Khamsin Virtual Racecar challenge 2013 (CFD model racing)

[astracrazy]

not a bad idea actually i'd pay if it was reasonable

obviously it can't be expensive it has to still be open to those who want to compete (like last year) and has to be reasonable (price to number of races - for example).

I still think some more publicity work could be done to gain some sponsorship or something from the cloud companies or whatever they are called.

If you can get at lease 1 or 2 f1/motorsport websites to at least do an article on it once a month or every race (roughly the same) then your get some exposure to more of the public. its just an idea, but this is something which isn't heard of before so there must be some interest in the right places from somewhere. Your more likely to get some sponsorship if you can say it will be published on a website which has hits of 5000 per day. Just an idea.

i'm gonna do and release my car anyway, but i'd prefer to to be for this competition so i hope a solution is found

[julien.decharentenay]

I agree that last year did not make much in terms of exposure - partly due to Nick and my inexperience in this area. It is a bit of chicken and egg thing. From memory, last year computation cost over the total 3 races was around $700 (~ 3500hours x 0.2$/h) - I did not count exactly but I think the order of magnitude is correct. It could be reduced by running only one CFD model per race as has been suggested.

What do you think of a crowd funding option such as?

x$ -> participation to 1 race
y$ -> participation to 5 races
z$ -> participation to 5 races + t-shirt

[astracrazy]

i don't think there would be any point doing 1 race, it should be 1 price for the whole competition

[RicME85]

that sort of model would work but each tier needs to have the same amount of races IMO.

£X - Championship entry
£Y - Championship entry + merch
£Z - Championship entry + merch + 3D print-out of your car

[julien.decharentenay]

Thanks for the feedback. Let's nail the season concept down so that everyone is happy and is aware of what they are signing for:

To clarify: the season fee will be set to cover computational costs only - no aim to make any profit on the back of participants. I will aim to provide a financial report at the end of each race to make it clear to everyone. I will aim to get the costing right so that there is as little left over as possible and any left over will be redistributed to first (50%), second (35%) and third (15%). Is this ok?

There are a number of aspects to be narrowed down:

Number of races?
Number of simulation per races? 3 as per last year or 1 to reduce cost?
Accuracy of the simulation? Is what was done last year acceptable or do you expect more refined simulation?

As an order of magnitude computational cost is around 12h x $0.22/hour + $0.07 (storage) (~ $2.7) per simulation using last year setting.

[astracrazy]

i don't think 100mph at 3deg of yaw is entirely needed, the difference isn't that huge enough to warrant the extra cost, the 100mph 0 deg figures may as well just be used for it. (i know its not entirely accurate but its keeping the cost down)

any idea where Nick is? Do you need help making some different engines, reg boxed etc?

I think last years sim was acceptable, perhaps some clarification though so we can create slot gaps i.e what size does the gap need to be not to be covered up. if that makes sense

I understand people will want as accurate as possible, but with that will obviously amount to extra cost so there needs to be a balance

So that will be $5.4 per car per race? 5 races = $27. So people pay $30 then that will create some prize fund at the end? just and idea. Roughly £18 isn't it

[cdsavage]

Good to hear that 2014 might still happen. My thoughts:

-The mesh settings from last year were too coarse. Any small details were basically erased by the mesher, and I think that with the mesh resolution used a lot of important vortex effects were missed.

-IIRC all 3 cases, 180 @ 0 yaw, 100 @ 0 yaw and 100 @ 3 yaw were performed with the whole car, which was twice the computation necessary for the 0 yaw cases.

-The three cases from last year were probably more than what is needed, especially given the mesh settings. I would definitely trade at least one of the 3 (maybe 2 of them) in exchange for a finer mesh.

-While it conflicts with the idea of cutting down on the number of cases to run, it might be worth considering different ride heights. It seemed from the '13 results that the figures from the yaw simulations didn't end up affecting the standings, IE they were pretty similar to the 0 yaw cases. Maybe a different ride height case would show some larger variations. If I get around to it I will try a different ride height with my car in Khamsin.

-The regulations were a bit vague last year. Personally I would take the '14 F1 rulebook and copy just the rules that will be enforced in the competition, add any KVRC-specific rules, then distribute that as a standalone rulebook. I felt last year that there were a number of extra FIA rules that my car complied with, but that weren't enforced in the competition.

-I think it was already mentioned at some point, but I think it would be fair for the first submission (maybe one before every race if time allows) to be for judging rules compliance only, allowing time to make changes before the actual simulation. Last year a DQ/time penalty in one of the rounds was disproportionately harsh. Also it would help if there was a more robust method for checking ahead of time if a specific concept will be judged to be legal - I submitted a couple of questions last year through the website but they were never answered.

I see astra has offered to help on reg boxes etc. I'm also happy to help on that stuff, as well as on compiling a rule book. I have already drawn up the reg boxes from the 2014 tech regs, it should be pretty easy to adapt that to something that will work for KVRC.

I think $30-$50 is a reasonable amount to ask for an entry fee. I am happy to commit to that if there are enough competitors.

