How To Study Animals With Wind Tunnels

ane. Introduction

Wind tunnels are essential tools for studying both vehicle and fauna flying in laminar and turbulent flows. The cardinal advantage of wind tunnels for biological research is that airspeed can exist finely controlled over an beast that remains stationary in the laboratory frame. This stock-still frame allows direct measurements of forces, kinematics and flow fields that would otherwise be unattainable. Inspired by these advantages, several laboratories have congenital open-circuit wind tunnels designed for studying bird flight in the past 50 years [1–9]. Open-circuit tunnels typically produce moderate turbulent intensities (ane–3%) and therefore provide valuable data virtually animal flying in moderate turbulence. Flying environments, however, span a broad range of turbulence, from the extremely laminar flows found in the upper atmosphere (less than 0.i% [10,xi]) to the extremely turbulent flows plant in forests and street canyons (10–xl% [12–14]). Turbulence significantly affects flight performance, because it promotes boundary layer transition and affects elevator/elevate measurements, particularly if lift-generating surfaces are operating most stall [xv]. Equally a result, turbulence has been shown to subtract the flight speed of bees [sixteen,17] and moths [18] and increase flying costs in hummingbirds [19].

To written report flying in extremely laminar environments, some laboratories take built specialized closed-circuit wind tunnels with very low turbulence levels. While modern open-circuit tunnels can produce turbulence intensities as low as 0.ane% [20], they cannot achieve the extremely low turbulence and temperature command possible in airtight-circuit tunnels. The Regal Institute of Applied science (KTH) [21,22] and Texas A&M [23,24], for case, take closed-circuit tunnels with centric turbulence intensities of less than 0.040% and less than 0.047%, respectively (measured at x one thousand southward⁻¹). A look at the properties of electric current country-of-the-art animal flying tunnels shows the merchandise-offs associated with low-turbulence tunnel design (table i). The tunnel at the University of Western Ontario produces college turbulence intensities (less than 0.three%) but is able to simulate flight at altitudes up to 7 km [25,26]. The tunnel at Lund University produces low turbulence (less than 0.06%) and tilts to simulate ascending (up to 8°) and descending (up to half-dozen°) flight [27–29]. None of these tunnels, however, can produce both highly laminar and highly turbulent flows.

Tabular array 1. Recent closed-excursion current of air tunnels produce low turbulence intensities. Modern closed-excursion tunnels utilise a combination of honeycombs, screens and contractions to create extremely low turbulence intensities. Centric turbulence intensity is the root mean foursquare fluctuation in streamwise velocity compared to the bulk airspeed. Measurements take been reported with different methods and positions within wind tunnels, and then identical weather for comparison are not bachelor in every case. The filigree over which measurements were made too varies; the grid at KTH [22], for instance, includes points closer to the wall than those reported here. Where possible, tunnels were compared at airspeeds relevant for animal flight (approx. ten k s^−ane). n.a., no information available.

Plummet
	location	test section west × h × l (k)	contraction ratio	max speed (m s^–1)	axial turbulence intensity (%)	boosted features
open circuit	Harvard U. [4]	1.ii × one.2 × ane.4	6	28.5	<1.28^a	animal flying
	U. Montana [6]	0.6 × 0.vi × 0.85	6	12	1.ii^b	animal flight
	U. Birmingham [5]	2.5 × two.1 × three.one	n.a.	northward.a.	∼0.viii^c	animal flight
	U. Illinois [20]	0.9 × one.ii × two.4	7.five	71.5	∼0.08^d
closed circuit	Saarland U. [7,eight]	one × one × 1	northward.a.	14	∼2^east	animal flying
	U. Western Ontario [25,26]	1.v × 1 × ii	ii.5	21	<0.3^f	beast flying
						hypobaric
						temp. control
						hum. control
	Lund U. [27–29]	ane.ii × 1.1 × 1.7	12.25	38	<0.06^g	fauna flight
						tilting section
						temp. control
	U. Southern California [thirty]	i.4 × 1.4 × 2.one	due north.a.	n.a.	∼0.025^h
	Texas A&Yard U. [23,24]	1.4 × 1.4 × 4.9	5.33	31	0.047ⁱ
	KTH [21,22]	ane.ii × 0.8 × seven	9	69	<0.040^j	temp. control
	Stanford	1 × 0.eight × 1.vii	7	50	<0.030^j	animal flight
						open/closed-jet
						turb. control
						temp. control

To accelerate the capabilities of fauna flight wind tunnels for biomechanics and aerodynamics studies, we built a multi-purpose tunnel that produces low streamwise turbulence (less than or equal to 0.028% at the centreline over the full operating range), but also high turbulence using an active grid of spinning vanes (up to approx. 45% at the centreline). The low turbulence levels are made possible by half-dozen seamless screens, a seamless honeycomb and a vii : 1 contraction ratio. The tunnel is besides barely audible at depression airspeeds. Measured racket levels in the test section reach 76.4 dB at 20 m due south^−ane, allowing quiet natural flight weather for animals and minimal acoustic disturbances to the flow. The low noise is a result of acoustic handling throughout the tunnel and an acoustic wall that separates the large diameter fan from the testing area. The tunnel test section (ane.0 one thousand wide, 0.82 m tall and 1.73 m long) can exist removed to operate the tunnel with an open jet of air in the testing area, allowing the placement of big diagnostic equipment, such every bit biplanar fluoroscopes for loftier-speed musculoskeletal recordings. A water-chilled heat exchanger keeps the examination section temperature steady (1σ = 0.007°C at x m s⁻¹) over a range of 10–30°C. The tunnel is washable, and all sections are attainable via hatches, making the tunnel easy to keep sanitary for animals. A system of redundant screens and netting keeps animals safely inside the testing area. The loftier-quality flow and the special features of the multi-purpose tunnel make information technology the kickoff of its kind. Hither, we give a complete description of its layout and its aeroacoustic operation. To ascertain the performance metrics for the air current tunnel, we used the KTH MTL wind tunnel as our principal reference, because it performs well and is relatively well documented [21,22].

2. Tunnel layout

The new current of air tunnel is a closed-excursion tunnel with a footprint of xv.1 × five.2 m and a maximum height of iv m. The components of the tunnel are designed to work together to create catamenia with uniform speed, uniform temperature, low turbulence and a depression dissonance profile (effigy 1).

