23 Ginzel
23 Ginzel
23 Ginzel
Ed GINZEL
1
1
Materials Research Institute, Waterloo, Ontario, Canada
e-mail: [email protected]
2011.07.11
Abstract
Civa simulation software has long been a tool for elaborate technique development and validation. But it is also
a very convenient teaching tool. Basic principles are easily illustrated and phenomena better explained by
setting up simulation models. Examples provided would be very time consuming to setup in a lab. It is
concluded that Civa provides an excellent tool for ultrasonic training.
Keywords: Civa, training, simulation, modelling
1. Introduction
Civa is a well known software that provides simulation and modelling tools for ultrasonic,
eddy current and radiographic NDT methods. A search of the papers on NDT.net quickly
identifies hundreds of references to applications of this software. Mostly the software has
been used in technique qualifications, for probe selection and optimisation and for
understanding the origins of signals in complex structures.
Although Civa has components for three NDT methods, only the ultrasonic module is
discussed in this paper.
Civa ultrasonic module is based on a semi-analytical solution to the wave equation. It has
been validated both in-house at CEA and at international trials (QNDE Benchmark). These
trials verified that the predicted responses in Civa are very close to what an operator would
see in the field. Based on the results of these trials we have confidence that Civa simulations
can represent the realities of UT.
With its ability to provide or import complex shapes, Civa is often touted as a great method of
saving costs on building mock-ups. But not all mock-ups need to be complex for Civa to
provide a significant cost benefit. The sort of simple blocks and targets used in basic UT
training can also benefit from this tool.
Often, this writer has found himself as an instructor of ultrasonic testing. Explanations using
equations or line drawings can only go so far. Having Civa at hand has allowed us to provide
much better explanations than could be afforded by pencil and paper. A few slides on a
presentation in a classroom are easily prepared in an hour or two and explanations made far
clearer. An instructor often gauges the effectiveness of their explanation by the look on
students faces. Having presented basic principles in several venues around the world where
the instructors English is often not the native language of the students, it has become very
apparent that the explanations based on good graphics are the most effective. In such
situations the instructional benefits of Civa cannot be understated.
Fabricating simple blocks with simple targets could be a tool used in a classroom. But
permutations and combinations of the parameters (e.g. varying probe frequency and size with
flaw height and tilt) could take many hours of preparation and does not present well in a
classroom. Instead, Civa can provide a series of modelled results plotted on graphs with the
stored signals available to show how the plots were derived.
The remainder of this paper is to simply illustrate how Civa can be used to provide graphic
illustrations of some of the principles in UT.
2. Instructional Modelling
Models provided in this paper are based on some of the aspects of UT that the author has been
asked about by various clients and students. By setting up a simple CAD specimen with a
basic calibration target, the rationale for test methods can be provided. To keep the paper
reasonably short, only five examples have been selected.
Concepts illustrated include:
o The effect of frequency on detection of tilted flaws (manual UT weld
inspection)
o Comparing calibration sensitivities (SDH diameters and manual UT)
o Comparing calibration sensitivities (SDH versus Notches)
o The effect of zone target size on zone separation (pipeline AUT)
o S-scans on SDHs versus notches (Phased-array UT)
2.1 The effect of frequency on detection
Basic level 2 instruction for manual UT of welds states we should use lower frequency probes
to detect flaws that are not oriented perpendicular to the beam. But what is not explained in
this simple guidance is the effect of the flaw size and what happens to the signal relative to
the reference level.
Two probes are compared (one 12mm diameter 5MHz, the other 12mm diameter 2.25MHz).
These are placed on a curved steel surface with a radius of 80mm (80mm is much greater than
the near zone so we are working well into the far field). Two targets are used and placed at
the centre of the modelled steel block. One target is an infinite plane (e.g. the block is a solid
round bar cut in half), the other is a 3x10 rectangle (air boundary) simulating a small flaw.
In each case the probe modelled is positioned over the centre point for maximum return off a
perpendicular incidence from the infinite plane surface or 3x10 rectangle. The probe is
moved in 1 degree increments around the cylinder. Monitoring the echo from the target at the
centre of the cylinder the effect on amplitude can be plotted as an echo-dynamic curve.
The amplitudes are plotted for each case in Figures 1-4.
