Boeing Sells 0 Planes on First Day of Paris Air Show v. 123 for Airbus, Yet Boeing Still Pushing for More Weakening of FAA Standards Via Greater Use of Computer Testing
Boeing execs must be smoking something very strong.
The manufacturer first tried blaming pilots for its two 737 Max crashes, even after groundings by airline regulators showed they thought something was wrong with the plane. Even as of this week, CEO Dennis Muilenberg has only gone as far as admitting to being “not consistent” in how Boeing conveyed information to regulators, airlines, and the public. And the apology was limited to the aircraft lights!
Muilenburg is still standing because he has the support of his board…a situation reminiscent of John Stumpf’s final days at Wells Fargo, where the embattled CEO stayed long after his sell-by date thanks to misplaced loyalty.
Muilenburg and his allies on the board appear to think that the Boeing 737 Max debacle will eventually blow over despite major customers continuing to suffer during peak travel season by having 737 Maxes mothballed, the continuing uncertainty as to when the plane will be allowed to fly again (even if the FAA gives a green light relatively soon, what about foreign regulators?), and continuing doubts about the airplane’s safety among high-profit business travelers.
And the latest stories don’t offer any cheer. For instance:
But even though Boeing directors have been telling themselves that the FAA is joined to Boeing at the hip and customers have no where to go (at least for years) given that Airbus can’t displace Boeing any time soon, the critical second assumption is not looking so good. Boeing’s most important audience, its buyers, have given the airline maker a loud wake up call by refusing to place any orders at the first day of the Paris air show.
Boeing knew the Paris Air Show would be slow, but this may have been worse than they expected. On day one of the show, Boeing did not announce a single new order for any of its airplanes, while Airbus recorded orders and options for 123 planes, according to the aviation consulting firm IBA.iQ.
On top of that, some customers are coming awfully close to calling for Muilenburg’s head. From the Seattle Times:
One of Boeing’s biggest 737 MAX customers sharply criticized the company’s handling of the crisis resulting from two recent crashes, and raised the stark question of whether Boeing Chairman and Chief Executive Dennis Muilenburg will be forced out as a result.
In an interview at the Paris Air Show, Avolon Chief Executive Dómhnal Slattery, who runs the third-largest airplane leasing company in the world, said the additional bad news in Paris on Monday of a delay to the 777X program will only feed “the swirling debate: whether Muilenburg will survive all of this.”
“I think he has the support of the Boeing board,” Slattery said. “But our view here is that Boeing have failed to win the media communications battle. They are forgetting about the most important constituency, which is the hundreds of millions of potential passengers.”
He said the refusal of Boeing to fully and publicly accept its share of the blame, an approach that has produced awkward moments with Muilenburg seemingly bound by legal restrictions from being too plain-spoken, has damaged Boeing’s reputation, especially overseas.
“You can understand that as a legal strategy,” Slattery said. “But if you play that out to its Black Swan scenario, there’s a scenario that says the aircraft program gets canceled.”
And in a continuation of Boeing’s tone deafness, the manufacturer is acting as if nothing has happened and it can continue down the path of further weakening FAA oversight and streamlining, as in reducing, its safety checks by relying more on computer models. From Reuters:
Boeing Co engineers are reducing the scope and duration of certain costly physical tests used to certify the planemaker’s new aircraft, according to industry sources and regulatory officials.
Several Boeing 777X aircraft are seen in various stages of production during a media tour of the Boeing 777X at the Boeing production facility in Everett, Washington, U.S., February 27, 2019. REUTERS/Lindsey Wasson
But the strategy could be at risk if regulators and U.S. lawmakers probing two deadly Boeing plane crashes require even more rigorous safety tests before certifying new aircraft as passenger-worthy.
As Boeing kicks off the year-long flight testing process on its new 777X, its engineers will cut hours off airborne testing by using computer models to simulate flight conditions, and then present the results to the U.S. Federal Aviation Administration (FAA) as part of the basis for certification, according to two people with direct knowledge of the strategy….
For Boeing’s proposed twin-aisle jetliner, known internally as NMA, Boeing’s Test & Evaluation group is developing the technology to replace costly and labor intensive physical safety tests used for decades – such as using machines to bend the wings to extreme angles and shaking the fuselage until it cracks – with computer modeling, according to three people with knowledge of the matter, including an FAA official.
Boeing really seems not to get how much trouble it’s in. The last thing it needs to be moving forward right now are changes in its certification process when it just had a massive de facto self certification failure. Recall an important Seattle Times story that described how the FAA greatly weakened supervision in 2004, and how over time, Boeing became aggressive in exploiting the new system.
It’s stunning that Boeing seems not even to understand that the multiple failures of its MCAS software (plane trimming down more aggressively than pilots could reverse, resulting quickly in a 40 degree nose down; barmy decision to rely on only one angle of attack sensor) resulted at a minimum from the software developers not understanding the engineering issues remotely well enough to devise sound remedies. For the purposes of this post, it doesn’t matter precisely how that came about, merely that it’s evident at Boeing that the the coders don’t understand how planes work remotely well enough to put their hands up and say “Does that make sense?” And it also suggests that the coders were given bad specs, which points to breakdowns on the engineering side.
Moreover, with it having become obvious that Boeing has captured the FAA, if Boeing were to succeed in getting the FAA to agree to reduce its physical testing any time soon, it would give foreign regulators an excuse to insist on doing their own certifications, not just for the return of the Max 737, but going forward. Having to get approval from multiple regulators (even if they were “certification lite” on specific issues) would put Boeing at a huge disadvantage to manufacturers who would face only one certification because their regulators were still trusted. As Marshall Auerback said via e-mail, ” I’m sure they are popping champagne at Airbus today.”
