How should we define the initial shape of the airfoil?

Which foil would you find more interesting?

Why do you find this shape more interesting?

What factors do you think would define good airfoils and why?

What parameters do you think are important for an algorithm? Do you have any advice about optimizing them?

How should we scope learning about evolutionary algorithms whilst coding a program to evolve airfoils? How should we balance the time for each?

Do you foresee any issues in this idea or our project in general?

Do you have any other feedback to give us?

2019/04/12 11:13:56 AM AST

Research some common shapes, while contrasting them with some *very* bad ones.

It seems like it would be more difficult to make work well.

I think that in my experience, immediately trying to use knowledge you find and then experimenting with it works well.

Might end up being over-scoped. On the other hand, also quite doable in a basic form.

2019/04/12 11:15:13 AM AST

Symmetry is beautiful

One that has had a lot of good effort and hard work put into it

Unbiased parameters are good

Maybe try and make the overarching concept more understandable to people unfamiliar with foils

2019/04/12 11:17:26 AM AST

A cool idea is to first do specific, then do random.

It would be interesting to see different shapes' interactions. It could come up with airfoils for different purposes

Initial shape that faces the front because that is where the initial interaction begins.

Maybe start simple with one point of focus in evolving, then add more points at a time.

Maybe not having enough time/ over scoping

The presentation was informative and seemed to flow well.

2019/04/12 11:17:43 AM AST

this question is confusing

colin "overscope" snow

i still don't know what an airfoil is

2019/04/12 11:18:06 AM AST

I think it depends on what your goals are. If you are looking to develop an airfoil that is standard and is more likely to work then the equation is probably what you want to use. You can also use the equation as a baseline and tune your parameters.

2019/04/12 11:18:17 AM AST

I would start with either an existing hydrofoil or an elliptical shape, but it might be nice not just for this, but for some other parameters offering user input.

Honestly I don't really think I would use this and I don't have the need to use it, but for your user base I would consider what is more valuable to them or more common

2019/04/12 11:18:40 AM AST

Starting with a circle would be nice. Something that is actually manufacturable. Random points wouldn't make sense in a real-world context.

Not sure. I don't know much about this topic but from what I've seen, symmetric things tend to be better? At least, symmetric along one of the axes.

How quickly does your algorithm cycle through generations? Is your analysis dynamic or static (as in, are you calculating static drag and stuff, or is the fitness function across time and space?)

2019/04/12 11:19:17 AM AST

With using a set of points, you can use discrete points and create more complex shapes for your initial shape which can be better.

I just think this would look interesting as well as you could make more complex shapes.

2019/04/12 11:19:17 AM AST

Probably a curve, but I don't know enough about this

More parameters to experiment with maybe

2019/04/12 11:19:37 AM AST

use the equation -- especially since random points might overscope your project; start off with preexisting hydrofoil

thoughts on UI/making it visually more intuitive

2019/04/12 11:19:59 AM AST

Having the initial hydrofoil as a set of points would be interesting in the sense that you never know what kind of a shape it can be generated!

I think probably because I can quickly picture how symmetrical airfoils would look in general, but I think I can't come up with a defining/typical asymmetrical airfoils right at the spot.

Being able to float/fly in most conditions?

If you focus like 70% of the time learning about evolutionary algorithms, do you think you will be better at coding a program to evolve airfoils?

2019/04/12 11:20:03 AM AST

stay within the limit of the math, but maybe break out a bit if you have time by doing things asymmetrically

I think all of the factors you defined are important in a good airfoil, I would figure out what you can validate and go from there

sounds like its easy to overscope, start small and move up

2019/04/12 11:20:17 AM AST

A circle would be cool

It would have a more practical application, rather than just being for fun.

Minimal drag and maximal lift, I guess. I don't know a whole lot about them.

This is sort of a vague question that I'm not sure how to answer.

Probably start off with a basic evolutionary algorithm and then adapt it to airfoils along the way.

Unclear at this stage

2019/04/12 11:20:44 AM AST

The point method

I'm unsure, but it seems like a fun design

coefficient of drag and lift seem very important, but you should probably do more research to figure that out

angle of attack seems most important to vary

you could do division of research and then share those notes and help

There could be several issues with doing high tech analysis. Testing small cases first and then building up to more complex analysis seems like the best option

The presentation was very high level as well as the questions, which made it difficult for me to fully answer them. I wish that was less explanation of high level things and more engagement with the audience

2019/04/12 11:21:44 AM AST

Randomized would challenge your program the most. I would be interested to see how your initial random foil, the sawblade shape, evolves into something that can actually be used as a foil.

Good parameter for evaluating: lift/sq footage? If you were using this to build a boat or plane, then you would want to minimize material costs as well.

Focus on the evolutionary algorithms. Building complexity with the airfoil calculations is easier later on than building complexity with the algorithm.

2019/04/12 11:24:33 AM AST

Of the three examples, I like the 3rd example most. I think using a standard looking foil with a thicker leading edge and thin trailing edge would be a good place to start. Either choose that (a nearly optimized example) or the opposite end (a flat plane), as it will make clear what direction you have to move in.

It narrows the number of parameters you have to sweep over, and should simplify analysis. If you finish, then move onto asymmetric foils.

I think angle of attack will be a significant factor, and one worth looking into more. Also, it's worth thinking more about what specific cases you're designing for before trying to identify what factors matter most to you.

For your project, with how long it takes to run aero simulations, I would try to optimize for runtime and simplicity on the simulation end.

I would have a conversation as a team about what your goals are for the project. I know learning about aerodynamics often requires massive amounts of activation energy, so I'd avoid diving toooo deep into that hole.

It's a tough project, but you all seem dedicated, so I have faith. Don't get too sucked into every parameter; pick two or three that matter to you and optimize for those, and then if you have time branch out.

2019/04/12 11:28:28 AM AST

I feel like using the equation would be less lines of code and probably easier to implement and easier to scope, so I would probably start with that and then add more parameters or switch to a graph if there's time.

Symmetrical seems like it would be easier to implement mathematically. In terms of function between symmetric and asymmetric it seems like google thinks the main difference is that asymmetric produces positive lift and symmetric does not--is this a concern in your design?

I think that if you go with the equation method above, you should start with optimizing the leading coefficient because that will have the most effect on the shape of the curve. You should pick which parameters you want to optimize them against--I know next to nothing about airfoils, so I have little advice on which parameters you should use, but no matter what you pick there will be something you can't consider.

I think you should decide as a team which one you want to focus more on and then change your goals based on which one you choose. For example, if you want to make evolutionary algorithms, focus on making those algorithms and then if your airfoil generator doesn't work as well, that's okay. Vice versa, you could focus on making as good an airfoil generator as possible, and put less work into making evolutionary algorithms.

I think this is possibly a project that is overscoped and you may not be able to implement everything you talked about.

2019/04/12 11:28:57 AM AST

I think you should be able to explain what the algorithm is doing at a good level; even if you don't implement an algorithm like that yourself

My biggest concern is relying on the evolutionary algorithm package to do all the interesting computation for you, make sure you have a good understanding of what it is doing/how it works before applying it to your project

2019/04/12 11:30:00 AM AST

Traditional airfoil is not symmetric. Also if you look at the cross-section of a whale's fin (which was the inspiration for designs of air turbine's blades) along with other animal wings, the "airfoil" shape of wings are not perfectly uniform. This is assuming nature knows what she/he/it is doing.

Does the cross-section of the wing change over the length of the wing?