| Popis: |
The approach described in this paper provides a method to adjust a predicted reliability to account for risks. The benefit of the method is to increase customer confidence in the adjustment while reducing the adverse effect on the reliability prediction. The method reduces the potential for a false positive adjustment to reliability predictions, which may cause costly failure investigations or rework. Typically when a risk is identified the easiest method to adjust the reliability prediction is to multiply the reliability prediction with a discrete probability. The discrete probability is the result of dividing the number of successes over the number of trials. The discrete adjustment is shown in Equation Number 1. Adjusted Reliability= Predicted Reliability×Successes/Trails (1) Although the adjustment is easy, it is a conservative adjustment that does not account for all information that results would cause a critical failure. The goal of the method is to inform the adjustment much like a Bayesian assessment to ensure all information and consequences are included in the assessment. Although the method includes more information and terms, it will ultimately reduce reliability predictions less than the simplified easy discrete model. The method, much like a risk assessment is broken into two parts; assess the risk consequence and assess the risk likelihood. When assessing the consequence, the risk is broken into an Event Trees (ET) and fault tree used to define the calculation model used to calculate the total risk likelihood of a critical failure. The likelihood is then used to adjust reliability predictions. The ultimate goal of the assessment method is to break the risk into as many AND conditions using non-discrete probability models to reduce the total risk likelihood with limited data. |
