Validation of simulation software for NDE applications in utility industry

Validation of simulation software for NDE applications in utility industry Thiago Seuaciuc-Osorio, George Connolly, Feng Yu and Mark Dennis Electric Power Research Institute The 5th International CANDU In-Service Inspection Workshop in conjunction with the NDT in Canada 2014 Conference June 16-18, 2014 Eaton Chelsea Hotel Toronto, ON (Canada) Outline • Background • NDE Simulation Software: CIVA • Validation of CIVA Simulation Results Summary â 2014 Electric Power Research Institute, Inc All rights reserved Our History… • Founded by and for the electricity i d t iin 1973 industry • Independent, nonprofit center for public interest energy and environmental research • Collaborative resource for the electricity sector • Major j offices in Palo Alto,, CA;; Charlotte, NC; Knoxville, TN – Laboratories in Knoxville, Charlotte and Lenox Lenox, MA © 2014 Electric Power Research Institute, Inc All rights reserved Chauncey Starr EPRI Founder Our Members… • 450+ participants in more than 40 countries ti • EPRI members generate more than 90% of the electricity in the United States • International funding of more than 15% of EPRI’s research, development and demonstrations • Programs funded by more than 1,000 energy organizations © 2014 Electric Power Research Institute, Inc All rights reserved Challenges & Opportunities Associated with NDE Modeling &Simulation • Increasing scope of NDE – Long Term Operation/License renewal – Buried piping; Concrete, etc • Ph Physical i ld demonstrations t ti off NDE ttechniques h i are increasingly expensive • Modeling can be used as a training tool for new work force • Theoretical justification through modeling is considered as a possible acceptable way of meeting the regulatory requirements NDE simulation codes must be validated against experimental data to determine their suitability for industrial application! © 2014 Electric Power Research Institute, Inc All rights reserved CIVA: Software Dedicated to NDE Simulation – Developed by Commissariat l’Energie Atomique (CEA), France – Multiple techniques and modules • UT : Ult Ultrasound d • RT : X Rays • ET : Eddy Currents processing data reconstruction reconstruction…)) • Analysis tool (signal processing, – Generic Simulation Procedure of ET • Specimen • Probe • Inspection • Flaws • Acquisition A i iti Run Analysis â 2014 Electric Power Research Institute, Inc All rights reserved Off-axis Detection • This study observes detection of reflectors away from the central axis of ultrasonic beam (skewing) • A circular 0.5” 2.25MHz conventional probe is used; scanning performed f d using i ttransverse waves att 45° (steel) ( t l) via i a plexiglass l i l wedge GE SE1057 • Data collected by Zetec Omniscan MX 16-128 – controlling software: Zetec Ultravision 1.2R7 • ATCO LPS-1000 encoder used for motion control along g two axes © 2014 Electric Power Research Institute, Inc All rights reserved Experimental Apparatus • A 304 SS reference block is used for experimentation and simulation – Overall dimensions 101.6mm×76.2mm×304.8mm (H×D×W) – Nine side-drilled holes as reflectors (Ø1.5875mm), ranging in depth from 6.35mm to 88.90mm (the ninth is not used) – Side-drilled holes are not though-holes; they are drilled ⅔ of the way through – x is the scan direction and y is the index direction © 2014 Electric Power Research Institute, Inc All rights reserved Experimental Procedure • Calibration for wedge delay, exit point from wedge front and shear wave velocity • Raster scanning g is p performed in 1mm steps p in both scan (x) and index (y) directions – Five different skew angles are used, varying from 135° to 195° – two cases are shown here: 150° 150 and 195° 150° positive skew © 2014 Electric Power Research Institute, Inc All rights reserved index scan 195° negative skew Comparison at 150° Positive Skew • CIVA simulations are run in “Direct” mode; no reflections nor mode conversions are included – cumulated side views: 150° 150° 4 5 6 7 SIM EXP CUMULATED SIDE VIEW CUMULATED SIDE VIEW • Comparison is favorable; third through seventh SDHs detected experimentally • Differences – first two SDHs are not detected experimentally but are strongly present in the simulation – CIVA predicting response along the length of the hole (was also the problem at the negative skew) instead of only at the corner © 2014 Electric Power Research Institute, Inc All rights reserved 10 Austenitic Stainless Steel Piping Sample • Piping sample from 10.0” NPS pipe – contains two circumferential flaws whose CL are at θ=30.0° and θ=78.1° © 2014 Electric Power Research Institute, Inc All rights reserved 16 Experimental Procedure • A circular 0.25” 3.5MHz conventional probe is used; scanning performed using transverse waves at 45° (steel) via a plexiglass wedge – coupling p g between p probe and wedge g achieved by y mineral oil – coupling between wedge and part achieved by running water • Data collected by Zetec Omniscan MX 16-128 – controlling software: Zetec Ultravision 1.2R7 • ATCO LPS-1000 encoder used for motion control along