The processing of natural gas requires the implementation of new technologies in a context of increasing demand around the world. The natural gas liquids (NGL) separation using supersonic devices is a novel and efficient way to reduce volume of installed equipment and operation costs using the effects of highly turbulent and circular flows. This research implemented Computational Fluid Dynamics (CFD) modeling to improve the efficiency of typical NGL recovery process using the supersonic approach. In this research, a novel turbulence modeling approach was implemented aiming to minimize the processing time, and the results obtained were validated with experimental data available. This research is based on the model called k-epsilon RNG modified for swirl flow, this model has not been used and validated previously in highly compressible, turbulent and circular flow systems. The efficiency of the process was improved by 11% in comparison to the efficiency reported in past studies, and the processing time for the modeling was reduced by 40% with the proposed and adjusted turbulence approach. During the validation of the model k-epsilon RNG modified for swirl flow the swirl factor, part of the turbulence model, was adjusted to an optimum value for compressible, turbulent and circular flow systems involved in supersonic NGL separation process, allowing accurate results to be obtained with lower processing time than with other typical and common approaches as RSM and LES.