User login

Navigation

You are here

A Better Alternative to RVE Analysis

Wenbin Yu's picture

RVE analysis is popular for computational homogenization. It can be used independently for virtual testing or as a module for multiscale modeling. Its popularity is mainly due to the maturity and acceptance of commercial finite element software. RVE analysis usually requires a 3D domain to obtain 3D properties and local fields. If a 2D RVE is used, only 2D properties and local fields are obtained. To obtain the complete set of properties, multiple analysis is needed. For example, to obtain the complete stiffness matrix, six 3D RVE analyses are needed. The main drawbacks are the computational cost, and difficulty in applying the right boundary conditions. 

The recently discovered mechanics of structure genome (MSG) and its companion code SwiftComp, when specialized to 3D structures, can provide a general-purpose micromechanics theory. Many examples, including the micromechanics simulation challenge have been used to demonstrate that MSG/SwiftComp is more versatile, efficient, and simpler than RVE analysis with out lossing any accuracy and geometric modeling flexibility.  More detailed comparison between MSG and RVE analysis can be found hereSwiftComp can be freely launched in the cloud at https://cdmhub.org/resources/scstandard. In other words, one can run a super-efficient "RVE analysis" on any devices including smart phones and tablets connected to Internet via a browser. Various GUIs are available for users to choose from including Gmsh, TexGen, ANSYS, and ABAQUS4, all of which are free available on cdmHUB  

Comments

Dear Prof. Yu,

Thank you for a very nice post and the comparison between RVE and MSG. I have read your blog and in that blog you have mentioned that "periodic BCs are the best BCs to be used for RVE analysis unless the one or more dimensions of the RVE are the same as the macro structural dimension such as laminate made of one woven fabric through the thickness"

Could you please provide me some articles that support this statement? At the moment, I am doing homogenization for plain woven fabrics and I have applied the PBCs on the meso scale model only in the in-plane directions. In macro scale, I use the membrane element instead of shell, because the bending stiffness in the considered material is negligible. I wonder if my approach is correct. Would you please kindly let me know your opinion?

Thanks in advance.

Kind regards,

Ken

Wenbin Yu's picture

Dear Ken, thanks a lot for your interest. If you only apply in-plane PBC to a woven fabric, which is done by most studies, you can only obtain in-plane properties and the homogeneous body is in a plane-stress state, which is ok if you structure is very thin. If you want the complete set of 3D properties and 3D local fields, then a mix boundary conditions should be applied. Please refer to my recent presentation on MSG, the last example. You might also be interested in trying our code at  https://cdmhub.org/resources/texgen4sc. We integrated TexGen with SwiftComp allowing others to compute the complete set of properties for any woven fabrics with one click.  

Subscribe to Comments for "A Better Alternative to RVE Analysis"

Recent comments

More comments

Syndicate

Subscribe to Syndicate