PATTERN RECOGNITION MODEL TO ESTIMATE INTERGRANULAR STRESSCORROSION CRACKING (IGSCC) AT CREVICES AND PIT SITES OF 304 SS INBWRS ENVIRONMENTS

摘要

Many publications have shown that crack growth rates (CGR) due to intergranular stress corrosion crackingrn(IGSCC) of metals is dependent on many parameters related to the manufacturing process of the steel and thernenvironment to which the steel is exposed. Those parameters include, but are not restricted to, the concentration ofrnchloride, fluoride, nitrates, and sulfates, pH, fluid velocity, electrochemical potential (ECP), electrolyte conductivity,rnstress and sensitization applied to the steel during its production and use. It is not well established how combinations ofrneach of these parameters impact the CGR. Many different models and beliefs have been published, resulting inrnpredictions that sometimes disagree with experimental observations. To some extent, the models are the closest to thernnature of IGSCC, however, there is not a model that fully describes the entire range of observations, due to the difficultyrnof the problem. Among the models, the Fracture Environment Model, developed by Macdonald et al., is the mostrnphysico-chemical model, accounting for experimental observations in a wide range of environments or ECPs.rnIn this work, we collected experimental data on BWR environments and designed a data mining patternrnrecognition model to learn from that data. The model was used to generate CGR estimations as a function of ECP on arnBWR environment. The results of the predictive model were compared to the Fracture Environment Modelrnpredictions. The results from those two models are very close to the experimental observations of the arearncorresponding to creep and IGSCC controlled by diffusion. At more negative ECPs than the potential corresponding torncreep, the pattern recognition predicts an increase of CGR with decreasing ECP, while the Fracture Environment Modelrnpredicts the opposite. The results of this comparison confirm that the pattern recognition model covers 3 phenomena:rnhydrogen embrittlement at very negative ECP, creep at intermediate ECP, and IGSCC at more positive ECP. ThernFracture Environment Model only covers the two later phenomena.rnThe pattern recognition model does not bring any physical chemical understanding of the problem. However, itrnis useful for smart extrapolation and, to a certain extent, can be used to delineate the range of operation of the physicalrnchemical model or to account for different phenomena needed to fully describe the IGSCC problem.