Search

This project aims to develop a generic, adaptable device able to emulate the distinct aerodynamic wake characteristics corresponding to different closed-wheel automotive vehicle categories, avoiding the use of a full-body vehicle model(s). The intention is to reduce complexity in both wind tunnel experimental testing and computational simulation, enhancing the capability of experimental facilities to study interference effects between vehicles in close proximity (e.g. platoon formation, drafting) and allowing faster build, meshing and computation time for CFD studies. The overall approach uses both computational and experimental investigationson generic automotive configurations (e.g. DrivAer models), particularly in relation to wake development and detailed flow structures. CFD simulations are mainly used for preliminary flow analysis and aerodynamic design, while the wind tunnel established the experimental, benchmark datasets. The three standard DrivAer vehicle configurations were adopted for studies relating to passenger cars, while the proposed 4th configuration ('high-performance') was developed and proposed as a benchmark model for race car aerodynamics. Two extensive, experimental wake datasets on the four DrivAer configurations have been created, the first using a pressure-based technique (total pressure rake), and the second applying an image-based method (Stereoscopic Particle Image Velocimetry). Additional experimental analyses support each wake dataset, such as (i) investigation of the effects of experimental conditions, (ii) Reynolds number characteristics and (iii) statistical analysis .These automotive wake benchmarks were used as the basis for the development of a conceptual wake generator, which is intended to be adaptable and able to ii generate representative wakes equivalent to automotive bodies. The development and key aerodynamic characteristics of this device are outlined, and each configuration experimentally validated using wind tunnel data. The complex wake structures associated with each of the four vehicle configurations are categorised, and the resulting wake emulator is seen to be capable of reproducing the key features within the wake development region of interest (i.e.0.666 < x/W < 1.33, where x/W is the distance behind the vehicle normalised by the DrivAer model width, W). This thesis is concluded with the proposal of a wake generator device as a new tool to support further development on automotive aerodynamics.

Local air quality is one of the factors constraining the development of airports. In countries of the European Union where new, stricter regulations for emissions of nitrogen oxides (NOx) have been introduced since 2010, the limits of mean annual concentrations are already exceeded at certain ground monitoring locations of large airports. This research project investigates the possibility of abating the aircraft exhaust plume at take-off by placing an array of aerodynamic windbreaks (‘baffles’) in the runway end safety area close to the aircraft starting position. The undertaken experimental investigation comprised sub-scale wind tunnel tests and full-scale field trials with a BAe 146-301 aircraft, performing take-off and landing cycles at Cranfield Airport. The initial wind tunnel experiments investigated the effect of a solid baffle row, placed in the path of a buoyant nozzle jet, on the development of the plume downstream. Using flow visualisation, the positive effect of the baffle row of promoting buoyant rise of the plume away from the ground was demonstrated successfully without the presence of wind tunnel flow. The investigation highlighted the importance of the distance of the baffles relative to the jet source on their effectiveness. In the presence of wind tunnel flow, the baffles caused an increased vertical spread of the plume downstream, but the plume was not observed to separate from the ground. In preparation of the field trials, the spatial arrangement of the baffle array was investigated by means of wind tunnel drag measurements, performed with a skinfriction balance. The experiments focused on key parameters such as the baffle slope angle and row spacing, favouring a configuration of three rows of baffles of increasing height. Based on the wind tunnel measurements, full-scale baffle prototypes were designed and manufactured at Cranfield University and were deployed in the field trials. Lidar and point sampler measurements during the field trials suggested that the plume had risen away from the ground on one occasion when the aircraft was located close to the baffles. A positive effect was shown in terms of reduced concentrations downstream of the baffles. This result was not replicated when the aircraft was further away from the baffles. The subsequent wind tunnel experiments focused on replicating the field trials at 1:200 scale in Cranfield’s Atmospheric Boundary Layer Wind Tunnel. The aircraft was represented at sub scale with a single stationary nozzle while the jet speed and buoyancy were modelled using similarity parameters such as the Froude number and the ratio of ambient and jet density. Mean concentration measurements were performed using a Flame Ionisation Detector method releasing methane as tracer gas. The effect of the baffles was observed to be mainly local in terms of reduced concentrations close the ground due to their sheltering effect. A more prolonged effect was found to be the increase of the plume’s vertical spread resulting in an increase in mean concentrations away from the ground.

