- Email: [email protected]
A feasibility study of micro air vehicles soaring tall buildings
J. Wind Eng. Ind. Aerodyn. 103 (2012) 41–49
Contents lists available at SciVerse ScienceDirect
Journal of Wind Engineering and Industrial Aerodynamics journal homepage: www.elsevier.com/locate/jweia
A feasibility study of micro air vehicles soaring tall buildings C. White, E.W. Lim, S. Watkins n, A. Mohamed, M. Thompson School of Aerospace, Mechanical and Manufacturing Engineering, RMIT University, GPO Box 2476, Melbourne, Victoria 3001, Australia
a r t i c l e i n f o
abstract
Article history: Received 27 May 2011 Received in revised form 6 February 2012 Accepted 14 February 2012 Available online 14 March 2012
Micro Air Vehicles (MAVs) are person-portable platforms, which due to their small size and corresponding manoeuvrability show great potential for surveillance in urban environments. Achieving useful endurance using on-board electric power plants remains challenging despite rapid advances in battery energy storage density. This paper examines experimentally the feasibility of using orographic ‘slope’ lift in urban built environments to enhance the operational capability of MAV platforms by increasing range and endurance. A representative building in an urban environment was selected for investigation. The flow-field velocity in the upwind region of a 1:100 scale model of this building was mapped in a wind-tunnel with a reproduced vertical velocity profile. The vertical velocity component was found to be in the order of 15%–50% of the mean wind velocity at building height. These results were compared with data measured on the full-size building and found to agree well. The sink-rate of a soaring MAV was measured through flight-testing and found to be less than the available vertical velocity component in the upstream flow field for average wind speeds which indicates that soaring is feasible, provided that controllability challenges in high-turbulence environments can be overcome. Crown Copyright & 2012 Published by Elsevier Ltd. All rights reserved.
Keywords: Micro air vehicle (MAV) Unmanned air vehicle (UAV) Urban soaring Building Orographic lift Range Endurance Flow topology
1. Introduction Micro Air Vehicles (MAVs) are essentially Unmanned Aerial Vehicles (UAVs) but smaller in size and dimension (personportable). Recently, there has been an increasing interest in utilising MAVs as surveillance platforms in urban environments for clandestine operations, or search and rescue coordination roles in the immediate aftermath of natural or man-made disasters. The performance of small aircraft, in particular range and endurance, is currently restricted by on-board energy storage limitations set by size, weight and type of battery/ electric sources (Kahveci et al., 2008). However, significant energy is readily available in the environment in which MAVs fly. Since the early investigation into flight, it was noticed that some birds were able to sustain flight for long periods of time without expending energy through wing flapping. Larger birds developed soaring techniques to extract energy from relative wind motions to reduce energy consumption (Lawrance and Sukkarieh, 2009). Humans have used the vertical air motions in the atmosphere for flight, but so far this has been limited to sport aviation, be it radio controlled models (such as slope and thermal soarers) or hang-gliders and
n
Corresponding author. Tel.: þ61 3 9925 6084. E-mail addresses: [email protected] (C. White), [email protected] (E.W. Lim), [email protected] (S. Watkins), [email protected] (A. Mohamed), [email protected] (M. Thompson).
sailplanes. The use of such energy is now becoming of interest for UAVs and MAVs as well (Langelaan, 2009). It has been estimated that the endurance of a small glider UAV (4.3 m wingspan) can be extended from 2 to 8 h in winter and to 14 h in summer using soaring flight (Allen, 2005), suggesting that soaring may be a way of reducing the limitations in on-board energy storage for MAVs. Soaring is generally divided into two types: static soaring and dynamic soaring. Static soaring is the process of flying in air that is rising, either from the vertical component of wind blowing up a slope (arising from natural features, such a hills or ridges), known as orographic lift or from thermals. Dynamic soaring involves flying trajectories through distributions of wind speed that have vertical components to increase kinetic energy (Lawrance and Sukkarieh, 2009). While dynamic soaring has been proven to be viable to increase the endurance of UAVs (Lawrance and Sukkarieh, 2009; Deittert et al., 2009a, 2009b); the flight trajectories do not appear to be possible in densely built urban environments or conducive to surveillance. Static thermal soaring has been explored in recent years as a way to improve UAV range and endurance. Thermals occur when air near the ground becomes less dense than the surrounding air as a result of heating or humidity changes at ground level (Allen, 2007). Birds ‘ride’ these thermals to remain airborne without flapping their wings. This technique is widely used by sailplane pilots to gain altitude. Many strategies for extracting energy from thermals for UAVs have been published (Allen, 2007; Cowling et al., 2009; Edwards and Silverberg, 2010).
0167-6105/$ - see front matter Crown Copyright & 2012 Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.jweia.2012.02.012
42
C. White et al. / J. Wind Eng. Ind. Aerodyn. 103 (2012) 41–49
The principle of building soaring appears to be feasible for MAVs and is most similar to slope soaring or the use of orographic lift. Sailplanes utilise vertical air motions related to wind over geographic features to stay in flight or gain height when slope soaring. Snyder and Zimmerman, 1975a stated that a sustained wind speed of 15 knots (7.72 m/s) or more perpendicular to a slope usually generates enough lift to support a sailplane and that the height of the lift is usually two or three times the height of the rise from the valley floor to the ridge crest. Wolters (1971) stated that the position of maximum lift will be found at about a 45-degree angle upwind and above the slope. While Langelaan (2007) provided a model for trajectory optimisation or duration extension of MAVs using such orographic lift, no work has been published regarding soaring of MAVs in urban environments using vertical components of the wind over buildings. However, examples do exist in nature of bird making use of orographic lift generated by man-made structures—such as herring gulls flying on the upwind side of a ferry or cruise ship (Tennekes, 2009). Extensive studies have been conducted to understand the flow around buildings in urban environments. This has included fieldwork studying the wakes of structures and turbulent flow around take-off and landing of aircraft (Cermak et al., 1977; Frost and Shahabi, 1977), dispersion of pollutants downwind of buildings (Zhang et al., 1996; Leuzzi and Monti, 1998; Kim and Baik, 2004), pedestrian wind comfort and numerous studies on wind loads on structures (Holmes, 2001; Peterka et al., 1985). According to Ricciardelli and Polimeno (2006), the characteristics of the wind flow in the urban environment are known to be somewhat different from those of a Turbulent Boundary Layer (TBL) naturally developed over a homogeneous rough surface. The mean wind speed will differ based on the macro-meteorological events, elevation and also the physical location and surroundings of the point of interest. Sharan (1975) investigated the characteristics of the flow around building models and found out that as long as the boundary layer thickness is greater than the model height, the average location of the front stagnation point lies around 85 per cent of the model height. These studies have predominantly focused on pressure distributions or downstream topology models, rather the velocity components immediately upstream of the building—an understanding of which is required for MAV building soaring.
A number of platforms exist that meet the specifications of MAVs that are person portable and may be capable of building soaring. These are recreational models used by radio control enthusiasts to either soar in thermals or in slope lift. However, it is not known if the orographic lift in urban environments is adequate to be soared by such platforms. Langelaan (2007) indicated the theoretical sink-rate of a small MAV similar to these platforms (with a wing area of 1 m2 and a mass of 10 kg) of 0.2 m/s is attainable, although was not measured. While many of these platforms are available at low cost, no work appears to have been published on measuring and characterising the glide of such—particularly with respect to sink-rates, flight speeds and the vertical air movements required to keep them airborne in typical wind speeds. The focus of this research is to build towards an understanding of the vertical flow field (updraught) in front of a building to investigate the feasibility of MAV soaring in an urban environment. A representative building within the RMIT University Bundoora Campus; ‘Building 201’, was selected due to its strategic position and the environment around it matched the topography of a suburban terrain (Fig. 1). A scale model of the building was used for wind-tunnel testing and the results validated by measurement from the roof of actual building. Results from these studies are then compared with the measured sink-rate and airspeed data of an MAV in order to study the feasibility of orographic soaring of MAVs in urban environments.
2. Materials and methods 2.1. Scale building testing Suburban terrain velocity and intensity profiles (based on Walshe, 1972) were generated at 1:100 scale in the RMIT University Industrial Aerodynamics Wind Tunnel. This involved modification from the nominally smooth, uniform test section flow by an array of upstream barriers and floor-mounted roughness elements to artificially thicken the boundary layer within a reasonable stream-wise distance. The test section is 9.2 m long, 2 m high and 3 m wide. The region of the test-section used for testing was 40 characteristic lengths (based on maximum barrier height) downwind from the array of barriers and over ten
Fig. 1. RMIT University ‘Building 201’: (a) north and (b) west face.
