Laser doppler velocimetry measurements in the Limerick bubbly flow rig

The reliable reproduction of the complex gas-liquid multiphase dynamics of large-scale bubble columns is quite challenging for two-fluid Computational Fluid Dynamics (CFD) simulations due to insufficient accurate experimental data available to improve the hydrodynamic models underpinning these simulations and to validate the simulation results.

The primary goal of this thesis was to efficiently employ the Laser Doppler Velocimetry (LDV) technique to study the liquid-phase velocity characteristics in the near-wall region of a large-scale rectangular bubble column (LimBuRig) to create a reliable extensive database for the CFD community for validating various 2-fluid CFD models (used to simulate bubbly flows in bubble columns) and provide suggestions for modifying those. To achieve these objectives, initial LDV measurements were extensively carried out on a small-scale 70x70 mm transparent square flow channel equipped with a single-needle sparger. Conducted mostly during the pandemic times, these experiments were aimed at gaining hands-on experience with the LDV data acquisition process for bubbly flows.

After that unique experiments were performed to acquire extensive quantitative data on liquid velocities in a wall layer of the pilot-scale Limerick Bubbly Flow Rig which is 2.5 m in height and has a horizontal cross-sectional area of 400 mm x 200 mm. The liquid velocity data were obtained using 2-D Laser Doppler Velocimetry along several horizontal lines at distances up to 5 mm from the front wall, with and without any co-flow of liquid. Different uniform and non-uniform flow configurations were created by varying the aeration rates and/or liquid co-flow conditions on two square sections on the bottom part of the column, thus allowing the study of the interaction of two different bubbly flows at a certain height above the gas distributors.

Data collection was challenging due to the frequent obstruction of the laser beams by the bubbles, noise due to multiple scattering events in the system, and the uneven distribution of seeder particles in the flow, thus giving rise to a very poor data rate at most locations. At a relatively small proportion of the measuring points, the data rate was sufficiently high to allow the reporting of meaningful velocity data including PDFs, spectra, and ACFs.

For both with and without liquid co-flow, the velocity pdf histograms of liquid velocities follow a Gaussian-type distribution centered around 0 m/s, suggesting no net liquid circulation at these locations.

At most locations, when the data rate is considerably high (~500-1000 Hz) the power spectral densities of liquid velocity fluctuations represent an energy-cascading process (obeying an almost -5/3 power law with deviations at certain frequencies). At other locations, no energy cascade is noticed from the power spectral densities.

So we can conclude that near the front walls, a liquid boundary layer has formed and shows locally and intermittently a quasi-turbulence type process where large-scale non-coherent bursts penetrate to take part in the energy transfer process from bigger to smaller scales.