A best-estimate-plus-uncertainty (BEPU) method is being developed for analyzing pressurized thermal shock (PTS) for the Euratom-funded research and training project on long-term operation (LTO) called “Advanced PTS Analysis for LTO” (APAL). The most important thermal-hydraulic phenomena were identified in a phenomena identification and ranking table (PIRT) for a 1300-MWe four-loop pressurizedwater reactor (PWR) of a German design. The accident scenario leading to PTS consists of a 50-cm2 break in the hot leg, leading to a loss-of-coolant accident (LOCA) and a coincident loss of offsite power (LOOP). The availability of high-pressure safety injection into the cold legs was limited to two cold legs. The resulting phenomena associated with water levels in the downcomer above the bottom of the cold leg included the formation of a cold-water layer along the bottom of the cold leg and a cold-water plume in the downcomer. The non-uniformity in the temperature of the water combined with the associated nonuniformity in the heat-transfer coefficients lead to non-uniform cooling of the reactor pressure vessel (RPV) wall, which leads to the thermal stresses in the RPV wall that lead to PTS.
After the important thermal-hydraulic phenomena and initial and boundary conditions were identified, the associated parameters in the computer simulation codes were identified. The important identified plant parameters and initial and boundary conditions included the capacity of the injection pumps and the temperature of the injected water. Because of the break size, the secondary-side cooldown rate also is an important boundary condition during the LOCA. The uncertainty in the plant parameters and initial and boundary conditions were obtained from the range of plant operation data and manufacturer’s specifications for plant equipment. For the important model parameters in the computer code, experimental data were collected, and they were compared with the values from the computer codes. Differences between the experimental data and the computed values were used for developing uncertainties distributions for the model parameters.