CFD sail trim optimization

Thanks!

Sorry, Mikko: I was inatentive and somehow missed your comments. You have perfectly eхplained the situation and provided all the necessory info in the very beginning of the discussion.

 

The actual values of V in the logarithmic formula strongly depends on the choice of z0. Back in the post #31 I had assumed z0=0.5 m, which corresponds to a choppy sea which quite slows down the air in the layers closest to the water surface. A much smaller value of z0, meaning smooth flat sea surface, would get the values calculated through the logarithmic formula much closer to the data measured in Nick's reference papers.

That's the nature of semi-empirical mathematical expressions. They are based on a rational theory, but need the measured data to form the basis for future calculations.

And, by the way, it also probably means that ORC TWS formula is valid only for the case of a near-flat sea surface. Which makes sense, considering that it is a VPP, which needs standardized and uniform conditions in order to create comparable data for different boats.

 

Thanks, Joakim!

 

According to the linked page, at sea with fetch > 5km it is given as 0.0002. http://www.webmet.com/met_monitoring/663.html . 0.5 would be parkland, forest or suburbia.

I have a formula of z_0 = 5e-5 * (V_T10^2/g) but don't know the reference. (V_T10 is the reference true wind speed at 10m).

 

Very good, thanks! Evidently, I had misinterpreted the values of z0 to be used in the calcs. We are all learning new things today.

 

That is what I'm talking about.

Let's come back to the article http://www.publish.csiro.au/?act=view_file&file_id=PH560511.pdf where you made perfect Exel calculations.

The wind speed was approximately 10m/s, experiment was performed in the ocean 5 miles away from the nearest island. And to get the perfect fit with the logarithmic model we need to take the surface roughness as several mm.

Update. tdem, thanks for the useful link Yes, less than 1 mm for the open sea: "Open sea, fetch at least 5km: 0.0002m"

 

Another good reference on downwind sail CFD. They used a roughness length of 0.25mm, but established by trial and error a value that would give the correct twist and velocity profile at the mast.

WIND TUNNEL AND CFD MODELLING OF PRESSURES ON
DOWNWIND SAILS by Richards and Lasher.

http://bbaa6.mecc.polimi.it/uploads/validati/TR08.pdf

 

So now that we have agreed that the wind gradient is likely 10-20% from mast top to boom the twist of apparent wind for a boat like J/32 is (with 20%)

About 2 degrees close hauled

4-6 degrees at 90 TWA

6-20 degrees at 150 TWA

Since your model also has the hull, the twist will be much higher there.

 

And hence, definitely not negligible.

 

What I've learned

I'll put my head back on the block and attempt to summarize what I've learned from all of this excellent discussion.

In setting up a 3D CFD simulation there seem to be a couple of approaches:

1) You can essentially model a full scale wind tunnel. Since the boat is stationary you have to adjust the boundary conditions to get the apparent wind vertical profile for a moving boat in a wind field. This sounds like the approach most people are taking. This requires that you create a different wind field for each combination of boat speed and wind speed/direction, but that is easy to do.

2) You can attempt to model a full scale boat actually sailing, where you create the wind field gradient within the CFD model itself. Now you have to have the boat moving relative to both the wind and the water. This was appealing to me as the simulation appears closer to the physical problem. However, the vertical velocity profile using flat simulated 'water' does not match the data people have presented here (the simulated velocity gradient with flat 'water' resolves almost all the velocity gradient in the bottom two meters of air, so the terms essentially vanish.) This might be fixed by adding simulated waves to the water, but that remains to be seen.

I'm going sailing tomorrow. I might just measure the wind speed at a couple of elevations while on the hook...

 

If you really want to get the wind gradient from convergence, you need to have several kilometers of domain before the boat and also you may need to have a very high domain. So it is really not worth it. Another option is to use periodic boundary conditions (like having several boats in a row). Then you could get the wind gradient. You can adjust it by defining the surface roughness.

But I don't see any point in using the option 2) unless you are including unsteady features like boat movement in waves, gusts and acclereration. Then you need to have a full 6 DOF model for the boat. That is commonly done with potential flow programs like Splash, but not so often with full RANS.

If you have a boat moving steadily (constant speed, heading and heel), option 1) is just as accurate and physically correct.

 

Joakim,

I'm sure you are right, but since I'm doing all of this just for my own enjoyment, I think I will pursue option 2 a bit further just to see what it takes to create a more realistic simulation. Have you seen any literature on simulations of the effect of ocean surface conditions on wind gradient?

 

Like Joakim explains, this is really a frame of reference question. The first option, emulating a wind tunnel test, with the pre-constructed apparent wind, happens in the frame of reference of the sailor moving with his boat. If the sailor could see the streamlines around the sails, they would look like in your simulation, bending strongly ahead of and behind the sail.

