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

© 2014 Electric Power Research Institute, Inc. All rights reserved.

2


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

3


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.

4


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.

5

CIVA: Software Dedicated to NDE Simulation
– Developed by Commissariat à 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

© 2014 Electric Power Research Institute, Inc. All rights reserved.

6


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.

7


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.

8


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
9


Comparison at 150° Positive Skew

• CIVA simulations are run in “Direct” mode; no reflections nor mode conversions are
included
– cumulated side views:
150°

150°
3
4

1
2
3
4

5

5

6

6
7

7

8

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


Comparison at 195° Negative Skew
• Cumulated side views:
195°

195°

1
2
3
4
5
6
7
8

EXP

• No SDH is detected experimentally; though there are blurred indications
for upper SDHs
• Simulated data show strong detection of every SDH
•S
Simulated results need further
f
investigation to determine the reason for
f
these signals
© 2014 Electric Power Research Institute, Inc. All rights reserved.

SIM

CUMULATED SIDE VIEW

CUMULATED SIDE VIEW

11


Notched Block
• Notched block is modelled as homogeneous isotropic steel
– Dimensions: 255.6mm×152.4mm×25.298mm (10”×6”×1”)
– Notches
N t h vary ffrom 1.27mm
1 27
(5% TWT/TWE) to
t 22.86mm

22 86
(90%
TWT/TWE) in height from back surface
• Probe is 0.5” 1.5MHz transverse; wedge at 45°
10

5

9

4

SHALLOW NOTCHES

8

3

DEEP NOTCHES

© 2014 Electric Power Research Institute, Inc. All rights reserved.

12

7

2

6

1


Experimental and Simulated Results
• (top) cumulated VC top view, filtered
by time to remove backwall reflections
and (bottom) cumulated VC side view

• CIVA simulations performed using
single contact element at 1.5 MHz
– Si
Simulated
l t d scan performed
f
d iin 8 rows (15
mm apart); in each row, 456 data are
collected (0.5 mm apart)

• Responses from notches 1, 9 and 10
not discernible due to interference
6

5

7

4

8

3

6

5

7

4

8

3

2

2

EXP

SIM

CUMULATED TOP VIEW

CUMULATED TOP VIEW

CUMULATED SIDE VIEW

EXP

6
7
8

© 2014 Electric Power Research Institute, Inc. All rights reserved.

1

CUMULATED SIDE VIEW

2
3
4

SIM

5
6
7
8

13

1
2
3
4
5

Comparison Summary
• Normalized echodynamic curves of cumulated top view normalized
by (left) amplitude of response from second notch and (right)
amplitude of response from sixth notch
– Simulation tends to overestimate amplitudes of subsequent
notches
DEEP NOTCHES SHALLOW NOTCHES

2

3

4

5

6
7

8

© 2014 Electric Power Research Institute, Inc. All rights reserved.

14


Comparison Summary
• Comparison of measured and actual depths of notches
– Both simulation and experiment tend to overestimate notch
depth i.e., the notch TWT/TWE is slightly underestimated

– Error slightly worsens for shallowest notches
7
5

6

6

5
4

2

3
2

© 2014 Electric Power Research Institute, Inc. All rights reserved.

15

3

4

7


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

2

1

EXP

1

CUMULATED TOP VIEW

SIM

CUMULATED TOP VIEW
CUMULATED END VIEW

CUMULATED END VIEW

1

2

EXP

2
1

© 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

1

flaw 1 CL

flaw 1 length

flaw 2 CL

flaw 2 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°

2

© 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 2 mA; focus-to-film distance : 25”: Exposure Time: 30 s

© 2014 Electric Power Research Institute, Inc. All rights reserved.

25