Closed-loop reactor network synthesis with guaranteed robustness

This paper presents a systematic methodology to design closed-loop reactor networks with guaranteed robustness. The methodology is based on the superstructure approach for reactor network synthesis and extends our previous work (Zhao and Marquardt, 2016) to simultaneously design the process and the control system structure. The spectral abscissa of the Jacobian matrix of the closed-loop reactor network is chosen to measure the response speed of the designed process. A mixed-integer nonlinear program (MINLP) with complementarity constraints and a robust eigenvalue constraint is formulated and solved sequentially by a two-step solution strategy. Structural alternatives of the process and the control system as well as parametric uncertainties are considered in an integrated framework. A case study involving continuous stirred-tank (CSTR) and plug flow (PFR) reactors is presented to illustrate the novel approach and compared with an established two-step design approach.