Most heavy goods vehicles in service today are fitted with add-on aerodynamic devices. The most common of which is the cab-mounted roof deflector. Such devices provide appreciable drag savings, however, they are often not optimised for the trailer. When a wind yaw angle is present, their savings also diminish as the yaw angle increases. The work conducted within this thesis investigated the possibility of using an adjustable deflector for active flow control. The optimum deflector height for a given trailer height was initially investigated using wind tunnel testing. The variation of this optimum with yaw angle and container separation was then investigated. From the results a 3D look-up table was generated. A control scheme was proposed that used the 3D look-up table requiring only three measurable inputs. The three inputs required were: the wind yaw angle, the container height and the container separation. A pressure differential located on the deflector was found to linearly relate to the wind yaw angle. This relationship allowed on-road measurement of the wind yaw angle and therefore enabled the development of a prototype controller. Extensive on-road testing and unsteady computational simulation were conducted. The results obtained indicated a mean yaw angle magnitude of around 5 perturbed by four fundamental low frequencies. These frequencies were identified in the runs conducted over the test period and an average frequency established. Higher frequency disturbances were attributed to the wakes of leading heavy goods vehicles and were filtered by a suitably chosen numerical filter. Finally, an estimation of the efficiency of the active device was made using a combination of simulation and full scale testing. From the results obtained, an optimised deflector generated an average drag reduction of 7.4%. An estimated additional drag reduction of 1.9% over the optimised deflector was predicted through use of an active system.

When a vehicle travels at hypersonic speeds during launch, cruise or atmospheric re-entry it is subject to extremely high surface flow temperatures. As well as on the vehicle forebody, extreme heating can take place close to surface protuberances which are almost impossible to avoid in a real flight vehicle. These disturbances interfere with the freestream flow and result in complex viscous interactions which induce a local heat flux augmentation that can become detrimental to the integrity of the vehicle. A greater understanding of these flow phenomena is required. This thesis develops the understanding of the behaviour of the flow around surface protuberances in hypersonic vehicles and presents an engineering approach to predict the location and magnitude of the highest heat transfer rates in their vicinity. To this end, an experimental investigation was performed in a gun tunnel at freestream Mach numbers of 8.2 and 12.3 and Reynolds numbers ranging from Reoo/m=3.35xl0 ⁶ to Reꚙ /m=9.35xl0 ⁶. The effects of protuberance geometry, boundary layer state, freestream Reynolds number and freestream Mach number were assessed. Further understanding of the flowfield was obtained through oil-dot visualisations and highspeed schlieren videos taken at frame rates of up to 50 kHz. Results show the local interference interaction is strongly three-dimensional and is dominated by the incipient separation angle induced by the protuberance. In subcritical interactions - in which the incoming boundary layer remains unseparated upstream of the protuberance - the highest heating occurs adjacent to the device. In supercritical interactions - in which the incoming boundary layer is fully separated ahead of the protuberance - the highest heating generally occurs on the surface just upstream of it. An exception is for low-deflection protuberances under low-Reynolds freestream flow conditions in which case the heat flux to the side is greater.

An investigation into the current methods employed to conserve vorticity in numerical calculations is undertaken. Osher’s flux for the artificial compressibility equations is derived, implemented and validated in Cranfield University’s second order finite volume compressible flow solver MERLIN. Characteristic Decomposition is applied as a method of vorticity conservation in both the compressible and artificial compressibility MERLIN solvers. The performance of this method for vorticity conservation in both these solvers is assessed. Following a discussion of the issues associated with application of limiter functions on unstructured grids three modified versions of the method of Characteristic Decomposition are proposed and tested in both the compressible and incompressible solvers. It is concluded that the method of Characteristic Decomposition is an effective method for improving vorticity conservation and compares favourably in terms of increased computational cost to vorticity conservation through grid refinement.