Figure 1. The closed-circuit current of air tunnel uses flow conditioning to decrease turbulence and tin operate in both open- and closed-jet configurations. (a) The components of the current of air tunnel are shown past color. A large diameter fan drives air through a series of menses conditioning elements, then through a contraction into the test section (green). A heat exchanger and temperature probes (orangish) are used to regulate temperature in the tunnel. Airspeed is measured using static pressure ports in the stilling chamber and contraction (nighttime blue). The turning vanes and sidewalls shown in lite blue are filled with mineral wool to attenuate fan dissonance. To facilitate discussion in the text, the corners are labelled 1–4 equally shown. Electronic supplementary material, figure SF.ane shows a photo of the tunnel taken from inside the laboratory. (b) But upstream of the test department, an active turbulence grid can be positioned to inject turbulence into the menses. For low-turbulence experiments, a filler with identical dimensions is used instead. (c) The test section can exist removed, and collector flaps can be installed on the inflow of corner 1. The resulting 'open-jet' configuration allows easy access to an exposed jet of air in the testing surface area.

ii.1. Fan

The airflow is driven past a single stage, 18 bract, centric fan. The fan was selected based on its low acoustic signature, which helped to minimize acoustic levels in the test department. A i.5 g diameter fan was chosen to maintain the meaty size of the air current tunnel while meeting the wind tunnel'due south depression noise requirements. To reduce vibration transmission, the fan's steel housing was mounted on its own massive concrete foundation, separate from the building foundation and connected to the tunnel through expansion joints. The fan is powered past an 80 kW consecration motor (891 r.p.m. max speed; ±0.1 r.p.m. resolution), which is housed in an enclosed nacelle. A tail-cone is provided to farther ameliorate fan efficiency as well equally to provide practiced airflow performance downstream. The motor is h2o-cooled and controlled by a variable frequency drive which uses a 100 : 1 turndown ratio to regulate fan speed based on user input. The drive-motor associates is equipped with braking in the consequence of an emergency stop.

ii.2. Flow conditioning

A series of flow conditioning elements are used to straighten and laminarize the flow before it enters the test section. Turning vanes were designed to straighten the flow while minimizing induced turbulence and circuit pressure loss. A honeycomb in the stilling chamber (jail cell length/diameter = 16) straightens the flow before it enters the menstruation workout screens. To maximize menses alignment, the honeycomb was made seamless and compatible past welding together precisely crimped stainless steel strips. Downstream of the honeycomb, v seamless, fine mesh, stainless steel screens are used to reduce turbulence in the stilling sleeping room. The mesh screens are spaced 350 mm apart have wire diameters of 0.16 mm, nominal apertures of 0.5 × 0.5 mm and open areas of 57.four%. A sixth identical screen is located in the crossleg between corners iii and 4 for boosted turbulence reduction. Each of the screens is pre-tensioned to reduce sag and lateral deportation. To prevent menstruum perturbations in the stilling chamber, fairing plates were used to conceal the screen and honeycomb attachment points. Post-obit the honeycomb and flow conditioning screens in the stilling chamber, the flow passes through a 7 : 1 fibreglass contraction, which reduces turbulence intensity by accelerating the flow. The contraction length and curvature were designed with computational fluid dynamics (CFD) simulations to reduce velocity gradients at the contraction get out, thereby minimizing the anisotropy of turbulence in the test department.

two.3. Testing area

The testing area can be operated in closed jet, with the test section installed, or open jet, with the test section removed and collector flaps installed (figure 1c).

The closed-jet configuration allows maximum airspeeds with minimal turbulence, menses angularity and pressure variation. When operating in closed-jet configuration, a blackness anodized aluminium examination section (airline dimensions i.0 thou wide, 0.82 k alpine and 1.73 1000 long) connects the wrinkle to corner one. A breather gap downstream of the exam section assures that testing pressures are near atmospheric, allowing access for probes and supports without the demand for closed admission ports. Streamwise gradients in pressure are minimized by diverging test section walls that business relationship for boundary layer growth. The floor and ceiling divergence angle is stock-still (0.v°) while the sidewall departure bending is variable to adjust for different blockage levels (see the electronic supplementary fabric, SM.one).

The test section is designed for efficient high-fidelity diagnostics on creature and small vehicle flight. All components in and around the test section are matt black to facilitate high-speed videography and particle tracking. In improver, each wall of the test section contains a panel that can exist removed for repair, replacement, modification or for quick access to the exam section interior. Ii sets of side panels were built: translucent acrylic, designed for durability during training flights; and thin optical glass, designed for the high transparency required for laser-based flow measurements. The area downstream of the test section is equipped with side and flooring hatches (with windows) for easy access while working with animals (electronic supplementary textile, figure SF.ten).

The open-jet configuration allows fifty-fifty easier access to live animals during preparation sessions in the airflow. It also allows large diagnostic machinery, such as biplanar fluoroscopy emitters and receivers, to be positioned around the flow. To switch into open-jet mode, the test section (approx. 270 kg) is removed with a hydraulic cart, and matt black aluminium collector flaps are fastened to corner one. The angle of the flaps can be adjusted to maximize force per unit area recovery and minimize flow divergence.

two.4. Airspeed control

Airspeeds upwardly to l m southward⁻¹ in the test section are adamant by measuring static force per unit area readings in the stilling bedroom and the wrinkle. For both readings, pressure is mechanically averaged over two sets of iv permanently installed static pressure ports. I gear up of ports from each section connects to a differential force per unit area sensor (Rosemount 3051S) tuned to measure airspeeds from 0 to 15 thou s⁻¹. The other fix connects to a differential pressure sensor tuned to measure out airspeeds from 15–l m s⁻¹. This multi-sensor approach was called to ensure high-precision airspeed control at both depression and high speeds. A third pressure level sensor calculates guess pressure in the contraction by comparing the static pressure to atmospheric force per unit area. During commissioning, a Pitot-static probe (Kanomax 1 m) continued to a high-accurateness pressure transducer (MKS 690A/698A) was placed at the centreline in the test section. A second-order polynomial bend fit was generated betwixt the facility force per unit area port readings and the readings from the Pitot-static probe. This bend fit is used to correct the facility pressure port readings to the true dynamic and static pressures in the test section. Airspeed is calculated from these corrected pressures by using the isentropic period equations for an ideal gas, including correction factors for temperature and humidity (electronic supplementary material, SM.ii).

The airspeed is controlled in open loop by setting the fan speed (r.p.m.), which is determined from a user-divers airspeed via a lookup tabular array. The default fan speed lookup table for an empty test section was calibrated with the Pitot-static probe during commissioning. This lookup table tin be modified when blockage levels change—such as when operating in open jet—where the mapping from fan speed to airspeed is different. The lookup tables are updated through a custom Human Machine Interface written in LabVIEW that sends commands via Ethernet to a Programmable Logic Controller. Once the table has been uploaded, a user can enter an airspeed ready point into the Human Machine Interface for automated airspeed control.

two.5. Temperature control

The temperature in the tunnel is controlled by a Proportional-Integral controller that regulates input to a water-cooled heat exchanger. The controller compares a user-defined ready indicate temperature to the average reading of four resistance thermometers in corner iv. Based on this comparison, the controller operates two control valves that regulate the exchange of chilled water (approx. 7°C) with water pumped at constant rate through the oestrus exchanger. This utilise of a pocket-size chilled water loop to regulate the primary coolant loop was chosen to simplify the controller and thus reduce time to reach thermal equilibrium. The controller gains are automatically adjusted based on airspeed to ensure thermal equilibrium at all speeds. Because the tunnel is in a climate-controlled laboratory, a separate heater was non necessary; ambience heat and friction losses in the fan are sufficient to warm the tunnel to the airspeed-dependent temperature range (electronic supplementary cloth, figure SF.11).

