Pergamon 0967-0661 (95)00086-0 Control Eng. Practice, Vol. 3, No. 7, pp. 1017-1021, 1995 Copyright © 1995 Elsevier Science Ltd Printed in Great Britain. All rights reserved 0967-0661/95 $9.50 + 0.00 CONDITION MONITORING AND DIAGNOSIS IN A SOLID FUEL GASIFICATION PROCESS Y. Majanne*, P. Lautala* and R. Lappalainen** *Tampere University of Technology, Control Engineering Laboratory, P.O. Box 692, FIN-33101 Tampere, Finland **Enviropower Inc., P.O. Box 35, FIN-33701 Tampere, Finland (Received September 1994; in final form May 1995) Abstract: The goal of the project is to define and develop a condition monitoring and diagnos- tic system for the solid fuel gasification process and implement the system in a standard digital automation system. The monitoring system is based on static non-linear models and statistical properties of measured signals. The system includes monitoring displays and parameter displays for operating and maintaining the system. Model parameters and alarm limits are determined automatically during the process operation. System development and testing work was carried out in a two-year project (93-94) and the participants in the project were process developer Enviropower Inc., automation system manufacturer Valment Automation Inc. and the control engineering laboratory of Tampere University of Technology. Key Words: Condition monitoring; model based diagnostics; pressurised gasification; automation system 1. INTRODUCTION Combined cycle energy production connected with the solid fuel pressurised air gasification process is a promising new technology to improve the efficiency and reduce harmful emissions of electricity genera- tion. Gasification of solid fuels makes it possible to use coal, biomass, etc. as a fuel for a gas turbine and by that means to build combined-cycle power plants even in places where natural gas is not available. However, the pressurised air gasification process is still under development, and there is not much empi- rical knowledge about the operation of the process on a large scale. For this type of new processl real-time condition monitoring and diagnostics helps plant staff to detect and analyse anomalies in the operation of the process. Some of the main problems in model-based condition monitoring and diagnosis in the process industry deal with system implementation and maintenance in a real process environment. Processes usually have several operating modes and points, and for this reason large and complex models are required. Most of the existing automation systems do not support the tasks needed for modelling and computing dynamic multivariate systems. More problems will rise if and when non-linear systems are involved. New methods like expert systems, neural networks and fuzzy logic have been studied in order to find solutions for these problems, but general shortcuts to overcome these problems have not yet been found. (Gilmore and Gingher, 1987; Ulieru, 1993; Venkatasubramianian and Chan, 1989) When dealing with industrial applications, the prob- lem is not only to find the theoretical solution for the diagnostic problem. Several constraints coming from the application environment must also be taken into account. Usually there are not enough measurements from the process, the existing ones are not accurate enough, the automation system has no reserve capacity for new applications, etc. So the testing and development of new systems will also require investments in process instrumentation and automation. This is problematic, because it is important that new systems like condition monitor- ing and diagnostics should be introduced to the plant staff in a real process environment so that they can see the benefits of these systems in their own processes. 1017