Skip to main content

BASIC RESEARCH INTO IMPROVED MANAGERIAL OPERATIONAL CONTROL AND PERFORMANCE FEEDBACK SYSTEMS FOR PROCESS PLANTS

Objective


A machinery diagnostic system (MDS) has been produced for the monitoring of complex or critical machinery using pattern based vibration techniques. The system extracts component specific data from a raw vibration signal and reduces this to a vector of fault relatable indicators using intensive numerical analysis. The technology provides early and reliable fault detection and diagnosis for effective operational planning and targeted maintenance.
The MDs consists of a data acquisition and analysis system (DABS) linked to a 386 based personal computer (PC) running plant operators maintenance management system (POMMS) executive software. The DAAS is a ruggedised computer performing data acquisition and analysis local to the monitored machinery. POMMS provides system configuration, database, display, trending and data interpretation facilities.
The MDS offers a powerful capability for online diagnostic monitoring of important machinery. Versions of the MDS are being for the health monitoring of a gas turbine used for industrial power generation, for the monitoring of machine tools in a production plant and for the end of production testing of gearboxes for quality control purposes. The POMMS software has already been used in a number of applications in addition to the MDS. The software has been installed in a workshop test system for large gearboxes used in the mining industry. It has formed the basis of the groundstation software for a helicopter health and usage monitoring system.
The operating environment is as follows :
Operating system: Xenix multiuser multitasking;\Software language: C

Intelligent sensors have been developed for online monitoring of multiple machines in remote locations. These vibration based sensors incorporate signal conditioning, analogue to digital (AD) conversion and data processing capabilities, outputting high level diagnostic information to a central system. Multiple sensors can be connected in series on a serial line providing 2-way communications. Sensors can be individually addressed and configured, can compute spectra, can extract features from spectra and can automatically learn fault detection thresholds. The sensors normally only send messages when a fault has been detected but they can be interrogated from the central systems.
The main capabilities of the sensors are bidirectional data exchange capabilities and data and parameter memory. The sensor delivers messages containing high level information on machinery health to a plant management system with raw or processed data available if necessary for complete interpretation at the central site.
The sensor is an intelligent black box with channels dedicated to acoustic analysis and vibration analysis for machinery monitoring. In addition the sensor has memory, data exchange capabilities and software development facilities for signature analysis.

An intelligent operator support system (IOSS) has been developed to provide a process supervision and feedback capability. This is a real time knowledge based data interpretation system running on a micro VAX II GPX workstation. Logical models representing a mapping of the process were developed from a knowledge base generated by a changed mode and effect analysis (CMEA). This covers both normal and distributed stated of the process plant and allows the direct extraction of rules describing and behaviour of the process.
The IOSS acquires process sensor data and generates a time stamped data stream. The data is passed to the knowledge base containing the logical models for time dependent rule interpretation. Whenever a parameter exceeds a predetermined threshold rules are fired in a forward chaining process until a diagnosis is produced. The event driven system updates a pictorial display and generates a diagnostic message giving the operator real time information on the current state of the process.
TO PRODUCE A COST-EFFECTIVE ALARM PROCESSING AND PROCESS DIAGNOSTIC SYSTEM FOR SMALL TO MEDIUM SIZED ENTERPRISES, CAPABLE OF SOME DEGREE OF SELF-LEARNING AND FEEDBACK OF OPERATIONAL DATA TO DESIGN SO THAT PROCESS FAULTS MAY BE MORE EFFECTIVELY CORRECTED.

PROCESS PLANTS, ESPECIALLY ONES OF NOVEL DESIGN, CAN SUFFER FROM CONTROL AND MECHANICAL FAILURE PROBLEMS THAT SERIOUSLY COMPROMISE PLANT AVAILABLITY, AND OCCASIONALLY THESE ARE SO BAD THAT THE VIABILITY OF THE PLANT IS JEOPARDIZED. THE PARTICULAR OBJECTIVE OF THE CONSORTIUM IS THEREFORE TO DEVELOP A SET OF COMPUTER TOOLS THAT WILL ALLOW AN OPERATOR TO (A) MONITOR DISTURBANCES AND ALRMS WITHIN THE PLANT MORE EFFECTIVELY AND MEANINGFULLY (HENCE OFFERING THE CHANCE OF BETTER CONTROL), AND (B) FOLLOWING THE FAILURE OF CONTROL OR COMPONENTS WITHIN THE PLANT, INSPECT A DATABASE OF PLANT OPERATIONAL, MAINTENANCE DATA ETC. FOR EARY SIGNS OF THAT FAILURE AND THEREFORE A DIAGNOSTIC FOR FUTURE USE.

Funding Scheme

CSC - Cost-sharing contracts

Coordinator

Stewart Hughes Ltd
Address
Chilworth Manor
SO9 1XB Southampton
United Kingdom

Participants (2)

GESELLSCHAFT FUER ANLAGEN- UND REAKTORSICHERHEIT (GRS) MBH
Germany
Address
Schwertnergasse 1
50667 Koeln
METRAVIB RECHERCHE DÉVELOPPEMENT SERVICE SA
France
Address
Chemin Des Ormeaux 200
69760 Limonest