Accepted Manuscript

Development of an Innovative Design of a Composite-Sandwich based Vehicle

Roof Structure

Soroosh Borazjani, Giovanni Belingardi

PII: S0263-8223(17)30205-2
DOI: http://dx.doi.org/10.1016/j.compstruct.2017.02.015
Reference: COST 8234

To appear in: Composite Structures

Received Date: 20 January 2017

Revised Date: 1 February 2017
Accepted Date: 2 February 2017

Please cite this article as: Borazjani, S., Belingardi, G., Development of an Innovative Design of a Composite-
Sandwich based Vehicle Roof Structure, Composite Structures (2017), doi: http://dx.doi.org/10.1016/j.compstruct.
2017.02.015

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Development of an Innovative Design of a Composite-Sandwich based Vehicle Roof Structure

Soroosh Borazjani* and Giovanni Belingardi

Department of Mechanical and Aerospace Engineering (DIMEAS), Politecnico di Torino, corso Duca degli Abruzzi, 24, 10129 Torino, Italy
[email protected] [email protected]

Abstract

Obligatory standards are dictated to vehicle manufacturers for decreasing the high number of road dead tolls in
rollover crashes; at the same time demands to produce light-weight vehicles have increased substantially to reduce
the toxic gases emission. This paper presents a new design of the roof strengthening configuration in which
sandwich material has been used. This type of configuration improves the energy-absorption capacity of the vehicle
roof system and dissipated impact energy in a controlled manner. In reality, carrying out roof crush test is not cost
effective, thus numerical analysis of vehicle roof crush test has been performed according to standard FMVSS 216
test. Sandwich structures with unidirectional carbon /epoxy composite face-sheets and Expanded Polypropylene
(EPP) foam core have been used to model different configurations for vehicle roof structure. The effects of
increasing the foam core density and face-sheets thickness on the energy absorption and strength-to-weight ratio
(SWR) of vehicle roof structure have been investigated. Results revealed that, the optimized sandwich solution type
6 with the face-sheets of 0.8mm and foam core of 70 Kg/m3 density reduces the vehicle roof panel by 68% while it
has almost the same structural performance with the steel solution having equal value of SWR.

Keywords: Rollover, roof panel, sandwich material, energy absorption, SWR

1. Introduction

The high rate of crash accidents fatalities urged the car manufacturer companies to improve the safety of vehicles. It
is reported that around 33000 people died in United States due to different type of accidents in 2010 [1]. It is quite
important to mention that although rollover crashes allocate small portion of total crashes, around 2.1%, this is the
most deadly type of crash. Unfortunately 35% of people involved in rollover crashes died, that is around 7600
people. Since, even if passengers are fastening their seatbelts (but in this particular case seatbelts are less effective
due to their particular geometry), there is a high chance of fatalities, 31%, in rollover accidents, it is vital to design
the vehicle roof component structure with an high standard of safety [2].
In spite of the existence of various roof crush tests, government institutions are pushing even higher safety margins.
There are different types of static and dynamic roof crush tests such as the Federal Motor Vehicle Safety Standard
(FMVSS) 216 [3], Inverted Vehicle Drop Dynamic Test (Society of Automotive Engineering, SAE, J996) [4] and
Dolly Rollover Test (SAE J2114) [5] that are aimed to evaluate the vehicle roof structure performance. The initial
rules of FMVSS prescribed that the vehicle roof should sustain a force equal to 1.5 times the unloaded vehicle
weight (UVW). The new updated standard, FMVSS 216a, [6] increased this margin by considering the force to be
applied to the vehicle roof structure equal to 3 times the UVW with the new obligation of making test on both driver
and passenger sides. There is also more conservative type of vehicle roof crush test standard which is considering
the applied load equal to 4 times UVW, set by Insurance Institute for Highway Safety (IIHS) [7].

Composites as energy absorber, light-weight and anti-corrosion materials are the perfect substitutions for metallic
structure specifically in the case of impact. Although these materials have not the possibility for plastic deformation
due to their brittle nature, they have high stiffness and strength to weight ratios. Several works have been done on
investigating the energy absorption and crashworthiness of composite and sandwich structures. Mamalis et al. [8]
studied the collapse modes of sandwich panels made of composite face sheets and a foam core under axial
compression force. Three collapse modes were observed. The first collapse mode occurred with foam core shear
failure and sandwich fragmentation. The second mode was characterized by face sheets delamination and buckling
and the third one was the progressive crushing mode. It was proved that the third mode is the most important type of
sandwich collapse mode due to energy absorption capacity of the structure; it depends on the foam core properties.

