1. Vibration Monitoring and Diagnostics of the Reactor Coolant Pump
22 October 2005
Center for Advanced Reactor Research
Jun-Seok Lee

2. Contents Introduction Current Monitoring and Diagnostic Systems
Problems and Root Causes
Diagnostic System of KEPRI
Conclusions & Recommendations
References
Introduction
Current Monitoring and Diagnostic Systems
Problems and Root Causes
Diagnostic System of KEPRI
Conclusions & Recommendations
References

3. I. Introduction Background
KEPRI-EPRI signed a joint research in late 1995.
KEPRI joined NMAC project on reactor coolant pump (RCP) condition monitoring managed by EPRI.
They found that condition monitoring and diagnostics (M&D) for RCP is one of the biggest issues in nuclear power plants (NPPs).
Problems in RCP
Seals
Motor stators
Vibration
Fixing those problems often results in significant outages of NPPs.
Monitoring systems have become more widely used, understood, and refined.
Most of them detect RCP problems by monitoring RCP vibration amplitude.
Few of them can diagnose the root causes of the problems.
Background
KEPRI-EPRI signed a joint research in late 1995.
KEPRI joined NMAC project on reactor coolant pump (RCP) condition monitoring managed by EPRI.
They found that condition monitoring and diagnostics (M&D) for RCP is one of the biggest issues in nuclear power plants (NPPs).
Problems in RCP
Seals
Motor stators
Vibration
Fixing those problems often results in significant outages of NPPs.
Monitoring systems have become more widely used, understood, and refined.
Most of them detect RCP problems by monitoring RCP vibration amplitude.
Few of them can diagnose the root causes of the problems.

4. I. Introduction M&D usage for RCP Oil analysis
It uses magnetic classification and spectrometric analysis of oil to detect problems with rotating.
Analyzing the data requires a lot of time when compared to the time between the detection of the problem and the failure of the machinery.
Bearing-oil temperature analysis
It uses a probe to measure bearing-oil temperature.
The temperature increases significantly, the severity of the problem has also increased significantly.
It is generally used with vibration analysis.
Acoustic data analysis
It uses the trend of the overall mean value and the spectral signatures.
The results can be affected by noise from other sources.
Vibration analysis
It is currently the most effective method for RCP M&D because it takes most of the physical factors that are affected by vibration into account.
It would be very helpful to relate the vibration characteristics to the corresponding factors.
M&D usage for RCP
Oil analysis
It uses magnetic classification and spectrometric analysis of oil to detect problems with rotating .
Analyzing the data requires a lot of time when compared to the time between the detection of the problem and the failure of the machinery.
Bearing-oil temperature analysis
It uses a probe to measure bearing-oil temperature.
The temperature increases significantly, the severity of the problem has also increased significantly.
It is generally used with vibration analysis.
Acoustic data analysis
It uses the trend of the overall mean value and the spectral signatures.
The results can be affected by noise from other sources.
Vibration analysis
It is currently the most effective method for RCP M&D because it takes most of the physical factors that are affected by vibration into account.
Relating the vibration characteristics to the corresponding factors would be very helpful.

5. I. Introduction Objective
To review the RCP diagnostic system currently being used in NPPs
To identify the major RCP vibration problems and their causes
To use the information to create knowledge base for diagnosing these problems
Objective
To review the RCP diagnostic procedures currently being used in NPPs
To identify the major RCP vibration problems and their causes
To use this information to create knowledge base for diagnosing these problems

6. II. Current Monitoring and Diagnostic Systems
RCP description
Contents
Specification
Kori 1
Kori 2
Kori 3&4
Young gwang 1&2
Uljin 1&2
General
Model
93-A
93-AS
93-A1
Model-100
No. of pumps per unit 2 3
Pump
Type
Vertical, 1 stage, centrifugal
Design pressure & temperature
(kg/㎠ & ℃)
174.7/343
No. of blades 7
Head (m)
79.9
86.3
84.7
85.3
80.0
Flow(㎥/min)
336.9
385.7
390.2
389.1
378.3
Motor
Squirrel-cage induction motor
Speed(rpm)
1190
1188
<Reactor Coolant Pumps and Motors Used in KEPCO(KHNP)>
<Reactor Coolant Pump (Westinghouse 93A Model)>

