AmericanChemicalSociety.com

Washing of oil-contaminated soils, extracting bitumen from oil sands, recovery of oil from storage tank sludge and washing of drill cuttings containing oil-based drilling fluids are processes that depend on the principles of oil- particle-water separation. The following is a summary of some of the key findings of computer modeling and experimental approaches to understand oil-particle-water separation.

A computational fluid dynamics (CFD) model has been developed to simulate the separation of oil from a sand particle via the external shear of water.  The flow is assumed axisymmetric, and so the model has been implemented in a two dimensional spherical coordinate system.  The governing Navier-Stokes equations are discretized using the finite volume method.  The oil/water interface is tracked using the Volume of Fluid (VOF) method that is usually implemented in a Cartesian coordinate system; our implementation in a spherical system is novel.  The finite volume flow solver for a single fluid was validated by simulating the uniform flow over a solid sphere, comparing the length of the wake, the vortex position, and the separation angle at various Reynolds numbers, with published results.  The VOF interface tracker was validated by reconstructing a number of simple interface geometries, and by advecting interfaces in various specified velocity fields, and then comparing results with known answers.  The complete flow model has been validated by predicting the equilibrium positions of a fluid drop that partially wets a solid particle under gravity and comparing these configurations with published experimental results. One article has been submitted to the Journal of Computers and Fluids dealing with the VOF implementation in spherical coordinates.

Turning to the application of the model to predicting the separation of oil from coated particles immersed in an aqueous shear flow, we have obtained preliminary results that predict the fraction of oil retained by the particle, as a function of various operating conditions and fluid properties.  We first considered the following base case scenario:  the properties of oil were assumed the same as those of water – density ρ=1000kg/m³ and dynamic viscosity μ=0.001kg/ms; an oil/water interfacial tension γ = 10mN/m; a contact angle α = 90°; a solid particle diameter 0.1mm; an initial oil volume fraction of 20%; and a flow velocity of 0.01m/s, which is on the order of the terminal velocity of a sand particle in water.  This base case corresponds to a capillary number Ca=0.001 and a Reynolds number (based on solid particle diameter) Re=1.  The particle is initially assumed to be uniformly coated; in the presence of the shear flow, a contact line is assumed to form when the film thickness at the leading edge thins to less than 2.5 microns.  For this case, the model predicts that all of the oil is retained and that an equilibrium configuration exists.  This suggests that the shear forces are not strong enough to overcome the interfacial tension holding the oil together, or on the particle.  Next, γ was reduced to γ =0.01mN/m, at which point the oil film behaves very differently.  At this value of γ, the effect of external shear is comparable to the interfacial tension, and deforms the oil drop such that a tail forms at the rear, that ultimately ruptures.  The same simulations were run for a smaller contact angle of 30°.  For γ =10mN/m, again an equilibrium configuration exists, but the oil wets a larger fraction of the particle surface than when the contact angle is 90°.  This suggests that the lower the contact angle, the more difficult it will be to remove the oil.   The validation work and the preliminary simulations described above and others were presented at the World Congress of Chemical Engineering in Montreal, Canada (August, 2009). We are preparing an article on the subject to be submitted to the Canadian Journal of Chemical Engineering.

In order to validate the CFD model a single particle flow chamber was constructed using a polycarbonate enclosure. A 5 mm metal particle was tethered to a flow distributor located on the top of the chamber. The oil was coated with a neutrally buoyant mixture of bitumen, chloroform and toluene. Due to design limitations (large pressure drops) Re< 1, a condition that, as predicted by the CFD model, does not induce oil detachment.  We are in the process of producing an alternative experimental setup involving free falling particles. 

We have also pursued an alternative approach involving batch soil washing experiments. Beach sand was sieved and contaminated using bitumen (2.5, 5 and 10 wt%) mixed with the toluene and left in a fume hood in order to evaporate the toluene and age the sand. This contaminated sand was then washed using either water (γ ~ 10mN/m) or a 0.1% sodium dihexylsulfoccinate (SDHS, anionic surfactant) solution at optimal salinity (γ ~ 0.01mN/m).Toluene was used as a solvent to aid in the separation.  Consistent with the predictions of the CFD model, using the surfactant formulation (γ ~ 0.01mN/m) and a small volume of solvent (to reduce the viscosity of the bitumen) it was possible to remove more than 97% of the bitumen from the sand. Using water and a large volume of solvent (γ ~ 10mN/m), more than 95% of bitumen was removed.  This is also consistent with CFD model predictions for high interfacial tension systems with high oil (or oil + solvent) to particle volume ratios.  These batch studies were presented at the 2009 World Congress of Chemical Engineering in Montreal, Canada (August, 2009).  

The next step in this work is to run many computational and batch washing experiments, also varying the oil viscosity and the initial volume fraction of oil, in order to define a "process map" that identifies the oil film behaviour as a function of the Reynolds and capillary numbers.