C. White et al. / J. Wind Eng. Ind. Aerodyn. 103 (2012) 41–49
characteristic lengths away from the roughness element. In order to rapidly measure and depict the velocity profiles arising from the many configurations of barriers and elements trialled, a stagnation tube rake and static pressure tubes were positioned at the location of interest (where the leading edge of the building was going to be placed), see Fig. 2. The pressures were displayed on a multi-bank inclined manometer which was found to be an efficient way of investigating the effects of the configurations. Once a suitably scaled profile was achieved it was documented in more detail with a TFI Cobra probe and traversing gear. Cobra probes are four-hole pressure probes offered commercially by Turbulent Flow Instrumentation (TFI); for details see TFI (2011) and Milbank et al. (2005). The probe used had a 2.6 mm multifaceted head, which was considered sufficiently small to have a minimum influence on the flow field. Cobra probes are able to resolve the three orthogonal components of velocity at up to 2 kHz and provide the static pressure, providing that the flow vector is contained within a cone of 7451 around the probe x-axis. Details of the system and examples of use can be found in Watkins et al. (2002) and verification and further details including dynamic
43
capabilities, probe operation methodology and calibration techniques can be found in Hooper and Musgrove (1997). The velocity profile is presented in Fig. 3, velocities were normalised to the mean velocity at building height. Generally the agreement is good except very close to the ground. For this study the locations of interest are around the top of the building and thus well above the stagnation point. Since small changes in the profile close to the ground make only minor changes to the position of the stagnation point (Sharan, 1975) this effect is considered to be negligible. The turbulence intensity profile is illustrated in Fig. 4. Turbulence intensity in the wind tunnel was measured using a TFI Cobra probe. Cobra probe reads turbulence intensity in all three orthogonal directions (Iuu, Ivv, Iww) and provides the overall turbulence intensity. As turbulence intensity is not the main focus of this paper, only the overall turbulence intensity was considered. Similar to the velocity profile, Fig. 4 shows a relatively good comparison between the two profiles except close to the ground. All testing were conducted at a nominal tunnel velocity of 10 m/s. A Reynolds number sensitivity test was conducted and indicated that the non-dimensional results were Reynolds number insensitive.
Stagnation tube rake
Array of barriers
Flow direction Building model
Connected to multi-bank manometer
Roughness element
Fig. 2. Building model (stagnation tube rake), roughness element and barriers (not to scale).
Fig. 3. Velocity profile reproduced in wind-tunnel.
44
C. White et al. / J. Wind Eng. Ind. Aerodyn. 103 (2012) 41–49
Fig. 4. Wind-tunnel turbulence intensity profile (no building present).
z x Wool tuft indicating mean flow direction
Fig. 5. Position of measurement grid upstream from model (not to scale).
The 1:100 scale building model had a width of 0.38 m, a depth of 0.38 m, and a height of 0.43 m. While the model replicated the building floor-plan, details such as windows, trimmings and ventilation systems were omitted. Since the blockage ratio was relatively low and all data were non-dimensionalised, no corrections were applied. Flow visualisation with helium bubbles was used for positioning Cobra Probes for local velocity measurements. The bubble wand was located at 0.5 m upstream from the building on the floor of the tunnel to minimise interference with the upcoming flow. The size and density of the helium bubbles and the lighting and exposure time were adjusted to achieve short streak lines, see Fig. 9. A vertical upstream plane aligned with the centreline of the building was established to measure the velocity field upstream of the building (Fig. 5). The spacing between measurement points was decreased in the region of greatest vertical velocity (as indicated with helium bubble tracks). Velocity components for the x, y and z axes were denoted by u, v and w respectively and conform to the sign convention specified in Fig. 5. The lateral distance (y) 0 m is at the centreline of the building, the height (z) of 0.43 m corresponds to the top of the building, and the longitudinal distance (x) of 0.02 m is upstream from the front face of the building. A Cobra probe was used for flow measurement at 35 points on the measurement grid indicated in Fig. 5. The angle of the tip of the Cobra Probe was varied at each point to ensure all data lay on the calibration surface of the probe (within the 901 cone of acceptance). This angle of inclination was measured at each point
Fig. 6. Cobra probe and mast on roof of Building 201.
with an inclinometer to allow later resolution of the velocity components. For each point there were three samples of 60 s. 2.2. Full-size building testing A northerly wind (bearing of 3601) was necessary for the analysis to reproduce the flow simulated by tunnel testing; at the time of testing the wind was approximately northerly with a small westerly component (3551–3601). The same Cobra probe used in the wind-tunnel testing was mounted on a mast with an adjustable angle of incidence on the roof of the building. The height of the mast was adjusted to 1.5 m above building top edge, which coincided with a point on the measurement grid from wind-tunnel testing. A wool tuft was tied to the rig to visualise the flow angle. The frame of the mast was then tilted in the direction of the flow with the angle measured using an inclinometer. An image of the mast during testing is presented in Fig. 6. Two samples of 15 min were used. 2.3. MAV flight-testing A robust MAV, yet posing minimal possibility of damage to the building and personnel, was desirable for testing; both for calibration flights from a flat field (see later) and for planned flights launched from the top of the building. The MAV selected
C. White et al. / J. Wind Eng. Ind. Aerodyn. 103 (2012) 41–49
was based on the Alula (evo series), a commercially available model (produced by Dream-Flight in California). This is a small discus-launch radio-controlled flying-wing glider with a wingspan of 0.9 m, wing-area of 0.167 m2, length of 0.48 m and approximate flying mass of 0.17 kg. A three-view drawing of the model is presented in Fig. 7. This was chosen due to the manufacturer’s claim of it being an ‘urban soarer’ from anecdotes that it was capable of soaring in unconventional locations due to low wing-loading (1.02 kg/m2) and full trailing-edge elevon control system, giving good manoeuvrability. These features mean that weak lift (flows with small vertical velocity component) can be successfully exploited in confined spaces. Additionally, the platform area of the Alula is analogous to that of a soaring raptor (hawk), lending itself to low recognisability as a man-made platform. This is potentially an important quality for MAVs used urban environments for surveillance purposes—easy detection of the platform can be reduced with such bio-mimicry. The discus launch technique (also referred to as a side-arm launch, whereby the platform is gripped by a wing-tip and accelerated in a circular path—similar to that of the discus-throw athletics event) enables good height ( 20 m) to be achieved by an operator—conceivably adequate to reach a lift band adjacent to a low rise building. Data on the range of flight speeds and associated sink-rates can be achieved through flight test or wind-tunnel measurements. Despite being the more challenging of the two, flight testing was selected for the measurements presented here, since not only could it generate data obtained under a range of turbulence intensities, it also offered much-needed insight into the control issues of flying in the turbulence inherent in the lower levels of the Atmospheric Boundary Layer (ABL). The relative turbulence characteristics experienced by MAVs flying through the ABL and the associated aerodynamic issues are significant and are considered further in Watkins et al. (2009) and Watkins et al. (2010). The craft was instrumented with light-weight data logging equipment (Eagle Tree Systems, LLCTM) to record barometric airspeed, altitude and position parameters using a modified eLoggerTM V4, Airspeed MicroSensor V3, Altitude MicroSensor V3 and GPS Expander Module respectively. A commercially available 2.4 GHz radio control system was used for flight control. A Prandtl-type pitot-static tube was installed on the right wing for airspeed and altitude measurements, positioned suitable far
GPS expander
45
Pitot tube
eLogger
Altitude MicroSensor
Airspeed MicroSensor
Fig. 8. Data-logging equipment installed in Alula.
forward of the aerofoil to give minimal interference for stagnation and static pressure. The installed equipment increased the mass of the craft to 0.212 kg from the original 0.17 kg. This installation is presented in Fig. 8. While it is acknowledged that increasing the flying mass increases the sink-rate (compared with a non-instrumented craft) the installation of onboard mission-specific equipment would likely result in a similar increase. Flat field flight tests (i.e. with no orographic lift) were conducted under a range of atmospheric winds; however, the data presented here are restricted to those obtained under very light winds (o2 m/s). Light wind conditions were desirable to ensure that airspeed could be stabilised and held constant for as long as possible. One author performed the launch, while another operated the flight controls. A number of flights were made with the discus-launch technique achieving a maximum height of approximately 15 m above the ground at the apex of the launch. Flight duration was typically 25–30 s. All parameters (airspeed, altitude and GPS) were sampled at 10 Hz. Each flight followed a similar profile that involved visually stabilising airspeed after launch and maintaining this constant airspeed for as long as possible (to obtain useful sink-rate data collection time) before reducing the airspeed and sink rate immediately prior to landing (to minimise damage). It was not possible to achieve a nominated airspeed as there was no telemetry to the pilot; however, provided the airspeed was in the normal flight envelope (above stall and below maximum) this was considered adequate, as determining sink-rate values for a range of airspeeds was useful.
