Application Of Flow Control To A Disc-Wing UAV
APPLICATION OF FLOW CONTROL TO A DISC-WING UAV
Jonathan R. Potts & William J. Crowther
Fluid Mechanics Research Group, School of Engineering,
University of Manchester, U.K.
ABSTRACT
A spin stabilised axi-symmetric disc-wing has potential application as a highly manoeuvrable, unpowered unmanned air vehicle or guided projectile. This paper considers the means by which aerodynamic control moments can be generated on such a disc-wing. The present experimental investigation outlines the aerodynamics of a non-spinning axi-symmetric disc-wing with and without forced transition strips installed on the disc rim. Fences fixed normal to the surface and the local flow direction provide promising control moments of useful magnitude. It is proposed that the control moments produced by passive transition strips or flow obstructing fences on a non-rotating disc-wing are achievable by a similar active flow control method such as active turbulent strips or on-demand fences. A simple analysis of disc-wing flight manoeuvres predicts the turning radius and time period of a 90° banked turn based on experimental data.
BIOGRAPHIES
Jonathan Potts (jonp@fs1.ae.man.ac.uk) studied mathematics as an undergraduate at the University of Manchester, U.K. (1994-98) gaining experience in a varied cross-section of the discipline, with special interest in applied mathematics. A grounding in fluid mechanics followed exciting his mind with the concept of fluid motion. Project work on Ocean Waves, Boomerangs and Tornado-like Vortices developed an interest in research. He chose to continue his education at the University of Manchester as a postgraduate in the Division of Aerospace Engineering since 1998. He is currently investigating the aerodynamics and control of disc-wings using experimental techniques, based at the experimental fluids and aerodynamics facility, the Goldstein Research Laboratory.
Bill Crowther (w.j.crowther@man.ac.uk) has an undergraduate degree in Aeronautical Engineering (1990) and a PhD in High Angle of Attack Aerodynamics from the University of Bath, U.K. (1994). He was at the Fluid Power department of the University of Bath from 1995 to 1997 working on the application of neural networks to fault diagnosis of hydraulic systems. He has been a lecturer at the School of Engineering, University of Manchester, U.K. since 1997 where his research interests include modelling and control of non-linear flight vehicles, flow control and neural networks.
LE - Leading edge
TE - Trailing edge
AoA - Angle of attack (°)
AdvR - Advance ratio (ΩπDd / V)
AR - Aspect ratio
Lf - Lift (N)
CL - Lift coefficient
CD - Drag coefficient
CDo - Profile drag coefficient
CM - Pitching moment coefficient
CR - Rolling moment coefficient
∆CL - Change in lift coefficient
∆CD - Change in drag coefficient
∆CM - Change in pitching moment coefficient
Dd - Disc diameter (m)
Rd - Disc radius (m)
Td - Disc thickness (m)
g - Acceleration due to gravity (ms-2)
H - Angular momentum (Kg m2 s)
I - Moment of inertia (Kg m2)
Md - Mass of disc-wing (Kg)
q - Precession rate (rad s-1)
Rc - Radius of circular flightpath (m)
S - Surface planform area (m2)
Re - Reynolds number
T90 - Time taken for a 90° turn (s)
V - Wind velocity (ms-1)
V1,V2 - Trailing vortices
x,y,z - Roll, pitch, yaw axes
L,M,N - Rolling, pitching, yawing moments (Nm)
p,q,r - Rates of roll, pitch, yaw (rad s-1)
ρ - Density of air (Kg m-3)
ωo - Angular velocity (rad s-1)
ωs - Spin rate (rad s-1)
Ω - Spin rate (Hz)
Introduction
Disc-wing based flight vehicles fall into two distinct categories:
- non rotating, non axi-symmetric body
- spin stabilised, axi-symmetric body
The first type will typically have a conventional airfoil cross-section when viewed from the side with a rounded leading edge and sharp trailing edge and thus defined flight orientation. This type of vehicle has the characteristics of a flying wing aircraft with low aspect ratio and as such is relatively conventional. The second type of flight vehicle by definition has an airfoil section with fore and aft symmetry and a centre of gravity at the centroid of the disc. This configuration will typically be unstable in pitch and for practical purposes must be inertially stabilised by spinning. Such a disc has no predefined flight orientation and offers potentially novel flight characteristics.
