Towards a Parallel Multiphase Solver Based on Potential Ordering

Abstract

This thesis presents our work towards developing a parallel multiphase solver based on potential ordering. We begin the thesis by introducing the Fast Multiphase Solver, as developed by Natvig, Lie et al. . Then we, in turn, study the parallel algorithms developed by Fleischer et al. and Bader. As a part of the study we have implemented the algorithms and give an overview of these implementations. Our implementation of the algorithm due to Fleischer confims the serial complexity but are unable to achieve parallel speed-up. Results based on our implementation of Bader's algorithm discourage further development with this approach. Finally we discuss further possibilities and propose our own ideas on how to adapt the parallel algorithms for use in a parallel multiphase solver.