Aerodynamic Development of the New Jaguar XF Adrian Gaylard Jaguar Land Rover, Aerodynamics Abstract The new Jaguar XF represents the latest instalment of Jaguar's progress towards a more contemporary and relevant style language, first seen in the XK. It is also the first Jaguar to have a drag coefficient (CD) of less than 0.3; which makes it the most 'aerodynamic' Jaguar to date. This paper describes the challenges posed by both the new style and the 29 month time-scale required for the complete project. To address these challenges Jaguar employed a radical aerodynamics development process, which used CFD to drive early design decisions and then shifted focus to wind-tunnel testing of a full scale multifunctional ‘buck’, to finalise and refine the aerodynamics and related attributes. 1 INTRODUCTION

The XF replaces the S-Type in Jaguar’s product line-up, as well as utilising its well regarded platform. It is, however, radically different: aiming to fuse the style and performance of a sports car with the space and refinement of a luxury saloon. The sports car heritage is seen in the generally aggressive stance; epitomised by large 'haunches' over the rear wheels along with an upright and aggressive front grille. This styling poses significant aerodynamic challenges. In addition, maintaining competitive performance whilst compensating for an increased frontal area, needed to enhance passenger accommodation, required a drag coefficient (CD) lower than the outgoing model. This was a major objective of the aerodynamic development programme. 2 OUTGOING MODEL

The S-Type is a full-sized saloon with ‘notchback’ style, rounded rear geometry. It also has a relatively steep backlight and sloping boot deck; as can be seen from the side-view shown in Figure 1.

Figure 1 Side View of a 2006MY Jaguar S-Type

These features contribute to a drag coefficient (CD) of 0.31, for the 2.7L diesel variant, and lift coefficients of 0.04 (CLf) at the front axle and 0.16 (CLr) at the rear. (Based on a 2.27 m2 frontal area)

Figure 2 provides an overview of the structure of the flow around the S-Type, based on a CFD simulation utilising Exa PowerFLOW. Surface contours of local (non-dimensionalised) drag force highlight the near-wake as a key source of drag (a). This ‘notchback’ wake structure [1] is also seen in the surface streamlines (b). Flow separates off the backlight and rear pillars developing a complex vortex structure in the ‘notch’; along with a rear-pillar vortex. In this particular case, it would seem that the flow separations arise preferentially from the flow around the rear pillars and not at the roof trailing edge. (For further details, including observations on wake asymmetry, see Gaylard et al. [2].)

Figure 2 Plan View of a 2006 Jaguar S-Type. CFD simulation of (a) surface contours of local drag (Red indicates positive drag and blue shows regions of negative drag i.e. thrust. -0.15[ CX [ 0.15) and (b) surface streamlines.

3 THE XF

The geometry of the new XF is illustrated in Figure 3. This shows a car that incorporates a number of (potentially) aerodynamically beneficial features: a well raked screen, cambered roofline, glass house tumblehome, reasonable front tyre-face coverage, squared-sills and a ‘fastback’ style rear with a sharp boot deck trailing edge condition.

Thus a CD target of 0.29 for the 2.7L diesel variant (Af = 2.33 m2), though competitive, might not appear to be particularly challenging. In a similar vein, the implied drag coefficient reduction target (∆CD) relative to the outgoing model of 0.02 would appear to be modest.

Figure 3 Line drawings of the XF from the front and side.

(b)

(a)

First, as we shall see, not all of these features were included in the initial styling concept. Second, some XF design features pose a significant challenge: large haunches, increased track and overhangs, larger cooling apertures (with enhanced cooling flows), bigger wheels, lower sump and re-designed exhausts. It was estimated that an S-Type with these features would have a CD of 0.34. This implies the actual ∆CD required was 0.05, a much more demanding task; particularly as XF’s under-floor and drive train were to be similar to those of the S-Type. Likewise, the change from a classic ‘notchback’ to a shape approaching a ‘fastback’ would appear to make the task of providing both lower total lift and an improved lift balance between the axles straightforward. However, boot deck height and length, along with trailing edge shape are very sensitive parameters – both from an aerodynamics and styling perspective. It required careful design optimisation to achieve the target CLf and CLr values of less than 0.1.

