Sei sulla pagina 1di 5

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/241173314

Simulation on High Velocity Impact Process of Windshield by SPH/FEM Coupling


Method

Article · August 2010


DOI: 10.1109/ICIE.2010.268

CITATIONS READS
7 294

3 authors:

Mengyan Zang Hu Chen


South China University of Technology UNSW Sydney
81 PUBLICATIONS   638 CITATIONS    8 PUBLICATIONS   44 CITATIONS   

SEE PROFILE SEE PROFILE

Zhou Lei
Los Alamos National Laboratory
53 PUBLICATIONS   688 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Development of numerical algorithms for the impact fracture simulations of laminated glass View project

FEM Application View project

All content following this page was uploaded by Hu Chen on 17 July 2014.

The user has requested enhancement of the downloaded file.


2010 WASE International Conference on Information Engineering

Simulation on High Velocity Impact Process of


Windshield by SPH/FEM Coupling Method
M.Y. Zang H. Chen∗ and Z. Lei
School of Mechanical & Automotive Engineering School of Mechanical & Automotive Engineering
South China University of Technology South China University of Technology
Guangzhou 510640, People’s Republic of China Guangzhou 510640, People’s Republic of China
Email: myzang@scut.edu.cn ∗ Email: kochin2008@126.com

Abstract—The impact resistance of windshield, especially un- results are presented. Compared with the experiment results,
der the high velocity impact, is very important for passenger’s the high impact fracture process of windshield is discussed in
safety. In this paper, a SPH/FEM (Smoothed Particle Hydrody- detail.
namics and Finite Element Method) coupling method was used
to analyze the high velocity impact process of windshield. The II. N UMERICAL M ODEL
glass of impact side was discretized by SPH particles to describe
the splash performance of glass fragments, while the other layers A. SPH/FEM Analysis Model
of windshield were discretized by hexahedron elements. SPH and In this paper, SPH/FEM coupling method is used to analyze
FEM were combined by defining a contact algorithm. Compared the high velocity impact process of windshield. As shown in
with the experiment results, it is shown that the simulation results
are consistent with experimental results well, and the SPH/FEM Fig. 1, a windshield plate, which is supported by rubber gas-
coupling method can be applied to analyze the high velocity kets at the four corners, is impacted by an aluminum ball at the
impact failure problem well. center of the upper surface. The windshield plate, which size
Keywords-SPH/FEM coupling method; windshield; high veloc- is 850 mm×650 mm×21 mm, is a laminated composite with
ity impact; numerical simulation four layers: the first and third layers are two piece of glasses,
the second layer is polyvinyl butyral membrane (i.e.PVB), the
I. I NTRODUCTION forth layer is polythylene terephthalate membrane (i.e.PET).
Windshield is an important component of high speed train, The diagram of the section and the thickness of each layer are
which affects riding comfort, permeability and safety. Usually shown in Fig. 2. The aluminum ball, which initial velocity and
it is required to have good shock resistance, penetration mass are 586 km/h and 1 kg respectively, is come to contact
resistance and nice permeability. Nowadays, both conventional with the windshield plate at the initial time.
experimental methods and modern computational simulation
methods are used for high velocity impact research of the
windshield. Timmer and Kolling [1] of the Daimler-Chrysler
Corporation ever presented a simplified finite element wind-
shield model, and verified it by the experiments, but it could
not describe the splash performance of glass fragments effec-
tively. Zang and Lei [2] applied 3-dimension Discrete Element
Method (DEM) to simulating penetration process of laminated
glass under high velocity impact, and studied the extension
of fracture and the splash performance of glass fragments,
but it needs to do more research for engineering application
because of the low computational efficiency of DEM and the
incompletion of material model.
To describe the splash performance of glass fragments dur-
ing impact process, the SPH/FEM coupling method [3]-[5] in
the commercial software Ls-Dyna [6] is applied to simulating Fig. 1. The half simulation model.
the high velocity impact process of windshield in this paper.
The glass of impact side is discretized by SPH particles to In the present work, the upper layer of glass, with 4.0 mm
describe the splash performance of glass fragments, while thickness, is modeled by using SPH to catch the phenomenon
the other layers of windshield are discretized by hexahedron of the fractures dispersion, while the rest parts are analyzed
elements. In the following sections, the SPH/FEM coupling by using FEM. In the SPH calculation region, all of the
model as well as the constitutive models and failure models particles are uniformly distributed, and a relative small initial
will be described at first. Then, the numerical simulation particle spacing is set to make sure the analysis precision.

978-0-7695-4080-1/10 $26.00 © 2010 IEEE 381


DOI 10.1109/ICIE.2010.268
Fig. 2. Diagram of the section of the windshield.

Meanwhile, hexahedron elements are used in the FEM domain.


