[Jorlin]

hello everyone,

During the numerical simulation of 1.5stage Turbine, I came across a description of a negative pressure value and I hope I can help.
When I simulate this turbine, the inlet is given Total pressure and total temperature. The outlet boundary condition is a fixed mass flow rate, and some cooling fluids also specify mass flow rate and temperature. When I was simulating using the harmonic balance method, I found that the pressure value at the outlet had a very negative value, and because I needed to use this value to perform exponential calculations, the solution failed. I would like to ask how to solve this problem and want to know the reason(But neither RANS nor URANS simulations have this problem)
The formula involved: pToalNGV2Outlet in: (1-(pTotalNGV2Outlet/pTotalNGV1Inlet)^((gammaAVG-1)/gammaAVG) is a negative value
pTtalNGV2Outlet=0 [Pa] + massFlowAve(Total Pressure in Stn Frame)@NGV2 Outlet
Error massage:
+--------------------------------------------------------------------+
The problem was: |
Follower: 2 | INVALID-EXPONENTIATION |
Follower: 2 | |
Follower: 2 | FURTHER INFORMATION |
Follower: 2 | |
Follower: 2 | The problem was encountered in executing the expression for: |
Follower: 2 | dhTotalIsTurbine |
Follower: 2 | The complete expression is: |
Follower: 2 | CpAVG*( mFlowTotalNGV1Inlet*tTotalNGV1Inlet*(1-(pTotalNGV2Outle \|
Follower: 2 | t/pTotalNGV1Inlet)^((gammaAVG-1)/gammaAVG))+mFlowTotalNGV1RIDNI \|
Follower: 2 | nlet*tTotalNGV1RIDNInlet*(1-(pTotalNGV2Outlet/pTotalNGV1RIDNInl \|
Follower: 2 | et)^((gammaAVG-1)/gammaAVG))+mFlowTotalNGV1RODNInlet*tTotalNGV1 \|
Follower: 2 | RODN |
Follower: 2 | The error occurs on sub-expression: |
Follower: 2 | (pTotalNGV2Outlet/pTotalNGV1Inlet)^((gammaAVG-1)/gammaAVG) |
Follower: 2 | |
Follower: 2 | BACKGROUND INFORMATION |
Follower: 2 | |
Follower: 2 | The error was detected at one location. The same problem may be |
Follower: 2 | present at other locations - that has not been investigated. |
Follower: 2 | The following values are for the first location which has the |
Follower: 2 | problem. |
Follower: 2 | |
Follower: 2 | In this expression, these sub-expressions have been calculated: |
Follower: 2 | |
Follower: 2 | Sub-expression = Value |
Follower: 2 | |
Follower: 2 | mFlowTotalNGV1Inlet*tTotalNGV1Inlet = 15587.8 |
Follower: 2 | These values were set before reaching the current expression: |
Follower: 2 | |
Follower: 2 | Name = Expression = Value |
Follower: 2 | |
Follower: 2 | tTotalNGV1TECInlet = massFlowAve(Total Temperatu= 320.922 |
Follower: 2 | tTotalNGV1RimSeal...= massFlowAve(Total Temperatu= 323.294 |
Follower: 2 | tTotalNGV1RailFlo...= massFlowAve(Total Temperatu= 301.310 |
Follower: 2 | tTotalNGV1RODNInlet = massFlowAve(Total Temperatu= 324.617 |
Follower: 2 | tTotalNGV1RIDNInlet = massFlowAve(Total Temperatu= 323.324 |
Follower: 2 | tTotalNGV1Inlet = massFlowAve(Total Temperatu= 330.457 |
Follower: 2 | pTotalNGV2Outlet = -4.275651E+14 |
+--------------------------------------------------------------------+

+--------------------------------------------------------------------+
| ERROR #004100018 has occurred in subroutine FINMES. |
| Message: |
| Fatal overflow in linear solver. |
+--------------------------------------------------------------------+

This is my first time asking. If there is anything unclear in the description, please forgive me and tell me and I will add it in time.
Thanks

[ghorrocks]

There are many problems this could be.

It could be numerical - your simulation is inaccurate, or during convergence some parameter is far enough out that it causes this problem

I cold be real - is this a compressible fluid flow? Have you checked it is not choked flow? Have you checked your boundary conditions are correct and well-posed?

[Jorlin]

hello,

The whole process can be regarded as an incompressible fluid. I also checked the boundary conditions several times. The same situation did not appear in the RANS and URANS simulations, although they were the same boundary conditions. The same for chocked flow.
Is there a way to look at specific numerical issues?

Thanks for your reply

[ghorrocks]

The CFX documentation states: "it is essential to use the smallest number of harmonics at which the solution accuracy is acceptable.". So have you tried increasing the number of harmonics?

[Opaque]

May I ask why you are running a transient simulation with outlet mass flow?

For a turbomachine the outlet mass flow is not a steady value, it fluctuates and it is a function of the motion of the rotor upstream.

In addition, are you using a fixed mass flow, or an expression-based mass flow and adjusting it as you go.

Keep in mind that during convergence, the solution fields are not guaranteed to be physical (it helps if they are).

[Jorlin]

Thanks for this idea, I will try it and give you a feedback.

[Jorlin]

hello，

This setting is based on a turbine test rig in Uni. This may be the real situation, but a fixed outlet mass flow rate value is specified as a boundary condition in the numerical simulation, which will definitely include approximations. So this part should not be a physical problem, because many previous simulations have verified it.
Thanks

[Opaque]

Originally Posted by Jorlin

hello，

This setting is based on a turbine test rig in Uni. This may be the real situation, but a fixed outlet mass flow rate value is specified as a boundary condition in the numerical simulation, which will definitely include approximations. So this part should not be a physical problem, because many previous simulations have verified it.
Thanks

That is your call. Not in my book. Fixed outlet mass flow after a rotating component makes no sense for a transient simulation. For incompressible flow, the pressure wave generated in one-time step has to be let go through the outlet. Fixed mass flow cannot do that.

We can make similar cases for compressible flow.

Summary: using fixed outlet mass flow for unsteady flow after rotating components is not a good practice.

[Jorlin]

got it, I will think about that. Thanks