0.00

0.05

0.10

0.15

0.20

0.25

0.30

0.35

XF S-Type XK XJ XJ220

Car Line

For

ce C

oeffi

cien

t

1.80

1.90

2.00

2.10

2.20

2.30

2.40

Pro

ject

ed F

ront

al A

rea

/m²

Drag Coefficient Lift Coefficient Front Axle Lift Coeffiicent Rear Axle Lift Coeffiicent Projected Frontal Area

Figure 4 Comparison of principal aerodynamic parameters for Jaguar Vehicles As shown in Figure 4, the outcome of this programme is the most ‘aerodynamic’ Jaguar yet, in terms of drag coefficient – the first with a CD below 0.3. It is clearly a step forward compared to its immediate predecessor, both in terms of drag and the balance of lift forces. This was achieved using a re-modelled aerodynamics development process. 4 THE AERODYNAMICS DEVELOPMENT PROCESS

4.1 Relationship With The Product Development Proce ss The approach generally followed in developing a vehicle is illustrated by Figure 5. This maps the aerodynamic toolset onto general programme timing gateways. During the initial pre-programme (AP) phase, concept studies are performed using a combination of expert judgement, CFD and (on occasion) wind-tunnel tests of full scale concept clay models. This toolset runs through into the programme as it passes the ‘Programme Start’ (<PS>) gateway. During this phase multiple themes are evaluated and progressively

(a)

eliminated from consideration. Around the ‘Program Strategy Confirmed’ (<PSC>) gateway a single theme is selected and a full scale foam buck may be developed; depending on the body style and scope for change. By the ‘Programme Target Compatibility’ (<PTCC>) gateway a full-scale test property (‘aerobuck’) is generally available and the balance of aerodynamic development shifts to favour the full scale wind tunnel. From the Final Data Judgement (<FDJ>) gateway onwards there is little scope for significant change. After <FDJ> a limited number of Validation Prototypes (VP) are tested in the wind-tunnel. This is mostly for monitoring and correlation purposes. Before the vehicle is launched (<J#1>) production tooled (1PP) and evaluation units (FEU) are also tested to ensure that programme targets have actually been met.

PHASE

<FDJ> <J#1>

TO

OLS

AE

RO

DY

NA

MIC

S

AP <PTCC><PSC><PS> PTC

Concept Theme Development

CFDTheory / Knowledge Based

FSWT GRP Buck

Concept Studies C

ED

FG

BA

ADF

F (Upper)F (Under)

VP

F (Over-check/Correlation)

1PP/FEUFSWT Foam Buck Concept CFD (Ranking)

Concept Clays

Figure 5 Jaguar Land Rover’s Generic Aerodynamics Development Process

Map The bulk of the aerodynamics development process combines CFD with full-scale wind-tunnel testing. The XF programme represents a significant milestone for Jaguar: for the first time a new product was designed without the use of reduced-scale wind-tunnel testing; this being replaced by CFD. 4.2 CFD A strategic decision was taken by Jaguar to move away from reduced-scale model testing. This was prompted, in part, by the variability seen in the results from available model wind-tunnels (MWT) along with improved CFD correlation and the extra insight provided by this approach.[3] However, during the XF programme the primary focus of the CFD simulations was to provide directional guidance and estimates of the implications of design actions for drag and lift, rather than necessarily absolute figures. As the programme progressed the CFD method was demonstrated to be capable of delivering worthwhile aeroacoustic simulations [4]. Therefore Jaguar’s Wind Noise team completed a limited programme based on aerodynamics’ models [5] providing further value from pursuing the computational route. All CFD simulations are conducted using the Exa PowerFLOW software.

4.3 Full Scale Wind-Tunnel Testing Approximately 18 months into the programme a multi-functional aerodynamics ‘buck’ (‘aerobuck’) was fabricated. This was constructed from fibre-glass upper surfaces built onto a frame that mounted mated it to an S-Type ‘go-cart’. This enabled a new styled surface to be developed on a realistic floor, with a complete engine bay. The bulk of the development work was carried out using the well-known MIRA Full Scale Wind Tunnel (FSWT) [6]. This is a closed test-section open-return tunnel. It is a fixed-ground facility with no floor boundary layer control. It has a 35m2 test section and has a maximum test speed of 130 km/h. A limited programme of moving-ground development, principally to develop the aerodynamic features of the under-floor was carried out using the Volvo [7] and Pininfarina [8] facilities. As the programme progressed the ‘aerobuck’ was updated so that other attributes could be developed or evaluated. These included cooling airflow, aeroacoustics and wiper performance along with water and dirt deposition. 5 XF DEVELOPMENT

