Robust optimization of periodically operated reactors with stability constraints

This contribution presents a method for the nonlinear programming based optimization of periodically operated reactors with uncertain parameters. The method uses normal vectors to stability boundaries to ensure a safe distance to these boundaries in the space of the system parameters. While an optimization without normal vector constraints may result in a point of operation that indeed is economically optimal, but unstable, optimization with the normal vector constraints guarantees a robust stable optimal result. The normal vector constraints ensure robust stability even if some or all of the model parameters of the optimized dynamical system are uncertain. An example of a simple chemical reactor that admits periodic operation is considered. We use the normal vector constraints in order to avoid unstable oscillations. The resulting optimal robust stable mode of periodic operation is compared to the steady state operation. A measure for the cost of stability and robustness is given.