Tomographic identification of gas bubbles in two-phase flows with the combined use of the algebraic reconstruction technique and the genetic algorithm.

Combined use of the algebraic reconstruction technique (ART) and the genetic algorithm (GA) shows highly accurate and efficient tomographic reconstruction of line-of-sight projection images of two-phase flows compared with reconstructions obtained by separate use of these methods. A modified GA-based tomography uses the ART reconstruction result as preliminary information on the number, shapes, and sizes of bubbles to be reconstructed. This combined use of the two methods exploits faster convergence of the ART to the approximate solution space and more robust and accurate optimization of the GA to the ultimate solution space. In the investigation a computer-synthesized phantom field that consisted of five elliptical gas bubbles in liquid or solid surroundings was used.