Content area
Components in reusable rocket engines, jet engines, and gas-fired and nuclear power plants are repeatedly exposed to high temperatures and loadings for prolonged periods, making them susceptible to creep-induced damage and/or failure. Creep is a time- and path-dependent failure mechanism that is a leading cause of unexpected failure in hot parts. There is a need to predict the useful life and identify new materials with superior creep properties. To that end, this study focuses on developing advanced models for creep-resistant alloys–an endeavor poised to transition into industrial application.
Creep exhibits uncertainty that spans logarithmic decades and demands hefty safety factors for creep-resistant designs. Knowledge concerning the long-term creep behavior of existing and novel materials is needed to make design decisions; however, only limited creep data is available. A lack of correlation between the legs of material science pyramid (chemistry-processing-microstructure-performance) in the existing life prediction models poses a challenge to predict accurate lifespan of critical components. Hence, an advanced Sine-hyperbolic (Sinh) creep-damage model is developed incorporating the threshold strength. Threshold strength is defined as the lower limit of creep activation below which the creep life is assumed infinite. Next, a probabilistic lifetime prediction model is developed that simulates the sources of uncertainty encountered during a part's lifespan. The probabilistic model requires very little data and enables the reliability-based design of rocket, jet, and power components. Finally, Machine Learning (ML) is coupled with human expert knowledge that offer the opportunity to discover complex relationship that relate the chemistry-processing-microstructure of materials with their performance. Multigene genetic programming (MGGP) with symbolic regression; a biologically inspired ML method, is employed to derive human interpretable creep equations. The optimal creep equation is observed to be a function of stress, temperature, alloy chemistry, and strengthening mechanism within the microstructure. The equation enabled the formulation of an optimized chemistry that exhibit improved rupture strength by order of magnitude. These interpretable equations enable an engineer to design new materials with optimized creep resistance properties. The synergistic interaction between theoretical modeling, probabilistic modeling, and machine learning will foster optimized material design, reduce development-to-deployment timeframe of candidate materials, eliminate design conservatism, and a rapid implementation in industries including energy, power, and aerospace.