Decision making Wind Tunnel, CFD or track testing ?

[cerojo]

In lieu of the great discussion that generates aerodynamic implementation on a racecar, I want to ask the 'geniuses' (me included of course) at the forum for their definite approach on the issue of aerodynamic optimization: Wind Tunnel and/or CFD and/or Track testing.

What path do you follow if:

1) The rules allow limited track testing before and during season.
2) In addition to the above, the budget is modest, ie: only 4 hrs of wind tunnel
3) In addition to the above, Your engineering resources are also modest.

I cannot think of more limitations to a somewhat serious racing class than these, but soon the rulers will put us under similar conditions ...EVERYWHERE, so we better be prepared.

Thanks, for your contributions.

[Steve Demeter]

If you are a club racer, accesxs to a full scale tunnel is rare.

CFD is beyond most of our means.

Track testing does return defintiive results for the low budget guy, assuming that the driver is good as a test driver.

F1 guys use CFD and tunnel time, but the ultimate test is performance on the track.

Does anyone think that any of the F1 cars that have not performed well were taken to the track after tunnel and CFD with the expectation that they would not perform?

[greg pizzo]

[FONT=Verdana]Seriously ?? at any level that has a modest budget .. (like anyone reading this forum) .... wind tunnel time is RARE would be an UNDERSTATEMENT [/FONT]
[FONT=Verdana] [/FONT]
[FONT=Verdana]CFD takes an enormous amount of time MODELING the parts (even to get started ... ie. THE CAR .. and is well beyond most people’s abilities first of and secondly TIME availability it would take 100's of hours to get a useful model to start with and THEN the CFD results would be somewhat suspect if the modeling was not 110% accurate .. as in YOU BUILD parts right off the Models ... ask my why I know this .... [/FONT]
[FONT=Verdana] [/FONT]
[FONT=Verdana]USE up some TIRES and get some TRACK time ... THAT will do more to decrease the time it takes you to complete a lap at any given track than ANY of the above .. and tell if a change helped or not, UNLESS you get 40-50 laps a day 4-5 days a week ... the BEST thing you can do is DRIVE the CAR as much as you can .... and that would be the cheapest as well, Unless you have a relative with a full scale wind tunnel on the estate or a CAD modeling genius that works for free and loves racing and has LOTS of free time... AND you have the $$ to get GOOD CFD software and understand how to USE it ... [/FONT]
[FONT=Verdana] [/FONT]
[FONT=Verdana]TRACK time is the most efficient by far ! most people reading this if not all would benefit the most from MORE track time ... [/FONT]
[FONT=Verdana] [/FONT]
[FONT=Verdana]As far as real data there is ALOT you can do with a decent Data acquisition system to get A to B to A comparisons .... and its pretty imperial at that point easy to tell if it make the car faster or NOT [/FONT]

[Brands]

For the club racer seat time is by far the most important factor, but there's no point running around lap after lap without a well planned test schedule. Having help from someone with aero experience is invaluable, and with simple data logging and a stop watch it's possible to quantify gains. There's a lot to be said for 'tin and tape' aero testing and even in the most advanced scale model tunnels this method often opens up new avenues for optimization.

[provamo]

track time trumps everything

most of us do not get enough seat time yet love to tinker with our toys

full scale wind tunnel testing is A LOT OF FUN but none of that counts after going off road and ripping up bodywork

just drive the f*****g car

[Mike Allison]

There is a low cost (relatively speaking) wind tunnel available to those in the South East and others if you don't mind the tow. I no longer have any financial involvement with the particular tunnel so this is an unpaid announcement. Some of you are aware of this tunnel but here's the web site address for anyone wanting to do a comparison.

www.a2wt.com

If you call, ask for Dave Salazar and tell 'em, Mike Allison sent you.

Mike

[JBlock]

To answer the question I need to ask a question or 2 or 3. What do you expect to learn? Do you want to map the car or are you looking to further develop the car (granted there are rules but developing can easily fit within the rules with fit and finish plus some creativity).

No to be a kill joy, but truthfully, there are much greater returns in other departments such a driver development. But if aero is burning concern let me ask, have you exhausted what you can learn from you data system, first?

