Artificial neural network application for multiphase flow patterns detection: A new approach

Abstract Multiphase flow measurement is a very challenging issue in process industry. There are several techniques to estimate multiphase flow parameters. However, these techniques need correct identification of the flow patterns first. Artificial Intelligence is one of the promising technique for identifying the flow patterns. In this paper, we used an Artificial Neural Network (ANN) for flow pattern identification but with pre-processing stage using natural logarithmic normalization. This pre-processing stage helps to normalize large data range and to reduce overlapping between flow patterns. Thus, the validity of the model was extended by using dimensionless inputs to be implemented for horizontal pipes of various diameters, liquid densities and viscosities. The concept was validated by building and testing the model using experimental data as well as well-known multiphase flow models. An ANN model was built using three dimensionless parameters only, namely, Liquid Reynolds Number, Gas Reynolds Number and Pressure Drop Multiplier. The present model achieved more than 97% accuracy in classifying the flow patterns for wide range of flow conditions.

[1]  Emad A. El-Sebakhy,et al.  Flow regimes identification and liquid-holdup prediction in horizontal multiphase flow based on neuro-fuzzy inference systems , 2010, Math. Comput. Simul..

[2]  J. P. Brill,et al.  Unified model for gas-liquid pipe flow via slug dynamics: Part 1: Model development , 2003 .

[3]  Martin Fodslette Møller,et al.  A scaled conjugate gradient algorithm for fast supervised learning , 1993, Neural Networks.

[4]  Cem Sarica,et al.  Unified Modeling of Gas/Oil/Water Pipe Flow - Basic Approaches and Preliminary Validation , 2006 .

[5]  Robert A. Jacobs,et al.  Increased rates of convergence through learning rate adaptation , 1987, Neural Networks.

[6]  William J. Bailey,et al.  Steady-State Multiphase Flow—Past, Present, and Future, with a Perspective on Flow Assurance , 2012 .

[7]  Kurt Hornik,et al.  FEED FORWARD NETWORKS ARE UNIVERSAL APPROXIMATORS , 1989 .

[8]  S. M. Ghiaasiaan,et al.  Artificial neural network approach for flow regime classification in gas–liquid–fiber flows based on frequency domain analysis of pressure signals , 2004 .

[9]  A. Murat Ozbayoglu,et al.  Analysis of gas–liquid behavior in eccentric horizontal annuli with image processing and artificial intelligence techniques , 2012 .

[10]  Takaaki Ohishi,et al.  Performance comparison of artificial neural networks and expert systems applied to flow pattern identification in vertical ascendant gas-liquid flows , 2010 .

[11]  J J Hopfield,et al.  Neural networks and physical systems with emergent collective computational abilities. , 1982, Proceedings of the National Academy of Sciences of the United States of America.

[12]  P. Vassallo,et al.  Two-phase frictional pressure drop multipliers for SUVA R-134a flowing in a rectangular duct , 2006 .

[13]  A. Dukler,et al.  A model for predicting flow regime transitions in horizontal and near horizontal gas‐liquid flow , 1976 .

[14]  Kurt Hornik,et al.  Multilayer feedforward networks are universal approximators , 1989, Neural Networks.

[15]  Simon Haykin,et al.  Neural Networks: A Comprehensive Foundation , 1998 .