Consider this corroborating evidence from the Boeing employee on KIRO:
Stuart argues that Boeing CEO Dennis Muilenburg inherited a range of issues from previous leadership. These issues include bad relationships with suppliers. He said previous leadership “despised engineers.” And he says he would hate to see Muilenburg go. Still, he doesn’t not describe a pleasant environment at work.
The way management kind of works is we never really know who our managers are sometimes,” he said. “We get shuffled around so much that our job codes, our job titles, everything changes. Because they are trying to make progress.
Now in fairness, Boeing does have a point. The physical tests are so crude and costly that rather than run them enough times to know where the tolerances are, engineers would build in a large margin of safety so as to not have to run tests a ton of times. As software developer and pilot Gregory Travis explained by e-mail:
It’s much cheaper to use simulators to train pilots, to use computers to analyze structures, etc. The proponents of simulation don’t openly talk about the economic benefits but they will argue, with some justification, that the computer allows a much higher frequency of testing. For instance, you can simulate a dozen engine failures at takeoff in a simulator an hour whereas you’d be hard pressed to practice two real engine failures an hour in a real airplane.
Same thing with fatigue cycles — a computer can simulate decades of cycles on an airframe in a few hours of CPU time whereas doing so with real airframe can take years (as it did for the Comet airliner)
The REAL payoff from computer analysis and simulation is the opportunity to value engineer the crap out of the airframe. By that, I mean to say to get the structure as close to the certification requirements (for strength, etc.) as possible without violating them.
For example, most commercial airliners must be shown to withstand a positive G-factor of 2.5 with a safety margin of 1.5. Meaning the airframe must be able to tolerate 2.5 Gs without any damage and it must be able to go to 3.75 Gs (2.5 times 1.5) before anything starts to break.
In the old days, engineers would just kind of guess at how strong they needed to make certain structures, like the wings, in order to meet those requirements. And then they would test their assumptions by putting the wings in a big tool that bent the wing until the wing broke, hopefully well in certifiable territory.
Now you can imagine that building a wing and the jig to bend it until it breaks is a terribly expensive thing to do. Why not have a computer:
a) Spit out the numbers that tell you EXACTLY how thick the wing spar needs to be so that it won’t fail at 3.75 Gs but will fail at 3.76 Gs — meaning it meets the certification requirements and not an inch more.
b) Self-certify that it’s calculations (the computer’s) are correct.
Planes like the venerable DC-3 were significantly “overbuilt” with regard to the certification requirements and regulations because the engineers at the time had no easy way of knowing how close they were to the requirements. And what they could do involved destructive testing (breaking wings, drop tests, etc.) that were hugely expensive. So they had a habit of coming up with numbers and then adding a generous amount of “Jesus Padding” to them. For example, if they thought that the wing spar needed to be 1/4” thick to satisfy the requirements (much less survive what they hoped would be only ONE drop test*), they’d make it a 3/8”th thick instead.
This added weight and expense to each aircraft produced. But at the insurance (for the engineer’s career) that it was going to be strong enough not only to pass the destructive tests but to not come apart in flight at loads below the certification criteria.
Modern computer modeling now tells them that the wing spar needs to be exactly 3/16”th thick to (barely) meet the certification criteria. And there’s no necessity to “drop test” the airplane to make sure that’s right. The computer is never wrong :0-(
* Drop tests are just one of many possibly destructive tests. It involves literally dropping the entire aircraft from 9 feet up in the air onto the ground and ensuring that nothing breaks. The engineers want the first drop test to go well so that they don’t have to do any more. Thus there is a lot of encouragement to overbuild things like the wing spars, the landing gear (onto which the airplane is dropped), etc. If you don’t have to worry about failing the drop test (because the computer is going to simulate it with numbers the computer comes up with) that incentive disappears.
Drop tests are of course not the only test. There are others like the bend-the-wing-until-it-snaps test. The manufactures hate these tests because they are so expensive. Thus the HUGE impetus to be able to replace physical tests with computer simulations and call it a day.
That’s what we’re moving to. The end of the “overbuilt” aircraft. Because overbuilt isn’t efficient.
But the simulations aren’t always what they are cracked up to be. Travis points out:
The problems with the 737 MAX’s aerodynamics didn’t fully manifest themselves until they actually flew the plane. Initial wind tunnel testing and computer simulation showed that the pitch up problem wasn’t bad. It was only when an actual test pilot flew the actual plane that they realized how significant it was.
And that “inefficient” overengineering has saved lives. For instance:
China Airlines 006 comes to mind. This was a 747 that experienced engine failure in clouds. Because the pilot did not counteract the failed engine with sufficient rudder input, the aircraft rolled, inverted in the clouds, and passed well out of its never exceed speed. During the recovery, the airplane experienced G-loads in excess of 5 Gs — well beyond the certification criteria requirements.
The aircraft managed to land at San Francisco where several critically injured passengers were treated. On inspection, most of the left elevator was found to have dynamically disassembled (meaning it was gone) and the wings themselves were both bent upwards by two inches.
Nevertheless, the airplane was restored (the bent wings were left as-is) and returned to revenue service for another 12 years.
If the structure had been designed and built to only meet the certification requirements (i.e. by a computer), as opposed to being overbuilt, I would venture to say the wings would have come off.
Boeing is still relentlessly out to squeeze fatter margins out of its airplanes when it ought to get its house back in order on its commitment to safety, which is in turn the result of a commitment to good engineering. It’s becoming clear that that will take a considerable effort given how engineering is no longer valued and worse, people are shunted about to make which makes knowledge transfer even harder.
No one wants to believe that national champions can go into a tailspin, but Boeing is testing that assumption awfully hard. If it doesn’t get its priorities sorted out soon, expect its dive to accelerate.