Eventos Anais de eventos
JEM 2019
5th Multiphase Flow Journeys
Gas Void Fraction Prediction for Air-Water and Air-Oil Two-Phase Flows via Artificial Neural Network
Submission Author:
Tiago Ferreira Souza , SP
Co-Authors:
Tiago Ferreira Souza, Cáio Araújo, Maurício Figueiredo, FLAVIO SILVA, Ana Maria Frattini Fileti
Presenter: Tiago Ferreira Souza
doi://10.26678/ABCM.JEM2019.JEM19-0017
Abstract
This work purpose was to assess the ability of the artificial neural network to predict the gas void fraction for air-water and air-oil vertical upward two-phase flows. The fluid properties and operating conditions were used as input parameters. To obtain the training and testing dataset, two-phase flow experiments were carried out in a 10.4 m long pipe of 0.053 m inner diameter, built in the Experimental Laboratory of Petroleum (LabPetro) at the Center for Petroleum Studies (Cepetro), located at the University of Campinas (Unicamp). A quick closing valve system was installed in the two-phase pipeline to measure the gas void fraction. The measurements carried out by this system presented 8.2% as the highest standard deviation. The artificial neural network (ANN) that best predicted the gas void fraction had six input parameters and one hidden layer with six neurons. The backpropagation was used as the training algorithm. The results showed a fine agreement between the ANN predictions and actual measured values.
Keywords
artificial neural network, Two-phase Flow, Gas Void Fraction
