This study utilizes molecular simulation techniques to investigate the properties of fluids, including binary mixtures of n-hexane and n-octane, CO2 and CH4 in MFI-type zeolites, n-alkane adsorption and diffusion in carbon nanotubes, and water-in-oil microemulsions using CO2-expanded mixtures as the oil phase. The simulations are performed using GROMACS and LAMMPS software packages, and the molecular models are parameterized using experimental data and theoretical calculations. Molecular dynamics simulation is used to predict the behavior of individual molecules in simulated fluids, allowing for the calculation of a range of properties, such as density, pressure, viscosity, and diffusion coefficients. The use of machine learning techniques is also explored to improve the accuracy and efficiency of the simulations. Recent advancements in the development of new force fields, enhanced sampling methods, and the use of machine learning have significantly improved the accuracy and reliability of molecular simulations, enabling the design of new materials and optimization of energy systems.