Using statistical analysis to understand creep modelling error and the ramifications for Solar Tower receivers.

Concentrated Solar Power (CSP) receivers, especially those adopted in Solar Tower (ST) configurations, operate at high temperatures and low stresses making them susceptible to creep damage. Creep modelling used to predict components' life and therefore the accuracy of creep modelling can significantly impact life estimation, prediction of failures, and ultimately the design. The accuracy of creep modelling is highly variable and depends on the choice of model, the choice of fitting functions, and sometimes depends on 'judgement' decisions by the modeller. In this article the Larson-Miller and Wilshire creep models are fitted to three common power generation materials (Grade 22, SS316 and Inconel 740H) with differently sized datasets. Region splitting via activation energy is also examined. The fits are examined for accuracy and bias for stress, temperature and rupture time by a statistical process of bootstrapping to provide a better understanding of the impact of the creep model and dataset size on the design and lifing of CSP receivers. • Considers the impact of dataset size on creep model accuracy and biases. • Examines model accuracy w.r.t. rupture time, stress and temperature. • First paper to use boostrapping to examine the stability and biases of creep. • Traditional creep models and region splitting examined. • Provides authors with awareness of the impact of creep model selection. [ABSTRACT FROM AUTHOR]