Size Distribution for Oil Droplets Dispersed in Water: An Experimental Study by Image Analysis

Abstract

The work in this thesis is based on image analysis of oil droplets dispersed in water. It uses a public domain image processing program called ImageJ to analyse distribution of oil droplets generated by a breaking wave in a flume and by injecting trough a nozzle in a tank. The droplet distribution will evolve with time, due to buoyant forces in the water. One goal for the work was to get quantitative knowledge on how the distribution evolves and compare these to the expected values for the rising speed. Large droplets is expected to rise faster than small ones, and as a result the droplets will be sorted with time. A second goal was to compare the volume of droplets dispersed through a nozzle to predictions based on existing theory.Some challenges were encountered while developing the design for the tank experiments. The use of a globe valve failed to disperse the oil into droplets. Only by use of direct injection through a nozzle was a distribution of droplets generated. One issue experienced was the resolution of the images used for image analysis. Focus and lighting used in the experiments affected the results, as the functions used to process the images gave large errors when analysing small droplets. For dense droplet distributions the image analysis failed completely because it could not separate droplets from each other. The values found in the tank experiments for droplet volumes and rising speed deviated considerably from the expected values. Together with errors from the handling of the images the oil viscosity and inter facial tension with water could have considerable effect on the droplets. For tests in the wave flume the suggestions made by Grammeltvedt(2013) worked with regards to lighting, but further improvements regarding sideways motion should also be handled.