Molten salt systems have experienced renewed interest recently for their applications in energy storage and next generation of nuclear reactors. Molten salt reactors are considered to be a leading candidate for the next generations of nuclear reactors; they have a great potential to help with the energy problems the world is facing. Molten salt reactors have enhanced safety figures and higher energy production efficiency over traditional nuclear reactors. Molten salts can serve as both the solvents of nuclear fuels and the heat transfer fluids. Experiments on molten salt systems can be difficult and even dangerous, due to the high temperature and highly corrosive environments of molten salts. Molecular dynamics (MD) simulation is a powerful tool to study molten salts, since MD simulations not only avoid the extreme environments in the experiments, but also provide scientific insights on the molecular level. In this thesis, we have applied MD simulations to study physical properties of molten salts. We have specially focused on a fundamental topic of MD simulation - the force field. A fixed charge model and a polarizable force field are studied to understand the importance of electronic polarization effects in the salt systems. A newly introduced approach - the machine learning force field - is also applied and its usefulness is discussed.