* Corresponding author.
E-mail address: [email protected] (S. Borazjani).
Many papers such as [9-11] studied sandwich tubes under tension and compression. Tarlochan et al. [11] tested
sandwich structures

* Corresponding author.
E-mail address: [email protected] (S. Borazjani).
made of different cores and face sheets materials and designs under axial loading. Tubes made with a sandwich
structure with composite skin reinforcement made of carbon, glass or Kevlar fibers along with Expanded
Polystyrene (EPS) or Polyurethane (PU) cores have been tested. These studies have demonstrated that carbon fiber
and EPS are the best solution for face sheet and core respectively. There are some papers available in which
innovative designs have been presented. Zhang et al. [12] designed honeycomb sandwich circular tube to improve
energy absorption efficiency. This design is made of two circular aluminum tubes filled in between with a core
shaped as a large-cell lattice. It was found that the buckling of lattices controls the folding of the outer and inner
tubes which results in an increment of the plastic deformation of the tube. It was observed that the intrusion of the
folded tube walls into the honeycomb cell gaps, strengthen the crashworthiness of the honeycomb cell. Zangani et
al. [13] proposed the foam-filled twin-walled composite hollow box sections with internal reinforcement.
Progressive failure of the suggested tubes in respect to different corner designs and arrangements of the corrugation
configuration inside the foam core were studied experimentally and numerically. Strain rate effects on the material
properties of foam structure has been determined by quasi static test just at low strain rates in [14]. Also damage and
failure of the considered foam structure have been studied under the drop tests and obtained experimental results
have been compared with numerical results.

Li et al. [15] evaluated the crashworthiness of hollow and foam-filled circular tubes under three-points bending test.
During test the failure modes of hollow, foam-filled single and foam-filled double tubes were observed and the total
energy absorption, the specific energy absorption and the energy –absorbing effectiveness factor were determined.
This research proved that the foam-filled double tubes has higher resistance to the bending load and absorbed the
highest energy. Steeves and Fleck [16] investigated the collapse modes of sandwich structures consisting of woven
glass fiber face sheet and PVC foam core under bending load by experiment and simulation. Different types of
collapse modes such as face sheet microbuckling and indentation, core shearing and crushing were observed
depending on the sandwich geometry and foam core density. Collapse modes of sandwich structures with respect to
the slenderness ratio of the beam and the relative thickness of face sheet to core were investigated. Damage
indentation of sandwich structure with rigid foam core under compressive load has been investigated and tested by
Rizov et al. [17]. Typical dimensions of the damaged zone on the sandwich face sheet have been measured and
load-displacement curves have been obtained. Using finite element codes residual stress and strains after unloading
were calculated and compared with the experimental results. Johnson and Li [18] investigated the effects of puncher
nose shape and the foam core density on the indentation of sandwich structures. It was found that the foam core
density has a high influence on the energy absorption of sandwich structures.
Previous works have discussed different methods in order to increase the strength of vehicle roof structure during
rollover crash. Mao et al. [19] strengthened the vehicle roof structure by increasing roof components thickness,
placing steel tubes inside the initial design and foam filling of the roof panel based on the requirements of the
FMVSS 216 and inverted dynamic tests standards; simulations have been done with the finite element code LS-
DYNA. Although these techniques increased the vehicle weight from 1% to 2.3%, they could increase the energy
absorption of roof components. Pan and Zhu [20] proposed surrogate-based design to optimize the roof strength
efficiency. Different types of computational optimization methods such as kriging, radial basis function, support
vector regression and various weighted average surrogates have been tried to select the most effective optimization
method. After having applied different optimization algorithms, weighted surrogate method proved the highest
efficiency by raising the crush resistance force of 41.7% while the vehicle weight reduced by 5.3%. A new design of
thermoplastic countermeasure concept to reinforce vehicle roof structure has been introduced in [21].
Countermeasure concept modifies the original structure by adding longitudinal ribs along the bending direction
stiffened by cross ribs to strengthen local buckle points. The proposed design presented in a full plastic
countermeasure and a plastic/metal hybrid countermeasure placed inside the B-Pillar. The efficiency of the design
has been evaluated by numerical simulation and it was shown that while achieving the same strength of the
traditional steel design, countermeasure concept reduced the structure weight by 40-50%. Rahul and Mirdamadi [22]
improved the roof structure stiffness by adding structural foam inside the vehicle body structure. After having
observed the efficiency of the design, the model was optimized to maximize the roof structure strength and
minimize the vehicle mass. Bambach [23] has proposed to improve the vehicle roof system performance by using
bonded composite metal structures. After roof crush quasi static tests, his studies showed that the suggested model
has the potential to contribute to higher roof strengths and light-weighting of vehicles.
This paper presents a new design of vehicle roof structure strengthening using sandwich structures with different
types of configuration to evaluate and optimize the performance of the suggested design.
2. Test Procedure and Development of finite element model

The experimental procedure is implemented according to FMVSS 216 test. Based on FMVSS 216 [3], a flat
rectangular rigid plate with the dimensions of 1829 mm by 762 mm applies the intended load in the specified
direction (see Figure 1) on the vehicle roof structure while the floor is clamped and constrained in all the degrees of
freedom. The rigid plate should be oriented in the way that its longitudinal axis is below the 5 degrees of the
horizontal (pitch angle) and its transverse axis is below the 25 degrees of the vertical (roll angle) of the centerline.
The rigid plate lower surface must be in contact with the vehicle roof centerline at the initial position and is placed at
the 254 mm from the forward most point of roof centerline. The FMVSS standard test configuration is shown at
Figure 1. Based on the standard, the load is applied to the vehicle roof with the velocity of no more than 13 mm/s in
maximum 120 seconds.

