Document
The Multirotor Test Bed – A New NASA Test Capability for Advanced VTOL
Rotorcraft Configurations
Carl Russell and Sarah Conley Aeromechanics Office NASA Ames Research Center Moffett Field, CA ABSTRACT In November 2019, NASA completed the first wind tunnel test entry of the Multirotor Test Bed (MTB), a new test capability for advanced VTOL rotorcraft configurations. The MTB had been under development since 2017 when the need arose for an easily reconfigurable test stand for multirotor aircraft configurations. With the wide-ranging assortment of aircraft currently targeted at Urban Air Mobility and Unmanned Aircraft System applications, there is a need for validation data that will increase confidence in the computational modeling tools being used to develop these platforms. The MTB fills this need. This paper describes the key features of the MTB as well as its first wind tunnel test entry. A selection of results from the test is presented here, demonstrating the flexible configuration of the MTB and the types of data researchers can generate using this new test capability.
Because of the highly interactional nature of the airflow in NOMENCLATURE multirotor systems, both steady and dynamic loads are of interest. In addition, the collection of acoustic data, where ESC Electronic Speed Controller possible, is desired. The data collected will be used to validate MCS MTB Control System simulations of multirotor systems and eventually lead to MTB Multirotor Test Bed better predictions of multirotor performance.
MUAS Multirotor Unmanned Aircraft System PWM Pulse-Width Modulation Two previous wind tunnel tests of multirotor UAS vehicles TTL Transistor-Transistor Logic were conducted in October-December 2015 [Ref. 1] and January-February 2017 [Refs. 2 and 3] – referred to as the A Rotor disk area, ft MUAS1 and MUAS2 tests, respectively. The MUAS tests C Power Coefficient, C = P /ρ A (Ω R ) P P measured the aerodynamic performance of five quadcopters C Thrust Coefficient, C = T /ρ A (Ω R ) T T (3DR SOLO, 3DR Iris+, DJI Phantom 3 Advanced, SUI dy Lateral rotor spacing, ft Endurance, and the ARL Overlapped Quadrotor), a tilt-wing dz Vertical rotor spacing, ft (Elytron 4S UAV), and an octocopter (Drone America x8).
P Rotor power, hp The MUAS1 test entry generated a high-quality set of q Dynamic pressure, lb/ft performance data for these vehicles and also raised additional R Rotor radius, ft questions, particularly related to vibrations, blade deflections, ρ Air density, slug/ft aerodynamic interference, acoustics, and trim strategies. The Ω Rotor rotational speed, rad/s MUAS2 test expanded on the first by attempting to better characterize vibrations, interactional aerodynamics, and blade INTRODUCTION motion.
The Multirotor Test Bed (MTB) is a new test capability The MUAS tests had two main limitations: recently developed by the Aeromechanics Office at NASA
-
Ames Research Center with primary funding from the Loads were only measured for the entire vehicle, so a full Revolutionary Vertical Lift Technology (RVLT) Project. The picture of interactional aerodynamics could not be gleaned purpose of the MTB is wind tunnel and hover testing of from the data arbitrary multirotor aircraft configurations with a focus on
-
Testing was limited to existing vehicles with no ability to individual rotor loads data.
alter the configuration ________________________ The MTB program built upon the knowledge gained during th Presented at the Vertical Flight Society’s 76 Annual Forum the MUAS tests. By measuring individual rotor loads for a & Technology Display, Virginia Beach, VA, October 6-8, multirotor system and allowing for adjustments to rotor 2020. This is a work of the U.S. Government and is not position and attitude, the MTB will provide a wealth of data subject to copyright protection in the U.S.
on the aeromechanics of arbitrary multirotor configurations.
Indeed, data collected during the first wind tunnel test entry Table 1. MTB Capabilities and Limits are already being used to validate CFD predictions [Ref. 4].
This test capability will also be available for future tests of Characteristic Range new multirotor aircraft concepts. The modularity of the MTB Maximum rotor 24.5 in (nominal size of the KDE- hardware was demonstrated in [Ref. 5], which used a large size CF245-DP) portion of the test rig hardware for a propeller test Maximum rotor 4,500 (for 24.5 in rotor) immediately following the MTB tunnel entry. RPM Lateral spacing 24.7 – 38.7 in, adjustable in 1 in increments (2 in if keeping HARDWARE DESCRIPTION symmetry about centerline) An overview drawing of the MTB is shown in Figure 1, and * Longitudinal 25.5 – 72 in, adjustable in 1.5 in Table 1 lists its capabilities. Figure 2 shows the MTB installed spacing increments in the U.S. Army 7- by 10-ft Wind Tunnel at NASA Ames Vertical position 9 in of travel, adjustable in 1 in Research Center. Figure 3 shows a close-up photo of the rotor increments assembly hardware with the main components labeled. Figure Individual rotor tilt 90 deg forward to 5 deg aft, 4 shows some of the key adjustable dimensions of the test rig.
adjustable in arbitrary increments The full description of the MTB design is given in Ref. 6, but Full MTB pitch 20 deg nose down to 10 deg up, a short description is given here. The MTB consists of a adjustable in arbitrary increments strongback made of two 81-in long steel plates, between Maximum wind 40 ft/s, for initial testing; more if which are sandwiched six support blocks. The strongback is speed vibratory loads are low enough intended to support up to eight rotors, but for the first test * Minimum when 24.5-in rotors are installed; less if rotor entry, the maximum number of rotors was six. The strongback diameter is smaller is supported by a single strut, and the pitch of the entire MTB is controlled via a pitch link that attaches to a jackscrew- actuated movable lug on the strut. The details of the pitch and tilt actuation are explained in the Control Systems section.
