I ran a test model and found that EPTO,EQV < EPPL,EQV, which seems counter intuitive, since, from equation 19-170 in the theory manual:
EPTOeqv = EPELeqv + EPPLeqv + EPCReqv
Is there something fundamental about the way these results quantities are related that I am missing? Is it possible for EPPL,EQV > EPTO,EQV? Does the choice of EFFNU in the AVPRIN command affect the reported values?
|
The short answer to this question is yes. But the explanation for why this is so is quite involved. Please see the "Appendix" below, and in particular, the findings summarized in the presentation attached to this solution. The attachment also includes a model and some macros which I used in my investigation. APPENDICES: A1: Initial findings/sources of confusion the user may see: ANSYS is getting only one set of strains ( 6 components)/ element from LSDYNA and stores these in the slots used for EPPL. This value is also assigned to all the nodes of that element (for PRNSOL). Then when EPTO is requested ANSYS adds up the EPPL,EPCR,EPEL etc and gives the total strain. In the case of LSDYNA only EPPL exists. Hence EPPL should be the same as EPTO. We are not sure that the strains from LSDYNA are plastic strains only. Hence we do not want users to look at EPPL. I have requested the developer to block out the EPPL strains since they may or may not be correct. I also found mention in the knowledge base of the fact that AVPRIN,,EFFNU affects reported equivalent strains, which I found to be the case with my test model. When I set EFFNU = 0.5, I created contour plots using PLNSOL, and found that: EPTO,EQV = EPPL,EQV They are also equal to each other in contour plots when I set EFFNU = 0 (though the values are very different from those when EFFNU = 0.5). But when I issue AVPRIN (with NO arguments), contour plots show: EPTO,EQV < EPPL,EQV Which I believe is wrong. Also, I found that AVPRIN,,EFFNU affects the value returned by *get,parm,node,node number,EPTO,EQV But it does NOT affect the value returned by: *get,parm,node,node number,EPPL,EQV I also found that in the variable viewer in POST26, EPPL,EQV is available in the GUI but NOT EPTO,EQV. I also found that in POST26, AVPRIN,,EFFNU does not affect EPPL,EQV. I saw a few articles advocating that we deny user access to EPPL,EQV but what seems to have happened is that we instead denied user access to EPTO,EQV (at least in POST26). I apologize for all the confusion. I will run this past my coworkers for further comment. You have indeed found a very troubling (or at least confusing!) issue. I truly hope that some development work can be done to fix this problem, but I cannot guarantee this will happen. Thank you for bringing this to our attention. I will try to learn more about this problem and/or get something done to fix it. Kind regards, Bill A2: Query I put to development: **** Entered By: bbulat @ 06/09/2005 05:28 PM **** Hi Tongru, I`ve spent quite a lot of time trying to figure out (unsuccessfully) what ANSYS is doing with the data LS-Dyna puts out for SHELL163. When you get a chance, please inspect the attached presentation and model. The original user question was `is it possible for EPPL,EQV > EPTO,EQV, and if so, under what circumstances does this occur`? For now, I`ll buy off on what you said in KD 12738, However, in attempting to answer this question, I found inconsistencies in data retrieved for these elements that I have so far been unable to resolve. It is these inconsistencies that I`m reporting to you now for comment. I can see how this would really confuse users. Thanks, -Bill A3: Development's response to my query: Bill, After digging in many old e-mails and Ansys source codes for two days, now I can give you an answer. First, let me explain what kinds of strains lsdyna writes to the Ansys result file. Lsdyna always writes the total strain components (x,y,z,xy,yz,zx). For the elastic-plastic material models, it also writes the elastic strain components (x,y,z,xy,yz,zx) and only the equivalent plastic strain (NO plastic strain Components). When the user requests the total strains (x,y,z,xy,yz,zx,prin,eqv,etc), it reads the rst file and follows the regular Ansys strain conventions. For example, eqv is Calculated from eq (19-165). But if the user requests the plastic strain, Ansys only gives equivalent plastic strain stored in the rst file. For the other plastic strains (except eqv), Ansys blocks the requests (because rst file doesn`t have them). Based on the above, the answer to your question is the following. `Is it possible for EPPL,EQV > EPTO,EQV, and if so, under what circumstances does this occur` YES. `EPPL,EQV` comes from lsdyna calculation. It is more like (maybe same) the accumulated plastic strain (ref eq (19-168)). `EPTO,EQV` comes from the calculation based on eq. (19-165). EPPL,EQV > EPTO,EQV happens when unloading occurs. In cyclic loading, the former can be very larger. The latter can be very small or zero. If no unloading occurs, EPPL,EQV < EPTO,EQV. ` In POST1: ` Data retrieved in POST1 for EPPL,EQV and EPTO,EQV using *GET were unequal. EPPL,EQV data retrieved with *GET was not affected by EFFNU, whereas EPTO,EQV was affected. The reason is explained above. `EPPL,EQV` comes from lsdyna directly, while `EPTO,EQV` comes from eq (19-168). ` Plots (PLESOL) of EPPL,EQV and EPTO,EQV were identical I think they should be different. Probably the difference is hard to see and because for many elements `EPPL,EQV` and `EPTO,EQV` are close. ` In POST26: ` ANSOL and ESOL results were identical to each other and were unaffected by SHELLP26 command (this model used NIP=5 defined through the real constant set and EDINT,5). Resul |
||
|