g-force - the design of an unlimited class reno...

(c)2000 American Institute of Aeronautics & Astronautics or Published with Permission of Author(s) and/or Author(s)' Sponsoring Organization.

AOO-39948

AIAA-2000-4341G-Force: The Design Of AnUnlimited Class Reno RacerE. Ahlstrom, R. Gregg,J. Vassberg & A. Jameson

18th AIAA Applied Aerodynamics Conference14-17 August 2000 / Denver, CO

For permission to copy or to republish, contact the American Institute of Aeronautics and Astronautics,1801 Alexander Bell Drive, Suite 500, Reston, VA, 20191-4344.


G-Force: The Design Of An UnlimitedClass Reno Racer

E. Ahlstrom* (Star Aerospace LLC), R.D. Gregg1^ (Independent Consultant),J.C. Vassberg* (Independent Consultant), & A. Jameson* (Stanford University)

IntroductionIn May 1999, Renaissance Research contractedStar Aerospace LLC to design an UnlimitedClass racing aircraft. At present, modifiedpiston-prop fighters designed before andduring WWII dominate this form of racing.These aircraft are now over 50 years old andhave very little service life left. Enginereliability is so poor that a winning Unlimitedaircraft will spend less than one hour at fullthrottle between engine rebuilds. Overhaulcosts range from $150,000 to $250,000 for the1940's technology engines due to thedisappearing parts supply. Finally, the historicvalue of flyable WWH fighters has increasedto over $1,000,000, well beyond the levelaffordable for racing use. Replicas of existingdesigns face the same engine limitations astheir predecessors. If Unlimited racing is tocontinue through the next decade, newUnlimited race plane designs are required.

Previous attempts in the last twenty years atfielding a dedicated Unlimited racer have beenlimited to two conventional designs (onesingle-engine and one twin), and one warbirdderivative. The single engine aircraft,Tsunami, was basically a downsized version ofthe P-51. Freed of the military requirements ofarmament and long range, Tsunami was thefirst Unlimited to unofficially post a 500 mphlap at the Reno air races. This design did notmake use of advances in aerodynamic methodsor concepts. Furthermore, it used a Merlinengine with the same dwindling, expensiveparts supply as the modified Mustang racers.Unfortunately, Tsunami and pilot were lost in

a landing accident in 1991 due to a flap failure,limiting the influence of this design onUnlimited racing.

The twin-engine aircraft, Burt Rutan's PondRacer, was similar in configuration to the P-38with twin-engine booms and a center podcockpit. The small auto racing engines usedhad never seen continuous duty at over 500 HPand quickly overheated under race conditionswhere 1000 HP was needed just to becompetitive. Engine reliability of this aircraftat Reno continued to be an issue. In threeyears, this aircraft only finished one low speedBronze category race coming in second to astock P-51. The aircraft and the pilot were lostin 1993 after an engine failure.

Last, Miss Ashley II consisted of a replica P-51 fuselage mated to a Lear 23 whig and tailalong with a Griffon engine-powering counterrotating propellers. While fast, it suffered acontrol failure in 1999; the aircraft wasdestroyed and the pilot lost.

The record of built-for-purpose Unlimitedracers is even worse than the aircraft they weresupposed to replace. Any new design mustaddress the issues of these failures and proveitself reliable and safe enough to take the placeof the tried and true warbirds.

This paper discusses the design workcompleted for a new Unlimited Class RaceAircraft named the G-Force Racer.

President, Star Aerospacef AIAA Sr. Member, also Sr. Manager, The Boeing Company

AIAA Associate Fellow, also Associate Technical Fellow, The Boeing Company§ AIAA Fellow, T.V. Jones Professor of EngineeringCopyright © 2000 by E. Ahlstrom, R.D. Gregg, J.C. Vassberg, A. Jameson. Published by the AmericanInstitute of Aeronautics and Astronautics, Inc. with permission.

(c)2000 American Institute of Aeronautics & Astronautics or Published with Permission of Author(s) and/or Author(s)1 Sponsoring Organization.

G-Force: The Design Of An Unlimited Class Reno Racer

Aerodynamics TeamEric Ahlstrom, president of Star AerospaceLLC was responsible for the conceptualvehicle design and the preliminaryaerodynamics. Based on the Mach limits,Reynolds numbers, and drag requirements ofthe concept a team of applied aerodynamicistswith experience in state of the art transportaerodynamics was chosen for the detailedaerodynamic design and analysis. Theaerodynamics team led by Robert Gregg andincluding John Vassberg, Dennis McDowell,and Mark DeHaan refined the design. CFDsolvers and optimization methods as well asadditional support were provided by Prof.Antony Jameson (References 1 -7).

Design ObjectivesThe ideal Unlimited racing aircraft wouldextend the performance envelope to the limitsof current propeller technology; Mach .80 and35 HP/ft2 of propeller disk area. Reliabilityshould allow not only several races per year,but also several hours of practice time at eachrace without failure. In the coming years, therace schedule may be expanded to 6 to 10events. To do this, the engine must be reliable,light, and powerful enough to last an entireseason without failure.

The following is a list of requirements andlimitations agreed upon by RenaissanceResearch and Star Aerospace LLC for thedevelopment of the G-Force Unlimited raceaircraft.

