Modeling the phase behaviour of bitumen/n-alkane systems with the cubic plus association (CPA) equation of state

Abstract

The cubic-plus-association equation of state was applied to model the phase behaviour of bitumen/n-alkane systems including saturation pressures, liquid-liquid boundaries, yields, and phase compositions. Yield is defined here as the mass of bitumen in the heavy phase divided by the mass of bitumen in the feed. To implement the model, the bitumen was divided into a set of pseudo-components based on a distillation assay and either the n-pentane insoluble content of the oil (CPA-C5 approach) or the propane insoluble content (CPA-C3 approach). The pseudocomponents in the solvent insoluble part of the oil were defined as self-associating components, all other pseudo-components were non-associating. The critical properties and acentric factor for each pseudo-component were determined from established correlations. A set of CPA parameters was then developed to fit the available phase behavior data. The self-associating pseudocomponents were assigned a distribution of self-association energies in order to capture the sequential partitioning of asphaltenes to the heavy phase upon solvent addition or in different solvents.Both approaches matched the phase behavior data for mixtures of bitumen with n-pentane and higher carbon number n-alkanes almost to within the experimental error. The CPA-C3 approach also matched the phase behavior data for mixtures of propane and bitumen. The CPA-C5 approach could not match the yield data for propane diluted bitumen but was more straightforward to implement and was less computationally intensive because it employed fewer self-associating components. To apply either approach to another oil, only the self-association energy of the selfassociating pseudo-components need be adjusted. The cross-association energy between the solvent and the self-associating pseudo-components must be tuned for any new solvent.

Extending the Modified Regular Solution Model To Predict Component Partitioning to the Asphaltene-Rich Phase

“…The CPA EoS approach assumes that polar interactions and hydrogen bonding control asphaltene phase behavior and uses a non-zero association term. The asphaltenes are treated as monomers with a uniform average molecular weight. , Both approaches can fit phase boundaries, onset, asphaltene yields, and heavy phase compositions. The main disadvantage of both approaches is their limited predictive capability because their parameters (particularly the asphaltene parameters) must be tuned using asphaltene phase behavior data specific to the given oil.…”

Regular solution theory applied to asphaltene related phase behaviour

“…The advantages of this approach are that it: is accurate over a broad range of temperatures, pressures, and compositions; is quantitatively predictive for mixtures of n ‐alkanes and oils; is qualitatively predictive for mixtures of other solvents and oils; requires a limited amount of input data (for dead oil and solvent, a SARA or distillation assay plus C5‐asphaltene content, solvent content, temperature, and pressure); follows a straightforward oil characterization procedure; is easily tuned to available data; and is suitable for process simulators, particularly the distillation‐based version of the model. The MRS model is also computationally faster than equation of state models because the partitioning coefficients are explicit and do not require the equation of state roots or derivatives for their calculation. A comparison of the computational efficiency was performed between the CPA equation of state [ 41 ] and the distillation‐based version of the MRS model as they use a similar fluid characterization approach, and both have been used to model asphaltene and pitch* yield precipitation. The flash calculation speed of the MRS model was 300 flashes per second compared with 50 flashes per second for the CPA model with the same fluid and conditions.…”

“…CPA has been tuned to match saturation pressures, asphaltene onsets, and heavy phase amounts and compositions, but the number of tuned parameters is double that of a cubic equation of state. [ 39–41 ] PC‐SAFT has also been tuned to match similar data without making use of its association term. [ 19,42,43 ] CPA and PC‐SAFT provide the best match to the complete set of phase behaviour data but both require a more complex flash calculation and must be tuned for each oil and solvent.…”

Understanding bitumen partial upgrading through process modelling and simulation

“…The fundamental problem is that the pseudo-components representing the oil phase do not have a sufficiently defined structure to generate meaningful binary interaction coefficients for liquid-liquid equilibrium calculations, in addition to the inability of this equation to capture the association of asphaltene molecules. There are other thermodynamic models outside the current simulation environment, such as the regular solution model or the cubic-plus-association equation of state, [24][25][26] that might be better for this purpose.…”