Critical heat flux (CHF) for water flow in tubes—II.: Subcooled CHF correlations

Abstract

The proliferation of critical heat flux (CHF) prediction methods typifies the lack of understanding of the CHF phenomenon and makes it difficult to choose a suitable correlation, look-up table, or model. In fact, an exhaustive literature search by the authors of this study identified over 100 CHF correlations applicable to subcooled water flow in a uniformly heated round tube. The accuracy of the latest versions of these correlations was ascertained using the PU-BTPFL CHF database from Part I of the present study. This database contains the largest volume of subcooled CHF data (5544 data points) of any database available in the literature. In response to the many inaccurate and inordinately complex correlations, two nondimensional, subcooled CHF correlations were formulated, containing only five adjustable constants and whose unique functional forms were determined without using a statistical analysis but rather using the parametric trends observed in less than 10% of the subcooled CHF data. The correlation based on inlet conditions (diameter, heated length, mass velocity, pressure, inlet quality) was by far the most accurate correlation, having mean absolute and root-mean-square errors of 10.3% and 14.3%, respectively, and is recommended for water flow in a tube having a uniform axial heat flux. The outlet (local) conditions correlation was the most accurate correlation based on local CHF conditions (diameter, mass velocity, pressure, outlet quality) and may be applied to water flow in a tube having a nonuniform axial heat flux. Both correlations proved more accurate than a recent CHF look-up table commonly employed in nuclear reactor thermal hydraulic computer codes. The parametric range for the inlet conditions correlation ( 0.25×10 −3≤D≤15×10 −3 m, 2≤L/D≤200, 300≤G≤30,000 kg m −2 s −1, 1×10 5≤P≤200×10 5 N m −2, −2.0≤x i ≤0.0 , −1.0≤x o ≤0.0 ) was chosen so as to include those regions where data were most abundant, containing approximately 85% of the subcooled CHF database. Superiority of the correlations was attributed to the systematic development of the functional forms of the correlations from the CHF parametric effects; thus, re-optimization of the constants, when additional subcooled CHF data become available, is not expected to produce an appreciable increase in accuracy.