Considering the distribution of stress, the density of meshes is
different in different place. For example, refined meshes are
Fig. 3. Variations of Young’s modulus and tensile strength of PVB on strain
used in the center of plate, while coarse meshes are used in rate.
the periphery. To catch the failure in the inner glass layer, the
approach of TNWF (tied nodes with failure) is used [7]. Here,
coincident nodes are generated in the inner glass layer and Assuming the experiment result of 0.425E-3(1/s) (Young’s
then tied together with a constraint relation. These tied nodes modulus is 5.88 MPa and tensile strength is 18.62 MPa)
remain together until the volume-weighted effective plastic as the static value, Young’s modulus and tensile strength
strain, averaged over all elements connected to the nodes in a corresponding to strain rate can be calculated approximately
given constraint, exceeds a specified value. Once this value is as follows:
exceeded, all nodes in that constraint are released to simulate E = a1 eb1 ˙ − c1 ed1 ˙ − e1 (2)
the initiation of a crack, fracture or penetration. Self-contact
T = a2 eb2 ˙ − c2 ed2 ˙ − e2 (3)
is defined here to monitor the possible contact after failure.
In the Ls-Dyna, SPH particles and finite elements can be Where a1 = 4684.0, b1 = 2.056E−7, c1 = 4600.0,
combined by the definition of ”Node-to-Surface” contact al- d1 = −1.367E−4, e1 = 70.0, a2 = 2253.0, b2 = 2.571E-7,
gorithm. Where, SPH particles are defined as slave nodes, and c2 = 2004, d2 = −1.751E−4 and e2 = 130.0 are material
finite element’s surfaces linked to SPH particles are defined as parameters, respectively; E, T and ˙ are Young’s modules,
master surfaces. The SPH particles and finite element meshes tensile strength and strain rate, respectively.
are tied together until meeting a failure critical law. Material parameters for analysis are given as below:
Considering the symmetry, a half numerical model, which
TABLE I
contains 22,100 hexahedron elements and 22,634 SPH parti- MATERIAL PARAMETERS.
cles, is established here to improve simulation efficiency.
Density Young’s Poisson’s Failure
B. Constitutive Models
Material Module Ratio Strain
The material of glass is assumed as isotropic elastoplastic, (Kg/m3 ) (GPa) (-) (-)
and Cowper-Symonds model is used here. Cowper-Symonds Aluminum 2700 71.7 0.33
model is a strain rate dependent elastic-plastic model. In this Glass 2500 74.1 0.225 0.007
model, strain rate is accounted and the yield stress is scaled PVB 101 0.006 0.42
by the strain rate dependent factor as shown below:
  1/P 
PET 1390 3.6 0.37
˙ Rubber 1200 0.61 0.45
σ = σs () 1 + (1)
D
Where σs is the initial yield stress, ˙ is the strain rate, D and C. Failure Models on the Interfaces
P are the Cowper-Symonds strain rate parameters. Values of In the impact fracture process, especially in the high velocity
D=1000.0 (1/s) and P =100.0 are used here for glass material. impact fracture process, the materials translate from continuum
Because glass is brittle material with large hardness, it is to noncontinuum uninterruptedly. To catch the failure on the
usually setting a plastic strain failure value to judge whether interface between each two layers, two failure models are used
glass is damaged or not. The failure strain value of glass in the present work.
is 0.007. The same constitutive model is also used for PET, The first failure model is used between the SPH particles
where the values of D=846.0 (1/s) and P =87.0 are set. and finite element meshes, when meeting the following failure
Compared to glass, the Young’s modulus of PVB is greatly critical law:  α  β
related to strain rate in high velocity impact process. Besides, |fn | |fs |
+ ≥1 (4)
it is predicted that the tensile strength also changes corre- Fn Fs
sponding to strain rate. In order to find the relation, tensile The failure takes place. Where fn and fs are the normal and
experiments of PVB were carried out with three strain rates shear interface force; and Fn and Fs are the normal and shear
(0.425E-3, 0.425, 156.5 (1/s)) by Oda and Zang[8]. Fig. 3 failure force; α and β are the exponent for normal and shear
shows the results. force.

