- 25 Feb 2021 01:23 PM
- 0
Controlling Interest
By Frank Colucci
The PAV-ER distributed electric propulsion testbed enables Embry-Riddle researchers to investigate safe control schemes for urban air mobility.
[About the image on the Top: The PAV-ER was built by the Eagle Flight Research Center to establish distributed electric propulsion control modes and help size future vehicles. (All images courtesy EFRC)]
Designers of electric vertical takeoff and landing (eVTOL) aircraft typically envision multi-propeller air taxis controlled by RPM or collective blade pitch. The Eagle Flight Research Center (EFRC) at Embry-Riddle Aeronautical University built a hovering multi-rotor testbed to try rotor speed, collective pitch and helicopter-style cyclic pitch in normal and emergency situations. The Personal Air Vehicle — Embry-Riddle (PAV-ER) has electric motor-and-rotor pods arrayed like those of a notional urban air mobility (UAM) platform. It hovers on battery power and simulates failures to reveal how its self-teaching control system best restores stability. EFRC director Dr. Pat Anderson (highlighted in last issue’s “Leadership Profile” Vertiflite, Jan/Feb 2021) explained, “The output would be a tool that would help the UAM people determine if they meet FARs [Federal Aviation Regulations] for losing an engine or propulsive pod — do you have the capability to keep flying — and knowing [the tool] was calibrated to real hardware.”
PAV-ER hardware and software mimic and counter servo failures, motor failures and other malfunctions, according to principal investigator Dr. Kyle Collins. “Part of our research plan is to look at failure modes and effects. Then we prioritize those based on the probability that they would occur using something like a fault tree.” Lessons learned may enable multicopter flight computers to compensate for failures with remaining effectors. “That’s the path we’re heading down,” said Collins. “It will be a fault-tolerant control system.”
The air-taxi-sized testbed can also help formulate scaling laws for bigger non-traditional eVTOL aircraft. Anderson explained, “One of the things we want to see as we fly these vehicles with fixed-pitch blades is can they go from the little models to the Avengers [movie] aircraft carrier-size with fixed pitch. We don’t think you can take a fixed-pitch blade and simply scale it up.” According to Anderson, as UAM designers increase rotor speed (RPM) to generate more control thrust, they also have to increase motor torque to overcome rotor blade inertia. “We’ve seen with 6-to-10-ft [1.8-to-3-m] rotors that to hold control power constant, the electric motor grows disproportionately — the thrust-to- weight ratio of the pod goes down.”
Variable-pitch, constant-speed rotors potentially avoid the trap. “What we’re trying to show is you’re not going to have an octocopter or quadcopter with [fixed-pitch] 40-ft [12.2-m] rotors, because the motor from a control standpoint is too heavy,” said Anderson. “If you use a constant-speed rotor, even if just with collective, that torque needed to overcome inertia goes away. You’d be close to a helicopter rotor with that goodness.” Anderson added, “I believe the real answer is cyclic. A lot of the UAM people don’t want to do to that because it’s ‘complicated.’ I think [cyclic’s] the right direction to get the control power in hover and forward flight the other systems don’t afford you. From the control standpoint, it makes a lot of things easy.”
In-House for Now
The Eagle Flight Research Center in Daytona Beach, Florida — an extension of Embry-Riddle’s College of Engineering — built the PAV-ER in-house. Anderson noted, “This is all, to date, financed out of our internal research and development budget.” The research center is looking for external funding and plans to evolve the testbed, notionally adding an aerodynamic shell, composite frame and hybrid electric propulsion. “This is what we typically do here,” explained Anderson. “I like to say it costs more money to fly higher than you’re willing to fall.” EFRC currently owns all data generated by the eVTOL testbed. “The data that comes from this would mostly leave Embry-Riddle as technical papers. If someone wants to come in and interface with us, we would entertain that.”
EFRC has long experimented with alternative aviation propulsion including work on the battery-powered e-Spirit of St. Louis and hybrid-electric Eco Eagle piloted airplanes. The twin-rotor/twin- motor Heurobotics Mk. 2 eVTOL unmanned aircraft system (UAS) — see “Lift Where You Need It,” Vertiflite, Nov/Dec 2016 — provided the electric motor-and-rotor pods now used on the PAV- ER testbed. Anderson said, “We wanted to leverage everything we learned in developing those pods to look at how those pods would be used, how many and in what configuration.”
Each propulsion pod generates about 55 lb (25 kg) of thrust and turns its rigid rotor head at up to 2,200 RPM. The PAV-ER uses eight pods, four forward and four aft, to model an air taxi that takes off and lands on rotors and cruises efficiently on tilting fore and aft wings. “There are a lot of configurations out there that are eight,” observed Anderson. “You can do more interesting failure cases with eight than four.” Collins added, “Part of the research we want to do is understand design changes like the number of rotors, thrust-to-weight ratios of the pods, thrust to weight of the entire vehicle, and fixed versus collective pitch.”
