Finite element analysis of creep-fatigue-oxidation interactions in 316H stainless steel

Damages caused by the effects of cyclic loading (fatigue) and high temperature (creep and oxidation) have been considered critical and need to be appropriately evaluated. A series of strain‐controlled fatigue and creep‐fatigue tests are performed on P92 at 873 K under oxygen‐containing environment. The creep‐fatigue life prediction results are summarized using models based on strain‐range partition, Manson–Coffin equation and linear damage summation (LDS) rule. Obviously, the models based on the LDS rule show relatively good performance with an error band of ±2.5. In view of the adverse effects of oxidation on creep‐fatigue endurance, this paper further develops a physically‐based oxidation damage equation, which is incorporated into LDS rule for the improvement of life assessment. The predicted and experimental results falling into ±1.5 error band proved the accuracy of the proposed oxidation damage equation in the LDS rule. Additionally, model selection criteria are recommended to evaluate the model prediction capabilities.

Section: Introductionmentioning

confidence: 64%

Section: Introductionmentioning

confidence: 64%

High‐temperature creep‐fatigue‐oxidation behaviors of P92 steel: Evaluation of life prediction models

Wang

Wen

Xia

et al. 2022

“…The larger should be used. 21 Figure 3 illustrates a Design Chart, that is, a plot of probability of creep-fatigue crack initiation in a widely used material in nuclear plants, 316H stainless steel, 22 versus 3T-RDM, obtained based on a benchmark problem and validated by three EDF case studies. 21 For this Design Chart, normal and lognormal distribution types were considered for input variables based on plant history data or on reasonableness of fit in the case of materials data.…”

Section: • Step (B)mentioning

confidence: 99%

A review of probabilistic structural integrity assessment in the nuclear sector and possible future directions

Chavoshi

Booker

Bradford

et al. 2021

There always exists uncertainty and variability in structural integrity assessments arising from lack of knowledge, modeling approximations, or differences between as-manufactured components and as-operated components. This may ultimately impact on reactor lives. Such uncertainties and variabilities can be understood and quantified using probabilistic techniques together with reliability-based acceptance criteria, facilitating the quantification and management of risk. This review article provides a systematic appraisal of the latest worldwide literature and also unpublished reports from EDF UK nuclear plants, giving an overview of the existing knowledge of probabilistic/reliability structural integrity methodologies and tools from a multifaceted stance, including failure modes, problem types, material types, employed codes, correlations, and probability distributions. Structural reliability analysis at different levels is discussed, and pertinent issues on convergence testing, required number of trails, sensitivity analysis, verification of individual analyses, extremely small probabilities, reactor-wide failure probabilities, and digital twin structural integrity monitoring are elaborated.

“…Stainless steel 316H is frequently used in boiler components, for example in pipes and their bifurcations and supports. [1][2][3][4][5][6] The probability of crack initiation in AGR components with different geometries can be predicted by physically-based numerical simulations and Monte Carlo (MC) analyses, see example by Kroese et al 7 The underlying deterministic approach is used to define the performance function in MC probabilistic assessment. Nevertheless, building MC probabilistic models and its Abbreviations: AGR, advanced gas-cooled reactor; AKNN-MCS, active learning-based K-nearest neighbors-Monte Carlo simulation; BRG, Bayesian ridge regression; CART, classification and regression tree; CoV, Coefficient of Variation; DCNNM, distributed-coordinated neural network metamodel; DCTKS, decomposed collaborative time-variant Kriging surrogate; DCWNNR, distributed collaborative wavelet neural network regression; DT, decision tree regression; EDF, Electricité De France; EN, elastic-net regression; GTB, gradient tree boosting regression; HAZ, heataffected zone; HTBASSs, high temperature behavior of austenitic stainless steels; KNN, K-nearest neighbor regression; LCF, low cycle fatigue; LHS, Latin hypercube sampling; LR, ordinary least squares linear regression; LS, Lasso regression; MAE, mean absolute error; MC, Monte Carlo; MLP, multilayer perceptron regression; PDF, probability density function; R 2 , R-square; RF, random forest regression; RG, ridge regression; RMSE, root mean square error; SGD, stochastic gradient descent regression; SVM, support vector machine; SVR, support vector regression; TIG, tungsten inert gas; WSEF, weld strain enhancement factor.…”

Section: Introductionmentioning

confidence: 99%

“…Rigorous and comprehensive probabilistic analyses are required to predict the structural failure of such components, some of which may operate at temperature around 650°C and are exposed to the risk of creep–fatigue deformation and rupture. Stainless steel 316H is frequently used in boiler components, for example in pipes and their bifurcations and supports 1–6 …”

Section: Introductionmentioning

confidence: 99%

Data‐driven prediction of the probability of creep–fatigue crack initiation in 316H stainless steel

Chavoshi

Tagarielli

2022

Self Cite

Stainless steel components in advanced gas‐cooled reactors (AGRs) are susceptible to creep–fatigue cracking at high temperatures. Quantifying the probability of creep–fatigue crack initiation requires probabilistic numerical simulations; these are complex and computationally intensive. Here, we present a data‐driven approach to develop fast probabilistic surrogate models of creep–fatigue crack initiation in 316H stainless steel. We perform a set of Monte Carlo simulations based on the R5V2/3 high temperature assessment procedure and determine the sensitivity of the probability of crack initiation to loads and operating conditions. The data are used to train different supervised machine learning models considering Bayesian hyperparameter optimization. We discuss the relative performance of such models and show that a gradient tree boosting algorithm results in surrogate models with the highest accuracy.