The starting point for the programme was set by a combination of styling aspirations, engineering constraints and aerodynamics advice. Early releases of styled surfaces were assessed using CFD simulations. Empirical corrections were used to account for features not present in the model (e.g. under-floor componentry, cooling flows, panel gaps and shut-lines). Such approximations are not ideal, so as the programme progressed more detailed CFD models were developed; incorporating more realistic geometric features thus reducing the number of corrections required. This approach revealed that the cognizance taken, by the designers, of ‘best practise’ aerodynamic guidelines along with the desire for more contemporary styling cues, had mitigated the ‘worst case’ starting point for CD of 0.34 to just under 0.33 (Figure 6). However, the rear geometry of this design still reflected the S-Type’s ‘notchback’ rear style.

Figure 6 Geometry used for the evaluation of an early XF Theme shown in (a)

side, (b) plan, (c) front and (d) rear elevations.

(a)

(b)

(c) (d)

From this point the programme aerodynamicists devised proposals to close the gap between estimated vehicle status and target. These proposals were evaluated using CFD, negotiated and refined with the designers and built into a ‘route map’ demonstrating what would be required for a particular theme to meet the target.

0.270

0.280

0.290

0.300

0.310

0.320

0.330

Them

e M

- Bas

e

+ Fro

nt B

umpe

r #1

+ Sill

Cover

age

+ Fro

nt P

lanfo

rm #

1

+ Rea

r Dec

klid

Raise

#1

+ Bac

kligh

t Ang

le #1

+ Rea

r Plan

form

#2

+ Hea

dlam

ps B

lende

d

+ Scr

een

Rake

Incr

ease

d

+ Rea

r Ligh

t Trip

s

Figure 7 A Typical CFD-Based Theme Development Cycle As the programme progressed more detailed CFD models were used – incorporating a realistic under-floor, powertrain, drivetrain and wheel styles. The CFD programme was, notwithstanding its limitations, able to provide the vehicle programme with a set of design directions that played a major part in the vehicle ultimately meeting its aerodynamic targets. For example, CFD simulations were able – at a very early stage - to demonstrate the benefits of a shallower wind screen rake Figure 8 shows iso-surfaces of total pressure deficit for (a) baseline and (b) increased rake configurations. Increasing screen rake has reduced the size of the a-pillar vortex, which benefits both drag and wind- noise. As has been demonstrated, the CFD programme focused mainly on drag reduction. Guidance was given on lift

(b)

Figure 8 The Influence of screen rake on the a-pillar vortex.

(a)

balance, but the results were known to be less robust. The effect of yawed onset flows and the aerodynamic implications of cooling flows were not assessed during this phase. (It should be noted that other CAE methodologies were employed by Jaguar’s cooling management team to ensure that cooling flows delivered the required thermal performance.) Therefore, the transition to the full-scale wind-tunnel programme was vital to delivering the overall aerodynamic performance. Once the ‘aerobuck’ was available aerodynamic changes suggested by the CFD programme were validated. As the design moved forwards modifications could be tested in larger numbers. The experimental programme also delivered more robust estimates for lift forces and the effect of cooling flows on vehicle aerodynamic performance. Finally, it also supported other related attribute development (e.g. cooling airflow, wiper systems, soiling and water management.)

Figure 9 Verification Prototype testing in the MIRA Full Scale Wind-Tunnel. However, CFD still played a valuable supporting role: allowing detailed flow field assessments to support the wind-tunnel development programme. Balancing the use of CFD and wind-tunnel methods during the development of the XF allowed key aerodynamic changes to be identified at an early stage, developed and implemented in a robust manner. The following sections illustrate the key changes that account for the improved aerodynamics of the XF, compared to the S-Type. This is done using data extracted from the relevant CFD simulations.

6 DESIGN IMPROVEMENTS

As a result of the development programme, the XF is considerably improved over the S-Type in a number of key areas (Figure 10). An obvious area in which the outgoing model could be improved was the exposure of the front tyre faces to the airflow. The lateral pull-in of the lower front corner on the S-Type directs air flow onto the front face of the tyres, resulting in significant pressure drag. By comparison, the lower front corner on the XF has been pulled outboard and an improved front wheel deflector is also included in the design. This results in a lower drag on the tyre front faces.

