Here, I'll get through some of many issues I'm currently facing with SOLID65 element type.
......
I'm modeling a torque experiment with concrete of compressive strength 48 MPa.
To be honest, the only reason that I'm still making trials with SOLID65 is the cracking/crushing capability.
However, it's resilient, rigid, cumbersome and very bad.
1- At the final load substep before the solution loses convergence, there's no apparent plastic strain, most of the time there's none.
2- The deformations are so small, this is mostly dictated by the pure elastic behavior with a high Young's modulus.
3- Providing the multilinear stress-strain values of the simplified curve does not yield the material to enter the plasticity zone, or, the plastic strains (if they emerge) are so small.
Here's an article that gives the equations to construct this curve.
Article Shear Strength Prediction of High Performance Reinforced Con...
4- ANSYS APDL crashes when William and Warnke material model is applied with a multiliear stress-strain relation, and it works when dilatancy is not supplied, however, the deformations and strains are mostly to fully elastic.
5- With SOLID185 I can insert purely plastic strain, but with SOLID65 an error arises.
Increasing the value of strains leads to premature convergence loss, and to excessive deformation at the moment of convergence loss.
Here are URLs about this error.
https://forum.ansys.com/discussion/9219/error-the-value-of-ux-node-exceeds-any-solution
https://www.researchgate.net/post/Why_have_I_received_an_error_message_in_ANSYS_Degree_of_freedom_limit_exceeded
........
Feel free to tell us about the limitations and issues of SOLID65, and how to properly manifest the plastic behavior of concrete.
Perhaps my article is interesting for you - I used Solid65 elements.
Article Modelling and quasistatic simulation of the stiffness degrad...
Jürgen Ries
I've implemented a multilinear elastic stress strain curve close to the way described in your article.
Sadly, when all the points are provided, the same DOF excessive value error occurs.
So, I've removed some of the lowest points and left the values to strain 0.004, however, the solution loses convergence in the mid of the process.
Currently I'm still clueless with SOLID65.
In my experience, convergence problems have usually been caused by meshing failures.
Jürgen Ries
I've toyed a little bit with MELA, I think I started to figure it out.
What matters is the "positive" elastic slope.
The material starts to be more ductile (apparently softer) when it enters a phase of lower "but positive" slope.
The issue of your softening graph is that it contains "negative" slopes which leads to hyper displacements which leads to the error of DOF excessive value.
ANSYS, then, reports the first node that exceeds the limit (defaults to 1E6) and terminates the solution.
____________
With that in mind, my next focus is on how to make ANSYS determine the ultimate load.
- Badges
- Science topic