Experimental and simulation studies of the shape and motion of an air bubble contained in a highly viscous liquid flowing through an orifice constriction. (12th October 2019)

• Record Type: Journal Article
• Authors: Hallmark, B.; Chen, C.-H.; Davidson, J.F.
• Is Part Of: Chemical engineering science. Volume 206(2019)
• Page Start: 272
• Page End: 288
• Publication Date: 2019-10-12
• Subjects: Multipass rheometer -- Two-phase flow -- Bubble motion -- Volume of fluid method -- OpenFOAM -- Viscosity averaging rules; Chemical engineering -- Periodicals; Génie chimique -- Périodiques; Chemical engineering; Periodicals; Electronic journals; 660
• DOI: 10.1016/j.ces.2019.04.043
• Languages: English
• ISSNs: 0009-2509

Abstract

Graphical abstract: Highlights: Gas bubbles in highly viscous fluids can take a wide variety of bubble shapes. Initial predictions of bubble shape & velocity differ from experimental measurements. A new viscosity averaging rule is presented and implemented within OpenFOAM. Simulation accuracy depends on mesh resolution and viscosity averaging method. Additional insight into bubble structure can be obtained from validated simulations. Abstract: This paper reports an experimental and computational study on the shape and motion of an air bubble, contained in a highly viscous Newtonian liquid, as it passes through a rectangular channel having a constriction orifice. The magnitude of the viscosity ratios, λ, and capillary numbers, Ca, explored is high: 5.5 × 10 5 < λ < 3.9 × 10 6 and 2.9 < C a < 35.9 respectively. A multipass rheometer is used for the experimental work: air bubbles are suspended in 10 Pa s and 70 Pa s polybutene viscosity standards and passed through an orifice-plate geometry constructed within an optical flow-cell. High levels of bubble distortion are observed, including bubbles that resemble 'crescent moons'. Simulation work is carried out using an implementation of the volume of fluid method in the freely-available finite-volume computational fluid dynamics code OpenFOAM. Quantitative data pertaining to the motion and shape of the bubble was extracted from both the experimental and simulation work. Initially, a good match between numerical simulation and experimental work could not be obtained: this problem was alleviated by changing the viscosity averaging method from an arithmetic mean to a logarithmically-weighted arithmetic mean. Medium- and high-resolution simulations using this new viscosity averaging method were able to match experimental data with coefficients of determination, R 2, typically 0.898 < R 2 < 0.985 .