Hi,
I'm self-developing explicit 3-D FEM code using GPGPU for my research regarding Penetration of projectile into rock. In the code, I'm incorporating JH-2(Johnson holmquist) constitutive model.
The question is how can I efficiently calculate effective plastic strain rate in the context of Explicit FEM like LS-DYNA and Abacus Explicit?( I don't know what kind of algorithm is implemented in the commercial software because it is black box to me.)
For this, I reviewed several literature but I cannot find the detailed description regarding how to calculate the effective plastic strain rate. For example, I found the flowchart of implementation of JH2 from Table 1 in
http://handle.dtic.mil/100.2/ADA402347
But the table only says "calculate effective plastic strain rate".
I did my verification simulation based on
http://www.dynalook.com/european-conf-2003/implementation-and-validation-of-the-johnson.pdf
and successfully passed the Case A. But I failed in the Case B in which evaluation of plastic strain rate is the key.
Any help would be sincerely appreciated.
Thank you in advance.
Best regards,
Daisuke Fukuda
Hi Daisuke,
Did you write subroutine to achieve your research? Maybe you can find some clues in the documentation of Abaqus.
Xiangxiong
Dear Xiangxiong
Thank you for your valuable comment.
Yes I am writing my code as can be found in the attached file "JH2.f90(fortran code)".
This is just a CPU version of my code (I will convert this to GPU version after I find this code can work!)
Following your advice, I found the documentation of Abaqus in http://abqdoc.byv.kth.se:2080/v6.11/books/exa/default.htm?startat=ch02s01aex77.html
but this page also treats the plastic strain increment as given.(But how..)
ln my code, I modeled all the plastic deformations by associative flow rule, which seems to be not working because my implementation of the model gave too rapid increase of damage, meaning that plastic strain increment is different from the one in Johnson and Holmquist's original paper. Or I think I have to reconsider the definition of Damage evolution law, i,e, strain linked to the (change of yield stress due to) damage increase. But the question is how???
Actually same problem has been reported in https://www.diva-portal.org/smash/get/diva2:613844/FULLTEXT01.pdf where the author used Autodyn. Following this author's approach, my simulation shows the closer result as in the Johnson and Holmquist's original paper. But still some modification seems to be necessary...
Thank you very much
Daisuke
Dear Aziz
Thank you for your comment.
Yes, as you mentioned, calibration is always necessary step.
However, calibration has a meaning once you "ensure" that the algorithm itself has no problem. My question posed here is for the stage before the calibration, i.e. the formulation and algorithm adopted here has, first of all, OK or not, especially for my formulation of plastic strain calculation and damage increment. Currently my friend is kindly trying to do the same calculation based on same geometry of the model as me using Auto-dyn(I am not interested in commercial software, which cannot analyze dynamic crack propagation as in http://rock.eng.hokudai.ac.jp/daisuke/index.html). Anyway if Auto-dyn shows same results as me, I can proceed. Otherwise, I come to deadlock..
Thank you for your concern.
Daisuke
@ Arijit: No. By same problem, you mean you passed test A but couldn't for tests B&C? Using https://www.diva-portal.org/smash/get/diva2:613844/FULLTEXT01.pdf , the difference between the original JH2 paper and my code became closer but still non-negligible difference. Maybe to compute plastic strain, I should have used Newton-Rhaphson iterative method to compute it in a implicit way....
Hello there,
I am using SPH approach to solve the same test cases but I still did not get any results. I followed your code and I rewrite into LAMMPS but as I said, there is no result yet.
I am wondering if I can discuss with you the steps or if you have a flow chart to the final code of JH-2.
Thank you so much for your time.
What do you mean by "there is no result"??? Literally no result?
My code would pass cases(tests) 1~2 while not for 3. So, I think the treatment of plastic strain rate should be carefully checked. See eqs. 45-51 in
https://apps.dtic.mil/dtic/tr/fulltext/u2/a327659.pdf
Regards,
- Badges
- Science topic