Sailing in an aquarium

A free surface simulation of the 470 sailing at 5,2 kn in an "aquarium". In CFD, the space your model is placed in is called the domain. This simulation is a test about the effect of the size of the domain on the flow & forces over the model, the "aquarium" representing a minimalistic domain. For perfect accuracy, your domain would ideally be as large as the sea. However, domain size is very costly in computational resources and time, hence the smaller the domain, the more simulations we can run in a given time. This is even more true for a free surface simulation (simulation involving the water-air interface), as the water surface has to be represented with very small elements to capture the effects of wave making and breaking.

The domain size affect the flow in several ways. Waves generated by the boat cannot develop completely before they reach the edge of the domain, and once the water exits from the back of the domain it enters again at the front, and in the end the 470 is sailing in waves by its own creation. Depth is very important - if the water gets too shallow, strange things start to happen, but that would be the case on open sea, too.

One particularity of simulating a hollow, shell-like boat such as the 470, is that when starting at the rest there is water inside the boat. Initially the water surface level is the same (zero meters) in the whole domain. Only when the simulation begins does the software start looking for a model-air-water interface. With the water accelerating to 5,2 kn, it first sloshes against and over the half open transom of the boat. As the flow stabilizes, the inside of the boat is drained.

 

Mikko,

If this were abinitio quantum chemistry you would consider running a lower level of theory outside your domain. I'm not a CFD specialist, but could you run a larger mesh size outside the domain?

Jim.

 

Not really - in CFD the domain is the domain. However, normally the domain would be much, much larger (10x), and you would use different sizes of elements (cells), very small close to the model & much larger further away. Actually, we are doing that here, too, but for the water surface, and up to a certain depth, you need a higher resolution everywhere for the water to behave like water (for this particle based solver).

Also, this is not really a aquarium at all - it only has a solid bottom. The side walls are not solid, there the boundary condition is what is called periodic, which greatly enhances the accuracy for such a small domain. With a solid side wall, the bow wave of the model would bounce back much more, yielding bad results.

 

How is the flow acceleration done? Stepping up sources and sinks? I ask because I was wondering if the solitons that show up at the start are artifacts of the strength steps.

 

No stepping up - it's full speed (2,8 m/s) from the start. It takes about 2s for the flow to establish itself and about 3s before the boat stops bouncing in heave.

 

Thanks!
your CFD is my favorite modern art. I subscribed to your channel.

Have you ever run analysis with negative heel? As in veal heel or a mast canted to windward? I wonder about the tip vortex and induced drag insights.

 

I haven't, no. I would love to do a Moth simulation, if someone would provide me the model.

 

could you put a tanker in your aquarium and calculate the added drag that results from a side force of a sail?

 

Nice simulation Mikko. I particularly like the c/board turbulence that shows at times, presumably from the slab sided section the rules require. Also a bit short and small an area, for hull size, again more down to age of design.

Skyak, you might like this.

If you look at the video link below, you will find near the end of the piece a demonstration of a real canting mast. So at least someone has been playing with this concept!. I sailed in this class a while back, but never got to play with that kind of toy.

http://www.youtube.com/watch?v=uL3TbMTpH4o

 

I would think that many would like to step up and exploit your interest. I have some modeling capabilities but suspect you only want validated class participants. The other question I have is would you do a full air and water simultaneous equilibrium, or do you do fix the boat attitude and do air separate? Current T-foils are out of my league.

My interest is in a solid wing sail that closes off against the hiking wing. Is it possible I could interest you in a model of mine?

 

Somewhere on the web I have found and read the testing of 3 diferent foiling Moths over 100 meters in a proper test tank in the US. Very interesting paper and they included a dummy helmsman, which was found to be 40% of the drag. There was quite a discussion about whether this dummy should be included, but after the results of the runs were done it was agreed by the testers that it had been a good idea. As this was full size and real speeds quite a few points of interest were discovered, It was good work, done I believe in 2009.

The shapes of the hulls, curve of areas, foil configuration etc are all given in the paper. Of course drag/lift and speed data is also given. One of the most enlightening pieces of work I have seen publicly available on the foiling Moth. If I can refind the paper, I'll put up the link.

 

I could yes, I even have a nice tanker model, but the problem is one free surface case runs about 7 days at a reasonable resolution... and I have many more priorities. I will keep in mind. A simple RANS simulation without the free surface (mirror model) could be done, probably the free surface will not change much the ratio of forces.

 

You mean probably the Bill Beaver Chesapeake paper - yes, that's a very nice reference, but I never have found the time to do the CAD model. Someone do it, and I will do the sim for you.

 

Thanks Mikko

 

Mikko, when you get the time, Could you explain a bit about how the program operates. I was wondering how the inflow velocity is maintained. Is there an upstream plane where any freestream velocity defect is repaired. I assume forces are integrated off the hull in this code, so that dynamic forces could be handled directly.

The periodic boundary condition can be done in may ways corresponding to different tessellations. Can you elaborate on the implementation here? The simplest one would be no different than a mirror condition, yes? Does the boundary condition on the sides change near the surface?

Can the boundary conditions be set to let the wave energy exit the system at the boundary and not reenter, yet maintain the mass and momentum continuity needed for force computation?