Dead-Time Compensation and Performance Monitoring in Process Control

Sammanfattning: The thesis contains two parts, dead-time compensation and performance monitoring. The first part on dead-time compensation is about robust tuning procedures for dead-time compensating controllers (DTC). Both stable and integrating processes are considered. Simple experiments are performed to obtain process models as well as bounds on the allowable bandwidth for stability. The DTCs used have few parameters with clear physical interpretation so that manual tuning is possible. In the second part on dead-time compensation the performance of PID controllers is compared to the performance of DTCs. The aim is to answer the question: ``When can a simple dead-time compensator be expected to perform better than a PID?''. The performance criterion used is the integrated absolute error (IAE). It is compared for PI and PID controllers and a simple dead-time compensator (DTC) when a step load disturbance is applied at the plant input. The topic of the second part of the thesis is performance monitoring of lambda-tuned feedback controllers. A lambda tuned loop is a loop with an one-degree-of freedom controller whose set point response is of first order plus dead time. Lambda is the time constant of the set-point response. For these loops a non intrusive performance monitoring methods is described which is thought to be a first indicator of bad performance after which existing diagnosis algorithms could be applied. For online implementation in distributed control systems a simple recursive algorithm to estimate the index is presented. All parameters of the monitoring method are set by using the lambda tuning. The method applies equally to stochastic or deterministic disturbances. Finally an algorithm to estimate a synthetic gradient of a quadratic cost function is presented. It is demonstrated that the gradient can provide valuable information for maintenance of controllers as the gradient gives information about the disturbances affecting the loop.