[cerojo]

To answer the question I need to ask a question or 2 or 3. What do you expect to learn? Do you want to map the car or are you looking to further develop the car (granted there are rules but developing can easily fit within the rules with fit and finish plus some creativity).

JB, Thanks, this is a question in which it is already understood that you WILL use D.A. to help set the car to the optimized parameters of the developed aeromap.

No to be a kill joy, but truthfully, there are much greater returns in other departments such a driver development. But if aero is burning concern let me ask, have you exhausted what you can learn from you data system, first?

Again, the driver is not a concern here, just what is the effective way to get the most out of the aero development...given the cost and other constrains.

[stigwhiteCF24]

For many years I've used a 'hot air tunnel' (patent pending). You get a friend or two to come over to the garage, you have a few Miller Lites (note: it works better with inexpensive beer; if you get something more up-market your results are not as good, and if you get craft or specialty beers it will not work at all because all the smug in the garage ruins the test), and you stand around staring at the car BS'ing about what might work aerodynamically. A couple hours of bench-racing hot air will yield dramatic aerodynamic improvements (real or perceived) even if you don't touch the car. Then, when at the track, DILYSI ...

[Becker Motorsports]

FWIW' go with seat time. The best Aero advice we received when we went fom FV to FC was "set all wings at "O" and work on the mechanical handling set up of the car. This provides insight into suttle changes that can be made to improve whatever might be ailing. Secondly it gives the driver a learning curve also which is also invaluablle. Thirdly most of your fellow racers will be glad to share info with you, and finally there is a no better source of info than this web site. Have a problem ?, ask the question and you get specific feedback!

Regards Ernie

[petewo]

As Kas Kastner wisely said many years ago...." if you want 0.5 seconds work on the car; if you want 5 seconds, work on the driver".

I should know.

Pete, CFF 42

[cerojo]

What I want to talk in this thread is Aerodynamics development and how to go about it the best way you know given the constrains.

None other subject is or was asked.

So please stay on task: aerodynamics thinking.

Thanks.

[Peter Horvath]

The consensus of opinion suggests that we place our racing time and money in places other than the wind tunnel. But wind tunnels have fascinated me since my model rocket days, when I would dangle a rocket on a string in front of a fan to make sure it would point into the wind.

Looking at my March issue of Racecar Engineering, page 21, their wind tunnel testing resulted in a 21.9% increase in downforce with a 3.3% increase in drag on their DJ Firestorm car.

The car appears sleek but frankly not nearly as complex and aero-intensive as an F1 car.

I can't help but wonder if this type of result is possible by doing little more than wing adjustments. I'm sure that our wings have a sweet spot and then a dead spot a fraction of an inch later.

The expense could be justified if you really tightened the testing parameters, and the results were dramatic.

[Fred Michael]

Design using CFD then take your prototypes to the tunnel for verification. Then take the finished product to the race track to validate the information you have received in the design and testing phases.

[ASRF1000]

Fred Michael's answer was right on the spot! There's no getting around it. This is the only true way to develop proper aerodynamics

[Neil_Roberts]

The giant-budget racing operations do all 3: CFD, wind tunnel, and track testing. Each testing method has a different set of strengths and weaknesses, which is why it takes all 3 to cover all of the bases.

In my opinion, the recent surge in CFD utilization is premature. CFD has matured into a high-confidence design tool for smooth shapes in steady state free flight at a high Reynolds number, such as an airliner in cruise flight. Race cars, especially open wheelers, are the exact opposite of that. Toyota F1 spent zillions on both CFD and wind tunnels, and found that the correlation of gains found in CFD to track test results was no better than 50%. That is, half of the CFD gains did not produce positive track test results. That's the major downside of CFD. The upside is that if you believe the results, then you have complete knowledge of the entire flow field that cannot be generated in a practical way in a wind tunnel or track test environment. CFD can also model yawed motion through a curving flow field, simulating driving a corner. Other downsides include the modeling and gridding time required, and the computational resources required for an acceptable rate of development. Swift recently bought a CRAY supercomputer to supplement our rolling road wind tunnel, so we have a foot in both camps.

