Simulation of a Turbine Cascade

General
1

Dear Dr. Witherden,

I am trying to simulate a low pressure turbine cascade like the T106c case in your paper and I have couple of questions regarding to the results.

  1. How should the residuals be evaluated? In my case the results seem logical in ParaView but the residuals are too high (for instance the amount of the residual of mass conservation Rho is about 1500-1800) even after a long time of simulation. In my understanding the residuals in PyFR are normalized with the time step( which is 5e-5 in my case), which means the R.M.S residual is in order of about 0.1, am I right?. I have also changed the controller option from PI to none and used a smaller time step but there has been no difference in the residuals!

  2. How can I Plot the time and span average pressure distribution? Actually I wanted to plot the time- and span average isentropic mach number to evaluate the aerodynamics of the blade. In order to do this I used the time average pressure (using the [soln-plugin-tavg] option), however I was not able to extract the wall boundary(the blade) in Paraview. If it is possible in Paraview, could you give me a tip, which filters I should use to extract boundaries?

I am attaching the config file used, in case you want to take a look at.

Much appreciated,

Amir Matin

Config.ini (1.44 KB)

2

Hi Amir,

In regards to your questions:

  1. I am not sure what you are referring to as residuals in the context of your case. Since you are using PyFR to perform an unsteady simulation of the Navier-Stokes equations (not for example a steady state RANS solution) then the concept of residual does not apply. Instead, what you want to do is run the unsteady simulation for a long enough period of time that a time-average of the forces and flowfield stops changing. If you can clarify what you are calling the residual that might be helpful.

  2. This is a bit fiddly to do with Paraview. However, one pipeline that has worked for me sometimes (not always) in the past is: merge blocks → clean to grid → extract surface → then clip just the surface of the blade out.

Thanks,

3

Hi Brian,

Thanks for the reply.

I was monitoring the Residuals printed from [soln-plugin-residual]. So, as you said, since this is an unsteady simulation, I have to monitor the fluid force form [soln-plugin-fluidforce-blade]
to evaluate the convergence.

I did exactly as you said in Paraview, still could not extract the Boundary Surface I needed. However in the new release It is possible to extract the Boundary Surface in the first place.

Thanks again for your help.
Amir