Probabilistic assessment of a reactor pressure vessel subjected to pressurized thermal shocks by using crack distributions

• Probabilistic methods are used to analyze a reactor pressure vessel.
• Crack distribution data from the decommissioned plants, Shoreham and PVRUF is used.
• Weld type, size and its manufacturing process are also considered.
• Embedded and surface short cracks result in the highest probability for failure.

Probabilistic methods are used to analyze a reactor pressure vessel (RPV) subjected to pressurized thermal shocks (PTSs) initiated by a small loss-of-coolant accident (SLOCA) and a medium loss-of-coolant accident (MLOCA). The FAVOR code is applied to calculate the probabilities for crack initiation and failure by considering crack distributions based on cracks observed in the Shoreham and PVRUF RPVs in the U.S. The crack parameters, i.e. crack density, depth, aspect ratio, orientation and location are assumed as random variables following different distributions. The Vflaw code is used to generate FAVOR input files for the crack distribution data from the decommissioned plants. Weld type, size and its manufacturing process are also considered in the calculation.In this paper it is shown that the calculated failure probability of the RPV subjected to the SLOCA is higher than that subjected to the MLOCA due to different loading. The failure probabilities are compared with those based on a different crack assumption. Among the analyzed cracks, the embedded crack with a depth of 6.83 mm and surface crack with a depth of 5.13 mm result in the highest probability for failure. Maximum stress intensity factors of the simulated cracks range from 36 MPa m0.5 to 91 MPa m0.5 for the MLOCA and from 30 to 41 MPa m0.5 for the SLOCA, respectively. We conclude that considering the observed crack distribution in probabilistic PTS analyses may lead to higher failure probabilities than by assuming cracks of specific size.