The concept of a circular planform flying wing (XF5U) was developed during the 1930s & 40s based on a three dimensional circular wing with cross- section based on the Clark Y family of airfoils (1). In 1972 the U.S. Navy commissioned a project to further the development of a spin stabilised self-suspended flare (2,3), which was essentially an axi-symmetric cambered wing with circular planform. Recent work by the authors outlined the aerodynamics of a spin- stabilised, axi-symmetric Frisbee-like disc-wing configuration (4,5). Independently Higuchi et al (6) investigated the flow over a similar disc-wing using smoke wire flow visualisation and PIV (particle image velocimetry) measurements aiming to provide information for possible UAV (unmanned air vehicle) applications.
Some practical applications for a disc-wing UAV of the spin-stabilised type are discussed below: The current development in small scale propulsion technology for UAVs and MAVs (micro air vehicles) provides challenges which will take time to fully overcome. A more short term application for a disc-wing UAV would be to use assisted propulsion, launching a disc-wing UAV from larger airborn vehicles or as barrel-launched munitions. The disc-wing UAV could be deployed from a height and manoeuvred towards a moving target or used to survey a target area and send information back to the operator. It has no protruding parts so could be carried in a back-pack on the ground over rough terrain until needed and then throw-launched from the hand.
Disc-wing Flight Dynamics
As an introduction to the dynamics of disc-wing flight consider Fig. 1a. Note that for a Frisbee-like shape at typical flight angles of attack, the centre of pressure (cp) of the disc-wing is ahead of the centre of the disc i.e. ahead of the disc cg. This results in an
untrimmed nose up pitching moment. If the disc is rotating, gyroscopic effects dictate that this pitching moment results in a precessional rolling rate, p. Thus spin provides enhanced pitch stiffness at the expense of roll stability. Using the conventional body fixed axes definition (Fig 1b), for a disc rotating in the direction of positive yaw then a positive pitching moment will generate a negative roll rate.
Flow Control Solutions
Aerodynamic control modifies the flow around a vehicle providing useful forces and moments to alter the flight trajectory and attitude in a desired way. For a rotating disc-wing there are two classes of flow control applicable:
Collective control – the deployment of an axi-symmetric set of control surfaces/jets which collectively provide a constant control moment throughout the rotation.
Cyclic control – the deployment of one or more control surfaces/jets which individually provide an oscillatory/impulsive control moment over each rotation.
Conventionally aerodynamic control is achieved by varying body geometry, for example the aerodynamic pitching moment acting on a conventional lifting surface is controlled by the deflection of a trailing edge flap. If this method was applied to a rotating disc-wing, the equivalent would be the cyclic deflection of an array of control surfaces on the trailing edge rim. The following is an outline of proposed flow control methods for further investigation.
Reaction Jets – The simplest means of providing control moments is through provision of reaction jets distributed around the perimeter of the disc. Whilst this is likely to be effective the performance penalty due to the power requirements is likely to outweigh other advantages.
Geometric Modification – Geometry can be modified locally through deflection of aerodynamic surfaces or globally through shape control applied to the whole body. Global geometric control has the advantage that no extra aerodynamic surfaces are required, however an actuation system able to cyclically camber and twist the disc is likely to be complex. A potential solution is to design a flexible structure with active stiffness provided by embedded piezoceramic actuators.
Fluidic Control – With fixed geometry, control moments can be generated either through boundary layer separation control or direct control of circulation around the body. Boundary layer control can be used to delay separation through increased mixing or to promote separation through the generation of an adverse pressure gradient. Mixing devices can be geometric e.g. vane vortex generators (7) or fluidic e.g. synthetic jets (8). An adverse pressure gradient can be generated through geometric modification of the local surface e.g. due to spoiler deflection or through blowing. Circulation control is a distinct type of boundary layer control in that momentum is directly injected into the boundary layer through tangential blowing such that the leading and/or trailing edge attachment lines are moved and the circulation changed. Circulation control can produce large control moments, however boundary layer separation control is potentially much more efficient.
