This paper presents selected results on heat transfer to supercritical water flowing upward in a 4-m-long vertical bare tube. Supercritical water heat transfer data were obtained at pressures of [approximately]24 MPa, mass fluxes of 200 to 1500 kg/m2s, heat fluxes up to 874 kW/m2 , and inlet temperatures from 320 to 460°C for several combinations of wall and bulk-fluid temperatures that were below, at, or above the pseudocritical temperatures.

In general, the experiments confirmed that there are three heat transfer regimes for forced-convection heat transfer to water flowing inside tubes at supercritical pressures: (a) normal heat transfer regime characterized in general with heat transfer coefficients (HTCs) similar to those of subcritical convective heat transfer far from the critical region, which are calculated according to Dittus-Boelter-type correlations; (b) deteriorated heat transfer (DHT) regime with lower values of HTC and hence higher values of wall temperature within some part of a test section compared to those of the normal heat transfer regime; and (c) improved heat transfer regime with higher values of HTC and hence lower values of wall temperature within some part of a test section compared to those of the normal heat transfer regime.

These new heat transfer data are applicable as a reference dataset for future comparison with supercritical water bundle data and for a verification of scaling parameters between water and modeling fluids.

Also, these HTC data were compared to those calculated with the original Dittus-Boelter and modified Bishop et al. correlations. The comparison showed that the modified Bishop et al. correlation (i.e., the Bishop et al. correlation with the constant proposed by Kirillov et al.), which uses the cross-sectional averaged Prandtl number, represents HTC profiles more correctly along the heated length of the tube than the Dittus-Boelter correlation. In general, the modified Bishop et al. correlation shows good agreement with the experimental HTCs outside the pseudocritical region; however, it underpredicts the experimental HTCs within the pseudocritical region. The Dittus-Boelter correlation can also predict experimental HTCs outside the pseudocritical region but deviates significantly from experimental data within the pseudocritical region by up to four times.

A reason for this deviation is that the Nusselt number in the Dittus-Boelter correlation and corresponding HTC values closely follow the regular Prandtl number (i.e., based on data from thermophysical properties tables), which in turn closely follows the peak in specific heat within the pseudocritical region. However, experimental HTC values show just a moderate increase within the pseudocritical region possibly due to significant variations of fluid temperature within the tube cross section. In this case, the bulk-fluid temperature might not be the best characteristic temperature at which all thermophysical properties should be evaluated. That is why the cross-sectional averaged Prandtl number is used in many supercritical heat transfer correlations instead of the regular one.

A simple empirical correlation was proposed for calculating heat flux at the starting point of the DHT regime. However, it should be noted that both these correlations, i.e., the Dittus-Boelter and modified Bishop et al. correlations, cannot accurately predict HTCs within the DHT regime.