Performance of the Self-Referencing Model on the Determination of Viscosity versus Pressure and Temperature of Various Hydrocarbons and Their Mixtures

This is the authors' abstract. We don't add key points for this paper.

The predictive model of the viscosimetric behavior studied in this article is the self-referencing model developed in our laboratory by KANTI et al. The dominant idea of this method is to characterize the dynamic viscosity (P,T) of a fluid just knowing its viscosity at P0=0.1MPa and at a reference temperature T0 (generally ambient temperature). For this purpose, we used the database developed by us (more than 15000 experimental couples P, T of 56 references) from which we extracted 4304 viscosity values (P,T) corresponding to 1866 experimental values of pure substances and 2213 of mixtures with 216 values of petroleum cuts Our study shows that the absolute average deviation respectively for pressures ≤ 1000 bars and ≤ 5055 bars (7.3% and 10.5%) obtained for pure substances (≥6 carbons) and (6.8 and 7.5%) for mixtures is satisfactory. Moreover when coupled with mixing rules (GRUMBERG-NISSAN and KENDALL-MONROE equations) without adjustable parameters it was able to account satisfactory for the variation of the viscosity (P,T,x) of the mixtures on the simple knowledge of the viscosity of the components of the mixture considered, at reference temperature T0 and at 0.1MPa. For all the mixtures studies an absolute average deviation respectively for pressures ≤1000 bar and ≤5055 bars of 8.6% and 9.8 without binary C1+C10 and 12.5% and 13% with binary C1+C10 was observed. It should also be noted that the self-referencing method seems to be particularly interesting for the restitution of the viscosities of systems containing heavy hydrocarbons in the liquid state and finds its limits for substances on gaseous state or light hydrocarbons (<6 carbons). The method depends also on the quality of the adjustment of the coefficients. If we consider that the coefficients were initially adjusted to moderately heavy substances, it might be interesting to resume the calibration procedure by including lighter substances.