For more discussion of the application of possible flow control solutions the reader is referred to (9). This paper presents an experimental investigation into the aerodynamics and control of a spin stabilised axi-symmetric disc-wing. The following briefly describes disc-wing aerodynamics and presents results from the application of two passive flow control methods on a non-spinning disc. Load data and flow visualisation results for a disc-wing with installed turbulence strips i.e. trip wires to force transition and another with flow obstructing fences are compared with those from a clean configuration. These results are compared to the clean configuration and analysed mathematically to predict the manoeuvrability achievable from similar methods of active flow control implemented on a rotating disc-wing. The suitability of active turbulence strips and fences as practical methods of disc-wing flow control are assessed based upon this comparison and a simple mathematical analysis of a banked turn manoeuvre.
EXPERIMENTAL METHODS
Wind Tunnels & Apparatus
The disc-wing was tested in two low speed wind tunnels: The first had an open-circuit with a test section of 0.9´1.1m, a top speed of 50m/s and a turbulence level of 0.5%. The second wind tunnel had a closed-circuit with a test section of 2.1´2.7m, a top speed of 70m/s and a turbulence level of 0.1%. The second tunnel was used for the smoke wire experiments due to the superior flow quality at very low speeds.
Fig. 2 The rig supporting a disc-wing, at incidence, in the wind tunnel.
A number of metal frames were used to mount the disc-wing in the wind tunnel in various configurations. The first (Fig. 2) was an L shaped arm with the disc mounted vertically on a horizontal axle supported by a vertical strut. The disc was mounted on a motor driven axle to test at various spin rates. Another arm was used for flow visualisation and held the disc in the horizontal plane.
The disc-wing cross-sectional profile can be seen in Fig. 3. The aspect ratio, AR, for a circular planform is 4/p » 1.27. For the discs tested, the centre line thickness to chord ratio Td/Dd was 0.14, Dd = 0.275m.
Fig. 3 Cross-sectional disc-wing profile.
Fig. 4 Disc-wing with installed turbulence strips.
Fig. 5 Disc-wing with installed turbulence strips and fences, LE at top of figure.
Transition or turbulence strips were fixed onto the upper surface of the disc-wing rim in the form of seven boundary layer trip wires, see Fig. 4. The copper wire rings of 1/2mm diameter were initially fixed onto the surface with super-glue laid 5mm apart, the first at a radius of 112mm. A thick coat of paint was applied to the surface to affix the wires further.
Fences were fixed onto the upper surface in combinations of four arced array’s of six at a radius of 119mm (LE) and 102mm (TE), see Fig. 5. The fences were cut from brass plate shim 1/4mm thick, with dimensions w´h = 14´7mm. They were set normal to the surface and the local flow direction, using super glue. The fences were spaced 20mm apart except over the centre line at 30mm apart.
Experimental Techniques
The aerodynamic loads were measured by a six component overhead balance (4). Surface paint flow visualisation used a kerosene-fluorescent powder mix (4,5). Vertical smoke wire for continuous oil supply and laser light sheet illumination (5).
RESULTS AND DISCUSSION
From previous work the authors have found that the effect of spin on disc-wing aerodynamics is relatively small. Due to the complexity of implementing an on board cyclic control solution in the wind tunnel, it was initially decided to investigate the application of various passive flow control methods to a non-rotating disc-wing. The results from this study provide an estimation of the control forces/moments achievable by a similar control system implemented on a rotating disc-wing. For example, the installation of passive vortex generator surfaces as a method of separation control for the non-rotating case offers a prediction of the control moments which could be reproduced by an active array of vortex generator jets for the rotating case.
Disc-wing Aerodynamics
The aerodynamic force and moment coefficients acting on the disc-wing are dependent on angle of attack and rotation rate, with the effect due to rotation tending to be small. The loads were measured for the range of conditions that the disc-wing will experience in free flight.
