Desenvolvimento de modelagem múltiplas soluções sólidas para prever precipitação de parafinas e inferir sua composição a partir de termogramas

Abstract

Paraffin deposition is a challenge faced by the oil industry. On the last 25 years, several thermodynamic models were proposed for this phenomenon. In this work, three models were evaluated: the solid solution one, that considers the formation of one solid solution (SS); the multisolid phase model (MS), that assumes that each solid phase consists of a pure component; and the so called multisolid solutions (MSS). This thesis proposed an algorithm to simulate MSS model, that enables the formation of rotator and orthorhombic phases simultaneously. The performance of these methodologies was examined through comparison of wax appearance temperature (WAT), amount of wax precipitated at each temperature (WPC), and the global solid composition. Besides that, the modeled calorimetric curves were compared with DSC results from the literature. An algorithm to predict the calorimetric curve was also proposed in this work. Around sixty correlations to predict n-paraffin properties were tested, by comparison between experimental data of pure n-alkanes and the prediction of WAT, WPC, and calorimetric curves. It was concluded that the correlations choice had a great impact on these predictions, even altering which solid phases were present in MSS model. MSS model obtained better results for WPC, occurring overestimation of solid content by SS model and underestimation by MS. Different representations of liquid phase were tested. Results show that the use of Peng-Robinson equation overestimates the amount of n-paraffins precipitated in comparison with the utilization of Flory free-model combined with UNIFAC residual term. Three representations for rotator phase were investigated in MSS model: ideal solution, Wilson, and Chain Delta Lattice (extended for multicomponent mixtures in this work). Chain Delta Lattice model did not achieve good results for mixtures with great size differences among its components. The rotator phase ideality overestimated the solvent content in the solid on WPC prediction, nevertheless, obtained good results for calorimetric curves when components have similar sizes. When the size difference among mixture components was little, the use of Wilson s model for the rotator phase lead to smaller errors in WPC and global solid phase predictions, as well as for calorimetric curves. In these cases, the use of Wilson for the rotator phase lead to the occurrence of only orthorhombic phases for the majority of correlations tested. When Wilson model was applied for rotator phase in mixtures with little size difference among their components, the model predicted poorly the solid-solid transition temperatures. MFS model did not properly predict the peaks location in calorimetric curves when it was experimentally observed the occurrence of components mixture in solid phases. Furthermore, an algorithm to predict mixtures compositions through comparison between modeled calorimetric curves and DSC essays is proposed. With this algorithm it was possible to obtain satisfactory estimation of binary n-paraffin mixtures by MS and SS models