3. Results and discussion 3.1. Velocity measurements for scale building model
0.9 m
0.48 m Fig. 7. Alula evo MAV platform (courtesy Dream-Flight).
A representative upstream flow field shown with helium bubble streak lines is presented in Fig. 9. The outline of scale ‘Building 201’ (lower left corner) has been enhanced for clarity. The grid for flow measurement has been added. The helium bubbles are seen as streaks due to the fine-tuning of the exposure time of the camera. Longer streaks indicate a higher velocity. The stagnation point can be observed in Fig. 9. The position of the stagnation point coincides with the Cobra probe measurements, which is approximately a height of 0.35 m. The mean overall velocity in the upstream flow field grid normalised to the mean velocity at the building height in the velocity profile (6.55 m/s) is presented in Fig. 10. In Fig. 11 the mean vertical velocity component (w) normalised to the mean velocity at building height is presented. A vector velocity field is shown in Fig. 12.
46
C. White et al. / J. Wind Eng. Ind. Aerodyn. 103 (2012) 41–49
Measurement grid
Measurement grid
z
x
Scale ‘Building 201’ Scale ‘Building 201’ Stagnation point
Fig. 9. Flow visualisation with helium bubbles.
1.0
0.8
0.45
0.50 Height (m)
Height (m)
0.50
0.4
0.45
0.6
0.4 Scale ‘Building 201’
0.40 0.2
Scale ‘Building 201’
0.05
0.08
Upstream distance (m) Fig. 10. Mean overall velocity compared with mean velocity at building height.
3.2. Full-size building measurements and comparison with model scale Velocity measurements are presented in Table 1 for Building 201 and a corresponding point from wind-tunnel testing. For purposes of comparison (as the wind speed during full-size building testing was different to that in the wind tunnel) velocity components are normalised to the mean flow velocity at this position. The maximum vertical component of the flow velocity was found to be approximately 50% of the mean velocity at building height, this region occurs approximately 451 upstream from and above the building height. This represents a peak vertical velocity component (w) of around 3.3 m/s. Substantial regions of vertical velocity exist of at least 1 m/s. The vertical velocity decreases towards the stagnation point of the building at approximately zE0.35 m on the upstream face of the building. This stagnation
z
x
0.40 0.1
0.35
0.02
0.3
0.2
0.0
0.35
0.5
Flow direction
0.0
0.02 0.05 0.08 Upstream distance (m) Fig. 11. Mean vertical velocity (w) compared with mean velocity at building height.
point occurs at approximately 80% building height which agrees with the studies of (Sharan, 1975). These features are confirmed by the flow visualisation in Fig. 9. 3.3. Consideration of errors Similar sources of error exist for the model and full-size building testing due to the use of the same velocity measurement instrumentation used. The Cobra probes are calibrated regularly and are temperature compensated and typical angle errors are negligible (less than one degree). The time-averaged velocity errors are very much a function of the mean velocity squared (since the velocities are deduced from pressure differences); however, considerable experience has shown an error of less than 0.3 m/s at the typical measured velocities for this work (Pagliarella, 2010). The main source of error occurs in the measurement of the inclination
C. White et al. / J. Wind Eng. Ind. Aerodyn. 103 (2012) 41–49
of the tip of the Cobra probe; these errors are of the order of 731 which will not significantly affect the conclusion drawn here. For full-size building measurement, error may have been introduced by the velocity deviating from northerly (with 51 of westerly).
47
3.4. Flight-testing sink rate results Flight data were analysed for each flight to identify a region of stabilised airspeed where the MAV was gliding under nominally equilibrium conditions in still air. Typical flight airspeed and altitude data is presented in Fig. 13, indicating the region used for sink-rate analysis. The discrete steps in altitude and velocity indicate the digital resolution of the data logger and the sharp peak in airspeed at two seconds is the result of the hand launch technique. Data from the other four flights also analysed with this approach are presented in Table 2. The airspeed is the mean for the analysis interval. Sink-rate was evaluated from the altitude loss for the interval divided by the time.
4. The possibilities of soaring The normalised velocity results of the full-size building test compare well with those measured for a similar point from the wind-tunnel testing (Table 1). The more significant lateral component (v) found for the full-size building data is attributed to the specific geometry of the roof of Building 201, which was simplified for the wind-tunnel tests. Although the number of samples is limited, the wind-tunnel tests appear to conservatively predict (under-predict) the vertical component of velocity at this point by a small amount. This indicates that the remainder of the flow field measurements from the wind-tunnel test are reliable (and can be used to predict the vertical velocity of this flow field), meaning that in the region measured the rate of lift will be between 15% and 50% of the mean velocity (wind speed) at this height, which
Fig. 12. Velocity vector plot.
Table 1 Velocity components. Case
Full-size
Table 2 Summary of flight test and sink-rate data. Normalised velocity component Sample Mean normalising v w velocity (m/s) u
1 2 Scale ‘Building 201’ 1 (wind-tunnel) 2 3
7.29 7.88 5.87 6.04 6.04
0.639 0.667 0.774 0.758 0.753
0.294 0.284 0.021 0.047 0.014
0.711 0.689 0.633 0.650 0.657
Flight Stabilised mean Airspeed number airspeed (m/s) standard deviation (m/s)
Analysis interval (s)
Altitude loss (m)
Sink-rate (m/s)
3 5 6 7 8
14.7 10.3 6.7 7.6 14.8
8 5 4 3 8
0.54 0.49 0.60 0.40 0.54
4.8 5.3 6.1 5.7 6.1
Fig. 13. Example airspeed and altitude data from flight 5.
0.29 0.18 0.30 0.34 0.34
48
C. White et al. / J. Wind Eng. Ind. Aerodyn. 103 (2012) 41–49
depending on the wind strength and direction may be adequate to keep a soaring MAV airborne. It is not possible to make direct comparisons or examine a mean of the measured sink-rates for different airspeeds as each airspeed will have a unique sink-rate determined by the parabolic glide– polar relationship. For every fixed-wing platform there will be an airspeed that corresponds to a minimum sink-rate, with greater sink-rates encountered at lower and higher airspeeds. This airspeed corresponding to minimum sink-rate typically occurs 10% above the stall speed (minimum flying speed) and as the Alula was flown near stall (indicated by control unresponsiveness and nosedrop) it can be assumed that airspeeds at which data were recorded are close to that for minimum sink-rate. While airspeed measurement resolution was in the order of 0.28 m/s, averaging the airspeed over the analysis interval significantly reduces low resolution of this error. The low standard deviation of airspeed for the analysis intervals (0.18–0.34 m/s) indicates reasonable airspeed reliability for the analysis intervals selected. Airspeed measurement calibration was confirmed by placing the Alula platform with logging equipment in a windtunnel and altitude calibration by placement at a range of known heights. The resolution of altitude (1 m) measurements limits the accuracy of the sink-rates determined, as such the maximum error was evaluated to be between 0.14 and 0.30 m/s (assuming the worst case altitude resolution error of 2 m) and also dependent on the length of the analysis interval; the longer the analysis interval the smaller the corresponding error (as it is altitude error is evaluated by dividing the error by the interval length). Because of this error it is not possible to base accurate glide–polar performance analysis based on these results—many more data at a greater range of speeds would be required to do this satisfactorily (to be the subject of further wind-tunnel testing). However, these results are sufficient to provide an indication of the sink-rate of the Alula platform at nominal airspeeds for the purpose of answering the question of the feasibility of soaring urban orographic lift. It may be concluded that for the airspeeds measured (between 4.8 and 6.1 m/s) the sink-rate was in the order of 0.5 m/s. With the maximum error added the sink-rate would still be in the order of 0.7 m/s. As the available vertical velocity component (orographic lift) component of the flow upstream of the building (Fig. 11) is between 15% and 50% of the mean flow speed at building height it may be concluded that this in fact possible—provided that the mean wind speed at the height of the building is in the order of 3 m/s—and the platform may be flown continuously a region of sufficient lift backwards and forwards along the windward face of the building (perpendicular to the wind direction) or directly into wind (provided the airspeed matches the horizontal component of the wind). An additional consideration is that of the statistical distributions of mean wind speeds. Clearly this soaring would only be possible on days of the year where the wind speed is greater than 3 m/s (at building height). In stronger winds the Alula and MAV platforms of higher sink-rate would be able to soar, conversely at lower wind speeds the Alula would not be able to maintain height, an alternative platform with a lower sink-rate would be needed. Additionally, the velocity distribution should be nominally similar to that found here when the wind direction is closely orthogonal to any of the four faces of the building (within 7101). It is expected that the least potential for lift exists when the wind is 451 to any vertical face. The influence of yaw angle on the available lift is an important consideration, which is to be the subject of further investigation now that the feasibility of soaring has now been indicated. It has been demonstrated that increasing the turbulence intensity from nominally smooth to 12%, increases the timeaveraged performance of an airfoil at similar Reynolds numbers
(Watkins et al., 2009). However, those results were obtained via a series of wind-tunnel tests where the wing was held stationary and the effect of continuously varying the control inputs in order to maintain steady flight in turbulence is not known. To keep an MAV in the area of maximum lift in front of a building would require precise maintenance of placement via an active turbulence rejection system and well as specific guidance and navigation algorithms. The influence of turbulence on MAVs is an area of on-going research, particularly the relative span of the craft to the scale of the turbulence (Watkins et al., 2006 and Thompson et al., 2011). If these control challenges can be overcome additional ‘‘energy harvesting’’ possibilities exist using an on-board electric power plant (brushless motor) and windmilling propeller as a generator to power systems or change batteries to keep a platform airborne—provided the wind is blowing.