Clean Disc-wing – The lift and drag trends for a clean disc-wing are shown in Fig. 7a,b for Reynolds number of 3.78´105, equivalent to a flow speed of 20m/s, and advance ratio (rim speed to flow speed) 0. The lift curve has slope 0.05 and the drag curve shows a minimum, CDo, of 0.09 at the zero lift angle (3°). The pitching moment curve (Fig. 7c) is non-linear and displays a negative (nose down) coefficient of -0.02 at the zero lift angle of attack. Zero pitching moment occurs at 8° incidence and with a nose up pitching moment for higher incidence. The rolling moment is essentially zero, which is as expected for a symmetrical body (Fig. 7d).
A general description of the surface paint flow visualisation over a non-spinning disc-wing, at 5° angle of attack, follows with reference to Fig. 8. The flow over the upper surface of a non-rotating disc is characterised by separation at an arc of constant radius (L1) on the leading edge rim, followed by reattachment at a line of similar geometry (L2) and trailing edge separation (L3) thereafter. This initial separation and subsequent reattachment forms a separation bubble (B). Trailing vortices detach from the upper surface at two symmetrical positions (V1, V2). The cavity flow (Fig. 8b) is characterised by separation at the leading edge lip. This separated shear layer impinges on the inside of the trailing edge rim with reversed flow beneath the shear layer (F). A ‘dead air’ pocket exists aft of the leading edge rim (E). The reader is referred to previous work by the authors (4,5) for a more detailed aerodynamic analysis of a disc-wing with smooth upper surface.
Disc-wing Flow Control
The following presents results from the application of two passive flow control methods on a non-spinning disc-wing. Load data and flow visualisation results for a disc-wing with installed turbulence strips and another with flow obstructing fences (or spoilers) are compared with those from a clean configuration.
Disc-wing with installed Turbulence Strips – The lift and drag trends for a disc-wing with installed turbulence strips are shown in Fig. 7 for a Reynolds number of 3.78´105 and advance ratio of 0. The lift curve has slope 0.05 and the drag curve shows a minimum, CDo, of 0.09 at the zero lift angle, -1° incidence. Zero pitching moment occurs at the zero lift angle of attack –1,2° incidence and provides a nose up pitching moment for higher incidence (Fig. 7c). The rolling moment is essentially zero, which is as expected for a symmetrical body (Fig. 7d).
A disc-wing with turbulence strips applied to the rim of the disc (Fig. 4), forces boundary layer transition on the leading edge. The flow over these boundary layer trip wires equates to the flow over a number of surface bumps with a small separation bubble behind each. These are of minimum height, totally submerged in the boundary layer and therefore create minimum drag. They are of sufficient height, however to force transition to a turbulent boundary layer. This increases mixing and energises the boundary layer preventing separation on the leading edge (Fig. 9a) so that the flow remains attached. The central cross-section of the near wall flow at zero AoA is shown in Fig. 13, illuminated by a laser light sheet. The laminar separation bubble is visuailised for a disc with smooth upper surface (Fig. 13a) and the effect of the turbulence strips causing attached flow on the leading edge rim (Fig. 13b).
On the trailing edge however the trip wires promote separation and the stagnation regions are seen to extend further upstream and towards the wing tips than for the clean configuration, Fig. 8a,9a. This earlier trailing edge separation is the dominant force in the positive (nose up) change in pitching moment seen in Fig. 7c for typical AoA ~ 0° to 5°. It is also responsible for a reduction in lift seen at this incidence range, Fig 7a. For a clean disc-wing with smooth upper surface the separation bubble will have a drag increment associated with it, therefore the attached flow for a disc with turbulence strips reduces the drag slightly, Fig. 7b. The rolling moment for both configurations is zero for the non-rotating case (Fig. 7d).
A direct comparison of the surface flow with and without turbulence strips is shown in Fig. 9b, which shows a disc-wing with turbulence strips on the starboard side only (not a superposition of two images). This surface flow directly contrasts the two flow regimes namely the arced leading edge separation, reattachment and well defined trailing edge separation line for the clean port side, and leading edge attached flow and promoted trailing edge separation for the starboard side.