Figure 1. FMVSS 216 test configuration

The numerical modeling is the same with the experimental procedure except for the rigid puncher velocity and that
is directly determining the simulation process time. The finite element model of TOYOYA YARIS 2010 has been
provided from the National Crash Analysis Centre (NCAC) at George Washington University in the United States
[24]. Some modifications such as deletion of unnecessary parts for roof crush modeling (e.g. Tires), and addition of
the windscreen model and of sufficient constraints between the components of the model have been done. Figure 2
shows the finite element model of the vehicle that consists of 974388 elements and 1009131 nodes with the loading
and boundary condition.

Figure 2. Load case and boundary condition

Due to the very high computational cost for modeling quasi static tests with the LS-DYNA explicit package in
actual process time, the simulation is performed by the rigid plate velocity higher than in the physical test but in
very small period of time. In order to prevent instabilities causing by small step size during the analysis, mass
scaling is used to increase the time step size in each cycle by adding artificial mass to the model.
To define the normal direction for the rigid plate is one of the difficulties in modeling the roof crush test. The
mathematical method for defining the normal plane has been presented at [25]. The rigid puncher normal plane is
obtained after the reference coordinate system rotates about the two planes of Z-X and Z-Y by the rotation angles of
β and α respectively. The final normal plane is presented as:


n = {cosα sin β ,sinα ,cosα cos β }

where β and α are pitch and roll angles respectively.

Different types of windshield laminated glass strategy are presented in Figure 3 by [26].Windscreens have been
modeled with single layer shell elements using ‘MAT_LAMINATED_GLASS’. The windshield consists of two
layers of glass bonded with PVB. Using MAT_LAMINATED_GLASS to model the windshield, it is possible to
capture the failure of glass layers based on predefined fracture plastic strain but failure modeling of the innermost
polymer is not feasible. Due to this fact, although failure occurs in glass layers, the windscreen failed elements are
not deleted. The main material properties of glass and PVB are presented at Table 1.

Figure 3. Different methods to model the windshield laminated glass

 Table 1. Glass and PVB main material properties

Mat_Rigid (Mat20) has been used to model the rigid plate while boundary_prescribed_motion_rigid has been used
to apply the load to roof structure. Automatic contact surface to surface between the rigid plate as the master surface
and the roof structure as the slave surface has been used to model the contact and to extract the reaction force
history. Mat24, Piecewise Linear Plasticity, has been used to model the steel roof with the thickness of 1.2 mm.
Mat24 is an elasto-plastic material model that can consider the failure based on the equivalent plastic strain and
capable of taking into account the strain-rate effects by using the Cowper-Symonds model, which scales the yield
strength. The two types of non-linear element formulations 2 and 16, named “Belytschko-Tsay” shell element and
“Fully Integrated Shell Element” respectively, have been used to model the roof panel. Possible Hourglass (HG)
modes, which are nonphysical, zero-energy modes of deformation, should be taken under careful control using
hourglass control algorithms [27].

3. Material selection for sandwich solution

The sandwich solution is made of two composite face sheets and a foam core. The composite consists of four layers
of unidirectional carbon fiber in cross-ply format and Expanded Polypropylene (EPP) is used as a core. The
schematic of sandwich structure is shown in Figure 4. The material properties of Carbon/Epoxy fiber are presented
in Tables 2 and 3.

Figure 4. Schematic of sandwich panel

Table 2. Composite material elastic properties

Table 3. Strength parameters of composite

Wide range of information about different composite material models has been presented at [30]. Mat54/55 has been
selected to model the composite face sheets due to the ability of this model to capture the failure for unidirectional
composite fibers. The thickness of both composite face sheets are from 0.6-0.8 mm; complete information about the
modeling of unidirectional composite fiber has been presented at [30].

Generally, compression is the main mode of deformation for foams since they are weak against shear or tension
[31]. When foams are under compressive force, three phases are detectable in their stress-strain curves. At the first
phase, linear compression occurs followed by the plateau phase determining the mechanical behavior of the foam.
The ideal foam is the foam which has long plateau region due to higher energy absorption ability and controllable
reaction stress. The last phase is densification which is a nonlinear compression. Using the same foam base material,
foam with higher density has higher plateau strength and higher amount of absorbed energy [32]. During the
compression test, neither lateral displacement nor cross section changes are expected: so the Poisson’s ratio is
almost zero for crushable foam [14].
There are different material models available to model the foam structure in LS-DYNA. Comparison of material
models for modeling foam structure along with their application is presented at Table 4 based on the manual and
material model help [33, 34].

