ML are fast and very stable. no seeming hiccups. difference is noticeable
1 Like
I don’t really think this is true. As far as I know, everyone is using the same foil analysis code (XFOIL) developed at MIT in the late 90s and put out as open source in 2000. There have been improved wrappers for it, but still the best analytical solver. Do you know something more about this and the software developers are using? If so, I’d love to hear what people are using
I don’t know what software designers are using. I’m going by what a friend of mine told me. He has been involved in foil development in conventional boats as well as foiling boats for a long time. He is regarded as one of the top foil designers in the world, working for an America’s Cup team and SailGP. I’ve worked with him on a few boat projects, a couple of them improving sailboat rudders. During my conversations with him, I understood him to say that relatively new CFD software allowed them to model foils as well as structures much better and faster than they could say 10 or 15 years ago. Maybe he was referring more to the structural modeling rather than the specific aspects of the foil section itself. The two are interrelated. My understanding was the software was what was making it possible. Listening to Adrian Roper talk about designing foils I get the same feeling. I think Adrians comments that I remember regarded being able to run a lot of modeling without having to prototype foils and how much that opened up the design space and the time and effort it saved.
1 Like
I still find surprising that small brands are keeping up and often kicking bu## of way bigger companies in foil design.
I appreciate the time and cost savings of CFD and theorethical knowledge but in many cases i see a missing link in the last phase: on water testing and refining.
Having a guy like KDW,who can both foil and design foils, is probably what makes magic happen or not.
1 Like
The point on KDW’s involvement is interesting. I think as athletes improve they will drive design, but probably not always from behind the computer. Look at how surfing evolved… short boards are not hydrodynamically efficient. They are in fact pretty hard to surf in many ways (unstable, awful in small surf). But pro talent and desired radical turns, barrels, etc drove the design that way. Foils are probably going to continue to get better and more specific based on needs.
I think foil brands are getting better at describing “who this foil is for” in their marketing too, which will help the sport grow. When I started looking at foils as a beginner it was like trying to solve a math problem to choose the right one. Now it’s much easier. North does a great job with their positioning in my opinion
1 Like
We’ve just barely scratched the surface. I fully anticipate new styles of kites/wings coming out. For example, I have ideas for a wing with retractable sections. Imagine having a 4M wing with a retractable section that can get you to 5/6M. More wing to get up, less to rip upwind. Imagine foils that are dynamic. They could change camber, section. span, angle of attack on the fly. People have said my ideas are dumb or impossible. But then someone does it.
For example I asked friends years ago if you could foil and skydive. They brushed it off as stupid. Someone recently set a world record skydive/base jump/wingsuit with a foil mounted to him. He went the farthest distance and was in the air the longest.
I can’t wait to see where it goes. And I already know my life isn’t going to be long enough to see the best of it.
2 Likes
Someone’s going to make a “parawing-specific” foil in the next year
I mean…at what point is he just wearing a glider?
1000%
Will there be a pocket size, electric pump? that would be a parawing killer.
4 Likes
Trust the Swiss to be on it already
1 Like
What about a looped carbon rod to give a parawing a LITTLE structure. Like a pop-up tent
4 Likes
As far as I know, everyone is using the same foil analysis code (XFOIL) developed at MIT in the late 90s and put out as open source in 2000.
I’m an amateur builder/designer, using XFoil. I know multiple brands that use it too.
But I’m using it as a python module and running it 100.000s of times to optimize a single section. That was possible, but very expensive and slow 10 years ago. Optimizing a section takes over one hour for me on a very fast 16 core workstation. And I’m doing 6 or 7 span locations for a front wing.
2 Likes
Have you heard of NeuralFoil? Supposedly pretty close to Xfoil accuracy but considerably faster solve time.
I’d love to play with XFoil (and NeuralFoil) in more of a scripted way like you are, any links or tips that could help fast track getting a tool chain up and going?
1 Like
My guess is: 90% of the innovation is done, now it’s just refining for specific purposes. Half century old Cessnas are good planes and still comprise a good portion of the recreational marker. Newer airplanes are better but 90% of the light airplane innovation was done by 1975, just like foils.
The Lift 120 is the Beechcraft Bonanza, introduced in 1947 and still a wonderful plane.
4 Likes
I feel like the influx from the race and windsurf foils will start to really impact surf foils positively.
Man that is cool, did you see the demo here? I looked at this a few years ago and Xfoil looked like a pain, now these look way more accessible (the functionailty that is, the science is well beyond me)
Hah, was just following the links from NeuralFoil into the Aerosandbox repo, but had not seen that demo, thank!
1 Like
I had heard of it but did not really try running it as part of my optimizer to see how much faster it is. You are right that XFoil creates a lot of trouble that NeuralFoil potentially avoids. XFoil is hard to get running in-process and multi-threaded. I had to mess with fortran code to get the data I needed and recompile the XFoil python module. Probably best to ask Claude to do that for you nowadays. XFoil can sometimes just refuse to converge, my code has a lot of error handling and retrying to deal with that.
The most difficult thing is getting the “cost function” right: knowing what you want to optimize and how to calculate it. That is the really hard part, and NeuralFoil will not help at all with that.
I also had to solve the problem of parametrization. I am using a modified NURBS method. I see NeuralFoil uses Kulfan parametrization, I do not know how well Kulfan works, but it’s important to verify that Kulfan is able to express all relevant Hydrofoil sections: I wasted a lot of time using poor parametrization methods when I started.
I have it working well now, so I did not feel the need to try something different to get similar results.
To get really better results, I think I need to go into full 3D CFD simulations and optimizations.
4 Likes
I think we are a long way from optimization of foil designs. When Kane DeWilde is still hand shaping foils and winning races and Aspect ratios are mostly 16 to 1 and under the sky is the limit at least for performance downwinding, racing and gliding.
Take a look at performance gliders with aspect ratios of 30+, some as high as AR = 50.
Higher aspect ratio means getting on foil easier, staying up on foil easier and keeping up with faster more slopey swells.
2 Likes
1 Like
As others have said, I think the limiting factor at the moment is construction material. Some potential foil designs (like the glider @Beasho posted) just aren’t possible with the materials currently available.
Maybe the more appropriate question is how much better can foils get with the current materials available? Is the dynamic switching from design limitations to materials limitations?
1 Like