When computational resources advance to the performance level required to do time-accurate solutions of Navier-Stokes equations on a regular basis, that will produce a giant leap in CFD's usefulness for blunt-body analysis. Wakes are always turbulent, and if those wakes interact with other parts of the car, then a time-accurate solution is the only way to produce high-confidence results. The difference in solution time between time-accurate and Reynolds-Averaged Navier Stokes solutions is roughly a factor of 10. We can do it on our CRAY, but we rarely do because the solution takes days instead of hours. Come on Moore, it's time to step on the gas!

Wind tunnels use real air, and with a moving ground plane and a high enough Reynolds number, correlate well to track test results. Of course wind tunnels have their own weaknesses. Closed-return tunnels always have a higher freestream turbulence level than the race track environment, so if the item that is being tested is particularly sensitive to boundary layer transition, then wind tunnel results can't be trusted. A significant downside to wind tunneling of race cars is that it is difficult to accurately replicate yawed travel through a curving flow field like real world cornering creates. Another downside is that the power required to pump air around is huge and therefore expensive. It costs real money to run an electric motor at 400 horsepower for hours per day, and also to run the cooling tower to remove 400 hp of heat from the air at the same time in order to produce a constant air temp in the tunnel. Since moving ground plane belts have to be sucked down to a platen to prevent the car from lifting them, there is an additional power demand associated with moving the belt and cooling the platen.

Track testing includes all of the real world effects, and that is both its major strength and weakness. It's impossible to hold all variables constant except for the one that you are testing. Driver performance, tires, wind, track temp and grip, air temp and pressure, and the conditions of almost everything on the car are all variables that confound even the most diligent attempts to learn useful things from track testing. If the measurement basis is lap time, then you need to be able to produce a string of consecutive laps within 0.5% all day long in order to reliably detect 0.5% gains. The major upside to track testing is that it can be done during the normal course of operating the car if the data system is complete and accurate enough. That can make track testing a no-cost addition to the normal course of events.

[Larry]

Neil has covered the subject well. The only thing I might add is that four hours of tunnel-time would barely give you enough to establish a baseline, let alone do any real testing. If you aren't using a full size tunnel, your testing is reliant upon the accuracy of your model--meaning you'll need to spend about as much money on a good model as you would on the car.

If you want to play with a tunnel, OK, but don't assume that you'll get anything that is truly relevant to your car unless you're prepared to put in a lot of money, time and some more money.

Larry Oliver
International Racing Products

[Wren]

You've put on a restriction that you don't have enough wind tunnel time to do anything meaningful and you don't have the engineering resources to do good CFD and you have a driver that is not a concern. Unsurprisingly, it looks like the answer to your question is to do on-track testing.

You didn't even mention the outrageous amount of time, effort, and money it takes to develop the parts to do aero development.

So, I would use your engineering resources to develop parts, use your wind tunnel budget to actually build the parts (although ~$4k doesn't go far), and do on track testing to figure out whether or not they work.

[rperry]

FWIW, I do CFD for a living so consider that potential bias when you read this post ;-)

Toyota F1 spent zillions on both CFD and wind tunnels, and found that the correlation of gains found in CFD to track test results was no better than 50%. That is, half of the CFD gains did not produce positive track test results. That's the major downside of CFD.

Toyota are also looking at incredibly small gains/details that I doubt the OP is looking at. If you change the placement of a side mirror by 10mm of course CFD will have a hard time capturing it, but I doubt that's the kind of aero development we're looking at in this case. If you swap out a wing profile, change the suspension, or make other "big" changes like that, inexpensive CFD can and will capture it.

Wakes are always turbulent, and if those wakes interact with other parts of the car, then a time-accurate solution is the only way to produce high-confidence results. The difference in solution time between time-accurate and Reynolds-Averaged Navier Stokes solutions is roughly a factor of 10. We can do it on our CRAY, but we rarely do because the solution takes days instead of hours. Come on Moore, it's time to step on the gas!

You need LES to capture the side mirror or tire wakes perfectly, but you don't need LES to capture the forces on properly designed wings accurately enough to see trends/deltas. After all, in my experience most CFD work is done with RANS anyways, not LES. Obviously CFD methods change to suit the problem, but I don't think the thread starter is sitting with the Boeing commercial airplane/F1/Lockheed Martin crowd. A 15M cell full car half model with RANS and wall functions should get him what he wants. And that's quite easy to do; with clean CAD I can mesh and solve that in one day.