5. Conclusions The velocity of the flow field immediately upwind of a building has been mapped using at 1:100 scale building in a simulated vertical velocity gradient and validated by measurements taken from the full-size building. It was found that vertical component of velocity in this region (up to 8 m upwind of the building) was between 15% and 50% of the mean velocity component at this height, which depending on the wind strength means that it would be feasible to soar a MAV platform if it has a sink-rate of less than approximately 0.5 m/s for wind direction nominally onto the face of the building and speeds of 3 m/s (at building height). Additional challenges of guidance, navigation and turbulence rejection exist to keep the MAV flying in this region and these are to be the subject of further studies, particularly the development of an autopilot for active turbulence control to maintain position.
Acknowledgements The authors would like to acknowledge and thank P. Peterson for assistance with flow visualisation, Prof. C. Wang, Director of the Wackett Centre for providing E.W. Lim with internship funding, and E. Wettstein for assistance during wind-tunnel testing. The enthusiasm of M. Richter and support of DreamFlight is also greatly appreciated. References Allen, M., 2005. Autonomous soaring for improved endurance of a small uninhabited air vehicle. In: Proceedings of the 43rd AIAA Aerospace Science Meeting and Exhibit. 10–13 January, Reno, NV. Allen, M., 2007. Guidance and Control of an Autonomous Soaring UAV, NASA/ TM2007-214611. National Aeronautics and Space Administration, Washington, DC. Cermak, J., Peterka, J., Woo, H., 1977. Wind-Tunnel Measurements in the Wakes of Structures, NASA CR-2806. National Aeronautics and Space Administration, Washington, DC. Cowling, I., Willcox, S., Patel, Y., Smith, P., Roberts, M., 2009. Increasing persistence of UAVs and MAVs through thermal soaring. Aeronautical Journal 113 (1145), 479–489. Deittert, M., Richards, A., Toomer, C., Pipe, A., 2009a. Engineless unmanned aerial vehicle propulsion by dynamic soaring. Journal of Guidance, Control, and Dynamics 32 (5), 1446–1457. Deittert, M., Richards, A., Toomer, C., Pipe, A., 2009b. Dynamic soaring flight in turbulence. In: Proceedings of the AIAA Guidance, Navigation, and Control Conference. 10–13 August, Chicago, IL. Edwards, D., Silverberg, L., 2010. Autonomous soaring: the Montague crosscountry challenge. Journal of Aircraft 47 (5), 1763–1769. Frost, W., Shahabi, A., 1977. A Field Study of Wind Over a Simulated Block Building, NASA CR-2804. National Aeronautics and Space Administration, Washington, DC. Holmes, J.D., 2001. Wind Loading of Structures. Spon Press, London. Hooper, J.D., Musgrove, A.R., 1997. Reynolds stress, mean velocity, and dynamic static pressure measurement by a four-hole pressure probe. Experimental Thermal and Fluid Science 15 (4), 375–383.
C. White et al. / J. Wind Eng. Ind. Aerodyn. 103 (2012) 41–49
Kahveci, N., Ioannou, P., Mirmirani, M., 2008. Adaptive LQ control with antiwindup augmentation to optimize UAV performance in autonomous soaring applications. IEEE Transactions on Control Systems Technology 16 (4), 691–707. Kim, J., Baik, J., 2004. A numerical study of the effects of ambient wind direction on flow and dispersion in urban street canyons using the RNG k-e turbulence model. Atmospheric Environment 38, 3039–3048. Langelaan, J., 2007. Long distance/duration trajectory optimization for small UAVs. In: Proceedings of the AIAA Guidance, Navigation and Control Conference and Exhibit. 20–23 August, Hilton Head, SC. Langelaan, J., 2009. Gust energy extraction for mini and micro uninhabited aerial vehicles. Journal of Guidance, Control, and Dynamics 32 (2), 463–472. Lawrance, N., Sukkarieh, S., 2009. A guidance and control strategy for dynamic soaring with a gliding UAV. In: Proceedings of the 2009 IEEE International Conference on Robotics and Automation. 12–17 May, Kobe, Japan. Leuzzi, G., Monti, P., 1998. Particle trajectory simulation of dispersion around a building. Atmospheric Environment 32 (2), 203–214. Milbank, J., Loxton, B., Watkins, S., Melbourne, W., 2005. Replication of Atmospheric Conditions for the Purpose of Testing MAVs: MAV Flight Environment Project: Final Report, USAF AOARD 05-4075. United States Air Force, RMIT University, Melbourne. Pagliarella, R., 2010. On the Aerodynamic Performance of Automotive Vehicle Platoons Featuring Pre and Post-Critical Leading Forms. PhD Thesis. School of Aerospace, Mechanical and Manufacturing Engineering, RMIT University, Melbourne. Peterka, J., Meroney, R., Kothari, K., 1985. Wind flow patterns about buildings. Journal of Wind Engineering and Industrial Aerodynamics 21, 21–38. Ricciardelli, F., Polimeno, S., 2006. Some characteristics of the wind flow in the lower urban boundary layer. Journal of Wind Engineering and Industrial Aerodynamics 11, 815–832.
49
Sharan, V., 1975. On characteristics of flow around building models with a view to simulating the minimum fraction of the natural boundary layer. International Journal of Mechanical Sciences 17, 557–563. Snyder, C., Zimmerman, J., 1975a. Thermal soaring, Aviation Weather, FAA Flight Standards Service & National. Weather Service (NWS), Washington DC. Tennekes, H., 2009. The Simple Science of Flight: From Insects to Jumbo Jets, second ed. The MIT Press, Cambridge, MA. Thompson, M., Watkins, S., Homes, J., White, C., 2011. Span-wise wind fluctuations in open terrain as applicable to small flying craft. The Aeronautical Journal 115 (1173). TFI, 2011. /http://www.turbulentflow.com.auS (accessed 23.05.2011). Walshe, D.E.J., 1972. Wind-Excited Oscillation of Structures. Her Majesty’s Stationery Office, London. Watkins, S., Milbank, J., Loxton, B., 2006. Atmospheric winds and their effects on micro air vehicles. AIAA Journal 44 (11), 2591–2600. Watkins, S., Mousley, P., Hooper, J., 2002. Measurement of fluctuating flows using multi-hole probes. In: Proceedings of the Ninth International Congress on Sound and Vibration. 8–11 July, Orlando, FL. Watkins, S., Ravi, S., Loxton, B., 2009. The effect of turbulence on the aerodynamics of low Reynolds number wings. Engineering Letters 18 (3). EL_18_3_09. Watkins, S., Thompson, M., Loxton, B., Abdulrahim, M., 2010. On low altitude flight through the atmospheric boundary layer. International Journal of Micro Air Vehicles 2 (2), 55–67. Wolters, R., 1971. The Art and Technique of Soaring. McGraw-Hill, New York. Zhang, Y., Arya, S., Snyder, W., 1996. A comparison of numerical and physical modelling of stable atmospheric flow and dispersion around a cubical building. Atmospheric Environment 30 (8), 1327–1345.