382
Another failure model is implemented in the finite element III. S IMULATION R ESULTS
region of the windshield plate between each two layers. The The impact process of windshield simulated by using
failure critical law is: SPH/FEM coupling method is shown in Fig. 6. At the be-
 2  2
|σn | |σs | ginning of impact, the shear failure mainly takes place at the
+ ≥1 (5) centre of windshield. During the impact process, compression
σn0 σs0
wave translates into extension wave by reflecting, which
Failure occurs when left side is larger than 1. Where σn and makes failure area become large and extend to periphery,
σs are the normal and shear stresses; σn0 and σs0 are the and circumferential fracture takes place around the centre of
normal and shear failure strengths. windshield leading to the splash of glass fragments. Compared
D. Prestress Analysis to experimental results (as shown in Fig. 7), it is obvious that
the numerical simulation results can clearly describe the splash
As mentioned above, the windshield plate is supported by performance of outer glass and is consistent with experimental
rubber gaskets at the four corners. At each corner, there are results well.
two pieces of gaskets, of which one supports the upper surface
and another one supports the lower surface of the windshield
plate. The windshield plate is fixed by clamping those gaskets.
To consider the influence of the prestress in the windshield
plate, the clamping process is simulated and the prestree is
taken. Here, the lower surface of the rubber gaskets under the
windshield plate are fixed, and the upper surface of the rubber
gaskets above the windshield plate are pressed by rigid plates,
as shown in Fig. 4. The displacement of the rigid plate is the
same as that in the experiment (will be shown in the next
section) and Fig. 5 shows the distribution of von mises stress
in the PVB layer. The stress concentration only takes place at
four corners and is almost symmetrical.

Fig. 6. Simulation results of impact process.

Time variation of normal impact force of aluminum ball


is shown in Fig. 8. In the process of impact, the normal
impact force of aluminum ball increases rapidly, and reaches
Fig. 4. The simulation model of prestress analysis. to the peak value at time 0.4 ms. It also means the normal
acceleration of aluminum ball reaches to the maximum. After
that, the normal resistance of windshield is decreasing because
of the failure of glass elements and the cracks propagation. The
separation between windshield and aluminum ball happens at
time about 1.5 ms. Since 3.2 ms, the normal impact force of
aluminum ball increases again for the springback of PVB.
The penetration usually happens in the case of large de-
formation. Generally, the large deformation is related to the
displacement in the normal direction of windshield. Fig. 9
shows that no penetration takes place in PVB and PET in the
case of achieving largest displacement in the normal direction
of windshield.
During the impact process, large deformation takes place
in both PVB and PET. After the rebound of aluminum ball,
Fig. 5. The distribution of von mises stress in the upper surface of PVB. the PVB layer is springing back, but a convex deformation is
reamined in PET caused by plastic strain (Fig. 10), and the

383
Fig. 10. The distribution of plastic strain.
(a) (b)

(c) (d)

Fig. 7. Experimental results. Fig. 11. Plastic deformation of PET.

50 IV. C ONCLUSION
This paper applies the SPH/FEM coupling method to nu-
40
merical analyze the high velocity impact process of wind-
Impact Force (KN)

30
shield. The splash performance of the glass on the impact side
and the convex of the plastic deformation of PET are compared
20 with experimental results, and it indicates that SPH/FEM
coupling method can analyze high velocity impact problem
10 effectively and accurately.
0 ACKNOWLEDGMENT
1 2 3 4 5 6
Time (ms) This work was supported by the International Cooperation
Project of the Ministry of Science and Technology of China
Fig. 8. Time variation of the normal impact force.
(No. 2008DFA51740) and the National Natural Science Foun-
dation of China (No. 10972079).

corresponding experimental result is shown in Fig. 11. R EFERENCES


From the above analysis, it is shown that the SPH/FEM [1] M. Timmel, S. Kolling and P Osterrieder,“A finite element model for
impact simulation with laminated glass,” International Journal of Impact
coupling method can be applied to analyze the high velocity Engineeringvol.34, 2007pp.1465-1478.
impact failure problem well. [2] M. Y. Zang, Z. Lei and S. F. Wang, “Investigation of impact fracture
behavior of automobile laminated glass by 3D discrete element method,”
Computational Mechanics vol.41, 2007pp.73-83.
[3] S. Attaway, M. Heinstein and J. Swegle, “Coupling of smooth particle
hydrodynamics with the finite element method,” Nuclear Engineering
and Design, vol.150, 1994, pp.199-205.
[4] T. D. Vuyst, R. Vignjevic and J. C. Campbell, “Coupling between
meshless and finite element methods,” International Journal of Impact
Engineering, vol.31, 2005, pp.1054-1064.
[5] L. Aktay, A. F. Johnson, “FEM/SPH coupling technique for high velocity
impact simulations,” Advances in Meshfree Techniques, 2007, pp.147-
167.
[6] J. O. Hallquist, “LS-DYNA theretical manual,” Livermore Software
Technology Corporation, 1998.
[7] N. F. Knight, N. Jaunky N, R. E. Lawson, et al, “Penetration and
simulation for uncontained engine debris impact on fuselage-like panel
using LS-DYNA,” Finite Elements in Analysis and Design, vol. 36(2),
2000, pp.99-133.
[8] J .Oda, M. Y. Zang, “Analysis of impact fracture behavior of laminated
(a) PVB (b) PET glass of bi-layer type using discrete element method,” Key Engineering
Materials, vol.145, 1998, pp.349-354.
Fig. 9. Displacement distribution in the normal direrection of windshield.

384

View publication stats

Potrebbero piacerti anche