Table 4. LS-DYNA available foam material models

Mat63 (Mat_Crushable_Foam) has been selected to model the foam core. This material model specifically is used
for crashworthiness applications where cyclic behavior is unimportant [34]. This material model is not capable of
considering strain-rate effects and need extra keyword command to capture the failure. In order to capture the
realistic mode of foam failure, it is necessary to use the keyword Mat_Add_Erosion. This option can predict the
foam failure based on plastic strain and tensile stress [35]. Tetrahedron solid elements with one integration point
have been used to model the foam core.
Hourglass control algorithm type 2 has been applied to control the hourglass energy [36]. Instead of adhesive joints
to attach composite face sheets to the foam core, Contact_Tied_Nodes_To_Surface has been applied. The contact
between the foam core and the composite face sheets has been detected using
Contact_Automatic_Surface_To_Surface.
There are few challenging difficulties in modeling foam structures under compressive force. When foam is under
compression load, large deformation occurs and may cause the “negative volume error”. In order to solve this
problem it is recommended to extend the initial stress-strain curve specifically at large strains [37]. In [38] after
implementing the uniaxial compression test data based on ASTM D1621-94, the stress-strain curves and the
extended curves for the two selected types of foam are presented in Figures 5 and 6. It was proved that based on the
Gibson equation [39] there is a relation between the compressive Young’s modulus and the density as shown in
Figure 7. The Young’s modulus for the EPP 45 Kg/m3 and 70 Kg/m3 result to be 2.1 and 5 MPa respectively.

a. Original curve b. Extrapolated curve

Figure 5. a. Original and b. exponentially extrapolated stress-strain curves for EPP 45 Kg/m3

a. Original curve b. Extrapolated curve

Figure 6. a. Original and b. exponentially extrapolated stress-strain curves for EPP 70 Kg/m3

Figure 7. EPP foam density vs. compressive Young’s modulus

 Another difficulty in modeling foam structure is penetration of elements into the neighbor elements. To avoid this
problem the use of the contact interior type 2 option is needed. This option can handle the elements penetration
problems in both compression and shear modes [35]. In the case of modelling a sandwich structure, it is generally
recommended to activate the laminate shell theory option. This option corrects the wrong assumption of constant
shear strain through the thickness of the shell; this parameter is very important when modelling sandwich structure
because it helps to prevent a too stiff behavior of the structure.

4. Results and discussion

As a first step, the analysis of the response dependency on velocity has been performed to find out the effects of the
increment of the rigid plate speed on the vehicle body stiffness; then the windscreens contribution to the vehicle
stiffness has been assessed. Then after the energy absorption capability of the different parts of the roof components
has been investigated and at last the results for different proposed designs are presented and discussed.

4.1. Velocity dependency

In order to decrease the computational time and cost the loading velocity increased in each step. There is a scale for
increasing the velocity based on energies ratio comparison. According to [20] during the quasi-static analysis the
ratio of kinetic energy to strain energy should be preserved less than 15%.
To analyze the speed dependency, the initial velocity (846.6 mm/s) of rigid plate has been progressively increased
by 50% in each step. Results of vehicle roof structure velocity dependency are presented in Figure 8 by means of
force-displacement diagrams for the normal production (steel) solution. As it is clear in Figure 8, by increasing the
rigid plate initial velocity the peak load and energy absorption are increased.

Figure 8. Dependency on velocity of the roof crush response

Although all the curves obtained for the different speeds are quite similar in their trend and almost converging at the
right extremity of rigid plate displacement, fluctuations appeared and curves progressively lost the smoothness due
to resonance caused by higher velocity. Since the target displacement for the puncher is set to 127 mm, the velocity
of 846.6 mm/s (that is the base one) is selected. Although it is possible to use higher velocities in order to decrease
the simulation time, the observed force oscillations will make the analyses convergence much more difficult and
may hidden other relevant effects.

4.2. Windscreens contribution to the vehicle stiffness

The effects of the front and rear windshields and of the side-screens on the vehicle stiffness are shown in Figure 9,
again by means of force-displacement diagrams. As it reported in Figure 9, three analyses situations have been
considered one without any screens, another one without front and rear windshields and the third one without side-
screens and results have been compared with all screens situation. It comes out that while the front and rear
windshields have high influence on the vehicle stiffness, the side-screens have small effects on the total stiffness. In
the case of no windscreens the peak load is reduced of 42% and this confirms the important role of the windscreens
in vehicle body stiffness.

Figure 9. Windscreens effect on vehicle body stiffness in roof crush test

4.3. Comparison of roof components energy absorption
The roof structure consists of 31 parts subdivided into the six main categories of A-Pillars, B-Pillars, C-Pillars, roof
panel, roof rails and side panels. As it is clear from the diagram of Figure 10, side panels absorb the highest amount
(about 50%) of energy during the roof crush test. The rest of absorbed energy is subdivided between B-pillars, roof
panel, roof rails, A and C-Pillars. The roof panel absorbs around 12% just by itself and this proves the importance of
the roof panel in the crushing test.

Figure 10. Energy distribution in roof structure components

4.4. Comparison of stiffness response between different designs of the roof panels

Different types of the roof panel design are proposed in Figure 11. Type 1 is the steel normal production solution,
type 2 is the composite solution that consists of one composite shell with 1.2 mm thickness, type 3 is the composite
sandwich solution with two composite skins (0.6 mm thickness each) and a full foam core, type 4 is the sandwich
solution with three longitudinal foam strips, type 5 is the sandwich solution with four longitudinal foam strips, type
6 is the sandwich solution with two longitudinal and one middle transverse foam strips and, finally, type 7 is the
sandwich solution with two longitudinal and two transverse foam strips . At the first step the sandwich structure
design is made with the EPP 45Kg/m3 foam core.

