Issue with crack propagation in LS-DYNA simulation of bullet impact on laminated glass sample

[Rajat Bhaisare]

Dear LS-DYNA Users,

I am simulating bullet impact on laminated glass to validate my numerical results against experiments. The laminated glass configuration is 6 mm float glass + 1.52 mm EVA interlayer + 6 mm float glass. The glass is modeled using MAT_110 (Johnson–Holmquist) with material properties taken from literature, and the EVA interlayer is modeled using MAT_024 (piecewise linear plasticity), also based on literature data. The model uses Lagrangian solid elements, and the projectile is modeled as a rigid bullet impacting at 135 m/s. I have also varied the impact velocity, but the damage pattern remains similar.

The issue is that the damage remains highly localized near the impact point, forming a circular radial damage zone, with no crack propagation towards the panel edges and no clear in-plane radial cracks on either the front or back glass face. This behavior does not match my experimental observations, where long radial cracks extending towards the edges and cracking on both front and back faces are clearly visible.

To address this, I have tried adaptive FEM-to-SPH conversion and full SPH modeling for the glass. Although fragmentation and the number of cracks increase, the cracks still do not propagate towards the edges, and the overall crack pattern does not correlate well with the experimental results, even after changing the impact velocity.

I would appreciate guidance on how to achieve realistic crack propagation and crack patterns in laminated glass under bullet impact in LS-DYNA, particularly regarding damage and failure parameters in MAT_110, mesh strategy, erosion criteria, glass–EVA contact definition, or alternative modeling approaches suitable for this problem. I will attach images of the experimental specimen and the numerical results for comparison.

Thank you all for your time and support 

[Евгений Калентьев]

Hello Rajat Bhaisare

How to connect glass and EVA? Change you material property for MAT_110 vs literature source? Share your k file

Sincerely,

Eugene Kalentev

пн, 12 янв. 2026 г., 19:48 Rajat Bhaisare <rajatbh...@gmail.com>:

--
You received this message because you are subscribed to the Google Groups "LS-DYNA2" group.
To unsubscribe from this group and stop receiving emails from it, send an email to ls-dyna2+u...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/ls-dyna2/657b10e3-5821-4bfd-85e9-c9e9a3a0e632n%40googlegroups.com.

[Rajat Bhaisare]

Hello Eugene Kalentev,

The Glass and EVA is connected with using Automatic Surface to Surface tiebreak and then the bullet and LG panel with Eroding surface to surface, Keyword for this work is attached. https://doi.org/10.1177/1369433219845691 - Float Glass properties are taken from this paper 

for EVA properties i was not able to find a specific paper in which  EVA is used as interlayer for laminated glass panel, so i have taken help from this website for properties of EVA - https://engineercalculator.com/polymer-plastic-properties-and-overview/eva-polymer-plastic-various-properties-and-overview/
if you have any suggestions for EVA properties or material model please let me know,in the attached keyword i used PVB properties for interlayer still the cracking pattern remained same i.e no crack was propogating towards the edge and a localised failure similar to previously attached image.

Thank you
Rajat Bhaisare

 keyword_for_the_sample.k

[Евгений Калентьев]

Hello Rajat Bhaisare.
I looked at your model and noticed that some of your material model constants differ from those specified in the literature. I tried using these (as in the screenshot), and the cracks began to look more realistic. The result depends heavily on the combination of specific parameters; try changing them. To simplify the model, I left only the top glass layer. You might also want to carefully examine the contact between the glass and EVA.

среда, 14 января 2026 г. в 11:14:36 UTC+4, rajatbh...@gmail.com:

[Valery Anpilov]

Dear  Rajat,

I don't think using material model 24 for EVA is a good idea.

If I'm not mistaken, EVA has viscous properties and doesn't fail according to the elastic-plastic failure mechanisms.

Try using material models from the PolyUMod library?!
It has the ability to take loading rates into account.

However, your problem involves very high strain rates, and I don't know how applicable the material models from the PolyUMod library are for such high strain rates.

I also don't understand the possibility of accurately modeling fracture at such rates.

Also, the question of whether it's possible to accurately model EVA failure at such high rates is unclear.

Perhaps it would make sense to increase the number of cells, at least in the active deformation region?

[Emin Akca]

Dear Rajat,

First of all, check your warning messages all the time during the initialization of your runs. There is a warning message about your *MAT_024 implying that your curve has a greater slope than the elastic modulus. When using *MAT_024, you should define your curves as true stress vs. true plastic strain so you should remove the elastic portion for your EVA material definition. DYNAmore has a great webinar about *MAT_024 on YouTube, you can watch here: https://www.youtube.com/watch?v=J5Qjj_9Sot0.

Second, a material model parameter set provided in the literature does not have to be precisely true for your material. Regarding crack propagation and element failure, check the damage equations of JH-2 constitutive model and understand D1 and D2 mathematically. FS parameter is also contributing to the element failure. Therefore, simply you can change D1, D2, and FS to optimize your element deletion settings.

