The knowledge of the phase equilibrium of refrigerant mixtures is crucial for the optimization of the production processes of refrigerant molecules, and the design of energetic systems (heat pumps, Organic Rankine Cycles, …). In this work, different predictive thermodynamic models (COSMO-RS and COSMO-SAC-dsp) have been used to determine the phase behavior of mixtures of refrigerants involving alkanes, fluorinated and chlorinated compounds. The COSMO-RS model leads to rather satisfactory predictions of the vapor-liquid equilibria of mixtures and is able to capture the azeotropic behavior for most systems. However, the original COSMO-SAC model as well as the COSMO-SAC 2010 version lead to unsatisfactory predictions of VLE for these systems, as they are unable to predict the azeotropic behavior in alkane + refrigerants. By combining COSMO-SAC with a dispersion term (COSMO-SAC dsp model) and readjusting the universal parameter for the F atom, it is possible to get very satisfactory predictions of similar accuracy as COSMO-RS. In order to predict the phase behavior at high pressure, the COSMO-SAC-dsp model can be combined with the Peng-Robinson equation of state and the MHV1 mixing rule. Excellent predictions of relative volatilities can be obtained with this approach over wide temperature and pressure ranges. COSMO calculation can also be used to predict saturated liquid densities, vaporization enthalpy and boiling points of pure refrigerants.

制冷剂混合物的相平衡知识对于优化制冷剂分子的生产过程以及设计高能系统（热泵，有机朗肯循环等）至关重要。在这项工作中，已使用不同的预测热力学模型（COSMO-RS和COSMO-SAC-dsp）来确定涉及烷烃，氟化和氯化化合物的制冷剂混合物的相态。COSMO-RS模型可对混合物的气液平衡做出令人满意的预测，并且能够捕获大多数系统的共沸行为。但是，原始COSMO-SAC模型以及COSMO-SAC 2010版本导致这些系统的VLE预测不令人满意，因为它们无法预测烷烃+制冷剂中的共沸行为。通过将COSMO-SAC与色散项（COSMO-SAC dsp模型）结合并重新调整F原子的通用参数，可以获得非常令人满意的与COSMO-RS相似的准确预测。为了预测高压下的相行为，可以将COSMO-SAC-dsp模型与Peng-Robinson状态方程和MHV1混合规则结合。通过这种方法，可以在较宽的温度和压力范围内获得相对挥发度的出色预测。COSMO计算还可用于预测纯制冷剂的饱和液体密度，汽化焓和沸点。为了预测高压下的相行为，可以将COSMO-SAC-dsp模型与Peng-Robinson状态方程和MHV1混合规则结合。通过这种方法，可以在较宽的温度和压力范围内获得相对挥发度的出色预测。COSMO计算还可用于预测纯制冷剂的饱和液体密度，汽化焓和沸点。为了预测高压下的相行为，可以将COSMO-SAC-dsp模型与Peng-Robinson状态方程和MHV1混合规则结合。通过这种方法，可以在较宽的温度和压力范围内获得相对挥发度的出色预测。COSMO计算还可用于预测纯制冷剂的饱和液体密度，汽化焓和沸点。