Optimization of acid gas sweetening plant based on least squares — Support Vector Machine (LS-SVM) Model and Grey Wolf Optimizer (GWO)

Natural gas is an energy resource that is widely used as energy and raw material in many industrial processes. It is contaminated some impurities such as CO2, H2S and water, hence, removal of the contaminant processes are required. One of the natural gas processing is Acid Gas Sweetening. The purpose of this process is to eliminate H2S and CO2 compound from natural gas. H2S tend to corrosive and CO2 will reduce the thermal efficiency. In this research, the goal of optimization that had to be accomplished is to minimalize the energy consumption on a condenser and re-boilers in regenerator process. Least Squares — Support Vector Machine (LS-SVM) is used to modeling a Qcondenser, Qre-boiler and CO2 on lean amine, Grey Wolf Optimizer (GWO) is used to find the optimum value of energy consumption in a condenser and re-boilers, based on training process, obtained the value of Root Mean Square Error (RMSE) for Qre-boiler, Qcondenser and CO2 on lean amine respectively are 0.0909, 0.0916 and 0.1011, from validation process, RMSE values obtained for Qcondenser, Qre-boilers, and CO2 on lean amine respectively of 0.0680, 0.0587 and 0.0850. The optimum values of energy consumption in a condenser and re-boilers using GWO obtained value are 1.287E+05 kJ/h, the value of Particle Swarm Optimization (PSO) as a comparison are 4.781+05 kJ/h.