(240b) Automated Control Structure Design for Complex Energy Integrated Networks
High energy efficiency has become a critical need in modern chemical plants, driven by the increasing energy cost. Energy integration in chemical plants has been an active topic of research, as it enables reduction of external energy consumption, and thus, leads to lower operational costs. However, such integration poses several challenges to the operation and control of a chemical plant, especially in the context of transitions between different operating points. To this end, extensive research activity is currently being pursued on the control of large-scale energy integrated plants.
In our previous work , we developed a generic, scalable graph-theoretic framework which can be used to analyze such plants with segregation of the energy flows in different orders of magnitude. Such plants exhibit multi-time scale dynamics and are amenable to a hierarchical control approach. Control structure design in each time scale is facilitated by the developed graph theory method. On the other hand, in many integrated plants, tight integration does not necessarily result in segregation of the energy flows in different orders of magnitude. Examples include cases whereby a large amount of energy is recovered and used to preheat different process streams through several process units, leading to energy recycle and throughput flows of comparable orders of magnitude. Control structure design in these plants is a non-trivial task which greatly affects their operability and controllability.
To this end, in this work, we propose a framework for the design of multi-loop control structures for general complex energy integrated networks, in an automated manner. We first transform such networks into an energy flow graph, where nodes represent individual process units and edges represent energy flows. Afterwards, an equation graph, which captures the interdependencies among the process variables, is constructed, exploiting the connectivity information of the energy flow graph. Next, the proposed algorithm identifies all the paths connecting the inputs and the outputs in the equation graph, and computes the corresponding relative degrees. Finally, an integer programming (IP) formulation allows obtaining optimal input/output pairing sets, which minimize the “structural coupling”, expressed in terms of the relative degrees, in the network. The optimization formulation can be augmented with constraints that dictate input/output pairs that should be included in or excluded from the control structure based on practical considerations. The method is also extended to allow for the synthesis of block decentralized control structures. The entire method is automated within an object oriented programming environment.
The application of the proposed framework is demonstrated through a case study in a representative chemical plant. Specifically, we consider an energy integrated solid oxide fuel cell (SOFC) system coupled with a reformer for hydrogen production . First, we apply the proposed framework to the network to propose an optimal control structure. Then, we perform closed-loop simulations to illustrate the effectiveness of the proposed control structure, and compare the results with those presented in .
 Seongmin Heo, Sujit S. Jogwar, Srinivas Rangarajan and Prodromos Daoutidis, “Graph Reduction for Hierarchical Control of Energy Integrated Process Networks”, Decision and Control (CDC), 51st Annual Conference on, IEEE, 6388-6393, 2012
 Dimitrios Georgis, Sujit S. Jogwar, Ali S. Almansoori, and Prodromos Daoutidis, “Design and control of energy integrated SOFC systems for in situ hydrogen production and power generation”, Comput. Chem. Eng., 35(9):1691-1704, 2011