Wind tunnels use real air, and with a moving ground plane and a high enough Reynolds number, correlate well to track test results.

Chances are any wind tunnel within reach of the thread creator is going to be totally "unreal."

If the measurement basis is lap time, then you need to be able to produce a string of consecutive laps within 0.5% all day long in order to reliably detect 0.5% gains. The major upside to track testing is that it can be done during the normal course of operating the car if the data system is complete and accurate enough. That can make track testing a no-cost addition to the normal course of events.

I think track testing and CFD are the way to go, though I think CFD is the most affordable. Building and testing a one off wing on the track would cost a lot more than just running it in CFD, IMO.

[Larry]

All tracks are different. In fact, different conditions exist for the same track, too. I would expect that a wing design would be optimized for best results at a particular track and under specific conditions. Therefore, changing tracks (or significantly changing conditions) would call for changing the wing if you want to truly optimize. There is no "best" wing to fit all cars/tracks/circumstances. Each wing has deficiencies in a particular realm, and is therefore a compromise.

Having said that, I don't see many folks changing wings. In fact, I don't see many even changing the angle of attack. I have seen various wings referred to as "high-speed" or "low drag" without the purchaser having any idea what the context of the term means.

For the engineering types, many of these things would be a lot of fun to play with, but as such, data is useless unless it is turned into information which can be used to improve performance.

Larry Oliver
International Racing Products

[swiftdrivr]

I was wondering if the tail on my DB-1 was sagging down so far it was actually causing lift. I am planning to use tuft-testing as suggested in Fred Pughs book, but it will only tell me something if the tufts lift off the top surface [And my wife can photograph them doing it.] What would be more useful would be some sort of pressure manometer device like cells that could be taped in places on both upper and lower surfaces. I have never heard of anything like this. A wind tunnel could tell me, but buying a new body part and back-to-back testing would be cheaper. You might re-visit the simple stuff like these old methods. Crude, but if you are looking for big tuning as opposed to fine tuning, it might help.---Jim

[BILL LOMENICK]

Jim,
I use a presure sensor sold by www.jameco.com number 196103. I calibrate it by making a manometer out of clear plastic tubing and water. I use them for air speed and pressure in the air box.

Bill Lomenick

[Evl]

There are a lot of inexpensive video options that could probably be useful to monitor low-tech indicators.

[AJWALKER]

I spent 20+ years working in an aerospace wind tunnel and flow vizualization was a main stay even with all the load balances and pressure data. My first FF (Citation Z16)had a new nose designed by previous owner. After having extreme overheating problems that could not be resolved I checked the flow field entering the side pod ducts with a thin plate (painted white) hose clamped to the lower arms and some oil paints diluted in gear oil. The flow patterns showed the nose deflected the air to far away from the radiator inlet. Using tuff strips, tuff grids, smoke or surface vis (colered oil, or fluoresent power suspended in oil and viewed with black light) you can get a quick view of what a flow field is doing. I recommend reading the Flow Vis chapters in "Low Speed Wind Tunnel Testing" by Barlow, Pope, Rae (ISBN 0471557749)

ajw
81 Z16
92 SC92

[Neil_Roberts]

Especially compared to wool tufts, oil streaks are very useful because they stay put after the car stops. While a wide variety of fluids can be used for oil flow viz, I have found that used engine oil works well enough and it costs nothing.

The down side to flow viz is that sometimes it takes an experienced eye to interpret the results correctly. Swift offers this service with oil formulations that are tailored specifically for the test environment. We have even done this on commercial airliners. One would expect airliners to be fully optimized down to the last decimal place on everything, but our testing has shown many areas that could be improved.

[AJWALKER]

One of the most useful home formulas for low speed testing is a mixture of baby oil, STP and flouresnet dye (for engine oil) from NAPA for finding leaks with black light. This mixture is put in a small pressure camber and injected through small tubes on the flow path suface when your at set point conditions. This keeps the witness trails from developing before set point conditions. Then you turn of the lights and record the patterns with a open shutter and painting (moving) the surface with black light. This was used very successfully with prop fan blades. This can be done with race cars just make sure no one is behind you during a test session when you release the viz. The show and tell later that night in the garage makes for great bench racing, beer in hand.