Flow regimes identification and liquid-holdup prediction in horizontal multiphase flow based on neuro-fuzzy inference systems

Numerous techniques have been used to identify flow regimes and liquid holdup in horizontal multiphase flow, but often neither perform well nor very accurate. Recently, neuro-fuzzy inference systems learning scheme have been gaining popularity in its capability for solving both prediction and classification problems. It is a hybrid intelligent systems scheme that is able to forecast an output in the uncertainty situations. This paper investigates the capabilities of neuro-fuzzy TypeI in identifying flow regimes and forecasting liquid holdup in horizontal multiphase flow. The performance of neuro-fuzzy modeling scheme is implemented using different real-world industry databases. Comparative studies were carried out to compare neuro-fuzzy systems performance with the most popular existing approaches in identifying flow regimes and predict liquid holdup in horizontal multiphase flow. Results show that neuro-fuzzy is flexible, reliable, outperforms the existing techniques and show bright future capabilities in solving different oil and gas industry problems, namely, rock mechanics properties, water saturation, faceis classification, and distinct bioinformatics applications.