Figure 10 A comparison of the local drag forces acting on the front of the XF and

the S-Type. (Red indicates positive drag and blue shows regions of negative drag i.e. thrust. -0.15[ CX [ 0.15)

It is also clear that other features also contribute to the XF’s lower drag coefficient: the outboard front corners chin and door mirror. The improvement seen on the outboard corners of the XF relative to the S-Type is the result of careful development of the plan shape of the front corner (‘planform’); pulling the fender front corner rearward to provide an angled surface which starts to ‘turn’ the flow around the front corners of the vehicle.

S-Type XF

Figure 11 Comparison of the local drag forces acting on (upper) S-Type and

(lower) the XF, in plan view. (Red indicates positive drag and blue shows regions of negative drag i.e. thrust. -0.15[ CX [ 0.15)

Further improvements can be seen on the screen; as a result of increased screen rake (Figure 11). The shallower backlight and raised boot deck moves the XF away from the classic ‘notchback’ shape of the S-Type towards a ‘fastback’ style. This has resulted in lower drag forces over the backlight and boot deck - i.e. lower base drag (Figure 12).

Figure 12 A comparison of the local drag forces acting on the rear of (a) the XF

and (b) the S-Type. (Red indicates positive drag and blue shows regions of negative drag i.e. thrust. -0.15[ CX [ 0.15)

S-Type

XF

S-Type XF

On the S-Type, flow separation over the backlight and (rear) c-pillar induces significant drag compared to the XF. In the latter case a relatively modest drag increment is seen, due to the c-pillar vortex. Similarly high drag levels are seen on the rear face of the S-Type’s boot and bumper; due, in part, to the formation of relatively strong vortices in the wake. On the underside of the vehicle improvements to both ‘form’ and hardware have paid dividends. The redesigned front bumper undercover (valance) and chin serve to reduce flow separation under the front of the vehicle. The tuning of the shape in this region highlights one of the benefits of CFD: being able to acquire flow visualisation in a relatively inaccessible region of the vehicle. This is demonstrated in Figure 13.

Figure 13 Computed surface flow streamlines over the underside of (upper) the S-

Type and (lower) the XF. In addition, the XF has an improved system of under-trays: As seen in Figure 13 - (a) an extended valance; (b) a lengthened engine under-tray and (c) a new tunnel-tray. In order to both reduce drag and rear lift a boot deck ‘flip’ was incorporated into the design. The XF also features a shallower backlight angle and raised boot deck, compared to the S-Type. These two features combine to reduce the effective backlight angle; this also helps reduce drag and rear lift forces. 7 DISCUSSION

This programme has been the first new Jaguar developed without the aid of reduced scale model wind-tunnel testing. In many ways, replacing this development tool with CFD was a bold step. This approach did not match the number of individual configurations that could have been tested in a typical model wind-tunnel programme. It also delayed the evaluation of the vehicle at yaw. However, the relative richness of the CFD data set did permit changes to be proposed based on a detailed consideration of the flow structure, as well as any changes in the overall aerodynamic coefficients.

(c) (a) (b) S Type

XF

The result of this was that by the time the project progressed to the full scale wind-tunnel a clear route to target had been established. The wind-tunnel programme was able to verify that this had been the correct design direction, for drag in particular. Even so, lessons were learnt as the programme progressed – particularly with regards to the use of CFD.

Figure 14 Backlight surface flow visualisation. (a) Flow is attached over the

complete backlight and boot deck in experiment (wool tufts). (b) Flow separation is seen in the CFD simulation as a result of vorticity convected from the a-pillar vortex (*).

The correlation between experiment and CFD obtained on Jaguar vehicle before this programme had mostly focussed on the ‘classic’ Jaguar style language, typified by the S-Type. The near-wake flow structures for this style of vehicle are complex (actually asymmetric). This had obscured an over-production of vorticity in the a-pillar vortex being convected downstream, merging with the near-wake over the backlight. The change to a ‘fastback’ rear geometry meant that this solver artefact was no longer obscured: the CFD simulations appeared to show (weak) flow separation on the backlight of the XF when none was seen in the wind-tunnel. This is shown in Figure 14. The results of wool-tuft flow visualisation are contrasted with CFD simulation. The image for the simulation combines iso-surfaces of λ2 [9] (to capture vortex cores) with surface streamlines. A vortex core shed from the a-pillar clearly feeds into the spurious flow separation on the backlight. Refinements to the spatial resolution strategy adopted in the CFD simulations improved, but did not entirely remove, this solver artefact. This highlights the need for attention to the expected flow structures when assembling a CFD correlation set: changes in design language can lead to dramatically different flow structures,

