Simulation Model for SG Eddy Current SG Inspection

Transcription

1 Simulation Model for SG Eddy Current SG Inspection Saptarshi Mukherji, Anton Efremov, Pavel Roy, Portia Banerjee, Anders Rosell, L.Udpa Michigan State University & Rick Williams, Nathan Driessen, James Benson EPRI 35 th Annual EPRI Steam Generator NDE Workshop, Clearwater Beach, Florida July 2016

2 Outline Benefits/Motivation Applications of Simulation Model Objectives Summary of SGTSIM Features Simulation Results Current Ongoing work Conclusions and Future Work 2

3 Benefits of Simulation Models What can simulation models do? Predict EC probe signals for different defect geometry Test bed for generating defect signatures Effect of probe wobble, frequency, sludge characteristics on probe measurements Training tool Optimization of sensor/system design Useful in Probability of Detection (POD) Models at low cost Use in Reverse engineering models for finding root cause of complex signals Key Advantages of Simulation Model: Provides an inexpensive and fast method to simulate realistic defect geometries 3

4 Practical Applications of Simulator Software Utility Engineers Assist in complex signal interpretation & Tube Integrity Assessments Assist in POD calculations Inspection vendors Assist in signal interpretation Assist in SSPD development Probe developers Aid in probe design NDE instructors Training tool Generate signals for training data Researchers / Qualifying Institute Generate signals for probe technique qualification (ETSS s) Generate signals for performance demonstration (QDA/AAPDD) 4

5 Project Objectives Continue development of a software tool that is capable of accurately simulating signals representing SG EC inspection data from various: Tube degradation mechanisms SG tube geometries Eddy current coil configurations Sludge compositions Foreign objects Generate simulated EC signals in a format that is representative of signals generated by field SG eddy current test equipment. Simulated data will be formatted so that it can be read and processed by commercial data analysis software 5

6 SGTSIM v4.0 (Beta) Features -Summary Tube Geometry Support Plate Tube Sheet Free Span Bobbin Pancake (.115) Probes MRPC X-Probe Point Location Freespan TSP TTS ID/OD Defect Geometry Shape Circular Elliptical Rectangular Real Cracks Orientation Circ Axial Free span TTS TSP Bobbin probes - (0.510 TF, TF, HF,.610 MR, TF, HF, MR, MR, MR, TF, MR, MR) Other features: Data Export in MIZ Format Batch Processing Generate noise added signal All results are experimentally validated 6

7 Example of User Interface Window Data Input screen Choose defect type as Real Crack Choose Defect orientation as ID or OD 7

8 GUI for Real Cracks Axial Notch - ID Axial Notch - OD Circumferential Notch - ID Circumferential Notch - OD 8

9 Develop export modules for multichannel probe configurations Using SGTSIM batch export feature, multi-frequency and multi-coil data can be generated and combined into a multi-channel data file in csv format This csv file can be used to generate a corresponding MIZ-80 datafile by the C2D converter from Zetec, and viewed in Eddynet display. Simulated signals User selected files CSV file Generator.csv file C2D Converter Channel configuration Process & Other Header Information Generated MIZ-80 Data File 9

10 Validation of SGTSIM Database of simulated calibration standard flaws eddy current signals Position Depth (%) Defect Type Probe Frequency (Khz) O.D./I.D. 100,60,40 O.D./I.D. 100,60,40 Axial Circ Axial Circ.610 X-Probe 400, 300, 100, TF Bobbin 550, 280, 140, 35 O.D. 100,60,40, 20(4) FBH.610 TF Bobbin 550, 280, 140, 35 O.D. 100,60,40, 20(4) FBH.560 HF Bobbin 650, 320, 170, 35 O.D. 100,60,40, 20(4) FBH.510 TF Bobbin 750, 380, 200, 50 O.D./I.D. 100, 60, 40, 20 n/a Radial exp (0.016 ) Axial Circ 360 deg Expansion Rotating +Pt (PP11A) Rotating Pancake (P115A) 300, 200, 100, X-Probe 400,300,100,50 Calibration Flaw Simulations in Progress Position Depth (%) Defect Type Probe Frequency (Khz) O.D. 30 Groove.610 X-Probe 400,300,100,50 10

11 Exported Bobbin Signals displayed in Eddynet 40% Circular Flaw- Diff channels Experimental Data Simulated Data 750kHz 380kHz 200kHz 50kHz 750kHz 380kHz 200kHz 50kHz 11

12 Exported Array Signals displayed in Eddynet 140kHz Experimental Data 400kHz 100% Axial Flaw 140kHz Simulated Data 400kHz 12

13 Exported MRPC Signal displayed in Eddynet 080 PAN +Pt 115 PAN Experimental Data 40% O.D. Axial Flaw +Pt Simulated Data 115 PAN + Pt + Pt 115 PAN 115 PAN 13

14 Capability to Read Crack Profile from Excel File Axial Notch Profile from MET data - mesh Profile Table +Point Probe ; 300KHz ; FARLEY-1 # Length (mm) % TW Experimental Signals Simulated Signals 14

