Hi
I am working on a Hyperelastic material. My material model is based on strain energy function (W) = U(J) + W(b)
where, U(J)=k/2*(log(J))^2; k=bulk modulus; J:Jacobian
W(b)= mu/w*(tr(b)-3) ; mu=Shear modulus; tr=trace; b=deviatoric deformation tensor
I have written UMAT for my material model and I am trying to simulate punched problem (rectangular block being subjected to roller on three sides(two lateral sides and bottom) and on top, pressure is being applied in middle half length). For smaller load, everything works fine and I get the result (Abaqus works)- smaller load meaning small deformation. So now when I increase load to higher value, there is convergence problem in Abaqus. Abaqus shows message that the solution is diverging and could not be converged. Have any one of you encountered with this problem? What are the necessary steps to be taken in case of having convergence problem? Is this the problem due to my material model or there is a way to get away with this? I have NLGEOM on, and my material is compressible.
Thank you for your input.
Anup
PS: I validated my material model with 1-D uniaxial test where analytical solution matched the Abaqus solution.
Which elements are you
Which elements are you using?
Also, you might want to use a UHYPER instead of a UMAT; I am not sure, but it might be that the formulation in UMAT does not work that well beyond the linear regime.
convergence problem
Hi Francisco
Sorry for the type. I have UHYPER, not UMAT and I wrote this UHYPER based on Abaqus documentation. I am getting large deformation in 1-D case. The deformation gradient is as high as 100 in tension and as low as 0.05 in compression. But for 2-D case, there is a convergence problem when large load is applied.
I tried both linear and quadratic element. My simulation is based on plane strain so my element type is basically CPE4(linear) or CPE8(quadratic). Mesh refinement also didnot help.
Thank you for your time.
How compressible is the
How compressible is the material? Have you tried using hybrid elements?
Also, you just might be getting to deformations that are too high... Does the deformed geometry look very weird already when you look at it in CAE? How do the deformations when it breaks down compares to the one in the 1D case?
convergence problem
I tried using Hybrid element (CPE4H) but still facing the same problem.No change even with hybrid elements. My material has E= 15 MPa and Poisson's ratio of 0.25. The deformation is not that wierd. The maximum Jacobian value is nearly 0.65. In 1-D, the rectangular bar subjected to uniaxial tension is nearly changed to a straight line shape( meaning highly deformed and no problem of convergence). In 2-d, as I am compressing the geometry on the middle half length , and the two lateral sides have roller, there is some region that is in tension and most of the domain is in compression. I do not know this combination is hurting the modeling. I have attached herewith the deformed shape of my geometry.
Thank you.