This BEM vs CFD propeller tutorial pits the two calculations against each other on the same machine and teaches you to interpret their gap.

Start with tutorial no. 1

The BEM vs CFD propeller debate is misframed when it looks for a winner. The two methods don't answer the same question, don't cost the same time, and don't get things wrong in the same way. In Heliciel PRO CFD/BEM they coexist on the same project, and the whole point is precisely to read their gap.

The exercise : record the BEM results of a propeller, run a CFD simulation on it, compare, then investigate the gap instead of just enduring it.

• Prerequisites : a propeller designed and calculated in BEM in Heliciel ;
• Duration : the time of one Phase 1, plus half an hour of reading ;
• What you get out of it : the ability to tell whether a gap is normal, or whether it signals an error.

• Tutorial outline:

1. What each method actually calculates
2. Recording the BEM values before running
3. Running the CFD on the same propeller
4. Reading the gap
5. Which gaps are normal
6. Which gaps are suspect

1: What each method actually calculates

The BEM method splits the blade into radial elements. For each one, it establishes the apparent velocity and angle of attack, looks up the lift and drag coefficients of the corresponding profile in a polar database, and sums the contributions along the blade. It accounts for three-dimensional effects through corrections established experimentally.

CFD knows neither blade elements nor polars. It discretises the fluid into tens of thousands to several million cells depending on the preset (from 15,000 for a boat propeller at Level 1 to 15 million for a wind turbine at Level 7) and solves the equations of motion in each one. Three-dimensional effects are not corrected : they appear, because they are part of the solution.

Heliciel's two windows side by side: on the left the BEM designer and its immediate results, on the right the CFD bench with its field map and flow lines. An overall view : the values are not legible at this size.

2: Recording the BEM values before running

A matter of method : write down the BEM values before knowing the CFD result. It's the only way to avoid reinterpreting after the fact what you thought you knew.

In Heliciel's permanent results panel, record the axial force, the torque about the rotation axis and the shaft power. Also note the complete operating point — flow velocity, rotation speed, fluid, diameter, number of blades — because it is this, not the propeller, that must be identical on both sides.

This last point is comparison mistake number one : comparing a BEM at 3 000 rpm with a CFD at 2 800, and drawing a conclusion about the method. The CFD bench inherits the operating point of the open project ; check it in the Flow tab before meshing.

3: Running the CFD on the same propeller

File > New CFD case (from the current model), preset Free Level 5 — Study at minimum, then Mesh.

Why Level 5 and not less ? Because the comparison only makes sense if the blade is actually resolved by the mesh. At the Express and Trend levels, the blade thickness is of the same order as the cell : the gap you would measure would be that of the mesh, not that of the method.

The Phase 1 convergence monitor, in fact, already makes this comparison for you : it reads the BEM thrust of the blades from the case data and permanently tracks the gap with the value it calculates.

The Logs tab at the start of a calculation, on the marine propeller of tutorial no. 13 : the « BEM targets injected » line shows the monitor receiving the thrust (−1,156 N) and the torque (−253.2 N·m) from the designer, with the 1 % stabilisation threshold it will apply to the calculated forces.

4: Reading the gap

At the end of Phase 1, open the numerical table with the Matrix Me! button and compare line by line with your BEM readings.

You need to know what you're comparing. Heliciel's BEM returns the blade forces ; the CFD returns the forces integrated over all the declared surfaces, hub included. The hub produces no thrust but drags, and it exerts a friction torque. Comparing a BEM blade thrust to a CFD total thrust amounts to subtracting a contribution from only one side.

Second precaution : the CFD separates, for each boundary, the pressure share and the viscous share. A well-designed propeller draws most of its thrust from pressure ; an abnormally large viscous share on the hub or on a wall indicates parasitic friction, not performance.

An extract of the CFD bench results : at the top the BEM reference (Fx −1186.43 N, Mx −804.67 N·m), then the CFD thrust (−794.41 N) and torque (−639.61 N·m) of the blades, the efficiency and the coefficients. The pressure / viscous split, boundary by boundary, is read in the Matrix Me! table, not shown here.

5: Which gaps are normal

A gap between the two methods is not a flaw : it is the consequence of different assumptions. The recurring sources, in the direction they push :

• The BEM polars are established for an infinite-span profile, at a given Reynolds number. On a real blade, each section works at a different Reynolds number, in a flow that is not parallel to its chord ;
• The BEM tip corrections are empirical. The CFD forms the tip vortex, and the resulting loss is not necessarily what a correlation predicts ;
• Numerical diffusion in the CFD, especially with a first-order upwind scheme, smooths gradients and tends to overestimate drag ;
• The turbulence model does not predict the laminar-to-turbulent transition. At low Reynolds numbers, the CFD drag is generally pessimistic.

No order of magnitude is guaranteed. On the real cases of this collection (free rotors, Level 3 and above), the blade thrust gap between Phase 1 and the BEM runs from −4% to −35%, and the torque gap is scattered even more. A gap is something to investigate : your mesh convergence, and a check of the operating points, tell whether it comes from the method or from your case.

6: Which gaps are suspect

Some gaps aren't explained by physics. They signal an error, and you need to learn to recognise them :

• A factor, not a percentage. A thrust that is half what it should be is not a modelling effect : first look for a different operating point on the two sides ;
• A reversed sign. The flow direction differs by case family — toward −X for a propeller, toward +X for a wing — and the rotation axis stays at +X. A force whose sign is surprising is almost always explained by a convention, not by a discovery ;
• A gap that changes with the mesh preset. As long as the result moves when you refine, there is nothing to conclude : you are measuring the mesh ;
• A calculation stopped without convergence. A solver that stops at its iteration limit returns a value, and that value is not a result.

Where to go next : tutorial no. 7 to tell a converged calculation from a finished one, and tutorial no. 18 to bring in a third judge — experimental measurement.

The CFD bench PDF reports for a single wind turbine case (Level 5), MRF Phase 1 at the top and Sliding Phase 2 at the bottom : thrust goes from −228.98 N to −231.21 N, torque from 18.99 to 20.67 N·m. This is not a BEM / CFD graph : it compares one phase with the other, and does not include the BEM reference.

The collection of twenty CFD tutorials

« Previous: Phase 2 with sliding mesh  |  Next: Reading the convergence of a run »

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 (you are here)
7. Reading the convergence of a run
8. Turbulence models
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 accompanies the Heliciel design tutorials, which cover the BEM part: blade design, choice of rotation speed, performance curves. CFD comes after them, to verify and to see.