The rut exchanger was designed to maximize water–air rut transfer, maximize temperature uniformity and minimize turbulence generation. The exchanger consists of two sets of vertical copper tubes connected to thin continuous horizontal aluminium fins, all of which are housed in an insulated galvanized steel frame to eliminate oxidation at the edges. To increment temperature uniformity, water is pumped at a constant flow rate and the menstruum direction reverses between the two sets of tubes. The exchanger is positioned just downstream of the diffuser following the fan. This position was called for three reasons: (i) the resulting dorsum force per unit area helps to keep the menstruum attached in the diffuser, (ii) any resulting turbulence has fourth dimension to disuse before reaching the exam section and (iii) the resulting dissonance and pressure drop are minimized by placing the exchanger in a department with wide surface area and thus low airspeed (predicted maximum of seven.1 m south^−ane).

2.6. Active turbulence grid

While the tunnel is designed to produce extremely laminar flow, it is also capable of simulating the highly turbulent weather condition that animals and vehicles encounter in the atmospheric boundary layer. Loftier turbulence is introduced with the use of an active filigree of spinning diamond-shaped vanes, a concept introduced by Makita & Sassa [32]. Our grid in particular was modelled after a like organisation adult by Cekli et al. [33,34]. The grid is made up of 7 rows and 8 columns of vanes, which are actuated independently by 150 W DC motors (Maxon RE40). For low turbulence conditions, the filigree of vanes can be replaced by an empty filler section (figure ib).

The active filigree of vanes can also be used to simulate not-uniform flows in the testing surface area. For example, past adjusting the fixed position of the vanes, a jet or wake profile can be created downstream of the grid. This technique could be used to investigate flight in turbulent wakes or jets and compare the results with depression turbulence conditions. The positions of the vanes can be controlled automatically based on loftier-speed marker tracking. This enables the grid to react in real time to the kinematics of animals or vehicles flight in the test expanse.

ii.7. Additional features

Expanding/contracting corners. To limit the overall circuit length while maintaining catamenia quality, corners 1 and 2 aggrandize with expansion ratios of 1.17 and 1.28, respectively. The corresponding diffusion half-angles are 1.5° and 2.6°. The estrus exchanger pressure drop immediately downstream of the backleg diffuser was included in the design to prevent menses separation at the expansion points. The values for the expansion ratio were chosen to minimize the tunnel footprint while maintaining flow uniformity and minimizing fan power consumption. Similar expansion ratios have been used in previous successful designs by Jacobs Engineering and were verified via CFD for the present wind tunnel.

Acoustic treatment. Acoustic fabric is used throughout the tunnel to create a low-noise environment in the test section. The internal side walls of the corners and crosslegs are lined with mineral wool (100 mm thickness) encased in perforated canvass metal, designed to attenuate sound in the mid-to-loftier frequency range. The turning vanes in corners i–3 are filled with mineral wool besides. The vanes in corner 1 are also covered with acoustically transparent carbon fibre material that redirects the flow smoothly while maintaining the acoustic absorption properties of the vane. For maximum noise reduction, synthetic washable fur covers tin be placed over the leading edges of the corner i turning vanes.

Cleaning and human condom. To maintain germ-free conditions for animals, the tunnel was designed such that it can be washed efficiently. All surfaces—both external and internal—tin can be cleaned with a low-pressure level wash of diluted bleach, which drains through ports in the tunnel floor. The purpose of the launder is to remove flow seeding residual, fauna faecal matter and/or feathers. A pulley system slides out the turbulence screens for like shooting fish in a barrel cleaning access. Ceiling panels are designed to support approximately 100 kg for piece of cake access to upper exterior sections and installation of equipment, and all internal sections are accessible through hatches on the sides or bottom of the tunnel. An interlock safety organisation prevents the tunnel from running when hatches near the fan are open up. Finally, the unabridged current of air tunnel laboratory is lead-shielded to meet 10-ray health and safe standards for loftier-speed biplanar fluoroscopy.

Animal rubber. A serial of protective screens, light amplification by stimulated emission of radiation-curtains and nets keeps animals confined to the testing area. First, an array of vertical steel strings (0.5 mm diameter, 10 mm spacing) prevents animals from flight upstream of the test department. Tensioned music wire (loftier-carbon steel) was chosen because of its high tensile strength and stiffness. These backdrop (i) minimize flow disturbance by ensuring that the strings are equally straight and thin as possible and (ii) minimize wire harm due to beaks and/or claws. Vertical wires were called to prevent animals from perching on the strings. The strings are matt black to facilitate high-speed particle tracking and avoid unwanted visual cues for animals flying in the test department. After birds are trained to fly reliably in the test section without inbound the pitch-black settling chamber, the array of wires tin be removed for experiments where minimal turbulence is required. Besides, engineering science aerodynamic research will exist performed without the array of wires. Therefore, the turbulence intensities reported here were measured with the array removed. Flexible netting was installed downstream of the test department and around the breather gap. Together, the wire array and the netting keep animals safely confined to the testing expanse. In addition, a concealed entrance hall effectually the tunnel and a permanent metal grid between corners one and 2 provide failsafe layers of secondary protection for animals. Finally, the laboratory can be hosed down, the air volume is refreshed 15 times per hr, and all lights are flicker-free to meet all modern animate being inquiry standards.

Humidity measurement. While relative humidity is not actively controlled, it is measured by a humidity sensor downstream of corner 4 and was found to stay relatively constant (±0.08%) over normal operating times. The humidity measurement is factored into the air density calculation to increase accurateness when recording airspeed (electronic supplementary material, SM.ii).

Fume purge valve. To allow fume visualization in the examination section, a fume purge valve was included in the ceiling of the diffuser downstream of the fan. The motorized valve is operated through the Human Machine Interface. Smoke is driven out by the fan force per unit area while fresh air is pulled in through the sabbatical gap.

3. Tunnel performance

Following installation, tests were performed to quantify the aeroacoustic properties of the wind tunnel. The flow in the test section is stable (1σ airspeed = 0.018 m s⁻¹ at the mid-range speed, 25 g due south⁻¹), uniform (1σ airspeed = 0.023 g due south⁻¹ at 25 thousand s⁻ⁱ), straight (1σ deviation angle = 0.14° at 25 m s⁻¹), depression turbulence (axial turbulence intensity = 0.021% at 25 1000 s⁻¹) and placidity (81.viii dB at 25 m s⁻ⁱ). The datasets resulting from the aeroacoustic testing have been uploaded equally office of the electronic supplementary cloth. Details of the testing procedures and results are given beneath.

three.1. Aeroacoustic testing traverse

A custom traverse was designed to measure the flow properties in the exam section (electronic supplementary material, figure SF.2). The base of the traverse was a steel strut positioned at ane of four horizontal positions (200 mm spacing) and four vertical positions (160 mm spacing) at the back end of the test section. A steel sting attached to the strut extended upstream to a point approximately i m downstream of the exam department entrance. The issue was a 4 × iv grid with 16 positions at which airspeed, pressure level, catamenia angularity, temperature and turbulence were measured. The strut can besides be positioned such that measurements are taken at the centreline of the test section. For each type of measurement, a different probe was mounted on the end of the sting: for airspeed and pressure measurements, a Pitot-static probe (Kanomax 1 m; sent to a MKS 690A/698A transducer); for flow angularity, a menstruum bending probe (CEA 5-hole; sent to a MKS 690A/698A transducer); for turbulence intensity, a hotwire probe (TSI 10-Motion picture Model 1241–20 with a CTA Thermal Anemometer AN-1003 system); and for temperature, a resistance thermometer. All measurements were conducted at a stable tunnel temperature of 20°C.