7. II. Current Monitoring and Diagnostic Systems
Current RCP M&D systems
No.
System
Company
Country
Type
Tool
Application
Remarks 1
DIAPO (DIAgnostic des POmpes)
EDF
France
Model-based
Vibration
Process signals
Part of PSAD 2
COMOS (Condition Monitoring System)
Germany
Vibration analysis
Noise
Grafenrheinfeld KWU PWR 3
LPMS (Loose Part Monitoring System)
Acoustic signal analysis 4
MAINS (MAINtenance Support expert system)
Toshiba
Japan
Offline prototype
Cause-consequence matrix
Interface with the software developed for Toshiba pump monitoring system that TEPCO is using 5
NSSS-DS (Nuclear Steam Supply System-Diagnostic System)
KEPCO(KEPRI)
Korea
Rule-based deduction with certainty factor operation
Seek the primary causal alarm among existing multiple alarms
Automatic actions
Probabilistic failure modes
Kori-2 PWR
IBM/PC (Prolog)
Diagnosis of 3 main systems: rod control system, RCP & pressurizer 6
Mechanical Diagnostic Expert System
G.E. Corporation R&D Center
USA
Rule-based symptom-fault matrix
Vibration analysis (FFT & time-based signal)
BWR
MS-DOS/UNIX (C) 7
Diagnostic System
Gensym Corporation
USA & Japan
Rule-based
Vibrations
Tokai-2 BWR
Expert shell GEIEMS/G2 8
Decision tree
Acoustic emission
Process data (pump head, motor temp. & current)
15 possible anomalies 9
Automated surveillance system
Argonne National Lab.
Pattern recognition
Sequential probability ratio test (SPRT)
EBR-2
MS-DOS (C)

8. II. Current Monitoring and Diagnostic Systems
DIAPO (DIAgnostic des POmpes) - Electrcité de France (EDF) and Jeumont Industrie
Information
Identified initial causes
Recognized problems
Recognized abnormal conditions
Locations of faulty components
Fault model
Associative model
Causal model
Prototypical model
Failure localization model
Part of PSAD (Poste de Surveilance et d’Aide an Diagnostic) system
An overall diagnostic system
Automatically monitors and examines the vibrations and operational data from turbine generators, primary circuits, and RCPs
DIAPO (DIAgnostic des POmpes) - Electrcité de France (EDF) and Jeumont Industrie
Information
Identified initial causes
Recognized problems
Recognized abnormal conditions
Locations of faulty components
Fault model
Associative model
Causal model
Prototypical model
Failure localization model
Part of PSAD (Poste de Surveilance et d’Aide an Diagnostic) system
An overall diagnostic system
Automatically monitors and examines the vibrations and operational data from turbine generators, primary circuits, and RCPs

9. II. Current Monitoring and Diagnostic Systems
Diagnostic System - Toshiba Corporation
Consists of 2 basic parts
Data Acquisition System (DAS)
Diagnostic Computer (DIC)
Diagnostic algorithm
Decision tree
Data evaluation function
To determine the cause of a problem, it uses
Data evaluation
Cause inference
Quantitative and trend estimation
Application : TEPCO
Diagnostic System - Toshiba Corporation
Consists of 2 basic parts
Data Acquisition System (DAS)
Diagnostic Computer (DIC)
Diagnostic algorithm
Decision tree
Data evaluation function
To determine the cause of a problem, it uses
Data evaluation
Cause inference
Quantitative and trend estimation
Application : TEPCO

10. II. Current Monitoring and Diagnostic Systems
Mechanical Diagnostic and Expert System - GE Corporation and Gensym Corporation
Diagnostic and expert system for BWR
Consists of 2 systems
General Electric Integrated Equipment Monitoring System (GEIEMS) : Used for collecting and analyzing problem-related data
G2 System : Used for diagnosing the causes of problems
Application : Tokai-2
Mechanical Diagnostic and Expert System - GE Corporation and Gensym Corporation
Diagnostic and expert system for BWR
Consists of 2 systems
General Electric Integrated Equipment Monitoring System (GEIEMS) : Used for collecting and analyzing problem-related data
G2 System : Used for diagnosing the causes of problems
Application : Tokai-2