Up to four lateral support beams are attached to the strongback to support the rotors. These beams can be moved in the longitudinal direction to any of the hole locations on the strongback (provided the rotors don’t interfere with each other). The lateral support beams have an “L-bracket” at each end, each consisting of a single machined piece of steel that slides over the lateral support beam to provide for lateral adjustment. A vertical support beam slides into the vertical leg of the L-bracket to provide for vertical adjustment of the rotor placement. A tilt mechanism using a linear actuator Figure 2. MTB installed in the U.S. Army 7- by 10-ft allows for adjustment of the rotor tilt between 90 degrees Wind Tunnel forward (airplane mode) and up to 5 degrees aft. The load cell, motor, and rotor are attached to the top of the tilt mechanism.
Figure 1. Overview of the Multirotor Test Bed Figure 3. MTB rotor assembly detail Figure 4. Different views of the MTB showing key adjustable dimensions devices. The system is, however, fairly limited in bandwidth CONTROL SYSTEMS compared with other serial protocols, such as USB and The MTB allows independent remote control of all six rotors, RS-232. For practical purposes on the MTB, the maximum both in rotor speed and in rotor tilt. Additionally, the pitch of baud rate was 10,000 bits per second, which limited the the entire MTB can be adjusted by the model operator from refresh rate of the control inputs to approximately 2/sec. In the wind tunnel control room. The following sections describe the six-rotor configuration, this update rate was sufficient, but the systems used for control of these different variables.
may not be satisfactory if more rotors are added to the MTB.
MTB Control System Rotor Speed The MTB is controlled by a LabView program called the The rotor speed on the MTB is controlled by off-the-shelf MTB Control System (MCS). The MCS controls and KDE electronic speed controllers (ESCs). As with other monitors all electronic functions of the MTB, including main commercial ESCs, those used for the MTB receive a pulse- bus power and temperature, rotor speed, rotor tilt, and MTB width-modulated signal, similar to that used for a radio- pitch. Various peripherals, described in subsequent sections controlled aircraft servo. A Pololu Mini Maestro 12-channel here, are used to drive the different actuators.
servo controller provides the PWM signals to the six individual speed controllers. The Mini Maestro is a low-cost The MCS communicates through a National Instruments data device that was previously used for the successful Multirotor acquisition board with the different peripheral systems over UAS tests conducted in the 7- by 10-ft Wind Tunnel in 2015 serial communication lines. The pitch, tilt, and rotor speed and 2017 [Ref. 1]. It communicates with the MCS over serial controllers are all made by Pololu Robotics & Electronics.
lines using the methods described in the previous section.
These different Pololu systems have the ability to send and This setup requires very little cabling between the MCS and receive messages over two TTL serial lines using a command the model, making the system fairly portable. This portability protocol called simply the “Pololu protocol.” The complete will be an advantage if there is a desire in the future to use the details of the Pololu protocol are beyond the scope of this MTB in a different wind tunnel facility where the cable run paper, but the basic functionality of the protocol is that a set lengths may be different between the test section and control of command bytes are encoded along with a device identifier, room.
which allows multiple devices to listen and receive commands on the same line. The Pololu devices are also able The motor speed on the MTB is measured using an infrared to communicate back to the MCS via a transmit line.
optical sensor that provides a 1/rev pulse train. This signal is captured by the MCS and processed into rotor RPM by From a cabling perspective, this system is very lightweight, measuring the time between pulses. In practice, these RPM as it only requires a single three-conductor instrumentation measurements were found to be accurate to within cable to simultaneously communicate with many different approximately 2 to 3 rpm, representing a maximum RPM angle. This angle is displayed on the operator console of the measurement error of 0.2% at the minimum rotor speed of MCS, but there is no closed-loop control. The operator enters 1,500 rpm. a step count that roughly corresponds to the desired pitch angle and then subsequently fine-tunes the angle. This A simple feedback controller was built using LabView’s procedure is a bit labor intensive, but in practice, there aren’t built-in PID control function. The controller was tuned many pitch angle changes during any given data run, so the initially with the Ziegler-Nichols method, but this led to an process did not significantly hinder testing. Future overly-sensitive RPM response. The controller gains were modifications to the MTB may include closed-loop control of then refined through trial and error until satisfactory the pitch angle.
performance was achieved. The controller as currently implemented can maintain rotor speed to within Power Systems approximately 10 rpm.
There are two main power systems for the MTB. The primary power for the rotors is provided by a 5 kW DC power supply Rotor Tilt that generates a maximum current of 100 A at up to 50 V. For The tilt angles of the individual rotors on the MTB are this test, the main rotor power was supplied at 48 V. The controlled by electric linear actuators. These actuators use a power limit of 5 kW is dictated by the available wall power in potentiometer to provide analog feedback proportional to the both the NASA model prep facility and in the wind tunnel.
actuator position. A Pololu JRK G2 Motor Controller is connected to each actuator. Like the RPM control system, The output of the primary power supply is controlled by the communication from the MCS to the motor controller is MCS via an RS-232 serial line. The current limit and voltage implemented via the Pololu protocol over a TTL serial line. can be remotely set by the model operator from the MCS, and The potentiometer feedback is constantly queried by the MCS the output of the power supply can also be remotely switched and reported to the model operator. on and off. In addition to the manual cutoff at the model operator station, the output of the power supply is interlocked There is no direct measurement of rotor tilt on the MTB. to the wind tunnel test section doors, removing the possibility Instead, the linear actuator position for each rotor has been of powering the model with a person in the test section.
mapped to an angle using a precision hand-held inclinometer.
