Different result from Texgen for the same input properties

[Hemanth_T_N]

Hi Everybody..!

I generated an FE Mesh of full plain weave textile RVE from Texgen. The input parameters of the yarn were as (elastic Modulii: Ex=138 GPa, Ey=9 GPa , Ez=9 GPa, Shear Moduli: Gxy=Gxz=Gyz 6.9 GPa and poisson's ratio of 0.3 (nuxy=Nuxz=Nuyz=0.3). The matrix was isotropic with E=3.5 GPa and Nu=0.35. I exported the model as an abaqus voxel file.

I imported the model into abaqus CAE and deleted the predefined steps, Boundary conditions & constraint equations. I specified the boundary conditions such that the RVE is pulled along one direction (say-x) and constrained along other directions (nodes on X faces are pulled along x direction while the nodes on y faces and z faces are constrained in y and z directions respectively). No MPCs were used in this analysis. I then submit the job to abaqus solver to get the microstresses acting on each element.

The macrostresses can be calculated as per formula:

Macrostress=sum of (elemental microstress * volume of corresponding element volume)/total volume of RVE.

The macrostress so obtained are divided by the applied strain to get first column of effective stiffness matrix. Similarly the second column and third column of stiffness matrix can be computed.
However the stiffness obtained from this procedure is different from Texgen( outlined in http://texgen.sourceforge.net/index.php ... and_Abaqus) The results and the comparison is as below:

Values are in Gpa
The final stiffness matrix without MPC The C matrix from Texgen
26.8877 3.4592 3.6674 24.4862 5.8594 3.6679
3.4592 26.8877 3.6674 5.8594 24.4862 3.6679
3.6674 3.6674 6.6833 3.6679 3.6679 6.6838


Error (assuming Texgen as correct)
9.81% -40.96% -0.01%
-40.96% 9.81% -0.01%
-0.01% -0.01% -0.01%

As can be seen, I have good comparison along the third column but not so much along the first and second column. The difference in C12 component (-40.96%) is particularly worrisome.
can someone tell me why I am getting different result and as to what would have been wrong in my analysis.

I would be grateful your inputs/advice/comment.

Have a great weekend...!

Thanks and Regards,
Hemanth T N

[Hemanth_T_N]

Hey all,

The stiffness matrices look messed up:
stiffness matrix without MPC:
26.8877 3.4592 3.6674
3.4592 26.8877 3.6674
3.6674 3.6674 6.6833

stiffness from Texgen (python scripts)
24.4862 5.8594 3.6679
5.8594 24.4862 3.6679
3.6679 3.6679 6.6838

difference w.r.t Texgen results:
9.81% -40.96% -0.01%
-40.96% 9.81% -0.01%
-0.01% -0.01% -0.01%

Sorry for the duplicity of messages.


Thanks and Regards,
Hemanth T N

[louisepb]

Hi Hemanth,

Is the difference due to the boundary conditions? The TexGen approach implements periodic boundary conditions, assuming that displacements on opposite faces are related to each other. They are described in full in Shuguang Li's paper here: https://www.sciencedirect.com/science/a ... 3603000620

I'm not an FE expert so am maybe not the best person to comment on the method that you've used.

Sorry not to be more help,
Louise

[Hemanth_T_N]

Hi Dr. Louise,

Thankyou for your reply. I am getting the same result as Texgen When I implement the periodic boundary conditions.

Regards,
Hemanth T N