I was having a look at this camera on board of JB´s McLaren in this last British GP:
Immediately I started to think about the fluctuation of undertray generated downforce and I was wondering if some of the aero forumers here can tell me how do you predict that in the windtunnel and what would common figures be of "delta" DF vs. a mean value.
"You need great passion, because everything you do with great pleasure, you do well." -Juan Manuel Fangio
"I have no idols. I admire work, dedication and competence." -Ayrton Senna
Idea would be to change damper settings to adjust the ride height in the wind tunnel, and do a parameter study to find DF vs. ride height. (considering the change in fuel load as well over the course of race distance)
Mapping those parameters and then set the ride height to a given range. The variation curve might still resemble a normal distribution curve..
maybe I am stating the obvious here..looking forward to more opinions..
n_anirudh wrote:Idea would be to change damper settings to adjust the ride height in the wind tunnel, and do a parameter study to find DF vs. ride height. (considering the change in fuel load as well over the course of race distance)
Mapping those parameters and then set the ride height to a given range. The variation curve might still resemble a normal distribution curve..
maybe I am stating the obvious here..looking forward to more opinions..
being nitpicky...not damper settings ,but ride height ...
but you would only get the aeromap for straight ahead driving ..so straightline and minute yaw angles and roll angles...so it nay get a bit more complicated to retrieve downforce maps under roll ,,yaw and wind influence...
you've tripped over a big problem with aero testing in general here - regardless of scale of testing. typically you articulate the model to a given position and hold there for a certain set time to collect samples of data from the balance and then smooth that data for a result. you do this for multiple heave, pitch, yaw, and roll positions to build an aeromap. there's big assumptions and data reduction involved and this always leads to opinions.
obviously air isn't that well behaved and there's significant dynamic issues when the position of the car is varying. it's difficult to quantify in a test, it's difficult to study in cfd and it's one of the hidden attributes that can make a gain in car design not show up in lap time. one method is to study the derivative of aero loads between two positions - in an ideal car you would drive this to zero for constant aero load regardless of position. pitch sensitivity gets thrown around a lot and that's simply the derivative wrt pitch change, but what about roll or yaw as well. F1 race engineers try to make this a minimal effect by running bare minimum wheel travels in their setups but the tires (tyres) move a lot more than the torsion bars do and you can't change that much.
the tire guru's when pressed will admit the same problem exists with tire data, the nice smooth force vs slip curves are a result of massive data reduction and assumptions.
I guess the skid block helps a bit here even when the car bottoms out their can still pass some air under the car.
the tire guru's when pressed will admit the same problem exists with tire data, the nice smooth force vs slip curves are a result of massive data reduction and assumptions
Shouldn't be there some failure indicator bars on these graphs?
It would give the designers a better idea about the range of the data.
During my research time (high viscous fluid flow) we did that but I must admit that I left them out most time because some test results got worthless by this.
stolenmojo wrote:you've tripped over a big problem with aero testing in general here - regardless of scale of testing. typically you articulate the model to a given position and hold there for a certain set time to collect samples of data from the balance and then smooth that data for a result. you do this for multiple heave, pitch, yaw, and roll positions to build an aeromap. there's big assumptions and data reduction involved and this always leads to opinions.
obviously air isn't that well behaved and there's significant dynamic issues when the position of the car is varying. it's difficult to quantify in a test, it's difficult to study in cfd and it's one of the hidden attributes that can make a gain in car design not show up in lap time. one method is to study the derivative of aero loads between two positions - in an ideal car you would drive this to zero for constant aero load regardless of position. pitch sensitivity gets thrown around a lot and that's simply the derivative wrt pitch change, but what about roll or yaw as well. F1 race engineers try to make this a minimal effect by running bare minimum wheel travels in their setups but the tires (tyres) move a lot more than the torsion bars do and you can't change that much.
the tire guru's when pressed will admit the same problem exists with tire data, the nice smooth force vs slip curves are a result of massive data reduction and assumptions.
in both areas you are defining teststhat refer to static non change situations
assume a position and wait for the values to stabilize..
In reality a race car is anything but static or stable .Andsystem response is highly depending on input ,not only force but also rate of change per time...
i could imagine this is one of the factors leading to the sims not matching reality in some cases.
I recall listening to a Racetech Symposium lecture by RUAG (http://www.ruag.com/en/Aviation/Innovation/Wind_Tunnel) who did some work on dynamic ride height measurements, with a Porsche sportscar IIRC. Like all things with aero, the more you try recreate real life situations, the harder you make life for yourself. Not surprisingly they found the lift curve differed for the same ride height depending curve depending if ride height was raising or lowering, as I recall this was still at low frequencies, the high frequency we saw on the McLaren at Silverstone must produce some very strange aero effects.