This angle mapping has been programmed into the MCS so Power is distributed to the rotors through a power bus the operator can command a tilt angle, and the actuator will mounted on the MTB strongback. The power bus has move to the corresponding extension length. It is also worth connections for 8 rotors, but for this test, only 6 of the power noting that the rotor tilt angle does not require the linear bus outputs were used. The bus contains two large diodes to actuators to be powered in order to hold position. The gearing protect the power supply from voltage spikes. Since these of the actuators is configured such that they cannot be back- diodes generate a significant amount of heat, active cooling is driven. The MTB pitch angle, described in the next section, provided by fans in the power bus case. The diode has this same characteristic, which is a useful safety feature temperatures are measured by two thermocouples and are that limits the potential consequences of any actuator power constantly monitored in the MCS.
failure.
The second power system is a 12 V DC power supply, which MTB Pitch provides power to the pitch and tilt actuators as well as to the Pololu control hardware and the pitch-sensing inclinometer.
As described in the MTB hardware section, the pitch of the All of the 12 V hardware is relatively low current and does entire MTB is actuated by a jackscrew connected to a pitch not have much potential for causing damage in the event of a link on the mounting strut. The jackscrew is turned by a small malfunction, so no remote control or interlock is implemented stepper motor that transmits power though an attached 9:1 for this hardware. There is an emergency stop button at the reduction gearbox. The resulting pitch actuation is very slow operator console, so power can be immediately cut to all of (a few seconds per degree), but allows for control of the pitch the 12 V hardware in the event of an emergency or mishap.
angle with only a 10-watt stepper motor.
Measurement Systems A Pololu Tic T825 Stepper Motor Controller provides both the drive power and the control function for the pitch stepper The primary measurements of interest on the MTB are the motor. Similar to the RPM controller and the tilt controller, individual rotor loads. Each rotor sits on top of a 6-axis load the Tic receives Pololu protocol commands over a TTL serial cell capable of measuring both steady and dynamic loads. The line and transmits current status back to the MCS. load cells used on the MTB are Interface model 6AR70A- S11, with a capacity of 75 lb in all three force directions and An analog inclinometer is secured to the upper surface of the 115 in-lb in all three moment directions.
MTB strongback to provide a direct measurement of the pitch The Interface load cells are paired with Interface Model BX8- MTB project was to measure the effect of relative vertical HD44 amplifiers that combine the signals from the strain placement of the forward and aft rotors in a quadrotor gauges in the load cell assembly and resolves them into six configuration. Previous studies have shown that there is an analog signals proportional to each of the six primary forces optimal vertical separation of the forward and aft rotors, and and moments. The amplifier also provides automatically this test sought to support that conclusion with experimental configured low-pass filtering for a maximum sample rate of data [Ref. 8]. In broad strokes, the main research goal of the 12,000/s. For this test, the maximum frequency of interest MTB is to provide experimental data to help validate corresponded to 12/rev at 4,500 rpm, or 900 Hz. The amplifier computational performance predictions for multirotor filter was set to a cutoff frequency of 2 kHz. The resolved aircraft.
analog signals are sent to the wind tunnel data system, which, for this test, was set to a sample rate of 4,000 samples per Test Matrix second.
The primary purpose of the first wind tunnel test entry was demonstrating the capabilities of the MTB. In particular there In addition to the load cell measurements, overall loads on the was a desire to show that the test rig could be rapidly MTB were measured using the wind tunnel’s external scales.
reconfigured while retaining all of its measurement Unlike the load cells, the scales are only capable of measuring capabilities. The test matrix was a reflection of the desire to steady loads. The overall MTB loads were not particularly of exercise as many of the configuration capabilities as possible interest for this test, because the only components being tested within the test window. It is expected that future tests would were the rotors. If a wing or fuselage is added to the MTB in have a narrower focus and would more specifically target the future, the tunnel scales could measure the additional individual parameter sweeps, such as rotor spacing. The test loads on these static components. For this test, the scales matrix is summarized in Table 2. For all test conditions, both merely provided validation of the load cell data.
weight and aerodynamic tare data were collected and applied during post-processing. Note that not all possible The force and moment measurements are the only scientific combinations shown in Table 2 were tested. Generally, the data reported in this paper. Acoustic data were also recorded parameter sweeps included the following values: using a phased microphone array system similar to that
-
Pitch, defined positive nose up: -10, -5, -2, 0, 2, 5 deg described in Ref. 7, and those data are still being processed.
