Khamsin Virtual Racecar Challenge 2015

[julien.decharentenay]

I can confirm 45:55 front to rear weight distribution, So you should be aiming for the same aero balance. I'll get some aero vs lap time charts sorted out over the weekend.

Richard provided charts prior to each race last year. This year we are trying something a little different. The charts will not be released to reduce the gap between those with access to a lot of computational resources and those with a more limited access. Having access to these charts provides a way to optimise the design to the optimum point given by the chart if you have access to enough computational power. Our intention, by not releasing the charts, is to reduce some of this competitive advantage.

The other changes that we are introducing relates to the weight distribution and aero-balance. The aero-balance is a very sensitive factor, which also requires a lot a CFD modelling to get it "just right". We will introduce a correction to the predicted COP to reduce the gap between the predicted COP and optimum COP to reduce the sensitivity of the aero-balance. The correction factor is not yet fully finalised but we are currently leaning towards a 0.5 weighting factor: the COP used in VirtualStopwatch is 0.5 * predicted COP + 0.5 * optimum COP. So there is still a big incentive in getting the COP close to the optimum, but the competitive gain from it is reduced. We are aiming at finalising this next week and incorporate the "KVRC corrected COP" in the OCCFD output for clarity.

[julien.decharentenay]

There is about 6 weeks left until the first race, and I feel that I am running around like a headless chicken. Does anyone feel the same?

[CAEdevice]

I don't agree with the cpu computing issue: it is a CFD challange not a CAD modeling contest.

About the COP correction: to be honest I don't like at all, even if I would accept it.

If a partecipant doesn't have enough computing power (it sounds unusual because with my 2007 old workstation with 8gb ram and an old Q6600 I run the 2014 season and the hardware was not a bottle neck, compared to my design errors), the solution is to provide one free run before each race (a practice session). A sponsor could help or a subscription fee.

Finally consider that providing the OCCFD with the framework is a very big help.

[RicME85]

Think that is a bit short sighted.
Last year and the year before when I had my Q6600 and 6gb RAM it was taking between 8 and 12 hours to run a simulation.
Now that wouldnt be a problem if I had another machine to use but I didnt so if I wanted to run a simulation it had to be when I was out all day (which isnt very often because of my health) or not use my PC for anything else because of slow down.
If I did it overnight that wasnt a problem but mostly wasnt an option as I use my PC for media streaming and transcoding and because I am often awake at night due to pain so watch media streamed from my PC.

Thats just my case. This year I no longer have my Q6600, I have an i3 4130 and 8gb RAM. I havent run a simulation yet so dont know how it will perform, yes the clock speed is better but there are less cores.

I know people who are still using really old Core2Duo's and similarly low performance CPU's so it is probable there are KVRC participants who arent in a position to run simulations or at least multiple runs to refine their model for COP.

Yes KVRC is a CFD competition but its more of a gateway to CFD rather than being a 'hardcore' CFD comp. It is the perfect place for CAD modellers to learn about aero which in turn will improve their CAD skills.

Sponsorship would be nice but as last year proved, its not easy to obtain.
Any company that puts their name to something (especially since the financial crash) wants to see a return on what they put in, what would any potential sponsor get from KVRC? Not a lot, the competition doesnt get coverage so any sponsor tied to KVRC wouldnt get exposure.

I think the KVRC guys are trying to keep the interest of everyone who signs up throughout the competition. Each year there has been a number of people sign up, a certain amount submit a model for the first race and then the number dwindles as they realise their model isnt up to the standard of the leaders. You might take great pleasure from working on your model and trying to extract as much performance from it as possible to close the gap to the front runners but the more casual modellers get bored, lose interest and never return.

This sort of allowance, which is a bit like rubber banding on racing games, could help bring in more participants. The KVRC team could go out to websites like racedepartment & grandprixgames and approach the hobby modders that draw up the new car models for games like rFactor & Grand Prix 4 and entice them to join the competition.

Just my two pence on the matter.

[machin]

About the COP correction: to be honest I don't like at all, even if I would accept it.

I'm of the opinion that the COP correction actually makes the challenge MORE realistic... Not even top F1 teams design their aero kits perfectly... They all "get it about right" in the CAD world and then build in enough adjustment to the aero devices to allow the actual balance to be perfected on the day via on track testing. The "COP correction" is aimed to replicate this on-track adjustment element of real world racing.

[CAEdevice]

About the COP correction: to be honest I don't like at all, even if I would accept it.

I'm of the opinion that the COP correction actually makes the challenge MORE realistic... Not even top F1 teams design their aero kits perfectly... They all "get it about right" in the CAD world and then build in enough adjustment to the aero devices to allow the actual balance to be perfected on the day via on track testing. The "COP correction" is aimed to replicate this on-track adjustment element of real world racing.

Your consideration about realism is interesting and I agree with you.

The thing I would change is the way to balance the car.
What would happen in a real track? Probably a team would work on flaps and wings to balance the car.
We could do the same: add a "virtual negative downforce (lift)" centered to the axle with too much aero load (leaving the same drag for simplicity). It would be quite easy to compute (momentum balance).

The best solution would be to have a practice session (practice run) and then allow to balance the car modifing only the rear wing (it's about what happens with LMP1 cars).

