I have been experiencing trouble with the amount of substeps required to obtain a solution. In my case I am analyzing the ultimate capacity of an unreinforced masonry building. I use Solid65 and the Willam Warnke model for concrete.
I am plotting a curve of load vs deformation but the problem is that the curve has a different shape depending on how many steps i use for the analysis.
It does not only affect the accuracy at the final point but actually affects the whole curve. Here I attach the plots. So if the amount of substeps seem to have an influence in the result not only in the accuracy.
Any idea of why this phenomena occur?
Mr.Andres,
You could see from load - deformation curve that there is no change in response in case of all increase in number of load steps. It is general understanding that the load can be applied even in one step or with number of steps to predict the elastic response. We need to apply load with higher load steps for predicting the nonlinear behaviour. so your results are ok. you apply load with 500 load steps that give better results and accuracy of result will be quite reasonable beyond the elastic response. all the best in your research.
Beside Claudio's answer which described well, it seams to me that you only used one time step for your analysis. for that reason I think your analysis won't be time efficient. I suggest that yoiu set two time steps for your analysis. Consider the first time step for the relatively linear portion of your analysis with maximum 10 substeps and the second time step for the nonlinear portion with somewhere between 100 to 200 substeps. This way you can get more accuracy where it is needed in your analysis and it will be more efficient as well.
- Badges
- Science topic