Other data recorded by the data system included the model
-
Tilt, defined positive nose down relative to the strongback, state (motor RPM, rotor tilt, MTB pitch) and wind tunnel with 0 deg representing the rotor plane parallel to the conditions (such as speed, density, temperature, etc.).
strongback (helicopter mode): -5, -2, 0, 2, 5, 10, 30, 45, 60, 90 deg TEST DESCRIPTION
-
Yaw, defined positive nose right: -10, -5, -2, 0, 5, 10 deg
-
Rotor speed: 1,500 to 3,500 rpm, in increments of 500 Test Objectives The test performed in late 2019 was the first wind tunnel entry The full test matrix and results from this wind tunnel test will for the Multirotor Test Bed. As such, a main focus of the test be documented in a NASA Technical Memo at a later date.
was ensuring that the MTB could successfully perform all of its designed functions while installed in the 7- by 10-ft Wind Because there are so many ways of positioning the rotors, a Tunnel. The primary systems checked out and used for this generalized numbering convention was used to describe the test entry were: rotor positions. This numbering convention assumes bilateral symmetry across the MTB centerline. For a more general
-
MTB control systems, including individual rotor tilt, full description of the rotor positioning that does not require pitch of the MTB, and rotor RPM control symmetry, a different numbering scheme would have to be
-
MTB hardware, including the strongback assembly, adopted. The rotor positioning is given by three ordered vertical and lateral adjusting beams, and pitch and tilt triples: mechanisms
-
𝑃 𝑃 𝑃 Force measurement systems, primarily consisting of the ଵ௫ ଶ௫ ଷ௫ six-axis load cells located at the base of each of the MTB 𝑃 𝑃 𝑃 ቐ ቑ ଵ௬ ଶ௬ ଷ௬ motors.
𝑃 𝑃 𝑃 ଵ௭ ଶ௭ ଷ௭ The primary scientific objective of the MTB project was the The first subscript indicates the row of rotors – the numbers measurement of forces and moments experienced by the 1, 2, and 3 correspond to the forward, middle, and aft rotors individual rotors of a multirotor system. In so doing, the respectively. The second subscript refers to the positioning highly interactional aerodynamics of multirotor systems were direction, with x , y , and z referring to the longitudinal, lateral, explored. One specific type of test that was targeted by the and vertical positions, respectively. The positions are given in and potential data quality issues due to wall effects on the terms of bolt-hole locations which are described here: rotor inflow.
-
Longitudinal, or x -position: numbered from 0 to 48. 0 is While there was not a narrowly targeted engineering question the forward-most position, while 48 is the aft-most addressed by this test, several parameters were systematically position. The longitudinal positioning bolt holes are investigated including the following: spaced at 1.5-in increments, with a maximum spacing of
-
72 in. The effects of the presence of multiple rotors, by investigating single-rotor performance, plus two-, four-,
-
Lateral, or y -position: numbered from 0 to 7. 0 is the inner- and six-rotor configurations under the same tunnel most position, while 7 is the outer-most position. The conditions and rotor settings.
lateral positioning bolt holes are spaced at 1.0 in
-
increments, with a maximum spacing of 38.7 in. The effects of lateral spacing on two side-by-side non- overlapping rotors. As an aside, the effects of lateral
-
Vertical, or z -position: numbered from 0 to 9. 0 is the spacing on side-by-side overlapping rotors are currently lowest position, while 9 is the tallest position. The vertical being investigated in another NASA test program.
positioning bolt holes are spaced at 1.0 in increments. For
-
practical purposes, 7 is the highest position that can be The effects of fore-aft vertical rotor spacing on a quadrotor used for the 7- by 10-ft Wind Tunnel installation. Positions configuration.
8 and 9 would place the rotors too close to the ceiling at
-
The effects of RPM and tilt transients.
certain rotor tilt angles, presenting both a safety concern Table 2. MTB Wind Tunnel Test Matrix Nominal Pitch Tilt Yaw Rotor Rotors Installed Rotor Wind Angles Angles Angles Speeds Positions Speeds (deg) (deg) (deg) (rpm) 1 2 3 4 5 6 (ft/s) -10 to 0 24 48 X X X X X X -10 to +10 0 1,500 to 3,000 20, 40 +10 ൝ 7 7 7 ൡ X X X X X X -5, 0 -5 to +90 0 2,000 20 7 7 7 X X X X X X -10 to +5 0 0 1,500 to 3,000 20, 40 X X X X -10 to +5 0 0 1,500 to 3,000 20 X X X X 0 -5 to +90 0 2,000 20 0 24 48 X X -10 to +5 0 0 1,500 to 3,500 20, 40 ൝ ൡ 7 7 7 X -10 to +5 0 0 1,500 to 3,500 20, 40 0 0 0 X 0 -5 to +90 0 2000 20 X -10 to +5 0 0 1,500 to 3,500 20, 40 X -2, 0 0 0 2,000, 3,000 20 0 24 48 X X X X -10 to +5 0 0 1,500 to 3,500 20, 40 ൝ 7 7 7 ൡ X X X X 0 -5 to +90 0 2,000 20, 40 0 7 0 X X X X -10 to +5 0 0 1,500 to 3,500 20, 40 0 24 48 X X X X 0 -5 to +90 0 2,000 to 4,000 20, 40 ൝ 4 7 7 ൡ 0 7 0 X X -2, 0 0 0 1,500 to 3,500 20, 40 X X -10 to +5 0 0 1,500 to 3,500 20, 40 0 24 48 X X 0 -5 to +90 0 2,000 to 4,000 20, 40 ൝ 1 7 7 ൡ * * 0 7 0 X -2, 0 0 to +45 0 1,500 to 2,500 20, 40 0 24 48 X X X X X X -10 to +2 0 0 1,500 to 3,500 20, 40 ൝ ൡ 1 7 1 *,† X X X X X X 0 -5 to +90 0 2,000 20, 40 0 0 0 * Sweeps included transients † Sweeps included tilting just Rotors 3 and 4 with the others remaining at 0 deg Figures 6-9 show the thrust and power on Rotor 3 for a RESULTS dynamic pressure of 0.48 lb/ft and rotor speed of 2,000 rpm.