Figure 11. Roof panel solution configurations

The comparison of load-displacement curves for all the considered solutions is presented in Figure 12. When the
roof structure is under compressive force, it deforms elastically up to the about 50 mm of the rigid puncher
displacement. In the range between 50 and 60 mm of displacement the maximum of the reaction force is reached.
After that, by continuing the loading, local panel buckling failures and screens deformation begin and all the curves
exhibit a decreasing trend. It can be noticed that the steel roof panel curve, due to the material ability of plastic
deformation, exhibits force values higher with respect to the other curves and this confirms the highest amount of its
energy absorption. All the curves are included into a narrow band, the maximum force and the absorbed energy
values are very similar.

Figure 12. Roof crush resistance comparison for solutions with EPP 45 Kg/m3 foam core

The deformed shape of vehicle roof structure is compared in Figure 13 with the picture of the NHTSA test result
[40]: they are in good agreement. The deformed shape of the steel roof panel is shown in Figure 14a; the maximum
equivalent stress at the roof panel is 431 MPa which is larger than the yield strength of the considered steel (220
MPa) and therefore the roof has been submitted to a quite large plastic deformation.

Figure 13. Comparison of real and FEM roof crush test

In the case of the full composite roof panel, a rupture occurs at the roof panel edge due to its lower strength and the
curve falls instantly with higher amplitude in comparison with the other curves exactly before reaching to
displacement of 60 mm as shown at Figure 12. As the rupture occurred the failed roof panel elements are deleted
based on the predefined failure criterion as shown at Figure 14b.

a. Steel roof panel b. Composite roof panel

Figure 14. a. Steel and b. composite roofs deformation

The solution type 3 shows a more regular deformation path along with higher energy absorption with respect to the
solution type 2. When the solution type 3 is under the compressive force, at first some cracks are appeared at the
face sheet and some elements at the longitudinal edge of roof are deleted; then the foam core is submitted to
pressure, rupture occurs at the structure and elements are deleted as shown in Figure 15.

a. Outer face sheet cracking b. Foam core compression c. Foam failed elements

Figure 15. Solution type 3 deformation and failure

According to Figure 12, although the other sandwich solutions behave similarly and absorb nearly the same amount
of energy, sandwich solutions have different mass. The total absorbed energy, maximum force and solution mass are
presented in Table 5.

Table 5. Solution results comparison (45 Kg/m3 foam core) Part I

The deformed shape of foam cores for the other different solutions are shown in Figure 16, on the left the deformed
shape and on the right the failed elements of the foam core. In the case of solution types 4 and 5, the foam strip at
the passenger side remains intact while the failure occurs at the other foam strips (driver side) made visible by the
blue elements. In the cases of configuration types 6 and 7, the transverse foam strips remain intact but the failure
occurs at both the side foam strips.
 a. Type 4

b. Type 5

c. Type 6

d. Type 7

Figure 16. Solutions a. Type 4- b. Type 5- c. Type 6- d. Type 7 deformation and failure

There should be a tradeoff between the absorbed energy and roof panel mass to identify the most efficient design. In
order to have an optimum design in the case of roof crush test, the strength to weight ratio (SWR) for vehicles is
defined. The SWR and roof panel mass reduction percentages for all solutions are reported in Table 6. It is clear that
steel panel (solution type 1) has the highest SWR and then the full foam sandwich solution. The best solution among
the sandwich solutions with strips foam is sandwich type 6 due to its higher SWR.

Table 6. Solution result comparison (45 Kg/m3 foam core) Part II

At the second phase the same sandwich solutions but with higher density (EPP 70 Kg/m3) foam cores have been
tested. The load-displacement curves for different solutions with higher density foam core are presented in Figure
17. The complete results for absorbed energy value, SWR and mass report are presented in Tables 7 and 8. Based on
the results presented in Table 7, the energy absorption capacity is improved for all the material solutions by applying
the higher density foam core.

Figure 17. Roof crush resistance comparison for solutions with EPP 70 Kg/m3 foam core
Although the roof panels mass increased by applying the higher density foam cores in the sandwich solutions, SWR
increased. This means that increasing roof panel mass using sandwich solutions with higher density foam cores is an
effective method due to the foam contribution to energy absorption. Although using the solution type 4 resulting in
higher strength-to-weight ratio in comparison with the other sandwich solutions, still there is a considerable
difference for this value between sandwich solutions and the steel one.

Table 7. Solution result comparison (70 Kg/m3 foam core) Part I

Table 8. Solutions results comparison (70 Kg/m3 foam core) Part II

At the next step, Inspiring from the method presented at [30] for increasing the composite layer thickness, composite
face-sheets thickness increased from 0.6 mm to 0.8 mm. The optimized results related to sandwich designs with the
face-sheets of 0.8mm and foam core of 70 Kg/m3 density are presented at Figure 18 and Tables 9, 10. As it is clear
at Figure 18 and Table 9, by increasing the face-sheets thickness, the values of maximum force and absorbed energy
increase resulting in strength-to-weight ratio raising. According to Table 10 using the optimized solution type 6
reduces the vehicle roof panel by 68% (from 15.7 Kg to 5 Kg), while due to its equal SWR value to the steel
solution it has almost the same structural performance with the steel ones and is introduced as a best solution.

