CFD sail trim optimization

Mikko, you are correct that an observer on the boat feels is the local flow as influenced by the boat, not the apparent wind. But if an observer on the boat measures the wind velocity away from the influence of the wind (perhaps with a laser device or an anemometer on a very long stick), they will be measuring the "apparent" wind. Also I agree with your comment about perturbation velocity and the difference between what is seen by the person at the side of the road versus by the bus driver.

The ability of one boat to "draft" when following close behind due to the influence of the wake can be understood from the sailor's frame of reference. The sailor on the boat just needs to lean over the bow and put their hand in the water. They will observe that when close behind another boat the oncoming water is slower than when away from another boat.

 

Boundary conditions for CFD are actually a difficult topic for nonuniform flow.

Can you be more specific as to how one would set this up? Typically, the inflow boundary conditions have the velocities specified, and the pressure gradient is driven to zero, while the outflow boundary conditions have the pressure specified and the velocity gradients driven to zero. Then there's the question of how to set the turbulence boundary conditions at the inflow boundary. And the initial conditions throughout the volume.

I suppose one way to establish the initial conditions would be to use a technique common to LES simulations, and that is to do a separate simulation of just a body of fluid with periodic boundary conditions, driven by body forces. Once it gets to steady state, a snapshot of one plane would set the boundary conditions and perhaps even the initial conditions for the desired simulation.

That is fine for the case of a wind gradient with no shear, or no shear to the apparent wind due to forward motion. I think that for the forward velocity case, the forward motion can be added to the velocity components from the periodic volume simulation to get the velocity boundary conditions. The pressures would be unchanged.

I'm tempted to say the turbulent kinetic energy would be unchanged, too, but I'm not sure that's right. Since the velocity magnitudes would be different, and the vorticity would be different, owing to the shear in the apparent wind, I would think the production of the turbulent kinetic energy would be quite different, too. But it's the same wind, with just a change in reference, as has been pointed out, so why would the turbulence be different?

Another method might be to literally reproduce the twisted flow wind tunnel, by inserting a cascade of twisted vanes into the model, with uniform flow upstream of the vanes. The grid density around the vanes would not have to be especially high, because the goal would not be to accurately compute the flow around the vanes. Just the averaged effect of them would be needed to twist the flow. The boundary conditions on the vanes themselves might be the slip condition, so as to not build up boundary layers on the vane surfaces that would shed wakes into the flow. But they would shed vorticity into the flow that would be representative of that in a sheared wind.

BTW, it's very interesting to see this kind of thing in action with landyachts. The wheels kick up dust that gets entrained in the trailing vortices. A yacht will lay down a brown trailing vortex that appears to sweep obliquely along the lakebed. It is laid down at the apparent wind angle from the yacht, but is convected with the true wind. It's really fascinating to see the vector relationships in action.

 

Nope. The Protocol prohibited teams from doing any meteorological measurements at the venue. They could only use data from public sources and ACRM.

The wind coming through the Golden Gate was believed to be well mixed, with less gradient than would be found in a fully developed wind on the open ocean, but I have no data to support that belief.