This propeller CFD optimisation tutorial runs a complete loop: spotting a flaw, correcting it, measuring the gain.

Start with tutorial no. 1

Propeller CFD optimisation is not a button. It is a loop: you calculate, you look at where the flow is misbehaving, you correct the geometry, you recalculate, you verify that the gain is real. This tutorial runs this loop once, completely.

The division of roles between the two methods is essential here. Heliciel's BEM is the tool that modifies: it recalculates a complete geometry in a fraction of a second, and knows how to search for an optimum by sweeping. CFD is the tool that sees: it shows the separation, the vortex, the dead zone that the BEM can only correct empirically.

• Prerequisites: a propeller already optimised in BEM, and a first CFD run at least at Study level;
• Duration: half a day for a complete loop;
• What you get out of it: a reproducible method, and the discipline that keeps you from mistaking numerical noise for a gain.

• Tutorial outline:

1. Step 1: looking at where the flow is misbehaving
2. Step 2: attributing the flaw to a parameter
3. Step 3: correcting in Heliciel
4. Step 4: recalculating identically
5. Step 5: measuring the gain, or the absence of gain
6. Not optimising a single point

1: Looking at where the flow is misbehaving

With the starting calculation finished, go to the Results tab. Four observations to make in order:

• Surface pressure: look for zones where the difference between the two faces is small. A portion of blade that is not loaded is not working, and its drag is a pure loss;
• Flow lines: look for a separation, that is, a place where the trajectories leave the surface and recirculate;
• cylindrical cut planes at several radii: they give the blade section in its real flow, with its apparent incidence. This is the view that lets you tell whether a section is pitched too high or too low;
• blade tip: look at the intensity of the tip vortex and the extent of the unloaded zone.

Take a capture of each observation with the Capture button: you will need them for the final comparison, and memory plays tricks.

The starting point: a zone where the flow separates, and a portion of blade that is not loaded

2: Attributing the flaw to a parameter

A flaw seen is only useful if you know which parameter controls it. The usual correspondence:

Only one correction at a time. It is tedious, and it is the only way to know what produced the gain: two simultaneous changes can cancel each other out and give the illusion that neither is useful.

Back in Heliciel: the pitch and chord distribution, where the correction is made

3: Correcting in Heliciel

The correction is made in the BEM project, not in the CFD test bench. That is where the chord distribution, twist, per-element profile choice, root radius and the shape of the generator line are found.

The CFD test bench's File > Open Heliciel project the current case originates from menu entry reopens exactly the project the case came from. This precaution is worth taking systematically: nothing is more frustrating than comparing a CFD run to a propeller that is no longer the one that was meshed.

Heliciel also has its own optimum-search tools by sweeping — rotation speed, number of blades, radius, chord distribution. They work in BEM, so in a fraction of a second per trial. The right division of labour is this: search in BEM, verify in CFD. Searching in CFD would mean paying hours of calculation per trial to explore a space the BEM sweeps through in a few seconds.

4: Recalculating identically

Create a new CFD case on the corrected geometry, and above all: identically. Same mesh preset, same solver preset, same operating point, same number of processes.

This is the rule of comparison, and it is easy to break unintentionally. Comparing a run at Study level to a run at Fine study level does not measure your correction: it measures the difference in mesh, which can be of the same order.

Keep the first case rather than overwriting it: File > Save CFD case as… before starting again. A case folder is self-contained, and you will want to come back to it.

Before and after, on the same colour scale: this is the only comparison that has any value

5: Measuring the gain, or the absence of gain

Open the Matrix Me! table of both cases and compare. Three questions, in this order:

1. Does the gain exceed the calculation's uncertainty? If you know, from your mesh convergence study, that the level used gives the forces to within a few percent, a one-percent gain is not a gain: it is noise;
2. Has the targeted flaw disappeared? Take the same views as in step 1, on the same colour scale, and check visually;
3. Has another flaw appeared? Correcting the pitch of one section shifts the load onto its neighbours. This is why the full observation is redone, not just the corrected zone.

A negative result is still a result: note it down. A correction that brings nothing saves you from redoing it on the next project.

The table that settles it: forces before, forces after, deviation — and the calculation's uncertainty alongside

6: Not optimising a single point

A propeller optimised for a single operating point can be mediocre everywhere else. This is a classic trap, and it costs dearly on a real machine that spends most of its time away from its nominal point.

The method: check each candidate at at least two or three points representative of its use — take-off and cruise for an aircraft propeller, nominal speed and gust for a wind turbine, speed range for a marine propeller.

Here too, the division of roles is self-evident. The multi-point sweep is done in BEM, where each point costs a fraction of a second. CFD only validates the retained configurations, and only at the points that really matter. A full sweep in CFD is not a design method: it is a verification method, used the wrong way round.

To anchor an optimisation campaign on an external reference, tutorial no. 18 shows how to calibrate against established data.

The collection of twenty CFD tutorials

« Previous: Aircraft propeller, cruise and static thrust  |  Next: Calibrating on a reference »

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
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 (you are here)
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 side: blade design, rotation speed selection, performance curves. CFD comes after them, to verify and to see.