Figure 18. Roof crush resistance comparison for the solutions with EPP 70 Kg/m3 foam core and face-sheets
thickness of 0.8 mm

Table 9. Solution result comparison (70 Kg/m3 foam core-0.8 mm face-sheets thickness) Part I

Table 10. Solution result comparison (70 Kg/m3 foam core-0.8 mm face-sheets thickness) Part II

5. Conclusions
In this paper a new design solution for the vehicle roof structure has been developed and checked in rollover crash
analysis. The considered new solution consists of a sandwich structure with unidirectional carbon/epoxy composite
face sheets and EPP foam core. Structural analysis has been developed by means of numerical simulation. Before
initiating to analyze the sandwich roof structure, speed dependency test has been implemented on the steel roof
structure and the appropriate test velocity for the rigid puncher has been selected. In particular, it has been pointed
out that high test velocities added instabilities to the analysis. Then the effects of front, rear and side-screens on the
total vehicle stiffness have been tested and it was shown that while front and rear windshields have high influence
on the vehicle stiffness, side-screens have small effects on the total stiffness. The next step was aimed to determine
the importance of roof panel among the vehicle roof structure in the perspective of the energy absorption. It has been
shown that the roof panel by itself absorbs about 12% of the total vehicle roof structure energy, in case of rollover
accident.
After having implemented the roof crush analysis with the sandwich structures panel, it was observed that although
all sandwich solutions have absorbed less energy than the metallic roof solution and have slightly lower SWR, they
give very large contribution to roof panel mass reduction from 70% to 77%. After the tradeoff between the energy
absorption, mass and feasibility of the design the sandwich solution type 6 was introduced as the optimum
geometrical design. At the second step the foam core density of all sandwich solutions was increased from 45 Kg/m3
to 70 Kg/m3. The new results revealed that the increasing of the foam core density has direct relation with energy
absorption capacity and SWR. Again although the solutions absorbed energy were less than the traditional design
(except the full-foam sandwich panel solution), they gave a contribution of 73% to 75% to vehicle roof panel mass
reduction, still there was considerable difference between the steel and sandwich solutions structural performance.
At the final step, the face-sheets thickness was increased from 0.6 mm to 0.8 mm to evaluate the effect of the
increasing the layer thickness. Results revealed that, the optimized sandwich solution type 6 with the face-sheets of
0.8mm and foam core of 70 Kg/m3 density reduces the vehicle roof panel by 68% while it has almost the same
structural performance with the steel solution.

References

[1] http://www.iihs.org/iihs/topics/t/general-statistics/fatalityfacts/state-by-state-overview/2010

[2] http://www.iihs.org/iihs/topics/t/general-statistics/fatalityfacts/state-by-state-overview/2010

[3] Federal Motor vehicle Safety Standard No. 216 - Roof Crush Resistance. CFR 49 571.216, October 1997.

[4] SAE J996, “Inverted Vehicle Drop Test Procedure,” August, 1967.

[5] SAE J2114, “Dolly Rollover Recommended Test Procedure,” February 2011.
[6] Federal Motor vehicle Safety Standard No. 216 - Roof Crush Resistance. CFR 49 571 and 585, Docket No.
NHTSA-2009-0093.
[7] Gatilao F, Roesser G & Reaume B. An Overview of FMVSS 216a-Roof Crush Resistance Testing, SAE Paper
No. 2010-01-1020.

[8] Mamalis A, Manolakos D, Ioannidis M, & Papapostolou, D. On the crushing response of composite sandwich
panels subjected to edgewise compression: Experimental. Compos Struct 2005; 71(2):246-257.

[9] Harte A, Fleck NA, & Ashby MF. Energy absorption of foam-filled circular tubes with braided composite walls.
Eur J Mech A-Solid 2000; 19(1): 31-50.
[10] Tuwair H, Hopkins M, Volz J, ElGawady MA, Mohamed M, Chandrashekhara K, et al. Evaluation of sandwich
panels with various polyurethane foam-cores and ribs. Composites Part B: Engineering 2015; 79:262-276.

[11] Tarlochan F, Hamouda A, Mahdi E, & Sahari B. Composite sandwich structures for crashworthiness
applications. P I Mech Eng L-J Mat 2007; 221(2):121-130.
[12] Zhang Z, Liu S, & Tang Z. Crashworthiness investigation of kagome honeycomb sandwich cylindrical column
under axial crushing loads. Thin Wall Struct 2010; 48(1):9-18.
[13] Zangani D, Robinson M, & Gibson A. Progressive failure of composite hollow sections with foam-filled
corrugated sandwich walls. Appl Compos Mater 2007; 14(5):325-342.
[14] Shah QH & Topa A. Modeling large deformation and failure of expanded polystyrene crushable foam using
LS-DYNA. Model Simul Eng 2014.
[15] Li Z, Zheng Z, Yu J, & Guo L. Crashworthiness of foam-filled thin-walled circular tubes under dynamic
bending. Mater Design 2013; 52:1058-1064.
[16] Steeves CA & Fleck NA. Collapse mechanisms of sandwich beams with composite faces and a foam core,
loaded in three-point bending. Part II: Experimental investigation and numerical modelling. Int J Mech Sci 2004;
46(4):585-608.
[17] Rizov V, Shipsha A, & Zenkert D. Indentation study of foam core sandwich composite panels. Compos Struct
2005; 69(1):95-102.