Contents lists available at SciVerse ScienceDirect
Journal of Wind Engineering and Industrial Aerodynamics journal homepage: www.elsevier.com/locate/jweia
A feasibility study of micro air vehicles soaring tall buildings C. White, E.W. Lim, S. Watkins n, A. Mohamed, M. Thompson School of Aerospace, Mechanical and Manufacturing Engineering, RMIT University, GPO Box 2476, Melbourne, Victoria 3001, Australia
a r t i c l e i n f o
abstract
Article history: Received 27 May 2011 Received in revised form 6 February 2012 Accepted 14 February 2012 Available online 14 March 2012
Micro Air Vehicles (MAVs) are person-portable platforms, which due to their small size and corresponding manoeuvrability show great potential for surveillance in urban environments. Achieving useful endurance using on-board electric power plants remains challenging despite rapid advances in battery energy storage density. This paper examines experimentally the feasibility of using orographic ‘slope’ lift in urban built environments to enhance the operational capability of MAV platforms by increasing range and endurance. A representative building in an urban environment was selected for investigation. The flow-field velocity in the upwind region of a 1:100 scale model of this building was mapped in a wind-tunnel with a reproduced vertical velocity profile. The vertical velocity component was found to be in the order of 15%–50% of the mean wind velocity at building height. These results were compared with data measured on the full-size building and found to agree well. The sink-rate of a soaring MAV was measured through flight-testing and found to be less than the available vertical velocity component in the upstream flow field for average wind speeds which indicates that soaring is feasible, provided that controllability challenges in high-turbulence environments can be overcome. Crown Copyright & 2012 Published by Elsevier Ltd. All rights reserved.
Keywords: Micro air vehicle (MAV) Unmanned air vehicle (UAV) Urban soaring Building Orographic lift Range Endurance Flow topology
1. Introduction Micro Air Vehicles (MAVs) are essentially Unmanned Aerial Vehicles (UAVs) but smaller in size and dimension (personportable). Recently, there has been an increasing interest in utilising MAVs as surveillance platforms in urban environments for clandestine operations, or search and rescue coordination roles in the immediate aftermath of natural or man-made disasters. The performance of small aircraft, in particular range and endurance, is currently restricted by on-board energy storage limitations set by size, weight and type of battery/ electric sources (Kahveci et al., 2008). However, significant energy is readily available in the environment in which MAVs fly. Since the early investigation into flight, it was noticed that some birds were able to sustain flight for long periods of time without expending energy through wing flapping. Larger birds developed soaring techniques to extract energy from relative wind motions to reduce energy consumption (Lawrance and Sukkarieh, 2009). Humans have used the vertical air motions in the atmosphere for flight, but so far this has been limited to sport aviation, be it radio controlled models (such as slope and thermal soarers) or hang-gliders and
n
Corresponding author. Tel.: þ61 3 9925 6084. E-mail addresses: [email protected] (C. White), [email protected] (E.W. Lim), [email protected] (S. Watkins), [email protected] (A. Mohamed), [email protected] (M. Thompson).
sailplanes. The use of such energy is now becoming of interest for UAVs and MAVs as well (Langelaan, 2009). It has been estimated that the endurance of a small glider UAV (4.3 m wingspan) can be extended from 2 to 8 h in winter and to 14 h in summer using soaring flight (Allen, 2005), suggesting that soaring may be a way of reducing the limitations in on-board energy storage for MAVs. Soaring is generally divided into two types: static soaring and dynamic soaring. Static soaring is the process of flying in air that is rising, either from the vertical component of wind blowing up a slope (arising from natural features, such a hills or ridges), known as orographic lift or from thermals. Dynamic soaring involves flying trajectories through distributions of wind speed that have vertical components to increase kinetic energy (Lawrance and Sukkarieh, 2009). While dynamic soaring has been proven to be viable to increase the endurance of UAVs (Lawrance and Sukkarieh, 2009; Deittert et al., 2009a, 2009b); the flight trajectories do not appear to be possible in densely built urban environments or conducive to surveillance. Static thermal soaring has been explored in recent years as a way to improve UAV range and endurance. Thermals occur when air near the ground becomes less dense than the surrounding air as a result of heating or humidity changes at ground level (Allen, 2007). Birds ‘ride’ these thermals to remain airborne without flapping their wings. This technique is widely used by sailplane pilots to gain altitude. Many strategies for extracting energy from thermals for UAVs have been published (Allen, 2007; Cowling et al., 2009; Edwards and Silverberg, 2010).
0167-6105/$ - see front matter Crown Copyright & 2012 Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.jweia.2012.02.012
42
C. White et al. / J. Wind Eng. Ind. Aerodyn. 103 (2012) 41–49
The principle of building soaring appears to be feasible for MAVs and is most similar to slope soaring or the use of orographic lift. Sailplanes utilise vertical air motions related to wind over geographic features to stay in flight or gain height when slope soaring. Snyder and Zimmerman, 1975a stated that a sustained wind speed of 15 knots (7.72 m/s) or more perpendicular to a slope usually generates enough lift to support a sailplane and that the height of the lift is usually two or three times the height of the rise from the valley floor to the ridge crest. Wolters (1971) stated that the position of maximum lift will be found at about a 45-degree angle upwind and above the slope. While Langelaan (2007) provided a model for trajectory optimisation or duration extension of MAVs using such orographic lift, no work has been published regarding soaring of MAVs in urban environments using vertical components of the wind over buildings. However, examples do exist in nature of bird making use of orographic lift generated by man-made structures—such as herring gulls flying on the upwind side of a ferry or cruise ship (Tennekes, 2009). Extensive studies have been conducted to understand the flow around buildings in urban environments. This has included fieldwork studying the wakes of structures and turbulent flow around take-off and landing of aircraft (Cermak et al., 1977; Frost and Shahabi, 1977), dispersion of pollutants downwind of buildings (Zhang et al., 1996; Leuzzi and Monti, 1998; Kim and Baik, 2004), pedestrian wind comfort and numerous studies on wind loads on structures (Holmes, 2001; Peterka et al., 1985). According to Ricciardelli and Polimeno (2006), the characteristics of the wind flow in the urban environment are known to be somewhat different from those of a Turbulent Boundary Layer (TBL) naturally developed over a homogeneous rough surface. The mean wind speed will differ based on the macro-meteorological events, elevation and also the physical location and surroundings of the point of interest. Sharan (1975) investigated the characteristics of the flow around building models and found out that as long as the boundary layer thickness is greater than the model height, the average location of the front stagnation point lies around 85 per cent of the model height. These studies have predominantly focused on pressure distributions or downstream topology models, rather the velocity components immediately upstream of the building—an understanding of which is required for MAV building soaring.
A number of platforms exist that meet the specifications of MAVs that are person portable and may be capable of building soaring. These are recreational models used by radio control enthusiasts to either soar in thermals or in slope lift. However, it is not known if the orographic lift in urban environments is adequate to be soared by such platforms. Langelaan (2007) indicated the theoretical sink-rate of a small MAV similar to these platforms (with a wing area of 1 m2 and a mass of 10 kg) of 0.2 m/s is attainable, although was not measured. While many of these platforms are available at low cost, no work appears to have been published on measuring and characterising the glide of such—particularly with respect to sink-rates, flight speeds and the vertical air movements required to keep them airborne in typical wind speeds. The focus of this research is to build towards an understanding of the vertical flow field (updraught) in front of a building to investigate the feasibility of MAV soaring in an urban environment. A representative building within the RMIT University Bundoora Campus; ‘Building 201’, was selected due to its strategic position and the environment around it matched the topography of a suburban terrain (Fig. 1). A scale model of the building was used for wind-tunnel testing and the results validated by measurement from the roof of actual building. Results from these studies are then compared with the measured sink-rate and airspeed data of an MAV in order to study the feasibility of orographic soaring of MAVs in urban environments.
2. Materials and methods 2.1. Scale building testing Suburban terrain velocity and intensity profiles (based on Walshe, 1972) were generated at 1:100 scale in the RMIT University Industrial Aerodynamics Wind Tunnel. This involved modification from the nominally smooth, uniform test section flow by an array of upstream barriers and floor-mounted roughness elements to artificially thicken the boundary layer within a reasonable stream-wise distance. The test section is 9.2 m long, 2 m high and 3 m wide. The region of the test-section used for testing was 40 characteristic lengths (based on maximum barrier height) downwind from the array of barriers and over ten
Fig. 1. RMIT University ‘Building 201’: (a) north and (b) west face.
