Document Type : Review Paper
Author
Department of Applied Chemistry, Faculty of Science, Malayer University, Malayer, Iran
Abstract
The reliable prediction of asphaltene precipitation is critical for mitigating operational challenges in oil production, yet existing thermodynamic models face persistent limitations due to the complex composition and self-association behavior of asphaltenes. This study evaluates two dominant modeling frameworks: solubility-based approaches, which treat asphaltenes as dissolved macromolecules, and colloidal stability-based models, which conceptualize asphaltenes as suspended particles stabilized by resin interactions. Solubility models, including Flory-Huggins’s theory and cubic equations of state, correlate precipitation with solubility parameters and Gibbs free energy changes, but are constrained by assumptions of molecular homogeneity and reliance on ill-defined critical properties. Colloidal models, such as micellization theory, link precipitation to resin depletion and micelle destabilization.
Statistical Associating Fluid Theory (SAFT) models, particularly the Perturbed-Chain (PC-SAFT) variant, bridge two modeling approaches by incorporating molecular-specific parameters—such as segment size, interaction energy, and chain length—to describe complex behaviors like molecular size variations and clustering in fluids. While SAFT models offer strong predictive accuracy, their reliance on carefully calibrated inputs and computational complexity highlights a trade-off between scientific rigor and practical application. Experimental studies demonstrate that no single model reliably predicts asphaltene precipitation across all crude oil types, underscoring the need for hybrid methods.
This article critically reviews model-based predictions in the context of asphaltene behavior, addressing their necessity, strengths, and limitations. It investigates the most widely adopted models in both academic literature and industrial software, analyzing the underlying reasons for their prevalence. Furthermore, the study explores the similarities and differences among existing models and their classification criteria.
Keywords