1. Performance (all performanceprojected at 5,000' MSL, ISA +20°C)a. >600 mph TAS top speed in level

flight,b. >550 mph lap speed on the 1999

Reno Unlimited air race course:19% at 1 G, 60% at 4 to 6 G, and21% in transition,

c. 9 G maneuver load, 5 G gust load;total 14 G limit load factor with1.5X safety factor to ultimate loadfailure.

d. Vne > 550 KEAS.e. Mmo> 0.80.

f. Roll rate > 200°/sec at 350 KEAS.g. Stall speed < 90 KEAS.h. Landing distance < 1,500 ft., dead

stick.2. Schedule

a. 1 prototype aircraft with sparessufficient for 10 race or air showevents tested throughout theperformance envelope and deliveredrace ready in 24 months from dateof contract award.

b. Production aircraft lead time of lessthan 12 months followingcompletion of prototype testing,building to a rate of 2 aircraft peryear.

It is interesting to note that the requirementsfor sustained turn rate and Vne exceed those ofsome state of the art jet fighters.

Design RequirementsThe design limitations,assumptions are;

requirements, and

1. Piston powered and propeller driven(this is the only real rule in UnlimitedAir Racing!)

2. Engine power/weight ratio: forreliability at continuous output, a limitof 2.5 HP/lb. was set for a modernturbocharged piston engine and gearreduction package with all accessories.

3. Stability and controla. Manual, unboosted control system,b. Positive static and dynamic stability

greater than current Unlimited raceaircraft,

c. Minimal change in stability, poweron vs. power off.

d. Minimal change in CG and handlingqualities, single vs. dual pilot.

4. Crew provisionsa. Dual pilot, tandem seating, dual

control for race training and PRflights,

b. Low altitude ejection or pilotextraction systems for both pilots,

c. 30° reclined angle for G tolerance,d. Provisions for mil spec oxygen and

G-suit connections.

2 of 18



Course AnalysisAn analysis of the Reno Unlimitedrace course yielded an optimum turnradius of about 1 mile for all threeturns. These turns comprise over 70%of the race course. The differentialbetween the pylon-to-pylon distanceand the projected ground track willcause the aircraft to travel a minimumof 4% farther than the measured lapdistance. Given the lap speedrequirement of over 550 mph, therequirement for a sustained turn rate of9.1°/second or 4 G was set at 550 mph.The increase in speed on the smallstraight areas of the course wascalculated to approximately offset the4% extra distance in the turns.Turbulence and other aircraft typicallyincrease G loads to 2 or 3 times normalmaneuver loads.

The 1999 and 2000 racecourses areillustrated below in Figure 1. The 1999pylon to pylon lap distance was 8.3miles. In 2000, the pylons before andafter the finish line were moved tomake the turns more uniform. The"deadline" at the south edge of thecourse was moved in to the north sideof the runway. These changes weremade to reduce the G required for agiven lap speed and improve spectatorsafety. No changes were made to theprogram specifications.

Conceptual DesignTo satisfy these requirements, Sam Bousfieldof Renaissance Research developed apreliminary configuration based on a midfuselage or body propeller system.

The body propeller concept originated in theearly era of aviation development to provideenergized airflow over the tail surfaces andplaces the heavy engine, gear reduction, andpropeller weights closer to the aircraft centerof gravity. Its use was abandoned due tomechanical complexity in 1924. Inventor Sam

NEW PYLON 1 LOCATION

NEW DEADLINE

NEW PYLON 8 LOCATION

Figure 1 -1999 & 2000 Reno Race Course

Bousfield has developed several innovativemethods to reduce the mechanical complexityrequired and has sponsored research into theperformance potential of a body propellerequipped aircraft for both racing and civil use.Patents on these applications are pending.

The body propeller configuration allowed avery low fuselage wetted area and provided thepotential for significant laminar flow over theforward fuselage. Engine size was chosen toprovide about double the thrust to weight andthrust to wetted area ratios of the currentUnlimited record holders, Rare Bear, Tsunami,and Dago Red.

3 of 18



Fuselage DesignThe fuselage was lofted as a minimum volumefairing around the required propulsion system.Due to the 80" length and 40 to 50" width ofthe engine package, a wide forward cockpitwas guaranteed and the addition of a secondcockpit posed virtually no aerodynamic orstructural penalties. The fuselage was loftedwith a fineness ratio of 5.7, which later provedto be too aggressive. The pressure recovery inthe region of the body propeller was tailored toprovide an area of decelerated flow to thepropeller and improve propulsion efficiency attransonic speeds.

Good structural layout was provided bylocating the engine in the upper half of the midfuselage, the wing below, and the coolingsystems faired into the lower fuselage. Thefuselage has been designed to be all compositeforward of the firewall; composite shell with astructural sub frame in the engine area formaintenance access; and all composite aft ofthe body propeller. Primary bending loads willbe carried through twin keel beams under thewing. All tail loads will be carried through thebody propeller gearbox.

The propulsion system mass fraction of 42%centered in the fuselage provided for minimalCG change with different pilot weights andfuel loadings. Both crew positions wereprovided with sufficient room for SKS-94 typepilot extraction systems reclined 30° for Gtolerance.

Conceptual Wing DesignReference wing area was set at 75 ft2 toprovide a wing loading range of 40 to 46Hx/ft2. Despite the high dynamic pressure ofthe racing environment and the opportunity forvery high wing loading, the lower wingloading was necessary to provide a slow stallspeed for safe dead stick landings. Based oncost and schedule constraints, a simple high liftsystem of single fowler flaps without a leadingedge device was chosen. This allowed thepotential for significant laminar flow, althoughthe design for Mdd and structural thicknesswould have priority. Unit Reynolds numbers

encountered during racing would range from 3to 4.5M per ft. This gave wing Reynoldsnumbers of up to 20M at the wing root and 9Mat the tip, comparable to a larger aircraft athigher altitude.

The maneuver loads, sustained turn rate, andthe gust load required no buffet at a CL of 0.64at M 0.72; an Mdd of 0.75 at a CL of 0.3; andan Mdd of 0.80 at a CL of-0.05 to 0.1.

