Prediction of temperature-entropy diagram classes for pure fluids in organic Rankine cycles using group contribution methods

The organic Rankine cycle is an indispensable technology in low-temperature heat recovery, where the working fluid is a critical element. It influences not only the type and design parameters of the equipment but, via the shape of the saturated liquid and vapor line, the configuration as well. Besides thermophysical and chemical properties, the selection of the ideal working fluid is also affected by the ever-growing stringency of environmental regulations. Hence, there is a growing demand to create unique, ecologically sound media. Addressing this challenge involves simultaneously optimizing molecular structure and process parameters —an endeavor often facilitated by computer-aided molecular design, applying group contribution methods. Given the frequent application of cubic equations of state (EoS) in this method, the present study investigates which of the 23 common cubic EoS should be chosen when the class of the shape of the newly designed phase curve is to be determined. The grouping is examined for the traditional 3-element and the more sensitive Györke classification. Interestingly, the findings indicate that the fewer material properties an EoS relies on, the more accurate the classification becomes, provided these properties are accurately estimated. Complex EoSs, however, may exhibit higher parameter sensitivity and, similarly to model validation, can become overfitted. Moreover, results show that achieving high accuracy in phase curve design does not necessarily equate to high classification ability, which may require the introduction of distinct indicators for classification and determining the precision of the temperature-entropy phase curve.