A research programme was initiated to examine three candidate high-lift technologies, which would, if implemented, simplify the mechanical complexity of the multiple component trailing-edge devices traditionally employed on civil transport aircraft. Experimental studies were undertaken with the aim of examining each technology in terms of its potential to favourably influence boundary layer development and improve the aerodynamic characteristics of a high-lift configuration. Preliminary studies of triangular serrated geometries, at the trailing edge of a modified flat plate, highlighted that the ability of the serrations to favourably influence the flow field development over an aft positioned single slotted flap was critically dependent upon the flap lap/gap and deflection angle. Under the test conditions, the serrations were most effective at low flap deflection angles, particularly serrations with a length corresponding to 13% flap chord. Extending these studies to a representative high-lift configuration significantly limited the range of flap laps/gaps and deflection angles over which the serrations were favourable. Furthermore, oil flow visualisation provided evidence of wake structures emanating from serration vertices, corroborating earlier hypotheses and suggesting the flow mechanism by which serrations favourably influenced boundary layer development over the upper surface of the downstream flap. Experiments indicated that when optimised, blowing tangentially from a slot at the trailing edge of the main element over the upper surface of a flap within a three-element high-lift configuration, provided a highly effective means of preventing boundary layer separation and increasing lift. This was corroborated by oil flow visualisation and computational simulations. Maintaining the same momentum coefficient and blowing through discrete orifices at the trailing edge of the main element, proved highly favourable, heightening the increment in lift in comparison to the corresponding tangential slot blowing configuration. Hence, the mass flow rate could be reduced in comparison to the tangential slot blowing configuration, without compromising the aerodynamic performance.

The potential for active control of low Reynolds number boundary layers using synthetic jet generators (SJG) has been established. The results from a four stage experimental study are presented in which the operational and geometric parameters of a rectangular slot choice SJG are optimised. A time-dependent, analysis of the SJG velocity profile is carried out in quiescent conditions prior to application of the SJG in (i) a nominally zero pressure gradient at plate boundary layer (l.54 > < l05 < ReX < 2.86 >

A combined experimental and theoretical study is described of the low speed flow over a highly swept and cambered wing that simulates the flow features of a transonic manoeuvre condition. The thesis is divided into two parts: Part I examines the research objectives from a customer perspective, with background information on the project history and funding sources. Since the research is aimed at improving the aerodynamic performance of low observable configurations, stealth technologies are discussed and their implications for combat aircraft wing flows. The management chapter of the thesis then discusses the influences affecting the decision making process for the acquisition of weapon systems in the UK. Part II describes the design of a highly swept and cambered wing that generates strong adverse pressure gradients near the trailing edge, leading to three-dimensional separations in this region. Using surface flow visualisation the nature of these flows is defined, indicating how the position of a separated streamline moves forward with increasing angle of incidence. These observations are confirmed by flow predictions using the SAUNA Computational Fluid Dynamics (CFD) method that solves the Reynolds Averaged Navier-Stokes equations, employing a two-equation turbulence model. The mechanism of the flow separation is also predicted using CFD, indicating that a separated stream surface reattaches at the wing trailing edge, forming a 'tunnel' of separated flow. To the authors knowledge this represents the first time that the main physical features of such a complex three-dimensional separated flow has been modelled using a CFD method. From an evaluation of the CFD methods employed, a design process has been proposed by which a wing designer can determine if wing flows over similar configurations remain attached. Additionally, the velocity magnitudes within parts of the separated shear layers and the wake are obtained using an optical non-intrusive measurement technique and give good agreement with the theory.

An experimental investigation has been carried out to study and understand the influence of an end plate boundary layer on half delta wing models at low Reynolds Number. The programme involved measurements in two facilities: a vertical water tunnel which was used for flow visualisation studies and a conventional closed working section wind tunnel for both flow visualisation and surface static pressure measurements. In both facilities dynamic and steady state or static measurements were made on half delta wing models with 55° and 70° sweep and varying thickness/chord ratio under the influence of a number of artificially generated end plate boundary layers. In both facilities, of all model configurations tested, for both dynamic and static test conditions, vortex burst was seen to move upstream, inboard and away from the wing surface as the angle of attack is increased and vortex core trajectory is seen to move towards the wing root, which is consistent with the findings of previous researchers. Vortex breakdown position is seen to move upstream, inboard toward the wing root and away from the wing surface as the end plate boundary layer thickness is increased. This is attributed to the influence of the interaction between the horseshoe vortex and the half delta wing leading edge vortex as a result of changes in the wall boundary layer thickness. In terms of vortex core trajectory, increases in end plate boundary layer thickness are seen to displace the vortex core towards the wing root. During dynamic tests an increase in wall boundary layer thickness is seen to suppress the hysteric behaviour of the vortex trajectory. Surface static pressure measurements at Reynolds Number of 479,000, during both static and dynamic tests, make it possible to see that the influence of changes in wall boundary layer thickness are small, often insignificant, at (x/c) locations greater than 0.45. This is consistent with an increase in wall boundary layer thickness promoting earlier vortex breakdown. Correlation between smoke flow visualisation (of both vortex breakdown and trajectory) and surface static pressure measurements, using the half-width of the suction peak as a parameter, was good. Differences between vortex characteristics in the water tunnel and wind tunnel were consistent with the influence of Reynolds Number.

