User login

Navigation

You are here

Simulating concrete failure in ABAQUS

Hi all,

I am making my PhD simulating composite concrete, that is, concrete reinforced with composite materials. Until now, I have made research on cohesive elements of ABAQUS. Now, I am focussing on the simulation of concrete. I was told that X-FEM was a great improvement on simulating fracture. Indeed it is, but in my case it is unsufficient. I could model perfectly the first crack, but the problem is that several cracks were expected and I have the impression that ABAQUS can just model a crack per enriched area. I tried to create so many enriched areas as elements I had of concrete, but a maximum of 20 enriched areas are supported by ABAQUS. Have you heard about any kind of user subroutine able to solve this limitation?

Furthermore, I am testing both the concrete smeared and the concrete damaged plasticity material models. They seem to work well in my one element test model, both uniaxial and biiaxial tests. I am working right now has around 400000 degrees of freedom. There, ABAQUS tends to converge slowly and the following warning is written:

 ***NOTE: MATERIAL CALCULATIONS FAILED TO CONVERGE OR WERE NOT ATTEMPTED AT ONE
          OR MORE POINTS. CONVERGENCE IS JUDGED UNLIKELY.

Does anybody can give me some feedback on how I can avoid this problem? The simulation stops suddenly at 18% of the final displacement because due to this warning the timestep is reduced down the minimum limit.

Thanks and best regards,

Mikel

Comments

Hi members,

following steps were performed: 

1) The possibility of preparing a user subroutine to overcome the limitation of the X-FEM will be studied, but it is on stillstand.

2) With the concrete smeared material model the calculation stops as soon as cracking is reached, in spite of the good performance in the one element model.

3) With the plasticity material model it works better but still it do not converge. I managed to understand the meaning of dt and dc as factors defining the plastic strain, having both of them a value between d (d=1-(sigma/(total_strain))/E0) and 1. Special attention has to be paid on doing both of them monotonically increasing with the crushing or cracking strain. I just did it with the help of Excel and defining a quadratic law between inelastic and plastic strain. I could see very well the differences in the unloading slopes when I restarted the one element model from different intermediate points, both in the cracking and crushing failure curves. Does anybody have material data that works of this material model? I found something but it is still to be proven.

4) I tried also the configuration *STATIC,STABILIZE and the configuration *DYNAMIC,HAFTOL with material damping but they converged very very slowly and I killed the simulations. I also made a test with *DYNAMIC,EXPLICIT but it was also very slow. I think I have to revise the material properties. 

I would thank any kind of feedback.

Best regards,

Mikel

Dear all,

I have gone further with my project. I have managed to converge the 2D and 3D models using a more regular meshing and using the unsymmetric solver of ABAQUS. This solver use more resources but  it is adequate for this type of problems. It was tested, with the same model and options changing the solver to symmetric and it did not converge. I have also reduced the number of points in my material model, so that the solver has more freedom. I also changed the concrete tension stiffening model from GFI to the default STRAIN. It seems to converge until a much further calculation point, but now I have to analyze the validity of the results. 

I will thank any kind of feedback. 

Regards,

 Mikel

Subscribe to Comments for "Simulating concrete failure in ABAQUS"

Recent comments

More comments

Syndicate

Subscribe to Syndicate