[astracrazy]

you make some good points, i like this open debate for setting up 2014

Mesh
agreed with cdsavage. If some compromise can be found it would be great and also advice on what will or won't be picked up with the mesh setting. I found last year some slot gaps were covered over (so advice on the minimum gap would be handy for example)

Rules
I'd like to see some clear rules as well. You can't go by the complete fia 14 rules (it would be too much for everyone) so the 14 rule 'boxes' and major rules (like the new nose height) should be used. Easily check because everyone works within the boxes anyway. Then KVRC makes there own rules for within those boxes for what they will check. Like CD said, create your own rule book and accept that beyond that its a free for all - everyone knows where they stand then.

i.e anything goes within the boxes unless stated

Sidepods
I know its come up before. An idea i have is there are say 3 different boxes (with the same area, just different shapes - long and thin etc etc). Within the rules at lease one of those boxes has to fit within the sidepods and not be visible externally. That should allow variation but cut out way too small sidepods

If cdsavage already has the rule boxes i don't mind looking into creating the engine and gearbox. I'd be happy to come up with a design, put it to the forum for suggested changes etc so everyone is happy

[machin]

Maybe a different ride height case would show some larger variations. If I get around to it I will try a different ride height with my car in Khamsin.

Presuming that the championship continues to use my Virtual Stopwatch program for the performance simulation, I thought it is worth pointing out that the program does not take into account varying ride-height throughout the lap. So you could do runs at different ride-heights and then pick the best one, but I personally think it is better that the championship organisers tell the teams what ride height/rake the car's will be set at and you design your cars around that figure.

This is quite realistic in that in real life the minimum ride height is normally dictated by mechanical aspects of the car design, the race track etc, and the aero design team have to work around that; so think of it as good training for real life engineering; working within defined constraints.

[astracrazy]

i say keep the ride height/rake as it is

[variante]

ok, i guess it's my round.

Cost:
_if we talk about 3-7$ per race (or so), i'm in.
_different packages available (as previously suggested)

Race conditions:
_5 races sounds good to me.
_1 scenario per race (looking at last year's data, final standings would be the same considering only one scenario: it would be weird to pay 3 times the price for pretty much the same thing, wouldn't it?)
_standard conditions for everybody in order to reduce the variables involved, thus cars' performances (same rake, same ride height, same wheelbase,...)
_increase accuracy of the simulations (cost wise)

Regulations:
_let's follow actual F1 regulations, with some flexibility. A regulations visualizer would be of great help: we could buid it together, maybe starting from the one made by cdsavage.
_Further ad hoc regulations need to be added: minimum air intakes-outlest area, minimum sidepods dimensions (the best one is to have sections of minimum area at given leghts of the car: allows the greatest design freedom),...

[julien.decharentenay]

Thanks a lot for the reply. I am going away this week-end, but will reply and start working on monday/tuesday.

For information, I am running a "webinar"/Google Hangout on Air on the use of SketchUp for CFD modelling on Wednesday morning (australian time). It is general, not car specific (and is not an advertising for Khamsin either). The webinar will be recorded and available on YouTube.

[astracrazy]

seems to be coming together

Race conditions:
_1 scenario per race (looking at last year's data, final standings would be the same considering only one scenario: it would be weird to pay 3 times the price for pretty much the same thing, wouldn't it?)
_standard conditions for everybody in order to reduce the variables involved, thus cars' performances (same rake, same ride height, same wheelbase,...)
_increase accuracy of the simulations (cost wise)

i think you need 2 scenarios though, you need the 100mph because then that can be used for 0 deg and 3 deg and you need 180mph. Maybe make up the 3rd scenario space with better accuracy

Regulations:
_let's follow actual F1 regulations, with some flexibility. A regulations visualizer would be of great help: we could buid it together, maybe starting from the one made by cdsavage.
_Further ad hoc regulations need to be added: minimum air intakes-outlest area, minimum sidepods dimensions (the best one is to have sections of minimum area at given leghts of the car: allows the greatest design freedom),...

the rules i still think its too much to use the whole 2014 rule book. Its too much to check as well and it will cause probs like last (this) year. I think you need to state the rules which will be used/tested (and where possible build them into the rule boxes that is given to everyone) and beyond that its legal. Obviously pull out the big rules, like nose height and like you say the intake area etc.

I just think we have to accept that there is going to be a greater amount of freedom than in f1, but its a level playing field on where they can be tested.

[machin]

i think you need 2 scenarios though, you need the 100mph because then that can be used for 0 deg and 3 deg and you need 180mph. Maybe make up the 3rd scenario space with better accuracy

Looking back over the Silverstone results there is very little difference in the aero coefficients between the two speeds... about 1% on average... if the cars had flexible wing elements then we might see a bigger difference in the coefficients, but as they don't I think it would be worth dropping the 180mph scenario for increased accuracy on the 100mph one....???

[astracrazy]

ok fair enough

so one scenario 100mph 0 deg could be used for everything but the accuracy can be increased