How should mesh and time step be chosen for VOF CFD?

VOF resolution is a space-time problem

The transported phase fraction changes over a finite number of cells. If those cells are too coarse, the interface is geometrically under-resolved. If the timestep is too large, the interface can cross too much of the mesh in one update, increasing numerical diffusion, oscillation or conservation error.

Because mesh refinement reduces Δx, it can also make the admissible timestep smaller. Refining the interface without revisiting Δt is therefore incomplete.

Refine the physics that shapes the interface

A uniform fine mesh is rarely necessary. Use local refinement or adaptive methods around the phase interface and important gradients while maintaining a controlled transition to the surrounding mesh.

Use Courant number as a screening metric, not the final criterion

A convective Courant number can be written as Co = UΔt/Δx. For VOF, the relevant velocity should represent local interface transport rather than only a domain-average inlet velocity.

Some solvers impose additional interface-Courant restrictions internally or through subcycling. These controls improve robustness, but they do not remove the need to demonstrate that the engineering result is insensitive to timestep.

Resolve the physical timescale, not only numerical stability

A timestep can satisfy a Courant constraint and still be too coarse to resolve the physical event of interest. Compare Δt with the dominant motion:

• wave period and encounter frequency,
• sloshing natural period,
• rotational or body-motion period,
• jet pulsation or forcing frequency,
• capillary timescale for small structures,
• opening/closing or prescribed-motion timescale.

For periodic signals, use enough samples per cycle to resolve amplitude and phase. If the quantity of interest is peak force or impact pressure, the required temporal resolution can be more restrictive than for mean liquid level.

Initialization and boundaries can dominate early interface behavior

A correct mesh and timestep cannot rescue an inconsistent initial phase field. Initialize the liquid level, hydrostatic pressure and velocity field so they are physically compatible. Sudden startup of flow, gravity or motion can create large artificial transients.

• Check hydrostatic pressure with gravity direction and reference pressure.
• Initialize phase fraction sharply and consistently with geometry.
• Define phase behavior at inlets and backflow boundaries.
• Use damping zones or non-reflecting strategies when waves should leave the domain.
• Ramp difficult motion or forcing when the physical system would not start impulsively.

Verify mesh, timestep and phase conservation together

1. Choose the quantity of interest.

   Examples: interface elevation, mean force, peak pressure, phase volume, frequency or wave amplitude.

2. Refine the interface region.

   Track whether interface thickness and important gradients change.

3. Reduce timestep independently.

   Do not change mesh and timestep simultaneously in the sensitivity comparison.

4. Monitor phase conservation.

   Integrate phase volume or mass and account for all open-boundary phase fluxes.

5. Check signal stationarity or periodicity.

   Exclude startup transients before calculating statistics.

6. Document the limiting resolution.

   Report local interface cell size and timestep, not only total cell count.

Common VOF mesh and timestep mistakes

• Using one global cell size when only the interface region needs high resolution.
• Refining Δx without reducing Δt.
• Using inlet velocity for Courant estimation when interface velocity is larger elsewhere.
• Accepting a stable but visibly smeared interface without sensitivity evidence.
• Checking total mass flow but not phase-specific mass or volume balance.
• Ignoring the physical wave, impact or motion timescale.
• Comparing transient cases before startup effects have decayed.