Disc-wing with installed Fences – The load data (Figs. 10,11) is from an experiment where four arrays of six fences were positioned in various combinations on the LE & TE (leading edge & trailing edge) rim of a disc with installed turbulence strips. Therefore, the baseline case used for comparison is the results from a disc-wing with installed turbulence strips only. The fences installed on the LE of the disc-wing did not change the aerodynamic lift, Fig. 10a,c shows the CL curve overlays the baseline case. The LE fences are positioned on the curved rim, slightly off the flat upper surface and therefore when fixed normal to the surface they become the equivalent of a flat plate at high AoA to the local flow direction. This means that there is a component of lift associated with each fence. The incident control surfaces create an adverse pressure gradient ahead of the array of fences but the lift destroyed by the presence of the fences on the surface seems to be compensated by the lift generated on the surfaces themselves. For the case where fences were installed on the TE only, a marked decrease in lift is observed in Fig. 10a,c particularly evident in Fig. 10c for typical flight AoA. The lift on the trailing edge rim is destroyed in the vicinity of the TE fences causing a change in CL, DCL = –0.08 for 0° to 10° AoA range. For a disc-wing with fences installed on both the LE & TE, the near surface flow quality is destroyed by the LE fences and as the flow then impinges on the TE fences it is already turbulent and unsteady. Therefore the lift decrease shown in Fig. 10c is less for fences on both the LE & TE than for a disc-wing with installed TE fences only.
The drag curve for a disc-wing with installed fences on both the LE & TE (Fig. 5) has a minimum CDo = 0.15 at –1° AoA (Fig. 10b). This drag curve shows an almost uniform increase from the baseline case over the tested AoA range, DCD = 0.06 for 0° to 10° AoA range. For the tests with fences installed on the LE or TE only, the drag generated by the LE fences is dominant for positive AoA. Conversely, the drag generated by the TE fences is dominant for negative AoA, see Fig. 10d. The two curves intersect at 0° AoA with CD = 0.13. At high AoA the TE fences are largely shielded from the free stream flow but the fluid impinges on the LE fences directly causing the increase in drag for positive AoA (Fig. 10d). At low (negative) AoA the local AoA of the LE fences reduces as does their associated drag. Below 0° AoA the TE fences become more exposed to the oncoming flow causing the increase in drag shown in Fig. 10d.
Fig. 11a shows the effect of LE fences on the pitching moment which provides a positive (nose up) increase throughout the AoA range tested, DCM = 0.008 for 0° to 10° AoA. This could be due to one or both of two reasons, lift generated by the LE fences and/or a decrease in lift on the TE due to a reduction of near wall velocity caused by the upstream disturbance from LE fences. Fig. 11c is a more detailed graph of the typical flight AoA range 0° to 10° which shows a couple of extra curves than Fig. 11a. The change in pitching moment caused by just one central fence on the LE causes over half that produced by twelve LE fences. This is close to the optimum position for a maximum change in CM due to one control surface, DCM = 0.005 for 0° to 10° AoA range. When fences were installed on the TE only, they have little effect at low AoA (0° to 4°) and become more dominant at higher AoA (>4°) showing a positive increase in CM.
For a disc-wing with installed fences on the LE port side only, a positive (starboard wing down) rolling moment results for low AoA (<15°) which becomes negative for higher AoA (Fig. 11b). When fences are installed on the TE port side only the CR is negative for the AoA range tested (positive and negative rotations correspond to conventional body fixed axes, see Fig. 1). For a disc-wing with fences installed on both the LE & TE on the port side only, the dominance of the LE fences is again apparent at high incidence, see Fig. 11b. Both test cases with installed LE fences overlay one another at high AoA (>10°). Whereas the two test cases with installed TE fences become more dominant at low AoA (<–5°) with the two curves overlaying one another (Fig. 11b). The baseline case for a disc-wing with installed turbulence strips only, is not shown on Fig. 11b,d but is CR = 0 for all AoA tested, see Fig. 7d. For typical flight AoA (Fig. 11d) the fences installed on the LE port side create a CR of around 0.002 whereas fences on the TE port side only, gives a CR or around –0.005 in the opposite direction. Fences on both the LE & TE create an approximately zero CR. Pressure distributions (future work) are needed to fully understand the generation of rolling moments for these cases.
