That’s a question that I receive very frequently, and that’s my usual answer:

The result you get are accurate **according to your input parameters** if you observe the convergence of the solution.

**BUT** don’t forget your output results will always depend on your model and your input parameters.

If the input is wrong, you cannot get accurate results. That’s what some people are calling: garbage in / garbage out.

What most people are doing is that they are calibrating the simulation results with real experimental results. If you get simulation results that are matching the experimental results then you know that your approach is correct.

Using only simulation and never compare with reality (experiment/thesis) will rarely provide accurate results… and you will never know if those results are accurate or not…

Manisha Sachdeva says

Hey…. I am working on error estimates…. basically ZZ error estimator ( Zienkewicz and Zhu error)… I have one question regarding that… I need to formulate it for a crack propagation problem… linear elastic material behavior…. We are actually not following what is suggested by the authors… rather following the same principle…

stresses and strains calculated anywhere within the element would be much better measure that at the nodes… rather than forming a patch for calculating SPR values at the nodes… I am using lumped weight to project stresses from the gauss points at the nodes….and error will be its difference with the values of stresses calculated directly at the nodes… calculating derivatives of shape functions at the nodes and using stiffness matrix… My CONCERN…. stiffness matrix is always calculated at Gauss points…. for linear problems… I guess this would work… Do u think it would also work for singularities….

Cyprien says

Hi Manisha, I have to say that I have no idea… If you find the answer please let me know as well.

Theliban says

Hi all,

I am the new learner to Ansys, doing my Masters in Engineering Design field. I am doing the VM250 to get some knowledge in Non- Linear Analysis. While solving the problem, I am getting the error as shown in the figure. By decoding the APDL code, I did the procedures and applied the BC. But continuously I get the error message. What should I do to have the converge solutions? Please help me to decode the APDL code.

Thanks.

The Error is starts like “Internal Solution magnitude exceed its value”

Cyprien says

An Internal Solution Magnitude Limit Was Exceeded.

… Please check your Environment for inappropriate load values or insufficient supports. Please see the Troubleshooting section of the Help System for more information.

In most cases this message will occur if your model is improperly constrained or extremely large load magnitudes are applied relative to the model size. First check that the applied boundary conditions are correct. In some cases, loads that are self-equilibrating with no support may be desired. To help in these cases, if this message occurs, consider adjusting the weak spring stiffness or turning on inertia relief.