Will a hexacopter survive a motor failure?

The belief

You are buying a heavy platform. It will carry something expensive, and it may fly over things you cannot afford to hit. Somebody tells you to buy a hexacopter rather than a quad, because "if a motor fails you still have five".

The belief is that motor count is redundancy — that six arms means the aircraft can lose one and carry on, and that eight means it barely notices. It is one of the most widespread ideas in multirotor buying, and it is wrong in almost every detail that matters. Hexacopters do fail after a single motor loss. Twenty-kilogramme hexacopters have come down after a single motor loss, and the forum threads describing it are all the same shape: the aircraft flew normally for a few seconds, then departed.

This article is about why that happens, and about the one number that actually decides whether your aircraft has any redundancy at all.

Why a quadcopter cannot survive it at all

Start with the quad, because it makes everything else make sense.

A quadcopter has exactly four actuators — four motors, four thrust values the flight controller can choose. It has exactly four things it needs to control: roll, pitch, yaw, and total thrust. Four knobs, four demands. The mixer is a square. Every demand is satisfiable, and there is nothing left over.

Now kill a motor. You have three actuators and four demands. There is no arrangement of three thrusts that independently satisfies four requirements — you are simply short of a control input. Something has to be given up.

The interesting part is what can be given up. In principle, the aircraft can abandon yaw. Let the airframe spin, and use the three surviving motors to keep roll, pitch and total thrust — the aircraft becomes a spinning platform that can still hold attitude and altitude in an averaged sense, and can be brought down under some control. This is a real result: research airframes have flown it, and some specialised firmware implements it.

Stock hobby firmware does not. Betaflight will not do it. iNAV will not do it. The mixer keeps solving for four demands with three working motors, the solution it computes is not achievable, the outputs saturate, and the aircraft rolls or pitches into the ground. In practice: a quad that loses a motor falls. Treat that as absolute. There is no configuration on a normal quadcopter that makes a motor-out survivable.

Why a hexacopter is better — but not in the way you think

Six motors, four demands. On paper, two spare degrees of freedom, so you can lose one motor and still have five actuators for four demands. That is genuine redundancy — the mixer problem is still solvable. This is the real and true part of the claim, and it is why hexes are recommended.

The problem is yaw.

A spinning propeller exerts a reaction torque on the airframe, opposite in direction to its own rotation. That is the only way a multirotor makes yaw: it is not aerodynamic, there is no rudder, there is only the net imbalance of propeller reaction torques. On a standard hexacopter the motors alternate CW, CCW, CW, CCW, CW, CCW around the ring. Three of one handedness, three of the other. In level flight the two groups' torques cancel, and yaw is commanded by deliberately running one group slightly harder than the other.

Now lose one motor. You have three of one handedness and two of the other. The torque budget is no longer symmetrical — the surviving three-motor group has more torque authority than the two-motor group, and that asymmetry is a function of thrust. Every kilogramme of lift the mixer asks that side to produce also comes with reaction torque it cannot cancel. Meanwhile, the missing motor's thrust has left a hole on one side of the ring, and the aircraft must re-shuffle thrust around the remaining five to keep the roll and pitch moments balanced.

Those two jobs fight each other. The thrust distribution that balances the attitude moments is not the thrust distribution that balances the yaw torques, and there is no longer enough freedom to do both perfectly. So the mixer trades: it holds attitude and lets the aircraft yaw away, or it holds heading and gives up some attitude authority, or it holds both and runs the surviving motors into saturation — at which point it holds neither.

Which of those happens depends on which motor died — its handedness and its position on the ring — and on how much thrust the remaining five have left to give. Losing a CW motor and losing a CCW motor are not the same failure, because the handedness of the survivors is not the same afterwards. This is why real hexacopters do sometimes spin up, drift off, or descend uncontrollably after one failure, and why the aircraft frequently appears to fly normally for a few seconds and then departs — it flies right up until the moment the mixer runs out of authority, and then it stops flying entirely.

Thrust margin is the whole game

Here is the sentence that matters more than anything else in this article.

Redundancy is not a property of the number of motors. It is a property of the thrust margin.

Work it through. A hexacopter hovering at 50% throttle loses a motor. The remaining five must now produce the same total thrust the six were producing. They have to work substantially harder — and on top of that, the mixer needs spare thrust above the new hover level to keep making control moments. If they cannot deliver it, the aircraft descends, and it descends no matter how clever the mixer is. The mixer cannot conjure thrust that the motors do not have.

