Many failure mechanisms can be traced to

underlying degradation processes. Degradation eventually leads to a

weakness that can cause a failure for products. When it is possible

to measure degradation, such data often provide more information

than traditional failure-time data for purposes of assessing and

improving product reliability. For some products, however,

degradation rates at use conditions are so low that appreciable

degradation will not be observed in a test of practical time length.

In such cases, it might be possible to use some accelerating

variables (e.g., temperature, voltage, or pressure) to accelerate

the degradation processes. In today's manufacturing industries,

accelerated destructive degradation tests (ADDTs) are widely used to

obtain timely product reliability information. In designing an

experiment, decisions must be made before data collection, and data

collection is usually restricted by limited resources. Careful test

planning is crucial for efficient use of limited resources: test

time, test units, and test facilities. The basic goal in designing

an experiment is to improve the statistical inference for the

quantities of interest by selecting appropriate test conditions to

minimize or control the variability of the estimator of interest.

Generally, an ADDT plan specifies a set of testing conditions and

the corresponding allocations of test units to each condition. In

this dissertation, we study the test planning methods for designing

accelerated destructive degradation tests from three aspects,

including non-Bayesian and Bayesian methods. First, Chapter 2

presents the non-Bayesian methods for accelerated destructive

degradation test planning when there is only one failure cause for

the testing products. Second, Chapter 3 describes the non-Bayesian

methods for accelerated destructive degradation test planning when

more than one failure cause (sometimes known as competing risks) are

induced for the produces which are tested at high-stress levels of

accelerating variables. Third, Chapter 4 shows the Bayesian methods

for accelerated destructive degradation test planning.