[18] Flores-Johnson E & Li Q. Experimental study of the indentation of sandwich panels with carbon fibre-
reinforced polymer face sheets and polymeric foam core. Composites Part B: Engineering 2011; 42(5):1212-1219.
[19] Mao M, Chirwa E & Chen T. Reinforcement of vehicle roof structure system against rollover occupant injuries.
Int J Crashworthiness 2007; 12(1):41-55.

[20] Pan F & Zhu P. Design optimisation of vehicle roof structures: Benefits of using multiple surrogates. Int J
Crashworthiness 2011; 16(1): 85-95.
[21] Kulkarni SC & Marks M. Thermoplastic Roof Crush Countermeasure Design for Improved Roof Crush
Resistant to Meet FMVSS-216, SAE Paper No. 2011-01-1119.
[22] Pathare R & Mansour M. Automotive Roof Crush, Structural Foam Enhancement Solution, SAE Paper No.
2009-01-0371.

[23] Bambach M. Fibre composite strengthening of thin steel passenger vehicle roof structures. Thin Wall Struct
2014; 74: 1-11.

[24] NCAC. Finite element model archive. National Crash Analysis Centre, George Washington University. 〈
https://www.ccsa.gmu.edu/capability〉. [Accessed 1.2013].

[25] Mao M, Chirwa E, Chen T & Latchford J. Static and dynamic roof crush simulation using LS-DYNA3D. Int J
Crashworthiness 2004; 9(5): 495-504.

[26] Peng Y, Yang J, Deck C & Willinger R. Finite element modeling of crash test behavior for windshield
laminated glass. Int J Impact Eng 2013; 57: 27-35.

[27] http://www.dynasupport.com/howtos/element/hourglass
[28] Tomblin J, Sherraden J, Seneviratne W, Raju KS. A-basis and B-basis Design Allowables for Epoxy-based
prepreg Toray T700SC-12K-50C/#2510 Plain Weave Fabric. AGATE-WP3.3-033051-131; September 2002.

[29] T700SC 12K/2510 Plain Weave Fabric. Composite Materials Handbook (CMH- 17), vol. 2. Rev. G. [chapter
4.2.38]

[30] Borazjani S & Belingardi G. Lightweight design: Detailed comparison of roof panel solutions at crash and
stiffness analyses. Int J Crashworthiness 2016; 1-14.
[31] Croop B, Lobo H. Selecting material models for the simulation of foams in LS-DYNA. Proceedings of the 7th
European LS-DYNA Conference, Dynamore Gmbh, Salzburg, Germany, 2009.

[32] Belingardi G, Duella R, Caminiti A. Optimal Choice of the Foam Design Parameters in Order to Meet the HIC
Index Limit of the FMVSS201 Standard, SAE Paper No. 2002-01-2095.
[33] LS-DYNA Keyword User’s Manual - Volume II. Material Models. Version R7.0. Livermore Software
Technology Corporation, February 2016.

[34] Ls-dyna theory manual, Hallquist J. O. (Ed.), California: Livermore Software Technology Corporation, 2006.
[35] Barsotti M. Comparison of FEM and SPH for modeling a crushable foam aircraft arrestor bed. Proceedings of
the 11th International LS-DYNA Users Conference, 2010; 16.
[36] K. Weimar and J. Day, “Negative volumes in brick elements,” 2003,
http://www.dynasupport.com/howtos/element/negativvolumes- in-brick-elements.
[37] S. Bala, “Best practices for modeling recoverable low density foams-by example,” 2006,
http://blog2.d3view.com/bestpractices- for-modeling-recoverable-low-density-foams-by-example/.
[38] Avalle M, Belingardi G & Ibba A. Mechanical models of cellular solids: Parameters identification from
experimental tests. Int J Impact Eng 2007; 34(1):3-27.
[39] Gibson LJ, Ashby MF. Cellular solids. Structure and properties. Cambridge: Cambridge University Press;
1997.

[40] NHTSA TestC0438.Toyota Yaris 4-door sedan with 5º pitch, 25º roll. National Highway Traffic Safety
Administration (NHTSA), Vehicle Research and Test Center (VRTC), Ohio, USA; 2007.
 Figures captions list

Figure 1. FMVSS 216 test configuration

Figure 2. Load case and boundary condition

Figure 3. Different methods to model the windshield laminated glass

Figure 4. Schematic of sandwich panel

Figure 5. a. Original and b. exponentially extrapolated stress-strain curves for EPP 45 Kg/m3

Figure 6. a. Original and b. exponentially extrapolated stress-strain curves for EPP 70 Kg/m3