15 Quantitative Metrics Flaw size Coil type Magnitude Difference Axial notch Depth 100% Length 0.38 Width Axial notch Depth 57% OD Length 0.38 Width % TW Circular hole X Probe (%) Phase Difference RPC +Point RPC Pancake RPC +Point RPC Pancake Axial X Probe Circumferential 100% TW hole.610 Bobbin probe (º) SG Exam Guidelines Data Quality Acceptance Criteria: Phase changes on normalized reference signal ±5 Amplitude changes on normalized reference signal ±20% 15

16 Current Ongoing work 1. Setup SGTSIM to run in EPRI HPC Linux based SGTSIM developed for testing HPC implementation Linux version installation at EPRI HPC 2. Training dataset- Signals generated where experimental data is unavailable in ETSS dataset 3. Reverse engineering - determine root cause of complex field signal 4. Noise incorporation in simulated signals for POD analysis 5. Simulation of : Loose parts (Carbon steel, Stainless steel, Copper) Complex signals 16

17 1. Development of a new SGTSIM GUI Cross- platform (Windows and Linux) Modular structure (Easy to extend the functionalities) Unified control over the Local and Remote machines 17

18 2. SGTSIM for Training data generation Flaw shape: Rectangular, Crack Width:0.005'', Tube OD:0.75'',Tube wall Thickness:0.043 Frequency Orientation Probe Position Length depth Axial Circ Pancake +Pt 20 OD ID OD : finished 100 Flaw shape: Rectangular,Crack Length:0.5'',Crack Width:0.005'', Tube OD:0.875'',Tube wall Thickness:0.052 Position depth Frequency Orientation Probe Axial Circ pancake +Pt ID 39 : finished Flaw shape: FBH of diameter:0.05'', Tube OD:0.875'',Tube wall Thickness:0.052 depth Frequency Probe pancake +Pt 100 : finished 18

19 3. Use of SGTSIM for Reverse Engineering of Complex Field Signals Complex Field Data A 150 khz +Pt Mag Bias coil showing volumetric indication of 0.09 Volts was observed 0.2 inch above the hot leg TTS in a one inch sludge collar ECT Graphics of 400 khz +Pt coil w/ volumetric indication 19

20 3. Use of SGTSIM for Reverse Engineering of Complex Field Signals Possible causes Loose part wear in the sludge pile region Pitting Hard sludge collars with a small area where sludge deposits have flaked off Lap signals Reverse Engineering Algorithm Develop mesh for each test case ECT Simulations of test cases using SGTSIM Comparison of calibrated simulated signals vs Experimental signals for closest match Determine possible cause of field data 20

21 Zoomed in Defect 3. Use of SGTSIM for Reverse Engineering of Complex Field Signals Case I. Loose part wear in the sludge pile region Mesh of the Geometry Tube Tube Sludge Sludge Sludge Sludge Properties: Permeability: 7, Conductivity: 0 Defect Dimensions:12 % TW 0.2 length, width 21

22 3. Use of SGTSIM for Reverse Engineering of Complex Field Signals Initial results Simulation results Field data Liz Plot shows defect indication 22

23 4. Capability to Inject Simulated Noise into Simulated Signal 1 D Random noise Horizontal and vertical components of the signal are affected by noise independently. Variance of noise is changed to control power of the injected noise. Noise added to Hor. Com. Noise added to Vert. Com. Noise Free SGTSIM simulated bobbin signal for 80% through-wall OD defect near TSP Noise variance = Noisy signal Noise variance =

24 4. Capability to Inject Simulated Noise into the Simulated Signal 2D Random & Periodic noise GUI for controlled Noise injected simulated data + point probe, 100KHz, Horizontal channel Experimental field data AAPDD, + point probe 100KHz, Horizontal channel User selected noise parameters Simulated Defect ANO Defect 24

25 4. Capability to Inject Simulated Noise into the 4. Capability to Inject Simulated Noise into the Simulated Signal 2D Random Periodic noise Simulated Signal 2D Random & Periodic noise Simulation flaw signal Random noise Zero-mean additive Gaussian noise Periodic noise Simulation signal with noise + + = 25

26 Summary/Conclusions A computational model for simulating SG tube inspection has been developed and validated using experimental measurements Experimental validation of Simulation results from a variety of probe geometry, tube geometry and defect geometry has been presented The model has several potential applications - The model can be used as test-bed to generate signals from defects that are expensive to fabricate The model allows user to make controlled variations of material properties, defect profiles and other operational parameters and observe effect of these changes on the measured signal (useful in POD calculations) The model can be used as a reverse engineering tool for determining root cause of complex signals 26

27 2017 Future work Develop capability to produce array probe signals from each of the individual array coils Develop EC signal simulation capability for: Simulation of AVBs and lattice grid geometries. A shielded rotating pancake coil probe Demonstrate use of SGTSIM for MAPOD applications 2018 Develop EC signal simulation capability for: Tube U-bend region Wear flaws and cracks in U-bend region Broached tube supports Develop SGTSIM Solver for faster performance Release SGTSIM Ver

28 Thank You Questions? 28