Calling all UAV Designers

[Mel Duval]

Hi guys,
I am trying to assist some young engineers in the design of a UAV senior project and need some help on failure rates. My current job is in Systems Safety (I have spent over 20 years in air to air and air to ground guided missiles on the program office side, Safety is a LOT different! :eek: ) and I am getting together the info to start a very basic Fault Tree Analysis. What in your experience are the highest failure rate items? Do you have any documentation, by chance?

A MIL-STD 217 analysis would give me data on overall reliability, but what I am really looking for is information on what fails in the field on real hardware. Anybody have some field experience to share? Things like battry packs (use a redundant power system), receivers (dual Rxs?), and servos (duals there too). What else is a good idea? Anything I am missing? Any ideas about a redundant data link?


I figured you guys would have the best knowledge around. I appreciate any help you can give.


Later,
Mel Duval

[Hovertime]

I'm not exactly the person you need but have some ideas.

Dual servo - not gonna work, if one servo has its electronics fail and fight another servo - they will not work both, or counteract each other. Provided that mech trans is intact 3 servos might add some redundancy or use more flight control surfaces , in that case small segment of control surface, even when deflected to full extreme would be counteracted by other control surfaces with working servos.

Batteries - motor trans system - same thing, use 2 closely positioned separate power systems with separate power sources, each must provide enough power to fly aircraft with good authority, in case one fails.

Normally batteries are pretty reliable, mil spec more so. Same goes for BL motors.
Servos are less reliable in theory since there are many components and moving parts, but we don't see them failing left and right in flight at all.

I know its nothing new but I thought I'd brake the ice here:)

[kd7ost]

I don't have any fact data sheets that shows it, but without a doubt the biggest source of failure is the pilot/builder. Bar none.

Dan

[Gary Warner]

(I'm not a UAV activist, but I do have many years flying R/C)

Reliability of most components is scaled with the stresses that component experiences. Put a small servo on a large control surface and it will be more likely to fail than if the opposite was done. How to quantify this is beyond my scope.

Reliability can also be related to non-fatigue issues, like a motor, gear or IC just taking a dive. In general, you get what you pay for. A cheap servo will most likely experience more anomaly failures than an expensive one. But again, it's too relative to quantify at the standards you are used to.

For the most part, off the shelf RC components would never reach the standards of FAA flight certification. (but I have seen experimental planes use giant-scale servos to run control surface trim tabs). There are some makers of exceptionally reliable components if the $ has no limits.

Some choices can be made to lessen the likelihood of there being a critical failure. Redundancy is good, as is over rating stress thresholds. In planes (model or real) we always have that trade off between what the engineers want for safety and what the marketing department wants for performance. In models one individual often has to play both roles. One only has to see that the door that came off the DC10 was a result of a previously designed part (designed well) being redesigned as part of an overall weight savings campaign.

So far a documents, that's going to be a tough one. As in real planes, they come from one of three places. The manufacture, field documenting and independent controlled lab facilities.

The manufacture is usually not happy to make public the failure rate statistics unless it's forced to or if their documents will show that they excel over their competition. Field documenting (in this case, "us") would require an orderly and methodical recording of failures, together with statistics data, for it to be of any statistical use. I'm not aware of anything like this being done in the RC world. In the full-scale world we have the FAA managed programs.

There may be some independent testing of RC devices (kind of like a "Consumer Reports" facility, but I would think that most of their work would revolve around performance testing (like does a O.S.40 really have a 1.1 BHP output).

You bring up an interesting point with these questions. As we are all so busy doing 'our own thing' we are not collecting statistical data about reliability, short of telling all of our friends about how your new servo failed and the plane crashed. Maybe as we move into larger and more complex airplanes, we need to consider a clearing house of sorts to gather and compile these failures. If UAV's and large RC planes become over regulated, it would be nice to have this data at our defense.

Just the ramblings of a man who needs to get a life :D .

Gary
--

[Unterhausen]

My experience in the Air Force is that real failure data is nearly impossible to determine, even in a system that is set up to collect such data. It was incredibly frustrating. There was one part that could lead to cabin depressurization, I swear one time a safety guy was close to hitting me over the failures. But when I tried to get the part that failed off of the list of qualified vendors, there weren't enough failure reports to justify it. Fortunately, the whole thing scared the vendor enough that they redesigned the part.

[Terry S]

Radio interference !

Terry

[Tuner]

In My experience the issues that have brought down my aircraft are as follows.

Airframe Construction errors & defects.(usualy only creates failur during high stress maneuvers) Broken wing and detached control linkages.

Radio Interference.
Gasoline Motor Failure/stall
Servo failure (very rare usualy with cheap servos)
Low Battery or Overly discarched Lipo's


Never had a bad reciever before and any time I have had a bad servo its been my own fault for not using the right servo for the job.

I think you should be more worried about flight control software bugs and turkeys.

Scott

[JettPilot]

I don't have any fact data sheets that shows it, but without a doubt the biggest source of failure is the pilot/builder. Bar none.

Dan

Dan could not be more correct. 99% is either defects in the design and building or pilot inexperience and error BAR NONE.

[lvspark]

Only component that has caused my planes to crash in the last 2 years is a bad speed controls with BEC. One was just bad and would drop BEC power after 5 minutes.. and the other a motor lead shorted and it blew the ESC to bits. Other than that I have had a couple motors go out, but glided to a safe and controlled landing. Servos have indicated problems and I changed them before they caused serious issues in flight.

Edit added link http://icat-server.mit.edu/Library/Download/205_Weibel%20R%20-%20Safety%20Considera.pdf

[CactusJackSlade]

Only component that has caused my planes to crash in the last 2 years is a bad speed controls with BEC.

So have you stopped using a BEC in your current configurations?

CJS

[lvspark]

I still use BEC but I make sure that when I am doing my preflight inspection to also inspect the motor leads and speed control very carefully. I also do not use E-flight brand speed controls anymore. I've had 2 E-flight go out on me for no appearent reason. One of those times included complete loss of control and crash due to loss of bec power.

[Mel Duval]

Thanks for the input!! Some interesting things to think about here.....

Later!
Mel D>

[Gary Warner]

Checklists are a must for large complicated aircraft. Here's the checklist I use for my electric powered Cherokee (109", 16 pound)

Gary
--