Thermophysical refrigerant property models play an essential role in dynamic models of vapor compression cycles, due to their highly nonlinear behavior, their coupling with many phenomena of interest, and the number of property computations required to simulate a cycle. As conventional iterative calculation methods are often too slow to be practically useful for such simulations, we compare two different approximation approaches, including one method that incorporates approximations of the liquid and vapor saturation lines, and another which approximates the entire property surface over a given domain. These methods are implemented in Modelica, and are demonstrated to successfully describe the nonlinear behavior of the refrigerant R-32 in a computationally and memory-efficient manner.