[MVRC] Mantium Virtual Racecar Challenge 2019 (Grand Prix Cars)

[machin]

....and use an optimization algorithm together with CFD to optimize...

There are people on here that probably weren't around at the time to know that when you were a competitor in the old KVRC days you used such a method... it caused quite a stir, me and Julien exchanged some private messages about it... you kind of brought a nuclear bomb to a knife fight. Luckily for everyone else you became the competition owner soon after!

[variante]

Hell, i'm already lagging behind CAEdevice, so i'm not giving away the schematics of my FW

But i can show you how a professional does it:
https://i.imgur.com/OOYPUPI.jpg
(taken from a SimScale webinar).

The way i make my wings is not too different. I find it important to decide a guide line for the low pressure side. It's an arc of circumference in the image here, which i also use for lower DF (and lower upwash) applications; for high DF, i use an arc of ellipse (as Machin essentially noticed).

The aerofoils arrangement, thickness and style from jjn9128 are extremely similar to mine.
CAEdevice's wing shape reminds me of a diffuser!

So the underside of the wing follows an arc (from a circle)? I would never ever start like that. Why would this arc be the best geometry? Do they stretch the geometry after they are done? To me it would more sense to create something like an ellipse and use an optimization algorithm together with CFD to optimize the two parameters of said ellipse. Circles are cool, but they are just a random shape.

Why the webinar guy uses a circle? I don't know...my guess is that he already knows that's the right shape for a given application. After all, some famous aerofoils use circular shapes (or nearly so).

Why exactly a circle? Correct me if i'm wrong, but since we're actually dealing with a huge number of complex shapes in 3D space, a full blown parametric optimization might take ages before converging to a result with residuals lower than the performance difference between a perfect circle and an almost perfect circle. So, in practice, a perfect circle might be a totally acceptable solution for certain applications.

Why i (sometimes) use a perfect circle? Because i find it more time efficient to optimize other shapes rather than fine tuning the ellipse. Basically, the same as point 2.

[jjn9128]

That's Nic Perrin. If you download the 2017 perrin from Onshape you can see the whole aerofoil configuration. Afaik it's not how f1 teams do it. As it's open source I don't think he'd want to give the very top secrets away so he's simplified the process a bit.

[variante]

Ah for sure it's a simplified process (there are missing parameters as well), but not necessarily wrong. It'd be very unfair to purposefully show a wrong process...
BTW Perrinn worked in F1 and other series, as a race engineer and aerodynamicist too. He's not Newey, but i guess he knows the basics.

I think that the arc -of whatever function you think is the best- is a legit and effective way to control the general shape of the wing. Some MVRC participants would benefit from implementing such a technique.

[CAEdevice]

I think you are evaluating things as more complicated or mysterious than they are in reality: the arc of circumference is only a local approximation of the parabolic function, which corresponds to a constant relative angular increment. It surely makes sense, but is it nothing else than a first attempt design rule.

[jjn9128]

Ah for sure it's a simplified process (there are missing parameters as well), but not necessarily wrong. It'd be very unfair to purposefully show a wrong process...

Without wishing to seem like an ass. Perhaps you have not spoken to many f1 engineers!? They don't give anything away if they can avoid it. Perrin wouldn't want to be the reason the Williams method got public. He'd likely never work in the sport again.

[variante]

You're assuming that there's always a "Williams method", a "Ferrari method",... While that's probably true when you get to a matter of details, on a general scale i think that after decades of aero in F1 all the teams have converged to -substantially- the same approach. You know, every year engineers go from one team to another... So much that the most generic techniques might be considered of public domain.

[CAEdevice]

Are you sure that exists something we can call "a method"?
When a design is so complex (we are not talking about extruded sections, but about 3D shapes), even much greater technical offices than a F1 design team don't have a method.
It seems that an F1 team should have something magical

[rjsa]

Thank you guys for the master class!

[LVDH]

....and use an optimization algorithm together with CFD to optimize...

There are people on here that probably weren't around at the time to know that when you were a competitor in the old KVRC days you used such a method... it caused quite a stir, me and Julien exchanged some private messages about it... you kind of brought a nuclear bomb to a knife fight. Luckily for everyone else you became the competition owner soon after!

Well maybe a laser guided bomb to a knife fight. This method cannot do magic but It worked very well. I never used excessive resources though. All I did was create an effective cfd setup that simply ran much faster than what was used for the official races. The morphing of the geometries was done in Blender, a very cool open source tool available to everyone. Then there was obviously some programming necessary, maybe not everyone would get that part done. The optimization then ran on the workstation, I used at the time. Nothing fancy but I know that some people here run on laptops.


What bothers me so much about using an arc for the wings, is that when you would take a look at an airfoil database, you would not find airfoils using arcs to describe to upper or lower surfaces. So why should an arc be good for a multi-element wing? But it is good to have a consistent basis to work from and gradually improve.

[variante]

I'd be interesting to see what a (qualified) human could achieve VS a fully automated optimization process of full car given a set number of runs
I know that in the future aeroboys will no longer have a job, eventually. But maybe the time hasn't come yet.

What bothers me so much about using an arc for the wings, is that when you would take a look at an airfoil database, you would not find airfoils using arcs to describe to upper or lower surfaces. So why should an arc be good for a multi-element wing? But it is good to have a consistent basis to work from and gradually improve.

Database aerofoils have presumably been hyper-optimized, and they're designed to work alone, without being bothered by rapidly varying surrounding elements. So, that comparison is not totally fair. And yet, some of those aerofoils do happen to have almost circular shapes.

Don't get me wrong, i don't really care about circle arcs. I often use different shapes myself. But it's wrong to aprioristically assume that the circle can't happen to be the most effective shape. And why not, as a practicle tip, we can say that it's hard to make a bad wing using the circle as a guide.

[CAEdevice]

When I first joined KVRC/MVRC I had my own "circle rule": I modeled wings extruding a section made of two arc circles with a round on the leading edge

Now I still use a simplified spline geometry but with 8 parameters for each airfoil.

[machin]

Hey guys, just a bit of a heads up; we’ve noticed that the rules don’t specifically prohibit “floating” parts (parts that are not connected to anything else) but suffice to say that this was not our intention

....If you have floating parts don’t worry; this round you’re OK... but I’ll shortly update the rules... next round warnings/penalties will be applied to anyone with floating parts.

[etsmc]

ooooh its Friday, does that mean we get scrutineering today??

[LVDH]

ooooh its Friday, does that mean we get scrutineering today??

Here you go:
https://mantiumchallenge.com/scrutineer ... zuka-2019/
Tomorrow, we will see some CFD close ups of your barge boards.