Description + Goal
A representative wind tunnel simulation of given car - to calculate drag/lift values, analyse airflow and identify problem zones
Problem:
Near end of simulation when states “100%” error occurs without any indication why. Here is error
link provided for solution approach: Fatal IO Error: An Error Occurred | SimScale Knowledge Base. Data seems to be fine since all force and momentum values have been calculated and graphed. Guess that error occurs while creating the solution field data/visualizations since there only the first iterations can be viewed (e.g. until iteration 180 of 500)
occured in last 3 runs -since new mesh was used, which was suppose to be more detailed therefore precise + also with different model using same parameters for mesh generation.
Not sure what to try since the error doesn’t speak and appears after everything seams to have worked fine.
I had a look at your simulation, and as you point out, the error happens after the solution is computed, apparently when processing the result data.
With that in mind, I noticed that your are specifying a large number of solution steps to be written:
This means that you will write every 15 out of the 600 steps, which is a relatively large amount of steps and data consequently.
For a steady state simulation this is not needed. As we are computing for the converged state of the flow, what we want is only the last computed time step. In other words, all of the intermediary steps can be discarded as not being the final solution.
Please write only the last time step by setting the Write interval to be equal to the End time, and try again.
I am sure this would work. However i do have a reason for writing so many solution steps: to recreate animations like this one from cobrien
This allows me to visualize the changing vectors of flows and their characteristics (turbulence, fluctuation, etc.) this is the only way I could think of how I could recreate them (since unfortunately he did not answer me when I asked him personally under his project nor I was able to derivative the answer from his simulation setup)
However fewer time steps would be sufficient, they would have to be located after divergence (which would be around after the 350 iteration). Is there a way to set a specific starting point from when onwards results should be written in a certain interval or simply listing the iterations which should be written (e,g, like 350; 351; 352; …)?
It seems to me that this animation comes from a transient simulation, compared to your steady-state solution:
In the steady-state solution, only the last solution step is actually relevant (because it has the lower residuals).
Thus, you must change to a transient simulation, which will compute a time history of the solution fields, which you can plot and extract the information that you mention. Just be aware that the computation times and core hour consumption will scalate!
I am sorry there is no a direct feature to achieve this in the Incompressible solver, but I noticed and raised the feedback to the dev team. A possible workaround is to use the ‘Continue run’ option to split the run in two parts:
The first run is for the period you don’t want to write. Here you can write only the last time.
The second run is for the period you do want to write. Here you can specify the write frequency.
I thought too that it might come from an transitent, however the project refered to did not seam to have done so and I was not able to set it up properly myself.
Nevertheless I like your proposed work around and will give it a try.
