We came up with the same issue, trying to figure out what value for erosion strain for a high speed concrete impact problem with the HJC and KC models. We've done some sensitivity analysis on this and determined that it makes a big impact on the results. In the end we used tuned the minimum principal strain for erosion using a calibration shot and got good results for other tests.
Erosion in Ls-Dyna is a kind of numerical trick which allows you to delete elements, which have a negative influence on your calculation. This erosion has nothing to do with real erosion of concrete. Parameters of this erosion you define 'ad hoc'.
Pleas look at appropriate comments in the Internet in http://www.dynasupport.com (for the basics) or http://www.dynalook.com - the site presents papers from European and International LS-Dyna User Conferences and papers provided by other users. Look also at the Yohoo Ls-Dyna users group, where you can find many information connected erosion in simulation of concrete.
Thanks for your reply, for example I am trying to perform 4-point bending test on the concrete specimen 32.5X0.075X0.075 cm. I want to simulate the crack when the concrete reach its ultimate state of stress. For this type of implicit analysis, unfortunately I cannot find sufficient information additionally for impact problems. I have read a lot of papers also I find some notes and papers which discuss the erosion and contact algorithm of impact problems but I cannot exactly find a method which determines these type of parameters.
If we talk about the criterion in * MAT_ADD_EROSION added to concrete material models (such as *MAT072R3), it does not have any specific methods for the selection of the criterion and its values.
The known expert told me: When forced to use erosion I select the maximum shear strain criterion (EPSSH) and set it to about 100%. The shear strain does not care if it occurs due to tension or compression, only a lot of element distortion.
You need to do a parameter study to see how the value changes your results of simulation. You should have some experimental data to compare results of simulations and laboratory tests.
Of course, you can use other concrete models, e.g. MAT084 (MAT085), MAT159, MAT272 or MAT273. In MAT084(085) there is a possibility of observe cracks propagation in concrete in LS-PrePost.
Thanks a lot, firstly, I have started using MAT_111_JOHNSON_HOLMQUIST_CONCRETE_MODEL. There is FS parameters which determines whether the failure is in tension or compression mode, Also I try to use VOLUMETRIC and EFFECTIVE strain in MAT_ADD_EROSION keyword, I have done some numerical analysis of concrete plate perforation and penetration. I have collected a lot of scientific articles , also read most of the notes released on the official website of ANSYS LS_DYNA. Most of the papers focused on the results and there is little information about the failure criterion of the concrete plate under high-velocity penetration.
We came up with the same issue, trying to figure out what value for erosion strain for a high speed concrete impact problem with the HJC and KC models. We've done some sensitivity analysis on this and determined that it makes a big impact on the results. In the end we used tuned the minimum principal strain for erosion using a calibration shot and got good results for other tests.
In the reality there is not any physical meaning to express what the erosion parameter is, you should calibrate the material parameters with the experimental findings in terms of the velocity, dissipated energy or the strength of your applied type of concrete. Erosion parameter can be expressed as the numerical trick to simulate the elements should be deleted during the simulation. Maximum strain can be a good threshold for determining the erosion parameter. It directly affect the residual velocity in the case of ballistic performance of the concrete material subjected to the impact of bullet or any type of missile. The other important parameters are the parameters related to the behavior of concrete under pressure. These constants can be determined through the experiments.