they don't get it...

[JohnRoberts]

I hate listening to talking heads on TV speculating about stuff without a clue (kind of like me here. )

The Boeing aircraft issue, appears to be bad sensor data driving the anti-stall algorithm to point the nose down. Could be air speed sensor under reporting speed... I vaguely recall issues in a small corporate jet where sensors could ice up and give bad data, but experienced pilots knew to ignore bad data.

It looks like Boeing will upgrade their software to make it even more idiot proof, as we have less experienced pilots to go around. Combining multiple sensor inputs to confirm or ignore bad data. It seems they could extract forward ground speed from GPS data too, but this must be factored with local wind speed to impute air speed. Of course they could probably tell if data was totally unreliable.

Of course this is pure speculation on my part so I may be wrong and have to eat my words.

JR

PS: BTW President Trumps tweet about this is complete nonsense (just like so many of his tweets to chum the news cycle). 

 

[dmp]

I don't know anything about planes, but my car has several traction control features that make automated responses to sensor data (toyota rav4). They are not something the driver can override immediately in a sudden situation.  The power to the individual wheels is automatically adjusted based on data.
Not sure this stuff has been extensively tested in areas that get a lot of bad road conditions (I don't think it has) 
Having grown up driving in the snow I feel pretty good about responding myself and the car's auto response is pretty strange sometimes .
I know someone who had to return a Subaru under the lemon law because it got so squirrely on snowy/icy roads.
I've read some people figure out how to pull fuses for some of the features on some model cars.

Anyway - point being - maybe the planes were tested on 99% of expected conditions and the tail outliers are not being responded to correctly.

 

[zamproject]

The Boeing aircraft issue, appears to be bad sensor data driving the anti-stall algorithm to point the nose down.

Hello JR

The black box have not been read so far (Ethiopia don't have the tools)
so nobody can say what happen for the moment

sensor problem was the AF447 issue IIRC

Best
Zam

 

[JohnRoberts]

I don't know anything about planes, but my car has several traction control features that make automated responses to sensor data (toyota rav4). They are not something the driver can override immediately in a sudden situation.  The power to the individual wheels is automatically adjusted based on data.
Not sure this stuff has been extensively tested in areas that get a lot of bad road conditions (I don't think it has) 

It was developed for stuff like that. I have seen some impressive demonstrations of cars like corvettes driving on frozen lakes with and without the cybernetic assistance. I have no idea about your specific example.

Having grown up driving in the snow I feel pretty good about responding myself and the car's auto response is pretty strange sometimes .

I did too (in NJ). There are useful experience based techniques (like maintaining forward momentum) and not using brakes unless absolutely necessary. I learned that lesson about brakes before crashing into a parked car, in a snow covered parking lot... (can't brake and steer at same time). Luckily it occurred at low speed and I had enough time to figure out what was happening and release the brakes before crashing.

I know someone who had to return a Subaru under the lemon law because it got so squirrely on snowy/icy roads.
I've read some people figure out how to pull fuses for some of the features on some model cars.

Anyway - point being - maybe the planes were tested on 99% of expected conditions and the tail outliers are not being responded to correctly.

I can't vouch for all auto designs but the brake steering approaches appear to have merit... I totaled my (limited edition) '93 in the rain when it hydroplaned in the rain, pushing me off the road surface at a 15' angle... I made multiple small heading corrections to avoid spinning out but I ran out of shoulder before getting it completely back up on the road surface. I ended up sticking it head first into a guardrail, totaling a car that was sure to become a collectors item.  Since then I have been envious of the advanced brake steering technology where they can brake one or the other back wheel to pull the car straight. That would have saved my bacon in the rain that day.

Even ABS brakes that I do have saved my sheet metal one time when some bubba, ran a red light and through an intersection cutting me off... I locked up my brakes and pulled the wheel hard right... the ABS slowed me while still allowing the wheels enough traction to steer, out of the path of a sure collision. The butt hole didn't even slow down and just sped away.

JR

 

[JohnRoberts]

Hello JR

The black box have not been read so far (Ethiopia don't have the tools)
so nobody can say what happen for the moment

sensor problem was the AF447 issue IIRC

Best
Zam

Perhaps similar but different aircraft... It appears related to anti-stall algorithm which may be different for different airframe. There are some military aircraft that are almost impossible to fly without computers because they are unstable (makes them easier to maneuver, faster, etc).

I suspect big airliners are tweaked to maximize efficiency so probably low drag that could involve different stall parameters.

Reportedly the FAA has  seen some GPS data that reveal similarities in altitude instability between the two crashes.

I suspect these planes are harder to fly by the seat of the pants, and modern pilots are perhaps less experienced.

===

I can imagine software tweaks to extract data from say the planes own GPS to compare ground speed with air speed sensors to confirm plausibility of the data.  They probably already do to read wind speed, but may not use it for checking double checking sensor data.

caveat lector... more speculation on my part....

JR 

 

[hitchhiker]

This article in the CBC news explains the software issue that not all pilots are trained for.

https://www.cbc.ca/news/business/boeing-737-business-impact-1.5054535

 

[abbey road d enfer]

There are some military aircraft that are almost impossible to fly without computers because they are unstable (makes them easier to maneuver, faster, etc).

All commercial planes designed in the last 20 years use supercritical wing profile, so yes, unstable.

 

[abbey road d enfer]

Anyway - point being - maybe the planes were tested on 99% of expected conditions and the tail outliers are not being responded to correctly.

That's the problem with testing vs. reality. Something happens that just couldn't have happened.

 

[JohnRoberts]

That's the problem with testing vs. reality. Something happens that just couldn't have happened.

