ADVANCED MANAGEMENT SYSTEM OF PIPE WALL THINNING BASED ON
PREDICTION-MONITOR FUSION
Fumio KOJIMA1 and Shunsuke UCHIDA2
1Kobe University, Kobe, 657-8501 JAPAN ([email protected]) 2Institute of Applied Energy, Tokyo, 105-0003 JAPAN ([email protected])
The 17th Symbio International Workshop 2012 on Advanced Condition Monitors for Nuclear Power and other Process Systems Kyoto University, Kyoto, Japan (September 3, 2012)
Talk of the Scenario (Part I)
Pipe wall thinning management Prediction by FAC analyses Reliability assessment of NDT New strategies of the management
2
Pipe Wall Thinning Management
Pipe wall management is to implement inspection programs in order to prevent catastrophic events due to leaks, raptures, and severe thinning, etc.
Flow-accelerated corrosion (FAC)
Liquid droplet impingement erosion (LDI)
3
Analyses for predicting wear rates and periodical inspections for the pipe wall thickness play essential roles in PWTM.
JSME Codes for Nuclear Power Generation Facilities
Rules on Pipe Wall Thinning Management for PWR plant
CA-2310 normal pitch
Evaluation by CA-3000
Yes
No CA-2320 Detailed survey
Measurement pitch with 20[mm]
( )srnsrm 32 tttt −+=
Standard thickness
tn :minimum thickness tsr:requirement
34
5
67
8
1
2
Thinning rate
mttmeasure >
Those have been performed by the conventional UT manually for the prescribed area of piping system. 4
Pipe Wall Thinning Management
Major tasks in PWTM include the selecting and scheduling components for inspection and the decision making for repair or replacement of the specific components of the piping system.
5
Talk of the Scenario (Part II)
Pipe wall thinning management Prediction by FAC analyses Reliability assessment of NDT New strategies of the management
6
FAC caused factors
FAC is determined by six parameters, 1.Flow parameter (mass transfer coefficient, MTC) 2.Material parameter (chromium content in materials) 3.Environmental parameters (temperature, pH, and O2 and Fe2+ concentrations in the water)
FACs occur on a piping internal wide range at an orifice, an elbow, and a reducer down stream.
7
Prediction code for FAC analyses
Step 1 • 1D CFD code
Step2 • O2-N2H4 reaction code
Step 3 • 1D FAC code
Step 4 • 3D CFD code
Step 5 • 3D FAC code
Step 6 • Total evaluation
Flow chart of FAC analyses:
8
Experimental results
9
Talk of the Scenario (Part III)
Pipe wall thinning management Prediction by FAC analyses Reliability assessment of NDT New strategies of the management
10
Reliability of Assessment
Q: Why is it necessary to evaluate inspection system for pipe wall management ? A: Performance of inspection should be evaluated. The recommended UT inspection process consists of marking a grid pattern on the components and of taking wall-thickness measurements at the grid points. Nevertheless, it is not easy to find where the maximum wall thinning occurs in the component
11
POD is a feasible measure
Probability of Detection A. P. Berenes, NDE Reliability Data Analysis, 1989
Probability that acceptable wall thickness from the population would be detected Given a defined inspection system Given a population of experienced wall thinning
Performance indices Hit/Miss analysis, - analysis Confidence bound for quantifying NDT.
12
Reliability Assessment using POD
Q: Why is it important to use POD ? A: POD provides the capability to develop a feasible inspection model used for quantifying inspection reliability for pipe wall management program.
13
POD Evaluation
Screening of pipe wall system (LDI、pipe under support plate )
Monitoring volumetric thinning (incident of FAC)
(Berens and Hovey, 1981)
(a: number of pits for LDI)
14
How to Calculate POD
Signal Response Analysis by
Collect signal response set corresponding to many different inspections for piping system with variety of wall thinning Taking the logarithm of the signal response relation, we obtain the formula: Then the regression parameters and can be estimated from those data set.
15
E-MAT Based NDE System
EMAT consists of a magnet and a coil of wire and relies on electro-magnetic acoustic interaction for elastic wave generation.
Using a Lorentz forces in conductor , (a direct interaction between the magnetization of the material and the applied field, and the magnetostrictive mechanism) , the EMAT and the metal test surface interact and generate an acoustic wave within the sample specimen.
16
Grand Test for Downstream of Orifice Compared with UT sizing results, maximum difference between UT and EMAT measurements became 0.28 [mm], average difference was 0.08 [mm], and median was 0.07 [mm].
We gratefully acknowledge the Institute of Nuclear Safety System Inc (INSS) and the Fossil Power Engineering Center, KEPCO for their assistance in this experimental works. 17
Signal Response Analysis
183.0ˆ010.1ˆ395.7ˆ
1
0
==
=
δσβ
β
18
How to Calculate POD
Compute the maximum likelihood estimate
The POD function can be then calculated by: where
540ˆ =deca
19
Contribution to the reliability
The “reliably” quantitative value for the applied inspection system can be detected by the inverse of POD(a) function:
Traditionally, those have been designated as
those can be derived from the “confidence bound” of POD;
20
POD (Straight pipe with pipe wall thinning) [ ]
−−
Φ=1
10
ˆ/ˆ
ˆ/ˆ)ˆln()ln()(βσ
ββ
δ
decaaaPOD
A1
B1
21
POD (Elbow)
A1
22
POD (Reducer)
A1
23
Talk of the Scenario (Part IV)
Pipe wall thinning management Prediction by FAC analyses Reliability assessment of NDT New strategies of the management
24
Link to Safety Measure
25
Wall thinning analyses could provide the user with the wear rates of components.
Inspections could provide the time remaining before a specified minimum wall thickness is reached
Prediction-monitor fusion:
Contribution to predictive plant model
Continuous surveillance using EMAT based NDE could provide the capability to use measured wear data to improve the accuracy of the wall thinning managements.
Predictive plant model for piping system could utilize the results of wall thickness inspections to enhance the wall thinning predictions.
26
Concluding Remarks
Current investigation on pipe wall thinning management was summarized.
The method for determining region of interest of inspection was shown based on the predictive model.
Under the use of latest inspection methods, it was shown that the probability of detection plays an essential role in the management.
Current and future works on the advanced management were considered.
27
Acknowledgements
This work has been performed as a part of the National Research Project for Enhancement of Measures against the Ageing of Nuclear Power Plants sponsored by the Nuclear and Industrial Safety Agency (NISA) in Japan. This research was also supported in part (F. Kojima) by the Asian Office of Aerospace Research and Development (AOARD) . NISA project member: Dr. A. Nakamura, Dr. Nagaya (INSS), Professor Y. Tsuji (Nagoya Univ.) Dr. M. Naito (IAE), Professor H. Nishino (Tokushima Univ.), Dr. H. Furukawa (JAPEIC), Dr. K. Toiyama (Industrial Res. Inst. , Hiroshima), Professor T. Takagi, Professor T. Uchimoto (IFS, Tohoku University), Professor K. Yamanaka, Professor Y. Ohara (Tohoku University), Professor H. Kikuchi (Iwate University), Dr. H. Nakamoto (Kobe University), Dr. D. Kosaka (Polytechnic University)
28