof outer sheet metal. These components are show in
Figure5. They made the roof structure stronger than the original model and eliminated structural
plastic deformation during the sensitivity assessment. The Strength to Weight Ratio (SWR) is
measured to be 3.9 times the vehicle’s unloaded weight .

Figure5. Roof structure components switched to elastic steel material to represent futuristic
strong roof

 
 

This representation was considered sufficient for the purpose of analyzing the near and far sides
of the roof. The near side of the vehicle is set to be the passenger side and the far side is the
driver side in this study. Finally, the simulation time of 350ms was sufficient to cover 3 quarter
turns of the rollover.

Selecting the initial conditions was based on the Santos test protocols for the JRS tests. These
values were considered as baseline model in this study. The approach is based on performing a
sensitivity study with multiple initial parameter variations on the validated FE model. The design
parameters addressed were the roadbed speeds of 0, 9, 18, 24, 30 and 36kph, roll rates of 190,
360 and 540deg/sec, and drop heights of 10, 20 and 30cm. The sets of simulations were
compared with each other.

Parametric Study and Baseline Model Comparison 

The baseline variables were: Roll angle of 145° at impact; Pitch angle of 5°; Yaw angle of 10°;
Roadbed speed of 24kph; Vertical drop of 10cm and roll velocity of 190deg/sec. The baseline
values of the JRS produced extensive roof crush in some vehicles but not in others.

The first set of parameters varied was the roadbed speeds. Roadbed speeds of 0, 9, 18, 30, and
36kph were investigated. A summary is shown in Table2.

Table2. Different roadbed speed simulations

Roadbed normal forces for different roadbed speeds are compared with respect to roll angle as
shown in Figure6. The initial roll angle of 145° is where the loading begins. A minimum roadbed
speed of 24kph was necessary to produce roof contact on both the near and the far sides. The
roadbed normal forces for roadbed speeds of 24, 30, and 36kph were about 40KN measured at
the near side (passenger or leading side), however, the load peaks around 60KN at the far side
(driver or trailing side). For roadbed speeds lower than 24kph, the far side impact was not
significant or complete. The roadbed normal forces for roadbed speeds of 9kph and lower were
higher at the near side but lower at the far side. For roadbed speed of 18kph, the far side did not
withstand high loading and the roadbed normal force curve did not converge to higher speed
curves.

 
 

This behavior corresponds closely to the following findings. For a stationary roadbed, the roof
crush was similar to a drop test. This is because the roll angular velocity could not alone
maintain sufficient speed for the vehicle to roll on its far side. For different roadbed speeds, the
affect of the roll rate of the vehicle depended on the speed. For roadbed speeds lower than the
initial roll rate times the shortest distance between the point of contact and the axis of rotation,
the vehicle roll rate was decelerated. For higher speeds, the vehicle roll rate was increased, as
shown in Figure7. When the vehicle was upside down (i.e. roll angle =180°), the distance
between the point of contact and the axis of rotation was the shortest. This condition decreases
the roll rate. This distance increases again when the roadbed contacts the far side of the vehicle
which, in turn, increases the roll rate.

Figure6. Roadbed normal force Vs Roll Angle for different roadbed speeds

 
 

Figure7. Roll Rate Vs Roll Angle for different roadbed speeds (SAE 060 filtered)

The second set of parameters varied was the roll rate. Roll rates of 190, 360, and 540deg/sec
were investigated. A summary is shown in Table3.

Table3. Different roll rate speed simulations

Roadbed normal force for different roll rates are compared with respect to roll angle as shown in
Figure8. As the roll rate increases, the roadbed force measured at the near side impact decreases.
The vehicle has a tendency to bounce off the ground at initial contact for higher rotational
speeds. The roadbed force effect at the far side impact sustains for a longer time in order to
manage the additional energy in the system.

Similar to roadbed speeds findings, the roll rate changes based on roadbed speed. This is
influenced by the initial roll rate times the shortest distance between the point of contact and the
axis of rotation. The vehicle roll rate of 190deg/sec is increased to match the roadbed speed and

 
 

decreased for the 360 and 540 deg/sec. After the near side hit is completed, the roadbed speed
controls the roll rate as shown in Figure9. This behavior is similar to all three roll rates
simulated.

Figure8. Roadbed normal force Vs Roll Angle for different roll rate speeds

Figure9. Roll Rate Vs Roll Angle for different roll rate speeds (SAE 060 filtered)

 
 

The third set of parameters varied was the drop height. Heights of 10, 20, and 30cm were
investigated. A summary is shown in Table4.

Table4. Different drop height simulations

Roadbed normal forces for different drop heights are compared with respect to roll angle as
shown in Figure10. As the drop height increases, the roadbed force measured at the near and far
sides increases. This is attributed to the increase in potential energy that has to be managed by
both sides of the roof.

Figure11 shows the roll rate versus the roll angle. The different curves lie on top of each other.
Therefore, the drop heights show no effect on the roll rate.

Figure10. Roadbed normal force Vs Roll Angle for different drop heights

 
 

Figure11. Roll Rate Vs Roll Angle for different drop heights (SAE 060 filtered)

Summary 

The focus of this study was to evaluate parameters sensitivity in the initial rollover conditions on
the JRS. The parameters would affect the roof loadings and dynamics of the vehicle with a
strong roof. An elastic roof structure was assumed. FE model were used to address this study.
The FE model was validated and reduced in order to reasonably evaluate the sensitivity study.
The simulation results showed that the roadbed speeds of 24kph is important for a 3 quarter turns
rollover. The roadbed speed influences the roll rate after initial contact. The increase in roll rate
resulted in a decrease roadbed normal force on the near side of the impact but an increase at the
far side. The roadbed speed controls the roll rate after the near side impact is completed.

This study is limited by the assumption made. Further work is recommended to identify more
variables that match the majority of the real rollover world accidents. In addition, other dynamic
rollover parameters should be investigated like different roll angle, changing multi-parameters at
the same time, as well as studying the effect of the JRS structure to the roof crush.