This research suggests two new group contribution methods to facilitate phase behavior calculation when reliable experimental data are lacking. The first method pertains to the implementation of an updated version of the Elliott and Natarajan method to the Statistical Associating Fluid Theory (SAFT) and Perturbed-Chain Statistical Associating Fluid Theory (PC-SAFT) equations of state. Shape factor parameters have been correlated for 878 compounds including different variety of families and the parameters from Elliott and Natarajan have been updated to improve accuracy for alcohols. Thereafter, thermodynamic properties such as boiling temperatures and vapor pressures have been predicted. We obtain 36%, 65%, and 32% AAD% in pressures for the ESD, SAFT, and PC-SAFT equations of state. Additionally, we have compared our GC-PC-SAFT to the one by Tihic et al., applying their suggested First-Order and Second-Order groups for 650 non-associating compounds. We observed higher accuracy for our method relative to the Tihic et al. The resulting P AAD% were 53% for Tihic FOG and 42% for Tihic SOG.
The second method suggests a new group contribution model for Tb at 760mmHg and Tb at 10 mmHg. These correlations recognize a finite limit in boiling temperature as infinite molecular weight is approached. The availability of two vapor pressures enables straightforward application of the Clausius-Clapeyron equation to estimate boiling temperatures at other points. In the presented approach, there are 3 parameters and 72 functional groups for each temperature which are regressed through a database consisting of 336 hydrocarbons and 642 non-hydrocarbons. The average absolute percent deviations (AAD%) between the correlated and experimental temperatures are calculated in comparison with Joback-Reid and Gani approaches. We obtain 3.5, 4.7, and 4.1 AAD% in temperature for the present work, Joback, and Gani methods, respectively. Additionally, the accuracy of the present work is evaluated by calculating the vapor pressures from the DIPPR correlation at the predicted temperatures of each model. We obtained 33.2, 104.3 and 48.1 AAD% in pressure for the present work, Joback, and Gani methods. Finally, the accuracy of the presented correlations are tested against Asher and Pankow model, UNIFAC-PL°, for 66 volatile compounds in the temperature range of 290 – 320 K. For the vapor pressure at the 10mmHg boiling temperature, we obtain 36.9 AAD% for the present work and 94.5 AAD% for the Asher method.
Overall, these group contribution methods establish a standard for comparison of more fundamental methods like molecular simulations with transferable potentials. Transferable potentials generally provide accuracy of 10-30AAD% in pressure, but have only been developed for relatively small databases over narrow temperature ranges.