[dariuszkoz]

Greetings,

My model is 3D problem of radial-inflow turbine with 6 manifolds (tubes) connected to it. It is a transient flow.
https://imgur.com/uAy4fsZ
https://imgur.com/6ZVn5Uf

- Rotor mesh motion is defined as radial velocity profile table where velocity increases constantly to 40 000 rpm. Boundary conditions on each manifold are set as mass flow inlet and also set as transient table profile with mass flow rate up to 1.0 kg/s. pressure 2.3 atm and temperature 1100 K
- Transient (first order implicit) /k-e SST/ PISO/ all equations first order except pressure/ non-iterative time advancement.
- fixed time step 1.1e-5 s
- pressure outlet and outlet with pressure 1 atm
- medium: air ideal gas density



Manifolds are opened one after another, and flow through turbine to the outlet.

Alfer ~ 3100 iterations temperature is limited to 5000 K and absolute pressure is also limited to 1.0 Pa


"absolute pressure limited to 1.000000e+00 in 11 cells on zone 30
temperature limited to 1.000000e+00 in 14 cells on zone 32 in domain 1
temperature limited to 1.000000e+00 in 23 cells on zone 30 in domain 1
temperature limited to 5.000000e+03 in 4 cells on zone 30 in domain 1
temperature limited to 1.000000e+00 in 14 cells on zone 32 in domain 1
turbulent viscosity limited to viscosity ratio of 1.000000e+05 in 126 cells "
and then divergence starts:


"# Divergence detected in AMG for pressure correction, temporarily solve with BCGSTAB!

Error: Divergence detected in AMG solver: pressure correction

Error: Divergence detected in AMG solver: pressure correction
Error Object: #f"



I've already tried to set Pressure AMG solver to flexible and AMG correction to selective but it didnt help.
Mesh has ~ 460k elements overall with mostly refined on rotor.
Mesh orto quality was ~0,25
Mesh skewness was ~0,80
Max aspect ratio ~14

Temperature and pressure are limited on rotor, near 4th manifold.
https://imgur.com/nszpOpZ
https://imgur.com/4kp4Yx0
Also, I noticed that even tho I've applied rotor velocity profile the rotors seems to firstly rotate in conter-colckwise directon than stops and starts to rotate clockwise.
Is there a way to monitor radial velocity maybe using momentum history?

I will be mostly appreciated with advices on how to overcome this problem.