Fly Forward Forecasting

[Randy Pirtle]

I would like to hear how others forecast product removals/failures when only total removals and corresponding flight hours in known. Removals forecasting as well as MTBUR (Mean Time Between Unscheduled Removals) forecast into the future is desired. We call this task Fly-Forword forecasting.

In Aerospace commercial aircraft world, I am very familier with using Weibull models, and a knowledge of aircraft fleet demographics (age, numbers, product mix), to estimate (using either Monte Carlo or deterministic methods) the number of failures to expect in the future months/years.

However, some products and aircraft applications to not afford the amount of information required to support Weibull Analysis and forecasting.

I am being compelled to forecast removals/failures when we only know the number of removals/failures per month and the flight hours accumulated per month. Note- that no flight hours per serialized product removal/failure is available!

I have been using Crow-AMSAA models for this purpose but my rational does not include provisions for aounting for product change introductions.

I would like to hear what others do or strike a dialogue with others similarly interested.

Thanks and Regards

[Pantelis RS]

Randy -- good question --.

Can you make up a simple example (using a generic sample of ‘what data is available’) including what results you want to get out of it?

I know you mentioned a case in the message, but if you can generically define the problem (or part of it) it may generate a better discussion. I know I have a few ideas…