Simultaneous integration of design and operation under uncertainty using dynamic perturbation and chance constraints

Conventional process design determines operating conditions and the size of main components first, and then later develops a control scheme. As this approach may lead to dynamic constraint violations during operation, hinder robust plant performance, or simply reduce the controllability of the process, integrative approaches were developed, which is known as integration of design and operation (or: integration of design and control). However, several challenges for its formulation and solution continue to exist, e. g., how to incorporate the variability of input variables, how to account for parametric uncertainty, or how to determine the length of the time horizon. The latter, in particular, has not been systematically studied so far. Within this contribution, we propose a framework that addresses these severe challenges and apply our framework on the case study of a continuously stirred tank reactor to demonstrate its successful implementation. The results show that both process dynamics and parametric uncertainty can have a pronounced effect on the optimal process design: reactor volume increases by a factor of 7 under uncertainty, by a factor of 1.6 if dynamics are considered, and by a factor of up to 14 if both dynamics and parametric uncertainty are taken into account. • Proposition of a framework for integration of design and operation. • Simultaneous solution of the optimization problem due to full discretization. • Incorporation of dynamic variability via amplitude-modulates pseudo-random binary sequences. • Inclusion of parametric uncertainty via point estimation methods and chance constraints. • Rigorous analysis of the impact of time horizon length on the obtained process design.