######## WORK NOTES COPIED FROM INCIDENT 599894 ########
**** Entered By: TS319341 @ 03/09/2007 07:56 AM ****
QUESTION:
Getting different results for plane13 and plane223
--------------------------------------------------------------
I created a simple structural-thermal static coupled analysis. In the first example a plane 13 is used, in the second one a plane 223. These are identical 2D axisymmetric models of a disk spinning at 200 revolutions per second and at 1000 C above the reference temperature. Material properties are orthotropic but linear. Using orthotropic properties, the results differ substantially (the PLANE223 model is correct). If materials are isotropic, there is no difference in results (both models are correct).
Further testing:
testOrtho_2D_af_13.txt: PLANE13 model
testOrtho_2D_af_223a.txt: PLANE223 model, matrix coupled thermal and structural matrices
testOrtho_2D_af_223b.txt: PLANE223 model, load vector coupled thermal and structural matrices
testOrtho_2D_af_223c.txt: PLANE223 model, load vector coupled thermal and structural matrices with NSUB,2 to force more than one iteration for (hopefully) proper coupling.
As you can see from the attached plots, only the direct coupled method ('223a' files) appears to give sensible results (we expect tensile radial stresses about a third of the way from the center of the disk, NOT negative ones as determined by all the load vector coupled models).
ANSWER:
The issue seems to be related to a misinterpretation of the TUNIF functionality. According to the TUNIF command description, its behavior varies depending on whether the analysis is simultaneous or iterative:
"Since TUNIF (or BFUNIF,TEMP) is step-applied in the first iteration, you should use BF, ALL, TEMP, Value to ramp on a uniform temperature load."
In other words, when the analysis is simultaneous (matrix coupling between U and TEMP using keyopt(2)=0 with PLANE223), the temperature nodal load is zero (TUNIF ramped) and
|
######## WORK NOTES COPIED FROM INCIDENT 599894 ######## **** Entered By: TS319341 @ 03/09/2007 07:56 AM **** QUESTION: Getting different results for plane13 and plane223 -------------------------------------------------------------- I created a simple structural-thermal static coupled analysis. In the first example a plane 13 is used, in the second one a plane 223. These are identical 2D axisymmetric models of a disk spinning at 200 revolutions per second and at 1000 C above the reference temperature. Material properties are orthotropic but linear. Using orthotropic properties, the results differ substantially (the PLANE223 model is correct). If materials are isotropic, there is no difference in results (both models are correct). Further testing: testOrtho_2D_af_13.txt: PLANE13 model testOrtho_2D_af_223a.txt: PLANE223 model, matrix coupled thermal and structural matrices testOrtho_2D_af_223b.txt: PLANE223 model, load vector coupled thermal and structural matrices testOrtho_2D_af_223c.txt: PLANE223 model, load vector coupled thermal and structural matrices with NSUB,2 to force more than one iteration for (hopefully) proper coupling. As you can see from the attached plots, only the direct coupled method (`223a` files) appears to give sensible results (we expect tensile radial stresses about a third of the way from the center of the disk, NOT negative ones as determined by all the load vector coupled models). ANSWER: The issue seems to be related to a misinterpretation of the TUNIF functionality. According to the TUNIF command description, its behavior varies depending on whether the analysis is simultaneous or iterative: "Since TUNIF (or BFUNIF,TEMP) is step-applied in the first iteration, you should use BF, ALL, TEMP, Value to ramp on a uniform temperature load." In other words, when the analysis is simultaneous (matrix coupling between U and TEMP using keyopt(2)=0 with PLANE223), the temperature nodal load is zero (TUNIF ramped) andthe thermal strain EPTH is zero (since TREF=0). These are the results the user wants to see. However, when an iterative analysis is activated (load vector coupling using keyopt(2)=1 with PLANE223 or PLANE13), the temperature nodal load is 1000 (TUNIF stepped) and thermal strain EPTH = ALPH*1000 is non-zero . To achieve the desired results with a load vector coupling, the user must either use the bf,all,temp,1000 command instead of tunif,1000 or issue tunif,1000 together with tref,1000. I have attached an additional input file (testOrtho_2D_af_223d.txt) which uses load vector coupling, replacing TUNIF with BF,ALL,TEMP,100. The results are now the same as they were for input file testOrtho_2D_af_223a.txt (which is to say they are now correct). |
||
|