Airlines scour the world for scarce 737 Max simulators | Hacker News Reader

Airlines scour the world for scarce 737 Max simulators

reuters.com

98 pointshhs58 comments

There’s some weird numbers in the article, which may either be a lost in translation situation or more evidence that airlines and Boeing are once again playing fast and loose with their plans.

For example, Southwest is apparently claiming they will be able to train their 10000 737 pilots on the simulator in 30 days. They have 3 simulators (still going through certification) and have ordered another 3 for late 2020.

Assuming they have all 6 on day one, that means each simulator will be training approximately 1600 pilots. Divided by 30 days, gives you 50 pilots a day? In other words less than 30 minutes a pilot assuming extreme efficiency. And training 24/7.

It's always two at a time, you train with a normal counterpart (captain or FO) next to you. But 1 hour isn't very long, although I guess they only need to do some MAX specific maneuvers to deal with the stall characteristics and MCAS. Not a full sim-check.

That's worrisome because by then the pilots will not have flown that particular aircraft for over a year. You'd expect some training time to be mandatory after such a long lapse.

Since they had minimal (iPad only) training time coming from the NG to the Max before, I too am curious to see on how this turns out. I assume that at least some authorities will take that into account.

Everything else about it is still a 737. The training is also not limited to just sim time, there should be extensive teaching on integration e.g. checklists and system design.

Also, when you're in the sim, it's not like its just take off, fly around, something fails, you land, kind of thing.. It's a situation they are provided and then you are observed as to how you handle that situation. In this case, I'm sure it will focus on random failures along this new MCAS path (what i'm saying is the MCAS system will be fully failable as a sim option). Your manager (a senior pilot) then decides, on all information not just the sim, if you are airworthy or need another go around of training.

> Everything else about it is still a 737.

Déjà vu

Not exactly wrong, though. It's the engine characteristics and MCAS system that make the difference, and those don't affect most regimes.

(Or in MCAS' case, aren't supposed to.)

I was surprised by that 10,000 pilot number. It seems huge until you look it up and realize they have 752 planes. Each plane averages 6 flights/day. So they have about 13 pilots per plane and an average of 2 pilots per flight. *Of course, many of those flights are flown by the same pilots which gives them coverage for the various shifts. The numbers are probably a little more off now too since some of their fleet is idle. Either way, 10,000 pilots doesn't even sound like enough after roughing up the numbers.

13 pilots per plane sounds like more than enough.

Each plane needs two pilots at a time. Each plane is manned maybe 105 hours a week (say 17 hours a day on average, less 5% of the time in maintenance), so it requires 210 pilot-hours per week. A full time employee works about 35 hours a week after sick leave and holidays. So you need about six full-time pilots per plane.

Pilots spend some time in training and admin outside of the plane, but it could hardly amount to enough to double the staffing requirements, especially for Southwest who are famously lean and efficient. I'd guess a significant number of the 10,000 are on standby or get limited hours (but of course need to be trained to the same level as full-timers).

You need 1 crew per flight, but really 2 crews per day per plane. Pilots work about half the days, so 4 crews per plane. That gets you to 8 pilots per plane, which is roughly 6k pilots without redundancy or pilots working very limited segments.

Pilots are limited to 100 hours flight duty per 28 days and 190 hours total duty time. SW had a fairly lower flight rate a few years back, something like 60 hours if I remember correctly.

Southwest only has 34 737 MAX planes. There is no way that all 10000 pilots will be trained in 30 days. Most likely, a couple hundred of those will be trained up front so that enough pilots are trained when the plane is returned to service. Airlines segregate pilots all the time based on ratings for specific planes so this should not present a logistics problem.

Southwest has another 200 MAX planes on order that have yet to be built, enough cushion to certify future pilots.

They will probably not train 10,000 pilots in 30 days. They will be able to train enough pilots in 30 days to cover the flights that will use a Max as equipment. Southwest reports that the max was scheduled for 330 out of 4,000 daily flights (https://www.southwest.com/html/air/737-MAX-8.html?clk=737MAX...). Training 5-10% of their 10,000 pilots in 30 days is much more feasible.

Southwest does (or did) intend to purchase more over time, but they will easily be able to keep pace to have enough trained pilots ready to fly as they replace older equipment.

Interesting that Boeing does no make the simulators directly but through partners. Having a delay in making the simulators might be because the makers of the simulators are having problems with the software replication from the real plane. Wonder what happens when the software on the simulators is in desync with the software on the real plane; or when software on the real plane has a bug that manifests only after 10 hour of continuous operation while the software on the simulator is continuously resetted for each new pilot. On the other hand, Boeing and the airlines should have provided more simulators if their new model is so different that it needs a dedicated simulator. One idea - Boeing should only sell new planes together with simulators - 1 simulator included for each x planes delivered.

> Boeing and the airlines should have provided more simulators if their new model is so different that it needs a dedicated simulator

I highly recommend reading this article:

https://onemileatatime.com/boeing-737-max-lion-air-simulator...

The first crash was with Lion Air. Boeing mocked them for wanting simulators. After the crash, Boeing claimed pilots of Lion Air were at fault. This while saying no simulator training was needed.

