I have Installed Abaqus swift comp Gui. After running the example which is available for 2D SG in Abaqus-Swiftcomp manual I got same homogenized properties as in manual but the dehomogenized result value of SN11 is not as same as in manual.
I have taken same properties of fiber and matrix as per the manual and homogenized result also same but there is difference in SN11 results after dehomogenization and I followed same way of manual for dehomogenization also.
I am attaching screen shots of manual process and my followed process with homogenized and dehomogenized results.
And there is one more doubt in manual why we put the value 0.0009 for e11 and e12, is these value came from macro model analysis or assumed.
The problem should be in the manual. The figure may refer to some other example with different global strains. You can keep working with the interface.
For Dehomogenization there are different values to enter like rotation displacement, strain. For this all value has needed or only one strain or one displacement value can dehomogenize if only one value can dehomogenize then what value will be good to choose for better results like displacement, rotation or strain. If one is choosing displacement and one is choosing strain and one is choosing two value of strain(these two also can be different because there are 6 values of strain) and one is choosing two value of displacement then what is the good option to get answer.
One more doubt to clear (I have read from Liu, X.; Rouf, K.; Peng, B.; and Yu, W and Rouf, K.; Liu, X.; and Yu, W) Macro model can be a normal block like in case of woven fabric if original dimension of model is 100*100*2 mm and SG(unit block) is 2*2*1 then for macro model analysis we have to just make normal 100*100*2 mm rectangular 3D block (in any FEA tool i.e. ANSYS ABAQUS) with homogenized properties from woven SG block(from swiftcomp texgen), is it correct.
You need displacements and rotations from macroscale analysis to get the local displacement fields, and you need averaged strain fields (6 components) from macroscale analysis to get local strain fields. You can read Dr. Yu's paper "Yu, W., 2016. A unified theory for constitutive modeling of composites. Journal of Mechanics of Materials and Structures, 11(4), pp.379-411." for more details especially the Eq. (48) and Eq. (50).
The first step to perform MSG-based multiscale analysis is to identify the SG according to the macroscale structure and the MSG models. For example, in your example the macro structure has dimensions 100*100*2, I would use MSG plate model and the corresponding SG should be 2*2*2 if you have 50 SGs repeated in the x and y direction respectively, and the macro model will be a 2D plane using shell elements instead of a block. A good way to help you understand the MSG theory is to try to reproduce the results in the papers you mentioned.
Vivek Dhimole @ on — Edited @ on
Dear sir,
I have Installed Abaqus swift comp Gui. After running the example which is available for 2D SG in Abaqus-Swiftcomp manual I got same homogenized properties as in manual but the dehomogenized result value of SN11 is not as same as in manual.
I have taken same properties of fiber and matrix as per the manual and homogenized result also same but there is difference in SN11 results after dehomogenization and I followed same way of manual for dehomogenization also.
I am attaching screen shots of manual process and my followed process with homogenized and dehomogenized results.
And there is one more doubt in manual why we put the value 0.0009 for e11 and e12, is these value came from macro model analysis or assumed.
Please find attachments for all.
Yours sincerely
abaqus-swiftcomp.docx
2 MBClick to download
Vivek Dhimole @ on
Hello sir,
Have you looked on that please resolve that issue?
Sincerely
Su Tian @ on
Hello Vivek,
The problem should be in the manual. The figure may refer to some other example with different global strains. You can keep working with the interface.
Vivek Dhimole @ on
Ok sir, Thank you.
For Dehomogenization there are different values to enter like rotation displacement, strain. For this all value has needed or only one strain or one displacement value can dehomogenize if only one value can dehomogenize then what value will be good to choose for better results like displacement, rotation or strain. If one is choosing displacement and one is choosing strain and one is choosing two value of strain(these two also can be different because there are 6 values of strain) and one is choosing two value of displacement then what is the good option to get answer.
One more doubt to clear (I have read from Liu, X.; Rouf, K.; Peng, B.; and Yu, W and Rouf, K.; Liu, X.; and Yu, W) Macro model can be a normal block like in case of woven fabric if original dimension of model is 100*100*2 mm and SG(unit block) is 2*2*1 then for macro model analysis we have to just make normal 100*100*2 mm rectangular 3D block (in any FEA tool i.e. ANSYS ABAQUS) with homogenized properties from woven SG block(from swiftcomp texgen), is it correct.
Xin Liu @ on
You need displacements and rotations from macroscale analysis to get the local displacement fields, and you need averaged strain fields (6 components) from macroscale analysis to get local strain fields. You can read Dr. Yu's paper "Yu, W., 2016. A unified theory for constitutive modeling of composites. Journal of Mechanics of Materials and Structures, 11(4), pp.379-411." for more details especially the Eq. (48) and Eq. (50).
The first step to perform MSG-based multiscale analysis is to identify the SG according to the macroscale structure and the MSG models. For example, in your example the macro structure has dimensions 100*100*2, I would use MSG plate model and the corresponding SG should be 2*2*2 if you have 50 SGs repeated in the x and y direction respectively, and the macro model will be a 2D plane using shell elements instead of a block. A good way to help you understand the MSG theory is to try to reproduce the results in the papers you mentioned.