Engineers in the manufacturing industries have used accelerated test (AT) experiments for many decades. The purpose of AT experiments is to acquire reliability information quickly. Test units of a material, component, subsystem or entire systems are subjected to higher-than-usual levels of one or more accelerating variables such as temperature or stress. Then the AT results are used to predict life of the units at use conditions. The extrapolation is typically justified (correctly or incorrectly) on the basis of physically motivated models or a combination of empirical model fitting with a sufficient amount of previous experience in testing similar units. The need to extrapolate in both time and the accelerating variables generally necessitates the use of fully parametric models. Statisticians have made important contributions in the development of appropriate stochastic models for AT data [typically a distribution for the response and regression relationships between the parameters of this distribution and the accelerating variable(s)], statistical methods for AT planning (choice of accelerating variable levels and allocation of available test units to those levels) and methods of estimation of suitable reliability metrics. This paper provides a review of many of the AT models that have been used successfully in this area.
Today's manufacturers face strong pressure to develop new, higher-technology products in record time, while improving productivity, product field reliability and overall quality. This has motivated the development of methods like concurrent engineering and encouraged wider use of designed experiments for product and process improvement. The requirements for higher reliability have increased the need for more up-front testing of materials, components and systems. This is in line with the modern quality philosophy for producing high-reliability products: achieve high reliability by improving the design and manufacturing processes; move away from reliance on inspection (or screening) to achieve high reliability, as described in Meeker and Hamada (1995) and Meeker and Escobar (2004).
Estimating the failure-time distribution or longterm performance of components of high-reliability products is particularly difficult. Most modern products are designed to operate without failure for years, decades or longer. Thus few units will fail or degrade appreciably in a test of practical length at normal use conditions. For example, the design and construction of a communications satellite may allow only eight months to test components that are expected to be in service for 10 or 15 years. For such applications, Accelerated Tests (ATs) are used in manufacturing industries to assess or demonstrate component and subsystem reliability, to certify components, to detect failure modes so that they can be corrected, to compare different manufacturers, and so forth. ATs have become increasingly important because of rapidly changing technologies, more complicated products with more components, higher customer expectations for better reliability and the need for rapid product development. There are difficult practical and statistical issues involved in accelerating the life of a complicated product that can fail in different ways. Generally, information from tests at high levels of one or more accelerating variables (e.g., use rate, temperature, voltage or pressure) is extrapolated, through a physically reasonable statistical model, to obtain estimates of life or long-term performance at lower, normal levels of the accelerating variable(s).
Statisticians in manufacturing industries are often asked to become involved in planning or analyzing data from accelerated tests. At first glance, the statistics of accelerated testing appears to involve little more than regression analysis, perhaps with a few complicating factors, such as censored data. The very nature of ATs, however, always requires extrapolation in the accelerating variable(s) and often requires extrapolation in time. This implies critical importance of model choice. Relying on the common statistical practice of fitting curves to data can result in an inadequate model or even nonsense results. Statisticians working on AT programs need to be aware of general principles of AT modeling and current best practices.
The purpose of this review paper is to outline some of the basic ideas behind accelerated testing and especially to review currently used AT modeling practice and to describe the most commonly used AT models. In our concluding remarks we make explicit suggestions about the potential contributions that statisticians should be making to the development of AT models and methods. We illustrate the use of the different models with a series of examples from the literature and our own experiences.
Within the reliability discipline, the term “accelerated test” is used to describe two completely different kinds of useful, important tests that have completely different purposes. To distinguish between these, the terms “quantitative accelerated tests” (Qua-nAT) and “qualitative accelerated tests” (QualAT) are sometimes used.
A QuanAT tests units at combinations of higherthan-usual levels of certain accelerating variables. The purpose of a QuanAT is to obtain information about the failure-time distribution or degradation distribution at specified “use” levels of these variables. Generally failure modes of interest are known ahead of time, and there is some knowledge available that describes the relationship between the failure mechanism and the accelerating variables (either based on physical/chemical theory or large amounts of previous experience with similar tests) that can be used to identify a model that can be used to justify the extrapolation. In this paper, we describe models for QuanATs.
A QualAT tests units at higher-than-usual combinations of variables like temperature cycling and vibration. Specific names of such tests include HALT (for highly accelerated life tests), STRIFE (stresslife) and EST (environmental stress testing). The purpose of such tests is to identify product weaknesses caused by flaws in the product’s design or manufacturing process. Nelson (1990, pages 37-39) describes such tests as “elephant tests” and outlines some important issues related to Qu
This content is AI-processed based on open access ArXiv data.