Sliding Mesh vs MRF for Rotating Machinery in ANSYS Fluent

The decision: sliding mesh or MRF?

You need to simulate a rotating machine—pump, fan, compressor, or turbine—and ANSYS Fluent offers more than one way to model rotation. The wrong choice wastes compute time or misses the physics you care about.

Useful takeaway: choose sliding mesh when you need transient torque, pressure pulsation, or blade-passing effects. Choose Multiple Reference Frame (MRF) when you only need a steady-state flow field and can accept an approximation of rotating effects.

Constraints that shape the choice

Before opening Fluent, fix these constraints:

• Quantity of interest: transient force or torque versus steady pressure rise.
• Compute budget: core-hours available and wall-clock deadline.
• Geometry complexity: simple annular gap versus overlapping or deforming parts.
• Setup time: sliding mesh requires interface definition and a time-step study; MRF is largely a steady-state setup.

Supported options compared

Option	Best for	Relative cost	Key limitation
Sliding mesh	Transient torque, pressure pulsation, startup	High	Fine time steps; interface mesh quality can degrade
MRF (Multiple Reference Frame)	Steady-state flow estimation, design screening	Low	Cannot capture unsteady blade-passing or startup
Overset (chimera)	Complex relative motion, overlapping geometry	Medium-high	Mesh generation overhead; interpolation errors at overlap

MRF solves each rotating zone as a steady frame with added source terms in the momentum equations. Sliding mesh moves the rotor mesh relative to the stator and exchanges fluxes across a sliding interface every time step. Overset uses overlapping meshes with interpolation between donor and acceptor cells.

Trade-offs in practice

Sliding mesh gives the highest physical fidelity for unsteady rotating machinery, but the cost is typically several times an MRF run. Planning heuristics suggest roughly 3–5× the cost for sliding mesh versus MRF, with MRF error on the order of 10–20% for transient phenomena. Treat those figures as rough guidance, not guarantees for your case.

MRF is attractive for early design because it reaches a steady solution quickly. Its blind spot is anything time-dependent: blade passing, vortex shedding, startup, or shutdown. If your design decision depends on peak pressure pulsation, MRF is not sufficient.

Overset mesh fits cases where parts move in complex ways, such as a valve rotating into a seat, but interpolation between meshes can smear sharp gradients, and generating the overset connectivity takes more effort than a single sliding interface.

Concrete implementation: centrifugal pump impeller

Scenario: model a centrifugal pump impeller rotating inside a volute. You need impeller torque and the pressure pulse at the volute cutwater.

Where to run: ANSYS Fluent, either the graphical user interface or the text user interface (TUI). You need write permission to the project folder where the case (.cas) and data (.dat) files live. Replace items in brackets with your own names.

1. Create two cell zones: one for the impeller ([impeller_zone]) and one for the volute ([volute_zone]).
2. Define a sliding interface between them. In Fluent, this is typically an interior boundary where the mesh on each side can rotate relative to the other.
3. Set the impeller zone to a rotating reference frame with angular velocity [omega] rad/s.
4. Choose a time step [dt] so the impeller rotates only a small fraction of a blade passage per step. A common starting check is to keep the moving-mesh Courant number below 1 near the interface.
5. Enable a transient solution with second-order implicit or second-order backward Euler time integration.
6. Initialize from a steady MRF solution if available, then run the sliding-mesh transient.

Example TUI-style placeholder (do not assume this exact syntax matches your Fluent version):

/define/boundary-conditions/zone-interior [sliding_interface] yes
/solve/set/time-step [dt]
/solve/transient ...

Risk: if [dt] is too large, the sliding interface may fail to converge or the solution may diverge. Reduce the time step and monitor the Courant number.

Expected check: torque on the impeller should oscillate at the blade-passing frequency once the transient is established. If torque remains flat, the setup may be behaving like a steady calculation or the time step may be too coarse to resolve the unsteadiness.

Validation and practical checks

Do not trust the first converged result. Verify with these steps:

• Mesh independence: refine the layers adjacent to the sliding interface and confirm the quantity of interest—impeller torque coefficient or volute pressure—changes by less than your chosen tolerance, commonly 1–2%.
• Reference comparison: run a simple rotor-stator case with an analytical or published reference and compare torque or pressure-rise coefficients.
• Courant number: plot the maximum Courant number in the moving zone each time step. Values consistently above 1 near the sliding interface suggest reducing [dt].
• Periodicity: verify that torque and pressure signals repeat over successive rotor passages after initial transients decay.

Practical way to check the result: export the impeller torque monitor to a CSV file and compute a fast Fourier transform. A clear peak at the blade-passing frequency confirms the sliding mesh is capturing the expected unsteadiness.

Limitations

• Sliding mesh requires small time steps, which increases wall-clock time and storage for transient data files.
• Mesh quality on sliding interfaces can degrade over long simulations; smoothing or local remeshing may be needed.
• MRF assumes the flow is statistically steady in the rotating frame. It cannot predict startup, surge, or blade-passing noise.
• Overset interpolation can introduce mass or momentum imbalances at the overlap region; check global balances if you use it.

Before you run: backup

Creating or editing a case file changes project state. Copy [project].cas and [project].dat to a backup folder before modifying zone definitions or time-stepping. If a run diverges or the interface setup fails, restore the backup rather than debugging an unknown case state.

Result check summary

Run is acceptable when:

• Torque coefficient varies periodically at blade-passing frequency (sliding mesh).
• Maximum Courant number stays below your stability limit, typically near or below 1 in the moving zone.
• Refining the sliding-interface mesh does not meaningfully change torque or pressure-rise coefficient.

If any check fails, adjust [dt], refine the interface mesh, or reconsider whether MRF is sufficient for your decision.