The Rise of the Swan Neck

Images copyright Simon McBeath and Michael J. Fuller
Text copyright Michael J. Fuller

This article first ran in Race Car Engineering's January 2012 issue
Audi R15, Sebring WDWT, 2009In 2009 the ACO introduced new rear wing regulations in response to a spate of frightening yaw induced blow over incidents that seemed to increase in frequency during the 2008 season.  At the September 2008 ACO press conference at Silverstone, the ACO's Jean-Claude Plassart  reflected on the reasoning behind the changes, “Safety is important, we have invested a lot in the track, but (the) cars are going faster and faster, and this has created accident(s) and (this) has concerned us, we have to reduce the speed of the cars.”  Ironically, Plassart added, ”And reduce costs, cars should be cheaper to build and cheaper to race.”
The rear wing changes were quite simple; a reduction in span from 2.0 meters to 1.6 and a shortening of wing chord from 300 mm to 250 mm.  The rear wing changes weren’t necessarily a direct response to the yaw incidents; the changes were made more out of a desire to simply reduce cornering speeds in general as that was felt to be a contributing factor to the blow overs.

The immediate effect was a loss of total downforce and a not so insignificant change in front to rear aerodynamic balance.  Between seasons development naturally produced balanced cars, but with slightly (perhaps) less downforce and slightly more drag.  Or so that was the goal of the regulations change.
And ultimately you can't argue against the results; lap times did slow in 2009.  Analyzing events that ran to full 2009 ACO regulations (Le Mans Series events, Le Mans, in addition to Sebring and Petit; events where full ACO regulations were adhered to) qualifying lap times increased an average of 2%.  But then again, how much of that time increase could be attributed to the 10% power reduction the diesels earned for 2009?  The narrow span rear wings, coupled with the 20 mm domed skids introduced at the beginning of 2009 as well, did seem to have a cause/effect relationship inasmuch as there haven't been any yaw induced flips since.  But what was more influential, the narrow span rear wing or the 20 mm domed skid that increased ride heights significantly?

In direct response to the changed regulations, two new trends emerged with one driving the other.  First, in order to recoup as much lost downforce as possible, aerodynamicists immediately began utilizing much more aggressive rear wing angles of attack in addition to more aggressive wing profiles and wing cambers.  The second trend was in response to the first and ultimately was much more intriguing, if perhaps only initially.  The intriguing bit was that nearly simultaneously, both Audi and Acura debuted their race cars, the R15 and ARX-02a respectively, with nearly identical specific details in the area of the rear wing.  Instead of utilizing a conventional bottom rear wing mount, both cars arrived with top mounts for the rear wing mount, so called “swan neck” mounts.  But how could two cars with completely divergent design philosophies come to the exact same design execution on one detail in a critical area?  What was going on here, another Stepneygate?

Actually the answer was comparatively boring and quite simple.  It turns out, as aerodynamicists started to go down the route of more aggressive rear wing assemblies, they stumbled upon one fundamental problem; flow separation in the area of the conventional bottom wing mounts.  And apparently the solution was pretty universal, hence Audi designers using an Italian scale wind tunnel agreed with Acura designers using a digital wind tunnel.
Acura ARX-02a, Sebring WDWT, 2009But in the end, how much downforce was really lost by the initial span and chord reduction?  And with development, how quickly was it gained back?  In the winter of 2008 it was obvious that the world economy was in the gutter.  Yet the ACO was proposing expensive safety changes for the following season with the singular objective of slowing the cars down.  But surely there were vastly less expensive alternatives?  And how effective were the narrow wing rules in reducing downforce?  We've assumed that they did as intended and reduced downforce.  But aerodynamicists are a clever lot, and it would be pretty naive to assume they accepted the loss and that was the end of it.
But with what methods at hand could we use to independently explore the effects of the ACO's 2009 rear wing regulations?  Could we also replicate what was seen in the development of the swan neck rear wing mounts?  If the cause and effect was so universal, could they be repeated?
Inquiring with insiders at various LMP manufacturers produced little in the way of concrete answers.  Apparently discussions of downforce lost are as short as discussions about downforce gained, even when only looking for a relative answer.  This would be telling in hindsight.

