Abstract
A methodology for the design of fault‐tolerant control systems for chemical plants with distributed interconnected processing units is presented. Bringing together tools from Lyapunov‐based nonlinear control and hybrid systems theory, the approach is based on a hierarchical architecture that integrates lower‐level feedback control of the individual units with upper‐level logic‐based supervisory control over communication networks. The local control system for each unit consists of a family of control configurations for each of which a stabilizing feedback controller is designed and the stability region is explicitly characterized. The actuators and sensors of each configuration are connected, via a local communication network, to a local supervisor that orchestrates switching between the constituent configurations, on the basis of the stability regions, in the event of failures. The local supervisors communicate, through a plant‐wide communication network, with a plant supervisor responsible for monitoring the different units and coordinating their responses in a way that minimizes the propagation of failure effects. The communication logic is designed to ensure efficient transmission of information between units, while also respecting the inherent limitations in network resources by minimizing unnecessary network usage and accounting explicitly for the effects of possible delays due to fault‐detection, control computations, network communication and actuator activation. The proposed approach provides explicit guidelines for managing the various interplays between the coupled tasks of feedback control, fault‐tolerance and communication. The efficacy of the proposed approach is demonstrated through chemical process examples. © 2005 American Institute of Chemical Engineers AIChE J, 2005