>Boeing mocked them for wanting simulators.

These were the conversations of two employees. I wouldn't say that Boeing outright mocked them but, in the processes going on above the employees paygrade, simulators were difficult to get to Lion Air

Lion Air asked, Boeing declined. Of course some person at Lion Air asked and some person at Boeing declined. That doesn't make a difference. These persons where talking for Boeing.

Ultimately the biggest mockery of the Lion Air request was Boeing's refusal of it.

You think because the employees weren't executives and they were talking to eachother it's all ok? What on earth are you talking about?

Boeing employees should be in prison, there shouldn't be bail while awaiting trial. This is a act of willful negligence and deceit and coverup that killed hundreds of people. It's one of the biggest crimes ever in the history of our country.

They're a defense contractor, they can't commit crimes.

Hahahahaha

Out of college I interviewed at a place that makes flight simulators, not at the scale of Boeing likely but I’d imagine at the same level of ‘best practices’.

I was surprised to learn that at least at that place, the actual aircraft software was running in the simulator, and that external sensor inputs to the software were being mocked out.

It was surprising to me because I had never thought of a simulator as simulating the aircraft itself, but rather only the experience of flying it.

So from this, I don’t think it the case that a simulator is having trouble recreating flight characteristics or reimplementing a ‘bug’ or other behavior because my guess is they are not reimplementing much/anything on the software level.

Though your point about bugs only surfacing after many hours of continuous operation could theoretically still occur without further controls, and of course it is hard to make a mock full-fidelity, especially so in the context of a full suite of analog flight sensors/actuators.

Simulators normally already run very close to 24x7. Not strange at all to have a 1am sim slot or 5am or similar at most airlines. So this is going to be chaos if the MAX is reauthorized to fly before everyone is trained.

I wonder how the validation of correspondence between the actual plane (with the unknown/unreleased MCAS related mods) and the simulator behaviour is performed, i.e. who (Boeing, the simulator manufacturers, the FAA, experienced test pilots) actually does what and - ultimately - which kind of "stamp" attests that the simulator is a valid representation of the plane behaviour (and how much time is it needed to perform this step).

Scrounging around for simulators or wondering how close it is to the real thing is moot because the 737 MAX is dead. The CEO is gone and there are more and more safety "glitches" being discovered. Even if they return them to service, who is going to fly on them? You? Sorry, but the cheapening-out on engineering and manufacturing over the years has eventually produced a New Coke, who's side-effect is that it kills the consumer. There's only "classic Coke" as a fix (737 NG), but that's not much better. In fact, Al Jazeera did an exposé in 2010 about the internal whistleblower who was ignored by Boeing management when it was revealed that substandard critical structural components made by subcontractor Ducommun were being crudely constructed by-hand and were grossly out of tolerances, yet Boeing management ordered them installed on customer planes anyhow. 737 NG's (-6xx, -7xx, -8xx, -9xx) have already been involved in hard landings and runway overruns where the fuselages broke apart, killing passengers, when previous similar airframes survived intact but were possibly damaged and needed inspections. These NG's are flying around above your head today, and it's unclear if the next landing or severe turbulence is going to rip the plane apart because it was either poorly engineered or poorly manufactured due to decades of lax "self-regulation"/regulatory capture and corporate greed. Some engineering areas and some planes were made better than others, but it's unnecessarily playing Russian roulette with people's lives because management used "creative" ways to cut corners.

If you want less micromorts, stick to well-maintained older 737's/777's and Airbus.

The capital cost of the 737 Max fleet is likely to be at least $20 billion. Do you think the airlines or Boeing are just going to write that off?

We are going to see normal fares for Max flights and increased fares for those who don't want to fly on Max.

Do you think they won't get bailouts?

To get a bailout they need to act as if it were important.

That's not really Boeing or the airlines call. If the relevant certification authorities refuse to certify it, then it isn't going to fly no matter how much Boeing or the airlines want it too.

It is going to be interesting to see how it goes. That it has taken this long to fix, points to it being, at a minimum a lot harder to fix that first suspected, and it might even be unfixable given the laws and regulation around airtravel.

>>Even if they return them to service, who is going to fly on them? You?

I never understood this. Are you ever in a situation where you have the luxury of chosing a plane you fly on? I fly several times a year, mostly with Ryanair/EasyJet/Jet2 and those are the only airlines that fly to the destinations I need to fly to. It's not like I can chose and say "oh no, Ryanair flies the MAX, so I'll fly Lufthansa instead". There is no choice in that matter at all. If Ryanair replaces their entire fleet with 737 MAX tomorrow, then that's what I will fly - the alternative is either a 30 hour long train journey, or flying with someone else with 1-2 changes for 5x the price.

On the flip side, the two destinations I regularly fly to have three airlines each, two of which use Airbus.

I absolutely do have a choice, and I'm not going to fly Boeing until further notice. Apart from the whole 737 Max thing, Airbus' planes are just much quieter internally.

Does anyone have any reading material on how the simulators work? Do they run the same computer hardware as actual aircraft, or is it all simulated? If it's simulated, how do they avoid the issue of a bug not being present in the simulator?