A trade study of wing thickness, sweep andtaper ratio was made using NASA SC(2)airfoils as a baseline. From this, a sectionthickness minimum of 13.5% at the wing rootand 12% at the tip was chosen combined witha quarter chord sweep of 28° for the Mddspecification of M 0.80. An aspect ratio of 8.3and a taper ratio of .45 were chosen to allow awing extension if race testing determined aneed for it. Conversely, a production breakwas included in the wing at ~87% span toallow a reduction in the wing area by 4 ft2,yielding an aspect ratio of 7.5. Theseadjustments will be used for different racecourses where either more or less turningcapability is required.

A planform Yehudi was incorporated into thewing trailing edge to provide space for themain landing gear. Because of the proplocation and diameter, a fairly long gear lengthwas required for propeller clearance. Laterlengthening of the fuselage required theYehudi to be blended with the outboard wingto provide more gear length. The Yehudi hadthe added benefit of reducing the wingdownwash angle leading into the propeller.

The wing is a two-spar design with spars at 15and 65% reference chord. A secondary sparbehind the main gear-well parallels the Yehuditrailing edge and structurally supports the gearpivots. One-piece wing box construction willbe used to reduce weight and complexity.

4 of 18



Conceptual Tail DesignTo provide prop strike protection, the verticalstabilizer was located below the tail cone.With high-dihedral horizontal stabilizers in thenormal position, this provided a Y-tailconfiguration. Additional benefits are that thevertical stabilizer becomes more effective at

Body Prop Design -Figure 2 shows an inboard side view of all ofthe primary systems of the body prop equippedUnlimited racing aircraft illustrated for volumeand balance. It is clear from this view that thebody prop allows a compact and aerodynamic

0 2 3 4 5 6 7 8 9 1 0 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27

Figure 2 — Body Prop Aircraft Profilehigher angles of attack and is not blanked bythe fuselage or horizontal stabilizer in a stall orspin.Tail sizing was set to provide approximatelydouble the power on stability of a P-51. Thiswas considered necessary due to the turbulentconditions of the race course and feedbackfrom experienced pilots.

Configuration ConceptsA comparison study was made of the bodyprop design vs. the more conventional tractorand pusher configurations for the mission ofUnlimited air racing. For the same wingloading and engine requirements, the studyshowed the body propeller configurationallowed the lowest wetted area, lowest drag,and highest propulsion efficiency for therequired single engine, single propellerconfiguration.

arrangement of engine, propulsion, structureand cockpit. The location of the wing spars,engine, and cockpit in close proximity allowsthe shortest load paths and the lighteststructure. The fuselage illustrates a 24" stretchin the aft engine area over the originalconceptual design to increase the wings criticalMach number (See discussion in Wing DesignSection). This leaves a great deal of unusedspace between the engine and body prop. Theexhaust and turbocharger installation may bemoved into this area to improve engineaccessibility. The engine block would bemoved forward 6" to compensate for the CGchange. Even so, this would provide 4 to 6"more room in the aft cockpit to carryinstrumentation or recline the aft pilot forhigher G tolerance. The proposed newturbocharger and engine arrangement isillustrated in the pusher version below.

5 of 18



The body propeller provides a very low dragdesign by allowing a laminar flow fuselage andwing. On takeoff, the prop accelerates theairflow over the tails providing positive controlearlier than is possible in a pusher. Thelocation of the prop near the CG provides avery small change in stability and handlingqualities power-on vs. power-off. Finally, byprofiling the fuselage at the body propinterface for a high degree of pressurerecovery, the body prop can be made to fly inslightly decelerated air. This allows a higheradvance ratio before the prop exceeds Machlimits and can be a large performanceadvantage at transonic speeds.

Mid-engine Pusher Design -

The mid engine pusher configuration is shownin Figure 3 and was derived from the bodyprop design by freezing the design forward ofthe wing trailing edge and changing only the

same power off stability as the body propversion. The pusher prop is located further aftof the CG than the body prop, so the power onvs. power off stability change is greater. TheCG impact of the aft propeller can becompensated for with either a more forwardengine mounting or ballast.

The mid-engine pusher thrust line was set tokeep the aft lower fuselage pressure recoverywithin airflow separation limits. The landinggear length is restricted by the wing design, soa full size-racing propeller would require atailskid and/or main gear skid to prevent tailstrike on landing. Alternately, a higherpropeller disc loading could be used anddiameter reduced.

The larger tail and increased fuselage wettedarea account for half of the performancedifference between the pusher and body propversions. The other half of the performance

I 2 3 4 5 6 7 8 9 1 0 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27

Figure 3 — Mid-engine Pusher Aircraft Profile

propulsion and tail. This enables the pusher toshare many of the advantages of the body propversion and use much of the same tooling. Theforward fuselage can achieve significantlaminar flow; the mid mounted engineprovides low pitch and yaw inertia; and the aftcockpit has excellent visibility. The aircraft islonger than the body prop version and hasmore wetted area in the fuselage due to thelarge aft prop spinner. The tail has beenmoved forward and enlarged to maintain the

difference is due to the lower flow decelerationavailable to the pusher at the propeller.

Tractor Design -The tractor layout shown in Figure 4 is themost traditional aircraft configuration and isthe most mechanically convenient. Placementof the engine, gear reduction, and propeller atthe front of the aircraft allows the simplestmechanical design and best engine



0 I 2 3 4 5 6 7 8 9 10 II 12 13 14 15 16 IT 18 19 20 21 22 23 24 25 26 27 28

Figure 4 - Tractor Aircraft Profile

accessibility. The tractor prop placementallows the shortest landing gear length andreduction of the internal volume required forthe retracted gear. A belly scoop coolingsystem was incorporated and did not accountfor a significant drag difference from theinternal cooling systems of the body prop orpusher.