Figure 1 Echo dynamic from infinite plane with 5MHz probe
Figure 2 Echo dynamic from infinite plane with 2.25MHz probe
Figure 3 Echo dynamic from small 3x10 flaw with 2.25MHz probe
Figure 4 Echo dynamic from small 3x10 flaw with 5MHz probe
The same 5 of rotation that caused a 3.6dB and 9.5dB drop of the backwall (infinite plane)
signal results is only 0.3dB and 1.3dB drop for the smaller 3x10mm target with the same
2.25MHz and 5MHz probes at the same 80mm soundpaths.
Perhaps more importantly, a flaw that is 5 from the perpendicular of the ideal beam provides
a larger response compared to the SDH reference target for a 5MHz pulsed compared to a
2.25MHz pulse (this based on a separate Civa experiment using a 3x10mm representation of
subsurface lack of fusion compared to a 3mm SDH)..
Concern for the rapid drop off of signal amplitude with changing incident angle is not as
relevant for small typical welding flaws as it might be for the large infinite plane (plate
backwall). This can be explained by the target being considered the new emitter. The large
(infinite plane) emitter has less divergence than the smaller (3x10mm rectangle).
2.2 Comparing calibration sensitivities (SDH diameters)
Scanning sensitivities are typically established using a reference target. A popular target for
weld inspection is the side drilled hole (SDH). This is a useful target because the same sort of
reflecting face is presented regardless of the angle of the beam. There is a range of diameters
used around the world and it is worth noting what the differences might be. Even within a
single code several diameters can be identified. E.g. ASME Section V stipulates that for up to
25mm a 2.5mm diameter SDH is used. Over 25mm wall to 50mm the target is a 3mm
diameter SDH and for inspecting wall thickness over 50mm through to 100mm the target is a
5mm diameter SDH. Implied is a decrease in sensitivity as diameter increases. But how
significant is the drop in sensitivity?
A scan using a 5MHz 12mm diameter probe on a 45 refracting wedge is modelled. A series
of SDHs is made from 5mm below the surface to 35mm below the surface of a steel block in
5mm increments.
The test is re-run for SDHs of increasing diameter from 1mm to 6mm in 1mm increments.
Figure 5 illustrates the scan pattern as a B-scan over the targets (3mm diameter used for the
illustration). Figure 6 provides the echo dynamic (maximum amplitude at each scan
increment) for the 6 hole diameters. The Civa software normalises the response to the largest
amplitude and all the others are scaled relative to this maximum.
Figure 5 B-scan of 3mm diameter SDHs with 12mm diameter 5MHz probe at 45
Figure 6 Echo dynamic from the 6 different SDH diameters(5mm deep SDH on left)
In Figure 6 the responses from the SDHs are seen to fit neatly one under the next starting at
the largest response from the 6mm SDH and incrementing down for each of the smaller
diameters down to 1mm diameter. Clearly the first three holes are inside the near field as the
peak amplitude occurs at the fourth target. Expanding the echo dynamic plot for the
maximum response (20mm deep SDH) we see the relationship between diameters in Figure 7.
Figure 7 Echo dynamic from the 6 different SDH diameters at 20mm deep hole
Relative amplitudes are as follows:
Dia. SDH
(mm)
Amplitude
(dB)
Ermolov
Predicted dB
drop
1 0 0
2 -0.6 -1
3 -1.4 -2
4 -3.5 -3
5 -5.7 -4.7
6 -8.7 -7.7
Added to the table are the dB drop values predicted by the Ermolov equations. This summary
demonstrates that the Ermolov predictions are very close to the semi-analytical solution. The
table also suggests that the importance of a 0.5mm increase in diameter between the lower
and middle thickness range for ASME sensitivities may not be as important as the Code
suggests. The difference between the 2mm and 3mm SDHs is less than 1dB (Ermolov
predicts a 0.8dB difference between 2.5mm and 3mm diameter SDHs just after the near zone).
This is less than the tolerance allowed between periodic calibration checks. However, the
difference between the 3mm and 5mm SDHs as used in ASME produces a notable dB
difference (4.3dB).
2.3 Comparing calibration sensitivities (SDH versus Notches)
Sensitivity calibrations are also made using notches (typically 10% of the wall thickness).
Unlike SDHs notches are not ideal and their reflectivity changes depending on the angle of
incidence on them. But many standards still use notches as an acceptable calibration option.
