Why Current Types of Accelerated Stress Testing Cannot Help to Accurately Predict Reliability and Durability? 2011-01-0800

This paper demonstrates the results of the analysis of the current practical situation in product reliability and durability as well as accelerated stress testing development. High stress testing is now the basic source for obtaining initial information to provide a prediction of a product's reliability and durability. This paper shows that this testing cannot offer information for the accurate prediction of reliability and durability, because the product degradation process during the testing differs from the product degradation process during the actual field situation. As a result, the time to failures also differs. It demonstrates that this difference depends on: a) loading during high stress testing is often higher than the maximum loading in the normal use conditions of the test subject; b) the simulation of only maximum loadings from the normal use conditions is not an accurate simulation of this loading process; c) the loading of a mobile product under normal use conditions has a random character, but stress testing is often used for testing methods of constant stress, step stress, or cycling stress that does not simulate this random character.

The above is changing the physics-of-degradation process of the product during testing in comparison with this process in normal use conditions. The simulation of only 1-3 stress factors from multiple field input influences during stress testing is another reason for causing the changes within the physics-of-degradation process in stress testing. The paper describes also other reasons for the above changes. Finally, the paper demonstrates how one can improve the above negative aspects of the current situation in stress testing.