The tractor is the longest of the threeconfigurations due to the high polar moment ofinertia caused by the forward engine placementand the negative stability impact of the forwardpropeller. To keep the tail area within draglimits required a longer fuselage, whichreduced the size of the tail surfaces and thetotal wetted area. A reduction in fuselagewidth aft of the engine combined with thecanopy extension for the aft cockpit served anarea ruling function. These factors improvedwing root critical Mach number over the midengine designs. However, the accelerated flowfrom the propeller increased the local Machnumber at the wing root by the same amountso there was no net improvement. The entirefuselage is in turbulent flow and cannot bemade laminar.

The tractor requires the largest tail volume ofall three configurations and has a large stabilitychange power-on vs. power-off. It should benoted that all of the current tractor Unlimitedracers are experiencing some amount of yaw

instability at race speed due to the stabilityderivative of the forward propeller. 28% of theperformance shortfall of the tractorconfiguration vs. the body prop is due to theincreased wetted area and the other 72% is dueto the decrease in propulsion efficiency and theincrease in fuselage drag due to the turbulent,accelerated flow and fuselage pressure dragfrom the tractor propeller.

Propeller ConsiderationsThe body propeller looks radical, yet theconcept of a prop on a body of revolutionoperating at transonic conditions has been wellproven. The integration of a transonicpropeller in the aft third of the fuselage provedto be very similar to the MD-80 and Boeing7J7 UnDucted Fan (UDF) engine nacelles ofthe late 1980's. The MD-80 UDFdemonstrated excellent efficiency andperformance up to Mach 0.8. The UDFconcept was eventually dropped as a result offuel prices not increasing as projected in theearly 80's. This concept would have likelybeen introduced into a production aircraftdesign had fuel prices remained high.

Even at speeds of Mach 0.75 to 0.8, a bodypropeller driven by a turbocharged pistonengine exceeds the efficiency of the best fanjets. This makes the body propeller design the

7 of 18



best choice for the next generation ofUnlimited racing aircraft.

Prop diameter is a primary driver of bothpropulsion efficiency and aircraft stability. Atractor aircraft requires the shortest landinggear for a given propeller diameter, but suffersthe greatest stability loss. The pusher gets thegreatest stability enhancement from thepropeller in the power-on condition, but loses

The performance comparison, shown in Table1, shows that the body prop could be capableof breaking the existing prop driven speedrecords on a power rating of 1500 HP, whilethe pusher would need 1800 HP; the tractordesign would barely scratch the current recordsat 2200 HP.Comparing lap speeds, we note that the pusherrequires 22% more power for the same

Dimensions

Empty weightFuel capacityRace takeoff weightLengthWing spanWing areaWetted areaAspect ratioWetted aspect ratioAft cabin widthPerformance atReno,NV; ISA+20CPiston-prop level speed recordLevel speed, 1500 HPLevel speed, 1800 HPLevel speed, 2200 HPPiston-prop lap recordLap speed, 1500 HPLap speed, 1800 HPLap speed, 2200 HPMax L/D (glide ratio withoutpropeller)

Mid-engineBody Prop

2,754 Ib.83 gal.

3,454 Ib.27ft.25ft.

86.9ft2

437ft2

7.21.4252"

Mid-engineBody Prop

529 mph564 mph606 mph


17.9

Mid-engine Pusher

2,956 Ib.83 gal.

3,656 Ib.28.5 ft.

25ft.86.9ft2

473ft2

7.21.3152"

Mid-engine Pusher



17.2

Front-engineTractor

2,794 Ib.83 gal.

3,494 Ib.29.5 ft.

25ft.86.9 ft2

460ft2

7.21.42

<35"Front-engineTractor

541 mph482 mph511 mph547 mph492 mph458 mph475 mph498 mph

13.4

Table 1 —Performance Comparison of Configuration Concepts

this benefit power-off. A large propeller on apusher aircraft can make the landing gear solarge as to be impractical. With its locationnear the CG, only the body prop aircraftcombines minimal stability change withacceptable gear length.

Both the body prop and pusher designs aremid-engine, and allow radical changes inengine weight and volume without significantchanges to the airframe. This makes eitherdesign a viable choice for new enginedevelopment. The tractor design wouldrequire new aero structure and ballasting toallow any significant change in engine weightor volume.

average lap speed. The tractor needs 47%more power to go 5 mph slower than the bodyprop.

Configuration RefinementThis section covers the refinement of thevarious components of the aircraft. For thepurposes of the paper, discussions regardingeach component are split into differentsections.

Wing Design -Any successful system design effort mustaccommodate a changing set of requirements

8 of 18



as the designers of the various subsystemslearn more about how their individual effortsimpact and are affected by the actions of theother designers. This was certainly the casewith this wing design as we integrated it withthe fuselage, propulsion system, stability &control, manufacturing and overall packaging.Most of the assembled team has workedclosely together for more than a decade. Ourstyle has been to allow the individuals of theteam to gravitate towards the work items thatthey feel most comfortable with, however,each member will loosely participate in allconcurrent activities in progress. Thisparticipation is usually in the form of dailydiscussions regarding the overall design of theaircraft. By disseminating everybody'sfindings on a very frequent basis, the group asa whole begins to understand how best tomaximize the system's performance, even if itmeans taking additional compromises at thesubsystem levels. These informal designreviews also serve the purpose for sanitychecks to occur frequently enough such that apoor design direction is not ventured too far.