Pipe weld inspections particularly prefer notches, probably because it is easier to mill a notch
into the pipe than try to place a SDH in a limited cylindrical section.
To compare the sensitivities achieved from SDHs and notches we model a simple scan across
2 SDHs and 2 notches. The holes are 3mm diameter and the notches 10% of wall (a 25mm
wall plate is used so the notches are 2.5mm high). The targets and scan path are illustrated in
Figure 8. Again, the 12mm diameter 5MHz probe on a 45 refracting wedge is used.
Figure 8 Model for comparing notch response to SDHs
The SDH centre positions are at 8mm and 16.5mm depths from the scan surface.
The dynamic scan results provide a series of peaks as each target is passed. These are seen in
Figure 9. Target responses from left to right in Figure 9 are as follows:
1 1.5 skip notch
2 Full skip notch
3 Half skip notch
4 Lower SDH on 3
rd
half skip leg
5 Lower SDH on 2
nd
half skip leg
6 Lower SDH on 1
st
half skip leg
7 Upper SDH on 3
rd
half skip leg
8 Upper SDH on 2
nd
half skip leg
9 Upper SDH on 1
st
half skip leg
Figure 9 Echo dynamic from 3mm SDH and 10% notch in 25mm thick plate
The 10dB difference between maximum responses is perhaps the first and most noticeable
difference between these two common sensitivity setting methods. Converting the dB
responses to percentage of screen heights we can construct an approximation of what the
Scan path
1
2
3
4
5
6
7
8
9
DAC might look like. This is done in Figure 10 where the SDH DAC is normalised to the
notch response.
0
20
40
60
80
100
120
0 20 40 60 80 100 120
%
F
S
H
Soundpath (mm)
DAC for Notch and SDH
3mm SDH
10% Notch
Figure 10 DAC curves for notch and SDH
2.5 The effect of zone target size on zone separation (pipeline AUT)
In pipeline AUT the use of zonal discrimination has demonstrated to be the most effective
option for rapid inspections. Zones are constructed by dividing the wall thickness into
intervals identified by the bevel with the fill region of the weld not to exceed zone heights of
more than the weld pass thickness. In fact the DNV standard, OS F-101, recommends that no
zone height exceed 3mm. Because the zones are addressed by individual probes or phased-
array focal laws in a fixed position, elaborate procedures are imposed to ensure that the zones
are well discriminated. This is done using the calibration block which uses flat bottom holes
(FBH) located at the centre of the zones and aligned along the bevel fusion line. ASTM E-
1961 defines the degree of discrimination required based on the relative responses of the
targets in the zones above and below the primary zone in a scan of the calibration block. It
requires that the response from the targets above and below in the same plane have an
amplitude not more than half the primary target and not less than 20% of the primary target.
Too much overtrace and the operator cannot be confident of the depth extents of a flaw and
without adequate overtrace the operator cannot ensure that there is adequate sensitivity in the
region between two zones.
The process has relied on a compromise of either larger targets in smaller zones or smaller
targets in larger zones to maintain the amplitude discrimination.
Some simple experiments in Civa allow us to examine the effects of varying these parameters.
A simple zone technique was modelled using a 5MHz 60 refracting wedge directing the
beam at the midpoint of three zones. Zone midpoints were separated vertically by 2.3mm.
Figure 11 Zone targets for 2.3mm zones in Single Vee bevel
In the first case 3mm diameter FBH targets were used and the second used the same zone
height but 2mm diameter FBHs. The bevel angle was 60 included so the refracted angle
makes perpendicular incidence at the targets. The two configurations are illustrated in Figure
11. With a 2.3mm vertical height for each zone the bevel surface length that lies in the zone
is 2.6mm long (the hypotenuse of the triangle with a height of 2.3mm). When a 3mm
diameter FBH is centred in each zone there is 0.4mm overlap of the adjacent targets above
and below. When a 2mm diameter FBH is placed in the centre of each zone there is a 0.6mm
gap between the tips of the FBH targets.