3.2. Airspeed stability and uniformity

The airspeed in the test department remained stable over the total operating range (0–50 m south⁻¹; figure 2a). To examination stability, airspeed was recorded for 240 due south at 1 Hz from the calibrated pressure ports. Seven set point airspeeds were considered over the operating range: five, 10, 20, 25, thirty, 40 and 50 m s⁻¹. Airspeed was more steady at the lowest speed (oneσ = 0.017 chiliad s⁻¹) than the highest speed (1σ = 0.060 m south⁻¹), just all airspeeds remained stable to inside less than 0.2% of the mean. The highest airspeed value (50 m due south⁻¹) was confirmed using measurements from the Pitot-static probe at the centreline.

The airspeed was spatially uniform in the test department (figure 2b). To test uniformity, airspeed was recorded from the Pitot-static probe (lx s at ten Hz) at each position of the iv × 4 testing grid. Airspeed uniformity was comparable at the two airspeeds considered: oneσ = 0.020 1000 s⁻¹ at 10 yard s^−one and iσ = 0.023 m s^−ane at 25 m south⁻¹. For completeness, we also present the total pressure uniformity past using data from merely the front port of the Pitot-static probe (electronic supplementary material, figure SF.three). Total pressure variation was slightly lower at 10 1000 s⁻¹ (oneσ = 0.344 Pa) compared with 25 one thousand s⁻¹ (1σ = 0.970 Pa). The uniformity in both airspeed and total pressure demonstrate that full force per unit area, static pressure and dynamic pressure level are all uniform in the test section over the total operating range.

The management of the flow was straight and uniform in the test section (effigy 2c). A menstruation bending probe measured flow direction past comparing pressure readings from ii ports on opposing sides of the probe. The probe was calibrated by performing a ringlet centrality angle sweep in both upright and inverted positions. Catamenia angle was then recorded from the probe (sixty south at 10 Hz) at each position of the 4 × 4 testing filigree. Yaw and pitch flow angles were measured separately, and so the total bending was calculated betwixt menstruum velocity and a streamwise unit of measurement vector (electronic supplementary material, SM.3). This full bending showed little variation over the test section at each airspeed considered: oneσ = 0.thirteen° at x m due south⁻¹ and oneσ = 0.14° at 25 grand s⁻¹. The low levels of flow angularity are made possible past the turning vanes and flow straightening elements upstream.

iii.3. Temperature and humidity stability and uniformity

The temperature and relative humidity remained stable over the full operating range (0–fifty k s⁻¹; effigy 3a). To test stability, temperature and humidity were recorded by the facility probes in the stilling bedroom (240 s at 1 Hz). 7 set up bespeak airspeeds were considered over the operating range: five, ten, xx, 25, 30, 40 and fifty g south^−ane. Temperature was more than steady at the lowest speed (aneσ = 0.007°C) than the highest speed (1σ = 0.073°C), but all temperatures remained stable to within less than 0.4% of the mean. The stability of temperature demonstrates the effectiveness of the manually tuned proportional-integral controller and the constant flow heat exchanger. Equally relative humidity was not actively controlled, its value is primarily a function of the ambience laboratory atmospheric condition. For example, the twenty m s⁻¹ condition was measured on a different day than the other airspeeds, resulting in the different humidity value recorded for that airspeed.

Figure 3. Temperature and humidity remain stable over the operating range; temperature is uniform in the test section. (a) Temperature in the stilling sleeping accommodation is shown while a proportional-integral controller regulates a water-chilled heat exchanger to continue the tunnel at twenty°C. Dashed lines testify the upper and lower limits of the tunnel'south operating range. The zoomed insert shows that temperature deviations from the mean temperature, $\bar{T}$ , vary depending on average airspeed, $\bar{U}$ . The standard divergence of T ranges from 0.007°C when $\bar{U}$ = 10 m s⁻¹ to 0.073°C when $\bar{U}$ = 50 m s⁻¹. (b) Relative humidity is recorded during tunnel operation. 1 trial ( $\bar{U}$ = 20 one thousand s^−one) was conducted on a different day, resulting in a unlike average humidity (approx. 63%) than the other trials (approx. 32–34%). The zoomed insert shows that relative humidity changes only slightly with no clear dependence on airspeed; the average standard deviation across the 7 airspeeds tested was 0.08%. (c) The deviation in temperature from the average tunnel temperature, ΔT, is measured on a iv × iv grid in the test department (black dots indicate filigree points). Deviations are slightly lower when $\bar{U}$ = 10 grand s⁻¹ ((i) 1σ of ΔT = 0.015°C) compared with when $\bar{U}$ = 25 m s⁻¹ ((2) 1σ of ΔT = 0.074°C). Colour indicates the relative temperature deviation, that is, $Δ T / \bar{T}$ .

The temperature was also spatially uniform in the examination section (figure threeb). To test uniformity, temperature was recorded by a resistance thermometer (60 s at ten Hz) at each position of the four × four testing filigree. The facility temperature changes slightly in the fourth dimension it takes to switch between grid signal trials. To split up the temporal and spatial variations in temperature, temperatures were corrected past comparing to an average facility temperature (electronic supplementary material, SM.four). Temperature was uniform at each of the two period airspeeds considered, though more uniform at the lower of the two airspeeds: aneσ = 0.015°C at 10 m s⁻¹ and aneσ = 0.074°C at 25 1000 s⁻¹. The loftier levels of uniformity demonstrate the effectiveness of the streamlined fins in the abiding menstruation rut exchanger.