(a) (b)

(*)

especially in the wake, effectively moving the simulation process outside the established ‘correlation space’. Following this programme improvements have already been made in the CFD process. These have included: using detailed models from the outset; improved spatial resolution strategy and improved model through-put by utilising a larger number of parallel CPUs per simulation. Future improvement to the CFD process will focus on more robust prediction of lift forces; interference drag due to cooling flows and providing simulations for yawed onset flows. The new Jaguar Land Rover aeroacoustics process will also deploy CFD more fully. 8 CONCLUSIONS

The Jaguar XF is the most ‘aerodynamic’ Jaguar to date – in terms of its drag coefficient (CD). Its aerodynamics have been developed using a radical process that replaces reduced-scale wind-tunnel testing with CFD simulation; before moving into a full-scale wind-tunnel test programme using a multi-functional aerodynamics test property (‘aerobuck’). The richness of the CFD dataset has been found to compensate for a reduction in the number of configurations that can be evaluated. Whilst the CFD method provided generally good directional indication (and insight) caution is still required. This is particularly the case when correlation datasets are predicated on the presence of particular flow structures which are a direct function of a particular design language. Therefore, care needs to be taken to ensure that correlation cases match the general flow structures produced by the vehicle under development. Using CFD alongside a wind-tunnel programme affords the opportunity to gain a much deeper level of insight into required design changes, whilst compensating for any weaknesses inherent in computational simulation. This synergy produces a more effective development process. The use of CFD from the early stages of development programmes also, increasingly, supports the reduction of wind-noise sources in parallel with the mainstream aerodynamic optimisation. ACKNOWLEDGEMENTS The author would like to thank Jaguar Land Rover for granting permission to publish this work. Also the author would like to point out that this paper is reporting the work of a number of colleagues, past and present, particularly: Paul Beckett, Ian Anderton, Joe Edge, Andy Sheppard and Jeff Howell.

REFERENCES 1 Gilhome BR, Saunders JW, and Sheridan J, “Time Averaged and Unsteady

Near-Wake Analysis of Cars.” SAE Paper 2001-01-1040, Vehicle Aerodynamics Design and Technology, March 2001, ISBN 0-7680-0747-X.

2 Gaylard AP, Howell JP and Garry KP, “Observation of Flow Asymmetry over

the Rear of Notchback Vehicles.“ SAE Transactions, Journal of Passenger Cars: Mechanical Systems, 6 (116), 2007-01-0900, pp. 993-1004.

3 Amodeo, J “The Development of CFD as a Primary Design Tool at Jaguar

Cars.” Fifth MIRA International Conference On Vehicle Aerodynamics, 13-14 October 2004, Heritage Motor Centre, Gaydon, UK.

4 Gaylard, AP, “CFD Simulation of Side Glass Surface Noise Spectra for a Bluff

SUV.” SAE Transactions, Journal of Passenger Cars: Mechanical Systems, 2006-01-0137, pp. 99-115.

5 Freeman, CM and Gaylard, AP “Integrating CFD and Experiment: The Jaguar

Land Rover Aeroacoustics Process.” 7th MIRA International Conference on Vehicle Aerodynamics, 22-23 October 2008, The Ricoh Arena, Coventry, UK.

6 Newnham P, Passmore M, Howell JP, and Baxendale AJ “On the Optimization

of Road Vehicle Leading Edge Radius in Varying Levels of Freestream Turbulence.” SAE Transactions, Journal of Passenger Cars: Mechanical Systems, 2006-01-1029, pp.994-1003.

7 Sternéus J, Walker T, and Bender T “Upgrade of the Volvo Cars Aerodynamic

Wind Tunnel.” SAE Transactions, Journal of Passenger Cars: Mechanical Systems, 6 (116), 2007-01-1043, pp. 1089 – 1099.

8 Cogotti, A, “The New Moving Ground System of the Pininfarina Wind Tunnel.”

SAE Paper 2007-01-1044, Vehicle Aerodynamics 2007, SP-2066, April 2007, ISBN 978-0-7680-1856-1.

9 Jeong, J. and Hussain, F “On the identification of a vortex.” Journal of Fluid

Mechanics, (1995), 285, pp. 69-94.