A homotopy continuation-based branch and bound algorithm for process synthesis and design using rigorous unit operation models

Abstract Synthesizing chemical processes using rigorous unit operation models often leads to a large-scale strongly nonconvex mixed-integer nonlinear programming problem, which is difficult to solve. In this work, we propose a homotopy continuation-based branch and bound (HCBB) algorithm in which an adaptive variable-step length HC method is used to navigate the NLP subproblem to gradually approach a feasible solution. The computational results indicates that the proposed HCBB algorithm provides a robust convergence to a higher-quality locally optimal solution from different initial points for an example of process synthesis compared to existing general MINLP solvers.