3.four. Turbulence intensity

Turbulence intensities are low in the examination section over the full operating range (0–50 m due south^−ane; figure 4a). 2 turbulence intensities were considered: the centric turbulence intensity, $q_{u} \equiv \bar{u'} / \bar{U}$ , and the transverse turbulence intensity, $q_{five} \equiv \bar{v'} / \bar{U},$ where $\bar{u'}$ and $\bar{v'}$ are the root hateful square fluctuations in streamwise and vertical flow speed, respectively, and $\bar{U}$ is the time-averaged bulk flow speed. To calculate turbulence intensities, velocity data were sampled at 10 kHz using an 10-film hot wire probe (TSI Model 1241–20). The frequency response charge per unit of the probe is airspeed dependent, just it is always higher up 100 kHz and thus more than 10 times our sampling rate. The Ac and DC components of the signal were sampled together (DC coupled), with no gain or offset, using a low-noise anemometer (CTA Thermal Anemometer AN-1003). The maximum bespeak to noise ratio for the sampled signal was 96 dB based on the 16 bit analogue-to-digital conversion. Notch filters were applied between sixtyi ± 1 Hz, where i = ane, 2, 3, … , to remove electrical noise from the velocity point (electronic supplementary material, effigy SF.4). The energy content of the turbulence was concentrated beneath 1 kHz (electronic supplementary material, effigy SF.5). At all airspeeds considered (10, 20, 25, 30, 35, 40, 45 and fifty m due south^−one), both axial and transverse turbulence at the centreline were establish to exist low compared with existing creature flying and small flight vehicle wind tunnels: q_u ≤ 0.028% and q₅ ≤ 0.030%.

Highpass filtering has a big effect on the axial turbulence intensities reported [15]. Previous studies accept used varying filters, documented with varying degrees of precision, making information technology difficult to compare the turbulence performance of existing tunnels. To overcome this complication, we applied five different filter types based on the entries of table 1 (see electronic supplementary material, SM.5 for detail on the filtering and analysis technique). Filtering has only a moderate consequence on transverse turbulence intensities, just a significant effect on axial turbulence intensities. This is to be expected since low-frequency streamwise travelling waves primarily bear upon axial flow speeds. A comparison of unfiltered data suggests that our facility has a college travelling wave contribution than the KTH current of air tunnel (boilerplate central q_u = 0.144% at ten m s^–1 compared with less than 0.1% reported at KTH at x m south⁻¹) [14].

Turbulence intensities were non only low at the centreline, but also throughout the examination department (figure 4b). Axial and transverse turbulence intensities were measured at each grid point using the same technique equally was used at the centreline (TSI Ten-Wire Model 1241–twenty). Both axial and transverse intensities showed little variation across the four × 4 filigree (1σ < 0.01%) for both speeds at which the uniformity was measured.

3.five. Acoustics

A dissever mount was built for the acoustic probe (electronic supplementary textile, figure SF.6). A forward-swept fibreglass strut extended through a slotted floor panel in the bottom of the test section. The strut was supported by an aluminium frame mounted to the bottom of the test section. Acoustic levels were measured at the cease of the strut using a 60 kHz microphone (B&K 0.5 in 4166). To improve accuracy, the signal was sent through a signal conditioner (B&K NEXUS 4-Channel) and a pre-amplifier (B&One thousand 2669), resulting in a hardware-induced sixty Hz highpass filter. The cooling pump was on during acoustic testing, and record was placed over spiral hole indentations in the frame, too as gaps between movable windows and the test department, as is mutual practice for audio-visual measurements.

Following Johansson [22], a streamlined GRAS custom pinhole nose cone was attached to the front of the microphone before measuring acoustic levels. The main result of the nose cone is to attenuate the sound pressure (SPL) at frequencies above 1500 Hz (effigy 5a), which are presumed to be a result of pocket-size turbulence and therefore non relevant to the acoustic signature of the air current tunnel [22]. The olfactory organ cone likewise introduced a pocket-size jump in SPL effectually eight kHz, presumably due to resonance in the nose cone. This effect is negligible, because the SPL contribution at these high frequencies (approx. 10 dB) is several orders of magnitude lower than contributions in the threescore–1500 Hz range (approx. 70 dB). However, at some frequencies beneath 1500 Hz, nose-cone resonance increased the acoustic response by up to 8 dB. This amplification is likely to cause the reported overall audio-visual levels to exist slightly overestimated.

Figure 5. — Effigy 5. Noise levels in the examination department are barely audible at low speeds, then increase with airspeed; racket decreases steadily with frequency except for localized tones. (a) SPL is measured at the centreline of the test department, both with a standard nose cone (standard mic) and with a GRAS custom pinhole nose cone mounted on the front. The olfactory organ cone was used to benumb high-frequency turbulence (higher up 1500 Hz), but it also caused a slight amplification of frequencies below 1500 Hz and a slight bound in SPL around 8000 Hz, presumably due to resonance in the olfactory organ-cone crenel. All further trials were performed with the nose cone attached. (b) Calculation pile fabric to the leading edge of the acoustic turning vanes in corner 1 causes a slight decrease in SPL, especially at high frequencies. (c) SPLs subtract with increasing frequency and have localized tones. The zoomed insert shows the tones more conspicuously, along with expected test section resonant frequencies: the blade passing frequency (coloured dashed lines, xviii blades × fan frequency = 107, 133, 158, 209, 260 Hz) and the audio-visual harmonics based on width and height of the test section (grey solid lines; contained of airspeed; i and j are the numbers of horizontal and vertical continuing waves, respectively, in the acoustic harmonic manner). (d) Overall SPLs increment with airspeed, ranging from 53.2 dB at 0 m s^–i to 94.5 dB at 50 thousand s⁻¹ with no filtering beyond the congenital-in lx Hz highpass filter. A 180 Hz filter, a nose-cone transfer office, and an A-weighting filter are used to demonstrate the event of highpass filtering and resonance correction (electronic supplementary fabric, table ST.iii).

To reduce noise farther, pile fabric (synthetic fur) was wrapped around the leading edge of the audio-visual turning vanes in corner 1 (electronic supplementary fabric, figure SF.7). This add-on of pile fabric led to a small only measurable decrease in the test department SPL, especially at higher frequencies (figure 5b).

Frequency contributions to sound levels in the test department showed a ho-hum decrease in magnitude with several localized tones (figure fivea–c). Many of the tones occur in the zero airspeed condition and are therefore attributed to audio-visual eigenmodes in the exam department. The i-jth transverse eigenfrequency, f_ij , of a rectangular air column is f_ij = (c/ii)[(i/w) + (j/h)²]^1/2 [35], where c is the speed of sound and westward and h are the width and height of the air cavalcade, respectively. For the test section, the get-go few eigenfrequencies occur in the 100–500 Hz range (effigy 5c), where many of the local SPL spikes occur. The spikes are not expected to line upwardly perfectly with the eigenfrequencies, because the tunnel walls diverge. Even so, the commencement 2 horizontal modes (i = 1, 2; j = 0) bear witness local spikes in SPL shut to their predicted values. Other spikes below 250 Hz tin exist attributed to the bract passing frequency, which causes a small-scale airspeed-dependent spike in the SPL.

