GA-SVM based study on natural gas hydrate phase equilibrium

Natural gas hydrate is endowed with such advantages as high gas storage rate and high reliability, thus it can be widely applied in natural gas storage and transportation. However, the presence of hydrates also incurs gas pipeline blocking and other negative impacts. Therefore, it is of great importance to investigate the genesis of hydrates. In this paper, the static genesis of Type I hydrates under additive- free conditions was studied. First of all, a series of experiments were conducted with the hydrate kinetic experiment apparatus. Then, the temperature and pressure data obtained from the experiments were predicted and optimized with the genetic algorithm and support vector machine (GA-SVM). After non-linear matching of the above data, the phase equilibrium curve and its equation were obtained, based on which, the phase equilibrium pressure when hydrate was generated at atmospheric temperature was computed to be 33.5 MPa, and the phase equilibrium temperature was estimated to be 237.1 K. Finally, data acquired with the SVM-GA model, the Chen-Guo model and the vdW-P thermodynamic model were compared with those classic experimental data. Results show that the average relative errors are 2.678%, 1.447% and 3.249%, respectively. It is concluded that the GA-SVM model is rather accurate and simpler than the Chen-Guo model and the vdW-P thermodynamic model, and can provide more date for the development and research of natural gas hydrates. © 2017, Natural Gas Industry Journal Agency. All right reserved.