In spite of the physical overlap of the adjacent targets, the 3mm FBH configuration provides a
better zone separation signal than the 2mm FBHs. Figure 12 shows the echo dynamic curves
for the probe centred on the middle FBH as a line scan moves the probe past the lower, then
middle and then upper FBH targets. To the left is the response that occurs for the 3mm FBH
targets and on the right is the responses expected for 2mm diameter FBHs. The 3mm targets
2.3mm
provide an acceptable separation between zones (-10dB and -13dB) whereas the 2mm targets
centred at the same locations produce an unacceptable response (the lower zone is only 5dB
lower than the main zone).
Figure 12 Zone responses3mm targets (left) and 2mm targets (right)
The increased response in amplitude off adjacent targets, in spite of their surface areas not
overlapping into the main zone, can be explained by the increased sensitivity to off-axis tip
diffractions that results when 6-7dB extra gain is added to the receiver in order to raise the
echo from the smaller target to the reference amplitude.
2.6 S-scans on SDHs versus notches (Phased-array UT)
In phased-array setups using S-scans the effect of angle dependence and target type for setting
sensitivity is perhaps more obvious than when mono-element probes are used. To illustrate
how extreme the effect is a series of S-scans is modelled using a scan over a 2mm high notch
and a 3mm diameter SDH. The probe is a 5MHz 16 element linear array with an active
aperture of 9.5mm and a passive aperture of 10mm. All elements are used to make an S-scan
from 45 to 65 in 2 increments. The scan is configured to pass over the notch and then the
SDH in scan increments of 2mm steps from left to right.
Figure 13 illustrates the probe positions for the direct interaction points of the beam at the
edges of the angular sweeps for the two targets. The first point of interaction with the notch is
had for the 65 beam at 8mm into the scan and the last direct interaction with the notch occurs
at scan position 32mm where the 45 beam detects the notch. As the scan continues towards
the SDH, the 65 beam makes the direct path at scan position 62mm and the last beam to
make a direct path to the SDH is the 45 at scan position 90mm. The echo dynamic curves
are placed above the S-scan images. Vertical cursors on the curves are used to indicate the
scan positions where the target interactions with the beams start and end. Horizontal cursors
indicate the maximum and minimum amplitudes associated with the responses from the
targets. Echo dynamic curves are made from the maximum response at each position without
regard for the angle that accounts for that peak amplitude.
Lower
zone
Lower
zone
Upper
zone
Upper
zone
Main
zone
Main
zone
Figure 13 S-scan responses from 2mm notch (left) and 3mm SDH (right)
Whereas in the single element example where we saw large variation between the sensitivity
settings used between a SDH and a notch, this example using a phased-array S-scan illustrates
how that situation is compounded when we use a range of angles instead of just a fixed angle.
Whereas for the SDH, the main contributing factor to the loss of amplitude in an S-scan is
echo-transmittance effects (i.e. reduction of coupled pressure with increasing angle), the notch
suffers from both echo-transmittance losses and mode conversion losses and a redirection of
the beam away from the receiver.
Software algorithms have been developed to compensate for echo-transmittance losses. The
so-called Angle Corrected Gain (ACG) works well on a uniform reflector (like the radius of
the IIW block). Had this been applied to the setup the only compensation required for the
response from the SDH would be that of increased travel time at increasing angles (i.e. time-
corrected gain). Figure 14 shows the difficulties that would be encountered using the
responses from the notch where the angle variations span 20dB and the angle-to angle change
is not a steady increase. Yet for the SDH the span of change due to all factors (echo-
transmittance, increased sound path in metal and wedge) is only 8dB.
-25
-20
-15
-10
-5
0
5
45 50 55 60 65
A
m
p
l
i
t
u
d
e
(
d
B
)
Refracted Angle ()
Amplitude response relative to angle
SDH
Notch
Figure 14 Comparing amplitude compensations required per angle for a SDH and a notch
This should indicate that a surface notch is an inappropriate target for TCG with phased-array
S-scan techniques.
3. Conclusions
Civa simulation software provides a useful training tool to explain many of the common and
some not so common aspects of UT.
Graphic outputs and the 3D images of probe position with available A, B, C and S-scan
presentations provide easy representations of concepts not easily described with the traditional
chalk-board available to instructors.
Acknowledgments
We would like to thank Roman Fernandez and Philippe Dubois of EXTENDE for their help,
time and patience in familiarising us with the many new features of Civa. It is a result of the
ease with which results can be achieved with Civa that I titled this paper Fun with Civa.