Overall, audio-visual levels in the test section are low; at the lowest airspeeds the wind tunnel is barely audible above the room background noise. The overall sound pressure level (OASPL) rises steadily with airspeed, with unfiltered values ranging from 53.2 dB at 0 k s⁻¹ to 76.iv dB at xx yard s⁻¹ to 94.5 dB at l m s⁻ⁱ (figure 5d). Equally with turbulence intensity, the frequency cutoff of a highpass filter has a pregnant result on the reported OASPL, making it difficult to compare existing wind tunnels. For this reason, we study our results with multiple filters. The first filter is the hardware-induced 60 Hz highpass filter from the signal conditioner. A 180 Hz filter was applied with the aforementioned routine that was used for turbulence intensity (electronic supplementary cloth, SM.5). For a third filtering option, we demonstrate the effect of applying a transfer function inspired past Johansson [22] to account for nose-cone distension below 1500 Hz (electronic supplementary textile, table ST.3). Finally, we use A-weighting (electronic supplementary textile, tabular array ST.three) to estimate dissonance levels based on human perception of audio.

3.6. Active turbulence grid

The active turbulence filigree can increase turbulence intensities upward to 16 or 45%, depending on the highpass filter used. To facilitate comparison with other studies using agile grids, we practical a grid motion that has been studied extensively [33,34]. For this motion, the vertical vanes stay at a fixed bending to generate constant blockage, and the horizontal vanes oscillate with a given maximum velocity, v _max, and aamplitude, φ _H. We studied the axial turbulence intensity, q_u , over a range of motion inputs spanning the operating range of the motors driving the vanes (5 × 6 grid equally spaced over the domain [v _max, φ _H] ∈ [0, 1500] r.p.thousand. × [0, 35]°). For each motion, 120 due south of information were sampled at x kHz from a single hotwire probe (Dantec 55P16; CTA module Dantec 54T42) placed on the testing sting at the centreline of the test department. Unfiltered centric turbulence intensities ranged from 12.5% ([5 _max, φ _H] = [0, 0]) to 45.1% ([v _max, φ _H] = [1200 r.p.m., 26.25°]) (effigy 6a). The low value occurs when the vanes are fully open and motionless, passively introducing turbulence into the menses. As with the intensities in the empty test section, applying a highpass filter changes the reported intensity. With a 0.i Hz highpass filter, the range changes to 12.4–44.6%; with a i Hz highpass filter, 12.0–twoscore.1%; with a 20 Hz highpass filter, 8.1–16.1%. The significant reduction in maximum turbulence intensity with the xx Hz highpass filter demonstrates the fact that the grid is primarily injecting turbulence at low frequencies, which practise not accept time to decay into broadband turbulence since the flow is not likely to be fully developed by the time it reaches the probe [36]. Nosotros chose to focus on unfiltered turbulence values for the active grid, because unlike the catamenia with no filigree, where low-frequency perturbations represent travelling waves in the tunnel, the flow with the filigree has low-frequency perturbations that take been injected purposefully.

Figure 6. An agile turbulence filigree can be used to create turbulence or non-uniform flows in the examination section. (a) A schematic of the agile turbulence filigree shows how 8 vertical and seven horizontal vanes can be independently rotated to adjy the flow upstream of the test section. (b) Grid motions based on those by Cekli *et al*. [33,34] produce centric turbulence intensities ranging from approximately 12.5% to approximately 45.1% (based on unfiltered velocity data). For these motions, the vertical vanes were held at a fixed angle while the horizontal vanes oscillated with a prescribed amplitude, φ _H, and max rotation speed, v _max. Grey boxes correspond to φ _H–v _max conditions across the physical limitations of the motors. These information are shown in tabular class in electronic supplementary material, table ST.4. (c) Filigree vanes were switched between fixed positions to change a uniform flow to a jet or wake flow (electronic supplementary material, figure SF.viii). The average airspeed at sixteen y positions shows a jet (i) and wake (ii) after effectually 1 s. The curves and shaded bands evidence an average and standard deviation of three identical trials. For each trial, airspeeds were averaged over i second before the vanes switched positions (−one s < time(t) ≤ 0 s) and each of the 3 southward after the vanes switched positions (0 s <t ≤ i s; 1 due south <t ≤ 2 due south; 2 southward <t ≤ 3 s). (d) Vane positions can conform based on the position of a tracked animal or vehicle. Here, the concept is demonstrated by opening i vertical vane upstream of a small quadcopter with a mark tracked by a motion capture system. Instantaneous airspeed profiles are shown at 2 fourth dimension steps (t _i and t ₂) corresponding with two frames from the video (electronic supplementary material, video SV.one; selected frames candy with Adobe Photoshop: Max Levels to 80 to improve contrast).

The active grid tin can also exist used to create non-uniform velocity profiles that include values upwards to ±40% of the mean airspeed. These profiles are created by adjusting the fixed position of the vanes. Changing the vane position tin alter the blockage and therefore the force per unit area drop beyond the active grid, which changes the hateful airspeed in the tunnel at a fixed fan speed. When the airspeed changes, it can take approximately 10 s to reach a new equilibrium as the mass of air is accelerated in the tunnel. I style to avert this delay is past switching between profiles with the same mean airspeed. We used this concept to demonstrate a switch from compatible flow to jet and wake profiles in approximately 1 s (figure 6b). The profiles were measured with an array of 16 wind sensors (Modernistic Device Rev P hot movie; 10 Hz [37]) spaced v cm apart horizontally, mounted on the testing sting centred in the exam department. Airspeed data for each sensor were nerveless at x Hz and averaged over 30 south.

While the 1 s filibuster is still longer than the timescales of wingbeats or rapid flight manoeuvres, the profiles can suit to the changing position of an brute or vehicle. We demonstrated this possibility by adding a 4 mm diameter retroreflective marker to the base of a small quadcopter (Estes Proto Ten) flying downstream of the active filigree. Six infrared cameras (Qualisys Oqus 7 plus; g Hz; iii MP) tracked the position of the quadcopter, and the position information were relayed to the agile grid control system. All horizontal vanes were held fully open, and all vertical vanes were closed except for the i closest to the lateral position of the quadcopter. The result is a jet with constant blockage that follows the position of the quadcopter (effigy 6c). While the quadcopter was moving as well quickly for the jet to friction match its position in real time, the successful tracking shows that the agile grid could send gust perturbations using a closed-loop controller based on the positions/orientations of animals/vehicles in the exam section.

4. Conclusion

The aeroacoustic operation of the new wind tunnel at Stanford is comparable to existing state-of-the-fine art wind tunnels. We obtained streamwise turbulence levels less than or equal to 0.030% throughout the test section at 10 m south⁻¹, compared with less than 0.04% at KTH [21] using the aforementioned filtering technique, demonstrating comparable turbulence levels. While we are unable to recreate the filtering used at Lund [27–29], their use of 1024 samples taken at 1 kHz suggests an effective highpass cutoff frequency of approximately i Hz. When using a ane Hz highpass frequency, our streamwise turbulence intensities were approximately 0.05%, comparable to the values reported at Lund. These values are also comparable to those at Texas A&One thousand, where 1 Hz was likewise used as a lower cutoff frequency [23,24]. The air current tunnel at KTH used similar acoustic handling and testing, allowing a straightforward comparison of our sound levels. Racket levels of 69 dB were reported at 35 m due south⁻¹ in the KTH tunnel, which is lower than our reported 84.4 dB [22]. The difference with KTH could exist somewhat exaggerated because of the transfer function used during acoustic measurements at KTH. When we apply a transfer function inspired by theirs, our reported levels decrease past approximately 6% (electronic supplementary material, table ST.3). Even with this filter the KTH tunnel is more than tranquility in comparison; regardless, both tunnels are barely audible in the test department at animal flight airspeeds.