The following section presents a small subset of the data that Recall from Fig. 5 that Rotor 3 is the middle, right-hand rotor were collected during the first test entry of the Multirotor Test when looking at the MTB from above. Also note that there Bed. The complete data set will be published at a later date in was no effort made to achieve any kind of trim, because the a NASA Technical Memo.
MTB is not representing an aircraft in this test case. The thrust and power therefore represent the measurements that result Throughout this section, the rotors are referred to by their from holding the desired rotor speed and attitude.
rotor number, from 1 to 6. Figure 5 shows a top-down view of the MTB with the rotors identified by number. It also The thrust is shown in Fig. 6 in dimensional units and in indicates the rotation direction of each rotor.
coefficient form in Fig. 7. The differences between the isolated rotor and side-by-side cases are fairly minor, with the side-by-side case resulting in a slightly higher thrust on Rotor 3. Once the front rotors (Rotors 1 and 2) are added, there is a significant drop-off in the thrust produced by Rotor 3, as it becomes impacted by the wake of the front two rotors. Adding 5.00 4.00 3.00 Figure 5. MTB rotor numbering and rotation directions Data Caveats 2.00 There are a couple of caveats on the data presented here, and Rotor 3 only values in the plots should not be interpreted as final results, 1.00 Rotors 3 and 4 but rather as representative of the type of data that can be Rotors 1-4 Rotors 1-6 generated with the MTB. These deficiencies will be corrected 0.00 before the complete results are published in the final data -10 -5 0 5 report.
MTB Pitch, deg Figure 6. Thrust for Rotor 3 with different numbers of The loads data presented here are based on a calibration that rotors present at varying angle of attack. RPM = 2,000, was provided by the load cell manufacturer. During and q = 0.48 lb/ft .
instrumentation checkouts, discrepancies were found between known check loads and the data obtained from the load cells.
Troubleshooting the instrumentation led to the conclusion that 0.015 the calibration matrix was inaccurate, but there was insufficient time to re-calibrate prior to the wind tunnel entry.
A post-test calibration of the six-axis load cells and re- reduction of the data are planned, but these activities have 0.010 been delayed due to COVID-19-related facility access restrictions.
The other main caveat on the data is that the wall effects of 0.005 the wind tunnel are still being investigated. Ref. 4 has shown Rotor 3 only that the impacts of the walls are non-negligible for the MTB.
Rotors 3 and 4 Further investigation of these impacts is warranted, and will Rotors 1-4 likely require corrections similar to those given in Ref. 9. Rotors 1-6 0.000 -10 -5 0 5 Effects of Multiple Rotors MTB Pitch, deg The main strength of the Multirotor Test Bed is its ability to Figure 7. Thrust coefficient for Rotor 3 with different generate independent loads for each of the rotors. The data numbers of rotors present at varying angle of attack.
presented in this section demonstrate the different results RPM = 2,000, and q = 0.48 lb/ft .
obtained with 1, 2, 4, or 6 rotors installed on the MTB.
the aft two rotors (Rotors 5 and 6) apparently decreases this coefficient calculation. The remaining mean loads data effect at nose-down pitch angles, with the thrust nearly presented in this section are expressed only in coefficient matching the isolated rotor and side-by-side cases. At more form.
horizontal or positive pitch angles, the results approach those seen in the 4-rotor case. The next four plots (Figs. 10 – 13) show the variation in measured thrust and power between the different rotors Results for the rotor power are shown in Figs. 8 and 9 in when all six rotors are installed. Figures 10 and 11 show the dimensional and non-dimensional units, respectively. For variation in thrust and power coefficient with all six rotors power, the isolated rotor and side-by-side cases match each operating at 2,000 rpm and a wind tunnel dynamic pressure other almost identically. Once the forward rotors are added, of 0.48 lb/ft . In general, it can be observed that the front the power required to maintain 2,000 rpm on Rotor 3 rotors draw slightly less power and produce slightly more increases. Except at the highest angle of attack, the measured thrust than the middle and aft rotors. This effect is likely due power on Rotor 3 is nearly identical between the 4- and 6- to the middle and aft rotors operating in the wake of the rotor cases. The power measured in horsepower is noticeably forward rotors. This result is consistent with the trends lower for the 4-rotor case, but this discrepancy disappears observed in Figs. 7 and 9.
once the effects of density are removed by the power 0.15 0.015 0.10 0.010 Rotor 1 0.05 0.005 Rotor 2 Rotor 3 only Rotor 3 Rotors 3 and 4 Rotor 4 Rotor 5 Rotors 1-4 Rotor 6 Rotors 1-6 0.000 0.00 -10 -5 0 5 -10 -5 0 5 MTB Pitch, deg MTB Pitch, deg Figure 10. Thrust coefficient for each rotor with six Figure 8. Power for Rotor 3 with different numbers of rotors present at varying angle of attack. RPM = 2,000, rotors present at varying angle of attack. RPM = 2,000, 2 2 and q = 0.48 lb/ft .
and q = 0.48 lb/ft .
