There are two basic ways to include rotation in pump, fan, compressor and turbine analyses. The cost difference between them is large; the accuracy difference depends on the problem [1] [2].
Moving reference frame (frozen rotor)
The rotating zone is solved as steady state in a rotating frame. Components do not move relative to one another, and a circumferential average is taken at the interface.
Where it fits: near the design operating point, with circumferentially uniform upstream and downstream flow, and where the quantity you want is an average — head, pressure-flow curve, efficiency.
It is cheap: a steady-state solution, typically a few hours.
Sliding mesh
The rotating zone physically rotates at every time step, and flow across the interface is computed at the true relative position. It is transient.
Where it is required:
- Rotor-stator interaction matters. Near a volute cutwater or guide vanes, blade passing produces a cyclic load fluctuation on the blade. A frozen rotor cannot show this by definition.
- Operation away from the design point. At part load, rotating stall and recirculation appear; these are transient phenomena.
- Vibration, noise or fatigue is the question. The quantity you want is the fluctuation itself.
- Cavitation inception is being investigated; the instantaneous minimum pressure falls well below the mean.
It is expensive: several full revolutions must be solved, typically ten to fifty times the cost of a frozen rotor.
A practical workflow
Produce the whole curve with a frozen rotor first. Then verify only the critical points — the design point and the extreme part-load point — with a sliding mesh. The difference between the two shows over what range the frozen rotor results can be trusted, and that is the valuable part of the report.
Common mistakes
Putting the interface in the wrong place. The rotating zone boundary should be far enough from the blade tip and where the flow has recovered.
Forcing a steady solution that will not converge. If a frozen rotor does not converge, that is usually a sign that the physics is transient, not that the relaxation factors need more work.
Forgetting the limits of a single-passage model. See the note on symmetry and periodicity assumptions.
References
- Menter, F. R. (1994). Two-equation eddy-viscosity turbulence models for engineering applications. AIAA Journal, 32(8), 1598–1605. DOI: 10.2514/3.12149
- International Towing Tank Conference. ITTC Recommended Procedures and Guidelines 7.5-03-02-03: Practical Guidelines for Ship CFD Applications. ITTC.