Our new current of air tunnel has comparatively low turbulence, low flow angularity, high menstruation uniformity and high temperature uniformity. Two additional features make our animal flight wind tunnel unique: the power to switch from open to airtight jet configuration and the ability to tune turbulence intensity with an active grid. Existing tunnels are uniquely suited for studying different aspects of animal flying. With the ability to vary climb angle, the tunnel at Lund [27,29] is ideal for studying migration and ecology. With the ability to vary pressure, the tunnel at Western Ontario [25,26] is platonic for studying high-altitude flight and bioclimatology. The abilities to vary turbulence and perform open-jet experiments with biplanar fluoroscopy make the new wind tunnel at Stanford ideal for studying comparative biomechanics.

Ethics

The new wind tunnel facility complies with all state and federal regulations governing the humane handling of research animals. The Authoritative Console on Laboratory Animal Care at Stanford endorsed the facility later commissioning, and they perform biannual inspections to ensure the facility continues to comply with all protocols. The laboratory animal care programme at Stanford is accredited past the Association for the Assessment and Accreditation of Laboratory Creature Care.

Data accessibility

Data available from the Dryad Digital Repository: http://dx.doi.org/10.5061/dryad.b65fj [38].

Authors' contributions

D.Fifty. provided custom pattern specifications from which the tunnel was designed. A.W. was the project manager for the current of air tunnel construction. D.B.Q. managed the completion and testing of the turbulence generation system. A.W. and T.N. led the main aeroacoustic testing and commissioning of the air current tunnel. D.B.Q., A.W., T.N. and D.L. interpreted the results and wrote the manuscript.

Competing interests

The authors declare no competing interests.

Funding

The wind tunnel was developed with financial support from Stanford University.

Acknowledgements

We would like to thank Jacobs Engineering Group for their aid with designing, edifice and commissioning the wind tunnel, specifically Ed Duell and Megan Krause for aeroacoustic testing and various analyses. We thank Brian Carilli, Kevin Manalili, Sandy Meyer and Chris Crismon for laboratory infinite pattern and project management. We thank Fritz Prinz, Ken Goodson and Stanford University ME kinesthesia for their support. Nosotros thank Yous van Halder and Sofia Minano for help with the pattern and testing of the active turbulence grid. For assistance building the brute condom screens, nosotros thank Craig Milroy, Ben Perlman, Mara Young and Lakhbir Johal. We thank Eric Chang for his help with the quadcopter tracking demo. For advice we thank Willem van de H2o, GertJan van Heijst, Mico Hirschberg, Arne Johansson, William Saric, John Eaton, Peter Bradshaw, Leo Veldhuis, Loek Boermans, Niels Rattenborg, Ninon Ballerstadt, Anders Hedenström, Florian Muijres, Cristopher Guglielmo, Bret Tobalske, Andrew Biewener and William Dickson. Nosotros give thanks Colin Pennycuick for current of air tunnel design advice from his wonderful book 'Modelling the Flying Bird'.

Footnotes

Electronic supplementary material is available online at https://dx.doi.org/x.6084/m9.figshare.c.3721822.

Published by the Royal Order nether the terms of the Creative Commons Attribution License http://creativecommons.org/licenses/by/4.0/, which permits unrestricted use, provided the original writer and source are credited.

