Questions and Answers

All Questions

0 Like 0 Dislike

Sadik Omairey

Tet and Hex element type results variation

Dear all,

I'm trying to use ABAQUS-Swiftcomp GUI to obtain homogenised RVE properties. It is my first use, and I'm happy with the ABAQUS interface and processing time, in addtion, matrix construction.

I used RVE method with periodic boundary conditions, Chamis equations (rule of mixture as a ref. point) will give relative results. On the other hand, with Swiftcomp, the results are:

  • Consistent with the above methods for E1, but significantly different for E2 and E3 (when using Hex elements);
  • Consistent in E2 and E3, but 50% less for E1 (when using Tet elements).

Kindly note that I did not use and special type elements. My RVE is set to be periodic, 3D.

Any advice would be great.

Thank you in advance…

Best regards

  1. Finite Element Analysis
  2. Homogenization
  3. RVE

Report abuse

2 Responses

  1. 0 Like 0 Dislike

    Wenbin Yu

    Please ask your questions related with SwiftComp at https://cdmhub.org/groups/yugroup/forum to get quicker response. 

     

     

    Cancel Report abuse

    Please login to answer the question.

  2. 0 Like 0 Dislike

    Bo Peng

    Hi,

     Thanks for your interest in SwiftComp and Abaqus SwiftComp GUI. With the provided information I cannot give you a definite answer. But my guess is that different mesh technologies result in that the periodicity of the nodes at the boundary surfaces change and thus different constraints are applied to the models with different mesh. 

    In another word, using Hex mesh for this geometry, most nodes at negative X surface can find their paired nodes at the positive X direction, but nodes at negative Y and Z surfaces cannot find their paired nodes. Therefore SwiftComp will apply weaker constraints (close to uniform traction constraint in the Y and Z directions but not exactly) compared with Periodic constraints and homogenize the properties related with Y and Z,  which result in the difference of E2 and E3. On the contrary, using Tet mesh, the opposite happened.

    You can check the *.echo file in your work directory and search for 'aperiodic' nodes. It will list all the nodes (and the directions) which cannot find a paired node at the corresponding opposite surfaces. To make the mesh periodic in each direction, you can apply some tricks in Abaqus.

    If this is not the reason, please send me the *.cae file and *.jnl file of your model. Thanks.

    Reply Report abuse

    Please login to answer the question.

Answer this question