High-Pressure Acid-Gas Viscosity Correlation
Résumé
Acid gases containing hydrogen sulfide (H2S) are often encountered in the petroleum industry. However, reliable experiments on their thermophysical properties in reservoir conditions, on viscosity in particular, are scarce. From a modeling point of view, H2S and carbon dioxide (CO2) are polar compounds and as such are often considered rather difficult to model accurately. In this work, we propose a correlation with a strong physical background based on a corresponding-states (CS) approach to predict the viscosity from the temperature and the density of a large variety of systems for all stable thermodynamic states (gas, liquid, and supercritical). In particular, this correlation is applicable to predict the viscosity of sour/acid-gas mixtures, whatever the thermodynamic conditions. This approach is based on the Lennard-Jones (LJ) fluid model, which has been studied extensively thanks to molecular-dynamics (MD) simulations over a wide range of thermodynamic conditions. This fluid model can be extended to deal with polar molecules such as CO2 or H2S without a loss of accuracy. First, we demonstrate that the proposed physically based correlation is able to provide an excellent estimation of the viscosity [with average absolute deviations (AADs) below 5%] of pure compounds, including normal-alkanes, CO2, or even H2S, whatever the thermodynamic conditions (gas, liquid, or supercritical). Then, using a one-fluid approximation and a set of combining rules, the correlation is applied to various fluid mixtures in a fully predictive way (i.e., without any additional fitted parameters). Using this scheme, the deviations between predictions and measurements are as low as those on pure fluids using temperature and density as inputs. The viscosity of natural- and acid-gas mixtures at reservoir conditions is shown to be very well predicted by the proposed scheme. In addition, it is shown that this correlation can also be applied to predict reasonably the viscosity of asymmetric high-pressure mixtures, even in the liquid phase. This physically based approach is easy to include in any simulation software as long as, apart from temperature and density, the only inputs--the molecular parameters of each species--can be estimated from the critical temperature and the critical volume when not known.