C. White et al. / J. Wind Eng. Ind. Aerodyn. 103 (2012) 41–49
characteristic lengths away from the roughness element. In order to rapidly measure and depict the velocity profiles arising from the many configurations of barriers and elements trialled, a stagnation tube rake and static pressure tubes were positioned at the location of interest (where the leading edge of the building was going to be placed), see Fig. 2. The pressures were displayed on a multi-bank inclined manometer which was found to be an efficient way of investigating the effects of the configurations. Once a suitably scaled profile was achieved it was documented in more detail with a TFI Cobra probe and traversing gear. Cobra probes are four-hole pressure probes offered commercially by Turbulent Flow Instrumentation (TFI); for details see TFI (2011) and Milbank et al. (2005). The probe used had a 2.6 mm multifaceted head, which was considered sufficiently small to have a minimum influence on the flow field. Cobra probes are able to resolve the three orthogonal components of velocity at up to 2 kHz and provide the static pressure, providing that the flow vector is contained within a cone of 7451 around the probe x-axis. Details of the system and examples of use can be found in Watkins et al. (2002) and verification and further details including dynamic
43
capabilities, probe operation methodology and calibration techniques can be found in Hooper and Musgrove (1997). The velocity profile is presented in Fig. 3, velocities were normalised to the mean velocity at building height. Generally the agreement is good except very close to the ground. For this study the locations of interest are around the top of the building and thus well above the stagnation point. Since small changes in the profile close to the ground make only minor changes to the position of the stagnation point (Sharan, 1975) this effect is considered to be negligible. The turbulence intensity profile is illustrated in Fig. 4. Turbulence intensity in the wind tunnel was measured using a TFI Cobra probe. Cobra probe reads turbulence intensity in all three orthogonal directions (Iuu, Ivv, Iww) and provides the overall turbulence intensity. As turbulence intensity is not the main focus of this paper, only the overall turbulence intensity was considered. Similar to the velocity profile, Fig. 4 shows a relatively good comparison between the two profiles except close to the ground. All testing were conducted at a nominal tunnel velocity of 10 m/s. A Reynolds number sensitivity test was conducted and indicated that the non-dimensional results were Reynolds number insensitive.
Stagnation tube rake
Array of barriers
Flow direction Building model
Connected to multi-bank manometer
Roughness element
Fig. 2. Building model (stagnation tube rake), roughness element and barriers (not to scale).
Fig. 3. Velocity profile reproduced in wind-tunnel.
44
C. White et al. / J. Wind Eng. Ind. Aerodyn. 103 (2012) 41–49
Fig. 4. Wind-tunnel turbulence intensity profile (no building present).
z x Wool tuft indicating mean flow direction
Fig. 5. Position of measurement grid upstream from model (not to scale).
The 1:100 scale building model had a width of 0.38 m, a depth of 0.38 m, and a height of 0.43 m. While the model replicated the building floor-plan, details such as windows, trimmings and ventilation systems were omitted. Since the blockage ratio was relatively low and all data were non-dimensionalised, no corrections were applied. Flow visualisation with helium bubbles was used for positioning Cobra Probes for local velocity measurements. The bubble wand was located at 0.5 m upstream from the building on the floor of the tunnel to minimise interference with the upcoming flow. The size and density of the helium bubbles and the lighting and exposure time were adjusted to achieve short streak lines, see Fig. 9. A vertical upstream plane aligned with the centreline of the building was established to measure the velocity field upstream of the building (Fig. 5). The spacing between measurement points was decreased in the region of greatest vertical velocity (as indicated with helium bubble tracks). Velocity components for the x, y and z axes were denoted by u, v and w respectively and conform to the sign convention specified in Fig. 5. The lateral distance (y) 0 m is at the centreline of the building, the height (z) of 0.43 m corresponds to the top of the building, and the longitudinal distance (x) of 0.02 m is upstream from the front face of the building. A Cobra probe was used for flow measurement at 35 points on the measurement grid indicated in Fig. 5. The angle of the tip of the Cobra Probe was varied at each point to ensure all data lay on the calibration surface of the probe (within the 901 cone of acceptance). This angle of inclination was measured at each point
Fig. 6. Cobra probe and mast on roof of Building 201.
with an inclinometer to allow later resolution of the velocity components. For each point there were three samples of 60 s. 2.2. Full-size building testing A northerly wind (bearing of 3601) was necessary for the analysis to reproduce the flow simulated by tunnel testing; at the time of testing the wind was approximately northerly with a small westerly component (3551–3601). The same Cobra probe used in the wind-tunnel testing was mounted on a mast with an adjustable angle of incidence on the roof of the building. The height of the mast was adjusted to 1.5 m above building top edge, which coincided with a point on the measurement grid from wind-tunnel testing. A wool tuft was tied to the rig to visualise the flow angle. The frame of the mast was then tilted in the direction of the flow with the angle measured using an inclinometer. An image of the mast during testing is presented in Fig. 6. Two samples of 15 min were used. 2.3. MAV flight-testing A robust MAV, yet posing minimal possibility of damage to the building and personnel, was desirable for testing; both for calibration flights from a flat field (see later) and for planned flights launched from the top of the building. The MAV selected
C. White et al. / J. Wind Eng. Ind. Aerodyn. 103 (2012) 41–49
was based on the Alula (evo series), a commercially available model (produced by Dream-Flight in California). This is a small discus-launch radio-controlled flying-wing glider with a wingspan of 0.9 m, wing-area of 0.167 m2, length of 0.48 m and approximate flying mass of 0.17 kg. A three-view drawing of the model is presented in Fig. 7. This was chosen due to the manufacturer’s claim of it being an ‘urban soarer’ from anecdotes that it was capable of soaring in unconventional locations due to low wing-loading (1.02 kg/m2) and full trailing-edge elevon control system, giving good manoeuvrability. These features mean that weak lift (flows with small vertical velocity component) can be successfully exploited in confined spaces. Additionally, the platform area of the Alula is analogous to that of a soaring raptor (hawk), lending itself to low recognisability as a man-made platform. This is potentially an important quality for MAVs used urban environments for surveillance purposes—easy detection of the platform can be reduced with such bio-mimicry. The discus launch technique (also referred to as a side-arm launch, whereby the platform is gripped by a wing-tip and accelerated in a circular path—similar to that of the discus-throw athletics event) enables good height ( 20 m) to be achieved by an operator—conceivably adequate to reach a lift band adjacent to a low rise building. Data on the range of flight speeds and associated sink-rates can be achieved through flight test or wind-tunnel measurements. Despite being the more challenging of the two, flight testing was selected for the measurements presented here, since not only could it generate data obtained under a range of turbulence intensities, it also offered much-needed insight into the control issues of flying in the turbulence inherent in the lower levels of the Atmospheric Boundary Layer (ABL). The relative turbulence characteristics experienced by MAVs flying through the ABL and the associated aerodynamic issues are significant and are considered further in Watkins et al. (2009) and Watkins et al. (2010). The craft was instrumented with light-weight data logging equipment (Eagle Tree Systems, LLCTM) to record barometric airspeed, altitude and position parameters using a modified eLoggerTM V4, Airspeed MicroSensor V3, Altitude MicroSensor V3 and GPS Expander Module respectively. A commercially available 2.4 GHz radio control system was used for flight control. A Prandtl-type pitot-static tube was installed on the right wing for airspeed and altitude measurements, positioned suitable far
GPS expander
45
Pitot tube
eLogger
Altitude MicroSensor
Airspeed MicroSensor
Fig. 8. Data-logging equipment installed in Alula.
forward of the aerofoil to give minimal interference for stagnation and static pressure. The installed equipment increased the mass of the craft to 0.212 kg from the original 0.17 kg. This installation is presented in Fig. 8. While it is acknowledged that increasing the flying mass increases the sink-rate (compared with a non-instrumented craft) the installation of onboard mission-specific equipment would likely result in a similar increase. Flat field flight tests (i.e. with no orographic lift) were conducted under a range of atmospheric winds; however, the data presented here are restricted to those obtained under very light winds (o2 m/s). Light wind conditions were desirable to ensure that airspeed could be stabilised and held constant for as long as possible. One author performed the launch, while another operated the flight controls. A number of flights were made with the discus-launch technique achieving a maximum height of approximately 15 m above the ground at the apex of the launch. Flight duration was typically 25–30 s. All parameters (airspeed, altitude and GPS) were sampled at 10 Hz. Each flight followed a similar profile that involved visually stabilising airspeed after launch and maintaining this constant airspeed for as long as possible (to obtain useful sink-rate data collection time) before reducing the airspeed and sink rate immediately prior to landing (to minimise damage). It was not possible to achieve a nominated airspeed as there was no telemetry to the pilot; however, provided the airspeed was in the normal flight envelope (above stall and below maximum) this was considered adequate, as determining sink-rate values for a range of airspeeds was useful.
