[Julian121]

I am simulating a stage of an axial compressor to analyse the effect of a casing treatment on the performance of the compressor.
The simulation is done close to the stall point and I am using steady state simulations.
After nearly fifty time steps, the isentropic efficiency and total pressure start to oscillate and they do not change at all even after 500 time steps.
The mesh quality is fine. I have also used smaller time steps but the oscillation still exists.
Does the simulation close to the stall point necessarily mean that the efficiency and pressure must oscillate?

[ghorrocks]

This question is essentially the same as the FAQ: https://www.cfd-online.com/Wiki/Ansy...gence_criteria

"The mesh quality is fine" - how do you know this?
"I have also used smaller time steps but the oscillation still exists." - yes, the recommendation is to use larger time steps, not smaller ones.

[Julian121]

Originally Posted by ghorrocks

This question is essentially the same as the FAQ: https://www.cfd-online.com/Wiki/Ansy...gence_criteria
"The mesh quality is fine" - how do you know this?
"I have also used smaller time steps but the oscillation still exists." - yes, the recommendation is to use larger time steps, not smaller ones.

My mistake, I used larger time steps but again the isentropic efficiency and pressure at outlet are not getting horizontal. Is the oscillation a sign of rotating stall?
In experimental data, stall begins at lower mass flow rates than this mass flow.
What is the sign of stall in numerical simulations?
As for choking, I lowered the pressure at outlet until the mass flow becomes constant.
All the criteria such as mesh expansion factor, minimum face angle, aspect ratio are satisfied.

[Julian121]

The attached photos are taken after using larger time step.

[Gert-Jan]

Isn't your solution just transient? Meaning, you told the solver to look for a steady state solution but it can't find it since it isn't there in reality?

[ghorrocks]

In support of what Gret-Jan said, the final few lines of the FAQ I quoted state "If you still cannot get the simulation to converge then try running it as a transient simulation."

[Opaque]

Your solution has not converged yet; therefore, you cannot make any conclusions about the physics of the flow.

You must understand why the algorithm "refuses to converge"/"unable to converge", and address the issue once known.

When residuals do not continue decreasing, the advice is to output an intermediate results file with Output Equation Residuals = All.

In CFD-Post, locate where the maximum residual is, and study the specifics of your problem. Poor mesh in that region? Recirculating flow and not good enough mesh to resolve it? Region at a domain decomposition boundary (if running in parallel), etc.

[Julian121]

So, the oscillations of total pressure and efficiency are not a sign of rotating stall?
Slightly larger mass flow than this no oscillation take place and the solution converges easily.
In numerical simulation, how can stall inception be identified?
CFD-Post shows that the maximum V-momentum residual lies near to the TE of the stator.
In this location, Mach number reaches zero.
Should the mesh be refined in this area?

[ghorrocks]

Looking at your plots I see the blade has a blunt trailing edge, high residuals nearby and a region of low flow velocity. This all suggests that there is a separation at the trailing edge of the blade which creates a separation and the separation is transient. You might be able to stabilise it using the comments described in the FAQ, but if not then you will have to model this as a transient model.

[Julian121]

I have decided to run a transient simulation to see how total pressure and isentropic efficiency would be in a transient simulation.
I changed the analysis type to transient and defined total time and timesteps.
Then I used the steady results which did not converge as initial solution for the transient simulation.
I had defined five mixing plane interfaces in steady simulation as:
Bellmouth - IGV
IGV - Rotor
Rotor - Stator
Stator - Outlet
Casing Treatment - Rotor
Should I change all these interfaces to transient?
I used the results which had oscillatory residuals at the same mass flow as the initial solution for the transient simulation. Is it ok?

[ghorrocks]

I do not know what you are trying to do so cannot say if it is OK.

[Gert-Jan]

Difficult to say indeed.
But if in the stationary run you had interfaces between rotating and stationary parts with frozen rotor interactions, you should use Transient rotor stator interactions on these interfaces in a transient run.


You can use any intial guess (interpolation) you like. But I would use a stationary result that is as close to your transient case as possilbe.

[Julian121]

I had used a frozen rotor interface between casing treatment and rotor. The other interfaces were stage.
Does a transient rotor stator interface require same pitch angle in both sides?
I have modelled 9 degrees of casing treatment and 9.47 degrees of rotor passage.
Does it require a transient blade row interface instead of transient rotor stator?
Does transient simulation affect pressure ratio of a compressor?
My simulations close to stall operating point did not converge even If I set the convergence criteria to 10e-4 and the residuals were oscillating.

[Gautham A]

Hello Julian121,
Sorry to open up an old thread, but did you get an answer to your own question "Does the simulation close to the stall point necessarily mean that the efficiency and pressure must oscillate?"
I am simulating a steady-state axial flow compressor in ANSYS CFX and near the stalling point mass flow, the pressure ratio and isentropic efficient plots are periodically fluctuating. As far as I know, this indicates that the flow is stalled.
My question is which point should we consider in the periodically fluctuating plot as the pressure ratio or the efficiency value for a particular mass flow. Is it the peak point or is there any other method?
Thank you

[ghorrocks]

The oscillations are a sign that the flow is starting to develop transient instabilities, and for airfoils and turbomachines that generally means the blade is just starting to separate (and the transient behaviour is the small separation bubble moving about). If you go any further you would expect a gross stall. But whether the small separation is enough to cause the overall lift to drop by enough that you can call it the stall point will depend on what you are modelling. Some airfoils stall very quickly, and some take a while for the tiny separation bubble to grow to a gross stall.