Development of measurement techniques based on image analysis for multiphase flows

Multiphase flows are ubiquitous in industrial processes, yet their inherent complexity makes accurate characterization challenging, often relying on empirical methods. This PhD thesis, conducted in collaboration with Nuovo Pignone - Baker Hughes, addresses the industrial need for advanced monitoring by developing two novel, non-intrusive techniques for multiphase flow characterization based on image analysis. The research aims to enable direct installation of a measurement system within a gas turbine enclosure, imposing stringent constraints on size and robustness. The first technique is a novel laser sheet attenuation method for the quantitative analysis of bubbly flows. By analyzing light intensity from a fast camera along two pixel lines, the algorithm resolves individual bubbles, measuring their velocity and estimating their volume by approximating their shape as an oblate ellipsoid. In its early development, the method demonstrated promising results, with uncertainties below 10$\%$ for bubble velocity and volume compared to reference values. The second methodology employs Electrical Impedance Tomography (EIT) for the qualitative identification of flow regimes in horizontal pipes. A combined approach using time-resolved 1D and temporally averaged 2D conductivity reconstructions successfully discriminated between stratified, slug, and bubbly flows in gas/liquid systems. This EIT-based approach proved particularly effective in opaque foam-laden flows, where traditional optical methods fail, and was also extended to provide qualitative analysis of more complex three-phase flows. Both techniques represent significant steps towards a compact, in-situ measurement system. Future work will focus on merging these methods to create a comprehensive Multiphase Flow Meter. This integrated system would simultaneously identify flow regimes and extract dispersed phase characteristics, providing a fundamentally different, entirely non-intrusive solution that reduces measurement uncertainty by preserving the natural flow pattern, thereby enhancing the optimization and safety of industrial processes.