NFRAMES command did not work.

[andrea...@gmail.com]

Hello,

   I ran into something strange.

The NFRAMES command did not work in a series of simulations.

I entered this command line:

& TIME T_END = 400 /

& DUMP NFRAMES = 4000, FLUSH_FILE_BUFFERS = .FALSE., DT_ISOF = 20.0 /

but if I look at the _devc.csv file, the last line, which corresponds to 400 seconds and which should correspond to the excel 4000 line, is instead on line 3651.

I noticed this when I compared this devc.csv file with the same simulation but with half mesh (sensitivity study).

In this second simulation, the NFRAMES command was strictly respected and worked properly.

Do you by any chance know what the problem may be?

thank you

Andrew

[dr_jfloyd]

You can't have more outputs than you have timesteps. 400 s / 4000 = 0.04 s.  In your csv file the intervals between outputs start at 0.35 s and end of 0.05 s suggesting that the FDS timestep is larger than 0.04 s throughout the simulation.  Was this the case?

[andrea...@gmail.com]

Thank you for your answer.

I had to read a few times your answer to understand it, but after searching  on the guide and reading the chapter "controlling time step" i think i get it now. 

Tell me if I understood correctly.

Simulation 1, 40x40x42.5 cm mesh

The default value for my simulation time step, calculated with the formula 5(dxdydz)^(1/3)/sqrt(gH) is 0.35 s, which is higher than 400/4000= 0,1 s.

Looking at the steps.csv file I see that the time step value remains higher than 0.1 s for most of the simulation, so there are probably overall less than 4000 available timestep (probably 3500, as suggested by the number of lines in my devc.csv file.

Simulation 2 20x20x20 cm mesh

The default value for my simulation time step, calculated with the formula 5(dxdydz)^(1/3)/sqrt(gH) is 0.17 s, which is still higher than 400/4000= 0,1 s.

BUT since the time steps value is slightly higher than 0,1 and only for few seconds of the simulation, there are overall more time steps than requested results (more than 4000), allowing the devc.file to display 4000 results.

I usually put a high NFRAMES amount to have fluid smv visualization, but this can be an issue with the comparison of DEVC results with different meshes. For example now I have this 2 simulation with a different amount of rows in the devc files. I will probably start using NFRAMES = T END. 

Any thought about this?

Thanks a lot again,

Andrea