Another characteristic of this group is that wehave been programmed to work on tightschedules and under small budgets. As aresult, you will notice that the group utilizesthe most cost-effective tools at every stage ofthe design effort. Initially, when the design isnot very well understood, design charts andrules-of-thumb dominate the effort. As thedesign begins to evolve, and these methods no

longer add value to the direction of the group,linear methods are drawn into the tool set.Then, as the ROI of linear methods begins toreduce, they are replaced by non-linear tools -starting with the simple and finishing with themost sophisticated. Using the right tool at theright time helps us manage costs and schedule,and allows our final designs to be competitiveat the highest level.

The design of the wing geometry occurred inseveral phases, with the basic requirementsdefined by Ahlstrom in phase I. Theserequirements included the general layout(planform, thickness distribution) of the wing,the design cruise condition (M=0.77,CLtot=0.32, Ren=14.5M), off-designcapabilities (CLbuffe,=0.64 at M.72, Mdd=0.80 atCL=0.1, CLmaxcw=l-6, etc.), and a pad on thedivergence Mach number to allow room forgrowth in out years. McDowell verified thatthe original planform sweep and thicknesscombination was appropriate for the desiredMach number capabilities of the race coursemission; this was done with design charts.

The baseline wing of phase II was defined byGregg, McDowell & Vassberg using airfoilsections derived from NACA 64 sections,scaled to conform to the original planform andthickness distribution established in phase I.Some cursory 2D optimizations wereperformed on these sections to better tailortheir characteristics for the initial designconditions; the 2D conditions and geometrytransformations used for this effort were based

Preliminary WingBuffet Boundary

0.850.8

0.750.7 -

1 0.650.6

0.550.5 -1

0.450.4

0.6 0.62 0.64 0.66 0.68 0.7 0.72 0.74 0.76 0.78

Mach

Figure 5 - Preliminary Wing Buffet Boundary

0.8



on simple-sweep theory. The method used forthis 2D activity was Jameson's SYN103 coderun in Euler, drag-minimization mode. At thispoint, the remaining unspecified geometricquantity was the twist distribution of the wing.To set this, we utilized Jameson's FLO22wing-only code, but we included pseudo-fuselage effects. The first pseudo-bodyinfluence is its acceleration on the on-set Machnumber at a critical station on the wing;typically this is around 50%-60% semi-span.Running the isolated fuselage geometry in asurface-panel method and interrogating theflowfield velocity at the critical wing stationdetermine this acceleration. The secondpseudo-body effect is how the presence of thefuselage at an angle of attack warps the on-setflowfield's local angle of attack as a functionof span location. The third pseudo-bodyinfluence is related to the carry-over lift of thewing's circulation onto the fuselage. This ratiois defined as Cunt/Cuving, is 1.22 for the G-Force configuration and was determined byrunning a surface-panel method on thewing/body combination. These pseudo-body

effects are included in FLO22's wing-onlysolution by running the exposed wing in thecode at the wing's CL, at a higher Machnumber and re-referencing the results back tothe original Mach, and adding a delta-twistdistribution to the wing to simulate theflowfield warping. Using this procedure, wespecified a twist distribution that yielded anear-elliptic span loading. This initial designwas done very rapidly, covering only a two-day period of after-hours work. It provided usa point to start the 3D design effort. Forreference, FLO22 runs in about 7 seconds onan AMD Athlon 800 MHz powered PC.

The initial analysis indicated that the designrequirements could be satisfied with a thickerwing than the conceptual layout of Ahlstrom.Some additional margin in buffet boundary,cruise Mach number and M^ was felt desirableto provide an additional margin for aileroneffectiveness at these speeds. Figure 5 presentsthe initial buffet analysis of the wingconfiguration as estimated by FLO22 at thisstage of the design. Table 2 shows the

DragrisePreliminary Wng.

QQ25

002

0015oaO

001

0005

-CL=0.1-CL=0.2

-CL=0.3-CL=0.4

-CL=0.5--Meld

045 05 055 06 07 075035Mach.

Figure 6 - Preliminary Wing Dragrise

08 086

10 of 18



CL0.10.20.30.40.5

Mdd0.7960.7900.7750.7570.738

Table 2 — Preliminary Wing MuchCapability

estimated Mach capability at different CL's,while Figure 6 shows the drag risecharacteristics. All of this analysis indicatedthat requirements were being met, howeverthere was a serious concern with the bodyeffects on this low fuselage fineness ratioconfiguration. The dragrise characteristicsindicate that this initial wing had a Machcapability at 0.3CL of about 0.775; however,this estimate was shown to be optimistic due tothe pseudo fuselage correction used in thisanalysis.

Without worrying too much about it, the teamwas relatively sure that our baseline wing wasgoing to have problems near the root regionbecause of the atypical contouring of thefuselage geometry, thus began phase III thatwas conducted by Vassberg & Jameson. Thefirst thing to do was to assess what issuesmight exist with the baseline wing geometry(designated shark 1), as it integrates with thefuselage. This analysis was performed on 9NOV, 1999 using Jameson's SYN88 Eulermethod and is illustrated in Figure 7. Thequantities in the legend of this figurecorrespond to the wing forces. Furthermore,the drag listed is only the inviscid drag, whichincludes the induced and shock components.If you will recall, the design lift was CLtot=0.32and that the carry-over lift ratio was 1.22 forthis configuration. Hence, the wing lift isCLwing

=CLtot/l .22=0.27. The other item to noteis that this analysis was performed at a Machnumber of M=0.78 rather than M=0.77. Thereason for this bump in freestream Machnumber was due to the on-set acceleration ofthe flowfield near the wing root from the

SI1AKK1 WING-BODY CONFIGURATIONMACH= .7*0

SYMBOL SOURCE ALPHA a. CDSTHK Otadcl I MO ISO.