The second option is in the frame of reference of an outsider standing ashore. I agree that this is more like the real world, after all apparent wind is just a construct to help calculate the forces. But this option will require moving boundaries, transient simulation and dynamic meshes - much more complex to set up and much more demanding in computational power.

While the forces will be exactly the same in both cases, the flow pattern will differ. The wind will bend much less as the boat slips through the true wind field, affecting mostly just the wind speed, less its direction. In this respect, most all textbook illustrations (for instance Gentry's well known streamline plots) are erroneous - wind really does not bend that much. See the video & attachment (the video resolution is not that good, but see the difference in the still pic).



The only time you really need to model the real wind and boat's motion through it is when there are several boats moving in different direction or at different speed, and you want to study their interaction. You only can construct the apparent wind for one boat at the time - the apparent wind for another boat, maybe on the other tack, would be all wrong.

 

Both frames of reference, the observer on the boat, and the observer on shore or an a platform fixed to the bottom, are equally "real". Which one is more useful depends on the situation, experimental/calculation/computation methods used, and user experience.

Apparent wind is not "just a construct to help calculate the forces". It is the wind which the boat and an observer on the boat sees and which determines the aerodynamic forces acting on the sails and other parts of the boat. Apparent wind is as real as the wind in any other frame of reference. An observer on the boat can only know what the wind is relative to the ground/water by calculating it using information about the velocity of the boat.

Frequently "perturbation velocity" is used with the frame of reference fixed to the boat. Perturbation velocity is the difference between the local air velocity and the undisturbed freestream air velocity. The use of pertubation velocity can lead to erroneous conclusions about the nature of the flow. For example the perturbation velocity may be pointed upstream which is interpreted as the air molecules are moving upstream when the air molecules are fact flowing downstream but slower than freestream. A classic example are the illustrations of the perturbation velocity around a lifting airfoil which appear to show the flow on the lower (pressure) side of the airfoil flowing upstream.

 

I'm sure you're right, as usual David. But I insist a little:

- Actually, what the sailor feels on board is the "local flow", the "apparent wind" being influenced by the freeboards, cabin top, sails etc. If the boat could feel something maybe she could feel the apparent wind, or the sum of its own velocity and the true wind velocity (the vector sum of 2 "real things", sounds a little of a mathematical construct to me ;-).

-To me the reference frame of the outside observer, fixed to the bottom of the sea, or the earth, is more objective, or "real", as long as we are living on this planet. The reference frame of the sailor is somewhat subjective: If you take my example of two boats beating to windward on opposite tacks, each will have their own, different apparent wind while the true wind remains the same for both.

- About the perturbation velocity (which I've erroneously called circulation in here): In the earth frame of reference, if the foil is moving and the fluid is still (like a keel through the water), particles are actually moving upstream with the perturbation velocity. This is one reason why I think the earth frame of reference better describes reality.

- An example about perturbation velocity: You are standing on a bus stop and a bus drives by without stopping. Just before the front end of the bus reaches you, you feel a (perturbation ;-?) wind in your face and you are pushed away from the driveway. About at the level of the rear tyres you are sucked in towards the bus and the driveway (scary, the reason mothers tell their children not to stand too close), and you feel the wind in your neck - the air that has been pushed away in front of the bus wants to get back to fill in the void behind the bus. So the perturbation is pushing the air particles that actually move back ond forth? In the bus driver's frame of reference, the air seems to be pushed aside, yes, but nowhere in front/forwards of the bus, as you say in your sailboat example. The streamlines around the bus, as drawn in the bus driver's frame of reference, don't suggest this at all. I still remember how this puzzled me a long time as a young student, waiting for the bus on my way to the Otaniemi Technical University, so keen to learn aerodynamics.

- Another example: For decades, I was equally puzzled about how is it that one can sail right behind in the wake another sailboat, only a boatlength or less away from its stern? This happens all the time, when boats are rounding the leeward mark and proceed upwind in a row. Looking at the streamlines from the textbooks (and my simulations), performed in the sailor's frame of reference (the "apparent wind world"), the boat behind should have such a header from the sails of the boat in front, that it could no way point as high as the one ahead. Yet we sailed all the time for extended periods on time in that position. If you look at the simulation in the earth frame, you can see that the wind is actually headed very little in the line behind the stern of the forward boat. Yet. there is a header and the wind force is down - how can the boat behind hang on in there at all, then? The explanation comes from the underwater perturbation - the boat ahead is dragging the water along in its wake, so that you experience less underwater drag in the boat behind. Just like driving close behind a truck, saving fuel. See the attachment about the Star seen from underwater - the perturbation in the wake behind near the surface is 1-2 kn. In the sailor's frame of reference, you completely miss this effect.