When analyzing the aerodynamic characteristics of a road vehicle, the flow around the basic body shape is complicated by the presence of the rotating wheels. Even though on most vehicles the wheels are partially shrouded their effect on the flowfield is still considerable. Despite this, very little is understood about the flow around a rotating wheel. This thesis describes the development of a validated steady state Reynolds Averaged Navier-Stokes CFD model to investigate the flow around automobile wheels. As all the previous investigations into the aerodynamic characteristics of wheel flows had been experimental, preliminary computational studies were performed. The basis of these was the 2D circular cylinder. The effects of cylinder rotation and ground proximity were modelled, and strategies for boundary conditions and mesh topology were developed. This work was extended into 3D with the modelling of an isolated wheel, both rotating and stationary. Using existing experimental data for validation, an extensive investigation into the effects of solver numerics, symmetry planes, turbulence models, and the method of turbulent closure was performed. An optimum solver configuration was developed which comprised of the RNG k-E turbulence model with full boundary layer closure. It was accurately predicted that the rotating wheel generates less lift and drag than the equivalent stationary wheel. A number of postulated experimental flow features were captured in the final solutions. Using a parallel experimental study to provide further validation data, the CFD model was extended to incorporate an asymmetric shroud containing a wheelhouse cavity. The influence of the rotation of the wheel, the geometry of the shroud, and the thickness of the stationary groundplane boundary layer were investigated. The rotating wheel now produced more drag than the equivalent stationary wheel. Reductions in wheel drag were found with a reduction in the ride height of the shroud, and with the addition of spoilers to the lower front edge of the shroud. Increasing the stationary groundplane boundary layer thickness also reduced the wheel drag. The effects of these changes on the wheel surface pressure distributions are presented.

Crosswind gusts have a continuous influence on the ride and handling of road vehicles. At low speeds the effect is negligible but as both car and wind speeds increase there is a reduction in refinement, ride quality is degraded and it becomes tiring to drive. Future environmental legislation concerning the reduction of carbon dioxide emissions will lead to a lighter road vehicle and a corresponding increase in crosswind sensitivity. The aerodynamicist's approach to understanding the fluid flow around a vehicle when subjected to a crosswind has conventionally been through steady state model tests where aerodynamic force and moment data are taken for different yaw angles. The accuracy of this data has previously been questioned because of a lack of simulation of the transient nature of the crosswind gust. Additionally, although force and moment data can tell the aerodynamicist which are the principle loads influencing a vehicles response in a crosswind, they fail to identify the specific regions on the vehicle that contribute to these aerodynamic loads. This can only be achieved by pressure mapping the model surface and although such a technique has been employed during steady state tests, no research has been presented with the correct modeling of the transient crosswind gust. To gain an initial understanding of the complex time dependent and separated flow fields around bluff vehicles, such as sports utility vehicles, when subjected to a crosswind, aerodynamic force, moment and surface pressure data of simple geometric shapes has been collected on the Cranfield crosswind track facility. Steady state data has been obtained from conventional wind tunnel tests and compared with the transient data. Unique pressure animations identify the growth and collapse of vortices on the leeward face as the primary transient characteristic and which produce peak aerodynamic yawing moments up to double that seen in the steady state.

