Hi,

I am getting some unexpected results running the attached cantilevered composite plate model in Mystran 11.0. :

- The model is a flat plate in the X-Y plane, with dimensions of 1.5m x 0.15m and fully fixed at one of the short ends.

- A Z-direction shear force is applied through an RBE3 at the unsupported end (Grid 737).

- I would expect a Z-deflection at Grid 737 close to T3 = 0.29625 (simple CLT calculation and in agreement with other FEM solvers)

- But I am getting T3 = 0.03613559 at Grid 737 instead.

- AUTO SPC forces are zero (constraining Z-rotations / drilling DOF on the CQUADs only)

Any ideas what could trigger this behaviour?

Do you have a simple test case for shell (CQUAD4) element bending that I can run to test the installation?

Thanks.

Can you check the reaction forces to make sure you are applying the correct load? There is a 0.1 scaling factor in the FORCE card so you may want to double check that. Have you tried an isotropic material to see if the results are correct for that first? I will let Bill know as well.

I'll take a look at it and get back to you

(08-27-2020, 06:26 PM)Admin Wrote: [ -> ]Can you check the reaction forces to make sure you are applying the correct load? There is a 0.1 scaling factor in the FORCE card so you may want to double check that. Have you tried an isotropic material to see if the results are correct for that first? I will let Bill know as well.

Reaction forces are correct - factor of 0.1 is intended.

The element node numbering is such that the vectors from grids 1 to 2 point in the x-direction for all elements - this is matched in the comparison models - so I don't think this is an input error? Besides, the results are not any closer if the plies are set to 90deg in the PCOMP cards.

As you suggested, I reran the model using an isotropic material - results are much closer (<2% error).

Comparing Zmax in the T3 deflection at grid 737:

Analytical (Bernoulli beam):

I = (1.25e-04*12)**3*0.15/12 = 4.21875e-11

F = 0.1

L = 1.5

E = 70.0E9

Zmax = F*L**3/(3*E*I) = 0.038095

Mystran (see file attached)

Zmax = 3.762719E-02

I just did a quick calculation treating the plate as a beam (should be good assumption since the aspect ratio is 20 to 1) and get a displacement at the end that is within 7% of the MYSTRAN plate problem result. I did run the problem in MYSTRAN and get exactly what you got, the displacement of 0.03613558. See the attached jpg which shows the beam displacement within 7% of the plate solution. Can you tell me what other sources gave you such a different answer?

O_Stodieck, can you show your hand calculations for the original problem as well? The composite version.

Sorry, I used the wrong width (.075 rather than .15). But where did the 70E+09 for modulus come from?

(08-28-2020, 12:05 AM)drbillc Wrote: [ -> ]Sorry, I used the wrong width (.075 rather than .15). But where did the 70E+09 for modulus come from?

Bill, there are two decks in the different posts and it was changed between them. Its a bit confusing, but maybe it was an attempt to try to make the composite more "isotropic-like". Not sure.

O_Stodieck, I was looking for a good QUAD4 example to send you and, in the process, uncovered an error in MYSTRAN that crept in between versions 10.1 and 11. It is a plate problem with a pressure load that has a theoretical solution so that you can see how well MYSTRAN gets that answer. I will fix the error and then send you the bdf, but in the meantime MYSTRAN may crash with a shell pressure load.

(08-28-2020, 12:02 AM)Admin Wrote: [ -> ]O_Stodieck, can you show your hand calculations for the original problem as well? The composite version.

Sure - I have posted a python implementation of the CLT approach here

https://gist.github.com/ostodieck/d0411f...2db4ccf512
I hope this is self-explanatory enough - let me know if something is unclear.

The laminate and materials are the same as in the first post, and the tip deflection is

0.29625 m as expected.
Thanks for checking this!