Fig.12 shows some surface paint flow visualisation results for disc-wings with installed turbulence strips and fences (flow direction from top to bottom). Fig. 12a shows the upper surface pattern for a disc-wing with installed turbulence strips and fences on both the LE & TE, see Fig. 5. The LE fences cause the flow to separate and the shear layer reattaches in the central region of the disc enclosing a large arc shaped separation bubble. The reattached boundary layer and near wall flow is turbulent and highly unsteady as it interacts with the TE fences. Spiral nodes are formed aft of the TE fences due to the vorticity generated by flow between the surfaces. Upstream of the disc centre the flow patterns are symmetric but downstream of the centre the flow pattern shows slight asymmetry with a stagnation point ahead of the TE fences drifting off to the starboard side. Fig. 12b depicts the upper surface pattern for a clean disc with one fence installed on the LE ahead of the initial arc separation line. The general form of the surface pattern for a clean disc and the patterns created by the complex interaction between a single fence & a separation bubble are both visible. The boundary layer separates ahead of the fence and reattaches downstream leaving a nodal point of reattachment on the centre line. A couple of stagnant regions move off to the sides in the near wake of the fence, which are linked, into a couple of nodes on the edge of the separation bubble. These spiral nodes probably indicate where a couple of vortices detach from the surface, each leaving a streamwise trail over the central surface. Further investigation is needed to better understand the interaction of flow structures due to the presence of the fence.
The central cross-section of the flow over a single LE fence on a disc-wing with installed turbulence strips is shown in Fig. 14, flow speed 3m/s. The laminar shear layer becomes unstable aft of the fence and then reattaches to the surface (Fig. 14a). A laminar spanwise vortex structure is observed ahead of the fence (Fig. 14b) aft of the initial separation, for 20° AoA.
Disc-wing Flight Manoeuvres
To get some idea of the flight dynamics of a rotating axi-symmetric disc-wing a simple analysis follows describing trimmed flight and a banked turn manoeuvre based on experimental values from Fig. 11, for a disc with installed turbulence strips. The bank manoeuvre is based on the mechanical principles of gyroscopic motion outlined by Barger and Olsson (10), neglecting drag throughout.
Trimmed Flight – For trimmed flight the lift Lf generated by a disc-wing must equal its weight Mdg,
[1] 
where r is the density of air, S the disc-wing planform area pRd2 based on the disc radius Rd, V is the flight speed and CL the lift coefficient. For a standard plastic Frisbee-like disc-wing of 160grams and diameter Dd = 0.275m the lift required for trimmed flight is Mdg = 0.16´9.81 = 1.57N. At a flight speed of 20m/s this is achieved for a lift coefficient Cl = 0.1 which corresponds to a 1° angle of attack.
Bank Manoeuvre – Again for a standard plastic Frisbee-like disc-wing flying at a roll angle of 60°, to maintain the same altitude for a banked turn the lift must support its weight,
[2] 
At a flight speed of 20m/s this is achieved for a lift coefficient CL = 0.1/cos60 = 0.2 which corresponds to a 2° angle of attack.
Consider the rolling moment generated by an array of fences positioned on the TE of a non-rotating disc starboard side only. The rolling moment coefficient CR = 0.005 at 2° AoA for a non-rotating disc-wing. Using this value as a prediction of the CR achievable from a similar array of active fences implemented on a rotating disc-wing, it is possible to predict the pitch rate q caused by gyroscopic precession. Practically this could be achieved with active on/off fences which could be raised in cyclic fashion over the TE region of interest during the rotation.
The mass distribution of the disc-wing is concentrated in the outer rim, therefore the moment of inertia I is approximated by assuming that the total mass is evenly distributed along the circumference such that, I = MdRd2. The angular momentum H in spin is therefore given by MdRd2ws about the spin axis, where ws is the spin rate. The introduction of a disc-wing rolling moment due to TE fences causes a pitch rate q (gyroscopic precession) given by,
[3] 
[4] 
where L is the rolling moment based on experimental parameter CR, the rolling moment coefficient. Therefore for CR = 0.005 at an AoA of 2° (Fig. 11d) and AdvR = 0.17 the pitch rate q is 0.265 which will take the disc 5.9 seconds to turn through 90°.
