Particle tracking INTEGER limitation warning
I have encountered a significant limitation in FLUENT in the DPM model. I am also aware that this problem may not be possible to solve at the user's side.
Here it is: setting the 'Max. Number of Steps' in 'Tracking Parameters', the highest possible number is ~2e9 which roughly equals the C/C+ signed int type (4B) –2,147,483,648 to 2,147,483,647.
This limitation would not be so severe if the same case did not appear when setting the 'Specific length Scale' to define the integration time step for steady tracking. Please note that the time step as given in the manual 25.2.3 reads
dt = L / (up+uc). Even setting L~1e-12 will limit the particle time step to the lowest number of 1e-9s, which I validated by plotting particle tracks with particle time step argument.
What I really don't understand is why there is this kind of limitation when running on modern PC's with x64 architecture? Shouldn't be used long/long long instead of int and double instead of float?
I will appreciate any fruitful discussion.
June 26, 2013 UPDATE
I found the FLUENT's GUI to be erratic. The limit values that are set in the GUI are not passed into the particle tracing module and the above problems occur. In addition, having a UDF for particle injections DPM_DEFINE_INJECTION, the parameters set in the UDF are not used when tracing steady particles by hitting 'Pulse' or 'Track'; instead the values set in the particle injection dialog are used.
However, when using unsteady particle tracking with defined time step and number of steps per steady continuous iteration, the time step can be set extremely small (1e-12s) and the initial injection parameters as defined in a UDF are used with all the benefits of large/smass numbers. In this case, the simulation can be proceeded only by 'Run Calculation --> Calculate' dialog.
Hope someone may find this helpful when struggling with FLUENT DPM.
|All times are GMT -4. The time now is 17:51.|