-3 -3 1.000 1.000 0.800 0.800 0.600 0.600 0.400 0.400 Rotor 1 Rotor 2 Rotor 3 Rotor 3 only 0.200 Rotors 3 and 4 0.200 Rotor 4 Rotors 1-4 Rotor 5 Rotors 1-6 Rotor 6 0.000 0.000 -10 -5 0 5 -10 -5 0 5 MTB Pitch, deg MTB Pitch, deg Figure 11. Power coefficient for each rotor with six Figure 9. Power coefficient for Rotor 3 with different rotors present at varying angle of attack. RPM = 2,000, numbers of rotors present at varying angle of attack.
and q = 0.48 lb/ft .
RPM = 2,000, and q = 0.48 lb/ft .
One curious observation in Figs. 10 and 11 is the poor between the power coefficient for Rotors 3 and 4, similar to agreement in loads between Rotors 3 and 4 as well as between the q = 0.48 lb/ft case. The thrust coefficient, on the other Rotors 5 and 6. Because the MTB was operating in a hand, is well matched between the left and right rotors at each symmetric configuration in this case, one would expect the longitudinal station for the q = 1.9 lb/ft case. The more left and right rotors at the same longitudinal station to exhibit pronounced differences in thrust values for the higher speed nearly identical results. This discrepancy is being investigated case is almost certainly due to the higher wake skew angle and may be related to the data caveats discussed previously. that would be expected at higher speed, leading to a stronger interference effect between longitudinally spaced rotors.
Figures 12 and 13 show the variation in thrust and power coefficient with all six rotors operating at 2,000 rpm and a Effects of Rotor Spacing wind tunnel dynamic pressure of 1.9 lb/ft . In this case, the The next sets of results demonstrate the types of studies that differences between the forward, middle, and aft rotors are can be performed using the MTB to reconfigure the rotor-to- more readily apparent. As seen in the lower-speed case, the rotor spacing. The first set of charts shows the effects of forward rotors generate more thrust and draw less power than lateral spacing on rotor thrust and power, while the second set the middle and aft rotors. There is again a discrepancy shows the effects of vertical spacing.
Figures 14 and 15 show the thrust and power coefficients 0.015 measured for the right-hand rotor at three different values of lateral spacing. The results show data for either Rotor 1 or 3 because the different 2-rotor configurations weren’t all run with the same side-by-side pair). The positions of the rotors 0.010 are 1 (inner-most), 4 (middle), and 7 (outer-most). These three positions correspond to a lateral spacing of 26.7, 32.7, and 38.7 in, or dy / R = 2.18, 2.67, and 3.16, respectively. Hole- Rotor 1 position 1 is actually not the inner-most location that a rotor 0.005 Rotor 2 on the MTB can be located, but with a rotor diameter of 24.5 Rotor 3 in, mounting the rotors at hole-position 0 would leave just 0.2 Rotor 4 Rotor 5 in between blade tips. This small clearance was judged to be Rotor 6 too close for safe operation, but with a smaller-diameter rotor, 0.000 this inner bolt-hole location could be safely used.
-10 -5 0 5 MTB Pitch, deg Adding to the comparison between the single rotor and the Figure 12. Thrust coefficient for each rotor with six two side-by-side rotors shown earlier in Fig. 7 and 9, Figs. 14 rotors present at varying angle of attack. RPM = 2,000, and 15 show that moving the two rotors closer together results and q = 1.90 lb/ft .
in more thrust being produced at lower power for the same wind and rotor speeds. The MTB does not have mechanical -3 rotor synchronization, so investigating intermeshing side-by- 1.000 side rotors is not feasible; however, another NASA experiment on a different test stand will investigate the 0.800 performance of intermeshed side-by-side rotors later in 2020.
The final results presented here for mean rotor loads show the 0.600 effects of adjusting the vertical separation between forward and aft rotors on a quadrotor configuration. Figures 16 and 17 0.400 Rotor 1 show power and thrust for the forward and aft right-hand Rotor 2 rotors (Rotors 1 and 3, respectively). The two configurations Rotor 3 shown are with all four rotors at the lowest possible location 0.200 Rotor 4 ( dz / R = 0) and with the aft two rotors at the highest possible Rotor 5 Rotor 6 location ( dz / R = 0.57). In both configurations, the rotor speed 0.000 on all four rotors is 2,000 rpm, and the tunnel dynamic -10 -5 0 5 pressure is 0.48 lb/ft .
MTB Pitch, deg Figure 13. Power coefficient for each rotor with six For the case with all four rotors at the same height, there is a rotors present at varying angle of attack. RPM = 2,000, large separation between the results for the forward and aft and q = 1.90 lb/ft .
rotor for both C and C . Clearly, there is a large interference P T 0.015 0.015 0.010 0.010 0.005 0.005 Isolated Rotor Isolated Rotor (Rotor 3, pos 7) Rotor 1, dz/R = 0 Wide spacing (Rotor 3, pos 7) Rotor 1, dz/R = .57 Mid spacing (Rotor 1, pos 4) Rotor 3, dz/R = 0 Close spacing (Rotor 1, pos 1) Rotor 3, dz/R = .57 0.000 0.000 -10 -5 0 5 -10 -5 0 5 MTB Pitch, deg MTB Pitch, deg Figure 14. Thrust coefficient for the right-hand rotor Figure 16. Thrust coefficient for the right-hand rotors with two rotors present at different lateral spacing and with four rotors present at different vertical spacing and 2 2 varying angle of attack. RPM = 2,000, and q = 0.48 lb/ft . varying angle of attack. RPM = 2,000, and q = 0.48 lb/ft .
