ABSTRACT
To this point we have focused on the design and interpretation of experiments in which no measurable degradation or failures have occurred. For systems where components are mission-critical single points of failure (failure of that component causes failure of the mission), NULL data of this sort are the best kind of data. In many cases, however, failures or degradation are observed even in life and are part of the design cost trade-off in building a system. It is necessary in those cases to be able to analyze the data. In accelerated testing problem there are several kinds of data:
1. Failure time data: For each experimental unit the observable is the time at which it fails.
