Newcastle University School of Chemical Engineering and Advanced Materials
 

ADVANCED PROCESS CONTROL

by: Mark J. Willis & Ming T. Tham

© Copyright

CONTENTS

  SUMMARY
1. WHAT IS ADVANCED CONTROL?
2. PROCESS MODELS
 
2.1. Mechanistic Models
2.2. Black Box Models
2.3. Qualitative Models
2.4. Statistical Models
3. MODEL BASED (MODERN) AUTOMATIC CONTROL
 
3.1. PID Control
3.2. Predictive Constrained Control
3.3. Multivariable Control
3.4. Robust Control and the Internal Model Principle
3.5. Globally Linearising Control
4. STATISTICAL PROCESS CONTROL
 
4.1. Conventional SPC
4.2. Algorithmic SPC
4.3. Active SPC
5. DEALING WITH DATA PROBLEMS
 
5.1. Inferential Estimation
5.2. Data Conditioning and Validation
5.3. Data Analysis
6. HIGHER LEVEL OPERATIONS
 
6.1. Process Optimisation
6.2. Process Monitoring, Fault Detection, Location and Diagnosis
6.3. Process Supervision via Artificial Intelligence Techniques
7. ADVANCED CONTROL
8. CURRENT RESEARCH AND FUTURE TRENDS
  BIBLIOGRAPHY
  APPENDIX A:
Examples of reported applications
 
  Control
AC1 Reactors
AC2 Separation processes
AC3 Power systems
AC4 HVAC systems
  Optimisation
AO1 Reactors
AO2 Separation Processes
SUMMARY

This report takes a non-technical look at the state-of-the-art in modern control engineering, focusing on techniques that are applicable to the process industries. As the rate of development in this field is phenomenal, the review is not exhaustive. What we have done is to draw upon the experiences of the Advanced Process Control Group at the School of Chemical Engineering and Advanced Materials, Newcastle University. which has been extensively involved in the fundamental development and application of modern control methods for nearly two decades.

It is also well known that any improvement in the performance of control strategies will result in more consistent production, facilitating process optimisation, hence less re-processing of products and less waste.

Process models underpin most modern control approaches. Depending on the model forms, different controllers can be synthesised. Even the prevalent Proportional+Integral+Derivative (PID) algorithm can be designed from a model based perspective. The performance capabilities of PID algorithms are limited though. More sophisticated strategies, such as adaptive algorithms and predictive controllers have been proposed for improved process control. Due to the emphasis on Quality, Statistical Process Control (SPC) techniques are also experiencing a revival. In particular, attempts are being made to integrate traditional SPC practice with engineering feedback control techniques. Each of these strategies possesses respective merits. Of special significance is the recent attention paid to developing practicable nonlinear controllers, in recognition of the fact that many real processes are nonlinear and that adaptive systems may not be able to cope with significant nonlinearities. There are two approaches. One attempts to design control strategies based on nonlinear black box models, e.g. nonlinear time-series or neural networks. The other relies on an analytical approach, making use of a physical-chemical model of the process. However, there are indications that the two approaches can be rationalised. Cheap powerful computers and advances in the field of Artificial Intelligence are also making their impact. Local controls are increasingly being supplemented with monitoring, supervision and optimisation schemes; roles that traditionally were undertaken by plant personnel. These reside at a higher level in the information management and process control hierarchy. Performing tasks that relate directly to overall plant management objectives, they effectively link plant business objectives with local unit operations. The result is an environment that is conducive to more consistent production.

Modern process plants, designed for flexible production and to maximise recovery of energy and material, are becoming more complex. Process units are tightly coupled and the failure of one unit can seriously degrade overall productivity. This situation presents significant control problems. The literature on relevant control, monitoring, supervision and optimisation techniques is voluminous, each article exhorting a certain solution to a particular problem. However, it is generally acknowledged that there is currently not one technique that will solve all the control problems that can manifest in modern plants. Indeed, different plants have different requirements.

A systematic studied approach to choosing pertinent techniques and their integration into a co-operative management and control system will significantly enhance plant operation and profitability. This is the goal of advanced process control.


Next Last

 

Copyright 1994-2009
Mark Willis and Ming Tham
School of Chemical Engineering and Advanced Materials
Newcastle University
Newcastle upon Tyne
NE1 7RU, UK.

 
Please email any link problems or comments to ming.tham@ncl.ac.uk