Modeling of Natural Gas Components Hydrate Formation by Using Neural Network

Please enable javascript in your browser.

Ameri, Azadeh | 2009

1029 Viewed

Type of Document: M.Sc. Thesis
Language: Farsi
Document No: 39190 (06)
University: Sharif University of Technology
Department: Chemical and Petroleum Engineering
Advisor(s): Ghotbi, Siroos; Taghikhani, Vahid
Abstract:
In this study two approaches were applied to predict the hydrate dissociation pressure of natural gas in the presence of aqueous water. One approach applied van der Waals and Platteeuw solid solution theory & PR EOS the other applied a feed forward multi layers artificial neural network (ANN) with 19 input variables (temperature, existence of hydrocarbon liquid and ice phase, gas phase composition, inhibitor composition in aqueous phase), and one hidden layer with 17 neurons. In comparison of both methods it was concluded that, in temperature above 12℃ , ANN is more accurate than thermodynamic model, but in lower temperature thermodynamic model is comparable with ANN. The trained network has less neurons in hidden layer than two other previously applied neural networks
Keywords:
Artificial Neural Network ; Van Der Waals Model ; Hydrate Phase Equilibrium ; Peng Robinson Equation of State ; Interaction Parameter ; Platteeuw Model ; Hydrate Dissociation Pressure

No TOC