Accelerated Life Testing for Reliability Validation

Abstract

Accelerated life testing of products, components and materials is to get information quickly on specific lives, life distributions, failure rates, mean lives, and reliabilities. Accelerated testing is achieved by subjecting the test units to application and operation stress levels that are more severe than the stress levels applied during normal use in order to shorten their lives or their times to failure. If the results can be extrapolated to the stress levels encountered during normal use, they yield estimates of the lives and reliabilities under use stresses. Such tests provide savings in time and expense, since for many products, components, and materials, life under use conditions is so long that testing under those conditions is not timely or economically feasible.Accelerated test conditions are typically produced by testing units at higher levels of temperature, voltage, wattage, pressure, vibration amplitude, frequency, cycling rate, loads, humidity, etc. (or some combination thereof) that are encountered under use conditions. Life data obtained from units tested at different elevated stresses are analyzed and extrapolated to obtain an estimate of the life and reliability at stress levels of typical use. The use of such accelerating variables for a specific product or material is dictated by experienced engineering practice. For example, when choosing the stress level at which a test will be performed, one must be cautious not to exceed the material limitations under the specific stress and to avoid introducing failure modes that would not be observed under use conditions.The following accelerated test models were covered by Kececioglu [1]: (1) Arrhenius; (2) Eyring; (3) inverse power law; (4) combination, when two or more accelerated stresses are applied, such as combinations of temperature and voltage; (5) generalized Eyring; (6) Bazovsky; and (7) Weibull stress life. This chapter covers the following additional accelerated testing models and their practical applications:1. Log-log stress-life 2. Overload-stress 3. Combined-stress percent-life 4. Deterioration monitoring 5.Step-stressThe log-log stress-life model uses a log-log plot of stress-life data to develop curves for specific reliabilities at various functional stress levels. The overload-stress model presents a unique equation for reliability, which combines the Weibull and the log-log stress-life models' results to determine the reliability of units at their use-stress level, having determined their reliability at a higher stress level.The combined-stress percent-life model analyzes the results when one unit is tested to failure at an elevated stress, and another unit is tested -first for a fraction of its life at the use-stress level, and subsequently at the high stress level used for the first unit -until this second unit fails. A first-approximation straight line is drawn to relate stress level and associated life, from which the use-stress life is determined. The deterioration-monitoring model involves monitoring, at both use and accelerat...

Failure time distributions for complex equipment

“…For details of the proof, which is quite technical, see Drenick 4 or chapter 13 of Kececioglu. 16 For additional commentary, see section 2.3 of Barlow and Proschan 10 and section 2.3 of Epstein and Weissman. 12 Figure 2 shows the celebrated "bathtub curve," a standard reliability engineering rule of thumb for complex components and systems.…”

Reliability Reallocation for Fuel Cell Vehicles Based on Genetic Algorithm

“…In the nonlinear optimization programming function of reliability allocation, the mathematical formulations are ensured based on certain parameters that must be supplied by engineers and designers. 25 The nonlinear optimization problem of reliability allocation is formulated as follows: Total FCE 22 36 29 17 104 BMS 1 21 29 0 51 EA 0 8 20 44 72 EDS 0 2 4 9 15 Total of entire FCV 23 67 82 70 242 FCE, fuel cell engine; BMS, battery and management system; EA, electrical appliance; EDS, electric driving system. In Equation (1), the formulation is used to achieve a minimum total vehicle cost, subject to R G , which is the lower limit on the overall vehicle reliability.…”

“…The feasibility of increasing FCV subsystem reliability can be obtained and quantified by weighing the factors that depend on FCV subsystem complexity, the advancement of technology, and the operational profile and technical level. 25,26 a i is the coefficient of subsystem cost function and adjusts the cost difference among subsystems. b i is the constant of the subsystem cost function and modifies the subsystem cost function.…”

Statistical approach in power cables diagnostic data analysis

“…1-σ, 2-σ or 3-σ can be calculated. In probability and stochastic process theory, a statistical distribution can be described by a number of different formulas [12][13]. Each formula is related to the other.…”