Flight Testing Summary

Apologies for delay. It has been an exciting, but incredibly busy month. As of September 4th when we finished our first round of flight testing, we switched gears to prepare this year’s high speed bicycle for the World Human-Powered Speed Challenge in Battle Mountain, Nevada. Following an intense week of fixing, filling, sanding, and blasting down the Nevada highway at 100+ km/hr, we headed straight to San Francisco to start work on an exciting project with Mythbusters’ Adam and Jamie! More on this later.

We’re now back in Toronto preparing for the next round of flight testing. Below is a summary of what worked, what didn’t, and how everything finally came together:

Day 1, Aug 21st: Final structural attachments. Finally, in a space big enough to put it all together, and floor level enough to properly measure from, we assembled the truss upside down and drilled the holes for the pins that hold it all together at the centre. In addition, the entire truss is externally braced with a series of lines, so these were cut, spliced to size, and for the first time, the entire truss structure was assembled and self standing.

Day 2, Aug 22nd: Structural adjustments. The length of all of the external bracing wire on the truss determines the position and angle of this highly flexible structure. Great care had to be taken to adjust the lengths of the bracing wires so that everything was in alignment. In addition, the bracing wires on the rotors themselves were adjusted so that the rotors would remain at right angles to the axis of rotation. The structure was loaded with the pilot weight, and held together! A small, but significant step towards successful flight.

Aidan attaching the external bracing lines that connect between truss arms.

Day 3, Aug 23rd: Spooling tests. Along with some additional structural adjustments, we performed tests on the spooling of the drive lines, trying to determine if there would ever be a scenario when a drive line might slip off a spool. Such a slip could easily cause a devastating crash. Fortunately we found that this should not be an issue as long as there is even a small amount of tension on the drive lines.

Day 4, Aug 24th: Rotor trimming. With the entire helicopter assembled, we began trim tests by pulling on the drive lines by hand, one rotor at a time. The rotors were trimmed by changing the angle of attack at the root attachment until both blades gave roughly the same amount of lift. If one blade is too high by even a small fraction of a degree, it will gain significantly more lift and drive the opposing rotor into the ground. Once the blades were balanced it was possible to have an entire rotor lift off and hover above the ground by pulling with a small force on the drive line. The moment the first rotor lifted was absolutely mesmerizing!

Day 5, Aug 25th: Adding pre-tension to the shear lines. We soon discovered that the Vectran braid used for all of the cross bracing lines tended to slacken after being loaded for the first time. This had some very unfortunate consequences for our truss structure, which uses Vectran lines as cross bracing throughout. Having these lines go slack meant that deflection of the entire structure was increased. With an extended team of volunteers we increased the tension on all of cross bracing lines by wrapping the base with Kelvar tow.

Day 6, Aug 26th: Taking in the drive posts. What we call the “drive post” is the axle of each of the rotors. It is connected to each rotor with a set of bearings and attached to the truss structure at the top. On the bottom it is attached to a set of lines that run along the bottom of the helicopter. These lines provide bracing for the entire structure so that it supports the load of a pilot as an arch would, rather than as a bridge would. This makes a much more efficient structure, but it means that the drive posts are being bent inwards under a fairly significant force. The bent drive post tilts the rotor outwards so that it contacts the ground making it impossible for the rotor to spin. The issue was fixed by cutting the truss structure where it connects to the drive post and deliberately setting the posts at an outwards angle. So when the structure is unloaded, the rotors are tilted, but once the pilot climbs aboard, the post are bend inwards and the rotors level out.

Day 7, Aug 27th: Rotors actuated by drive system. With the rotors levelled, we could finally apply the drive force from the pilot. The drive line from the rotor was pulled through the pulleys attached to the bottom of the bicycle frame and spooled onto the spools attached to the cranks. The drive system worked absolutely flawlessly, a tremendous accomplishment gives the magnitude of the forces that it must bear.

Unfortunately at this point we became fully aware of a significant structural design flaw. When the structure is loaded, half of the cross-bracing lines will be in tension, while the other half go slack. The cross bracing lines are meant to provide torsional stiffness as well, but if half of them are slack the torsional stiffness is almost completely absent. The result is that when the pilot loads the structure, each rotor post will tilt, and if it is not being held vertical, it will quite literally fall into the ground. After discussion a variety of options, we decided to externally brace the rotor posts with long vectran lines. For the first time the structure could stand on its own, loaded and ready for flight.

The cross-bracing lines can be seen marked with orange and green tape.

