Grim Reapers

January 19, 2016 7:40 PM

More U.S. military drones are crashing than ever... Driving the increase was a mysterious surge in mishaps involving the Air Force’s newest and most advanced “hunter-killer” drone, the Reaper, which has become the Pentagon’s favored weapon for conducting surveillance and airstrikes against the Islamic State, al-Qaeda and other militant groups. From the Washington Post, January 19, 2016.
posted by cenoxo (32 comments total)

More crashing? Yeah, sure, given how many are used now, I'd expect that.

Without comparison to previous years and the data on how many missions were flown? I literally cannot tell you if there is s trend, or if a particular model of drone crashes more than any other.
posted by eriko at 7:50 PM on January 19, 2016 [1 favorite]

While Air Force leaders have publicly bemoaned a lack of personnel and resources...
posted by bonobothegreat at 7:52 PM on January 19, 2016

eriko: "The Reaper’s mishap rate — the number of major crashes per 100,000 hours flown — more than doubled compared with 2014."

The article also describes documents obtained via FOIA that discuss electrical problems resulting in crashes.
posted by tobascodagama at 7:58 PM on January 19, 2016 [3 favorites]

So? Is that less or more than other drones?
posted by eriko at 7:59 PM on January 19, 2016

Imagine you're an Air Force general. Would you rather have double the drone crashes or just one dead human pilot that craters his F-16?

Who cares if the drones crash, besides the accountants?
posted by Cool Papa Bell at 8:08 PM on January 19, 2016 [6 favorites]

dude, have you ever had any contact with Federal purchasing? that's a hell of a thing to write off with handwaving.
posted by indubitable at 8:13 PM on January 19, 2016 [1 favorite]

Well, at $14 million per, you can buy at least 25 of them for less than the cost of one F-35.
posted by mhoye at 8:26 PM on January 19, 2016 [6 favorites]

The title "more drones are crashing than ever" doesn't even agree with the plot they show, which is basically constant since 2009.
posted by kiltedtaco at 8:34 PM on January 19, 2016 [1 favorite]

The bit that jumped out at me is that half of all Predators ever purchased by the Air Force have been destroyed or badly damaged. Half.

On one hand, yeah, they're drones and they're relatively cheap you don't lose the pilot. On the other hand... half.
posted by phooky at 8:36 PM on January 19, 2016 [2 favorites]

Again, this compares to other drones how? I can accept this drone crashes 50% of the time, sure. How do other drones do?

This is how science works. "X breaks 50% of the time. How often does Y and Z break?"
posted by eriko at 8:40 PM on January 19, 2016 [1 favorite]

The Air Force publishes this stuff.

Here's the Class A accident rate per 100,00 flight hours for the Predator:
FY10 - 3.46
FY11 - 5.01
FY12 - 4.18
FY13 - 4.53
FY14 - 4.20
FY15 - 3.65

Reaper:
FY10 - 1.78
FY11 - 2.31
FY13 - 1.93
FY14 - 2.23
FY15 - 5.0

For reference, the five year average rate for the F-16 is 2.14. Lifetime average is 3.48. For the F-22, the 5 year rate is 5.39.
posted by kiltedtaco at 9:05 PM on January 19, 2016 [12 favorites]

So in other words, the Air Force is more or less uniformly inept across all platforms, regardless of unit price?
posted by Greg_Ace at 9:24 PM on January 19, 2016 [1 favorite]

I've been waiting for someone to create an EMI gun for use against unmanned aerial vehicles.

A ten watt maser would work as long as it had a reasonable aiming mechanism. If you hit a drone with it, induced noise would cause the computer to crash, and shortly thereafter the drone would crash.

I serously doubt that the electronics in those drones is hardened.
posted by Chocolate Pickle at 9:41 PM on January 19, 2016 [3 favorites]

One last thing, just because the statistics here are very interesting. One might ask, how have half of all the predators been destroyed, while the accident rate isn't tremendously higher than the F-16? So there have been 328 F-16s destroyed out of about 2300 purchased; approximately 13%. According to the link I gave above, these aircraft have accumulated 10,491,752 flight hours, very roughly 4,000 hours per airframe. The predators have 1,820,212 flight hours over (and this number is much rougher, given the losses and additions) ~200 airframes, so more like 9,000 hours per airframe.

The lesson here is that you can fly drones much harder, to the point of failure, in a way that you would never do for a crewed aircraft. When an F-16 gets to ~6000 hours it goes to the desert, but the predator stays in service.
posted by kiltedtaco at 9:55 PM on January 19, 2016 [15 favorites]

A maser? Heck, as I've said here before quite a few years ago, figuring out a way to fire a net out of a cannon wouldn't be too hard at all, and if you made it out of fishing line, the cost would be tiny.
posted by lupus_yonderboy at 9:57 PM on January 19, 2016 [1 favorite]

So in other words, the Air Force is more or less uniformly inept across all platforms, regardless of unit price?

Or ept. I don't have numbers for platform failures in other air forces. But what that tells me is "Nope, they crash and burn no more and no less than anything else in the USAF inventory."

In other words? This FPP is frank propaganda, and fails badly at being that. Oops!

That's annoying. I was hoping for evidence that drones were worse platforms than planes with competent pilots aboard. But you know what? Hope is not science. It's clear by this that they aren't, in which case, why waste human lives on this?

Count me, as of today, as one of those saying that the correct answer is to ground F-15s, F-22s, and A-10s, and replace them with drones. Oh, and ground F-35s, but that's been my answer for the last five years, because that program has failed so hard for so long that you just can't save it.

B-52s? That's harder. We don't have evidence, and they carry a *whole bunch* of munitions, something drones currently can't do. But the smaller planes? Yep, we can replace them wholesale.

