Hi @WheelmanPEI,
I think there might be some confusion about the concept of MTBF.
MTBF doesn't need accumulation, as it uses a random variable. It's a bit mathematical to explain, but I can use an analogy to explain it.
Suppose we have a fair coin, and we flip it. The coin can land 50% on its tail, 50% on its head, which is fair.
If I will make a bet, that it will be head the next time I throw it, its an honest 50/50 bet, so far so good.
Now suppose I did flip the coin already 10 times, all resulting in head. I would like to make the same bet, the next will be head. This is still a fair bet, 50/50. As the previous outcomes do not impact the next outcome. (The coin doesn't care, it's still a fair coin)
This feels a bit weird, and I get that. The reason why it feels weird, is because the chances that I would throw 11 heads in a row are really really small. That is, if I pick up a coin now and try to throw heads 11 times in a row.
But, here we make the mistake. Saying: "I will flip 11 times head in a row" is fundamentally different than "I flipped 10 heads already, I will flip it again". See what I'm saying?
The previous outcomes of a random variable does not influence the next outcome of that same random variable.
Back to the beginning, MTBF uses that same concept. Mathematically speaking it translates: "The machine has a x% chance of failing" to a "How many days on average can this machine go without failure?". Or, back to the coin: "What is the chance of me throwing x times head in a row, starting now?".
Just as with the coin, the outcome of the next failure-free flight is not influenced by the previous flight. But we can make a statement about the average length of a "failure-free"-streak. And that statement is the MTBF (Mean Time Between Failures), it converts a percentage of failure, to a comprehensible number. I don't know what "magneto failure 0.0001%" means, but I know what "magneto failure on average once every 5000 hours" means.
That being said:
Increasing the MTBF to 1,000,000 will make the chance of a failure neglectably small
Purposely failing and repairing a component will not make a difference on the chance of failure
There are no logs on the "time without failures", as keeping track of this requires programmers to do more bookkeeping, while the previous outcome does not influence the failure rate.
@Black-Square Please correct me if I'm wrong. I could not let the opportunity pass, to spread the knowledge of my favorite statistics teacher.
P.S. Your car does not use MTBF, but lifetime of components, so please, don't use math to skip on maintenance
