su2code/SU2

Add more content to history.csv file

lcheng9 opened this issue · 6 comments

lcheng9 commented

Is your feature request related to a problem? Please describe.
No time and min/max value are available in the history file, which makes the post-processing module difficult to handle the unsteady results.

Describe the solution you'd like
Add a "Time" column in the history.csv file, this column shows the simulation time for the current "Time_Iter". Add minimum and maximum values for all available result quantities, such as pressure, velocity, density, etc.

Describe alternatives you've considered
An alternative way to get the min/max values is by reading each flow_.vtu result file and it is expensive for a many-step unsteady analysis. There seems no way to have the "Time" quantity for now.

Additional context
An example of the columns in history.csv is:
"Time_Iter", "Time", "Outer_Iter", "Inner_Iter", "Rho_min", "Rho_max", "RhoU_min", "RhoU_max", ......, "rms[Rho]", "rms[RhoU]", "rms[RhoV]", "rms[RhoE]", "rms[nu]"

lcheng9 commented

Looking into the source code, I found the keyword "CUR_TIME" can show the time in the history file. Is there any reason that CUR_TIME is not listed in the history.csv by default for transient analysis? Now I can output CUR_TIME column locally, but not sure if should I check in the code. Thanks.

bigfooted commented

You can show all available options for screen/history output by running SU2_CFD -d config.cfg

lcheng9 commented

Thank you for your response, @bigfooted.
I can see the available options using the flag -d as you suggested.

Is there any way that I can activate the CUR_TIME in the history file using the config.cfg? Currently what I did is adding the line below to certain source files, then build and run. It works but has to change the source code.

if (config->GetTime_Domain()) requestedHistoryFields.emplace_back("CUR_TIME");

Thanks.

lcheng9 commented

Thank you for sharing the Output details @pcarruscag. Now I can output the Cur_Time column with only changing of input script.

After reading the source code, I also found that the history field depends on the user settings, such as physics type, turbulence model, etc. If I manually assign the HISTORY_OUTPUT variable, will the default settings be wiped out? It seems that adding the "CUR_TIME" to the default history field for transient analysis is reasonable if not conflict with other settings. I will be proposing a PR, however, please feel free to comment or challenge. Thanks.

KillingVector commented

SOLVED see edit.

Hello! I seem to remember that it used to be possible to get angle of attack in the history file? Is this still possible? I would like to be able to just pull the angle of attack out of the history file, along with the aero coeffs.
Context:
I have looked through the custom output document and the reference config file, and (perhaps I am secretly blind), but there does not seem to be a simple output group for it.
Reason I need it: performing analysis for fixed Cl, so angle of attack is useful information.

At the moment I am getting probed data:

CUSTOM_OUTPUTS='m_vel_x : Macro{VELOCITY_X};\
                m_vel_y : Macro{VELOCITY_Y};\
                vel_x1: Probe{$m_vel_x}[-15, -15];\
                vel_y1: Probe{$m_vel_y}[-15, -15]'

And just getting the tan(y/x) of the probed values for angle. This works well enough.
The other option I have is stripping the angle of attack out of the forced breakdown file, would prefer to not do, but can.

EDIT:

It can be pulled out of line 2 of flow.meta