The results from a series of low speed wind tunnel tests on two half model highly swept wings (a symmetrical aerofoil section and a highly cambered aerofail section) are presented in order to examine the trailing edge flow separation mechanism and its development with wing sweep between 30' and 60'. The tests involved surface oil flow visualisation, smoke flow visualisation, surface static pressure and force balance measurements at streamwise chord Reynolds numbers from 1.5 x 105 to 5.2 x 106 and Mach number from 0.09 to 0.17. These results are used to assess two viscous-inviscid interaction CFD methods (BVGK and VFP) and two boundary layer methods (TAPERBL and WAKELAG) used to predict the flow over the highly cambered wing. A parametric study using cropped delta vane vortex generators in a co-rotating array was conducted on the 40' swept wing to investigate the effect of vane chordwise position, vane orientation, vane height relative to the boundary layer thickness and vane spacing on the prevention of the trailing edge separation. The performance of these flow control devices is assessed in terms of changes in; the wing surface flowfield, lift curve slope and the lift-dependant drag factor. In addition comparisons are made between the clean wing and flow control wing measured pressure distributions. The results and analysis show that the performance of the vortex generators is improved when the height of the vortex generator is approximately equal to that of the local boundary layer thickness and when the vane angular deflection to the local upstream flow direction is between 14' and 21'. The performance is also seen to depend on the vanes position ahead of separation and on the adverse pressure gradient to be restored and may also depend on a vane spacing made non-dimensional on the wing normal chord rather than the vane height. Similar performance improvements are observed with the wing swept to 50' using the positioning guidelines from this optimisation study. The performance of concave slats, canted cropped delta vanes, 'bent'wires and sub-boundary layer wires as vortex generating devices are seen to be not as effective as upright cropped delta vane vortex generators. To assist in the interpretation of the parametric vortex generator study a low speed wind tunnel technique is developed using shear stress sensitive liquid crystals to investigate the downstream development of vortices from cropped delta vane vortex generators. The results show that -- i) submerged vortices have less influence on the surface flow with downstream distance than vortices closer to the edge of the boundary layer, and ii) the primary increase in skin ffiction arises in the flow adjacent to the upflow side of the vortex. This area increases with vortex size. The results from this research programme are finally shown to be applicable in two market areas. The first is as a performance improvement on current highly swept winged military aircraft and the second is as flight controls on future aircraft from making the vortex generating devices active. The possible customers in these two areas are identified and marketing strategies developed for each case.

An experimental parametric sturdy of vane and air-jet vortex generators in a turbulent boundary layer has been carried out. Experiments were carried out in two facilities, one with a free-stream velocity of 20 m/s and a boundary layer thickness (6) of 41.5 mm, and one in a high speed facility at free-stream Mach numbers of between 0.45 and 0.75 and a boundary layer thickness of 20 mm. Cross-stream data were measured at a number of downstream locations using a miniature five-hole pressure probe, such that local cross-stream velocity vectors could be derived. Streamwise vorticity was calculated using the velocity vector data. In the low speed study, vortex generator parameters were as follows: ' Vane vortex generators: thin rectangular vanes with a vane aspect ratio of unity (2h/c = 1), free-stream velocity 20 m/s, incidence (cc = 10', 15', 18', 20'), height-to-boundary- layer- thickness-ratio (h/8 0.554,0.916,1.27,1.639), and strearnwise distance from the vortex generator (x/6 = 3.855,12.048,19.277,26.506). ' Air-jet vortex generators: circular jet nozzles, free-stream velocity = 20 m/s, jet nozzle pitch and skew angles (cc, P= 30', 45', 60'), hole diameter-to-boundary-layer-thickness-ratio (D/5 = 0.098,0.193,0.289), jet-to-free-stream-velocity ratio (VR = 0.7,1.0,1.3,1.6,2.0), and strearnwise distance from the vortex generator (x/8 = 3.855,12.048,19.277,26.506). In the high-speed study, the vortex generator parameters were as follows: Vane vortex generators: thin rectangular vanes with an aspect ratio of unity, incidence ((X 1505 20'), he i ght-to- boundary- I ayer-th i ckne s s-rati o (h/8 = 0.75), strearnwise distance from the vortex generator (x/6 = 8.755 16.25,23.75), and free-stream Mach numbers of 0.45,0.6 and 0.75. Air-jet vortex generators: jet pitch ((x = 30', 45'), jet skew angle (P = 30', 45', 60'), hole diameter-to-boundary-layer-thickness-ratio (D/8 = 0.15,0.3), j et-to- free- strearn-ve loc ity ratio (VR = 1.6), and strearnwise distance from the vortex generator (x/6 = 8.75,16.25,23.75, 31.25), and free-stream Mach numbers of 0.50,0.6 and 0.75. Streamwise vorticity data from the experiment was used to generate prediction techniques that would allow the vorticity profiles, downstream of vane or air-jet vortex generators, to be predicted. Both techniques are based on the approximation of the experimental cross-stream vorticity data to Gaussian distributions of vorticity through the vortex centre. The techniques, which are empirically derived, are simple equations that give the peak vorticity and vortex radius based on the vortex generator parameters. Use of these descriptors allows the assembly of the Gaussian vorticity equation. Both techniques are compared with the experimental data set and were seen to produce peak vorticity results to within 12% and 20% (for the vanes and air-jets respectively), 15% for the radius of the vortex, and 15% and 20% in vortex circulation (for the vanes and air-jets respectively). The two simple prediction techniques allow good prediction of the vortex structure at extremely low computational effort.