Day 8, Aug 28th: First flight. A series of trim test were first performed to ensure that all the blades were producing roughly the same amount of lift and that all of the external bracing wires were properly adjusted so that each rotor post was vertical. From here on, much of our time would be spent trimming and adjusting the structure and the rotors. On the forth trim test two of the rotors lifted off the ground, and on the fifth test we achieved a very brief moment of liftoff!

Atlas in flight.

Day 9, Aug 29th: First crash. The first flights were done without the control canards attached to limit the number of systems that we had to worry about. Now, the canards were attached and another flight was attempted. Unfortunately, as the rotors spooled up, one of the blades went wildly high, throwing the opposing blade into the ground and snapping the spar at the root. In the debrief it was determined that once the rotor attained a certain speed the aerodynamic pitching moment of the canard overpowered the control spring, deflected the control surface and drove the blade upwards. The carbon-fibre rotor spar was fixed on the spot and while the epoxy was curing a reflexed tail was added to the canards. Quick flight tests were performed by holding the canard out the car window and adjusting the reflex until the pitching moment was appropriate. This ensured that both the control spring and the pitching moment would always be in the same direction, and would always keep the control line under tension. The solution worked surprisingly well, and in the afternoon several more trim flights were performed.

Canard control surface with reflexed trailing edge.

Day 10: Aug 30th: Calling off flights. Even with the fixed control surfaces we kept on having to call off flights just as the helicopter was taking off. With all of the external bracing wires there is only a narrow space in which the blade can fly. If it goes too high or too low it will clip a bracing line. So if the trim was off by even a small amount someone would end up yelling stop before all four rotors had lifted off. We attacked the problem in two ways: first we moved the external cross bracing lines further up the structure to clear more space, and second, we replaced the Vectran rotor bracing lines with steel lines that would keep rotors closer to the ground.

Day 11, Aug 31st: Untethered flights. After a few quick trims and a short flight, we proceded to fly “untethered”. In previous flights, Cameron would be holding on to the bike frame to ensure that the helicopter wouldn’t drift, but now the helicopter was free to move, and we were surprised at how little drift there was. Unfortunately upon landing, the power was let off too quickly and one of the rotors sprung into the ground, snapping another spar. At this point I have to congratulate the team for keeping their composure through all of the ups and downs. Within a few hours the spar was fixed and we continued on with more tests.

Day 12, Sept 1st: Control adjustments, springs. On Sept 1st, the field was in use for a soccer tournament, so we took the time to clean up some of the control actuation. We replaced the control springs that had been giving us issues, and replaced some of the control pulleys that were a bit sticky.

Day 13, Sept 2st: Going for altitude. With time running out in the summer, we prepared for the series of test flights that would lead to an American Helicopter Society Sikorsky Prize attempt. We needed to establish that the controls were in fact functional and that we could climb and descend in a consistent manner. After a few trim flights we decided to aim for a 1.5m altitude flight. The helicopter lifted off quickly and began to accelerate upwards, but under the increased g-loading, one of the rotor spars snapped, tearing through the ribs and mylar skin. This fix took a little bit longer, but we managed to get in a few more tests in the evening. The conclusion was that the spar may have failed due to a manufacturing defect, but either way, the factor of safety is very small. To be safe, we decreased the length of the rotor bracing lines, putting more of the load on the line, and less bending load on the spar itself.


Day 14, Sept 3rd: Final flight tests. On the final day, we set up again for what could be a Sikorsky Prize attempt. We managed several trimming flights and a flight of 15 seconds where the controls were used to keep the helicopter within the 10 metre box. Unfortunately upon landing we had two ground strikes and two more broken spars. We immediately repaired the spars, but since this was our last day on the field we decided to put the helicopter away in one piece, take a step back and evaluate what we had learned.

Todd, Cameron, and Calvin at the centre as Atlas takes flight.


Given our ambitious time lines we are extremely pleased with what we were able to accomplish: we achieved flight, becoming the forth human-powered helicopter in history to do so; the final weight of the helicopter and the power required to fly turned out to be surprisingly close to the predicted values; and, although not fully tested yet, our system for control actuation looks very promising. Going forward, our plan is to build a single-rotor test stand and develop a better trimming strategy so that each rotor can be optimally tuned and accurately trimmed. This should reduced the chance of ground strikes, reduced the required power, and give us a chance to further refine the controls. Most of the team is back at school now, so the progress will not be as fast as it was in the summer, but within the next few months we should be flight testing again, with our sights set on the prize!

.h Visual Unity
.University of Toronto Aero+Engsci+Materials
Aero Club of Canada Trust Fund
CAD Micro
CD Adapco
Paterson Composites
A-Line Precision Silver
Avion Technologies
Pratt & Whitney
Proto 3000
Quest HP
Ultimate Workshop
HED Cycling
Metal Supermarkets