S-OJ

Figure 7-Sharkl Wing/Body Pressure Distributions11 oils



propulsion system. McDowell & DeHaanusing methods based on actuator-disc andblade-element theories determined thisacceleration. Since we were concerned mostabout the wing-root region at this stage in thedesign, we used the full level of propellereffects on the on-set Mach number. Referringto Figure 7 again, notice the strong shock thatunsweeps as it nears the side-of-body. As wehad suspected, the contouring of the fuselagecross-sections had an adverse effect on thewing aerodynamics, unfortunately it was worsethan we had feared. The inviscid drag(induced + shock) of the wing was-•DwinglNV—=180 counts for the baseline

configuration. For reference, SYN88 and adrag-minimization run with thousands ofdesign variables uses about 10X the CPU timeof an analysis calculation.Phase TTT continued by running SYN88 indrag-minimization mode, constraining thewing modifications to be thicker everywherethan the baseline geometry. Initially, we were

running these optimizations at a single point(cruise design), just to see what the potentialbenefit might be. Within 30 design cycles,SYN88 dropped the wing's inviscid drag from180 counts to 104 counts. This optimizationwas launched on the evening of 9 NOV, 1999and completed early 10 NOV, 1999... an over-night turn-around. Although fairly largeimprovements were realized, we felt we coulddo better if we were allowed to modify thefuselage contour near the wing trailing edge.In discussions between Gregg, Ahlstrom,McDowell, Vassberg & Jameson, severalconcurrent changes to the aircraft's generallayout were being considered. The team wasforming these ideas as we were beginning tolearn more about the complete system. Thechanges that were directly related to the wingdesign were the fuselage reshaping and atrailing-edge planform blending that wouldallow more room for stowing the landing-gearstructure. Gregg & Vassberg made theplanform modifications to the current wing and

3I1ARK52 WING/BODY CONFIGURATIONUACII= .7

SYMBOL SOURCE ALPHA CL. CDSVXU S_fc51 .7Li .1714 MTitSYXtt ShMfcl 1.000 .267K .01796

Figure 8 - Shark52 Wing/Body Pressure Distribution

12 of 18



Vassberg simultaneously generated anadditional three fuselage definitions thatparametrically stretched it by 1,2 and 4 feet aftof the wing-root mid-chord and consistent withthe engine packaging requirements. Otherfuselage geometries were created, howeverthey were not pursued because the fuselagestretching exercise achieved our goal toessentially eliminate the unsweeping of theshock system at the wing root. In fact, thetrailing-edge modification was also done in amanner to help alleviate the shock-unsweepproblem; as well as accommodate the landinggear. This planform change proved to bebeneficial as another triple-point dragminimization was performed on 13-14 NOV,1999, which dropped the wing's inviscid drag

data to show that we could completelyeliminate the shock unsweep with the 2-footstretch. This optimization reduced the wing'sinviscid drag from 92 counts to 72 countswithin 30 design cycles; the resulting winggeometry was designated shark52. Thepressure distribution for shark52 at M=0.78and CLwn^O.27 is shown in Figure 8. Itshould be emphasized that within the course ofone week, the wing geometry had evolvedfrom one that produced 180 counts of invisciddrag to shark52, which only has 72 counts atthe design point. This is an extremely largeimprovement accomplished in a verycompressed time! Furthermore, the databaseof CFD solutions (-100) had grown largeenough that very informed modifications to the

SI1ARK52 & SI1ARK1 WING/BODY CONFIGURATIONSUACH= .7*0

( Conliwri rf .05 Cf )

- .71 . Cl.- .17L4CD- .00711

Figure 9 - Comparisons of Initial and Final Wing Pressure Isobars

from 148 counts to 98 counts at the designpoint. For clarification, this wing redesign wasdone with the original fuselage to establish anew base for the parametric study, whichstretched the fuselage. Repeating similartriple-point optimizations on the 1-, 2- and 4-foot fuselage extensions provided sufficient

configuration could be made. This includedthe wing-planform change to better stow thelanding gear, as well as the fuselage reshapingto eliminate the shock unsweep issue. Figure 9shows a side-by-side comparison of thepressure isobars of the initial wing and theshark52 wing that clearly shows the reduction

13 of 18



in the shock strength across the entire wing, asa result of all of these important changes to theconfiguration. The final aspect of phase HIwas that we needed to rebalance the aircraft.This required a 6-inch stretch forward of thewing to compensate for the 2-foot stretch ofthe after-body. Once the fuselage geometrywas frozen, we polished the pressuredistributions some more by running SYN88 ininverse-design mode.

Not much else was done on wing design forthe remainder of 1999 because ofThanksgiving and Christmas holidays andother end-of-year commitments. However, theteam began to kick around the idea of alaminar-flow design. We did some quickcalculations on the attachment line Reynoldsnumber, Ree, of the current wing anddiscovered that we were within the rangewhere Tolmin-Schlichting waves along theattachment line would decay rather thanamplify. So we felt that the possibility oflaminar runs was achievable. The dilemma,however, was could we count on getting

laminar flow in the field. After all, this is arace plane whose mission is at very lowaltitude above ground level where bug strikesare sure to occur. What we decided to do wasto investigate whether or not we could tailorthe wing's pressure distributions to havefavorable gradients for up to 40% chordwithout adversely affecting the aerodynamicperformance of the fully-turbulent wingdesign. If we could, then we would change thedesign, yet not take credit for any laminar-flow-related drag reductions.

