Upper Shelf Energy Prediction Model for Irradiated Reactor Vessel Steels

Current USA regulations in 10 CFR 50, Appendices G & H ensure adequate fracture toughness and provide for the monitoring of radiation embrittlement of the ferritic components of the reactor pressure vessel (RV). Regulatory Guide (RG) 1.99, Rev. 2 provides guidance on acceptable methods for predicting the effects of neutron irradiation in order to meet the requirements of Appendix G. Specifically, RG 1.99, Rev. 2 provides an embrittlement prediction model for Charpy transition temperature shift (TTS) and a prediction model for decreased Charpy upper shelf energy (USE). The prediction model for USE decrease has remained unchanged since introduction of RG 1.99 in 1975. The objective of this study is to present new USE prediction model(s) developed using an international light water reactor database similar to the effort behind the recently-updated ASTM E900-15 TTS prediction model. A database of ASME and similar specification USE decrease information was developed from USA and select international light water reactor surveillance capsule data, including the latest surveillance capsule fluence, irradiation temperature, material chemistry and other information. The USE database has more than 1,500 USE change measurements of irradiated RV steels. Several best estimation models to predict irradiated USE of materials were developed based on data fitting. Two types of best estimation models were investigated; one model type uses the ASTM E900-15 predicted TTS as a primary input parameter, while the other does not, so that a USE prediction could be made independently of the ASTM E900-15 TTS prediction. By using the ASTM E900-15 TTS as a primary input, the models of the first type implicitly considered the embrittlement mechanisms of matrix damage and copper rich precipitation. In the non-TTS models, the effect of copper was expressed by a hyperbolic tangent curve that has both an upper value and lower value in order to consider the effect of copper saturation. Associated standard deviations as a function of predicted USE were also established so that bounding predictions could be made. Bounding models from each type that conservatively predict irradiated USE by bounding at least 95% of the USE decrease data in the database were identified. These bounding models are estimated to have relatively low impact on the number of USA plants that are projected to have RV steels that drop below 50 ft-lbs (68 J) relative to RG 1.99, Rev. 2. Finally, a non-TTS model was selected as the recommended model, because it does not require calculation of TTS by ASTM E900-15 and thus is simpler to implement.