Now take the same hexacopter loaded until it hovers at 80% throttle. It loses a motor. There is nowhere for the remaining five to go. That aircraft has no redundancy at all, and the six arms bought you nothing. It will come down, exactly as a quad would, and the fact that it is a hex is irrelevant to the outcome.

If you want a genuinely redundant aircraft, you must buy the thrust margin, and that means the aircraft must be over-powered for its normal mission — deliberately, with a large margin, at the maximum all-up weight you actually fly, not the empty airframe weight in the spec sheet. Redundancy costs money and payload. There is no way to have it for free by adding arms.

Coaxial X8 and octocopters

A coaxial X8 puts two motors on each of four arms, one above the other, turning in opposite directions. Lose one and the survivor is on the same arm, at the same position on the ring, with a partner of the opposite handedness. The thrust hole and the torque imbalance both stay local, and the remaining motors are geometrically better placed to cover them. That is genuinely more graceful than the hex case, and it is why serious heavy-lift platforms are so often X8s rather than hexes.

A true octocopter — eight motors on eight arms — simply has more actuators, more margin, and more freedom in the mixer, provided you have not eaten all of it with payload. Note that word "provided". The thrust-margin rule above does not stop applying because you fitted eight motors. An octo flying at 80% hover throttle is in the same trouble as everything else.

Also be honest about what a coaxial pair costs you: the lower prop works in the wake of the upper one, so a coaxial pair is less efficient than two props on separate arms. You are paying flight time for that graceful degradation. Sometimes that is exactly the right trade.

What the firmware actually does about it

This is a firmware decision, and the three mainline firmwares are not remotely equivalent.

ArduPilot treats motor failure as a first-class problem. It has thrust-loss and motor-failure handling whose purpose is to notice that the aircraft is not achieving the thrust and moments the mixer commanded, and to respond — reducing demands, flagging the condition, and in some configurations keeping a multirotor with sufficient spare motors flying well enough to be recovered. It is the only one of the three that has seriously engaged with this. Do not overstate it either: it is not a guarantee, its behaviour depends on the airframe, the margin and the configuration, and an aircraft with no thrust left over cannot be saved by any amount of software.

Betaflight and iNAV have essentially nothing here. They are not built for it. They will not detect that a motor has died. The mixer will carry on commanding output to a motor that is not turning, quite happily, all the way to the ground. That is not a criticism — neither firmware was designed for heavy redundant platforms, and pretending otherwise would be worse. But it means that if you are flying a heavy multirotor and telling yourself the extra arms make it safe, your firmware choice is a safety decision and you have already made it. The differences between the three firmwares are not just about features and configurator ergonomics.

What to actually do

• Buy the thrust margin. It is the only thing on this list that actually creates redundancy. Everything else is an optimisation of a margin you either have or do not.
• Know your hover throttle at your real all-up weight, with the real payload, on a real battery at a real state of charge. If you do not know this number, you do not know whether your aircraft is redundant, and you should assume it is not.
• Understand that a hex is not a safety guarantee. It is a possibility of survival, contingent on margin and on which motor fails. It is not a property you purchased with the frame.
• Consider the firmware deliberately. If motor-out survival is part of your safety case, Betaflight and iNAV do not provide it.
• Eliminate the failures you can. Most motor "failures" in the field are not dead motors — they are desyncs, ESC faults, connector failures, and motors that were running far too hot for a long time before anybody noticed. Redundancy is the last line, not the first. A properly tuned heavy platform that is not cooking its motors is a safer aircraft than an untuned one with two extra arms.
• Watch the per-motor outputs. An aircraft working one motor much harder than the others is already eating its own margin, and it is doing so before anything has failed.
• Test it, or do not claim it. For anything flying over people or property, a "redundant" aircraft that has never been tested for a motor-out is an assumption, not a safety case. If you have not seen the aircraft lose a motor — on a tether, over water, over an empty field, at a height and a weight you chose deliberately — then you do not know what it does. You have a belief about what it does. Believing is what the people in those forum threads were doing, right up until the aircraft departed.

The failsafe logic is subject to exactly the same rule, and for exactly the same reason: a safety feature that has never been exercised is not a safety feature. It is a hypothesis you have decided to test with the payload attached.