Boeing has been working on long term plans to replace expensive physical testing of aircraft with computer simulations. Those will then be presented to the FAA for sign off.
This is a Pandora’s Box scenario, first it’ll be simple flights and over time, it’ll be entire aircraft from end to end. Once you let them do it, they’ll have no excuse to ever stop.
Simulations are used extensively but are utterly reliant on the available information from real life being coded and processed. It’s a massively intensive process and requires leading edge super computers.
The issue is that unless there is adequate input and every last variable is available to permit processing things can be missed.
As a recent example Boeing admitted it had failed to use MCAS information on 737 flight simulators. If the computer doesn’t have the raw data it can’t process it.
There’s also a growing appreciation, especially in military aircraft simulators, that real world issues simply cannot be constantly planned for or simulated. According to recent reports the F35 design simulators showed no issue with a real world problem: nothing predicted that the stealth paint would peel off and blister at Mach speeds, or that despite every precaution, from other aircraft simulations and earlier problems on the F22, pilot blackout and disorientation would cause a fatal crash as recently in Japan, even with an experienced pilot.
While the examples above are unlikely to affect a civil airliner in the same way, it simply outlines that unless micro detailed levels of data are included in the simulation software, it can’t use it to predict an outcome.
Counterbalancing the argument are that simulated nuclear weapons tests have gone on for years and what could be more complicated? They have little to do with explosive force or radiation – they’re more about the relatively basic way the safety interlocks work, and the detonation mechanics, and they’re a lot more simplistic than many realise in what they set out to achieve. They’re not carrying passengers, or travelling 5,000 miles twice a day with parts that need to last 30 years doing so.
Arguments for computer testing are that it saves time, costs less, can be updated, and there is years of data to input and provide a baseline for a simulation. Yet there is always the possibility that complacency, over reliance and eventually institutional ignorance creeps in.What happens when the simulator designers retire, few people can ever know the full extent of a system like the original designer. Super computers have a habit of being able to adjust their own parameters and any AI that made a mistake in the future – would anyone know?
Humanity has a habit of being lazy and expects things to simply work. What if that stops happening? Who’s going to correct it? Who would even know where to start as the engineers who conducted the physical tests vanish over time?
I understand why Boeing want to do what they are planning, but it needs to stop now. Too many old school processes and practices have gone. That people who write code think they are engineers is a joke, skilled they might be but engineers they are not. Engineers are the men and women who work with physical structures and machinery, and we lose that understanding and comprehension of the physical world at our peril.
The 737Max is a warning, systems fail, human, IT based, organisational. Computers aid improving process, but they will never be mechanical engineers with a lifetime of human experience and basic intuition. And if the day ever comes when they are, I pity us all.