# FFT from simulation data - Are simulation time steps "real time"

 Register Blogs Members List Search Today's Posts Mark Forums Read

 April 15, 2020, 12:40 FFT from simulation data - Are simulation time steps "real time" #1 Member   Brandon Join Date: Jun 2018 Location: Germany Posts: 46 Rep Power: 7 My Simulation: Wall resolved LES around an airfoil section to study laminar to turbulent transition. Not relevant, but good to know: I have generated isotropic "atmospheric" inflow conditions inserted close to the airfoil and the divergence-free condition is satisfied. What I need: An FFT at different locations to observe how the energy at different frequencies change as the flow goes from laminar to turbulent. What I have: Time series data of the velocities, pressure (relative pressure) and Reynolds stresses recorded at certain monitored points after the simulation converged satisfactorily. I would be using the pressure to plot the FFT as this is what I would use had the experiment been done in real life using microphones. My Question: Since I am using non-dimensional units everywhere (I think), can I consider the time step of 0.001 to be 0.001 seconds (Just an example time step not the actual in case you think it is not sufficient for transition detection)? If not, how do I do this so that I get the frequency in Hertz when I generate my FFT using the time step difference and pressure data? This is why I think that I use non-dimensional units: I have a unit inflow velocity and have set the Reynolds number needed, so the viscosity is modified accordingly. The chord length is one unit (because based on the chosen Reynolds number the viscosity is modified considering a length of one). The time step is set as 0.001 which I want to believe (and my logic says) can be considered as real-time as though this was an experiment conducted in real-time using a unit chord length airfoil in a liquid of chosen viscosity where the pressure was sampled using a probe every 0.001 seconds or with a frequency of 1000 Hz.

 April 16, 2020, 03:29 #2 Senior Member   Tom Fahner Join Date: Mar 2009 Location: Breda, Netherlands Posts: 634 Rep Power: 32 Hi, As long as you are solving with a transient solver and not in Local Time Stepping mode a simulation time of 0.001 s is 0.001 s. With LES I guess you run with pimpleFoam or pisoFoam, so that works out. Even if you choose unit lengths or velocity, it is still 1 m and still 1 m/s. Best regards, Tom kepler123 likes this.

April 16, 2020, 07:54
#3
Member

Brandon
Join Date: Jun 2018
Location: Germany
Posts: 46
Rep Power: 7
Quote:
 Originally Posted by tomf Hi, As long as you are solving with a transient solver and not in Local Time Stepping mode a simulation time of 0.001 s is 0.001 s. With LES I guess you run with pimpleFoam or pisoFoam, so that works out. Even if you choose unit lengths or velocity, it is still 1 m and still 1 m/s. Best regards, Tom
Thanks a lot for the reply Tom, this was extremely helpful. I just had to make sure that it was right. Yes, I do use a transient solver and the piso algorithm