Determination of pressure drop through a circular orifice for two-phase flow with machine learning technique

Abstract

Flow orifice is a known technique for measuring and regulating the flow. The accurate prediction of mass flow rate (ṁ) along the orifice is a research area. Many numerical and correlation based studies are implemented to get an accurate solution for the orifice problem, but still a better solution for the two-phase orifice problem is a vital topic. The determination of Δp along the orifice is dependent on the inputs such as Reynolds number (Re), area ratio (AR), and volume fraction (α) for a two-phase analysis. The range of data taken from a numerical analysis conducted based on the variation of the above inputs. The range of input data considered such as Re up to 100000, AR is 0.2–0.7 and α is 0.1–0.8. The present studies considered no. of machine learning (ML) techniques used for the orifice problem to determine the Δp. The ML techniques considered for the analysis are Linear Regression, LASSO, Ridge regression and Elastic-net for evaluating ṁ. A comparison statement also presents for the existing database and the predicting result based on ML technique with the best possible solution. The ML based study will give a better analytical view for the two-phase based orifice study.