Software. Even in autos it’s getting too complex to predict.
Should be fairly easy to diagnose. If they can pull the valves, or even most of them, and cycle them - do they open/close when signaled to open close? You can find design/manufacturing problems that way, or rule out a hardware failure.
If the valves work, check the drive circuitry and wiring. When commanded, does the signal get there at the proper voltage and with enough current to actuate the valve?
If all that is working then it is a software problem - for some reason the system did not command the right state.
If it were me, (and yes I've designed/coded software for embedded systems) for every bit of external hardware outside the control system - both sensors and actuators - I'd have a very, very simple module that interfaced to them. As in dirt simple, too simple to screw up. Said module would always log any change of state.
That way you'd always be able to go back and figure out if a command was given to the hardware (or not), when, etc. Even though this is rocket science, this isn't "rocket science" - this is computers/software dealing with the outside world 101. The outside world is messy, you have to defend yourself against it at every touchpoint.
If you're not getting the commanded state when you expect, it is your control system. This should be fairly well partitioned into higher level functions and low level functions. "Hey, we're ready for engine startup sequence." - high level. "Engine startup? That means open valves x, y, z, shut vents p, d, q..." - low level response to change of state. In general this logic should be expressed simply and clearly.
If they haven't traced through these kinds of things by now, and don't have a "smoking gun" culprit for the failure, then they have serious problems. It says their software architecture and development standards/practices don't support this kind of analysis. They're probably dealing with a mess of their own making.
Yeah I got to admit I’m surprised they don’t know what did or failed to happen more or less when it happened. I’m shocked if they’re truly scratching their heads mystified this level of engineering. I’d have guessed that the thing would HAVE to audit itself at every level in the way you describe.
The car CO’s seem to just pile on code as they need it. Mopar wound up with a transmission problem that was, it turned out, being caused by…the dome light! It’d be a bitch if this sort of thing is happening with Starliners and such.