INTEGRATION OF FAULT DIAGNOSTIC TECHNOLOGIES INTO A COMPLEX CONDITION MONITORING SYSTEM AND ITS PRACTICAL RESULTS

We review the results of a large project for development a Maintenance Advisory System of a refinery with more than 2000 rotating machines. The system contains 18 on-line monitoring vibration diagnostic systems with a unique central database for 100 strategically most important machines and one offline diagnostic system with an independent knowledgeand database. The machines malfunction and fault detection is based on different diagnostic technologies like 3D Vibration Analysis, Thermography, Leakage Detection, Used Oil Analysis and Ferrography. The vibration analysis system is automatic expert software. Different interactive expert systems were developed for the rest analysis techniques. The results of fault detection and malfunction analysis integrated and displayed in one software system called BDES (Board of Diagnostic Expert System), which is a complex condition monitoring system for rotating machinery. We describe shortly the ExpertALERT artificial intelligence system, as well as ThermoALERT, LeakageALERT, OilALERT, and FerroALERT interactive expert analysis software, their applications and connection to SAP PM Maintenance module of the Refinery. Operative maintenance decisions and maintenance planning is grounded on integrated fault diagnostics and severity estimation. The IFSS is a webbased Information and Fault Statistic System, which helps strategic decision making of the maintenance management using statistical analysis of identified faults in territory allocation and trending in time. The Condition Based Maintenance strategy carried out. We describe the hardware and software solutions and show practical examples. Introduction This paper shortly describes the results of software development, what was realized in the framework of MOL’s (Hungarian Oil and Gas Company) On-line Diagnostic Project. The main goal of the project was the installation of on-line vibration monitoring systems for surveillance 100 strategically most important rotating machinery. The new on-line systems were integrated into one surveillance system, which means that, all of them work into one central database, placed on a central server computer in the server room of the Refinery Szazhalombatta. Parallel to the on-line systems have been installed an off-line vibration monitoring system with two ExpertALERT automated asset management diagnostic software with two independent database. These databases are synchronized by replication. The off-line and on-line databases of ExpertALERT contain the measured data, the results of data evaluation, the analysis results and the reports on the machines condition with machine faults, an estimation of fault severity, and recommendation for maintenance action. Goal of software development As the part of the “On-line Asset Management Project” the next software development goals were formulate: • Develop specific expert systems for analysis the next diagnostic data: thermo images of rotating machines, oil analysis, Ferrography and sealing leak detection, • Integrate the results of ExpertALERT and the newly developed expert software into a unique asset management system, • Develop a software, handling the risk matrix for rotating machinery risk classification, • Develop a web-based information system for spreading the diagnostic information in the company’s LAN, • Develop an interface between the diagnostic systems and the SAP PM module, www.dellta3n.hu 1 • Develop a software module for scheduling of measurements, • Develop a machine registry database developing software, Figure 1. shows the schema of the developed software system. Figure 1. Schema of the software system developed for MOL Refinery This schema does not contain the Machine Registry software, which is standalone software. The software module for scheduling measurement work is integrated into the SAP Interface. Characteristics of the software modules ExpertALERTTM software developed by AzimaDLI includes an imbedded rule based diagnostic system to help you screen through large amounts of data efficiently and focus on machines with problems. Although the diagnostic system is well proven and extremely accurate, a report editor is included should you wish to alter the reports or add comments. The ability to quickly analyze large amounts of data and accurately identify machines with problems is a key element for efficient vibration services. In a typical plant, about 10 20 % of all machines tested will have inherent mechanical faults. Within this group, far fewer will require immediate service. The distinguishing feature of AzimaDLI's rule-based, automated diagnostic system is that it identifies problem machines and focuses on manually reviewing the data from these machines. This approach is far more efficient than analyzing data from every single installed machine. Customers who use this automated approach typically receive a 20:1 benefit-to-cost ratio. The diagnostic system contains over 4,500 individual fault templates. These templates are based on empirical data acquired from hundreds of thousands of machine tests conducted over more than twenty years. They can be applied to more than forty general machine component types, including motors, pumps, fans, blowers, gearboxes, compressors, generators, turbines and machine tools. The system analyzes machine test data in a matter of seconds and produces a concise report that lists specific mechanical faults, the severity of each fault and an overall recommendation Compared with systems that simply indicate that a machine is in 'alarm' mode, this