The disc will describe a horizontal curved flightpath due to the precession caused by the rolling moment and the horizontal component of lift. To get an idea of the scale of this curved flight trajectory, assume that the disc-wing describes a quarter circle with linear velocity V = 20m/s. In a circular flightpath the horizontal component of lift balances the centrifugal force and the angular velocity w0 must match the precession rate q such that,
[5] 
From equations [4] and [5] a measure of the arc radius can be calculated rearranging for Rc,
[6] 
For a spin rate of 25 rad/sec (AdvR = 0.17) the circular flightpath radius is estimated to be Rc = 75m equivalent to 275 wing spans, 275Dd (diameter). From recent work by the authors (9), for maximum manoeuvrability i.e. a sharp banked turn, the spin rate should be minimised to describe a circular flightpath of minimum radius. However as the spin rate approaches zero the disc loses its gyroscopic stability. At large rotation rates the disc shows resistance to pitch q, due to increased angular momentum and is much more difficult to control. Therefore these values Rc = 75m and T90 = 5.9secs (T90 is the time taken for a 90° banked turn) are likely to be slightly low, the disc will probably require a higher spin rate for increased stability somewhere between 200 to 400 rpm and therefore (Rc, T90) ~ (100m, 10secs). Fig. 6 shows the
Fig. 6 Calculated disc-wing flight trajectories.
horizontal flight trajectories in the xy plane calculated from the above analysis, for both trimmed flight and a 90° banked turn, V = 20m/s. For both trajectories the points are 0.5 seconds apart except the last which terminates at 5.9 secs, the radius of the banked turn is Rc = 75m. The drag on a disc-wing would cause the flight speed to drop over time however this could be overcome by dropping in altitude during the turn.
Aerodynamic Control of a Disc-wing UAV
The TE fences tested during this preliminary investigation have generated pitch control forces capable of producing a banked turn manoeuvre as previously described. If this degree of aerodynamic control could be gained from active on/off fences or some other method of control then that would offer practical possibilities for aerodynamic control of a disc-wing UAV. For an active system of fences on a rotating disc-wing a more practical solution would be to have on/off control surfaces generating collective control rather than deflecting each one for a period of the rotation.
The testing of passive turbulence strips has also been promising with a change in pitching moment resulting from these sub-boundary layer devices (Fig. 7c). Micro balloon actuators (11) offer an active method of doing this, they are small arrays of inflatable surfaces which increase surface roughness on-demand. These could be installed on the rim of the disc and allow roll control by inflating the balloons to the required level. If the balloons were initially deflated i.e. flat then for a disc-wing UAV with positive spin (Fig. 1) a negative (nose down) CM (Fig. 7c) for typical flight AoA would result in a positive roll rate (starboard wing down) due to gyroscopic precession. Then as the balloon actuators are inflated they would cause a positive (nose up) change in CM and at some point reach a trimmed state of zero pitch i.e. zero roll rate. When inflated further the CM becomes positive and the disc-wing UAV would roll in the opposite (starboard wing up) direction. The authors also found that the rolling moment coefficient for a spinning disc-wing with installed turbulence strips could provide a precessional pitch rate capable of producing a banked turn manoeuvre with a turning radius of 85m completing a 90 degree turn in 6.7 seconds (9).
CONCLUSIONS
Proposed methods of both passive and active flow control have been outlined with specific application to a spin stabilised axi-symmetric disc-wing. Further investigation in the form of a feasibility study is required to determine the effectiveness of each.
The comparison of the aerodynamics of a disc-wing with installed passive turbulence strips, fences and the clean configuration, is insightful as a tentative prediction of the moments which could be produced by a similar method of active flow control implemented on a spin-stabilised disc-wing UAV.
The control moments generated by a disc-wing with installed fences are capable of producing a banked turn manoeuvre of 75m radius for the rotating case. If this degree of aerodynamic control could be gained from active turbulence strips or some other method of control then that would offer practical possibilities for aerodynamic control of a disc-wing UAV.
