Brought to you by EarthWeb
IT Library Logo

Click Here!
Click Here!

Search the site:
 
EXPERT SEARCH -----
Programming Languages
Databases
Security
Web Services
Network Services
Middleware
Components
Operating Systems
User Interfaces
Groupware & Collaboration
Content Management
Productivity Applications
Hardware
Fun & Games

EarthWeb Direct EarthWeb Direct Fatbrain Auctions Support Source Answers

EarthWeb sites
Crossnodes
Datamation
Developer.com
DICE
EarthWeb.com
EarthWeb Direct
ERP Hub
Gamelan
GoCertify.com
HTMLGoodies
Intranet Journal
IT Knowledge
IT Library
JavaGoodies
JARS
JavaScripts.com
open source IT
RoadCoders
Y2K Info

Previous Table of Contents Next


3.2.1. TIGER

Milne (1992) developed an online monitoring expert system for use with gas turbines that sample data every second. The initial system has two modes of operation, the first being online and the second a post-crash analysis.

The system provides the user with various information displays showing the current state of the turbine, including a turbine overview, real-time exhaust spread, a summary of current faults, a summary of current alarms, a history of past faults, trend displays, and displays showing data input in real-time. The system also provides facilities for the historical logging of data to disk for trending purposes.

The system is comprised of various subsystems: an interface to the turbine's instrumentation hardware, signal decoders for analog and digital signals, a 4-minute buffer of recent data (in case a crash occurs), a trending mechanism, and the interface to the expert system itself. The interface between the instrumentation hardware and the expert system is achieved by Intelligent Application's ANNIE package.

Diagnosis is broken up into three subareas:

  • Alarm diagnostics
  • Analog-related alarms
  • Trip log and analysis

Alarm diagnostics consists of two major groups of alarms that can be diagnosed: normal operation and start-up alarms. The normal operation alarms are of primary importance as the system will be able to give a more precise diagnosis of why an alarm has triggered as (Milne, 1992):

"One of the common problems of alarm messages on gas turbines is that many possible things can go wrong all leading to the same alarm."

Analog-related diagnostics are used as a predictive maintenance strategy to detect a developing fault before it triggers an alarm or trip situation. This diagnostic method uses relatively simple comparisons of analog values to detect significant changes in the data.

Trip log and analysis are used when an alarm or trip situation occurs. Data in the system's 4-minute buffer are stored onto disk for later examination. These data samples can then be used to identify faults. Fault logs are also stored to disk in order to provide a history of problems that have occurred with the turbine.

At the time of its implementation in 1992, TIGER was the largest online expert system in regular use with a gas turbine. The system was developed for and used by Exxon Chemicals and was found to be very effective. By 1995, the system was being expanded as part of a European project: TIGER (ESPRIT 6862) (Milne, 1995). The developed application is still considered to be the most advanced software tool available for condition monitoring of gas turbines. The system runs constantly, sampling key operational parameters every second, and uses these data samples to perform a wide variety of examinations, including limit checking, dynamic response, and consistency checks using model-based predictions. The system also provides an extensive graphical user interface to relay information to the system operator, including extensive data trending facilities, which are user configurable, to allow the user to specify what data to sample and when.

The application has been in constant use by Exxon chemicals for over 2 years and has been instrumental in identifying many problems with the gas turbine that it was monitoring. Turbine sensor data are constantly sampled, via a link to instrumentation systems, and evaluated every second. The package can also be configured to connect to any serial link equipment or any real-time database over a network.

The latest version of the TIGER system consists of three individual diagnostic tools:

  • KHEOPS
  • IxTeT
  • CA-EN

KHEOPS is a full-power, high-speed, rule-based system whose primary purpose is the limit checking of data. IxTeT is a powerful tool that has the capability to monitor the dynamic reaction of a gas turbine. This feature is unique to TIGER, in that no other condition monitoring system has this facility. The CA-EN tool is a model-based prediction and diagnosis system used for fault detection and the isolation of faulty components (Milne, 1995).

3.2.2. SA-VANT

In 1984, EPRI (the U.S. Electric Power Research Institute) started the SA-VANT project with the aim of initially studying the viability of expert system job aids for field use by maintenance engineers (Armor et al., 1993). First development models using expert system technology showed good results, with the system diagnosing problems in 25 minutes that normally took an hour for an expert technician. There were some problems that novice technicians could not solve without help, which only took them 26 minutes when they were aided by the system.

Due to the promising experiences with the early system, EPRI chose to use SA-VANT to aid technicians in diagnosing failure to start-up problems on large gas turbines. The initial reliability of the turbines was at a level of 85% or lower. Industry, however, set the required reliability to at least 95%. EPRI believed that, using its expert system, the 95% reliability level could be reached by avoiding aborted start-ups, which could possibly damage the engine and inflict costly delays at a critical time.

The knowledge base for SA-VANT was developed by gathering experts with knowledge on various different models of gas turbines, including utility "trouble shooters." The knowledge acquisition was done using a round-table format with leading technicians carefully going step-by-step through symptom-to-cause scenarios. EPRI also decided to include several multimedia concepts in the system, such as diagrams, video clips, and audio.

After development, the system was tested in the field to determine its effectiveness by a broad range of technicians ranging in experience from 6 months to 16 years. The use of SA-VANT resulted in a 25% decrease in time to diagnose a problem, with an 81% decrease in calls for assistance, and an 89% decrease in the number of aborted re-start attempts.

In the period from 1992 to 1995, the SA-VANT system was in use by Jersey Central Power & Light and has produced a savings of approximately $50,000, with a predicted 15-year (from 1992) savings of $235,361 (EPRI, 1996).


Previous Table of Contents Next

footer nav
Use of this site is subject certain Terms & Conditions.
Copyright (c) 1996-1999 EarthWeb, Inc.. All rights reserved. Reproduction in whole or in part in any form or medium without express written permission of EarthWeb is prohibited. Please read our privacy policy for details.