Thermodynamic Vapor-Liquid Equilibrium in Naphtha-Water Mixtures
Abstract
Naphtha is used to dilute the froth from bitumen treatment. Naphtha is recovered using a Naphtha Recovery Unit (NRU) and sent back to the froth dilution step. To minimize the environmental and economic impact of the NRU, it is imperative to maximize the naphtha recovery. It is, in this respect, that enhanced NRU Vapour-Liquid-Liquid equilibrium data is a significant value. The prediction of phase equilibria for hydrocarbon/water blends in separators, is a subject of considerable importance for chemical processes. Despite its relevance, there are still pending questions. Among them, is the prediction of the correct number of phases. While a stability analysis using the Gibbs Free Energy of mixing and the NRTL model for n-octane/water, provide a good understanding of calculation issues when using HYSYS V9 and Aspen Plus V9 software, this shows that significant phase equilibrium uncertainties still exist. In the case of multicomponent mixtures, the Tangent Plane Distance (TPD) is evaluated as a possible criterion for calculating the number of phases. Additionally, Paraffinic Aromatic Synthetic Naphtha (PASN) with a similar True Boiling Point (TBP) as typical naphtha can be used. Runs were developed in a CREC VL Cell operated with n-octane/water and PASN/water mixtures under dynamic conditions and used to establish the two-phase (liquid-vapour) and three-phase (liquid-liquid-vapour) domains. Results obtained demonstrate that the complete solubility is larger than the predicted by simulation software or reported in the technical literature. Furthermore, and to provide an effective and accurate method for predicting the number of phases, a Classification Machine Learning (ML) technique was implemented. Finally, traditional flash split calculations are reported explaining the challenges presented for the solution of the Rachford-Rice equations. A comparison of flash calculations between water/n-octane and PASN/water mixtures using SRKKD EoS is provided. The value of an ML approach developed based on the abundant experimental data available from the CREC-VL experimental Cell experiments is presented.