Anyway: it's a game and everyone of the staff is doing his best, so I would accept every decision =D>

[cdsavage]

Your consideration about realism is interesting and I agree with you.

The thing I would change is the way to balance the car.
What would happen in a real track? Probably a team would work on flaps and wings to balance the car.
We could do the same: add a "virtual negative downforce (lift)" centered to the axle with too much aero load (leaving the same drag for simplicity). It would be quite easy to compute (momentum balance).

The best solution would be to have a practice session (practice run) and then allow to balance the car modifing only the rear wing (it's about what happens with LMP1 cars).

Anyway: it's a game and everyone of the staff is doing his best, so I would accept every decision =D>

Maybe I am misinterpreting your suggestion, but isn't this pretty much how things worked in previous years? A car with non-optimal COP would keep the drag value but any excess downforce on the axle which was too highly loaded was ignored, I think this is basically how it worked.

[machin]

What would happen in a real track? Probably a team would work on flaps and wings to balance the car

Personally the way I would do it on a real track is if I had too much front balance I would reduce the front flap incidence a little and increase the rear a little (or add/remove dive planes, gurney flaps, etc, etc)... that would leave overall drag and overall downforce about the same, whilst getting the balance I want. If we assume that is how the "teams" would do the balancing in a KVRC race there is no need for any CFD re-runs, we would simply adjust the COP towards optimum (within reason), whilst keeping CL.A and Cd.A the same.

A car with non-optimal COP would keep the drag value but any excess downforce on the axle which was too highly loaded was ignored, I think this is basically how it worked.

What you have described here is, in effect, what happens in real life when a car goes around a corner... lets imagine that our one car is in fact two cars connected by a short chain between the rear bumper of the lead car and the front bumper of the chase car... if one car has twice the downforce of the other what speed will the pair of cars be able to corner at? They will only be able to corner at the speed of the lower downforce car... if they try and corner at the speed that the car with the higher downforce can achieve the low downforce car will slide off and drag the other car with it. So when you consider a single car the same thing happens... the cornering speed of the car is limited to whichever end has (relative to the mass/tyre balance) the lower downforce level, but on the straights the whole drag still applies.

[machin]

It is also worth bearing in mind that the balance isn't very critical on a high downforce track like the ones that the KVRC2015 will start off at... below I have simulated two cars to illustrate this point. Both cars have the exact same front downforce (denoted by Clf.A), but the top car has the rear perfectly balanced (relative to the mass/tyres) to the fronts. The second, lower, setup has its rear wing at a much higher incidence, this results in much more drag (+25%) for zero gain in cornering speed (see explanation in my post above)... but because drag has very low influence on lap times at the high downforce tracks the "penalty" for having way too much rear wing turns out to be only 0.2 seconds per lap...



i.e. for the earlier rounds at least you don't really need to worry too much about balance; just get as much downforce as possible at both ends of the car.

[CAEdevice]

Thanks to cdsavage and machin: actually the solution was the same of the last year "nihil sub sole novi"

[machin]

I don't want to get into too much of a debate about this but I do want to make it clear; The COP adjustment proposal is a new thing; it is not what we had last year...

If you start with a balanced car and add wing at one end of the car the cornering performance of the car will not increase; it is limited by the end of the car that has the lowest proportion of downforce. In effect you can simply "ignore" the additional downforce you put on at that one end although it is still actually "there", but it is "redundant". What you can't ignore is the additional drag; this is always there too, but this will slow your car. That is a fundamental of vehicle dynamics and is how Virtual Stopwatch has always, and will always, work.

For this year we're proposing that we'll automatically redistribute the downforce that you do have (within reason) so that it is closer to the ideal balance split... In effect that means there is less "redundant" downforce on your car. This is equivalent to the team "tweaking" the balance by playing with flap angles and gurney tabs during test sessions; it means you haven't got to be so accurate when you make your CAD model, it should make the racing closer and more fun as you'll be rewarded for improving your car, and make it less about achieving the best balance.

I hope that is clearer?

[cdsavage]

The track selection for 2015:

Round 1 (2nd April) - Nurburgring
Round 2 (14th May) - Monaco
Round 3 (25th June) - Nurburgring
Round 4 (6th August) - Magny Cours (non-championship)
Round 5 (17th September) - Sao Paulo
Round 6 (5th November) - Sepang

In practice, the first 3 rounds will require a maximum downforce setup as machin mentioned, and the final 3 rounds will require the same medium to low downforce setup (depending on the efficiency that we eventually reach with the new formula). While the downforce/drag/laptime charts won't be provided, we will be providing some information to give a better idea of the setup goal for the final 3 rounds, along with round 4 serving as a non-championship round.

[CAEdevice]

Well done!

Is the first race (2nd April) included in the championship (points giving)?

[cdsavage]

The plan is for the first race to be a normal round. Being a high-downforce track, it's less important to figure out the correct setup - just go for maximum downforce.

[CAEdevice]

The plan is for the first race to be a normal round. Being a high-downforce track, it's less important to figure out the correct setup - just go for maximum downforce.

Ok!

I would expect for the "fast" races tracks like Monza, but if I rember well last year Magny Cours was considered a very low downforce track (not by me ).