thanks for your attention. I visit the page but I can't find any to my problem.the advices were general. I want to know how can I model the soil and it's boundary in ABAQUS, exactly!
I believe the answer boils down to using a matrix of dynamic stiffness functions on your boundary.
You
Use analytical functions for the 'green field'
Exctract data at the cavity of your dig and/or surface of your interface
Invert the matrix that is output from the above (one inversion per frequency)
Insert the inverted result as BC to your Abaqus model
Analyze the response using direct matrix solution
The above can be rather time consuming, so it may be worth your while to see if you can make some shortcuts, e.g. by the use of reciprocity. It all depends on your analysis purpose.
You mean that according to my earthquake direction (X,Y,Z or multicomponent), I should use sweep mesh and then check the arrow of that to be in my earthquake direction?
I may be wrong here but to the best of my knowledge, infinite ('perfectly' absorbing elements) exist only for acouustic waves (Pressure-waves), while soil involves more wave types.
Aligning the mesh to be normal to the wave direction is good avice but I believe it cannot handle more than some situations as infinite elements handle most situations for the acoustic case.
Therefore, the much more laborious procedure where you use matrices as described above is the general procedure. Than again, if a simple mesh adapations is good enough, then you should use it.
In this spirit, here (see Figure 3) is another approach to model absorbing boundaries. It is simple to make but computationally costly. http://qringtech.com/wp-content/uploads/2014/01/SPIE205933-18.pdf
There exists also the FE code calles Code Aster which uses a 'rubber band mesh' on the boundary to, as best possible, absorb impinging waves. Code Aster has implemented the coupling matrix approach to a French analytical code for layered soils. http://www.code-aster.org/V2/spip.php?rubrique2
As far as I understand, a matrix on the boundary is the general method. Physical modelling as shown above using 'absorbing wedges' handle more wave types but leave some reflection. Infinite elements for one wavetype is bettwer than rho*c boundary conditions (rho*c corresponds to diagonal matrix, while the general approach involves a full matrix). The Code Aster 'rubber band' boundary mesh relects but is easy to work with and allow introduction of impinging waves as loads.
The 'best' approach is the one that works for you. What works for you depends on what you want to do and how well you have to do it.
Code Aster contains a coupling to a code calles MISS3D which implements the matrix method. Here is PhD thesis on the subject http://www.theses.fr/2013ECAP0006/abes
If you are running a very large model, the matrix method probably is your best bet. I was invoved with a project where we used this approach and still had to use a local supercomputer for the analysis.
Feel free to drop me a mail if you want to try Code Aster instead of Abaqus.