3. Results and discussion 3.1. Velocity measurements for scale building model
0.9 m
0.48 m Fig. 7. Alula evo MAV platform (courtesy Dream-Flight).
A representative upstream flow field shown with helium bubble streak lines is presented in Fig. 9. The outline of scale ‘Building 201’ (lower left corner) has been enhanced for clarity. The grid for flow measurement has been added. The helium bubbles are seen as streaks due to the fine-tuning of the exposure time of the camera. Longer streaks indicate a higher velocity. The stagnation point can be observed in Fig. 9. The position of the stagnation point coincides with the Cobra probe measurements, which is approximately a height of 0.35 m. The mean overall velocity in the upstream flow field grid normalised to the mean velocity at the building height in the velocity profile (6.55 m/s) is presented in Fig. 10. In Fig. 11 the mean vertical velocity component (w) normalised to the mean velocity at building height is presented. A vector velocity field is shown in Fig. 12.
46
C. White et al. / J. Wind Eng. Ind. Aerodyn. 103 (2012) 41–49
Measurement grid
Measurement grid
z
x
Scale ‘Building 201’ Scale ‘Building 201’ Stagnation point
Fig. 9. Flow visualisation with helium bubbles.
1.0
0.8
0.45
0.50 Height (m)
Height (m)
0.50
0.4
0.45
0.6
0.4 Scale ‘Building 201’
0.40 0.2
Scale ‘Building 201’
0.05
0.08
Upstream distance (m) Fig. 10. Mean overall velocity compared with mean velocity at building height.
3.2. Full-size building measurements and comparison with model scale Velocity measurements are presented in Table 1 for Building 201 and a corresponding point from wind-tunnel testing. For purposes of comparison (as the wind speed during full-size building testing was different to that in the wind tunnel) velocity components are normalised to the mean flow velocity at this position. The maximum vertical component of the flow velocity was found to be approximately 50% of the mean velocity at building height, this region occurs approximately 451 upstream from and above the building height. This represents a peak vertical velocity component (w) of around 3.3 m/s. Substantial regions of vertical velocity exist of at least 1 m/s. The vertical velocity decreases towards the stagnation point of the building at approximately zE0.35 m on the upstream face of the building. This stagnation
z
x
0.40 0.1
0.35
0.02
0.3
0.2
0.0
0.35
0.5
Flow direction
0.0
0.02 0.05 0.08 Upstream distance (m) Fig. 11. Mean vertical velocity (w) compared with mean velocity at building height.
point occurs at approximately 80% building height which agrees with the studies of (Sharan, 1975). These features are confirmed by the flow visualisation in Fig. 9. 3.3. Consideration of errors Similar sources of error exist for the model and full-size building testing due to the use of the same velocity measurement instrumentation used. The Cobra probes are calibrated regularly and are temperature compensated and typical angle errors are negligible (less than one degree). The time-averaged velocity errors are very much a function of the mean velocity squared (since the velocities are deduced from pressure differences); however, considerable experience has shown an error of less than 0.3 m/s at the typical measured velocities for this work (Pagliarella, 2010). The main source of error occurs in the measurement of the inclination
C. White et al. / J. Wind Eng. Ind. Aerodyn. 103 (2012) 41–49
of the tip of the Cobra probe; these errors are of the order of 731 which will not significantly affect the conclusion drawn here. For full-size building measurement, error may have been introduced by the velocity deviating from northerly (with 51 of westerly).
47
3.4. Flight-testing sink rate results Flight data were analysed for each flight to identify a region of stabilised airspeed where the MAV was gliding under nominally equilibrium conditions in still air. Typical flight airspeed and altitude data is presented in Fig. 13, indicating the region used for sink-rate analysis. The discrete steps in altitude and velocity indicate the digital resolution of the data logger and the sharp peak in airspeed at two seconds is the result of the hand launch technique. Data from the other four flights also analysed with this approach are presented in Table 2. The airspeed is the mean for the analysis interval. Sink-rate was evaluated from the altitude loss for the interval divided by the time.
4. The possibilities of soaring The normalised velocity results of the full-size building test compare well with those measured for a similar point from the wind-tunnel testing (Table 1). The more significant lateral component (v) found for the full-size building data is attributed to the specific geometry of the roof of Building 201, which was simplified for the wind-tunnel tests. Although the number of samples is limited, the wind-tunnel tests appear to conservatively predict (under-predict) the vertical component of velocity at this point by a small amount. This indicates that the remainder of the flow field measurements from the wind-tunnel test are reliable (and can be used to predict the vertical velocity of this flow field), meaning that in the region measured the rate of lift will be between 15% and 50% of the mean velocity (wind speed) at this height, which
Fig. 12. Velocity vector plot.
Table 1 Velocity components. Case
Full-size
Table 2 Summary of flight test and sink-rate data. Normalised velocity component Sample Mean normalising v w velocity (m/s) u
1 2 Scale ‘Building 201’ 1 (wind-tunnel) 2 3
7.29 7.88 5.87 6.04 6.04
0.639 0.667 0.774 0.758 0.753
0.294 0.284 0.021 0.047 0.014
0.711 0.689 0.633 0.650 0.657
Flight Stabilised mean Airspeed number airspeed (m/s) standard deviation (m/s)
Analysis interval (s)
Altitude loss (m)
Sink-rate (m/s)
3 5 6 7 8
14.7 10.3 6.7 7.6 14.8
8 5 4 3 8
0.54 0.49 0.60 0.40 0.54
4.8 5.3 6.1 5.7 6.1
Fig. 13. Example airspeed and altitude data from flight 5.
0.29 0.18 0.30 0.34 0.34
48
C. White et al. / J. Wind Eng. Ind. Aerodyn. 103 (2012) 41–49
depending on the wind strength and direction may be adequate to keep a soaring MAV airborne. It is not possible to make direct comparisons or examine a mean of the measured sink-rates for different airspeeds as each airspeed will have a unique sink-rate determined by the parabolic glide– polar relationship. For every fixed-wing platform there will be an airspeed that corresponds to a minimum sink-rate, with greater sink-rates encountered at lower and higher airspeeds. This airspeed corresponding to minimum sink-rate typically occurs 10% above the stall speed (minimum flying speed) and as the Alula was flown near stall (indicated by control unresponsiveness and nosedrop) it can be assumed that airspeeds at which data were recorded are close to that for minimum sink-rate. While airspeed measurement resolution was in the order of 0.28 m/s, averaging the airspeed over the analysis interval significantly reduces low resolution of this error. The low standard deviation of airspeed for the analysis intervals (0.18–0.34 m/s) indicates reasonable airspeed reliability for the analysis intervals selected. Airspeed measurement calibration was confirmed by placing the Alula platform with logging equipment in a windtunnel and altitude calibration by placement at a range of known heights. The resolution of altitude (1 m) measurements limits the accuracy of the sink-rates determined, as such the maximum error was evaluated to be between 0.14 and 0.30 m/s (assuming the worst case altitude resolution error of 2 m) and also dependent on the length of the analysis interval; the longer the analysis interval the smaller the corresponding error (as it is altitude error is evaluated by dividing the error by the interval length). Because of this error it is not possible to base accurate glide–polar performance analysis based on these results—many more data at a greater range of speeds would be required to do this satisfactorily (to be the subject of further wind-tunnel testing). However, these results are sufficient to provide an indication of the sink-rate of the Alula platform at nominal airspeeds for the purpose of answering the question of the feasibility of soaring urban orographic lift. It may be concluded that for the airspeeds measured (between 4.8 and 6.1 m/s) the sink-rate was in the order of 0.5 m/s. With the maximum error added the sink-rate would still be in the order of 0.7 m/s. As the available vertical velocity component (orographic lift) component of the flow upstream of the building (Fig. 11) is between 15% and 50% of the mean flow speed at building height it may be concluded that this in fact possible—provided that the mean wind speed at the height of the building is in the order of 3 m/s—and the platform may be flown continuously a region of sufficient lift backwards and forwards along the windward face of the building (perpendicular to the wind direction) or directly into wind (provided the airspeed matches the horizontal component of the wind). An additional consideration is that of the statistical distributions of mean wind speeds. Clearly this soaring would only be possible on days of the year where the wind speed is greater than 3 m/s (at building height). In stronger winds the Alula and MAV platforms of higher sink-rate would be able to soar, conversely at lower wind speeds the Alula would not be able to maintain height, an alternative platform with a lower sink-rate would be needed. Additionally, the velocity distribution should be nominally similar to that found here when the wind direction is closely orthogonal to any of the four faces of the building (within 7101). It is expected that the least potential for lift exists when the wind is 451 to any vertical face. The influence of yaw angle on the available lift is an important consideration, which is to be the subject of further investigation now that the feasibility of soaring has now been indicated. It has been demonstrated that increasing the turbulence intensity from nominally smooth to 12%, increases the timeaveraged performance of an airfoil at similar Reynolds numbers
(Watkins et al., 2009). However, those results were obtained via a series of wind-tunnel tests where the wing was held stationary and the effect of continuously varying the control inputs in order to maintain steady flight in turbulence is not known. To keep an MAV in the area of maximum lift in front of a building would require precise maintenance of placement via an active turbulence rejection system and well as specific guidance and navigation algorithms. The influence of turbulence on MAVs is an area of on-going research, particularly the relative span of the craft to the scale of the turbulence (Watkins et al., 2006 and Thompson et al., 2011). If these control challenges can be overcome additional ‘‘energy harvesting’’ possibilities exist using an on-board electric power plant (brushless motor) and windmilling propeller as a generator to power systems or change batteries to keep a platform airborne—provided the wind is blowing.
