Engineers strive for system reliability. Getting there is a surprisingly complex problem.
Episode Summary:
Bell’s prototype for an autonomous pod transport cargo drone suffered a failure during a recent test flight at Fort Benning, Georgia. No one was injured, although the aircraft sustained substantial damage. According to the NTSB report, a cracked ceramic capacitor in the circuitry of one of the motor controllers caused one of the four electric lift motors to fail, leading to the crash. All flight test programs have failures, but multi-motor, multirotor drone designs have different failure modes and reliability problems compared to conventional helicopters. It’s a tough challenge. Jim Anderton comments.
Access all episodes of End of the Line on Engineering TV along with all of our other series.
Transcript of this week’s show:
To see any images, graphs, charts, graphics and/or videos to which the transcript may be referring, watch the above video.
This is the Bell APT70, a large tail-sitting four-rotor cargo drone developed for battlefield supply and civilian cargo delivery service.
It is one of a new generation of electric multirotor UAVs that are expected to revolutionize the transport of critical cargo in military, law enforcement, medical and commercial services. It is big, with a gross weight of 400 pounds, and it translates to horizontal winged flight at up to 100 miles an hour and can carry 70 pounds of cargo in its current configuration.
Those are useful numbers, and the system has undergone extensive testing. Like all test programs, especially in aerospace, you have to accept some losses—and last year at Fort Benning, Georgia, an APT70 fell from the sky and sustained substantial damage.
Interestingly, the National Transportation Safety Board (NTSB) issued an aviation accident report similar to the ones they use for crewed airplane accidents and determined that the reason for the crash was the failure of the electronic speed control unit operating one of the four lift motors. Bell designed the system to hover and allow a controlled descent in case of a failed motor, but in this case the unit hit trees five seconds after the failure, before the ground-based pilot/operator could take remedial action.
The electronic speed control failure was tracked down to a cracked ceramic capacitor in the ESC circuitry. Now, anyone that has used electronics in any form to control anything from a military drone to a TV remote has seen circuit failures caused by discrete component failure. A failed ceramic capacitor is not unusual. But for engineers that work on programs like these, this accident shows how little things sometimes require complex thinking, and the answers to a lot of questions.
The APT70 system is designed to reduce thrust on the rotor opposite a failed motor to allow a controlled descent and landing. In this case, there was not enough time for the pilot to gain control, so it went down. What does an engineer do in a situation like this to improve reliability? Well, there are several options.
One would be to add system redundancy to the electronic speed controller to reduce the likelihood of this kind of failure in the first place. That sounds good; however, dual or triple redundant electronic speed controls would not only add weight and cost, but the extra systems would themselves be potential failure points—ones which could prevent accidents but decrease overall dispatch reliability. The unit wouldn’t go down, but it would likely have more down time.
Do you instead implement a tighter quality assurance system and add component level inspection to components such as capacitors? Do you pass that responsibility down to the capacitor manufacturer? Or do you make the ESC boards a quick-change unit and add an additional routine maintenance check, possibly automated, before flight? Or do you tackle the problem a different way, by accepting a statistically reasonable level of failures, while minimizing the impact of those failures.
A possibility would be to automate the emergency response system, perhaps with some kind of machine vision system that would pick a safe place to land and simply put the unit down. On a battlefield, it is likely that anywhere will do—but if the unit is flying over an urban area on a civilian mission, dropping onto a busy freeway or downtown intersection simply isn’t an option.
Do you use AI to make real-time decisions about that emergency landing? What would that take? And what would it cost? Or do you redesign the system to use more motors, and more rotors, to maintain flight control even with the loss of one propulsion unit? Bell’s own tiltrotor products cross-connect both rotors so that a single engine failure will still allow the safe landing of the aircraft. That is mechanically complex and expensive, but for crewed systems it is a necessity.
So how do you make an unmanned system like this simple, cheap, safe and reliable? It is very hard, and it is expensive. Right now, there are dozens of start-ups across America working to build systems like Bell’s APT70. In terms of vertical takeoff rotorcraft, Bell—along with Sikorsky—essentially wrote the book. They invented the modern helicopter industry. They know what they’re doing. But those basic questions about the statistical likelihood of failure, the consequences of those failures and the best strategies for addressing those failures, are the same for billion-dollar air framers and garage tinkerers.
Are we really getting close to a world where Domino’s Pizza drops a pie onto your front porch? I suspect we’re closer than we realize. But it’s going to take a heavy lift from a lot of talented engineers to make it reliable.
