Hi,
I have a question about acquiring the softening part of the compressive stress-stain curve of concrete under uniaxial static loading.
I have tried very different options: Concrete Damaged Plasticity model, different stress, strain, damage material parameters as input, Newton and Riks (arc length) calculation methods, but still have some problems getting the softening part of the curve.
Has anyone have any suggestions to get a realistic softening in a static uniaxial test? Actually, the problem is always the same. The solution does not converge after the peak or sometimes before the peak.It gives the error message: Too many attempts made for this increment Have tried Newton and Riks methods. Have decreased the min time increment down to 1E-25, decreasing it even further does not help either.Have tried both implicit and explicit calculations.I have seen numerous posts in the internet about not being able to get the strain softening part of the stress-strain curve of concrete but no answers on the solution.
Dear Akshay,
It is difficult to answer your question without knowing the details of your model. However, I would recommend the followings:
1. Carefully review your material model. You can use common stress-strain relations, such as Mander et al. 1988 model, for defining the uniaxial compressive behavior in the Concrete Damage Plasticity (CDP) model. Also, check the definition of your uniaxial tensile behavior, since it may be the cause of the convergence problem. Try to avoid decreasing the stresses in your stress-strain curves down to ZERO since it may also cause numerical issues. Use a slightly higher value instead.
2. Use dynamic/explicit or dynamic/implicit approach with a quasi-static loading scheme. Increase the duration of analysis and loading to reduce the dynamic effects in your results.
3. Apply DISPLACEMENTS instead of loads to have a displacement-controlled simulation. Don't apply loads to your structure, apply displacements instead.
If you explain your model details further, others may help you more.
I hope this helps.
Hi S.K. Azad,
First of all thank you for answering my problem in a short time span.
Actually, I am modelling simply supported RC T beam to show the shear lag effect in the compression flange at Ultimate condition (Limiting Condition). So as It is behaving good when the stress level in the concrete is low (till half of the ultimate strength of concrete). But as I am applying higher loading (UDL) which should induce ultimate stress in the concrete in compression, it is either not reaching that maximum level or sometimes even have crossed that ultimate level. But it is not showing the strain ductility part of the stress-strain of concrete and load deflection curve should also show that ductility nature.
I have used static loading and Riks static method of load application.
Is there any other approach for doing the same thing?
Dear Akshay,
No need to thank. As I said in the previous post, use dynamic implicit or explicit analysis method, and apply displacements instead of loads. I think it will help. Use online sources as well as ABAQUS documentation to find details of dynamic analysis as well as application of displacement.
Hello, the softening part of the curve means that fracture has to be taken into account. In order to obtain objective results one must need to try to model concrete using a non-local model or a fracture mechanics-based model.
Dear Akshay,
My colleague faced this problem before, after many trials, she decided to split the beams in to 3 parts and join then with interface elements, the locations and slope of the separators between parts is per-determined from the geometry ( from point of applied load to point of support reaction). this procedure gives good matching with experimental results. see attached paper.
best wishes
Hi Akshay,
The reason why you did not obtain the smooth curve in softening region is because you apply load in your model. For nonlinear static analysis, try to use displacement control instead of load control. I totally agree with Mr. Sina in his opinion. In addition, quasi-static analysis will also give you more smooth curve compared to static general. However, you have to define the increment of displacement as narrow as possible in order to attain the proper curve.
Regards
Hi - you may find the modelling approach and other useful information in of my student's thesis. The link is provided below.
http://lib.dr.iastate.edu/cgi/viewcontent.cgi?article=2133&context=etd
Hi , Had your problem solved? If yes then how you have done it plz let me know. i am also getting convergence problem after peak behaviour of load-displacement curve.
- Badges
- Science topic