Issue link: https://www.e-digitaleditions.com/i/1417310
26 OCTOBER 2021 Inhalation Conclusion and next steps e need for accurate thermophysical property pre- diction for diverse formulation mixtures is expected to grow, as next-generation formulations with low global warming potential (GWP) are introduced and as usage of predictive simulation increases during pMDI development. is need can be supported by appropriate experimental measurement campaigns for formulation mixtures, spanning the full range of temperature, pressure and composition that will be encountered by pMDI internal ows and aerosols. A promising set of methods for representing and predicting many of these mixture thermophysical properties is a framework of physically-based mod- els that use activity coe cients to describe non-ideal molecular interactions in the liquid phase, which can a ect such properties. ese models are particularly useful for solution formulations containing ethanol. Future work would require current and new exper- imental data for new propellants, e.g., HFA152a, with ethanol and other important constituents, to generate activity coe cient models that describe the physical property behavior of these mixtures. is may involve searching for the most suitable activity coe cient model that best describes the measure- ment data and underlying physics and chemistry. is ultimately could lead to a property prediction tool that can be integrated, with con dence, in one-dimensional and three-dimensional CFD sim- ulations, including usage at conditions outside the range of existing experimental data. e ability of the UNIQUAC model to predict physical property data is shown in Figure 3, where UNIQUAC parameters previously generated solely from tting to experimental SVP data [16] are used directly in the UNIMOD model for viscosity. e experimental data has a maximum of 2.3% relative standard deviation, from 10 repeats, and the maxi- mum deviation of the UNIMOD prediction from the experimental data is 4.9%, with a root mean square deviation of 3.5%. Figure 4 shows the e ect of incorporating both SVP data into the tting process (as standard) and desired thermophysical property data (here, surface tension is shown). When the parameters are solely from the SVP, the t is not satisfactory, with the root mean square deviation at 7.5%. When the surface tension data was incorporated into the tting process, this was reduced to 5%. is tting approach was then used to generate all the data presented in reference 18, to ensure a better representation of the surface tension data. ere was no adverse e ect on the quality of the prediction of the viscosity. e tted expressions that generated the examples shown in this article could readily be used in con- junction with 1-D modeling or CFD modeling to provide predictions of SVP, surface tension and liq- uid viscosity at all temperature and composition conditions encountered in such a model. Figure 4 Surface tension measurement [14] and UNIFAC prediction for HFA134a/ethanol mixtures, at 20.3°C. Parameters were generated solely from SVP data or from SVP and surface tension data. 0.025 0.020 0.015 0.010 0.005 0.000 Surface Tension (N/m) 0 0.2 0.4 0.6 0.8 1 Ethanol Mass Fraction UNIFAQ - SVP Fitting Experiment UNIFAC - SVP + ST Fitting Figure 3 Liquid viscosity measurement [15] and UNIMOD prediction (using UNIQUAC parameters) for HFA134a/ethanol mixtures, at 20.4°C. 1.4 1.2 1 0.8 0.6 0.4 0.2 0 Viscosity (cP) 0 0.2 0.4 0.6 0.8 1 Ethanol Mass Fraction Experiment UNIQUAC SVP Fitting only