[Julian121]

As I have written in my previous posts, I am doing a transient simulation of a passage of IGV/Rotor/Stator and a casing treatment.

I am evaluating the performance of the different casing treatment configurations.

I am using static pressure BC at the outlet.

I have tested many things including:

Adaptive time stepping to see if not having enough small time step causes not getting converged solution

Using static pressure at outlet instead of mass flow as was suggested in the CFX manual

Transient blade row and transient analysis

Passing of more than 5 full turns

I have waited so long but still a converged solution which should be a reapating pattern is not made.

Does not having equal theta in circumferential direction which is the case for my simulation (IGV 10.58 degrees/Rotor 9.47 degrees/Stator 9.72 degrees /CT 9 degrees) will cause difficulty in getting convergence?

Has anyone ever had such experience?

The attached image shows the mass flow at the inlet and the average over the passing period has been imposed on the mass flow.

[ghorrocks]

Fluids don't always converge to a repeating pattern. Fluids are infamous for generating complex fields in 3 space dimensions and time - turbulence for instance. So you may well have come across a case of a non-repeating flow.

In these cases you can't look for a repeating pattern, instead you have to wait until the time average converges to an accuracy you are happy with. You have to determine an appropriate time averaging time scale - you want it long enough to average the flow out, but not so long it includes the start-up transient.

[Julian121]

Like you said it seems that no repeating pattern is going to happen for my case.

I have decided to do FFT analysis of the unsteady pressure signal in the CT.

The rotor has 38 blades and the rotational speed is 3000 rpm, but after 4700 time steps the main frequency is about 950 Hz which is half of the rotational speed.

The probe is placed in the CT which is a stationary domain.

Could someone please explain the reason why the rotor blade passing frequency is not seen in the spectrum?

[AtoHM]

Seems a bit odd, that the passing is not visible at all. Anyway, FFT is not of much use when you have pitch change at the interfaces of your components. Through the pitch ratio, the profiles are scaled which shifts your frequencies as well leading to unexpected FFTs. I investigated this last year with a rotor-stator setup. The FFT showed a rotor passing peak which was exactly shifted by the pitch ratio inside the stator domain.

[Opaque]

Originally Posted by AtoHM

Seems a bit odd, that the passing is not visible at all. Anyway, FFT is not of much use when you have pitch change at the interfaces of your components. Through the pitch ratio, the profiles are scaled which shifts your frequencies as well leading to unexpected FFTs. I investigated this last year with a rotor-stator setup. The FFT showed a rotor passing peak which was exactly shifted by the pitch ratio inside the stator domain.

That is correct if you use the Profile Transformation method (pitch scaling)

If you use another pitch change model such as Time transformation, or Fourier transformation, you should see the correct frequency due to the rotor in the stator (better be or it is a problem in the models)

[Julian121]

I am not quite familiar with the passing frequency concept.

With the probe inside the casing treatment, is it expected to see just the rotor passing frequency or a combination of IGV Rotor Stator or their harmonics should appear in the frequency spectrum?

If a cyclic event happens in the CT, will it appear in the spectrum as well?

The blade passing frequencies I have are as follows:

IGV passing frequency = 1700 Hz

Rotor passing frequency = 1900 Hz

Stator passing frequency = 1850 Hz

Casing treatment frequency = 6000 Hz

In the frequency spectrum, there exist four main frequencies that are not related to these frequencies.

[Julian121]

I have placed a probe in the casing treatment. The unsteady pressure signal for this probe shows some dominant frequencies other than the rotor passing frequency.

In general, for the probe inside the casing treatment, might other frequencies than the rotor passing frequency (for example IGV’s or Stator’s) appear in the spectrum?

I have also placed another probe in the rotor which rotates with the rotor. When I analyze the FFT of this probe the rotor passing frequency do appear in the spectrum. Is it correct?