(461a) Proactive Fault-Tolerant Model Predictive Control

Fault-tolerant control methods have been extensively researched over the last ten years in the context
of chemical process applications and provide a natural framework for integrating process monitoring and
control aspects in a way that not only fault detection and isolation but also control system reconfiguration
is achieved in the event of a process or actuator fault. The motivation for this broad research direction is
provided by the strong industrial need to handle process and control system faults effectively before they turn into
major accidents. Most of the work in the area of fault tolerant control in the last few years is in the context of active
fault tolerant control where a fault tolerant control system is designed to detect and isolate the fault and then automatically
reconfigure the control system to maintain operation as close as possible to optimal conditions with the remaining control actuators.

In this paper, we embark into a new direction in the area of fault tolerant control; namely, we try develop algorithms that
calculate in real-time from process data and manufacturer guidelines the probability of a process control system component to fail,
and in the event this probability exceeds a certain threshold we proactively determine how to reconfigure the control system to guide
the process state in an operational mode that allows to fix or replace the component that is about to fail. We formulate and solve this
problem in the context of nonlinear chemical process models and use model predictive control as the feedback design technique. We use
results from stochastic stability to assess the stability of the closed-loop system. We will use chemical process examples to evaluate the
effectiveness of the approach using Monte-Carlo simulations.