Learning Forum

[Mohamed Farmaan]

I am doing a simple tensile strength analysis for blend A and blend B in static structural. I am comparing the results with the experimental results, and I found that the stress results are perfect with a marginable deviation; the strain and the total deformation are deviating by a lot. I couldn't figure out what might be the reason. I have done the mesh independence study. I have attached the setup of the tensile strength analysis.

 

column B and C are tensile load applied for blend A and blend B respectively. column D and E are for blend A and coloumn F and G are for blend B respectively.

The experimental deformation for the condition 'DP 0' is 0 mm for blend A and blend B respectively, and 1 mm and 0.082 mm for the 'DP 1' condition, respectively. But my result is far off in terms of total deformation for DP 1.

I can provide more information, but being new to the material analysis field, I am not sure what to attach here.

[Armin]

Hi Mohamed,

Do you have force versus displacement data from experiment to compare against the simulation? If so, it would be helpful to identify the issue.

As a general suggestion, you can apply displacement-based or velocity-based boundary conditions for loading a tensile sample rather than force-based boundary conditions (like in your case). Velocity-based boundary conditions are also better in line with how samples are tested in labs.

[Mohamed Farmaan]

 

Yes, I do, but still, the results are bad. I tried with displacement too. It’s the exact opposite when I take the displacement as the input boundary condition. The stress is far off, but the strain is close to perfect. 

 

[Armin]

I see... it is best to overlay the experimental and numerical force-displacement or engineering stress-strain curves to identify potential issues. For example, if the initial slope is off, it is likely related to Young's modulus or elastic region, while divergence at larger strains is likely related to plasticity parameters.  

[Mohamed Farmaan]

Hello Armin, I finally got the stress-strain curve results to match the experimental results. The mistake I made was not defining the elastic and plastic regions in engineering data for the material; for the elastic region, I took it as an isotropic property, while for the plastic region, I took it as a multilinear isotropic harndening. Thank you very much, Armin, but I am facing an other issue now, I am not able to read the fracture region of the stress-strain curve. From my research, if I am using static structural they have mentioned the use of EKILL APDL command. I couldn't find any script for it, if you are familiar with it or if you are familiar with any other ideas can you help me.

[Armin]

Hi Mohamed,
No worries. Could you include a plot of the result you have so far (compared with experiment)? 

[Mohamed Farmaan]

Sure, I have attached the result below.

[Armin]

Thanks Mohamed.
I think the discrepancy between the model and experiment is related to post-necking regime where the experiment localizes earlier than the model (evident by the load drop) followed by a complete fracture that is not resolved by the model. To capture the post-localization regime better, you can work on your hardening curve (input stress-strain curve) after the onset of necking. This will help to align your simulation data with experiment after the onset of necking. There are some techniques available for "post-uniform hardening characterization" that I suggest that you look them up online but the following paper may be helpful to you: https://www.sciencedirect.com/science/article/pii/S2666359722000208

[Mohamed Farmaan]

got it Armin, thanks a lot. I will get back to you once I do some research and try some of the methods. Thanks for the paper a lot.