Renato Feijo Evangelista
Modeling thermodynamic properties can be challenging when the data available for parameters identification is limited. Fully-predictive group contribution (GC) methods have been developed as an alternative to overcome data scarcity. Although providing a higher degree of accuracy, most recent GC approaches require detailed information of the molecular structure, which is not acquirable for systems with unspecified components. This work intends to establish the foundation to address this specific scenario. The proposed PC-SAFT approach assisted by a homosegmented group contribution scheme permits parameters calculation with accuracy due to one adjustable parameter without requiring meticulous information regarding the molecular structure, for instance, the relative position between carbon-centered groups. This semi-predictive approach is especially suitable for cases in which some data is often available, but cannot be taken into consideration by current GC models. Sensitivity analyses indicated good predictability in the extreme case where a single vapor pressure data point is provided to adjust the model parameter, whereas enhanced predictions may be achieved if more data are available.
This original modeling approach was further enhanced by redefining the groups based on carbon-13 nuclear magnetic resonance (13C-NMR). In this version, the groups are defined by spectrum segments delimited by specified boundaries, while the relative amount of groups in a molecule is quantified as the relative intensity of signals comprised within each specified range. Therefore, the 13C-NMR analysis of the fluid of interest may suffice as the sole source for the required structural information. The model parameter is fitted to at least one vapor pressure data point, which can be estimated using refractive index and molecular weight data, through a series of newly defined correlations. This approach is especially useful for cases where saturation pressure measurements are not practical, but refractive index is more easily quantified. Furthermore, these correlations represent an alternative method for calculating critical properties and acentric factor of non-polar hydrocarbons, as a function of molecular weight and refractive index, which has numerous potential applications.
The concepts established in this work have shown promising potential for some industrial applications given the simplified experimental and computational implementation. For instance, equations of state have been extensively used to predict the phased behavior of petroleum fluids in the oil and gas industry, however obtaining the model parameters for these systems is still a challenge given that they often consist of undefined components. Therefore, the application of the concepts presented in this work is especially valuable in this scenario. In this context, a methodology for applying the approach to the modeling of petroleum fluids and in a fully-predictive manner was proposed and exemplified over hydrocarbon mixtures. With this work, I aim to motivate further evaluation of the proposed methodology for application to actual petroleum fluids and to inspire the development of practical methods to perform thermodynamic calculations of complex multicomponent mixtures.