(21b) Modelling and Experimental Study of Viscosity and Density of Live Fluids at Reservoir Conditions
AIChE Annual Meeting
2024
2024 AIChE Annual Meeting
Engineering Sciences and Fundamentals
Thermodynamic and Transport Properties Under Pressure
Sunday, October 27, 2024 - 3:50pm to 4:10pm
Transport properties play a key role in both geological carbon storage in depleted hydrocarbon reservoirs and CO2-enhanced oil recovery. A thorough understanding of the viscosity and density of live reservoir fluids is necessary as these properties strongly influence fluid mobility, thereby affecting convective flows of fluids in porous geological formations.
In this study, a bespoke vibrating wire viscometer and a commercial vibrating-tube densimeter and were used for the simultaneous measurements of viscosity and density at temperatures between (298 to 423) K at pressures up to 75 MPa. The estimated expanded relative uncertainties are 2 % for viscosity and 0.2 % for density with a coverage factor of 2. The experiments were conducted on binary and ternary mixtures containing either decane or hexylbenzene with dissolved CO2 and/or CH4. The measurements were made in the single-phase compressed fluid region at various mole fractions of the dissolved gas, which was CO2, CH₄ or an equimolar mixture of the two. Subsequently, the results were correlated using the modified Tait equation for density and the Tait-Andrade and Vogel–Fulcher–Tammann equations for viscosity as functions of temperature and pressure. The data were precisely captured by both correlations, with deviations within the estimated experimental uncertainties.
The predictive capabilities of the residual entropy scaling approach were assessed by extending the universal correlation developed by Binti Mohd Taib and Trusler [1] which maps the scaled reduced viscosity onto a hypothesised mono-variant function of scaled residual entropy by means of two substance-dependent scaling factors. The performance of this approach was assessed for the mixtures studied using both predictive mixing rules for scaling parameters and fitting. The multi-parameter Helmholtz energy equations of state and the multi-fluid Helmholtz Energy approximations were used to calculate residual entropy for pure substances and mixtures, respectively. The effects of using less accurate thermodynamic models will also be considered. Lastly, the potential for enhancing the predictive accuracy of the residual entropy model for complex substances, including branched and isoalkanes, will be investigated through the integration of machine learning techniques with chemical descriptor methodologies.
In this study, a bespoke vibrating wire viscometer and a commercial vibrating-tube densimeter and were used for the simultaneous measurements of viscosity and density at temperatures between (298 to 423) K at pressures up to 75 MPa. The estimated expanded relative uncertainties are 2 % for viscosity and 0.2 % for density with a coverage factor of 2. The experiments were conducted on binary and ternary mixtures containing either decane or hexylbenzene with dissolved CO2 and/or CH4. The measurements were made in the single-phase compressed fluid region at various mole fractions of the dissolved gas, which was CO2, CH₄ or an equimolar mixture of the two. Subsequently, the results were correlated using the modified Tait equation for density and the Tait-Andrade and Vogel–Fulcher–Tammann equations for viscosity as functions of temperature and pressure. The data were precisely captured by both correlations, with deviations within the estimated experimental uncertainties.
The predictive capabilities of the residual entropy scaling approach were assessed by extending the universal correlation developed by Binti Mohd Taib and Trusler [1] which maps the scaled reduced viscosity onto a hypothesised mono-variant function of scaled residual entropy by means of two substance-dependent scaling factors. The performance of this approach was assessed for the mixtures studied using both predictive mixing rules for scaling parameters and fitting. The multi-parameter Helmholtz energy equations of state and the multi-fluid Helmholtz Energy approximations were used to calculate residual entropy for pure substances and mixtures, respectively. The effects of using less accurate thermodynamic models will also be considered. Lastly, the potential for enhancing the predictive accuracy of the residual entropy model for complex substances, including branched and isoalkanes, will be investigated through the integration of machine learning techniques with chemical descriptor methodologies.
References
- M. Binti Mohd Taib, J.P.M. Trusler, J. Chem. Phys. 152, 164104 (2020). https://doi.org/10.1063/5.0002242