About the RESTART, TIME_SHRINK_FACTOR and SPECIFIC_HEAT

[ESett5555]

Hi everyone,

I am simulating a 7x7x11 m room, with 1m thick concrete walls. The room is heated by 8 steel blocks using INTERNAL_HEAT_SOURCE=2500. I have two baseline setups: one with a 1 m^3/s supply of air at 20°C, and one without ventilation.

As I was interested in the first place in the steady-state results (mainly the temperature of the walls), I specified TIME_SHRINK_FACTOR=1000 on the &MISC line. I managed to get the steady-state results as I wanted, for both cases. Without TIME_SHRINK_FACTOR, the simulation takes too much time to be run.

Now, i want to run a transient case: starting from the steady-state solution with ventilation on, I want to switch off the supply fan via a RESTART while disabling TIME_SHRINK_FACTOR so that physical time progresses normally.

However, if I try to delete the MISC line, or if I try to reduce the value of TIME_SHRINK_FACTOR, FDS gives me an error when I try to restart the simulation. The error codes that I had were :

• forrtl: severe (67): input statement requires too much data
• forrtl: severe (157): Program Exception - access violation.

I think that this is due to the way the .restart files are made.

I tried to get around the problem by keeping TIME_SHRINK_FACTOR=1000 and scaling up the value of SPECIFIC_HEAT on the MATL lines (e.g. from 0.87 to 870 kJ/(kg.K)), but it gave me the same types of errors. I only managed to change it from 0.87 to 0.9.

Do you have an idea on how I could manage to get my simulation running?

Thanks in advance for your help !

[mcgratta]

Are you doing this on a Windows computer? I ask because users sometimes have trouble with restarts on Windows.

[ESett5555]

Hello again,

I managed to get the RESTART working with a bit of work to bypass the problem. In case anyone encounters the same issue, here is what worked for me:

1. Defining a constant CELL_SIZE=0.03 on all &SURF lines prevents the solid-phase mesh mismatch and allows RESTART to execute, even when adjusting SPECIFIC HEAT or TIME_SHRINK_FACTOR before the restart
2. To solve the time step issue, instead of using TIME_SHRINK_FACTOR, I initially set SPECIFIC_HEAT=0.00087 on the &MATL line to reach steady state quickly. Once converged, after stopping the run by placing a CHID.stop file in the calculation directory, I updated the material property to SPECIFIC_HEAT=0.87 and resumed the calculation with RESTART. The calculation continued with he same time step, as expected.

I am now facing an other issue regarding RESTART (and &CRTL logic), but as this is unrelated to TIME_SHRINK_FACTOR nor SPECIFIC_HEAT, I will open a separate discussion for it.

Thank you for the help !

[mcgratta]

Thanks for the information. Just to clarify, do you mean to say that items 1 or 2 are potential solutions to the problem? In other words, one could do either 1 or 2 to solve the problem?

[ESett5555]

In my case, I needed to do 1 and 2.

Fixing a constant CELL_SIZE (item 1) is required to use RESTART while changing SPECIFIC_HEAT (which is what I do in 2). So item 1 is mandatory, while item 2 might only concern my specific issue.

In other words, item 1 concerns the ability to use RESTART, while item 2 concerns what I did to solve the time step issue here :

However after the restart, the time step went from 16.1 s with TIME_SHRINK_FACTOR=1000 to 6e-4 s with TIME_SHRINK_FACTOR=1. I would expect a factor 1000 reduction of the time step (and thus a time step around 1.61e-2 s), but it seems that this is not the case.

[mcgratta]

Check the CFL and VN (Von Neumann) numbers before and after. For small solid phase cells, we might be seeing a VN constraint.

[ESett5555]

It seems to be both the CFL and VN numbers that limits the time step in my calculation (when changing TIME_SHRINK_FACTOR), but as both of them are proportional to the time step, I would expect a time step around 1.61e-2 s.

If I understand it correctly, the CFL number is only checked in gas cells. Is the VN number also checked in solid cells? With thermal diffusivity such as ? :