(452a) Modelling the Separation of Oil and Water in Pipelines

Study of oil and water mixing has a long history and surprisingly it is still of great interest in many industrial applications. Mixing and separation occur in the process constantly and the final state is usually determined by agitation, phase properties and interfacial physics. Application of CFD to solve the problem introduces several multitude of complexities, including the 3-dimensional variables, numerical convergence, droplet size distribution, interfacial properties, etc. A seemingly simple problem can become intangible very quickly and that's why semi-empirical 1-D correlations dominate the industrial applications overwhelmingly. In oil and gas industry, it is often argued that the high energy input at the subsea choke tends to generate tiny water droplets and such tiny droplets are stable in the turbulent flow field, which would minimize the likelihood of corrosion caused by water wetting of the pipe walls, and thus the limitations of semi-empirical models do not matter. But is this true?

More detailed modelling of oil–water flow has been performed to address this and identify any limitations of the semi-empirical approach. Such modelling involves representing water–oil interfacial phenomena and the interactions between droplets, i.e., break-up and coalescence. Traditional computational fluid dynamics (CFD) approaches use either the Eulerian method, where initial droplet size information is assumed, or the Lagrangian method, where individual droplets are tracked. The latter is usually restricted to low water concentrations to avoid expensive inter-droplet collision modelling. Recent advances in discrete element models show promise but creating these is computationally expensive.

Instead, our approach uses population balance models, which avoid those issues by including droplet size as an Eulerian variable [Ref. 1 and 2] and modelling its changes with collision probability and interfacial physics using the local turbulence information inherent to CFD. Note that gas flow was not included in the model, as it would add significantly to the complexity. This is an extension to the paper we presented last year with added validations to other experimental results and the comparison to empirical models. The results showed the model has good accuracy over a range of fluid properties and water concentrations. The use of the interfacial area concentration (IAC) population balance model [Ref 1] shows the advantage of including droplet size as a variable and demonstrates the necessity to have a simpler model with consistent predictions. The model was calibrated with the experimental work of Sommons et al. (2001) [Ref 3] and validated with the experimental results of Fairuzov et al. (2001)[ref 4] and Kee et al. (2016)[Ref 5]. The definitions of flow regimes and surface water-wetting are laid out for CFD results which can be useful for more precise quantification in other applications.

[Ref 1] Wu Q., Kim, S. and Ishii, M.: “One-group interfacial area transport in vertical bubbly flow,”

[Ref 2] Chesters, A. K.: “The modelling of coalescence process in fluid–liquid dispersions: A review of current understanding,”

[Ref 3] Simmons, M. J. H. and B. J. Azzopardi: “Drop size distribution in dispersed liquid–liquid pipe flow,”

[Ref 4] Fairuzov Y. V., Arenas-Medina, P., Verdejo-Fierro, J. and Gonzales-Islas, R.: “Flow pattern transitions in horizontal pipelines carrying oil-water mixtures: Full-scale experiments,”

[Ref 5] Kee, K. E, Richter, S., Babic, M. and Nesic, S.: "Experimental study of oil-water flow patterns in a large diameter flow loop - the effect on water wetting and corrosion," Corrosion (2016) 72(4), 569-582