This propeller CFD turbulence tutorial explains the automatic model choice and shows how to see the calculated turbulence.

Start with tutorial no. 1

Heliciel PRO CFD/BEM does not ask you to choose a propeller CFD turbulence model. It picks one according to the case family: Spalart-Allmaras for a wing, k-ω SST for a rotor. This tutorial explains why this choice is not arbitrary, and shows how to look at the turbulence field the solver produces.

The experiment: calculate an airfoil twice — once as a wing, once as a rotor blade — and observe that the model chosen differs, along with what it changes in the result.

• Prerequisites: an airfoil available in Heliciel's profile database;
• Duration: two Phase 1 calculations, plus reading;
• What you get out of it: the ability to know when the model limits your result.

• Tutorial outline:

1. Why turbulence must be modelled
2. Spalart-Allmaras, the wing model
3. k-ω SST, the rotor model
4. Seeing the turbulence field
5. Inlet turbulence conditions
6. What the model cannot tell you

1: Why turbulence must be modelled

A turbulent flow contains eddies of every size, from a metre down to a tenth of a millimetre. Resolving all of them would require a mesh and a time step out of reach for a real propeller: that is direct numerical simulation, reserved for research on elementary geometries.

The approach used in engineering is to calculate only the mean flow, and to represent the effect of the eddies with a model. This model adds a so-called turbulent viscosity, far higher than the fluid's viscosity, which captures the mixing produced by the agitation. The whole difficulty lies in calculating it correctly where it matters: near walls, and in separated zones.

Both of Heliciel PRO CFD/BEM's models fall under this approach. They do not differ in their ambition but in what they prioritise.

What the solver calculates and what it models: the mean flow on one side, the agitation on the other

2: Spalart-Allmaras, the wing model

For a wing or airfoil case, Heliciel uses the Spalart-Allmaras model. It transports a single quantity, a modified turbulent viscosity, with three mechanisms: turbulence is produced by shear, destroyed near walls, and diffused through space.

Its strengths are those it was designed for: it is robust, economical, and built for external aerodynamics — precisely the problem of a wing airfoil in a free-stream flow.

Its limits are known: it handles strongly separated flows less well, and it does not predict the transition between laminar and turbulent regimes. On a low-Reynolds-number airfoil, where a sizeable part of the chord stays laminar, it therefore tends to overestimate drag.

3: k-ω SST, the rotor model

For a propeller or rotor case, in Phase 1 as in Phase 2, Heliciel uses k-ω SST. It transports two quantities: the turbulent kinetic energy and its dissipation rate. Its distinctive feature is to switch from one formulation to the other depending on the distance to the wall, which gives it the accuracy of one model near surfaces and the stability of the other farther away.

On a rotating blade, three arguments point to it:

• it better predicts behaviour under adverse pressure gradient, a permanent condition on a blade's suction side;
• it better resolves boundary-layer separation, which the one-equation model underestimates;
• it is the industry standard for turbomachinery and rotors, which makes your results comparable to those in the literature.

It costs roughly 20 % more in computation time, and it is more sensitive to the turbulence conditions imposed at the domain inlet — a point covered below.

The choice table: one equation against two, and what the second buys you on a rotor

4: Seeing the turbulence field

Turbulence is not just a setting: it is a field you can look at. Results tab:

• Cuts frame: choose a Vertical cut, then the Turbulence content. A slice of the domain takes on colour according to the turbulent intensity;
• Iso-surfaces frame: check Turbulence and activate one or more volumes. You get equal-value surfaces, which trace the wake and the tip vortices in volume.

What you see then is instructive: turbulence is weak in the upstream flow, strong in the blades' boundary layer, and it fills the wake. On a well-designed propeller, this wake is a clean helical sheet; a turbulent zone that suddenly thickens on the suction side signals a separation.

Cylindrical cut of turbulent viscosity on a 3-blade rotor: the red ring at the rotor plane, the turbulent wake downstream, and the calm flow upstream (cyan).

5: Inlet turbulence conditions

The two-equation model needs to be told at what level of turbulence the fluid enters the domain. This inlet value is not trivial: set too high, it drowns the flow in artificial turbulent viscosity and smooths out everything you were trying to see.

Heliciel sets these conditions according to the configuration. In a closed duct in particular, where the inlet is close to the rotor, the setting is established from a controlled turbulent-viscosity ratio rather than an arbitrary intensity: excess inlet viscosity would lower the effective Reynolds number to the point of completely distorting the field.

You can check the conditions actually written: they appear in the case folder, in the initial field files, and Heliciel logs them when the solver starts. Tutorial no. 2 shows where to read them.

6: What the model cannot tell you

Three honest points to keep in mind when using a result:

• No transition. Neither model predicts the shift from laminar to turbulent. On a low-Reynolds airfoil — drone, model, small propeller — the calculated drag is pessimistic;
• No instantaneous structures. The averaged approach produces a statistical flow, not a movie of the eddies. A vortex shedding event is represented in its mean effects, not in its detail;
• Beyond stall, it is qualitative. Past roughly fifteen to twenty degrees of angle of attack, the results indicate a trend, not a value. For a stalled airfoil, the polar from Heliciel's profile database or a Phase 2 calculation is worth more than a steady-state calculation.

The model interacts directly with the mesh: it cannot do better than what the near-wall cells allow. That is the subject of tutorial no. 9.

The model chosen, announced in the logs at startup: nothing is hidden, nothing needs to be set

The collection of twenty CFD tutorials

« Previous: Reading the convergence of a run  |  Next: Boundary layer and y+ »

1. First CFD propeller simulation
2. Reading the OpenFOAM case folder
3. Choosing the mesh preset
4. The MRF zone and Phase 1
5. Phase 2 with sliding mesh
6. Comparing BEM and CFD
7. Reading the convergence of a run
8. Turbulence models (you are here)
9. Boundary layer and y+
10. Reading a pressure map
11. Where the forces come from
12. Checking an airfoil polar
13. Marine propeller and cavitation
14. Wind turbine and tidal turbine
15. Fan in a closed duct
16. Aircraft propeller, cruise and static thrust
17. The optimisation loop
18. Calibrating on a reference
19. From the hull to the propeller
20. The deliverable calculation file

This series complements the Heliciel design tutorials, which cover the BEM side: blade design, choosing the operating regime, performance curves. CFD comes after them, to verify and to see.