Subaru Impreza Pikes Peak

[wesley123]

agreed, those fences are pretty much ultra high df/drag, but it isnt much of a problem as they have power to turn into downforce/drag, opposed to other classes.

[machin]

So, Machin, does your program take into account those kind of dynamics too? (...honestly i fail to imagine how to translate "drifting" into simple algorithms)

Sorry Variante, I somehow missed this question before... the cornering algorithm used in Virtual Stopwatch is empirical in nature and based on a simplified tyre force distribution model; it takes overall information about the car (mass, CG position, front:rear roll stiffness distribution, track width, front and rear downforce, etc, etc) and then calculates the cornering forces at each tyre using this information and a set of "modifiers" that have been derived from actual on-track testing. Each different type of car/tyre/track surface has its own set of "modifiers". Because these "modifiers" are derived from real testing the cornering speed predictions have always proven to be very accurate for well-engineered/set-up cars, even for loose surface "drifting" conditions. Coincidentally, for the loose surface testing we actually used a Subrau Impreza rally car amongst others, although it was the old body shape.

What the program cannot tell you is the effect of small suspension changes such as an extra click of damping, a little extra toe-out etc, and cars which are not well engineered or set-up incorrectly will fall short of the program's predictions, but it is still a big improvement on the algorithms used in programs such as Optimum Lap which is based on "point mass" physics.

In summary the three main points for ensuring Virtual Stopwatch provides accurate simulations are:

1, Real life data.
2, Real life data.
3, Real life data.

[MadMatt]

I will give a sneak preview here of what to expect for the aero of my Pikes Peak models. I am still deep into software settings and parameters so take these with caution. They will probably vary a bit but not that much. Remember, this is at 160[km/h] !

1) Pikes Peak "v1" (the one presented here and on the blog):
- Downforce: 5600[N]
- Drag: 2400[N]
- L/D: -2.33:1
- Balance: 18% front
- CL: -1.700
- CD: 0.729

2) Pikes Peak "v2" (the next iteration that I will present in some time but just to give you an idea, it has lowered ride height by 10mm, longer front splitter, adjusted rear wing angle of attack, revised winglet):
- Downforce: 6200[N]
- Drag: 2300[N]
- L/D: -2.70:1
- Balance: 21.5% front
- CL: -1.815
- CD: 0.673

I can still massively improve on this as I still have lots of flow separation and version 3 is already on its way with vortex generators, sculptured side skirts, revised front winglet, extra front flap and revised rear flaps. I will post pictures and detail as soon as I have found the perfect CFD settings for the simulations, but already wanted to share some values there as it might interest you guys!

[Pierce89]

Sweet!Good job!Just need to add front DF or reduce it at the rear.

[machin]

=D>

Cool. thanks for sharing!

I see that the Peugeot 208 T16 has big vents on its bonnet to direct the air from the front radiators up and over the car; probably aiding front downforce.... however I guess if your model keeps the engine in its original position you won't be able to do this...

[MadMatt]

Cheers guys for the interest and comments!

Yes I will try to balance the car more without reducing the rear which will be a challenge but a good one! Regarding the Peugeot you are 100% right that it helps them to get downforce, but as you said, I will keep the engine at the front and I even put a limitation to myself on the fact that I don't want to cover the vent to the intercooler by a wing. I also put a limitation on not cutting the chassis (except the rear door and boot for the radiator) so it is really interesting because by doing so I can really say that these parts can just be "bolt on" the original WRC.

[MadMatt]

Another entry on my blog, with focus on the 2008 Subaru Impreza WRC and CFD results! Next step will be to present you the same but for the Pikes Peak v1!



>> HERE! <<

What do you think? Is there some CFD data of existing rally cars so we could compare the values?

[Alonso Fan]

Its awesome!

[PNSD]

For your rear wing, I think your setting angle might be wrong?

The flow is coming in at an angle. The real one a little more high-nose up which would possibly solve your separation problem.

http://image.modified.com/f/features/mo ... r_wing.jpg

[MadMatt]

It is difficult to say from the picture what angle has got the wing. The WRC model is a model that was made by a student using real measurements and blueprints. I agree with you on what you propose to solve the issue, but I would have to measure on a real car how is this angle. Again, the simulation was at 160km/h while rally stages are more in the region of 110-120 (over the season).

I will also not focus on the WRC model because it is not the aim of the project, but I just wanted to have a baseline to see what improvements I could make over it!

Also I cannot even compare the values with something existing unfortunately. It is a real shame that the teams are not even publishing the values of old cars. What would it cost them to do so? Nothing as they are obsolete but it would be nice for engineers or curious people to just get these values.