two axes © 2014 Electric Power Research Institute, Inc All rights reserved 17 Experimental and Simulated Results • CIVA simulations performed using • (top) cumulated VC top view, single contact element at 3.5 MHz filtered by time to remove – Simulated scan performed in 89 rows (0 (0.8 8° b k ll reflections backwall fl ti and d (b (bottom) tt ) apart); in each row, 35 data are collected cumulated VC end view (1.0 mm apart) 2 EXP CUMULATED TOP VIEW SIM CUMULATED TOP VIEW CUMULATED END VIEW CUMULATED END VIEW EXP © 2014 Electric Power Research Institute, Inc All rights reserved 18 SIM Comparison Summary • Flaws are well located by both experiment and simulation • Differences – CIVA overestimates length g of first flaw;; experimentally p y it is underestimated – Both methods underestimated length of second flaw – CIVA underestimates strength of reflection from first flaw relative to the second flaw flaw CL flaw length flaw CL flaw length actual 30.0° 10.6° 78.1° 14.8° experimental 30.1° 8.9° 77.4° 12.2° simulated 30.0° 11.8° 78.4° 12.9° © 2014 Electric Power Research Institute, Inc All rights reserved 19 UT Simulation Summary • Three comparisons have been observed: – Quality of CIVA off-axis predictions from SDH – Relative reflection strengths and depth estimations from notches cut into steel block – Quality of experimental and CIVA-estimated location of circumferential i f ti l flflaws iin austenitic t iti stainless t i l steel t l piping i i sample l • Good qualitative and visual agreement between simulation and experiment given the main limitations: – no noise present in CIVA simulations – user must be aware of CIVA simulation options, particularly those controlling number of modes and reflections – options are available to account for structural noise and other simulation phenomena but computation time is greatly increased • CIVA simulation performed adequately when compared against experimental measurements for notched block and austenitic stainless steel piping sample © 2014 Electric Power Research Institute, Inc All rights reserved 20 Eddy Current Inspection of Steam Generator Tube w/ Holes © 2014 Electric Power Research Institute, Inc All rights reserved 21 CIVA ET simulation 400 kHz bobbin coil, differential mode, ASME standard, IN 600, OD: 0.875” , WT: 0.05” © 2014 Electric Power Research Institute, Inc All rights reserved 22 CIVA ET simulation vs experimental Results 400 kHz bobbin coil, differential mode, ASME standard, IN 600, OD: 0.875” , WT: 0.05” Simulation results Experimental results Red: 100% thru; Black: 69%; Blue: 19% © 2014 Electric Power Research Institute, Inc All rights reserved 23 CIVA ET Simulation vs Experimental Results 400 kHz bobbin coil, absolute mode, ASME standard, IN 600, OD: 0.875” , WT: 0.05” Red: 100% thru; Black: 69%; Blue: 19% © 2014 Electric Power Research Institute, Inc All rights reserved 24 CIVA RT Screen Dump Tube Voltage: 220 kV; Tube Current mA; focus-to-film distance : 25”: Exposure Time: 30 s © 2014 Electric Power Research Institute, Inc All rights reserved 25 CIVA RT Simulation vs Experimental Results Experimental Carbon steel (0.125, 0.25” thick) Analytical, optical density (0-4) ( ) © 2014 Electric Power Research Institute, Inc All rights reserved Analytical+ Monte-Carlo, optical density (0-4) 26 Bimetallic Welds Specimen © 2014 Electric Power Research Institute, Inc All rights reserved 27 CIVA RT Simulation vs Experimental Results: Bimetallic Welds Experimental © 2014 Electric Power Research Institute, Inc All rights reserved Simulated 28 Summary • General good qualitative agreement was achieved between experimental and CIVA results for the simulations performed in this study – useful to interpret the underlying physics and signal observed in NDE measurements; – provide a useful tool when training inspectors; – determine the influential parameters thru parametric studies studies • Like any simulation tools for engineering applications, CIVA represents simplified and idealized NDE inspections – critical to obtain the accurate information of the input parameters needed in CIVA simulation; – Critical to validate CIVA models against experimental data for generic inspection or on a case-by-case basis for complex inspections with respect to technique justification and demonstration for plant operation © 2014 Electric Power Research Institute, Inc All rights reserved 29 Together Shaping the Future of Electricity Together…Shaping © 2014 Electric Power Research Institute, Inc All rights reserved 30 ... results for the simulations performed in this study – useful to interpret the underlying physics and signal observed in NDE measurements; – provide a useful tool when training inspectors; – determine... History… • Founded by and for the electricity i d t iin 1973 industry • Independent, nonprofit center for public interest energy and environmental research • Collaborative resource for the electricity...Outline • Background • NDE Simulation Software: CIVA • Validation of CIVA Simulation Results Summary â 2014 Electric Power Research Institute, Inc All rights reserved Our History… • Founded