The PAV-ER structure is a simple frame, 100 inch (2.5 m) wide at fore and aft “wing” positions. “There’s communication, but no centralized high-power management,” said Anderson. The structure can position the propulsion pods to model thrust- or tractor-rotors. “We’ve done both rotors-up and rotors-down,” said Anderson. “We’re now rotors-up on both sides.” The fore-and-aft wing configuration provides even more experimental variables. Anderson said, “Having something that is inherently unstable is not unattractive. Balancing lift loads between the two wings gave us a configuration.” The PAV-ER layout also provides more hints about what happens in non-optimal flight. “It’s more interesting to look at an eight-rotor vehicle when it comes to failure modes. That drove the eight-rotor idea. That closely matched the scaling laws.”
Since the start of tethered flight testing in 2019, the PAV-ER has risen only inches off the ground. A new hover cage 100-ft (30.5-m) square has been built for untethered flight. Current battery endurance is just 5–10 minutes. However, EFRC researchers have a hybrid-electric powerplant rig that promises greater endurance than the hobbyist-grade lithium polymer batteries now on the testbed and affords additional testing opportunities. Anderson explained, “We’re noodling over a wire from the hybrid powerplant to this to see how the integration of the upstream hybrid and downstream pods would work together.”
The PAV-ER has its propulsion pods arrayed on a structural frame made of 4130 chrome alloy aircraft tubing. Anderson acknowledged, “The current vehicle is rather flexible, which can itself be a research component. How do you use rotors to control deflection for a lighter vehicle?” Plans call for a stiffer composite tube structure later in the test program.
Hover Off-the-Shelf
Rotors, motors, rotorheads and other PAV-ER dynamics required no new development. “Everything above the firewall is commercial off-the-shelf aerobatic helicopter stuff,” noted Pat Anderson. The testbed flies today with the same two-bladed rotors used on the Mk. 2 UAS. The commercial rotors turn at less than 2,200 RPM. In addition, Embry-Riddle researchers designed and flight-tested quieter rotors with much higher disc solidity than those on the Mk. 2 UAS. Plans call for the quiet rotors turning at less than 1,500 RPM to be tested on the eight-pod PAV-ER.
The two-pod Mk. 2 proved able to land vertically when one rotor shaft sheared, thanks to the high hub moments generated by the remaining cyclic-pitch rotor. The off-the-shelf rigid rotor heads of the Mk. 2 carry over to the air taxi testbed. “We really see the ability in this rigid rotor head to make control moments using cyclic control.”
The Mk. 2 tail-sitter adopted iterative parameter identification flight controls that “taught” the autonomous vehicle to fly. The PAV-ER has an Ethernet-based, input/output data acquisition and control cube supplied by United Electronics Industries (UEI) in Walpole, Massachusetts. The vehicle control can work with MATLAB programming language and the Simulink graphical programming environment. “Those compile stuff natural to an engineer into a program that will operate hardware,” explained Anderson. “Aero engineers don’t learn C++ or lower-level languages anymore.”
With no fuselage or aerodynamic fairings, the PAV-ER was designed without elaborate computational fluid dynamics (CFD) to map airflows. “The focus has been more on control analysis using uniform and dynamic inflow models,” noted Collins. Anderson added, “We’re looking more at flight dynamics. There’s a lot of separated flow. There’s a lot of complex flow. To compensate for the uncertainty in the aerodynamics, we are using a learning-to-fly methodology with iterative parameter identification and analysis... We do analysis, but we’re not doing full Navier-Stokes CFD.”
The PAV-ER power pods will not provide direct measures of thrust, torque and control moments. “We don’t measure the lift directly,” explained Collins. “Electronic speed controllers at each pod give us information on the power sustained by each motor. Our initial flights are using a radio control transmitter connected to that UEI cube.” The data acquisition cube scans 1,000 input/ output channels in less than a millisecond.
A single-pod test stand outfitted with a six-degree-of-freedom load cell will allow the characterization of the pod thrust, torque and control moments as collective, cyclic and RPM are varied. The results will be folded into the Simulink analyses to better characterize the vehicle through simulation. Plans call for the testbed to be instrumented with a Global Positioning System (GPS) receiver and inertial measurement unit (IMU) for more autonomous flight control.
Collins concluded, “When we move to automated flight, we may have a controller that says, 'When this rotor over here fails, how do you maintain automated hover?’” The results will feed back into the vehicle design process. “You could loop back into the design process to help you design not only the control method but how many rotors you might want to have.”
Leave a Comment