ULTRASONIC CHARACTERIZATION OF DEFECTS
ULTRASONIC CHARACTERIZATION
OF DEFECTS
1
INTRODUCTION:
This report treats the ultrasonic measurements
performed on the new V-welded carbon steel blocks and development of algorithms
for feature extraction, flaw position estimation, etc. The software used below
is partly based on the software used for immersion testing of simulated and
artificial defects from the previous project Ultrasonic Characterization of
Defects [1, 2, 3]. However, there has been a substantial rewriting of the
algorithms, and new algorithms have also been added to fit the new contact
measurements of the V-welded steel blocks. A substantial effort has also been
put in complementary manual inspection related to interpreting the ultrasonic
data obtained from the scanner. Section 2 describes the manufacturing of the
new carbon steel blocks and the "natural" flaws that are implanted in
them. The main part of the text, Section 3, describes the B-scan (and D-scan)
measurements that have been performed and discusses the features found in the
measurements. In Section 4 the algorithms for: position estimation, region of
interest selection, feature extraction, and depth normalization are presented.
This section also contains a discussion of the different defect classes
proposed and a comparison of signal features between different flaw types. At
the end of the section there is also a comparison of artificial flaws contra
natural flaws. Finally, Section 5 gives a summary of performed work and Section
6 gives the conclusions.
2 .REALISTIC
TEST BLOCKS:
The
first step of the project Ultrasonic Defect Characterization includes designing
and manufacturing test blocks made of carbon steel (CS) containing realistic
flaws in the weld. The original planning concerned 12 welded blocks, each with
3 flaws in the V-weld. After the discussion with the manufacturer (Sonaspection
International Ltd.) the number of blocks has been reduced to four, mainly due
to the substantial difference in production cost. Other important factor
supporting this change was related to logistics in our laboratory which is not
well prepared for handling heavy steel loads. It should be noted however, that
despite the reduced number of blocks, the number of flaws has remained the same
as planned, i.e. total 36 flaws of different types. Thus this change does not
influence the extend of the experimental work. We decided to split block
manufacturing in two steps, first to get CS blocks and then, after acquiring
and processing ultrasonic data from the flaws, to manufacture similar blocks
made of stainless steel (SS). The main reason for that is the need of practical
experience before deciding what types of flaws should be manufactured in the SS
blocks. However, the manufacturing of the SS blocks has been postponed for
reasons which will be explained later in this report.
2.1
DESCRIPTION OF THE CS TEST BLOCKS:
Four
blocks, each with 9 various flaws, were designed in collaboration with, and
manufactured by Sonaspection International Ltd. All blocks have dimensions 42
mm x 400 mm x 600 mm and consist of two carbon steel plates, welded together (V-weld).
The defect types and sizes manufactured in the blocks are summarized in Table
1. From Table 1 can be seen that our flaw population consists of 24 sharp flaws
(various types of cracks and lack of fusion) and 12 soft type flaws (slag,
porosity and over penetration). Closer analysis shows that we have three
different types of cracks characterized by various sizes, angles and locations.
The cracks were manufactured by mechanical fatigue and were implanted by
semi-direct insertion (created before the welding process or at a
pre-determined stage during welding). We have also natural sharp flaws in the
form of lack of side wall fusion. This gives an idea of spread in the sharp
flaw class which should also result in the variation of their ultrasonic signatures.
On the other handsoft defects by their nature should result in similar
ultrasonic responses, independent of their location and size.
Sonaspection
delivered detailed drawings of the blocks, as built flaw details and
photographs of flaw signatures for each crack. Copies of Sonaspection block
drawings are in Appendix B. A copy of the Sonaspection report was delivered to
SKi. The test blocks have been subjected to careful ultrasonic inspected in our
lab and all the defects were localized according to the reports from
Sonaspection.
TEST
BLOCK MEASUREMENTS
This
section describes the used transducers, the measurement setup, and the
measurements performed.
3.1
TRANSDUCERS :
The
contact UT inspection of the blocks has been performed using mechanized scanner
and a digital ultrasonic system based on Saphir PC board. B-scans for each flaw
were acquired and a flaw data base has been created. Two miniature screw-in
transducers from Panametrics, with center frequencies 2.25 MHz (type V539-SM)
and 3.5 MHz (type A545S-SM) were used in the (shear wave) contact inspection:
Both transducers had nominal element size 0.5" (13 mm) and were assembled
by screwingdirectly into miniature angle beam wedges type ABWM-5T also from
Panametrics. Six different angle beam transducer configurations, listed in the
Table 2 above, were created in this way. Advantage of this solution is quite
obvious, by using the same active element we have obtained angle transducers
with very similar characteristics. Since ultrasonic response of a particular
flaw is determined both by the flaw type and by the transducer characteristics
it is essential for defect characterization to keep transducer characteristic
as constant as possible.