5. Conclusions The velocity of the flow field immediately upwind of a building has been mapped using at 1:100 scale building in a simulated vertical velocity gradient and validated by measurements taken from the full-size building. It was found that vertical component of velocity in this region (up to 8 m upwind of the building) was between 15% and 50% of the mean velocity component at this height, which depending on the wind strength means that it would be feasible to soar a MAV platform if it has a sink-rate of less than approximately 0.5 m/s for wind direction nominally onto the face of the building and speeds of 3 m/s (at building height). Additional challenges of guidance, navigation and turbulence rejection exist to keep the MAV flying in this region and these are to be the subject of further studies, particularly the development of an autopilot for active turbulence control to maintain position.
Acknowledgements The authors would like to acknowledge and thank P. Peterson for assistance with flow visualisation, Prof. C. Wang, Director of the Wackett Centre for providing E.W. Lim with internship funding, and E. Wettstein for assistance during wind-tunnel testing. The enthusiasm of M. Richter and support of DreamFlight is also greatly appreciated. References Allen, M., 2005. Autonomous soaring for improved endurance of a small uninhabited air vehicle. In: Proceedings of the 43rd AIAA Aerospace Science Meeting and Exhibit. 10–13 January, Reno, NV. Allen, M., 2007. Guidance and Control of an Autonomous Soaring UAV, NASA/ TM2007-214611. National Aeronautics and Space Administration, Washington, DC. Cermak, J., Peterka, J., Woo, H., 1977. Wind-Tunnel Measurements in the Wakes of Structures, NASA CR-2806. National Aeronautics and Space Administration, Washington, DC. Cowling, I., Willcox, S., Patel, Y., Smith, P., Roberts, M., 2009. Increasing persistence of UAVs and MAVs through thermal soaring. Aeronautical Journal 113 (1145), 479–489. Deittert, M., Richards, A., Toomer, C., Pipe, A., 2009a. Engineless unmanned aerial vehicle propulsion by dynamic soaring. Journal of Guidance, Control, and Dynamics 32 (5), 1446–1457. Deittert, M., Richards, A., Toomer, C., Pipe, A., 2009b. Dynamic soaring flight in turbulence. In: Proceedings of the AIAA Guidance, Navigation, and Control Conference. 10–13 August, Chicago, IL. Edwards, D., Silverberg, L., 2010. Autonomous soaring: the Montague crosscountry challenge. Journal of Aircraft 47 (5), 1763–1769. Frost, W., Shahabi, A., 1977. A Field Study of Wind Over a Simulated Block Building, NASA CR-2804. National Aeronautics and Space Administration, Washington, DC. Holmes, J.D., 2001. Wind Loading of Structures. Spon Press, London. Hooper, J.D., Musgrove, A.R., 1997. Reynolds stress, mean velocity, and dynamic static pressure measurement by a four-hole pressure probe. Experimental Thermal and Fluid Science 15 (4), 375–383.
C. White et al. / J. Wind Eng. Ind. Aerodyn. 103 (2012) 41–49
Kahveci, N., Ioannou, P., Mirmirani, M., 2008. Adaptive LQ control with antiwindup augmentation to optimize UAV performance in autonomous soaring applications. IEEE Transactions on Control Systems Technology 16 (4), 691–707. Kim, J., Baik, J., 2004. A numerical study of the effects of ambient wind direction on flow and dispersion in urban street canyons using the RNG k-e turbulence model. Atmospheric Environment 38, 3039–3048. Langelaan, J., 2007. Long distance/duration trajectory optimization for small UAVs. In: Proceedings of the AIAA Guidance, Navigation and Control Conference and Exhibit. 20–23 August, Hilton Head, SC. Langelaan, J., 2009. Gust energy extraction for mini and micro uninhabited aerial vehicles. Journal of Guidance, Control, and Dynamics 32 (2), 463–472. Lawrance, N., Sukkarieh, S., 2009. A guidance and control strategy for dynamic soaring with a gliding UAV. In: Proceedings of the 2009 IEEE International Conference on Robotics and Automation. 12–17 May, Kobe, Japan. Leuzzi, G., Monti, P., 1998. Particle trajectory simulation of dispersion around a building. Atmospheric Environment 32 (2), 203–214. Milbank, J., Loxton, B., Watkins, S., Melbourne, W., 2005. Replication of Atmospheric Conditions for the Purpose of Testing MAVs: MAV Flight Environment Project: Final Report, USAF AOARD 05-4075. United States Air Force, RMIT University, Melbourne. Pagliarella, R., 2010. On the Aerodynamic Performance of Automotive Vehicle Platoons Featuring Pre and Post-Critical Leading Forms. PhD Thesis. School of Aerospace, Mechanical and Manufacturing Engineering, RMIT University, Melbourne. Peterka, J., Meroney, R., Kothari, K., 1985. Wind flow patterns about buildings. Journal of Wind Engineering and Industrial Aerodynamics 21, 21–38. Ricciardelli, F., Polimeno, S., 2006. Some characteristics of the wind flow in the lower urban boundary layer. Journal of Wind Engineering and Industrial Aerodynamics 11, 815–832.
49
Sharan, V., 1975. On characteristics of flow around building models with a view to simulating the minimum fraction of the natural boundary layer. International Journal of Mechanical Sciences 17, 557–563. Snyder, C., Zimmerman, J., 1975a. Thermal soaring, Aviation Weather, FAA Flight Standards Service & National. Weather Service (NWS), Washington DC. Tennekes, H., 2009. The Simple Science of Flight: From Insects to Jumbo Jets, second ed. The MIT Press, Cambridge, MA. Thompson, M., Watkins, S., Homes, J., White, C., 2011. Span-wise wind fluctuations in open terrain as applicable to small flying craft. The Aeronautical Journal 115 (1173). TFI, 2011. /http://www.turbulentflow.com.auS (accessed 23.05.2011). Walshe, D.E.J., 1972. Wind-Excited Oscillation of Structures. Her Majesty’s Stationery Office, London. Watkins, S., Milbank, J., Loxton, B., 2006. Atmospheric winds and their effects on micro air vehicles. AIAA Journal 44 (11), 2591–2600. Watkins, S., Mousley, P., Hooper, J., 2002. Measurement of fluctuating flows using multi-hole probes. In: Proceedings of the Ninth International Congress on Sound and Vibration. 8–11 July, Orlando, FL. Watkins, S., Ravi, S., Loxton, B., 2009. The effect of turbulence on the aerodynamics of low Reynolds number wings. Engineering Letters 18 (3). EL_18_3_09. Watkins, S., Thompson, M., Loxton, B., Abdulrahim, M., 2010. On low altitude flight through the atmospheric boundary layer. International Journal of Micro Air Vehicles 2 (2), 55–67. Wolters, R., 1971. The Art and Technique of Soaring. McGraw-Hill, New York. Zhang, Y., Arya, S., Snyder, W., 1996. A comparison of numerical and physical modelling of stable atmospheric flow and dispersion around a cubical building. Atmospheric Environment 30 (8), 1327–1345.