A Routine Bayesian Assessment and Modification of Prediction

The author previously introduced an approach to the modification of mechanical component life predictions by utilization of Bayes Theorem in its discrete formulation. This paper shows how these previous ideas have been extended and developed into the production of a computer program BAMBI (Bayesian, Analysis Modified By Inspection) for the routine assessment and modification of belief.

Three important problems and solutions are discussed in depth:

1. (1)
   The introduction of experience from operational exposure when there had been either a very high or very low prior belief of success.
2. (2)
   Determination of sample size limitations.
3. (3)
   Concurrent evaluation of experience from different test times.

It is concluded that a computer program such as BAMBI can be extremely useful for evaluating an original design intent in the light of subsequent test and field experience. Further, the availability of such a routine Bayesian analytical technique in Gas Turbine engine design provides one example of Assurance Technology spin-off.