Uncertainty analysis on droplet size measurement in dispersed flow film boiling regime during reflood using image processing technique

•Droplet size and velocity are studied using advanced image processing technique.•Droplet measurement uncertainties are quantified with various statistical methods.•Measurement downstream of spacer grid has better accuracy than upstream location.•Droplet count lower limit of 30 gives statistically valid measurement results. In the dispersed flow film boiling (DFFB) regime during the reflood stage of a postulated loss of coolant accident (LOCA), liquid droplets are entrained by a continuous vapor phase moving upward within the rod bundle. The liquid droplets may interact strongly with the vapor flow and the resulting thermal–hydraulic behavior of the two-phase mixture could significantly affect the evolution of peak cladding temperature (PCT). Therefore, a clear and comprehensive understanding of liquid droplet behavior is crucial for reactor safety analysis. As an attempt to visualize and quantify the droplet behavior during reflood, an advanced image processing technique is used in the present study to capture the distributions of droplet size and velocity at the RBHT test facility. In order to ensure that the experimental droplet data obtained is reliable as well as to better utilize the data for subsequent thermal–hydraulic analyses, the image processing based measurement uncertainties have to be investigated and analyzed. In this paper, various statistical methods are explored to determine the minimum droplet sample size required as a function of standard deviation at a given confidence level. Accordingly, an appropriate droplet count lower limit is selected. Based on this lower limit, the 95% confidence intervals are determined for two typical reflood tests. In addition, the repeatability in droplet measurement of Rod Bundle Heat Transfer (RBHT) reflood tests is carefully examined by comparing data sets from two identical tests. Results show that a lower required level of precision and a smaller scattering in the measured droplet data generally require a smaller sample size. For typical RBHT reflood tests, it is found that the droplet size measured downstream of a spacer grid near the peak power location always has better accuracy compared with that at the upstream location. For the tests investigated, the maximum relative error in the liquid droplet Sauter mean diameter is found to be generally smaller than 0.15 (15%) when the droplet count lower limit is 30. Comparison of the results from two identical tests indicates that the RBHT reflood test resu