-3 -3 10 10 1.000 1.000 0.800 0.800 0.600 0.600 0.400 0.400 Isolated Rotor Rotor 1, dz/R = 0 Isolated Rotor (Rotor 3, pos 7) 0.200 Rotor 1, dz/R = .57 Wide spacing (Rotor 3, pos 7) 0.200 Rotor 3, dz/R = 0 Mid spacing (Rotor 1, pos 4) Rotor 3, dz/R = .57 Close spacing (Rotor 1, pos 1) 0.000 0.000 -10 -5 0 5 -10 -5 0 5 MTB Pitch, deg MTB Pitch, deg Figure 17. Power coefficient for the right-hand rotors Figure 15. Power coefficient for the right-hand rotor with four rotors present at different vertical spacing and with two rotors present at different lateral spacing and varying angle of attack. RPM = 2,000, and q = 0.48 lb/ft .
varying angle of attack. RPM = 2,000, and q = 0.48 lb/ft .
to investigate the effects of spacing and rotor placement on effect between the forward and aft rotors that increases the actual vehicle performance. These results are given to show power and decreases the thrust produced by the aft rotors.
the types of data and trade studies that could potentially be When the vertical separation is increased, the power required explored using the MTB.
by the aft rotor decreases, while its thrust increases. With the increased vertical separation, the aft rotors’ performance Dynamic Loads nearly matches that of the isolated rotor. This is a similar result to those observed in CFD predictions for the MTB The previous sections dealt with the results for mean loads, presented in Ref. 4, as well as in previous computational which are obtained by sampling the load cells at 4 kHz for 30 studies on other multirotor configurations, such as those seconds and averaging the data. This section shows sample presented in Refs. 10 and 11. At the same time, the forward results for dynamic loads data obtained from the six-axis load rotor becomes less efficient, producing less thrust at a higher cells. For every data point collected, both mean loads and a 5- power level than for the dz / R = 0 case.
second sample of the dynamic loads were collected for all 36 load cell channels. For the transient cases, the dynamic data This and the previous examples do not represent trimmed collection time window was increased as necessary.
aircraft configurations, so additional studies would be needed Figure 18 shows the vibratory loads in the x-direction for 1.4 Rotor 3 both in isolation and in a six-rotor configuration (the Isolated Rotor same test points as those shown in Figs. 6-11). The vibratory Six Rotors 1.2 loads are expressed at discrete harmonics of the rotor rotational speed, or N/rev. In the previous MUAS tests with 2-bladed rotors, described in Refs. 1 and 2, high vibratory 0.8 loads were observed, especially at 2/rev. A similar trend was observed with the MTB, which is not surprising given that its 0.6 rotors are also 2-bladed.
0.4 The isolated rotor case shows very high 2/rev loads, but the 0.2 loads at the remaining harmonics are nearly an order of magnitude less. In the 6-rotor case, the 2/rev loads are still 1 2 3 4 5 6 7 8 quite high, but the 4/rev loads are also significant, indicating N per Rev blade-wake interactions between the rotors. One area of planned future research on the MTB is an investigation of the Figure 18. Rotor 3 N/Rev vibratory loads in the x- effects of rotor azimuth phasing on the vibratory loads.
direction, in isolation and with all six rotors present.
Research has already shown that there are acoustic RPM = 2,000, and q = 0.48 lb/ft .
advantages to controlling rotor phasing [Ref. 12], and it is reasonable to expect that vibratory loads could be reduced using similar methods. Thus far, however, the MTB control Thrust (dynamic) system is not capable of controlling the rotor-to-rotor phasing.
Thrust (steady) The uncontrolled rotor phase tends to meander at a given rotor Rotor Speed speed, which leads to time-varying vibration magnitudes.
Another function the MTB is capable of is creating rotor speed transients, which can be investigated for their effects on rotor loads. Figure 19 shows the thrust response of an isolated 5 rotor undergoing a transient from 1,500 to 2,500 rpm at a 2 1500 dynamic pressure of 1.9 lb/ft and an MTB pitch angle of -2 deg. As shown, the vibratory thrust loads are quite high, with approximately the same magnitude as the mean load.
The red line shows a rolling average of the thrust, while the 0 5 10 15 20 Time, s yellow line, plotted against the secondary axis, shows the rotor RPM. One thing that can be observed in Fig. 19 is that Figure 19. Rotor 2 isolated rotor thrust response to an the thrust change appears to lag slightly behind the rotor speed RPM transient from 1,500 to 2,500 rpm. Pitch = -2 deg, change. This effect is not unexpected, as the rotor wake takes and q = 1.9 lb/ft .
a finite amount of time to respond to the change in rotor speed.
Understanding the rotor load and vehicle response to changes in RPM is an active area of research at NASA [Ref. 13], and this type of transient data could provide validation data for these efforts.
Finally, Fig. 20 shows a waterfall plot for the same transient thrust data shown in Fig. 19. In this image, the 2, 4, 6, and 8/rev harmonics are clearly visible, with 2/rev being the most prominent. The remaining harmonics are present, but are at a much lower magnitude.