11. II. Current Monitoring and Diagnostic Systems
COMOS (COndition MOnitoring System) and LPMS (Loose Part Monitoring System) – Germany’s nuclear power industry
COMOS
Specifically to monitor shaft and bearing vibrations
Using the power densities and the correlated functions of the pump signals
LPMS
Using knowledge base that relates acoustical signature data to pump components
Automated surveillance system – Argonne National Laboratory
An expert system
to provide early problem detection
to diagnose RCP problems
Using Sequential Probability Ratio Test (SPRT) and If-Then rules
SPRT : A statistical-based, pattern recognition technique
Continually analyzing digitized signals from sensors in RCP
Monitoring parameters : rotor speed, vibration level, power, discharge pressure
Comparison of two rotor-speed signals to determine the discrepancy of them.
COMOS (COndition MOnitoring System) and LPMS (Loose Part Monitoring System) – Germany’s nuclear power industry
COMOS
Specifically to monitor shaft and bearing vibrations
Using the power densities and the correlated functions of the pump signals
LPMS
Using knowledge base that relates acoustical signature data to pump components
Automated surveillance system – Argonne National Laboratory
An expert system to provide early problem detection and to diagnose RCP problems
Using Sequential Probability Ratio Test (SPRT) + If-Then rules
SPRT : A statistical-based, pattern recognition technique
Continually analyzing digitized signals from sensors in RCP
Monitoring parameters : rotor speed, vibration level, power, discharge pressure
Comparison of two rotor-speed signals to determine the discrepancy of them.

12. III. Problems and Root Causes
Key Problems and Root Causes
Searching the INPO Nuclear Plant Reliability Data System (NPRDS) database for problems from 1993 to 1996
Yielding 57 problems that resulted in plant trip, shutdown, and power reduction
Major causes of problems : vibration, leakage
Major causes of vibration problem: shaft, bearing, misalignment
Key Problems and Root Causes
Searching the INPO Nuclear Plant Reliability Data System (NPRDS) database for problems from 1993 to 1996
Yielding 57 problems that resulted in plant trip, shutdown, and power reduction
Major causes of problems : vibration, leakage
Major causes of vibration : shaft, bearing, misalignment
< Cause of RCP Failures >
< Root causes of RCP vibrations >

13. IV. Diagnostic System of KEPRI
Structure of NSSS-DS (Nuclear Steam Supply System-Diagnostic System)
< Probe System for Monitoring RCP Vibration >
< Structure of Monitoring System >

14. IV. Diagnostic System of KEPRI
General Approach
Classify the types of abnormal vibration by cause and problem, using information from NPRDS search and NMAC survey.
Determine the diagnostic methods to analyze the data.
Evaluate the abnormal vibration characteristics, using information from the following to help with evaluations:
Experts, field engineers, and analysts
Papers that have been published
Case histories on RCP failures
Experimental verification
Construct the knowledge base.
Develop a diagnostic algorithm that does the following:
Uses evaluation functions to quantify knowledge expressions.
Uses certainty factors to improve the reliability of diagnoses.
Shows the probabilistic diagnostic results of each abnormal vibration condition.
Estimates the vibration trend and the amount of time needed to shut down the RCP.
Identifies the component that may be defective and provides a repair procedure.
General Approach
Classify the types of abnormal vibration by cause and problem, using information from NPRDS search and from NMAC survey.
Determine which diagnostic methods to use to analyze the data.
Evaluate the abnormal vibration characteristics, using information from the following to help with evaluations:
Experts, field engineers, and analysts
Papers that have been published
Case histories on RCP failures
Experimental verification
Construct the knowledge base.
Develop a diagnostic algorithm that does the following:
Uses evaluation functions to quantify knowledge expressions.
Uses certainty factors to improve the reliability of diagnoses.
Shows the probabilistic diagnostic results of each abnormal vibration condition.
Estimates the vibration trend and the amount of time needed to shut down the RCP.
Identifies the component that may be defective and provides a repair procedure.

