Bubble's rise characteristics in shear-thinning xanthan gum solution: a numerical analysis

Abstract

The rise of gaseous bubbles in shear-thinning liquid is a fundamental issue in fluid physics, particularly the bubbles rise dynamics in water-soluble xanthan gum concentration have a strong link for enhancing the stability of foams that have been encountered extensively in oil recovery, methane hydrate formation processing, froth flotation, and food and beverage industries. Here, air bubble rise behavior in xanthan gum solution (XGs) was investigated by using a coupled volume of fluid with the level set approach in CFD (computational fluid dynamics) modelling. The Carreau-Yasuda rheological model was adopted to define the viscous properties of the XGs and the continuum surface force model was used to track the interface between bubble and XGs. Additionally, ANN (artificial neural network) modeling was demonstrated for estimating the outputs of CFD. Our estimated CFD results for different bubble terminal velocities in water and experimental data obtained from the literature showed that there was a maximum relative error of 4.51%. Then, the CFD setup was utilized to investigate the effect of different concentrations of XGs and liquid flow index (N) on bubble rise dynamics within the bubble Reynolds number (Reb) range up to 10.05 and Eotvos number (Eo) range up to 3.47. For a fixed bubble size, the dimensionless bubble terminal velocity decreased in increased flow index and concentration of XGs, which led to decrease Reb and increase Eo. For a given XGs, the dimensionless bubble terminal velocity significantly depends on Reb and Eo. It was found that XGs and flow index significantly affect the distribution of dimensionless liquid viscosity than that of the dimensionless liquid velocity close to the bubble's underneath region. In a comparison with the rheological power-law model, the Carreau-Yasuda model showed to predict more accurate results. The estimated drag coefficient showed a deviation from the empirical equation reported in the literature, in contrast, a more accurate estimation in drag coefficient was obtained based on modified Reynolds number and Archimedes number. ANN modelling outputs agreed with CFD results and indicated that the training and testing of ANN have great efficiency to predict unknown values.