As a comparison, with the same element size, you may also try tetrahedron elements for the sake of crack pattern but you should be careful since tets behave stiffer than hex elements geometrically and remember that they are less effective than hexes.

Best,

—Emin

[Rajat Bhaisare]

Hello Eugene,

I used the material properties you suggested, and the numerical results obtained were close to the experimental observations of the front and the back face. I have attached the simulation result for the front and back face. However, when I replaced those parameters with silica float glass properties, the crack patterns were not that realistic.

The parameters for silica float glass were taken from the following paper:
https://www.semanticscholar.org/paper/Implementation-and-Validation-of-the-Ceramic-Model-Cronin-Bui/185b0f0e3c4090132c6a1e427e3ec26450faa82a (Cronin & Bui).

At present, I am still working on achieving more realistic crack development by improving the experimental validation and refining the material properties. Regarding the contact definition between the EVA interlayer and glass, I plan to study the LS-DYNA manual in detail, try different contact formulations, and run a few trial simulations. I will update you once I obtain more consistent results.

Thank you for your guidance.

Best regards,
Rajat Bhaisare

 

On Thu, 15 Jan 2026 at 13:51, Евгений Калентьев <eugenek...@gmail.com> wrote:

Hello Rajat Bhaisare.

I looked at your model and noticed that some of your material model constants differ from those specified in the literature. I tried using these (as in the screenshot), and the cracks began to look more realistic. The result depends heavily on the combination of specific parameters; try changing them. To simplify the model, I left only the top glass layer. You might also want to carefully examine the contact between the glass and EVA.

Sincerely,

Eugene Kalentev

вт, 13 янв. 2026 г. в 16:27, Rajat Bhaisare <rajatbh...@gmail.com>:

Hello Eugene Kalentev,

the Glass and EVA is connected with using Automatic Surface to Surface tiebreak and then the bullet and LG panel with Eroding surface to surface the material properties are mentioned below for float glass and EVA, I am not allowed to share the keyword file for this, hope you understand and the literature from which the properties are taken is Zhou et al ,2019 for the float glass and EVA i got the mechanical properties from few papers and from a website and the stress strain curve for EVA is taken from a random paper if eva stress strain curve is required i can also share that.

*MAT_JOHNSON_HOLMQUIST_CERAMICS_TITLE
Float_glass
$#     mid        ro         g         a         b         c         m         n
         1    2530.02.10000E10       0.7       0.2     0.035       1.0      0.72
$#    epsi         t     sfmax       hel      phel      beta
       1.02.780000E7       0.54.500000E92.400000E9       1.0
$#      d1        d2        k1        k2        k3        fs
     0.043      0.853.05000E103.05000E101.03000E11    0.0024

*MAT_PIECEWISE_LINEAR_PLASTICITY_TITLE
EVA_interlayer
$#     mid        ro         e        pr      sigy      etan      fail      tdel
         4     950.05.000000E7      0.45   4000000       0.0    0.0024       0.0
$#       c         p      lcss      lcsr        vp
       0.0       0.0         4         0       3.0
$#    eps1      eps2      eps3      eps4      eps5      eps6      eps7      eps8
       0.0       0.0       0.0       0.0       0.0       0.0       0.0       0.0
$#     es1       es2       es3       es4       es5       es6       es7       es8
       0.0       0.0       0.0       0.0       0.0       0.0       0.0       0.0

Thank you 
Rajat Bhaisare

[Rajat Bhaisare]

Dear Valery,

Thank you for your comments.

Based on the literature, for short-duration blast loads (≤ 100 ms), changes in the shear modulus of polymeric interlayers are reported to be negligible, and they are often modeled as elastic–plastic materials under blast loading (Larcher et al.; Hidallana-Gamage et al.; Wei and Dharani). With this assumption, I initially used *MAT_024 for EVA as a simplified approach.

I also found a recent study where EVA was modeled using a bilinear isotropic hardening law in ANSYS AUTODYN, link for this is given below. The closest equivalent in LS-DYNA appears to be *MAT_003 (Plastic Kinematic), which can reproduce bilinear isotropic hardening behavior. I am currently reviewing this model and plan to use it for EVA.

 Paper - "Influence of interlayer types and thicknesses on the blast performance of laminated glass panels", DOI : https://doi.org/10.1016/j.istruc.2023.105231 

Regarding the PolyUMod library, I am not yet familiar with it, but I will read about it and evaluate its applicability for high strain-rate blast loading.

I will also refine the mesh in the active deformation region, as you suggested.

Thank you again for your guidance.

Best regards,
Rajat Bhaisare

--

You received this message because you are subscribed to the Google Groups "LS-DYNA2" group.
To unsubscribe from this group and stop receiving emails from it, send an email to ls-dyna2+u...@googlegroups.com.

To view this discussion visit https://groups.google.com/d/msgid/ls-dyna2/0cc221db-e6ef-4778-aa34-a18c58342398n%40googlegroups.com.