This study set out to gain a better understanding on the viscosity and diffusion coefficients of hydrocarbons with dissolved CO2 and CH4, relevant to enhanced oil recovery and carbon storage. Measurements of the viscosities and diffusion coefficients of the mixtures have been done at high temperature and pressure conditions where carbon storage happens, typically at temperature between (298 and 473) K and pressure up to 100 MPa.
Viscosity measurements were performed using two different types of viscometers where density was measured simultaneously. The viscosity of m-xylene + CO2 were measured using VW-VD apparatus where the wire was tensioned vertically by a sinker while the viscosity of DIDP + CO2 were measured using VW-VT apparatus where the wire was fixed at both ends horizontally. The viscosity of mixtures of m-xylene + CO2 involved were measured with mole fractions of carbon dioxide between 0 to 0.652 while for DIDP + CO2 mixtures, the mole fractions of CO2 studied were xCO2 = 0.207, 0.411, 0.610 and 0.810. The experimental viscosity of the mixtures was correlated using the Tait-Andrade equation and excellent fits were obtained with ΔAAD ≤ 1.0% for both systems. Surface fits as functions of temperature, pressure and compositions was also done for the m-xylene + CO2 system with a total of 18 parameters for viscosity and 13 parameters for density. All data points fit well with the correlation with ΔAAD = 1.9 % for viscosity and ΔAAD = 0.3 % for density except for density data at x = 0.464 where the values deviate greater than the other mole fractions and noticeably greater than their uncertainty. These correlations allow the interpolation of data with respect to temperature, pressure and composition which enables the comparison with the available experimental data in the literature reported at other conditions.
A simple model for viscosity based on the residual entropy scaling was developed to predict the viscosity of hydrocarbons with dissolved CO2. This model applied the highly-accurate Helmholtz equations of state to calculate the residual entropy of pure substances and the multi-fluid Helmholtz-energy approximation for the residual entropy of mixtures. For pure substances, decane, m-xylene, CO2, dodecane and octane were chosen based on the availability of both reliable wide-ranging viscosity data and a wide-range Helmholtz equation of state. Each substance collapsed on a single curve when two scaling factors were introduced: (1) a horizontal factor h, which scales the residual molar entropy; and (2) a vertical factor Rη, which scales the reduced viscosity, which revealed the mono-variant relationship over an extended range of temperature and pressure. The universal curve was represented by the third order polynomial function and optimised in a global regression where the deviations of the data from the model for each substance are all within a band of ±10 %. The model was expanded to the octane + dodecane system where linear mixing rule was used to calculate the molar mass, M, and the values of h and Rη . The result shows reasonable agreement with absolute average relative deviation of 2%. To investigate the behaviour of the model to asymmetric mixtures, decane + CO2 and m-xylene + CO2 mixtures were considered. Results showed that the model systematically underestimates the experimental data by approximately 30% for both systems with m-xylene + CO2 systems showed slightly better results than decane + CO2 systems. This is a rather encouraging finding considering that only simple mixing rules were used.
Experimental measurement on diffusion coefficient of methane at infinite dilution in methylbenzene and in heptane were done at temperatures ranging from (323 to 398) K and at pressures from 1 MPa up to approximately 65 MPa. The apparatus used was the Taylor dispersion apparatus (TDA) with overall combined uncertainty of 2.3%. Over the temperature investigated, the diffusion coefficient was found to increase more than 90% for all pressure conditions. To add to that, the diffusion coefficients of methylbenzene were observed to be much lower by not more than 14% than the diffusion coefficient of heptane. The experimental data were fitted using a simple empirical model where the parameters were fitted as linear and quadratic functions of temperature resulting in five parameters per solvent. The Stokes-Einstein model was also used to analyse the experimental data that resulted with just two parameters per solvent. Both approaches represent the data with ΔAAD of around 3.5%. The results from this work provide important insights for future measurements on this type of system.
The experimental data on the diffusion coefficients of CH4 in methylbenzene and heptane were used to develop an improved model based on the well-established rough-hard sphere model. Numerous experimental data were retrieved from the literature involving several gaseous solutes in hydrocarbon solvents, together with molecular dynamics simulations (MD) for systems of smooth hard spheres in order to establish a general correlation for gaseous solutes in non-polar liquids. The gaseous solutes considered in this work consists of light hydrocarbons, carbon dioxide, nitrogen, and argon and the liquid solvents were all hydrocarbon liquids. To apply this model to a particular system, two parameters are needed: (1) the molar core volume of the solute and solvent; and (2) the roughness factor. Having to have both parameters, the model may be used to predict the tracer diffusion coefficient over wide ranges of temperature and density. The model was found to correlate the experimental data with an average absolute relative deviation of 2.7%.
The results presented in this thesis extends the knowledge of the viscosity and diffusion coefficients of hydrocarbon liquids with dissolved CO2 at wide range of temperatures and pressures. In addition, it also provides powerful models that can predict the viscosity and diffusion coefficient. The insights gained from this study may be of assistance for future work.