Phase IV began in mid-January, 2000. Thisdevelopment was concentrated on promotinglaminar flow on the wing without degradingthe performance of the wing if it was fullyturbulent as compared with the fully-turbulentdesign of shark52. This objective was notlimited to the design point, but rather wasexpanded to include the Mach number range ofM>0.74 and a lift range of CLwmg<0.27. Thefirst task was to compute the viscous flowabout the shark52 configuration at variousflow conditions. This was accomplished using

FINAL SHARK WING/BODY CONFIGURATIONRBN= 14JO . UACH= .780

SYMBOL SOURCE ALPHA CI. CD3TX1077 «uifc!«7 .fflJ ItU JJ11U«X1077 SkldcM? .Oil .IIU .0111*

——.—— STYLET skuaar i.ut JIM

0.4 0.6 0.1 1.0X / C

0.2 0.4 0.6 0>->JjQX / C

4IJHSp.il

Fieure 10 - final Wine/Bodv Pressure Distributions14 of 18



Jameson's SYN107P Wing/Body Navier-Stokes method. Starting with the computedpressure distributions of shark52, a series ofinverse designs were performed, also withSYN107P. Vassberg & McDowell wouldmodify the pressure distributions and Jameson& Vassberg would run SYN107P in inverse-design mode. The design Reynolds numberwas Re=14.5M, based on the reference chord.What we found was that it was easier toredesign the wing at the higher Mach numberand accommodate the requirements at M=0.74,rather than the other way around. This studywas completed on 8 FEB, 2000 with the winggeometry designated sharkns7. Figure 10illustrates the pressure distributions forsharkns7 at M=0.78 and a lift range ofCLwing

=0.18-0.34. At the design point CD,̂ is128 counts which is composed of Cowing-pom=77 counts and Cow^f = 51 counts. Note thatfavorable gradients exist on both upper andlower surfaces for about the first 30%-40%chord, depending on span location and liftingcondition. For reference, SYN107P runs inparallel under MPI, and optimization runs costabout 10X-20X the cost of an analysis. Recent

SYN107P benchmarks on a 16 processorAthlon 650 cluster shows the analysis runningin 30 minutes.

For those unfamiliar with inverse design, weshould point out that a pressure modificationvery near the leading-edge "stagnation" pointcould dramatically alter the leading-edgegeometry. This occurred in the above exercise,which yielded a leading-edge radiusdistribution that was not smoothly varying outthe span, yet this was expected. Hence, ourfinal task of phase IV was to re-establish anappropriate leading-edge radius distributionwithout really changing the wing pressuredistributions at the cruise design conditions.We tailored the wing's leading-edge geometryfor low-speed characteristics as well as withconsideration to the requirements needed forpromoting laminar flow. These modificationswere accomplished with local explicitgeometry perturbations. During this leading-edge modification step, Ahlstrom andMcDowell also slightly reduced the requiredwing thickness distribution. After both ofthese changes were incorporated into sharkns7,

Final Shark WingAirfoil Geometry

30.0 1

20.0 -

10.0-

o.o-

-10.0 -

-20.0

SYMBOL AIRFOIL ETA———————— Oudxnid ft».0———————— Ko« l».l

0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 80.0 90.0 100.0Percent Chord

Figure 11 - Final Wing Root and Tip Airfoils

15 of 18



Final Shark WingSpanwise Thickness Distributions

8.0

7.0-

6.0-

2.0-

i.o-

0.0

SYMBOL AIRFOIL—————— Sknfc

0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 80.0 90.0 100.0Percent Semi-Span

Figure 12 - Final Wing Half-Thickness Distribution

the final wing was analyzed to verify that thesegeometry changes did not adversely effect thepressure distributions. When overlaid on thesame plot, the curves nearly appeared as one.

The final wing geometric characteristicsresulting from the process described above areshown in Figures 11 and 12. Shown are theroot and an outboard airfoil along with thehalf-airfoil thickness distribution across thespan. As indicated in Figure 12, the maximumthickness averages around 12%. Overall, thesenumbers are not significantly different fromwhat Ahlstrom had originally assumed in hisconceptual layout of the G-Forceconfiguration.

The estimated benefit of laminar run isbetween 10 and 20 counts of drag dependingon the Mach number. On the lower surface, thepositive gradient allows a laminar run toapproximately 30 to 40% chord before theadverse gradient will cause transition. On theupper surface the shock will trigger transitionprovided any attachment line contaminationfrom the fuselage boundary layer is removed

by a notch-bump, and that the Ree does notexceed about 200. The leading edge Ree variesfrom approximately 125 just outboard of thefuselage to around 80 at the tip. The amount oflaminar run on the upper surface will increaseas Mach increases due to the shock moving afton the airfoil as well as the pressure gradientsbecoming more favorable. As a result, it isexpected that at race conditions the wing willhave an appreciable extent of laminar flowprovided that the surface of the wing is smoothand free of particulate contamination.

Clean wing CLmax, Cunaxcw, was also monitoredduring the design process to ensure the wingsatisfied the clean wing stall speeds requiredby Star Aerospace. The design requirementswas to provide a CLmaxcw of at least 1.6 at aMach=0.2. This was determined by solving theflow field about the wing with FLO22 at anumber of angle of attacks (approximately 15to 16 degrees in most cases) and interrogatingthe wing to find Cpmjn at various spanwisestations. When a critical Cp^ value (for thisReynolds number, Cp^ is approximately -15)is exceeded, CLmax is defined. The wing

16 of 18



developed for the G-Force exceeds the CLmaxrequirement slightly, having an estimatedCLmaxCW Of 1.64.