15. IV. Diagnostic System of KEPRI
Construction of Knowledge Base
Survey
A survey on RCP condition monitoring and diagnostic techniques which was sent to all EPRI NMAC members of US and international plants
Database
The NPRDS database was searched for information covered from 1990 to 1996
Experts and case histories
Lots of RCP experts were consulted and several RCP case histories were reviewed to obtain information based on RCP experience
Literature survey
RCP related papers were reviewed to get the information of abnormal vibration characteristics of RCP.
Construction of Knowledge Base
Survey
A survey on RCP condition monitoring and diagnostic techniques which was sent to all EPRI NMAC members of US and international plants
Database
The NPRDS database was searched for information covered from 1990 to 1996
Experts and case histories
Lots of RCP experts were consulted and several RCP case histories were reviewed to obtain information based on RCP experience
Literature survey
RCP related papers were reviewed to get the information of abnormal vibration characteristics of RCP.

16. IV. Diagnostic System of KEPRI
Cause-Symptoms Matrix for Knowledge Base

17. IV. Diagnostic System of KEPRI
Diagnostic Algorithm
Possible method for the diagnostic system : Fuzzy set method, Bayesian method, Dampster shaper method, Certainty factor method
Using Fuzzy set method for the system
Defining the membership functions to be used to evaluate each fact
Calculating the weighting factor of each type of vibration data, using several evaluation function
Diagnostic Algorithm
Possible method for the diagnostic system : Fuzzy set method, Bayesian method, Dampster shaper method, Certainty factor method
Using Fuzzy set method for the system
Defining the membership functions to be used to evaluate each fact
Calculating the weight factor of each type of vibration data, using several evaluation function
< Procedure for Weighting Factor Calculations >
< Evaluations Functions >

18. IV. Diagnostic System of KEPRI
Flow chart
*Did not conduct in this research.

19. IV. Diagnostic System of KEPRI

20. IV. Diagnostic System of KEPRI
Output of the Diagnostic System

21. V. Conclusions & Recommendations
Most of RCP monitoring systems can detect RCP problems with monitoring abnormal vibrations, but few of these system can diagnose the root causes of problems.
From the survey on NPPs, the identified mechanical faults of the RCP are still on-going issues.
From the NPRDS search
Vibration is one of the most important symptoms of key failures on the RCP.
Major cause of vibration : cracked shaft, misalignment, rubbing, unbalance
Recommendations
Conduct an experimental verification for improving the reliability of knowledge.
Identify the vibration characteristics due to changes in the operation parameters.
Determine the relationship between the root causes and the RCP components.
Define the various evaluation functions and determine their relationship with the knowledge base.
Use probabilistic theory to estimate the optimum amount of time for repairing RCP problems.
Share the documented knowledge for each type of RCP with all nuclear power plants.
Conclusions
Most of RCP monitoring systems can detect RCP problems with monitoring abnormal vibrations, but few of these system can diagnose the root causes of problems.
From the survey on NPPs, the identified mechanical faults of the RCP are still on-going issues.
From the NPRDS search
Vibration is one of the most important symptoms of key failures on the RCP.
Major cause of vibration : cracked shaft, misalignment, rubbing, unbalance
Recommendations
Conduct an experimental verification for improving the reliability of knowledge.
Identify the vibration characteristics due to changes in the operation parameters.
Determine the relationship between the root causes and the RCP components.
Define the various evaluation functions and determine their relationship with the knowledge base.
Develop a means of using pattern recognition to automatically enter other parameters.
Use probabilistic theory to estimate the optimum amount of time for repairing RCP problems.
Share the documented knowledge for each type of RCP with all nuclear power plants.

22. VI. References
Yong Chae Bae, “Consideration for Vibration Monitoring and Diagnostics of the Reactor Coolant Pump”, EPRI NMAC Tech Note, TR , July, 1997.
Yong Chae Bae, “A Study on the Diagnostic System for Reactor Coolant Pump”, Journal of Korea Society for Noise and Vibration Engineering, Vol.8, No. 4, pp , 1998.
Se Woo Cheon, Soon Heung Chang, Hak Yeong Chung, “Development strategies of an expert system for multiple alarm processing and diagnosis in nuclear power plants”, IEEE Transactions on Nuclear Science, Volume 40, Issue 1, pp , February, 1993.