Since they have both black boxes we should have a lot of data to compare and contrast between the two crashes.

I can imagine modern pilots becoming reliant on the instrumentation.

JR

 

[cyrano]

There's some rage amongst pilots because they weren't told the system exists.

From the preliminary report about the first crash, it appears the pilots corrected the "nose down" movement 26 times before crashing. The second case seems to be the same, going from satellite data.

The "nose down" movement is used to prevent stalling. It shouldn't be active at all during take-off, or landing, according to pilots I've talked to.

 

[zamproject]

Since they have both black boxes we should have a lot of data to compare and contrast between the two crashes.

I can imagine modern pilots becoming reliant on the instrumentation.

JR

Yes black boxes are in France now, at BEA i guess.
There is multiple ground observation/report that say the plane was on fire before touching the ground...
we'll see

Best
Zam

 

[boji]

my car has several traction control features that make automated responses to sensor data

I've been in circumstances where I want the loss of grip so as to gauge sections of road, etc, or I want to hit the throttle to maximally avoid something in the rain, and I'm stuck swearing at the ATC as it feathers my control input.

If it's true about hidden overrides that 'correct' fly by wire inputs despite the pilot undoing those changes over and over then wow, talk about poor coding.  They were just asking for trouble.

 

[JohnRoberts]

There's some rage amongst pilots because they weren't told the system exists.

From the preliminary report about the first crash, it appears the pilots corrected the "nose down" movement 26 times before crashing. The second case seems to be the same, going from satellite data.

The "nose down" movement is used to prevent stalling. It shouldn't be active at all during take-off, or landing, according to pilots I've talked to.

Yes it sure sounds like a faulty air speed sensor.

I am not a pilot (either), but it seems to me that anti-stall warnings are most important when close to the ground where you don't have enough altitude to regain speed, so during take off and landing.

The good news is this seems fixable, the bad news is events like this are more likely to occur as aircraft become too complex to fly by humans without cybernetic assistance.

Of course autopilot crashing the plane is never good.

JR

 

[JohnRoberts]

I've been in circumstances where I want the loss of grip so as to gauge sections of road, etc, or I want to hit the throttle to maximally avoid something in the rain, and I'm stuck swearing at the ATC as it feathers my control input.

it is telling you that it lost grip... that is why it feathered. It doesn't predict wheel slip, it measures it.

Unclear if you would move faster without traction control. Maybe on a dry race course, unlikely in snow/wet. 

If it's true about hidden overrides that 'correct' fly by wire inputs despite the pilot undoing those changes over and over then wow, talk about poor coding.  They were just asking for trouble.

Still waiting for expert analysis, while due to liability I don't expect Boeing to stand up and announce a major FU.

JR

 

[cyrano]

Unclear if you would move faster without traction control.

Try ice. Some traction control systems can't even get the car running on ice. But even the better ones are far slower than a driver with extensive experience who knows how to handle that car on ice.

These systems are around because most people haven't got a clue...

Majority rules.

 

[abbey road d enfer]

Try ice. Some traction control systems can't even get the car running on ice. But even the better ones are far slower than a driver with extensive experience who knows how to handle that car on ice.

Different expectations, hence different optimizations. A commuter expects to get to destination with both his body and his car untouched. A rally driver doesn't care if the car is dinged and his body bruised.
A car manufacturer has to play it safe by all accounts.
Cybernetic systems on military planes have a very different optimization than civil airplanes. Even the best military pilots cannot outperform their embedded systems. Top Gun and Star Wars are myths.

 

[JohnRoberts]

Try ice. Some traction control systems can't even get the car running on ice. But even the better ones are far slower than a driver with extensive experience who knows how to handle that car on ice.

I generally try to avoid ice, one of the few things I like about living in MS. I only experienced "black ice" on interstate overpasses one time in three plus decades.  Many of the locals couldn't even negotiate those short interstate overpasses without losing control during that icing condition.

I grew up in the Northeast so have plenty of experience driving on snow, can't say I've ever mastered driving on ice (other than holding a straight line and avoiding my brakes).  :

These systems are around because most people haven't got a clue...

Majority rules.

Getting slightly back on topic... I heard another factoid about that aircraft involved in the two similar crashes. Apparently that newer model has engines mounted in somewhat different location as well as perhaps other changes to improve fuel efficiency, that changes the flight characteristics, or how the airframe responds to pilot flight control inputs. To make the aircraft behavior more like the previous versions of Boeing aircraft to reduce the need for additional pilot training /familiarization they tweaked the computer control interfaces to respond like the earlier airframes.

I heard this mentioned by a technology editorial writer not the company, and will try not to speculate too much from this, other than the reduced reliance on manual "hands on"  flying by human pilots.   

JR 

PS: Self driving cars will save thousands of lives, and the then rare accidents will become big news (already are).

 

[cyrano]

There are several other crashes surfacing that could be related. Not passenger planes, tho.

And it seems the FAA outsourced checking of the Boeing safety to... Boeing.

https://www.newyorker.com/news/our-columnists/how-did-the-faa-allow-the-boeing-737-max-to-fly

That's at least a bit strange, isn't it?

 

[JohnRoberts]

I don't trust the government to be smarter than Boeing's engineers, or that boeing would game human safety... This is not the same as VW cheating emissions testing.

Here is an interesting story I hadn't heard before... https://www.bloomberg.com/news/articles/2019-03-19/how-an-extra-man-in-cockpit-saved-a-737-max-that-later-crashed

It seems pilot training is important.

JR

 

[gridcurrent]

McDonnell/Douglas bought Boeing with Boeing's money.
While the MD engineers are retired or dead,
the culture that gave us the DC-10 permeates at Boeing today.