Reproductive accuracy of selected void fraction correlations for horizontal and vertical upflow

Abstract

Several mean void fraction correlations have been proposed in the past. Indeed, they differ with respect to the physical parameters included, the underlying data basis, the model assumptions etc. Their predictions also considerably deviate from each other. Thus, the rational choice of an appropriate model is not straightforward. On the basis of a large mean void fraction data bank, containing experimental results for vertical upward and horizontal pipe flow, some specific correlations have been checked for their predictive accuracy in relation to each other and in view of a generally accepted level of permissible inaccuracy. The selection of the correlations for the relative ranking is based on the definition of a set of physically indispensable equation variables and the inclusion of some theoretical limits, which have to be met by the relationships. The authoritative criteria for the ranking are statistical parameters describing the scatter of the differences beteen the measured and the calculated values of the mean void fraction and of the two-phase mixture density. In detail, the standard deviations of the logarithmic and absolute deviations are used with the same weight. The comparison confirmed that most of the void fraction correlations either reproduce the mean void fraction or the mean density with a rather acceptable accuracy. All in all, Rouhani’s first correlation and the older HTFS-Alpha model are the only ones meriting a general recommendation in the case of horizontal and vertical upward flow.