Folie 1

1
Increasing Power Output and Performance of
Nuclear Power Plants (NPPs) by Improved
Instrumentation and Control (IC) Systems

• 29 31 May 2007
• Prague

2

• A Program System
• for Measurement Validation
• in Safety-Relevant Components of NPP

J.Haenel, U.Gocht, M. Wagenknecht, A. Traichel,
R. Hampel (IPM), M. Wieland (Areva NP)
3
Contents

• Motivation
• Signal Validation
• Mathematical Methods
• Workflow
• Summary

4
1 Motivation

• Condition-based surveillance of NPPs with
  safety-relevant requirements
• Crucial importance of error control of
  components
• (in-service inspection)
• Minimization of shutdown period
• Decreasing of personal effort
• Minimization of malfunctions

5
Motivation

• Early error detection
• Redundant measurements without complication of
  models
• Observation of history (trends)
• Consideration of non-stochastic effects
• Analysis of time series as one basic instrument
• ? Validation

6
Motivation

• Program system for validation (with prior
  attributes)
• Consists of independent validation components
• Evaluation of redundant data before
  reconciliation
• Analysis, evaluation and validation of
  measurement data
• Displaying conditions of components and system
• Automatization of signal validation
• ? Software (operator-supporting tool)

7
2 Signal Validation
8
Signal Validation

• Intelligent process and system control
• Test of consistency of measurement information
  and real process values (sensor validation)
• Check of operability of components and system
  (process validation)
• ? detection and identification of malfunctions

9
Signal Validation
Deviation from the normal regime

• Measurement errors
• Malfunction of components
• Change of process state

10
Advanced Data Validation

• Redundant data measurement errors could possibly
  be recognized by simple value comparison
• Data reconciliation
• adjust process measurements with (random,)
  errors
• satisfy material and energy balance constraints
• Restricted or non-restricted Validation
  (additional bounds)

11
Advanced Data Validation

• Example Measurement errors in masses may be
  compensated by small changes in temperatures in
  plant components
• false conclusions by insufficient information on
  the values
• ? Minimization of insufficient information

12
Advanced Data Validation

• Applied mathematical methods
• Plausibility check
• Comparison of measured values with time series
  prediction
• Simple value comparison by redundant measuring
• Smoothing
• Classical data validation (reconciliation)
• Tests on stationarity and trends

13
3 Mathematical methods
Plausibility Check

• Test criteria predefined intervals (physical
  limits)
• Value out of range ignored in further
  calculation
• ? Detecting measurement errors or malfunction of
  
• components at an early stage

14
Comparison Of Measured Values With Time Series

• Simplifying classification
• restricted or non-restricted validation
  (additional bounds)
• Identifying
• measurement errors or change of process state

15
Simple Value Comparison By Redundant Measuring

• precondition normal distribution of values with
  mean µ and variance s
• test criteria distance of values with t?-
  quantile of normal distribution
• Results
• redundance calculation of mean
• no redundance single values

16
Simple Value Comparison By Redundant Measuring
17

• Tests On Trends

a) - linear or square regression equation -
coefficient of determination R2
0,5 with SSR - regression sum of
squares SSE - sum of squared errors. SST -
total sum of squares
18

• Tests On Trends (a)

19
Tests On Trends
b) Autocorrelation Checking of periodicity in
time series
20
Tests On Trends (b)
21
Tests On Trends (b)
22
4 Workflow

• Interconnection of modules
• Accounting the information flow
• Dependent on number of redundant sensors
• Classical reconciliation with partially validated
  data (if possible)
• ? Basis of program system

23
Workflow Of One Sensor Per Measuring Point
24

• 5 Summary

• Test of mathematical methods with real data
• ? Usefulness of our approach
• Implementation in software
• ? Automatization of signal validation