Learning Forum

[pchoopanya]

Dear all, 

This is my first post here. Please kindly help me with this issue. 

I am doing a simulation of an axial compressor of a gas turbine engine. It is the gas turbine used in a power plant, hence it is huge and high air mass flow rate around 340 kg/s and hence the flow inside is a sonic flow.  

I have chosen the Mixing Plane method to model my problem. Therefore I only need one blade from each stage. There are 6 stages (rotor + stator) hence total number of blade is 12 blades + 1 IGV blade. 

However, since the problem of interest is very similar to the Mixing Plane tutorial provided by ANSYS FLUENT. 

Normally, there exists a small gap between the rotor blade tip and its outer casing (rotor tip clearance). However, such gap does not exist in the geometry provided in the FLUENT tutorial. The rotor blade is basically connected to rotor hub and rotor shroud at its root and tip, respectively.  

In setting the boundary conditions for each of these walls; 

Cell zone condition >>> rotor is defined as one separated zone that rotates (rotating/moving frame of reference) while stator is another separated zone but is fixed in space (inertial frame of reference)

Boundary conditions >>> rotor blade and rotor hub are walls which rotate. So, they are defined as stationary walls (with respect to the rotating frame of reference) This totally makes sense to me. 

The problem arises when setting up the boundary condition for the rotor shroud (or the outer casing). Of course, in the real compressor, this wall is NOT rotating and fixed in space. The tutorial suggests defining it as a moving wall with respect to the ABSOLUTE frame and set the rotational velocity at 0 rpm. This way, the rotor shroud will be fixed in space >>> which totally makes sense to me, again. 

However, thinking about the actual geometry (the 3D cad) given, the rotor blade tip is connected to the rotor shroud..... but in setting up the boundary conditions for them, rotor is rotating, while the shroud (which is connected as the same entity) is not allowed to moved. How does this happen? It does not make any sense to me. Could you kindly clarify this to me? 

 

 

Best regards, 

 

Pat 

 

[Rob]

It's a tutorial, they don't always make sense but are usually trying to teach a specific feature.  

For a wall bounding a rotating zone we can have a speed relative to the zone (typically zero as the surface rotates at the same rate) or as an absolute value (typically zero for a stationary wall). We choose whatever makes most sense for the model. 

[pchoopanya]

Thank you very much for your answer. 

Yes, I doubted it that this might be working because it is only a single stage compressor (1 rotor and 1 stator).

However, when it comes to my very own problem containing 13 rows of blades, convergence is very very very difficult to achieve, if at all possible. My latest result has not shown any sign of convergence yet, especially the continuity residual where it rises to 1e+10 and flatten out there with no sign of coming down. 

Setting the rotor-shroud as fixed with respect to the absolute frame of reference, the method described in the tutorial makes absolute sense to me. >> This is understood as the rotor blade rotates, but its casing does not.  

The only thing I wonder if, can I say that, since the rotor blade actually merges/connects with the rotor shroud, this is not 100% correct? I also notice that the actual rotor-shroud has a hole in the middle due to the existence of the rotor blade tip - which does not really represent an actual rotor shroud.    

In my problem, having a rotor blade connected to the rotor-shroud and setting the rotor-shroud stationary while rotor blade rotating >>> Could this be the reason why I am having difficulty in getting the solution to converge? 

Is it worth going back to the geometries, make room for the blade tip clearance, and re-mesh everything? I have talked to an ANSYS support engineer in my area and he said this was totally fine and instead of going back to the start, I only needed to play around with numerical variables, like, under-relaxation factor, initialisation, deactivate some equations, and then re-activate, start with smaller pressure at the pressure outlet and then increase the value. I have tried all of these and none worked. 

Could you please suggest me what to do or try? I am really confused now. 

 

One more question.....

Using a Spalart-Allmaras model, I could obtain a converged solution at the desire pressure outlet, though the residual value fluctuates in a cyclic pattern and its value is quite high (1e-03). I also checked the mass imbalance by using mass flux through inlet and outlet..... the mass imbalance is significant. 

Reading the manual suggests using the right pair when defining a mixing plane. For this case, I used pressure-outlet (upstream) // pressure-inlet (downstream) 

This is the most robust pair, but does not guarantee mass conservation, is this correct? 

What about using     pressure-outlet (upstream) // mass-flow-inlet (downstream), will this fix the mass imbalance problem and ease the convergence? 

Could you please kindly help me with these? Thank you very much for your time.