Even better, we can replace them in the Navy as well, because those planes can barely carry anything!
posted by eriko at 9:59 PM on January 19, 2016 [5 favorites]

A commercial RF anti-drone weapon already exists. In my memory, but sadly not in my Google searches, I remember reading about similar devices in the military being built by (probably) Raytheon, as well as hardening against such devices, that are intense enough that they're not just jamming spectrum but locking up electronics with their spurious signals.
posted by traveler_ at 10:02 PM on January 19, 2016

Count me, as of today, as one of those saying that the correct answer is to ground F-15s, F-22s, and A-10s, and replace them with drones.

See, the problem then becomes having to put up with the terrible music that your AI idol singer produces when she falls in love with your last two test pilots
posted by Mrs. Davros at 10:27 PM on January 19, 2016 [8 favorites]

So in other words, the Air Force is more or less uniformly inept across all platforms, regardless of unit price?

Or ept.

Well yeah, but saying it that way lacks zing.
posted by Greg_Ace at 10:28 PM on January 19, 2016

...I remember reading about similar devices in the military being built by (probably) Raytheon...

via US Naval Institute

Anecdotally: I grew up in one of the cities where they build those ships, and folks said they never ran the old version of that radar at full power when in port for fear of scrambling nearby pacemakers.
posted by qbject at 11:20 PM on January 19, 2016

Count me, as of today, as one of those saying that the correct answer is to ground F-15s, F-22s, and A-10s, and replace them with drones.

Drones as we currently know them are designed for missions that don't require split second decision making: surveillance and standoff air-to-ground missile attack. Lag between drone and operator isn't an issue for those missions, particularly not when the enemy doesn't have an air force or SAMs.

You say you want drones that can take over air-to-air combat and close air support missions. Either of those tasks would require at least semi-autonomy, including target selection.
posted by justsomebodythatyouusedtoknow at 12:29 AM on January 20, 2016 [1 favorite]

If you hit a drone with it, induced noise would cause the computer to crash, and shortly thereafter the drone would

And what happens when some yahoo misses and hits the airplane far above it?
posted by anotherpanacea at 3:45 AM on January 20, 2016

I serously doubt that the electronics in those drones is hardened.

Why? I've been involved in design programs for much smaller, non-weaponized drones and they definitely had to be EM-hardened.
posted by indubitable at 4:22 AM on January 20, 2016 [3 favorites]

Because after it was discovered that the drones (and other things) were sending out their videos unencrypted, the problem still wasn't fixed 3 years later.

So something semi-obvious like EM proofing might've been skipped over if not in design then in implementation for reasons like costs. This is just me saying that there's precedence.
posted by I-baLL at 5:07 AM on January 20, 2016 [1 favorite]

"That's annoying. I was hoping for evidence that drones were worse platforms than planes with competent pilots aboard. But you know what? Hope is not science. It's clear by this that they aren't, in which case, why waste human lives on this? "

But that's not what the article seems to be about. The article seems to be trying to imply that there might be a flaw that's being used to take down Reaver drones. I mean, the crash rate for that drone doubled in a single year and the crashes are traced to an electrical problem whose trigger is as of yet unknown. This might be an indication that somebody's discovered a flaw that can be remotely exploited and is using it to take down drones whether through HERF or something else.
posted by I-baLL at 5:13 AM on January 20, 2016 [1 favorite]

Isn't this the sort of problem that was basically recognised and an approach defined for in WWII in operations research by R V Jones et al? You model expected losses under different operational conditions, spot anomaliesl establish reasons and suggest changes. You might expect higher losses for a particular kind of drone if it's doing different missions to others - but are those the actual losses you're getting? Is there a change in operations that will reduce the losses without reducing effectiveness? And so on.

I know that in WWII, this approach identified many problems (including the introduction of new weapons and procedures by the enemy) and helped optimise offensive and defensive approaches to realising objectives. I would have hoped, given the huge advances in modelling, statistical analysis and intelligence methods since then, we would have got rather better at it by now...
posted by Devonian at 5:18 AM on January 20, 2016 [2 favorites]

Because after it was discovered that the drones (and other things) were sending out their videos unencrypted, the problem still wasn't fixed 3 years later.

And? What does that have to do with EM hardening? I just told you, I've been on programs where EM hardening is a design requirement. They test it. What about this is not clear?
posted by indubitable at 5:29 AM on January 20, 2016 [2 favorites]

All you statistics warriors challenging whether the failure rate is meaningful might also find the other part of the fine article interesting

Investigators have traced the problem to a faulty starter-generator, but have been unable to pinpoint why it goes haywire or devise a permanent fix.

There's a specific engineering fault in their $14M aerial death robot. And it's causing them to crash. That seems worth attention.
posted by Nelson at 6:49 AM on January 20, 2016 [4 favorites]

Notably, this is an engineering fault. There's nothing to suggest that there's some kind of new countermeasure out there crashing our drones. It's just good old lowest-bidder construction.
posted by tobascodagama at 7:00 AM on January 20, 2016

When the Obama administration has discussed drone strikes publicly, it has offered assurances that such operations are a more precise alternative to boots on the ground and are authorized only when an “imminent” threat is present and there is “near certainty” that the intended target will be eliminated. Those terms, however, appear to have been bluntly redefined to bear almost no resemblance to their commonly understood meanings.
posted by bukvich at 7:53 AM on January 20, 2016

It is a going out of business war sale.
posted by Oyéah at 11:09 AM on January 20, 2016

I thought it was obvious that the starter-generator issue deserves attention and should be fixed. But the article chose to put it in a "drones are constantly falling out of the sky!" context rather than the context of failure rates for military aircraft, which I think was unhelpful and uninformative.
posted by kiltedtaco at 11:35 AM on January 20, 2016