Hacker News Reader

Comment by alexis_fr | Hacker News Reader

Airbus scandal won’t come “in a month”: Boeing’s problems became public in 2013 and it is the same problem over and over.

- The Ducommon airframes in 2013, problem is FAA delegated certification to Boeing,

- The batteries, albeit smaller,

- The MAX, 2 crashes due to oversight and to the FAA delegating certification to Boeing.

Boeing’s problem didn’t come in a day, but in 20 years since the merge with MD, who had the same lose and fast culture. Airbus is safe for the time being, even if risks always exist, political instability for example. Not the case for the moment.

View on HN

Well, for Airbus there is this where it looks like problems were swept under the rug https://en.wikipedia.org/wiki/Air_France_Flight_296

... though from the same era, there is this for Boeing which seems much worse https://en.wikipedia.org/wiki/Boeing_737_rudder_issues

> for Airbus there is this where it looks like problems were swept under the rug

Or this:

https://en.wikipedia.org/wiki/Qantas_Flight_72

This is the same basic design flaw that Boeing's MCAS has (allowing a single AoA sensor failure to cause the airplane to do unexpected and dangerous things), but unlike MCAS, it wasn't an ad hoc addition to deal with an issue in a redesign of an old airplane, it was designed into all Airbus fly-by-wire aircraft from the start.

You keep posting this, but it's simply not true- Qantas flight 72 did not experience an AoA sensor failure, and even if it had, the A330 uses all 3 independent AoA sensors and all 3 independent air data inertial reference units to command control movements.

The issue was caused by one of the ADIRU CPUs corrupting valid sensor data to produce a series of very specific spikes, hitting a specific edge case in the AoA cross-check logic that made the flight computer behave as if the erroneous data it was seeing was valid.

It's undoubtedly a serious design flaw, but it's not even remotely in the same ballpark as the Max's flagrant disregard for all process and well known design standards.

http://code7700.com/accident_qantas_72.htm

> a specific edge case in the AoA cross-check logic

What the article you linked to describes doesn't look quite like that to me. It looks like a design that does not use the "median of three values" method that the article says is used for "most" sensors (but not AoA). I agree, however, with your correction that it was not a direct AoA sensor failure such as occurred in the 737 Max incidents.

> It's undoubtedly a serious design flaw, but it's not even remotely in the same ballpark as the Max's flagrant disregard for all process and well known design standards.

It's not the same as MCAS in the sense that it's not an ad hoc addition to deal with an issue that arose in a redesign of an old airplane, true.

However, it is a case of (a) a flaw in an automated system causing the plane to automatically take an action that is unexpected and dangerous, instead of the automated system detecting the error; and (b) an automated system overriding the input of the human pilot in a case where the human pilot can plainly see that the automated system is doing something wrong.

Also, it illustrates a more general design philosophy with Airbus that has led to other incidents, which is to hide important information about what the automated system is doing from the pilots and limit their ability to interfere with its operation. The pilots have no direct readout of the AoA sensor values that the automated system is using, or indeed of most of its other inputs. It was also not clear to the pilots exactly what mode the automated system was in (they thought it was in Direct Law when it wasn't), and there was not a simple "stop the automation" button they could press to put the plane into a known manual mode with known behavior.

While I would agree that this is not the same specific failure mode as MCAS was with Boeing, I'm not sure it's not "remotely in the same ballpark" as far as long term implications are concerned.

Reply on news.ycombinator.com