Nanoparticles Stabilized Oil Water Emulsion Flow

Abstract

Abstract Nanoparticles (NP) stabilized oil-water emulsion flow is studied experimentally and theoretically for both hydrophilic and hydrophobic NP, under Water Sweep (WS) and Oil Sweep (OS) conditions. Novel and highly accurate experimental data are collected for different NP concentrations and different pump speeds. The Legacy-Texaco skid flow loop is modified to enable data acquisition, and the microscopic camera is modified with Vector Control Module (VCM), as well. The phase inversion is measured utilizing a Circular Differential Dielectric Sensor (C-DDS), a Rectangular DDS (R-DDS), pressure transducer and mass flow meter. Droplet size distribution data are acquired with a microscopic camera for dilute dispersed-phase volume fractions. Data are acquired from all 5 instruments simultaneously. Hydrophilic NP, even at low concentrations, consistently delay the occurrence of phase inversion as dispersed phase fraction is increased. Concentration of 1 wt% hydrophilic NP delays the phase inversion up to 93% of organic volume fractions for o/w dispersions. The transient phase inversion period is shorter (1.5 to 2 min.) for hydrophilic NP emulsions; longer phase inversion durations (greater than 2 min) occur for the hydrophobic NP emulsions. With increase of hydrophobic NP concentration, the viscosity of mineral oil increases significantly. Hydrophobic NP stabilized emulsions (in higher concentrations) exhibit shear thinning behavior indicating that yield stress must be applied in order to initiate flow. Concentrations of 0.05 and 0.11 wt % of Hydrophobic NP cause early phase inversion at all 3 pump RPMs. However, at concentration of 0.5, 0.8 and 1 wt % lead to delay in phase inversion. Extension of Pereyra (2011) correlation for dMAX is carried out for NP stabilized emulsion flow. Log Normal distribution and modified Rosin Rammler distribution are recommended for prediction of the cumulative droplet size distributions of NP stabilized emulsions for OS flow. Optimized parameter sets are generated from experimental data for in both these distributions. Comparison between the acquired data and both log normal distribution and Rosin Rammler distribution predictions show Normalized Root-Mean-Square Deviation (NRMSD) discrepancy values less than 13%.