[andylaurence]

Why are you simulating at 160kph? Are there lots of 160kph corners? The average speed for Loeb was about 145kph, despite clearly doing 250kph in places. That suggests that apex speeds are generally far lower. I'd wager 100kph would be more representative, although I have no data to back that up other than gut feel. 160kph is a lot of speed to be carrying through a corner - I've not cornered that fast all season.

[MadMatt]

Why are you simulating at 160kph? Are there lots of 160kph corners? The average speed for Loeb was about 145kph, despite clearly doing 250kph in places. That suggests that apex speeds are generally far lower. I'd wager 100kph would be more representative, although I have no data to back that up other than gut feel. 160kph is a lot of speed to be carrying through a corner - I've not cornered that fast all season.

That is a good point here Andy. I will explain why 160km/h:

1) As you said, Loeb's time was 145km/h average. I would think that it is possible to go faster, probably 150km/h average, but higher would be more difficult. So in that sense, 160 is too high. But then, I also wanted to have the car competitive in other hill climb races where average speeds are often higher than 160-170km/h.

2) I would say that Pikes Peak hasn't got a large variety of corners. It is either slow hairpin or fast corners, and we know that in slow hairpin the tyres are making more difference than pure aero. So it is not wrong to design the car for high speed corners.

3) I will also not "optimize" the car in the sense "trying to enhance the lift and drag". The aim of my project is to develop a Pikes Peak car, reach my objectives and then assess changes in aero performance depending on yaw and altitude. I also plan if I have time (which I think I will) to optimize the car for yaw condition, and of course that will be at 160km/h but it is not the main objective of the project (which again was to develop a Pikes Peak Subaru Impreza).

So yes, 160km/h might be a bit too high, I should have reduced it to probably 140km/h, but when I wrote the downforce objective, it looked like this: 1000lb @ 100mph which I found nice to present. Round numbers. If I really had to make for real a Pikes Peak car for sure I would adapt different things such as a limit on the downforce, lap time simulation to define aero parameters to achieve, etc.

Hope it helps

[andylaurence]

I have to say I'm not familiar with Pikes Peak other than having watched a bit of the coverage at the weekend. Maybe it is all high speed corners and hairpins, in which case, the speed in the high speed corners might actually be often higher than the average speed up the hill. In that case, my point is invalid. I'm just surprised at the speed, given that I race on circuits that are usually much quicker than road-based events and my highest apex speed all season is about 140kph, despite average speeds of about 130kph. Most apex speeds are 65-110kph.

As far as other hillclimbs go, I assume you're referring to European hillclimbs. Here in the UK, hillclimbs tend to be slow and medium speed corners with only a few high speed corners (Kennel/Crossing at Shelsley Walsh, for example).

Most importantly, I'm not saying you're wrong, just asking why.

[MadMatt]

I have to say I'm not familiar with Pikes Peak other than having watched a bit of the coverage at the weekend. Maybe it is all high speed corners and hairpins, in which case, the speed in the high speed corners might actually be often higher than the average speed up the hill. In that case, my point is invalid. I'm just surprised at the speed, given that I race on circuits that are usually much quicker than road-based events and my highest apex speed all season is about 140kph, despite average speeds of about 130kph. Most apex speeds are 65-110kph.

As far as other hillclimbs go, I assume you're referring to European hillclimbs. Here in the UK, hillclimbs tend to be slow and medium speed corners with only a few high speed corners (Kennel/Crossing at Shelsley Walsh, for example).

Most importantly, I'm not saying you're wrong, just asking why.

Don't worry Andy, I value your input, I am not upset or anything, any feedback I get has got all my attention!

The best for me would be to get the speed trace of Loeb's run. I have asked for it, we will see. You are right about the hillclimb thing, I was (implicitly) talking about European rounds, especially those in the Alps. My home round's average speed is about 182km/h over a 5km racecourse with a lot of high speed corners where 160km/h would be a really good fit.

Again, purely for Pikes Peak this 160km/h value might be too high, but I cannot change that now. I think it is still a good compromise between regular hillclimb races (not in UK as you said ) and Pikes Peak.

[MadMatt]

My project is still going and despite a lot of issues with software license and the university cluster, I was able to proceed to simulations. They will be posted tomorrow as I am now getting the last iterations done on the "revised base version" of Pikes Peak v1.



>> HERE! <<

I had to change slightly my plans of how I developed the car (regarding the different versions created) but I am very happy to say that even the first iteration (presented few weeks ago) was far above the objectives, except for the aero balance (which could be anticipated due to the large rear wing).

Also I can tell you that the balance issue has been completely corrected in a manner that I will explain next week. Stay tuned as it really gets interesting now that I will finally get results from CFD on the Pikes Peak models!

So how do you think I have improved the aero balance to shift it towards the front?