Two findings now pointed toward the crossfeed-valve assembly, but a question remained. Why was the valve opening every eight seconds in the first place? A butterfly valve did not repeatedly actuate without a command.
Faith asked permission to inspect the valve assembly physically. Dr. Sutter approved.
Faith pulled on nitrile gloves, took a penlight from Elton's tool roll, and reached into the access panel. She did not begin with the valve. She began with the wiring harness that carried the position-sensor signal back toward the engine-control computer.
She ran her fingers along the harness slowly, inch by inch, following it from the valve housing toward the bracket that should have secured it against the nacelle structure.
Then she stopped.
"Neil, quarter-inch-drive torque wrench."
Neil handed it over.
Faith placed the wrench on the bracket's mounting bolt and turned carefully until the tool engaged. She checked the digital readout.
"Three foot-pounds."
She held the wrench so the room could see. "Specification for this bracket is sixteen foot-pounds. This bolt is barely more than finger-tight. Somebody installed it and never torqued it properly."
The room remained silent because the importance was not obvious yet.
Faith explained. "This bracket keeps the wiring harness fixed in position. At sixteen foot-pounds, the harness stays locked down and the sensor line remains stable. At three, the harness has enough slack to move under vibration. Every time it shifts, electrical noise enters the position-sensor circuit."
She pointed toward the data traces. "The engine computer interprets that noise as a change requiring fuel balancing. It opens the crossfeed valve. The harness shifts back, the noise disappears, and the computer closes the valve. Then the vibration repeats the cycle about eight seconds later."
The full failure chain became clear.
One loose bolt allowed the wiring harness to move. The movement created false sensor noise. The computer opened the crossfeed valve unnecessarily. The left feed line briefly lost pressure.
Repeated cycling overheated the O-ring. The seal degraded, introducing trace contaminants. The left engine became progressively unstable. The control systems eventually recognized a flameout condition and shut down both engines.
One bracket bolt. Three foot-pounds instead of sixteen.
That was what had helped ground a $100 million jet for fifty days. It was what fifty mechanics, $8 million in diagnostics, and hundreds of pages of logs had failed to identify because everyone had focused on engines, software, control modules, turbine components, and standard checklists.
Nobody had knelt down and physically tested the secondary bracket.
Nobody had followed the wiring harness with their fingers.
Nobody had listened to the machine the way Faith did.