ACKNOWLEDGEMENTS
The authors would like to thank D. Mould at the Goldstein Research Laboratory for the skilled construction of the rig & disc. This research was funded by the EPSRC, award reference number 98317373.
REFERENCES
1. Zimmerman, C.H., Aerodynamic Characteristics of Several Airfoils of Low Aspect Ratio, NACA Technical Note, No.539, Aug. 1935.
2. Stilley G.D. & Carstens D.L., Adaptation of Frisbee Flight Principle to Delivery of Special Ordnance, AIAA 72-982, AIAA 2nd Atmospheric Flight Mechanics Conference, Palo Alto, California, USA, 1972.
3. Stilley G.D., Aerodynamic Analysis of the Self-suspended Flare, Honeywell Inc., NAD/Crane RDTR No.199, AD740117, 23 Feb. 1972.
4. Potts J.R. & Crowther W.J., The Flow Over a Rotating Disc-wing, RAeS Aerodynamics Research Conference Proc., London, UK, Apr. 2000.
5. Potts J.R. & Crowther W.J., Visualisation of the Flow Over a Disc-wing. Proc. of the Ninth International Symposium on Flow Visualization, Edinburgh, Scotland, UK, 2000.
6. Higuchi H., Goto Y., Hiramoto R. & Meisel I., Rotating Flying Disks and Formation of Trailing Vortices, AIAA 2000-4001, 18th AIAA Applied Aero. Conf., Denver, CO, Aug. 2000.
7. Harrison T., Wood N.J. & Zhong S., Turbulent Separation Control Using Vortex Generators, RAeS Aerodynamics Research Conference Proc., London, UK, Apr. 2001.
8. Crook A. & Wood N.J., Measurements and Visualisations of Synthetic Jets, AIAA 2001-0145, 39th Aero. Sci. Meet & Exhibit, Reno, NV, Jan. 2001.
9. Potts J.R. & Crowther W.J., Flight Control of a Spin Stabilised Axi-symmetric Disc-wing, AIAA 2001-0253, 39th Aero. Sci. Meet & Exhibit, Reno, NV, Jan. 2001.
10. Barger V. & Olsson M., Classical Mechanics: A Modern Perspective, McGraw-Hill, New York, 1973.
11. C. Grosjean, G. Lee, W. Hong, Y.C. Tai & C.M. Ho, Micro Balloon Actuators for Aerodynamic Control, 11th Ann. Int. Workshop on MEMS, Heidelberg, Germany, pp. 166-171, Jan. 1998.
(a) Lift coefficient. (b) Drag coefficient.
(c) Pitching moment coefficient. (d) Rolling moment coefficient.
Fig. 7 Force and moment characteristics at zero spin rate for a disc-wing with turbulence strips and the clean configuration, Re = 3.78105.
Fig. 8 The upper (a) and cavity (b) surface paint patterns for a clean disc-wing at 5° incidence, V = 15m/s.
Fig. 9 The upper surface paint patterns for a disc-wing with installed turbulence strips on (a) port & starboard and (b) starboard only, at 5° incidence, V =20m/s.
(a) Lift coefficient. (b) Drag coefficient.
(c) Lift coefficient (detail). (d) Drag coefficient (detail).
Fig. 10 Force characteristics at zero spin rate for a disc-wing with installed turbulence strips and various combinations of fence arrays, Re = 3.78105.
(a) Pitching moment coefficient. (b) Rolling moment coefficient.
(c) Pitching moment coefficient (detail). (d) Rolling moment coefficient (detail).
Fig. 11 Moment characteristics at zero spin rate for a disc-wing with installed turbulence strips and various combinations of fence arrays, Re = 3.78105.
Fig. 12 The upper surface paint patterns for (a) disc-wing with installed turbulence strips and fences (b) clean configuration with one fence central on the LE rim, at 5° incidence, V =20m/s.
Fig. 13 Central cross-section of the (a) laminar separation bubble and (b) turbulent boundary layer on the leading edge of a non-rotating disc-wing at 0° incidence, V = 8m/s.
Fig. 14 Central cross-section of the flow in the vicinity of one fence central on the leading edge of a non-rotating disc-wing at (a) 5° (b) 20° incidence, V = 3m/s.