Tail Design —An initial look at the horizontal tail sizing asdetermined by Ahlstrom indicated a largerhorizontal tail was required, but afterestimating the fuselage contribution to stabilityusing CFD analysis instead of "cookbook"equations, McDowell & DeHaan determinedthat the tail was sized appropriately for thistype of aircraft. From an aft C.G. of 30%, theaircraft is estimated to be 25.2% stable power-off and 33.2% stable at 550 mph and fullpower. The Neutral Point is estimated to be at54.5% cmac- Directional stability indicates thatthe G-Force configuration has over twice thedirection stability of stock WWII fighters inuse at Reno. The estimated Cnp for thisconfiguration is conservatively -0.002 poweroff and -0.00255 at full power. Typical WWHaircraft have a Cnp of approximately -0.001 atnormal operating power. This should providean advantage for the G-Force aircraft as typicalwarbirds exhibit low directional stability athigh speed racing conditions.The final horizontal and vertical tails utilizemodified NACA 4, 63 & 64 series symmetricalairfoils with a thickness of just less than 15%for the horizontal tail and 13% for the verticaltail. The tail Mach capability with theseairfoils and with tail sweeps of 35 and 36degrees (horizontal and vertical tailsrespectively) is estimated to be M=0.84. Twoconcepts to compensate for prop swirl are stillbeing considered. One approach is to twist thetail for the estimated swirl distribution asdetermined by blade element theory. Defining

a twist distribution on a symmetric airfoil tailstill allows for a non-handed part by flippingthe tail left to right. A second approach issimple to rig a slab tail asymmetrically tocompensate for the swirl. Multiple tails arebeing considered for flight testing to determinethe optimal configuration.

Fuselage and Prop Face Mach -The local Mach number was evaluated at theprop disc plane (without the effect of the propitself) by DeHaan using a surface-panelmethod (References 8-9). The analysisindicated that the propeller disc will see aMach number that is 4% less than thefreestream Mach number. For comparison, aprop at the front of the aircraft would seeapproximately the same freestream Machnumber. Work was performed to try to furtherreduce the onset Mach number seen at theprop; some reductions were possible but at atrade of causing flow separation along thebody near the prop. The concept of reducingMach number at the prop disc by furtherfuselage shaping was dropped.

Drag Build-up -A drag build-up for this aircraft was completedand indicates that the profile drag estimate forthe G-Force during the conceptual phase isreasonable. Table 3 summarizes the estimateddrag of the aircraft at race conditions of550mph. Not included in this estimate is anylaminar run on the body or wing. Therefore,approximately 5% better performance thanthese estimates is likely. Note that at thisspeed, max L/D occurs at 0.49 CL(approximately a 7g case) and is estimated tobe 14.78.

Table 3 - Estimated Drop Build-un

ConditionCL

g'sCDOCra

CDCCotrim

CototL/D

6000', 550 mph0.07

10.019280.000250.000230.000100.01986

3.52

6000', 550 mph0.30

40.019280.004620.000540.000120.02455

12.22

17ofl8



ConclusionsIf Unlimited racing is to continue in the futureas a viable and exciting sport, newly designedrace planes are required to replace the aging,but beautiful, piston powered fighter aircraft ofthe 1940's. This paper describes thedevelopment of a new Unlimited Race plane,G-Force, designed to fly 550 mph laps atReno. The processes utilized to develop thisaircraft range from very simple to N-SOptimization tools. The combination of thesetools, used efficiently and where each tool canprovide meaningful results, under the eye ofexperienced engineers, rapidly developed aconfiguration that satisfied a set of welldefined goals. It is interesting to note howaccurately conceptual methods can be atproviding global layouts of such specializedaircraft as the G-Force. However, without theuse of optimization tools, such as SYN107Pthis wing design would not have been rapidlydeveloped nor would it likely have satisfied allthese requirements as well as it does.

AcknowledgementsThe wing design effort would not have been assuccessful as it was without the participationand contributions of Antony Jameson ofStanford University. Professor Jamesonsupplied consulting time, aerodynamic designand analysis software as well as computerresources through his firm, IntelligentAerodynamics, Int'l. The authors would alsolike to acknowledge The Boeing Company forallowing the aerodynamics team to work onthis project after hours.

References

[2] P.A. Henne and R.M. Hicks. WingAnalysis Using a Transonic Potential FlowComputational Method. NASA TM-78464, July 1978.

[3] A. Jameson, W. Schmidt, and E. Turkel.Numerical solutions of the Euler equationsby finite volume methods with Runge-Kutta time stepping schemes. AIAA paper81-1259, AIAA 24th Aerospace SciencesMeeting, Reno, NV, January 1981.

[4] A. Jameson. Solution of the Eulerequations for two dimensional transonicflow by a multigrid method. AppliedMathematics and Computations, 13:327-356,1983

[5] A. Jameson. The present status,challenges, and future developments inComputational Fluid Dynamics. Technicalreport, 77th AGARD Fluid Dynamics PanelSymposium, Seville, October 1995.

[6] A. Jameson. Optimum aerodynamicdesign via boundary control. In AGARD-VKI Lecture Series, Optimum DesignMethods in Aerodynamics, von KarmanInstitute for Fluid Dynamics, 1994.

[7] A. Jameson, L. Martinelli, and N. A.Pierce. Optimum aerodynamic designusing the Navier-Stokes equations.Theoretical and Computational FluidDynamics, 10:213-237, 1998

[8] J.L. Hess and A.M.O. Smith. Calculationof potential flow about arbitrary bodies.Progress in Aeronautical Sciences, 8,1966.

[1] A. Jameson. Transonic Potential FlowCalculations Using Conservative Form.Proceedings of AIAA 2nd ComputationalFluid Dynamics Conf., June 1975, pp. 148-161.

[9] J.C. Vassberg. A fast surface-panelmethod capable of solving million-elementproblems. AIAA paper 97-0168, 35th AIAAAerospace Sciences Meeting, Reno, NV,January 1997.

18 of 18

g-force - the design of an unlimited class reno...

Documents