8. Assessing Product Reliability
8.2. Assumptions/Prerequisites
8.2.1. How do you choose an appropriate life distribution model?

8.2.1.1.

Based on failure mode

Life distribution models and physical acceleration models typically only apply at the individual failure mode level

Failure mode data are failure data sorted by types of failures. Root cause analysis must be done on each failure incident in order to characterize them by failure mode. While this may be difficult and costly, it is a key part of any serious effort to understand, model, project and improve component or system reliability. 

The natural place to apply both life distribution models and physical acceleration models is at the failure mode level. Each component failure mode will typically have its own life distribution model. The same is true for acceleration models. For the most part, these models only make sense at the failure mode level, and not at the component or system level. Once each mode (or mechanism) is modeled, the bottom-up approach can be used to build up to the entire component or system. 

In particular, the arguments for choosing a life distribution model described in the next 3 sections apply at the failure mode level only. These are the Extreme value argument, the Multiplicative degradation argument and the Fatigue life (Birnbaum-Saunders) model. 

The distribution-free (Kaplan - Meier) approach can be applied at any level (mode, component, system, etc.).