Study on Liquid Film Thickness Measurement of Annular Flow with Image Processing

Abstract

For the last decades, a lot of researches have been conducted on the study on gas-liquid two phase flow. The importance of the related researches stands on the basis of the widespread its industrial applications. An appreciable percentage of the researches are devoted to research on annular flow. Annular flow is a flow pattern of gas-liquid two phase flow. The presence of a liquid film flowing on the wall of tube and the gas flowing in the core of tube characterize it. Liquid film thickness is major variable to characterize the annular flow with mass flow rates of gas and liquid, densities of gas and liquid, viscosity of gas and liquid, pressure, temperature, tube diameter, tube material, etc. This paper presents the new measurement method for liquid film thickness of annular flow, which uses visualized result of the flow and image processing procedure. First of all, the optical path in the tube of gas-liquid two phase flow flowing is interpreted by analytical method. The variation of optical path in annular flow with varying liquid film thickness is the main objective of analysis. The result shows the five types of optical path exist in each case of liquid film thickness. Based on the result, the new thickness measurement method is suggested. For the new method, the image of annular flow with transparent tube and light blocking cover and the image processing procedure are required. As a configuration of the light blocking cover affects the image of annular flow, the effect of a configuration of the cover is investigated by analytical method. Additionally the effect of a refractive index of the materials is investigated by analytical method with same reason. The method using image processing was verified by experiment. For comparison, the sampling method is selected, which can measure spatial-temporal averaged liquid film thickness. Film thickness data are measured by two methods and those are compared to verify image processing method. The results of each method matched well with ±20% error.