Thread Rating:
  • 0 Vote(s) - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
CQUAD4 in bending test
#21
Just another way of putting it to see the issue:

Case1: E1=E2
If E1=E2=9.0E9, T3 at Node 22 is 7.03E-4 (classical is 7.11E-4)...so this is OK.

Case 2: Change E1 by half and hold E2
If E1=4.5E9 and E2=9.0E9, T3 at Node 22 is 9.08E-4

Case 3: Change E2 by half and hold E1
If E1=9.0E9 and E2=4.5E9, T3 at Node 22 is 9.37E-4

For Cases 2 and 3, I would have expected one of those to have a deflection of 2*7.03E-4 and the other to have a deflection fo 1*7.03E4. But the results are odd instead.

https://www.mystran.com/other/MAT8_issue.dat
Reply
#22
I believe Bill has identified the issue. He will report on how it will be addressed here.
Reply
#23
O_Stodieck, I have found the problem with the QUAD4 with orthotropic materials. The default QUAD4 in MYSTRAN is actually made up of 4 non-overlapping TRIA3 elements and these triangular elements are not coded to rotate their orthotropic material axes to align with the material axes specified (or implied) on the basic quad shape. I will have to correct this error but, in the meantime, try using:
 
1) Put PARAM, QUAD4TYP, MIN4 into your current bulk data file. The MIN4 version of QUAD4 is an actual quad, with transverse shear flexibility. It does have some issues with spurious modes when the shape is not rectangular. For your problem it is fine. I get agreement with bean theory to 4 significant figures (beam theory deflection = 0.2962, see attached image)

I will put a warning message in MYSTRAN and a stop to let user's know of this problem until it gets fixed


Attached Files Thumbnail(s)
   
Reply
#24
(09-09-2020, 05:35 PM)drbillc Wrote: O_Stodieck, I have found the problem with the QUAD4 with orthotropic materials. The default QUAD4 in MYSTRAN is actually made up of 4 non-overlapping TRIA3 elements and these triangular elements are not coded to rotate their orthotropic material axes to align with the material axes specified (or implied) on the basic quad shape. I will have to correct this error but, in the meantime, try using:
 
1) Put PARAM, QUAD4TYP, MIN4 into your current bulk data file. The MIN4 version of QUAD4 is an actual quad, with transverse shear flexibility. It does have some issues with spurious modes when the shape is not rectangular. For your problem it is fine. I get agreement with bean theory to 4 significant figures (beam theory deflection = 0.2962, see attached image)

I will put a warning message in MYSTRAN and a stop to let user's know of this problem until it gets fixed

Thanks drbillc, I will try that for now.
Reply
#25
The next step would be to make a test file with a PCOMP. I would first start with just a single layer and make sure a change of orientation is correct. Then I would try a multi-directional laminate.

I think the other consideration is the orientation of the element itself (local orientation vs an orientation specified with w.r.t. a coord system). This is important for general problems since for a general quad/tri mesh has local system in many directions (but the properties are usually desired to be w.r.t. a coord system (not local system in that case).
Reply
#26
Would you come up with a good test problem that has those features? I have a PCOMP test problem which gave answers matching those in the comparison solution. Still working on why that happened
Reply
#27
I can look into it further.

Btw, here is the paper for the formulation of the CQUAD4 in MYSTRAN

https://www.researchgate.net/publication...al_element
Reply
#28
For the recent version (11.2), there is now an error when the CQUAD4 and orthotropic properties are used, as follows.

*ERROR 1999: Current code for QUAD4 element 1 is incomplete for Orthotropic material properties.
Try a PARAM, QUAD4TYP, MIN4 for an alternate QUAD4 if your elements are nearly rectangular.
Otherwise you will have to use CTRIA3 elements
You can override this and continue with a bdf entry: DEBUG, 248, 1
Reply


Forum Jump:


Users browsing this thread: 2 Guest(s)