The crosswind sensitivity of a car is described as the sensitivity of a driver-vehicle system to vehicle movements provoked by wind influences. Research has shown that it is a contributory factor in accidents and that it is adversely affected by certain body shapes and by reductions in weight. Future legislation calling for a reduction in the amount of Carbon Dioxide produced by cars may well lead to this reduction in weight. Tests have been carried out on llth scale car models to compare the different results produced by static and dynamic tests. Although static tests are easier to perform, dynamic tests, in which the model is propelled across the wind tunnel, offer the advantage of an improved simulation of the skewed profile that the vehicle encounters at full scale. In these experiments a number of different skew profiles, turbulence profiles and model configurations including estates, hatchbacks and saloons, were tested at Reynolds numbers above 4x10⁵. The results showed that data from the more simple static tests were generally more conservative than those from the dynamic tests in a skewed profile up to approximately 30° of yaw. However, the static tests were unable to predict the peak yawing moment that occurs as a result of the transient flow over the model as it enters the gust. The dynamic tests also indicated oscillations in the flow over the rear end of the models in the hatchbacks with backlight angles of 22° and 36°, although less so in the latter. Other results suggested that there was an improved correlation between lift coefficients from different skew profiles, if the mean local velocity integrated over the height of the model was used to non-dimensionalise the data. Quality Function Deployment is not an appropriate framework to be used in the development of a complete vehicle but would prove useful if applied to the issue of crosswind sensitivity as a sub-component.

Motor car crosswind stability can be adversely affected by reductions in both vehicle mass and drag coefficient. As these are two likely results of future developments the importance of research into vehicle aerodynamic stability is set to increase, moreover, there is evidence that transient effects will be the critical. An experimental facility has been designed and constructed and tests have been carried out to investigate the implications of simulating dynamic flow-fields. Vehicle models of approximately 1/6th scale have been propelled along a test track, in the laboratory, to pass through a simulated crosswind gust of variable resultant yaw angle. Force and moment measurements have shown the aerodynamic inputs to be highly repeatable, though the technique has been restricted somewhat by the presence of mechanical "noise". Additional dynamic yaw experiments were conducted on a bluff-body model mounted in the College of Aeronautics' Oscillatory Facility. In some ways this technique is not as realistic as the Crosswind Track in its simulation of the full scale flow, however, despite its simplicity valuable aerodynamic data was derived from this test. Quasi-static tests have also been conducted and demonstrate that for certain model configurations a clearly defined yaw angle range exists where two different wake flow-structures are possible. At any given yaw angle, the dominant structure is determined by the flowfield history - essentially the direction in which the model is moved. This causes hysteresis in the forces and moments generated. In such a situation the flow is referred to as being bi¬stable. Both track and dynamic yaw tests indicate that the bi-stable flow phenomenon, witnessed in quasi-static experiments, can influence the dynamic forces and moments measured on a model. The flow structures associated with bi-stability are viscous-dominated and the slow development of viscous loads can be an important feature. It is possible that various vehicle configurations could induce bi-stable flow. If such flow behaviour is apparent then quasi-static forces and moment measurements will not provide an adequate engineering estimate of the transient aerodynamic loads. In this event it is imperative that the automotive engineer conducts investigations into the vehicle's dynamic performance.