Measurement
Setup Four different scanning methods were used. The aim was to make direct
measurements from the top side of the steel blocks shown in Figure la. However,
for smaller angles direct measurements were obstacled by the upper surface of
the V-weld. In such cases indirect measurements or measurements from the
backside were performed instead, which is shown in Figure lb and c. Another
reason for making indirect (or backside) measurements was low amplitude of thereflection
obtained in the direct measurement due to the angle between the transducer main
beam and certain flaws, like sidewall cracks. The fourth scanning method is
shown in Figure Id, which is so-called D-scans. That is, the probe is moved
along the weld side-wise. D-scans are interesting because they reveal how the
defect response varies along the defect. It is also interesting to see the
response from the weld itself, both with and without a defect present.
Typically the shap of the weld variessubstantially spatially and D-scan shows
this variationrather clearlyMeasurements The performed measurements are
displayed in Table 3 and Table 4. The measurements consist of B- and D-scan
data matrices and the total number of measurements are 2x133. The main part of
data comes from the welded steel blocks described above, but new measurements
have also been performed, for comparison, on the two old aluminum blocks with
artificial defects used in the previous project [3]. In addition to these
measurements all flaws have also been investigated manually. The artificial
flaws in Table 4 named SBH are side-drilled holes, and the ones named S are
cracks (notches).1 3.4 Results In this section a number of B-scans from each
defect type are shown for illustration. They are selected so that both common
features and feature variations are represented for each defect type. Note
also, that some of the images contain echos from non-defect parts of the weld,
like the topor bottom surface of the weld or the steel-weld junction. These
echos are explained (if possible) when they are encountered. In the figure
titles the name of the data files are given. An example is p28b 1 3 5, where
p28 means porosity flaw 28, b means backside measurements, 1 indicates that the
flaw is located in test block PL4501, and 3 5 is the used transducer frequency
(3.5 MHz). The B-scans presented in the following subsections are from
measurements with the 3.5 Mhz transducer. The reason for showing the 3.5 Mhz
transducer only is that the measurements is performed on carbon steel blocks
with very little material grain noise. This implies that the 2.25 MHz and the
3.5 MHz transducers should give similar results with the exception that the 3.5
MHz transducer has a higher center frequency and thus shorter wavelength and
therefore higher resolution (both transducers has approximately the same
bandwidth). There is, however, a comparison of the two transducers in Appendix
A. 3.4.1 Center Cracks The center cracks found in the steel test blocks can be
either straight or slightly tilted. Figure 2 shows indirect measurements using
the 45-degrees 3.5 MHz transducer. All echos around 60 mm stem from direct
reflections from the bottom surface of the weld and all echos around 120 mm
stem from indirect reflections from the top surface of the weld. One can see
that there are two rather strong peaks (too strong to be diffraction echos) in
all three B-scans. We can not find an unambiguous explanation for the presence
of the second echos, but we believe that it must depend on the structure of the
cracks implanted into the weld. These two peaks do not always occur in signals
from center cracks, if we for example scan the defect in Figure 2a from the other
side of the weld, we get only a single peak. This effect is also less
pronounced if a higher angle probe is used. Figure 3 shows B-scans of the same
defects obtained with a 60 degree transducer, and there is for example only one
peak in Figure 3b (see Section 5 for a further discussion on this topic).
3.4.2 SIDEWALL CRACKS:
The sidewall cracks
are located in the steel-weld junction and are therefore tilted with the same
angle as the weld (30 degrees). This makes it difficult to apply direct
measurements and all measurements are therefore performed from the backside or
indirect. Figure 4 and Figure 5 show backside measurements performed with the
3.5 MHz 45- and 60-degrees transducers. All echos around 60 mm for the
45-degree transducer, and around 84 mm for the 60-degree Distance Irom center
of weld to transducer (mm) (a) Distance Irom center ottransducer come from the
top surface of the weld. No double echos, like for center cracks, are noted for
the sidewall cracks. 3.4.3 Lack of Fusion The lack of fusion (LOF) defects are
located in the same way as sidewall cracks, and are therefore also tilted 30
degrees. This results in the same difficulty to make direct measurements here
as well. Figure 6 and show measurements with the 45- and the 60-degree
transducers, respectively. In the same way as for the previous backside
measurements, echos around 60 mmfor the 45-degree transducer, and echos around
84 mm for the 60-degree transducer come from the top weld surface. Some of the
LOF measurements have two peaks like the center cracks had (Figure 6b and c)
for the 45-degree transducer, however they are more separated than for the
center cracks. The double echos were only seen when the 45-degree transducer
was used.
3.4.4 SLAG: The echos from the slag inclusions
are rather distinct regardless of which transducer that is used. Figure 8 and
Figure 9 show four examples using the 45- and 60-degree transducers. No direct
measurements were performed because of practical reasons (the transducer is
obstacled by the weld surface), see Section 3.2.
Comments
Post a Comment