May be the next step for F1 teams is to run a full lap aero simulation based on real-time chassis attitudes, much like they do for the chassis on a four\seven post rig.
scarbs wrote:I recall listening to a Racetech Symposium lecture by RUAG (http://www.ruag.com/en/Aviation/Innovation/Wind_Tunnel) who did some work on dynamic ride height measurements, with a Porsche sportscar IIRC. Like all things with aero, the more you try recreate real life situations, the harder you make life for yourself. Not surprisingly they found the lift curve differed for the same ride height depending curve depending if ride height was raising or lowering, as I recall this was still at low frequencies, the high frequency we saw on the McLaren at Silverstone must produce some very strange aero effects.
May be the next step for F1 teams is to run a full lap aero simulation based on real-time chassis attitudes, much like they do for the chassis on a four\seven post rig.
I wonder if those gullwing activity of the Redbull Front wing is intentional ...and creating a flow pattern... Other teams ,namely Mercedes have much stiffer wings .
Last year the Toyotas ha also extreme flexingin their endplates -resulting in some
bars installed to stiffen the assembly a bit.
Thankyou very much scarbs!
I found that mag really interesting and want to reproduce a graph here:
As you can see, there is a substantial difference between static and dynamic windtunnel testing.
I wonder if the next step for for optimum aero performance should require shape patterns that comply with the wind waves generated by the heave and pitch.
"You need great passion, because everything you do with great pleasure, you do well." -Juan Manuel Fangio
"I have no idols. I admire work, dedication and competence." -Ayrton Senna
"Make the suspension adjustable and they will adjust it wrong ......
look what they can do to a carburetor in just a few moments of stupidity with a screwdriver." - Colin Chapman
“Simplicity is the ultimate sophistication.” - Leonardo da Vinci
Interesting read there Scarbs. i'm still skeptical mainly because of the balance and the load cell ranges - to shake a 40-50% model between the balance and the model would create a decent amount of inertial load. to measure the small changes in aero load in that same model requires you to use up the majority of the range of your load cells. it seems to me without putting pencil to paper that you would clip the outputs of the load cells if you shook the model too high in frequency or amplitude.
the wheels off approach means for an open wheeler they are moving the suspension arms up and down as they shake the main body too? and confining them to the interior of the wheels? impressive control system if that's the case.
i certainly don't doubt the reality that static smoothed wind tunnel data isn't the true picture as that graphic shows, i'm just not sure how to go get that answer and how vital that answer would be to a successful car design.
the lotus video looks odd because the front of the undertray seems to contact the belt. maybe that was just an installation verification of the model positions (certainly the wheels weren't spinning very quickly so it wasn't a full 'wind on' run). they pulled out the leica so i doubt they end up with the model way off in position to touch the belt during a run. lotus is a first year team but all the names involved aren't rookies.
I posted the video, just to show how they simulate different RH´s in a wind tunnel.
In my expirience (real Sport/Touring/Gt cars, in 1:1 wind tunnels), you never run any
springs or dampers in the car. Because you would get an ride height change due to downforce. Normally you try to hold one parameter (e.g. RH) in a know position to measure the other parameter(s) (e.g. downforce/split & drag).
Depending on the capacity of the wind tunnel you use, the car is either hold in a position via actuators. Then you measure at one condition, move to the next measure again, etc. you do this for a given ride height/pitch/ roll matrix.
If your tunnel (mainly tunnels without rolling road/turning wheels) don´t hold the car in position, you use ridgid dummy dampers, and change the length of them to simulate different RH`s. That´s a quite time consuming process.
Yaw is, in full scale tunnels, typical simulated, by turning the whole vehilcle including the running belt in the airstream. Depending on your tunnel, you can rotate the car up to 360°, where 90° and 180° positions are used to evaluate aero stability when the car spins at high speed (sports car). You do this to make sure your car does not get airborne when spinning at high speed. (some manufacturers take this test very serious)
In a full scale tunnel, I have never seen the attempt to simulate ocillation of the car. It´s more testing at different quasy steady state conditions.
But maybe some people attempt to do this in scale tunnels - I don´t know.
In the running vehicle you will be able to see the net effects in the push rod loads. But some analysis is needed, to seperate the aero content from the overall load variations.
Last edited by 747heavy on 21 Jul 2010, 16:17, edited 1 time in total.
"Make the suspension adjustable and they will adjust it wrong ......
look what they can do to a carburetor in just a few moments of stupidity with a screwdriver." - Colin Chapman
“Simplicity is the ultimate sophistication.” - Leonardo da Vinci
typical yaw test in a full scale tunnel with a real vehicle.
note: it´s a 5 belt tunnel and you see the actuators which hold the car in place.
5 belt tunnel = 4 belts to drive the wheels & 1 belt to simulate the moving road.
closeup on the actuators:
typical 5 belt full scale tunnel design used for road/race car development
Last edited by 747heavy on 21 Jul 2010, 18:21, edited 1 time in total.
"Make the suspension adjustable and they will adjust it wrong ......
look what they can do to a carburetor in just a few moments of stupidity with a screwdriver." - Colin Chapman
“Simplicity is the ultimate sophistication.” - Leonardo da Vinci