But short of a good sized budget and a wind tunnel, this investigation was coming to a rapid halt.  But of course there was CFD.  Could these questions be investigated accurately utilizing commercially available CFD?
Taping the talents of RCE's Mr. Simon McBeath and the CFD option quickly became reality.  The only thing required was the time required to generate the CAD files on my end, and all the meshing and case running on Simon's end.
We started with a 2008 wing profile that had been used by a named LMP effort who will remain anonymous.  The first case tested was the 2008 profile to the 2008 full width (2000 mm) span and 300 mm chord and mounted to a conventional bottom rear wing mount.  In isolation, this case generated 1739 lbs. of downforce and 226 lbs. of drag, 7.69:1 L/D.  
Up front let's mention that we shouldn't  get too fixated on the absolutes.  But for a reality check, the  numbers were put in front of someone with knowledge of what a contemporary LMP car rear wing should generate and their response was, “The absolute forces you calculated seems reliable to me.”  And throughout this process we had people with relevant knowledge looking over our work making sure we didn't loose the plot too much.

Next we lopped off 400 mm of wing span and scaled the “2008” profile to the 2009 regulated wing chord (250 mm).  Naturally this wasn't a bespoke wing shape given the ACO legality box, but we were simply looking to see what the downforce loss was if we took the old wing and made it fit the new regulations.  The results was a 593 lb loss in downforce (1146 lbs total) for a 70 lb loss in drag (156 lbs total).  Interestingly, efficiency stayed about the same at 7.36:1.  This was a 34% loss in rear wing downforce for a 31% reduction in drag but only a 4% loss in efficiency.  Dome's Hiroshi Yucchi commented, “34%, just by wing change, is almost the same as our wind tunnel results.”  He furthered, “It sounds quite accurate.”
2009 LMP rear wing CFD study, flow separation around conventional bottom mountsPlaying the part of a design team within a major LMP program, and with the “encouragement” of a nearly 600 lb downforce loss, it's pretty evident development would immediately commence to gain back much (or, as much as possible) of what we lost.  But obviously teams wouldn't merely just scale down their old 2008 rear wings.  They would look to optimize the wing to the new regulation box.  And this meant getting into the wing development business.  This wasn't for the faint of heart and the project could easily fall off the rails here.  At this point our experts were brought back into the fold to get an idea of what manipulations would produce the best “bespoke” wing for the 2009 regulations.  In discussions we came to understand that the mainplane angle of attack and camber were two basic methods used to modify the rear wing looking to gain back the lost downforce.

So with our wing modified as directed, and everything looking copacetic, CFD runs showed it nearly 720 lbs down (1019 lbs total) over the benchmark 2008 wing.  What was going on?  This should have been the ticket.  A clue was in the drag figures (214 lbs) as it was gaining even over the previous scaled 2008 wing case.  So we were losing even more downforce and gaining drag when at very least we expected increases in downforce.
We suspected the culprit was flow separation.  And indeed, flow visualizations showed a large disturbance in the area of the rear wing mounts.  Testing our theory, two additional runs were tested that backed the mainplane angle out first 1.5 degrees and then 3 degrees, rotating around the MP trailing edge, all while keeping the flap angle and all other parameters constant.  The 1.5 degree reduction showed little better than a repeat; +3 lbs downforce, -8 lbs. drag (1022 lbs and 206 lbs respectively).  But most interestingly, with the 3 degree reduction, suddenly the bespoke 2009 wing came alive with downforce increasing by 205 lbs and drag dropping 27 lbs. (1227 lbs and 179 lbs).

But in reality we had been tutored to look for this.  And this was the answer to why the swan neck rear wing mounts came into being.  With the use of higher camber rear wing mainplanes and higher angles of attack, the conventional method of mounting the rear wing proved to be a source of flow separation.

Flow separation on conventional mount rear wingAnd so it was with much anticipation that we tested the swan neck wing mount case.  Indeed things began to get even better:  1299 lbs of downforce for 186 lbs of drag.  The flow separation went away, and at this point we were “merely” 440 lbs down on the 2008 full-span case.  In terms of efficiency we weren't that badly off, only 8% down on wing L/D.  And matching drag through an increase in flap angle (+8 degrees) saw downforce further increase to 1413 lbs.  At that point we were within 4% of our 2008 rear wing drag level, so there was a tad bit more to be gained downforce-wise (perhaps, L/D was now down 15% compared to the '08 case suggesting we were petering out in this setup's potential), but we moved on to other areas of development.

