Development and Integration of Light Water Reactor (LWR) Materials Corrosion Degradation Codes into Grizzly

The primary objectives of the proposed research was the development of deterministic, physico-electrochemical models for predicting the accumulation of localized corrosion damage (pitting corrosion, stress corrosion cracking and corrosion fatigue) in the primary coolant circuits of the currently operating fleet of Light Water Reactors (LWRs) and the embedment of the models into the Grizzly code currently being developed at the Idaho National Laboratory as part of their program on nuclear power plant component aging. Localized corrosion in LWR (BWR and PWR) primary coolant circuits (PCC) is primarily an electrochemical phenomenon, augmented by mechanics and microstructure, the rate of which is determined by certain electrochemical properties, such as the electrochemical corrosion potential (ECP), solution conductivity, temperature, pH, flow rate, and the kinetics of the reduction of redox depolarizers (e.g. O<sub>2</sub>, H<sub>2</sub>O<sub>2</sub>, and H<sub>2</sub>) on the surfaces external to the crack, in addition to mechanical loading (stress intensity factor on the crack) and micro-structural/micro-chemical factors (grain size, precipitates, etc). Because the efficient control of environmentally-assisted cracking (EAC) damage accumulation requires the accurate control of these parameters, it is necessary to develop codes that can accurately predict ECP and crack growth rate (CGR) at any point in the primary coolant circuit (PCC) over wide ranges of temperature (25 <sup>o</sup>C to 320 <sup>o</sup>C), pH (6 – 8), ECP (-0.9 V<sub>she</sub> to 0.2 V<sub>she</sub>), solution conductivity, flow rate (1 – 6 m/s), and stress intensity factor (5 MPa.m<sup>1/2</sup> – 50 MPa.m<sup>1/2</sup>). Knowledge of these parameters, along with suitable damage prediction codes, would allow an operator to predict the accumulated damage in PCC as a function of the future operating history of the reactor (the “corrosion evolutionary path,” CEP). In performing this study, we have further developed our previous prediction codes in the form of BWR_MASTER and PWR_MASTER by upgrading all sub-models for calculating radiolytic species concentration, ECP, and crack growth rate (CGR) as a function of reactor operating variables (power, radiation density, temperature, location in the PCC, flow velocity, coolant pH and conductivity, and operating history). The codes have been used to predict the accumulation of IGSCC damage in Type 304 SS in the core shroud of a BWR over a fuel cycle and to estimate the damage at the same location during start-up, considering transients in reactor power, temperature, and conductivity (due to hide-out/hide-out return). The predicted damage is in good agreement with plant observation. Regarding PWRs, we have developed two new models for calculating CGR in mill-annealed, Alloy 600, by considerably upgrading the MPM (mixed potential model for estimating the ECP), the Coupled 2 Environment Fracture Model (CEFM) that were originally developed to predict ECP and CGR in sensitized stainless steels, to predict CGR in nickel-base alloys, as well as developing a micro- void pressurization model for also estimating CGR in MA Alloy 600. Both CGR models yield CGRs that are in excellent agreement with the experiment. We have also successfully developed crack initiation models for both stainless steels and mill-annealed, Alloy 600. That for stainless steels is based on a pit being the initiation site, whereas in that for mill-annealed, Alloy 600 postulates that initiation occurs at emergent grain boundaries that have been wedged open by internal oxidation. Experiments show that the crack initiation time (CIT) is highly distributed and is a seemingly, random quantity that exhibits trends with various system properties, such as surface stress, hardness, yield strength, etc. Although not identified in the SoW, we have developed a theoretical framework for describing the distributions in the CIT by assuming a normal distribution in the number of initiation sites concerning surface stress. It is well-known from experiments that CGRs are highly distributed quantities also, with almost all systems exhibiting log-normal distributions. In work outside of that proposed, we used the ANN and CEFM to confirm that a log-normal distribution in CGR is expected theoretically if the independent variables are normally distributed. This work essentially defines the accuracy that one might expect in the calculated CGR due to randomness in the independent variables. To provide fundamental, input data for the various models, we have made extensive measurement of the kinetic parameters (exchange current density and Tafel constants) for the oxygen electrode reaction (OER) and the hydrogen electrode reaction (OER) on stainless steels (Types 304 and 316) and nickel-base alloys (Alloys 600 and 690) in reactor coolant at temperatures to 300<sup> o</sup>C and as a function of pH, [O<sub>2</sub>], and [H<sub>2</sub>]. We have also optimized the Point Defect Model for passivity and passivity breakdown on measured electrochemical impedance spectroscopic (EIS) data for all alloys studied in