In this work, we present improved correlations to calculate PC-SAFT parameters for hydrocarbon components based on molecular weight and refractive index (or density) at 20 °C without prior knowledge of the hydrocarbon family. We use the correlations to develop a fully predictive approach for the modeling of the temperature- and pressure-dependence of density of crude oils and petroleum fractions (e.g. gasoline, diesel, and jet fuels). These hydrocarbon mixtures are treated as lumped-solvents and are modeled with PC-SAFT parameterized from simple laboratory measurements of molecular weight and refractive index (or density) at 20 °C without detailed knowledge of composition. The approach is also extended to the modeling of live oils under gas injection. The predictive capability of the model is tested against 591 density measurements and 116 bubble point measurements from 32 crude oils from around the world. The model can predict, without any tuning parameters, density and bubble pressure with an accuracy of 1.06% and 4.82%, respectively, for live oils and their gas blends. We also develop a hydrocarbon characterization factor based on molecular weight and refractive index. The new characterization factor, called the aromatic ring index (ARI), can clearly distinguish between different families of hydrocarbons and provide an indication of the number of aromatic rings in the component. ARI us used throughout this work as a measure of aromaticity. The developed correlations and ARI concept are implemented in a characterization procedure for modeling polydisperse asphaltenes from a crude oil produced in deep-water. We use probability distributions to represent the maltenes and asphaltenes which allows for the generation of any number of pseudo-components without the need of extra tuning parameters. We perform a sensitivity analysis to investigate the implications of treating asphaltenes as a mono- or a poly-disperse mixture. From the modeling results of the UAOP envelope for the deep-water oil, we observe that PC-SAFT predicts a minimum in the upper critical solution temperature (MUCST). We develop a semi-empirical model that accurately captures the low-temperature behavior of the LLE phase boundary. The model is based on a linear extrapolation of normalized cohesive energy (LENCE). Finally, in search for a better model than the widely-used classical cubic EOS, we conclude this research work by presenting a general formulation of the newly-developed cubic-plus-chain (CPC) equation of state which hybridizes the classical cubic EOS with the chain term from SAFT. The general formulation allows the use of any cubic EOS (e.g. vdW, SRK, PR,…etc.) as a reference with any radial distribution function (e.g. Carnahan-Starling, Elliott,…etc.) for the chain term. This formulation facilitates future research for the improvement of CPC EOS for the final purpose of modeling polymer and crude oil systems.