We tried a number of rear wing endplate iterations, but saw little benefit.  This isn’t to say this couldn’t be an area of successful development, but in our limited running (all straight line), we saw nothing promising.

We also tested a reverse swan neck, one that came up over the trailing edge of the wing.  Over looking the practicalities of locating such a mount on a contemporary LMP gearbox given rear overhang maximums and desired rear wing position, it essentially didn’t perform any worse than the standard swan neck, 1288 lbs downforce, for 186 lbs drag.
Swan neck mount, no flow separation: same Main Plane AoA, same flap AoA as conventional caseIn the end we were able to claw back 16% of the initial 34% loss when matching for drag and within our limited number of runs.  Certainly with further development on the wing, we only ever contemplated extruded 2D sections after all, as well as entertaining other areas of the car, and gains well beyond what we found were within reason.  And while we never we able to extract answers to how much downforce the major LMP efforts were able to gain back, in hindsight it could be seen as tacit admission that the 2009 LMP regulations did very little in actually stripping the cars of downforce.

Indeed, with Dome's withdrawal from Le Mans competition and subsequent release of aerodynamic figures for their S101 and S102 series of LMPs, we had laid out in front of us what the net effect was to one competitor; between the 2008 S102 and the unraced 2011 S102i, Dome saw a 24 lb gain in downforce for a 50 lb drag increase in their Le Mans configuration.  Dome's Hiroshi Yucchi indicated, “Due to the small rear wing, we initially lost around 4% efficiency. Then we managed to recover 3% by the rear fender, wing stay design, and so on.”  When all was said and done, Dome suffered a 1% decrease in efficiency, “It was estimated around 0.5 sec per lap (at Le Mans).”  The cost?  According to Yucchi, “...between JPY 20,000,000-25,000,000 ($239,500-$299,400) to produce one car set. This does not include the aero development costs.”  This covered tooling and one car set worth of update parts, not tunnel, CFD, or CAD time.

So was upwards of $240,000 worth a 1% reduction in efficiency which equated to a 2% increase in lap times and even less than that at Le Mans?  When the stated goal was to reduce cornering speeds, no.  Given that the 2011 regulations were coming on line, it made even less sense for the ACO to implement these changes when they did given the Rear view of swan neck mount, no flow separation: same Main Plane AoA, same flap AoA as conventional casesurrounding economic climate.  A much more cost effective change would have been the simple implementation of drastic inlet restrictor reductions aimed closer to the proposed 2011 power levels.  This would have been a reduction of between 100 and 150 hp (from 700 to around 550). The cost would have been negligible, and coupled with a regulation mandating an engine freeze up to 2011, there would have been no incentive for expensive engine development.  According to John W Judd, “changing restrictors is very cheap compared to the change in wing design, particularly as for some teams the first opportunity to test the new design at high speed may be the Wednesday of Le Mans week, 3 days before the race starts.”  Engines would have needed to be remapped, but as Judd points out, “...we are used to the restrictor size changing almost on an annual basis, so the work to optimize the engine for a new restrictor is something we are used to anyway and wouldn't be an additional cost to the teams.”

But Zytek's Tim Holloway offers a slightly different opinion, “You are right in that they could have proposed a simple large power reduction which would have reduced lap times. And as Judd says, it would not have been difficult to adjust engines to suit. But as always, we chassis people would then want to take drag out of the car... and that would lead to a high cost aero program.  So which ever route you take there is no cheap, simple solution other than where we started out.”  Perhaps, but the cost burden would have shifted from a high mandated cost to a more reasonable elected cost, the difference between need and want.  With the way the rules were implemented, the 2009 aero regulations became a $240,000 rules compliance fee.  Naturally the tooling and development cost wasn't a direct burden on the privateers (though it did effect the price they paid), but what did the Lolas and Zyteks sell their update kits for, $60,000-$70,000?  Regardless, it wasn't simply a $1000 fee, plus whatever you could afford to throw at it to stay competitive (as is done year to year), had the ACO had gone the restrictor route.

ęCopyright 2013, Michael J. Fuller