Assessing the Controllability of Heat Exchanger Networks

Improving energy efficiency is probably the most generic issue for the process industry today, and process integration is a common tool to successfully achieve this. However, an increased level of process integration, usually implemented by re-circulation and other physical feedback structures, can increase the complexity of the dynamics drastically. This, in turn, implies increased control difficulties and may make implementation unfeasible. There are methods to assess the energy efficiency of existing heat exchanger networks (HENs) and identify modifications of the current design to improve the heat recovery. However, they are generally based on steady state models and give no information about the dynamic performance. As choices concerning stream splits, heat exchanger size and placement may significantly impact the controllability of a HEN, it is of principal importance to not only optimize with respect to cost but also with respect to controllability when designing HENs. As the methods for design or retrofit of HENs may identify several alternative designs with similar economic performance, controllability can be the deciding factor for which solution to implement. Moreover, in some cases failure to take controllability into account may lead to that the proposed design, while perhaps optimal from a steady state economic stand point, may be impossible to control properly. To create a design of a HEN that includes controllability concerns there need to be strong metrics to assess the controllability of a design. When assessing the controllability of a HEN, the relative gain array (RGA) is generally used [1]. The RGA serves two purposes, it recommends an inputoutput pairing for decoupled control of the HEN, while also giving a measure of the HEN’s controllability. However the standard RGA is based on the static gain of the system, so it does not contain information about the dynamics of the system.