Tek-Tips is the largest IT community on the Internet today!
Members share and learn making Tek-Tips Forums the best source of peer-reviewed technical information on the Internet!
-
Congratulations MintJulep on being selected by the Eng-Tips community for having the most helpful posts in the forums last week. Way to Go!
Material Model in Ansys for Elastic Perfectly Plastic Analysis 2
- Thread starter kevin314
- Start date
ThomasH, inline6
I am not able to understand why are you both "sceptical" about the convergence. Yes different softwares adopt different aprroach. But I guess limit load analysis (or I can say to any analysis) if performed on different softwares should give same results keeping all the parameters same. Otherwise there will be no universality in the results. Also underlying algorithm for all softwares is Finite element method which is general. Doesn't it?
We need to understand what convergence means for this method. I am also doubtful about how it has been achieved. But still what I think is as follows.
First software will calculate deformations by following the slope path of the material model. If the load is higher than yield or dimensions are such that it will not hold on to, the component will yield and distribute the load to other sections and if the sections are able to take that load, it will either yield to Finite strain and solution is converged or it will go in loop of calculating infinite strain/deformations following the constant stress line of material model.
For all the softwares, I guess process will be same irrespective of programme used to achieve it. So software requirement are their on mesh, geometry, boundary conditions and assumptions. But I guess it is not software dependent.
Engineer before the idiot(sorry intelligent, I mean) box which runs these softwares need to define that.
rb 1957
No. We generally don't run limit load if the structure is passing in the elastic analysis. Logical. I have given an example where if your singularity is widespread(it can't be called singularity then) such that the linearisation of stresses is ambiguous and results in the failure to the limits defined by ASME, we go for limit load or strain hardening approach which you follow.
I am not able to understand why are you both "sceptical" about the convergence. Yes different softwares adopt different aprroach. But I guess limit load analysis (or I can say to any analysis) if performed on different softwares should give same results keeping all the parameters same. Otherwise there will be no universality in the results. Also underlying algorithm for all softwares is Finite element method which is general. Doesn't it?
We need to understand what convergence means for this method. I am also doubtful about how it has been achieved. But still what I think is as follows.
First software will calculate deformations by following the slope path of the material model. If the load is higher than yield or dimensions are such that it will not hold on to, the component will yield and distribute the load to other sections and if the sections are able to take that load, it will either yield to Finite strain and solution is converged or it will go in loop of calculating infinite strain/deformations following the constant stress line of material model.
For all the softwares, I guess process will be same irrespective of programme used to achieve it. So software requirement are their on mesh, geometry, boundary conditions and assumptions. But I guess it is not software dependent.
Engineer before the idiot(sorry intelligent, I mean) box which runs these softwares need to define that.
rb 1957
No. We generally don't run limit load if the structure is passing in the elastic analysis. Logical. I have given an example where if your singularity is widespread(it can't be called singularity then) such that the linearisation of stresses is ambiguous and results in the failure to the limits defined by ASME, we go for limit load or strain hardening approach which you follow.
A not uncommon selling point for nonlinear FEM-software is their ability to converge a solution. Convergence issues can be due to structural failure (instability) but also from numerical issues.
An example is if you have a structure that you want to load to the limit. If you apply the full load in a single step you can get convergence issues for a number of reasons. If you instead apply the load in smaller steps like 10%, 20% 30% and so on until you reach 100% the computations take longer time but you have a more stable approach to the critical level. So the convergence issues may come at a higher load level. I have seen this happen more than once. So the later approach can result in a higher ultimate load than the first.
And that is one parameter. In the nonlinear solvers there are usually several methods available depending on the exact application. I recently attended a seminar on Nonlinear FEA and there is a lot of pitfalls and parameters. But perhaps this particular code has very specific requirements now to perform the solution?
Thomas
An example is if you have a structure that you want to load to the limit. If you apply the full load in a single step you can get convergence issues for a number of reasons. If you instead apply the load in smaller steps like 10%, 20% 30% and so on until you reach 100% the computations take longer time but you have a more stable approach to the critical level. So the convergence issues may come at a higher load level. I have seen this happen more than once. So the later approach can result in a higher ultimate load than the first.
And that is one parameter. In the nonlinear solvers there are usually several methods available depending on the exact application. I recently attended a seminar on Nonlinear FEA and there is a lot of pitfalls and parameters. But perhaps this particular code has very specific requirements now to perform the solution?
Thomas
- Status
Similar threads
- Question