몬테카를로 시뮬레이션에 의한 Grade 91강의 크리프 균열 성장속도의 확률적 예측

Abstract

Grade 91 steel is used for the structural components for a Gen-IV nuclear reactors, such as a very high temperature reactor (VHTR) and sodium-cooled fast reactor (SFR) due to their high thermal conductivities, low thermal expansion coefficients and excellent irradiation resistances to a void swelling compared to austenitic stainless steel. Since the reactors are designed for up to life span of 60 years, their components are inevitable from the creep crack growth rate (CCGR) due to creep damage at the elevated temperatures. To assure the integrity of the components, the CCGR should be accurately evaluated from reliability viewpoints based on experimental data. In this study, to predict the creep crack growth (CCG) behavior of Grade 91 steel, a series of the tensile, creep, and CCG tests at 600oC was performed. The CCG tests were conducted using 1/2” compact tension (CT) specimens under various applied loads. The CCGR was statistically analyzed in terms of the empirical equation between the CCGR (da/dt) and C* parameter. In addition, the values of the C* parameter and CCGR at a specific load (a low applied load) were predicted using Monte Carlo Simulation (MCS). The results were drawn as follows;

(1) Experimental CCG data of the relationships between da/dt and C* parameter were compared using three methods such as the least square fitting method (LSFM), the mean value method (MVM), and the probabilistic distribution method (PDM). The MVM and PDM were clarified to be reasonable in determination of the CCGR lines.

(2) The values of predicted C* parameter and CCGR in each applied load followed lognormal distribution well compared with normal distribution and Weibull distribution. The MCS using lognormal distribution was demonstrated at a specific load condition of very low stress level.

(3) To predict the CCGR at a specific low-stress level, a new method was proposed using a MCS technique following a statistic distribution of CCGR data. The proposed method will be useful for estimating the CCGR behavior of very low stress level using only several CCG data.