Figure 7. EPP foam density vs. compressive Young’s modulus

Figure 8. Dependency on velocity of the roof crush response

Figure 9. Windscreens effect on vehicle body stiffness in roof crush test

Figure 10. Energy distribution in roof structure components

Figure 11. Roof panel solution configurations

Figure 12. Roof crush resistance comparison for solutions with EPP 45 Kg/m3 foam core

Figure 13. Comparison of real and FEM roof crush test

Figure 14. a. Steel and b. composite roofs deformation

Figure 15. Solution type 3 deformation and failure

Figure 16. Solutions a. Type 4- b. Type 5- c. Type 6- d. Type 7 deformation and failure

Figure 17. Roof crush resistance comparison for solutions with EPP 70 Kg/m3 foam core

Figure 18. Roof crush resistance comparison for the solutions with EPP 70 Kg/m3 foam core and face-sheets
thickness of 0.8 mm
 Tables

Table 1. Glass and PVB main material properties

ρ (kg/m3) E (GPa) ν Failure plastic strain

Glass 2500 74 0.227 0.001
PVB 1100 2.6 0.435 -

Table 2. Composite material elastic properties

ρ E1 E2 ν12 G12 G23 G13

(Kg/m3) (GPa) (GPa) (GPa) (GPa) (GPa)
Carbon/Epoxy -
Unidirectional [28, 1520 127 8.41 0.0205 4.21 4.21 4.21
29]

Table 3. Strength parameters of composite

Xt (MPa) Xc (MPa) Yt (MPa) Yc (MPa) S (MPa)

Carbon/Epoxy –Unidirectional 2200 1470 48.9 199 154
 Table 4. LS-DYNA available foam material models

Material Failure Strain-rate

Typical Applications
Card Consideration Effects
MAT53 No No Used to model polyurethane foam for automotive applications
Used to model urethane foam for seat cushions and padding on the
MAT57 Yes Yes
side impact dummies (SID)
MAT61 No Yes Used to model viscoelastic bodies such as foams.
MAT62 No Yes Used to model CONFOR foam on the ribs of side impact dummy
Used for foams in impact applications where cyclic behavior is
MAT63 No No
unimportant.
Used to model low Density Urethane Foam in seat cushions,
MAT73 Yes Yes
padding on the side impact dummies, bumpers and interior foams.
MAT75 No Yes Used to model isotropic crushable foams
Capable of rate effects consideration for low and medium density
MAT83 Yes Yes
foams.
Used to model low density, transversely isotropic crushable foams
to enhance automotive safety in low velocity (bumper impact) and
MAT142 No No
medium high velocity (interior head impact and pedestrian safety)
applications.
MAT144 No Yes Used to model isotropic crushable foams with strain rate effects.
MAT154 Yes No Used to model aluminum foam
Used to model crushable foam with optional damping, tension
MAT163 No Yes
cutoff, and strain rate effects.
MAT177 No No Used to model highly compressible foam
MAT178 No Yes Used to model highly compressible foam with strain rate effects
Used to model highly compressible synthetic foam that are used in
MAT179 Yes Yes
some bumper designs.

Table 5. Solution results comparison (45 Kg/m3 foam core) Part I

Absorbed Maximum
Solution Mass (Kg)
Energy (kN.m) Force (kN)
Type 1 3.40 33.8 15.69
Type 2 3.06 30.6 3.03
Type 3 3.14 33.1 4.80
Type 4 2.95 31.0 3.73
Type 5 2.96 30.9 3.73
Type 6 2.97 31.3 3.65
Type 7 2.96 30.7 3.65
 Table 6. Solution results comparison (45 Kg/m3 foam core) Part II

Strength-to-Weight Panel Mass Reduction

Solution
ratio (SWR) Percentages (%)
Type 1 3.087 -
Type 2 2.827 80.71
Type 3 3.053 69.42
Type 4 2.862 76.23
Type 5 2.853 76.24
Type 6 2.890 76.76
Type 7 2.835 76.75

Table 7. Solution results comparison (70 Kg/m3 foam core) Part I

Solution Absorbed Maximum Mass (Kg)

Energy (kN.m) Force (kN)
Type 4 2.98 31.6 4.12
Type 5 3.01 31.2 4.12
Type 6 3.01 31.4 3.99
Type 7 3.01 30.9 3.99

Table 8. Solutions results comparison (70 Kg/m3 foam core) Part II

Strength-to-Weight Panel Mass Reduction

Solution
ratio (SWR) Percentages (%)
Type 4 2.917 73.74
Type 5 2.880 73.74
Type 6 2.898 74.57
Type 7 2.852 74.57

Table 9. Solution results comparison (70 Kg/m3 foam core-0.8 mm face-sheets thickness) Part I

Solution Absorbed Maximum Mass (Kg)

Energy (kN.m) Force (kN)
Type 4 3.11 32.4 5.13
Type 5 3.13 32.4 5.13
Type 6 3.19 33.5 5.00
Type 7 3.17 32.8 5.00
Table 10. Solution results comparison (70 Kg/m3 foam core-0.8 mm face-sheets thickness) Part II

Strength-to-Weight Panel Mass Reduction

Solution
ratio (SWR) Percentages (%)
Type 4 2.993 67.30
Type 5 2.987 67.30
Type 6 3.087 68.13
Type 7 3.021 68.13