Modelling two-phase Z factor of gas condensate reservoirs: Application of Artificial Intelligence (AI)

Two-Phase Flow Pressure Drop Modelling in Horizontal Pipes with Different Diameters

“…To conclude the previous research, it has been confirmed that soft computing approaches are a promising tool for solving complex nonlinear problems (Faraji et al, 2022). Most of the previous studies are mainly focused on characterizing multiphase flow as a function of flow regimes, and multiphase flow frictional pressure drop has given less attention.…”

“…The training aims to build an ML model that can predict future events (outcomes). Such a model is purely data-driven, and learning is achieved by finding the relationship between the input and output parameters (Faraji et al, 2022(Faraji et al, , 2020. Some examples of the ML methods in the literature include Artificial Neural Networks (ANN), Support Vector Machine (SVM), Deep Learning (DL), Evolutionary Algorithms, Genetic Programming and Fuzzy Logic.…”

“…The first layer is known as the input layer corresponding to the input data; the middle layers are known as the hidden layers, representing the computational operation of the network and the output layer, which determines the network's output. An MLP with one hidden layer is accurate enough to generalize many problems; however, an extra layer/s might be added to the network (Azizi et al, 2016;Faraji et al, 2022;Louridas and Ebert, 2016). The number of hidden layers and neurons in each layer is entirely problem dependent and can be determined empirically (Haykin, 2005;Schmidhuber, 2015).…”

“…The exploration efficiency of the algorithm is assessed using updated weight function. The optimum weights are calculated utilizing searching global minima mechanism (Faraji et al, 2022).…”

“…To investigate the quantitative impact of each input variable on the output parameter (frictional two-phase pressure drop) for the developed MLP-GA model, a sensitivity analysis was performed. Several methods exist (e.g., Kendall, Pearson and Spearman) in the literature to perform this analysis (Chok, 2010;Faraji et al, 2022). Pearson's relevancy factor (r) method was used as one of the most popular techniques for sensitivity analysis (Mesbah et al, 2017;Shariaty et al, 2019).…”