IFEs Activities on Instrument Accuracy Monitoring at OKG

1
IFEs Activities on Instrument Accuracy
Monitoring at OKG

• Mario Hoffmann
• William Beere
• OECD Halden Reactor Project
• Institutt for energiteknikk

2
Introduction

• Signal Validation andCalibration Monitoring
• Thermal PowerUncertainty Determination

3
Feasibility Study on Calibration Monitoring

• Oskarshamn Unit 3
• Boiling Water Reactor
• 1198 MWe
• OKG Interests
• Calibration reduction
• Better knowledge of the instrument status
• Shortened outage periods
• PEANO Feasibility study
• See how On-line Calibration Monitoring can help
• Gain experience with the use of HRP systems

4
Motivation for On-line Monitoring

• Time-based maintenance
• Current practice to comply with regulatory
  requirements
• Limited insight or knowledge of sensor status is
  available as a basis for process operation
  decisions
• Calibration is performed regardless of the sensor
  status
• Condition-based Maintenance
• Calibrate only when needed, based on the
  instrument condition
• Perform On-line Calibration Monitoring

5
Motivation for On-line Monitoring

• On-line Calibration Monitoring
• More efficient maintenance strategy
• Increase calibration intervals (calibration
  reduction)
• Up-front condition-based actions

6
PEANO System

• On-line Calibration Monitoring and Sensor
  Validation system, based on Fuzzy-Neural Network
  models

7
PEANO Purposes

• Calibration MonitoringBetter knowledge of the
  instrument status will help to estimate and
  target your re-calibration efforts during
  maintenance planning and lead to reduced outage
  time
• Signal ValidationValidated and reconstructed
  measurements can help you to make better
  operational decisions and provide better input
  data for other Operator Support Systems

8
Fault Detection Capabilities
9
OKG Unit 3 Project Details

• Sensors from the heat balance system
• Well instrumented and calibrated part of the
  process
• Limited scale
• Modelling Data
• Covers 7.5 months31st May 2005 16th January
  2006
• 113 Signals
• 10 min sampling rate
• Start-up, shut-down and normal operation

10
Process Diagram
11
Process Data for Modelling

• Data files containted 113 signals
• Temperature, pressure, flow
• Electric and thermal power
• Redundant measurements
• 6 groups of redundant measurements
• 29 sensors
• Cross-correlation Analysis
• 7 signals showed low cross-correlations
• E.g. from the Heat Removal System
• Redundant and low correlated sensors were removed
• 77 signals selected for modelling

12
Models for the Project

• Heat Balance System
• Single model77 signals
• Models of Sub-Systems
• Condenser System10 signals
• Heat Removal System9 signals

13
Data for Model Testing

• Historic plant data with know problem
• 13. 20. March 2002
• 23. 30. September 2002
• 17. 24. April 2003
• 18. 25. March 2005
• Process data from 2006
• Up until the last maintenance period (June 2006)

14
Temperature Sensor Drift - 2005

• Sensor tag 312 KC502
• Feed Water Lines Temperature (C)Range 0 250
• Adjustment at -0.4C drift(i.e. 0.16 of the
  range)
• A new drift occursexceeding the errorbands
• Sensor is rewired to312 KA502 after a short stop

15
Flow Sensor Problem - 2002

• Sensor tag 312 KC301Feed Water Flow
  (kg/s)Range 0 1100
• Flow sensor drifted by 3 kg/s
• Resulting in a 3 MWh loss of produced power,
  which is also detected

16
Coast Down - 2006

• Coast down data was not available in the
  modelling data
• Coast down is initiated at2130 on 22nd May 2006
• The PEANO estimate for the power starts to
  deviate
• Confidence value for the model drops to low from
  2140 on 22nd May 2006

17
Future Developments

• Large scale application handling
• Preferred PEANO model size is 30-50 sensors
• Realistic application at OKG Unit 3 of 1500
  3000 sensors
• Automatic sensor grouping is needed

• Regulatory acceptance for calibration reduction

18
OKGs method for thermal power uncertainty
determination using PROBERA
19
Current Power Control

• Peaks not to exceed safety limit
• Average not to exceed full power limit

20
Proposed Power Control

• Peaks not to exceed safety limit uncertainty
• (Power unceratinty not to exceed safety limit)
• Average not to exceed full power limit

21
How to determine Uncertainty

• Analytical function
• Linearise
• Need measurment uncertainty matrix

22
How to determine dependencesAnalytical

• Determine contributions
• Turbine power
• Power lost to cleaning circuits
• Internal circulation pump
• Heat losses
• Determine formulations to contributions

• Can easily miss covariances
• Difficult to verify entire process

23
How to determine dependencesFlow sheet

• Small verifiable units
• Easily compared to actual process
• Function for Thermal Power is automatically
  generated

24
Example of importance of flow sheet
T3
T2
T1
Q1
Q2
W

• Q1 W ( T2 T1 )
• Q2 W ( T3 T2 )
• Qtot W ( T3 T1 )
• Qtot Q1 Q2

25
Probera Construction

• Pre-processing of measurements with physical
  redundancy
• Process flow sheet builder
• Non-linear flow sheet solver and optimizer
• Reconciled measurement values
• Parameter determination
• Linearization
• Linear data-reconciliation for
• Reconciled measurement uncertainties
• Parameter uncertainty determination

VDI-2048
26
What is VDI-2048

• Data Reconciliation Standard VDI-2048Uncertain
  ties of measurement during acceptance tests on
  energy conversion and power plants, October 2000
• Independent assessment of measurement
  uncertainties and correlations
• Generation of data-reconciled measurement values
  using constraint equations
• Convergence Criteria
• Acceptance Criteria (gross errors)
• Parameter confidence limits calculation

27
Review of Probera

• Review of methodology
• Suggested improvements to better conform to
  VDI-2048
• Convergence criteria
• Acceptance criteria
• Uncertainty (confidence) limtis
• Basic testing
• Testing out statistical calcultions for simple
  systems
• Analysis of thermal power results
• Direct calculation of main contributing
  measurement

28
Results

• O1 1.63
• Main contribution from feed water flow
• Contribution factor 0.98
• O2 1.25
• O3 0.4
• Uncertainties for O2 and O3 are not limiting to
  operations
• Unceratintiy for O1 is limiting operations
• Feed water flow identified as measurement to
  reduce uncertainty
• Power increase can be determined and compared to
  re-instrumentation cost

29
Summary

• Calibration Monitoring
• PEANO has been successfully applied to the OKG
  data
• Known historic sensor drifts have been detected
• Further development is ongoing to expand PEANO
  with support for large-scale applications
• Thermal Power Uncertainty Determination
• Use data-reconcilliation methodology
• Measurement uncertainty should be monitored
  through signal validation

30
HOLMUG

• A forum to advance implementation of on-line
  monitoring methods
• Co-ordinated by the Halden Reactor Project
• Participation from utilities, research
  institutes, universities, regulatories and
  vendors
• Next meeting October 3th-4th, 2007 in Olkiluoto,
  Finland