Solution of Chance-Constrained Mixed-Integer Nonlinear Programming Problems

Abstract In this contribution a framework for the solution of chance-constrained MINLP problems is described and tested to solve of process synthesis problems with strongly nonlinear and non-convex subsystems. The framework can handle the appearance of non-monotonic relationships between uncertain inputs and chance-constrained outputs, the appearance of multiple roots in the chance constraint evaluation, and performs extensive result recycling to ensure a robust performance despite the structural changes implemented by the MINLP optimization solver. The framework can be interfaced with optimization and simulation solvers programmed in C++, Fortran, and Python. As a first application the process synthesis of the oxidative coupling of methane with a focus on the removal of carbon dioxide from the product stream is investigated.