Standardized Polynomials for Fast Evaluation of Refrigerant Thermophysical Properties

Steady state and dynamic simulation and optimization are a key step in the design of heating, ventilation, airconditioning and refrigeration (HVAC&R) systems. It is well known that the computation time in such simulations is dominated by the refrigerant thermophysical property calculations. These calculations generally involve calculating all thermophysical properties given one or two independent parameters. Refrigerant thermophysical properties are typically calculated using some fundamental equations of state (EOS). The NIST REFPROP database is an industrially accepted standard for EOS implementation. Due to the iterative nature of the EOS calculations in REFPROP, the computation time is significant and sometimes not acceptable for optimization of HVAC&R systems. In this paper, a comprehensive approach for speeding up thermophysical property calculations is presented, including the functional forms as well as the implementation aspects. A set of polynomial functional forms are presented that allow for approximation of all the thermophysical properties in all the regions for a particular refrigerant (pure fluid or a blend) of interest. The polynomials can be easily scaled to make a judicious trade-off between computation time and accuracy. Analyses for refrigerants such as R1234yf, R32, R410A, R407C and R407F are presented. Using the proposed curve fits, the saturation properties for any refrigerant can be evaluated using less than 42 floating point operations (flops) and the flash calculations with less than 300 flops per property. The mean absolute error in predicted saturation properties is 0.001% and that of flash calculations is within 0.05%. Overall the individual property calculations are 100-5000 times faster than NIST REFPROP resulting in component and system simulation speed up factor of more than 100 for refrigerant blends. The use of two standardized and scalable functional forms for approximating all properties for all refrigerants of interest facilitates easy and robust implementation on a variety of steady state and transient simulation platforms as well as on hardware since limited data needs to be stored.