Multivariable Control Structure Design of Heat Exchange Networks Based on Mixed-Integer Quadratic Programming

Abstract

Abstract In this work, a new systematic methodology to address multivariable control structure (MCS) design for heat exchange networks (HENs) is proposed. This methodology is based on recent developments presented by Braccia et al. (2017) where the MCS design for medium/large-scale processes is performed based on a mixed-integer quadratic programming (MIQP) formulation. This methodology is based on an equation-oriented superstructure which addresses simultaneously: the selection of controlled variables (CVs) and manipulated variables (MVs), the input-output pairing definition, the controller topology, the initial heuristics considerations, and the stability test by using steady-state information only. Usually, in the HENs cases, all the outputs variables need to be controlled, i.e. there are no uncontrolled variables. In this context, the MCS design procedure presented in Braccia et al. (2017) requires a reformulation to quantify the manipulated variables instead of the uncontrolled variables. Furthermore, some modifications of the original MIQP algorithm related to the augmented functional cost are analyzed. The performance of different controller topologies, i.e. decentralized, sparse, and full, are evaluated. Two examples are given in this work to illustrate the potential of the proposed methodology and to support the final conclusions.