On the scaling of uncertainties in thermal hydraulic system codes

Validation is a key step in the field of thermal hydraulics. System codes rely on validation experiments carried out at smaller scales than the actual reactor size. Nowadays the extrapolation of the codes accuracy between scales is accepted. Validated codes are then used in the licensing process using best estimate plus uncertainties (BEPU) or conservative approaches. Nonetheless, limited efforts have been done on the validation of such methodologies as a whole. This validation would imply to evaluate the scaling of uncertainties or the scaling of the conservative margins. The present work focuses on the evaluation of the scaling of uncertainties for BEPU methodologies. Test 3 of the OECD/NEA ROSA-2 project from the LSTF facility has been used for the analysis. The LSTF facility, operated by the Japan Atomic Energy Agency, is a facility that replicates a scaled down 4-loop pressurized water reactor (PWR). The facility was designed following the power-to-volume scaling approach with a scaling factor of 1/48. The experiment represents a small break loss-ofcoolant-accident at the hot leg. A post-test calculation was performed in the past showing the validity of RELAP5 to reproduce the involved phenomenology. The RELAP5 nodalization has been upscaled to the Ascó NPP reactor size (1/39), and an intermediate scale (1/25) using the Scaling up Methodology (SCUP) developed at UPC for qualifying full-scale nodalizations. The comparison of the three calculations at different scales provided equivalent results when known scaling distortions like heat losses were omitted. The three calculations have been complemented by an uncertainty analysis following the GRS methodology for propagating uncertainties (Wilks' formula). The comparison of the propagation of the uncertainties at different scales provides an insight on the scalability of the uncertainties and on the validation process of BEPU methodologies.