A Methodology for CHF Prediction in VVER Rod Bundles

TL;DR

This paper presents a methodology for predicting critical heat flux in VVER reactor rod bundles using various correlations and LUTs, assessing their accuracy, computational efficiency, and effects of physical parameters.

Contribution

It introduces a comprehensive approach combining reactor modeling, correlation comparison, and algorithm assessment for CHF prediction in VVER reactors.

Findings

Groeneveld LUT closely reproduces Bobkov LUT CHF values.

Non-uniform axial heat flux does not impact CHF in subcooled conditions.

Bourke algorithm is more noise-resistant and similarly efficient as standard algorithms.

Abstract

The critical heat flux (CHF) is an important thermal-hydraulic parameter that must be determined during the design of water-cooled reactors to ensure safety of operation. In this paper, a methodology is proposed to predict the CHF in VVER rod bundles. A whole-reactor model of VVER-1000 is implemented in RELAP-SCDAPSIM, and, based on the bundle-average approach, the parameters of the hot channel are used to assess four CHF predictors (i.e., the 2006 Groeneveld look-up table (LUT), and the Bowring, W-3, Biasi, and OKB-Gidropress correlations) upon comparison with the 2011 Bobkov LUT. A brief critique is given of the correction factors for both the Groeneveld and Bobkov LUTs. The effects of channel diameter and non-uniform axial heat flux on the predicted values of CHF in subcooled conditions are also studied. The performance and time complexity of the standard and Bourke trilinear interpolation algorithms are assessed for use with LUTs in safety codes. It has been concluded that the non-uniform axial heat flux does not affect the CHF in subcooled conditions and that the Groeneveld diameter factor exponent gives better predictability of the CHF in VVER rod bundles over the Tanase and Wong exponent correlations. Additionally, the Groeneveld LUT has been found to nearly reproduces the Bobkov LUT CHF values when used in conjunction with the Bobkov heated length factor. Values of MDNBR equal to 1.83 and 1.96 can be used as thermal limits at 12\% overpower when the Bowring and W-3 correlations, respectively, are used in subcooled conditions. The standard and Bourke algorithms nearly exhibit the same average CPU time. The Bourke algorithm, however, is less affected by noise. Further research is needed to ensure the smoothness of the Bobkov LUT and to derive reactor type-dependent correction factor formulas for use with the Groeneveld LUT.