SUMMARY AND CONCLUSIONS The Multirotor Test Bed is a new capability for multirotor testing developed at NASA Ames Research Center. The goal of the MTB project was to create a testbed that could be used Figure 20. Waterfall plot for Rotor 2 isolated rotor for studies of advanced VTOL rotorcraft configurations, thrust response to an RPM transient from 1,500 to 2,500 rpm. Pitch = -2 deg, and q = 1.9 lb/ft .
particularly those targeted at the Urban Air Mobility and ACKNOWLEDGEMENTS Unmanned Aircraft Systems markets. The reconfigurable The authors would like to acknowledge the contributions of geometry of the MTB allows researchers to conduct trade the many people who helped make the MTB development and studies and parametric investigations of multirotor aircraft testing a success. In particular, from NASA, Gina Willink, performance without creating multiple wind tunnel models.
Tom Norman, and Bill Warmbrodt provided invaluable guidance and assistance. The machine shop crews, led by This paper summarized the design and capabilities of the Robert Kornienko and Vincent Derilo, machined the MTB MTB and described a small subset of the data that were hardware and provided guidance and helpful suggestions gathered as part of the first wind tunnel test entry of the MTB during the design process. Deividas Čelkys and Tomas in the U.S. Army 7- by 10-ft Wind Tunnel. The results shown Narbuntas provided operational support throughout the entire give examples of the types of parametric studies that could be wind tunnel test. The first wind tunnel test of the MTB would performed using the MTB. Once the complete data set has not have been a success without the many hours put in by the been validated, it will be made publicly available for U.S. Army 7- by 10-ft Wind Tunnel crew, particularly Gary researchers to use. NASA has already begun to use the data Fayaud, Bruce Gesek, Steve Nance, Dan Pruyn, and Jiawei generated in the first MTB wind tunnel test to validate Toh. This work was primarily funded by NASA’s predictions of multirotor performance. Furthermore, the Revolutionary Vertical Lift Technology (RVLT) Project, with expectation is that the Multirotor Test Bed will become a key seed funding from the NASA Ames FY18 Internal Research experimental capability for generating a wealth of multirotor and Development Fund.
validation data in the future.
Tunnel.” VFS Aeromechanics for Advanced Vertical Flight REFERENCES Technical Meeting, San Jose, CA, January 21-23, 2020.
1. Russell, C., Willink, G., Theodore, C., Jung, J., and Glasner, B., "Wind Tunnel and Hover Performance Test 8. Johnson, W. and Silva, S., “Observations from Results for Multicopter UAS Vehicles", NASA/TM—2018- Exploration of VTOL Urban Air Mobility Designs,” 7th 219758, February, 2018.
Asian/Australian Rotorcraft Forum, Jeju Island, Korea, October 2018.
2. Russell, C., et al., “Multicopter UAS Performance Test 2” NASA/TM—TBD.
9. Langer, H.-J., Peterson, R., and Maier, T., “An Experimental Evaluation of Wind Tunnel Wall Correction 3. Chen, G., Nuñez, G., Russell, C., Avera, M., and Methods for Helicopter Performance,” American Helicopter Dotterweich, J., “Wind Tunnel Test Results for an Society 52nd Annual Forum, Washington, D.C., June 4-6, Overlapped Quadrotor Configured UAS,” AHS International 1996.
74th Annual Forum and Technology Display, Phoenix, AZ, May 14-17, 2018.
10. Healy, R., Misiorowski, M., and Gandhi, F., “A Systematic CFD-Based Examination of Rotor-Rotor 4. Conley, S., and Russell, C., “Comparing CFD Predictions Separation Effects on Interactional Aerodynamics for Large of the Multirotor Test Bed with Experimental Results,” eVTOL Aircraft,” Vertical Flight Society 75th Annual Abstract submitted to VFS 76th Annual Forum and Forum & Technology Display, Philadelphia, PA, May 13- Technology Display, Montréal, QC, Canada, May 19-21, 16, 2019.
2020.
11. Ventura Diaz, P. and Yoon, S., “Computational Study of 5. Battey, L. and Russell, C., “Wind-Tunnel Testing a Small NASA’s Quadrotor Urban Air Taxi Concept,” AIAA Isolated Folding Propeller,” Vertical Flight Society 76th SciTech Forum 2020, Orlando, FL, January 6-10, 2020.
Annual Forum & Technology Display, Virginia Beach, VA, Oct. 6-8, 2020.
12. Schiller, N., Pascioni, K., and Zawodny, N., “Tonal Noise Control using Rotor Phase Synchronization,” Vertical 6. Conley, S. and Russell, C., “Mechanical Design of the Flight Society 75th Annual Forum & Technology Display, Multirotor Test Bed,” Vertical Flight Society Philadelphia, PA, USA, May 13-16, 2019.
Aeromechanics for Advanced Vertical Flight Technical Meeting, San Jose, CA, January 21-23, 2020.
13. Malpica, C. and Withrow-Maser, S., “Handling Qualities Analysis of Blade Pitch and Rotor Speed Controlled eVTOL 7. Burnside, N.J. and Horne, W.C., “Development of Quadrotor Concepts for Urban Air Mobility,” International Instrumentation and Methods for Time-Domain Powered Lift Conference 2020, San Jose, CA, January 21- Measurement of Rotor-Type Sources in a Hard Wall Wind 23, 2020.