Cyclic loaading - reg

[Prasanna Tatavarty]

Hello Reza and team,

   I'm trying to model the cyclic behaviour of Aluminum using PRISMS plasticity. In the prm file in cyclic loading folder, there are two extra paratmeters related to kinematic hardening. What are these parameters? Can you provide the law/equation related to them? And how could I obtain these from my experimental data?

Regards

Prasanna Tatavarty

[Chaitali Patil]

Hi Prasanna, 

You can refer to the articles using Prisms-fatigue, (references from thereof) for the model discussion, such as 
https://doi.org/10.1016/j.actamat.2021.117524

https://doi.org/10.1016/j.jmrt.2022.06.075

-Regards,

Chaitali 

[Prasanna Tatavarty]

Hello team, 

     Can anyone help with this? "I've gone through the given papers. Do PRISMS Plasticity and PRISMS Fatigue use the same kinematic hardening law (Ohno-Wang)? If yes, can I extract the parameters from my experimental LCF data? "

Regards

Prasanna Tatavarty

[Kamin Tahmasbi]

Hi Prasanna,

Please see the attached image clarifying the kinematic hardening parameters used in Ohno-Wang two-term back stress model for L-PBF AlSi10Mg as well as the calibrated tensile results to have a better explanation. The CPFE model parameters were iteratively adjusted to provide a best fit with the uniaxial and monotonic experimental results of L-PBF AlSi10Mg. As you can see the attached tensile calibration figure, the slope of region 1 was controlled by elastic modulus constants (C11,C12,C44). The stress value of region 2 was controlled by Initial Slip Resistance (Sa). Next, region 3 and region 4 were controlled and calibrated using kinematic hardening parameters; (h₁,r₁,m_1, h₂,r₂,m₂), where the stress value for region 3 was controlled by b1=h1/r1 and the higher slope of region 3 was controlled using larger values of h₁ and r₁. Similarly, the stress value of region 4 was controlled by b2=h2/r2 and the lower slope of region 4 compared to region 3 was because of using smaller values of h₂ and r₂. 

You can also refer to the below article for more information about the model that we used.
https://www.sciencedirect.com/science/article/pii/S0142112325001239

Best,

Kamin Tahmasbi

[Itsuki Fujita]

Hi PRISMS Team,

I am currently working on identifying material parameters for the Ohno–Wang (OW) model to develop a CPFE model for mild steel.
However, I noticed that changing parameters such as r and h does not significantly affect the shape of the stress-strain curve.
To verify this behavior, I conducted a simulation using the same material parameters as those reported in the paper shared by Kamin, to see if I could reproduce the same hysteresis loop.

Specifically, I used the BaseLine simulation files from MaterialCommons as the base, and modified the BCinfoTable to apply tension–compression cycles with a strain amplitude of 0.003, running for three cycles.

Unfortunately, I was not able to reproduce the same hysteresis loop shown in the paper.
At this point, I am unsure whether the issue lies in the simulation environment, the prm.prm settings, or elsewhere, and I would greatly appreciate your advice.

Here is a summary of what I have done:

• Material model: Compiled the three .cc files stored in src of prisms-fatigue

• Material parameters: Exactly the same as reported in the paper

• Boundary condition: TabularBCs for 3-cycle tension–compression loading

I have attached the hysteresis loop from the paper and the loop obtained from my simulation for comparison.

Any guidance would be sincerely appreciated.

Best regards,
Itsuki

2025年4月17日木曜日 23:26:17 UTC+9 Kamin Tahmasbi:

[Chaitali Patil]

HI Itsuki, 

Probably Kamin can have better insights for the differences in the results, I would suggest you download the whole folder from the materials commons  (for the specific condition you are interested in)  and run the same setup, and compare. 

-Regards, 

 Chaitali

--
You received this message because you are subscribed to the Google Groups "Prisms-CPFE-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to prisms-cpfe-us...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/prisms-cpfe-users/41771c60-dc1f-4578-b412-72fbae20f431n%40googlegroups.com.

[Itsuki Fujita]

Hi Chaitali,

You're right — I’ll try running the simulation exactly as provided on MaterialCommons and share the results afterward.

However, my PC cannot handle a mesh with 6 million elements, so I’ll reduce the values for number of mesh subdivisions and number of voxels slightly.

As long as I don’t reduce them too drastically, the overall triaxial response should remain roughly the same, correct?

Best regards,
Itsuki

2025年5月30日金曜日 22:57:22 UTC+9 Chaitali Patil:

[Itsuki Fujita]

Hi PRISMS Team,

I ran the baseline-strain0.002 simulation file uploaded to MaterialCommons, with only the mesh size scaled down to fit my computational resources.

