New association schemes for tri-ethylene glycol: Cubic-Plus-Association parameterization and uncertainty analysis

Abstract

Transportation of natural gas extracted from subterranean reservoirs is subject to several risks introduced by the presence of water. These risks can be mitigated through subsea dehydration, the design of which requires an accurate understanding of the thermodynamics of these systems. Having previously investigated the use of mono-ethylene glycol as a dehydration agent, this research focusses on the industry workhorse: tri-ethylene glycol (TEG). The Cubic-Plus-Association (CPA) equation of state has previously provided a good description for systems of equal complexity and in this work the literature 4C and 6D association schemes, as well as four newly proposed association schemes (4F, 5F, 6F and 5C) are evaluated for the description of TEG. These association schemes, which are explained in the manuscript, represent different ways of self-association for TEG.Pure component parameters are estimated by fitting to vapor pressure, liquid density, and liquid-liquid equilibrium data, with statistical uncertainty analysis implemented to determine the confidence intervals of the parameters. Binary interaction parameters are similarly fitted to binary vapor-liquid equilibrium (VLE) data. Optimized parameters are evaluated by testing their performance in predicting binary and ternary VLE systems.Due to the strong influence of certain experimental data sets, the literature 6D parameters are shown to be unsuitable for the applications in this work. The need for consistent experimental data quantifying both vapor and liquid phases, specifically for TEG – CH4 has been highlighted. The best overall performance was obtained using the newly proposed 4F association scheme. Along with improved overall description of various VLE and LLE systems, with average absolute relative deviations generally between 2-20%, single parameter sensitivity analysis shows improved optimality for several systems also not included in the parameterization. This is indicative of a better fundamental characterization of the TEG molecule. The value of the uncertainty analysis is shown for the binary interaction parameters, where for TEG – CH4 VLE a temperature-dependent correlation is shown to be unnecessary using the optimum association schemes.