References

i
Tucker VA, Parrott GC
. 1970 Aerodynamics of gliding flight in a falcon and other birds. J. Exp. Biol. 52 , 345–367. Crossref, ISI, Google Scholar
2
Pennycuick CJ
. 1968 A wind-tunnel report of gliding flying in the pigeon Columba livia . J. Exp. Biol. 49 , 509–526. Crossref, ISI, Google Scholar
3
Butler PJ, West NH, Jones DR
. 1977 Respiratory and cardiovascular responses of the pigeon to sustained, level flight in a wind-tunnel. J. Exp. Biol. 71 , 7–26. Crossref, ISI, Google Scholar
iv
Hedrick TL, Tobalske BW, Biewener AA
. 2002 Estimates of circulation and gait change based on a three-dimensional kinematic analysis of flying in cockatiels (Nymphicus hollandicus) and ringed turtle-doves (Streptopelia risoria). J. Exp. Biol. 205 , 1389–1409. Crossref, PubMed, ISI, Google Scholar
v
Ward Due south, Bishop CM, Woakes AJ, Butler PJ
. 2002 Heart rate and the rate of oxygen consumption of flying and walking barnacle geese (Branta leucopsis) and bar-headed geese (Anser indicus). J. Exp. Biol. 205 , 3347–3356. Crossref, PubMed, ISI, Google Scholar
6
Tobalske BW, Puccinelli LA, Sheridan DC
. 2005 Contractile action of the pectoralis in the zebra finch according to mode and velocity of flap-bounding flight. J. Exp. Biol. 208 , 2895–2901. (doi:10.1242/jeb.01734) Crossref, PubMed, ISI, Google Scholar
vii
Ward S, Möller U, Rayner JM, Jackson DM, Bilo D, Nachtigall Westward, Speakman JR
. 2001 Metabolic power, mechanical ability and efficiency during wind tunnel flight past the European starling Sturnus vulgaris . J. Exp. Biol. 204 , 3311–3322. Crossref, PubMed, ISI, Google Scholar
viii
Bruderer LU, Liechti Atomic number 26, Bilo DI
. 2001 Flexibility in flight behaviour of befouled swallows (Hirundo rustica) and house martins (Delichon urbica) tested in a air current tunnel. J. Exp. Biol. 204 ,1473–1484. Crossref, PubMed, ISI, Google Scholar
9
Mueller TJ
. 2000Aerodynamic measurements at low Reynolds numbers for fixed wing A1:U22-air vehicles. In Proc. of the RTO A VT/VKI Special Course on Development and Operation of UAVs for Military and Ceremonious Applications, thirteen–17 September 1999. Belgium: VKI. Google Scholar
10
Hoblit FM
. 1988 Gust loads on aircraft: concepts and applications . Washington, DC: AIAA. Crossref, Google Scholar
11
Reeh AD
. 2014Natural laminar menstruation airfoil behavior in prowl flight through atmospheric turbulence. PhD dissertation. Technische Universität Darmstadt, Darmstadt, Germany. Google Scholar
12
Clarke JF, Ching JK, Godowitch JM
. 1982 An experimental report of turbulence in an urban environment . Environmental Protection Agency, Research Triangle Park, NC: Environmental Sciences Research Lab. Google Scholar
thirteen
Watkins S, Milbank J, Loxton BJ, Melbourne WH
. 2006 Atmospheric winds and their implications for microair vehicles. AIAA J. 44 , 2591–2600. (doi:10.2514/i.22670) Crossref, Google Scholar
fourteen
Roth 1000
. 2000 Review of atmospheric turbulence over cities. Q. J. R. Meteorol. Soc. 126 , 941–990. (doi:ten.1002/qj.49712656409) Crossref, ISI, Google Scholar
15
Bradshaw P, Pankhurst RC
. 1964 The design of low-speed wind tunnels. Prog. Aerosp. Sci. 5 , 1–69. (doi:10.1016/0376-0421(64)90003-10) Crossref, Google Scholar
16
Combes SA, Dudley R
. 2009 Turbulence-driven instabilities limit insect flight performance. Proc. Natl Acad. Sci. 106 , 9105–9108. (doi:10.1073/pnas.0902186106) Crossref, PubMed, ISI, Google Scholar
17
Ravi South, Crall JD, Fisher A, Combes SA
. 2013 Rolling with the flow: bumblebees flying in unsteady wakes. J. Exp. Biol. 216 , 4299–4309. (doi:10.1242/jeb.090845) Crossref, PubMed, ISI, Google Scholar
18
Ortega-Jimenez VM, Greeter JS, Mittal R, Hedrick TL
. 2013 Hawkmoth flying stability in turbulent vortex streets. J. Exp. Biol. 216 , 4567–4579. (doi:ten.1242/jeb.089672) Crossref, PubMed, ISI, Google Scholar
19
Ortega-Jimenez VM, Sapir N, Wolf M, Variano EA, Dudley R
. 2014 Into turbulent air: size-dependent effects of von Kármán vortex streets on hummingbird flight kinematics and energetics. Proc. R. Soc. B 281 , 20140180. (doi:10.1098/rspb.2014.0180) Link, ISI, Google Scholar
xx
Selig MS, Deters RW, Williamson GA
. 2011Wind tunnel testing airfoils at low Reynolds numbers. In 49th AIAA Aerospace Sciences Meeting, Orlando, FL, four–7 January, newspaper 875. Reston, VA: AIAA. Google Scholar
21
Lindgren B
. 2002Flow facility pattern and experimental studies of wall-divisional turbulent shear-flows. PhD dissertation. Regal Found of Technology, Stockholm, Sweden. Google Scholar
22
Johansson A
. 1992A low speed wind-tunnel with extreme flow quality: pattern and tests. In Proc. 18th ICAS Congress, Beijing, China, 20–25 September, vol. two, pp. 1603–1611. ICAS. Google Scholar
23
Hunt LE, Downs RS, Kuester MS, White EB
. 2010Flow quality measurements in the Klebanoff–Saric Wind Tunnel. In 27th AIAA Aerodynamic Measurement Engineering and Ground Testing Conference, Chicago, IL, 28 June–1 July, newspaper 4538. Reston, VA: AIAA Google Scholar
24
Saric WS, Reshotko E
. 1998Review of flow quality issues in air current tunnel testing. In 20th AIAA Advanced Measurement and Footing Testing Technology Conference, Albuquerque, NM, 15–18 June, paper 2613. Reston, VA: AIAA. Google Scholar
25
Ben-Gida H, Kirchhefer A, Taylor ZJ, Bezner-Kerr Due west, Guglielmo CG, Kopp GA, Gurka R
. 2013 Estimation of unsteady aerodynamics in the wake of a freely flying European starling (Sturnus vulgaris). PLoS Ane 8 , e80086. (doi:x.1371/periodical.pone.0080086) Crossref, PubMed, ISI, Google Scholar
26
Gerson AR, Guglielmo CG
. 2011 Flight at low ambient humidity increases poly peptide catabolism in migratory birds. Scientific discipline 333 , 1434–1436. (doi:10.1126/scientific discipline.1210449) Crossref, PubMed, ISI, Google Scholar
27
Pennycuick CJ, Alerstam Th, Hedenström AN
. 1997 A new depression-turbulence wind tunnel for bird flight experiments at Lund Academy, Sweden. J. Exp. Biol. 200 , 1441–1449. Crossref, PubMed, ISI, Google Scholar
28
Spedding GR, Hedenström A, Johansson LC
. 2009 A note on wind-tunnel turbulence measurements with DPIV. Exp. Fluids 46 , 527–537. (doi:10.1007/s00348-008-0578-one) Crossref, ISI, Google Scholar
29
Engel Due south, Bowlin MS, Hedenström A
. 2010 The part of current of air-tunnel studies in integrative research on migration biological science. Integr. Comp. Biol. 50 , 323–335. (doi:ten.1093/icb/icq063) Crossref, PubMed, ISI, Google Scholar
30
Spedding GR, McArthur J, Rosen M
. 2006Deducing aerodynamic mechanisms from near- and far-wake measurements of fixed and flapping wings at moderate Reynolds number. In 44th AIAA Aerospace Sciences Meeting and Exhibit, Reno, NV, ix–16 January, newspaper 7783. Reston, VA: AIAA. Google Scholar
31
Barlow JB, Rae WH, Pope A
. 1999 Low-speed current of air tunnel testing . Canada: John Wiley & Sons. Google Scholar
32
Makita H, Sassa Thousand
. 1991 Active turbulence generation in a laboratory wind tunnel. In Advances in turbulence three (eds AV Johansson, PH Alfredsson), pp. 497–505. Berlin, Germany: Springer. Crossref, Google Scholar
33
Cekli HE, van de Water West
. 2010 Tailoring turbulence with an agile grid. Exp. Fluids 49 , 409–416. (doi:x.1007/s00348-009-0812-v) Crossref, ISI, Google Scholar
34
Cekli HE, Tipton C, van de Water W
. 2010 Resonant enhancement of turbulent energy dissipation. Phys. Rev. Lett. 105 , 044503. (doi:10.1103/PhysRevLett.105.044503) Crossref, PubMed, ISI, Google Scholar
35
Kuttruff H
. 2007 Acoustics: an introduction . Eq. nine.7. Boca Raton, FL: CRC Press. Google Scholar
36
Davidson PA
. 2015 Turbulence: an introduction for scientists and engineers . Oxford, UK: Oxford University Printing. Crossref, Google Scholar
37
Prohasky D, Watkins S
. 2014Depression cost hot-element anemometry verses the TFI Cobra. In Proc. 19th Australasian Fluid Mechanics Conf., Melbourne, Australia, 8–11 December. Victoria, Australia: Australasian Fluid Mechanics Society. Google Scholar
38
Quinn DB, Watts A, Nagle T, Lentink D
. 2017 A new low-turbulence wind tunnel for fauna and pocket-size vehicle flight experiments. Dryad Digital Repository. (http://dx.doi.org/ten.5061/damsel.b65fj) Google Scholar

Source: https://royalsocietypublishing.org/doi/10.1098/rsos.160960

Posted by: turpinbaxt1992.blogspot.com