A simple method of short‐term mechanical reliability prediction for ceramics in reactive environments

Abstract

A method is demonstrated for prediction of ceramic reliability limited by fracture in reactive environments. The simplicity of the method relies on the ceramic displaying ideal indentation‐strength behavior and on the applied loading leading to short‐term power‐law crack velocity behavior. Reactive failure strength measurements, treating flaw size as a variable and failure time as a constraint in a “pass‐fail” test, enable reliability, including both intrinsic and contact flaws, to be predicted. Little to no computation is required.

Long‐term ceramic reliability analysis including the crack‐velocity threshold and the “bathtub” curve

“…A subsequent experimental work omitted microstructural effects and showed that constant stressing‐rate loading and long‐term zero‐stress aging in moderately corrosive environments could significantly affect reliability predictions through flaw alteration effects. Another work simplified consideration even further by omitting all effects of microstructure, contact‐flaw alteration, and crack‐velocity thresholds, to demonstrate a method for short‐term reliability prediction in reactive environments for components under constant applied stress containing contact flaws. In the work here, different simplifications are made to arrive at long‐term analytical reliability predictions: Microstructural effects and contact‐flaw stress fields are omitted, components containing simple cracks under constant applied stress are considered, but the crack‐velocity threshold in the reactive environment is included.…”

“…In the work here, different simplifications are made to arrive at long‐term analytical reliability predictions: Microstructural effects and contact‐flaw stress fields are omitted, components containing simple cracks under constant applied stress are considered, but the crack‐velocity threshold in the reactive environment is included. In addition, the works cited made deterministic reliability predictions as only mean responses were considered. Here, probabilistic reliability predictions are made as the distribution of responses is considered.…”

“…In this case, threshold and inert strength proximity effects are small. As noted in Equation 23 and Figure 5C, dT/dC < 0, and hence direct replacement of C by T using Equation 23 in F(C) results in the complementary cumulative distribution function (ccdf) of failure times or the reliability of the population, R(T) = 1 À F(T). [7][8][9] The reliability R(T) gives the probability that a crack randomly selected from population will have a time-to-failure greater than T. The bounds of the reliability are R(T ≤ T P min ) = 1 and R(T ≥ T P max ) = 0.…”