Airbus issues recall of over 6,000 A320 aircraft for software rollback

[thewayne]

This is... interesting. In October, an A320 operated by Jetblue was en route from Cancun to Newark when it suffered an unexpected loss of altitude. It made an emergency landing in Miami. No injuries or damage to the aircraft. The FAA directive reported that the October flight "experienced a malfunction in its elevator aileron computer (ELAC), which is a computer that controls the plane’s pitch or nose angle. Airbus believes that solar flares—intense and concentrated streams of electromagnetic solar radiation—may have corrupted the data and caused the ELAC to malfunction, suddenly sending the aircraft plunging down."

The article goes on to say "The fix for the issue is a relatively quick revert to earlier software before the planes can fly again, except for some jets that may require a complete hardware replacement." (emphasis mine)

Now, this raises some questions. First, why does reverting the software to a previous version fix the problem? Obviously reinstalling software would fix a corruption issue, unless there was hardware damage, in which case you'd have to replace the hardware and then reinstall the software. Since you're reverting the software, that implies that the older software had some self-healing features that could detect if something had damaged the program and it could reload part or all of itself from safe storage, not unlike error-correcting memory. And personally, if I were designing software for aviation that would fly on aircraft, I'd like to have this feature. I have no idea if their software can do this.

But this is the big question: if the software can self-repair, WHY WOULD SUCH A FEATURE BE REMOVED? Clearly such a feature would take a lot of resources, both occupying computer memory (overhead) and processing power (CPU resources) with its monitoring. BUT THIS IS A FIELD WHERE YOU WANT BOTH BELT AND SUSPENDERS! I just don't get why you would dumb-down a program.

The other question is why the computer doesn't have increased shielding? Granted, you cannot completely shield equipment in aircraft against high-energy particles, it's just not practical. The particles are too energetic, the weight and size of such shielding would be prohibitive. And because aircraft fly at high altitude, you don't have as much atmosphere acting as an attenuator, slowing down the particles a little bit. This is why living at high altitude, such as Russet and I do at 9,000', people have increased rates of thyroid problems and cataract formation: we are exposed to harsher sunlight and more directly hit by higher energy sunlight, where as people living at sea level get the full benefit of a skyful of air slowing things down.

So a couple of questions linger over this. Reloading an older version of the software shouldn't take long: after it's reloaded, the flight crew will have to confirm the ELAC system is functioning as expected. And if it doesn't load properly, it's probably due to damage to said system and the plane will have to be taken out of service pending replacement of the computer. Disruptions to air travel to accommodate things like this will cost the airlines a lot of money and result in hordes of angry passengers whose travel plans are being disrupted.

https://gizmodo.com/how-solar-flares-could-have-corrupted-an-airbus-plane-2000693690

Comments

[graydon]

What I've seen about this is that the event itself was surprising; cosmic ray damage is expected, but the amount was outside the previously accepted scope.

(Which is to say, the spec for the possible problems they're designing against just changed. Which would explain why some physical hardware will need to be replaced; fine against the old spec, not fine against the new spec.)

It's also pretty easy for the most-current software to contain an optimization that turns out to fail when something unexpected happened, and for the previous version to be believed OK in that event.

Flight control software has speed constraints, and response margins; really easy to see how someone could be trying to improve the response window and introduce a case where the corruption checking got bypassed when something unexpected happened.

[disneydream06]

Here's hoping, for everybody, that it's a quick and easy fix. :o :o :o
Hugs, Jon

[richardf8]

I am, at this moment, in a hotel in Santiago, Chile because the Airbus 320 we took from Montevideo had the update applied as we sat on the tarmac and the plane rebooted 3 times over about 90 minutes causing us to miss us to miss our connecting flight.

[thewayne]

Wow.  I just read that a vast majority of the planes were updated over the weekend, leaving a few hundred of the oldest that needed computers replaced.  I have a feeling that your plane might have been amongst those older ones.  Hope your travels resume soon and safely!

[richardf8]

I can't guess, but I assume that's not the case, since the plane did ultimately take off rather than us being deplaned in Carrasco.

[silveradept]

And, of course, the people who are desperately trying to believe there's no situation at all where things might be getting exposed to more powerful radiation or that the planet may be changing in some way, and that we might need to try and predict those kinds of events are probably hoping that the problem will simply go away and not bother them or demand that they do something.