However, the resulting stress level was significantly lower than that shown in the paper.
Has anyone else observed this behavior or might know the reason for it?

Additionally, when I visualized the model in ParaView, the meshgrainID field appeared disordered or inconsistent. Is this expected behavior, or could it indicate a setup issue?

Any advice would be greatly appreciated.

Best regards,
Itsuki

2025年6月2日月曜日 16:51:22 UTC+9 Itsuki Fujita:

[Kamin Tahmasbi]

Hi Itsuki,

It seems the main reason of the different result is the Dream3D generated microstructure that you are using since it's not normal as you shown above. The hysteresis loop result in my paper was generated using GrainID and grain orientation files of Dream3D microstructure with 120x120x120 number of voxels. Reducing only the values for number of mesh and number of voxels without changing the grainID files would affect the result. If your PC cannot run simulations for 120x120x120 number of elements, you need to generate or use microstructure with much lower voxels (e.g. 32x32x32) to get a closer result. 

I could find the simulation files for my 32x32x32 voxels microstructure which was used in my initial calibration. If you run this attached folder, you should get a similar result to what I showed in the paper for 0.003 strain hysteresis loop. 

Best,

Kamin

[Itsuki Fujita]

Hi Kamin,

I ran the simulation using the files you shared, and I was able to reproduce the same output shown in your paper — thank you so much!
It turns out the issue was due to changing the mesh size without adjusting the grainID accordingly.

Thanks to this, I now feel confident in my simulation environment and ready to move forward.
Next, I plan to work on curve fitting for mild steel.
While I continue with the preprocessing of EBSD data, I’ll start estimating material parameters by modifying the parameter values in the simulation file you provided.

I have a quick question:
When running a uniaxial tension test simulation, do I need to switch to SimpleBCs, or can I simply adjust the loading conditions in TabularBCs to apply a single tensile cycle?
I feel like either should work, but I’d appreciate your thoughts on the best practice.

Thanks again, and best regards,
Itsuki

2025年6月4日水曜日 10:11:44 UTC+9 Kamin Tahmasbi:

[Kamin Tahmasbi]

Hi Itsuki,

You're welcome. Yes, I think it's better to use SimpleBCs for uniaxial tensile test calibration. I have attached my simpleTension simulation files to this email which was used for 32x32x32 voxelized microstructure in my initial calibration. Be careful to use slipDirections, slipNormals and LatentHardeningRatio files of BCC structure for the mild steel that you want to calibrate, as my files are for Aluminum alloys with FCC structure.

Best of luck,

Kamin

[Prasanna Tatavarty]

Hello Kamin and Team,

      I went through your paper explaining the kinematic hardening. I'm working with aluminium alloy and have calibrated the rate-independent parameters (slip resistance, initial hardening, modulus, saturation stress, power law exponent) using the tensile test results based on the method given in this paper: https://www.sciencedirect.com/science/article/pii/S074964192300181X 

Now I want to model the LCF behaviour of my material. I tried using the cyclic loading module (in the FCC folder of PRISMS plasticity), but it has two extra parameters - kinematic slip parameters. I'm unable to calibrate for them or find any equation related to them.

In your paper, you have used a rate-dependent model with backstress terms. In the image you attached (prm file), the slip resistance and saturation stress values are the same, and the initial hardening modulus is zero. Why is it so? If I want to use this model for my material, how can I calculate the parameters (h1, r1, etc.)?

Also, can I model the tensile behaviour with a rate-independent model and LCF behaviour with a rate-dependent (backstress) model? Any suggestions in this regard will be greatly helpful.

Regards

Prasanna Tatavarty 

[Prasanna Tatavarty]

Hello Kamin and Team,

      I went through your paper explaining the kinematic hardening. I'm working with aluminium alloy and have calibrated the rate-independent parameters (slip resistance, initial hardening, modulus, saturation stress, power law exponent) using the tensile test results based on the method given in this paper: https://www.sciencedirect.com/science/article/pii/S074964192300181X 

Now I want to model the LCF behaviour of my material. I tried using the cyclic loading module (in the FCC folder of PRISMS plasticity), but it has two extra parameters - kinematic slip parameters. I'm unable to calibrate for them or find any equation related to them.

In your paper, you have used a rate-dependent model with backstress terms. In the image you attached (prm file), the slip resistance and saturation stress values are the same, and the initial hardening modulus is zero. Why is it so? If I want to use this model for my material, how can I calculate the parameters (h1, r1, etc.)?

Also, can I model the tensile behaviour with a rate-independent model and LCF behaviour with a rate-dependent (backstress) model? Any suggestions in this regard will be greatly helpful.

Regards

Prasanna Tatavarty