Prediction of methane uptake on different adsorbents in adsorbed natural gas technology using a rigorous model

One of the most promising methods for transporting natural gas and overcoming its low energy density is adsorbed natural gas (ANG) technology. ANG technology is highly dependent on the perfect conception of adsorption isotherms in different operational conditions and on different adsorbents. In this study, the utilization of a novel mathematical model of least squares support vector machine (LSSVM) for accurate prediction of adsorption isotherm has been examined. The considered variables were temperature, pressure and type of adsorbents. A data set containing 670 experimental data points of methane adsorption on 10 different adsorbents in a broad range of temperature and pressure were used for training and testing of the LSSVM model. Results showed that the LSSVM model is capable to predict adsorption isotherm with an acceptable statistical parameters of 2.3058% and 0.9995 for AARD% and R², respectively. In addition, the leverage statistical algorithm indicated that the suggested model is statistically authoritative for prediction of methane isotherm adsorption and no outliers have been detected in the data set.