I don't think the simulator is designed to QA test the airplane. It's only for training pilots, and pilots are only trained on procedures that airlines deem worth training for. This is things like engine failures at inopportune moments, stall recovery, etc. The idea is not to simulate the outcome of strange circumstances, but to get pilots comfortable with procedures. So if there is some bug, it doesn't matter -- if you're training pilots to take off after v1* is called out and an engine fails at that exact moment, the instructor is not looking to see that "hey we applied the brakes and the airplane stopped OK with light rain and 3 knots of wind", they are looking to see that the pilot continues the takeoff anyway and doesn't try to stop on the runway. The pass condition is following procedure, not "well that's not the procedure but it looks like everyone lived anyway".

https://en.wikipedia.org/wiki/V_speeds#V1_definitions

The various data that feeds into training procedures comes from actually flying the real airplane.

Since the planes become more software dependent, the pilots should be trained on procedures to recognize/evaluate/react to software problems/glitches as well. The problems with the B-MAX were actually related to bad software (MCAS). There is probably some delay between how pilots are training and the importance of software in their job.

I think, functionally speaking, a software failure is no different to that of hardware-based control / instrumentation systems (think of all the odd behaviour that could be caused by a short circuit or a bad connection somewhere in the wiring). The difference is that software systems tend to be more complex, and failures are in theory less random and more systematic - it's more a case of stepping on an unlikely combination of inputs (i.e. a bug that occurs in an untested corner-case) than encountering a random component failure which is more likely the case for a hardware system.

Edit: actually I suppose I just made your point

same software as in the airplane. sometimes the same hardware, also.

I asked the same question a week ago and got some great answers: https://news.ycombinator.com/item?id=22032891

TLDR: There are many different types of simulators.

The article says Southwest has 10.000 737 pilots. Let's say all of them will be trained. The article also says that simulator training costs 0.5-1K USD per hour. Let's take the higher figure. So that means 5M USD per hour training for all pilots (two pilots per simulator). Let's then speculate and say that the FAA approved training package is 10 hours (I believe this will turn out to be less than 3).

We're looking at 50M USD cost of training pilots for the MAX, for an airline which ordered almost 300 MAX planes.

That is less than 200K USD per plane. It is depressing that this was being weighed against loss of human life.

There's other costs as well. An untrained pilot can't fly, meaning that the airline is shorthanded. If you don't have enough trained pilots, you have to cancel flights and lose revenue.

They would presumably start simulator training BEFORE the planes arrived, wouldn't they?

Which they now had to do for over a year. So that was some great cost-saving exercise again...

With all the effort Boeing put into making controls the same, I wonder why they couldn't just release a software update for existing 737 simulators that adds the ability to switch to Max characteristics. I get the no new training was a goal of the Max, but in existing simulators would at least be a consolation prize.

The issue they're getting bit by now is that technically, MAX can't be architected the same way as older 737's due to the Flight Computer representing a single point of failure with catastrophic consequences.

Older 737's are fine having two separate Flight Computers with only one in command at a time. MCAS changed that. So they now have to consider significant hardware/software rearchitecture which goes beyond a mere

   If(MAX){
      MCAS();
     } else {
      otherStuff():
     }

For instance, early on in the investigation, pilots had to use the NG simulators with manual fault injection to recreate the experience because the MCAS software simply wasn't there.

https://www.seattletimes.com/business/boeing-aerospace/newly...

The "newly-stringent" tests described are basically 101 level test cases in excluding catastrophic outcomes from a design, where you basically assume the perfect confluence of bit flips occur to ruin your day. If you've done your job right, that shouldn't end up being a problem.

Boeing didn't, therefore, it was a problem.

The title makes it sound like simulators are going extinct

In fact, now that training is going to be needed, new simulators are going to be built, but it takes time. Maybe some 737ng Sims can be converted

By the way, some airlines have their own but several have their crews train in outsourced training facilities (CAE, etc)

what surprises me is that there are even any simulators for this plane, when the manufacturer said such training isn't necessary. apparantly some airlines still put safety above everything.

It's worth it from a financial perspective. If a plane goes down it hurts the brand and public trust significantly. People in general are already afraid of flying, lets not give them more reasons :)

>> The 737 MAX has been grounded since March 2019 after two fatal crashes and cannot return to service until regulators approve software changes and training plans.

My understanding from the previous wave of news was that two planes crashed because they stalled in mid air because their engines were too large and too far forward and that somehow affected the plane's center of gravity... And the solution is a software update? You can't fix hardware problems with software.

I wonder if this is a trend. First Intel and now Boeing... Shipping defective products and then trying to hack together patches on top.

If you're going to comment in threads like these it would be nice if you familiarized yourself with the material at least at a basic level.

They did not stall because of the engines. The plane would be perfectly flyable without mcas, the plane only behaves differently. But in order to avoid recertification boeing decided to tweak the plane behaviour with mcas so that there is no noticeable difference for the pilots. In hindsight this obviously wasn't done right.

You most definitely can fix _certain_ hardware problems with software.

Reply on news.ycombinator.com