Collection: Software

Chapter (9): TOFD Data Collection Software
CHAPTER (9): TOFD DATA COLLECTION SOFTWARE
TOFD systems are used with computer controlled systems that divide the software activities into three
basic stages:
o Ultrasonic Setup
r Data Acquisition
. Analysis
Software menus provide the operator with options to control the probe and instrument parameters,
set the displays used to monitor acquisition in near real-time and to subsequently analyse the data for
identificatiory location and sizing of indications.
9.1 ULTRASONIC SETUP OPT]ONS

Setup menu options for TOFD are used to establish the ultrasonic instrument parameters associated
with the A-scan. As such, the UT Setup display includes parameter setting input and the A-scan,
thereby allowing the operator to see the effect of any parametric changes on the quality of the signal.
Figure 9-1UT setup window for A-scan display tao)
A typical menu for A-scan and probe parameter setup is seen in Figure 9-1. Since the TOFD technique
is used with a standard pulser-receiver system it is common for some of the controls to be unrelated to
TOFD applications. DAC, TCG and some gate aspects (e.9., amplitude, time, trigger-on-peak or
shoulder) may therefore be available but not used for TOFD.
Table 9-1 gives a description of the parameter controls seen on the setup display for the image in
Figure 9-1.
739
Table 9-1 UT setup parameters

ame
Menu ltem Description
Prb Del The probe delay for the displayed channel.
Velocity This field shows the programmed ultrasonic velocity for the selected
channel.
Gain This field shows the gain setting for the selected channel.
Ch The channel number of the A-Scan being displayed. This value may be
modified by using the mouse left button over the Up/Down arrow to the
right of the text, or by using the Up/Down arrow keys.
TX Transmit transducer number
RX Receive transducer number
Start This is the time at the left edge of the A-Scan display. Changing this value
scrolls the A-Scan LeftlRight. Values may be entered via the keyboard (the
Enter key must be pressed for the software to use the value), or the
Up/Down - Arrow/Page keys. The shift key may be used in conjunction with
the Up/Down - Arrow/Page keys to scroll faster. Left Click and hold mouse
button to scroll left or right.
Zoom Controls the amount of data displayed in the A-Scan display window.
Values of 1:7, x2, x3, x4, x5, x6, and FULL are available.
DAC view This list controls the display of the DAC curve on the A-Scan display. (Only
available for Pulse-echo channels).
Show Activates the Peak Search Gates and real-time Amplitude, Range & Depth
reading boxes
Cursors This button when clicked opens the cursor control dialog box.
Persist Activates a coloured fill-in on the A-scan display that remains on-screen
after the A-scan peak has moved
80 Sets the vertical scale to show 80% fuIl scale, with -6dB steps
400 Sets the visible scale to 400%
+2dB l -2dB Increases / Decreases the gain in 2dB steps
9.2 ACQUISITrON OPTTONS

Acquisition menu details are generally those items that relate to how the probes and scanner are
configured, so as to allow the operator to subsequently analyse the data acquired. There is no
consistency ftom one manufacturer to another as to how the division between "Setup" and
"Acquisition" are separated. Some of the items that are considered Ultrasonic Setup Parameters by
one company are grouped with Acquisition menus by another.
Menu displays for other features controlled include:

o Pulser parameters (voltage and pulse duration)
o ADC rates (typically 25,50 and 100MHz with some options for higher and lower)
740
r Bit rates (8, 10 and 12 bit sampling are the standard options, but for some equipment this is
fixed with no option for changes)
o Filters
. Averaging
o Data compression
o Instrument pulse repetition frequency
Windows and Tabs are the common displays for these menus. Figure 9-2 illustrates the "Collection
Hardware Setup" window for the A-scan seen in Figure 9-1. It is seen with the "Charurel" tab opened.
Figure 9-2 TOFD acquisition lngnss (40)
For TOFD applications r part of the configuration involves entering details of the probe and part.
Included in the extra probe details would be PCS, angle and crossing depth (this refers to the depth in
the part where the centre of beam rays cross). Part details would include the acoustic velocities, wave
type (compression or shear), thickness, curvature (when option for correction due to ID or OD testing
is available) and the scan direction (e.g., parallel or non-parallel). An example of some of the data
entry options for the probe and part options is seen in Figure 9-3.
Figure 9-3 TOFD probe and part entry menu (ao)
1.47
Associated with the Data Acquisition information are the scanner input details. This information is
required to identify scan direction(s), scan length, encoder accuracy and sampling interval (e.g., 1 A-
scan waveform every 1mm or 0.5mm, etc.) and the display seen during the data collection. Figure 9-4
illustrates an example of some of the scanner inputs on one tab of a Scanner Setup window. It is in this
window that some manufacturers might provide the option to set the data collection in "Free run"
mode. If operating in free run mode, the data acquisition is then triggered by the internal clock and
timed intervals are used to display the A-scan changes (e.g., 5 samples per second).
This can also be the menu used to calibrate the encoder. In the example menu in Figure 9-4 pressing
the "calibrate" button brings up the encoder calibration window. The operator marks'out a distance of
300mm and puts the leading edge of the probes on the start position. The encoder count is set at 0
(otherwise press the <Clear Count> button). The operator then moves the probes in a straight line until
the leading edge of the probes is on the end position of the marked 300mm. The probes are then held
steady and the <OK> button in the calibration window is pressed. This returns the operator to the
Scanner Setup window. The calculated pulses per mm will be shown in the field beside the Calculate
button. Once calculated correctly this figure is unlikely to change for the encoder used (minor change
may occur due to wear on the encoder wheel).
Figure 9-4 TOFD Scanner entry menu (ao)
742
Once all parameters have been input the entire setup is usually saved to a "setup file". This ensures
that if anything is seen to be incorrect or inadequate that only minor adjustments are necessary to the
setup instead of having to start over and make entries in each data entry box.
When the operator is satisfied that a good scan will result based on the settings used, the actual "Data
Acquisition" canbe started by simply pressing a "Start" function button on the software. This puts the
computer into a mode that is geared to collecting the A-scans from the gated region. The scan can be
stopped manually or will automaticalty stop once the distance entered has been reached or the timed
interval (for Free-run mode) has elapsed.
During the data acquisition there is usually a display of the scan in progress. A single display format
is not possible for all applications. The operator will need to determine what views are best for the
application at hand. Generally, for a single channel TOFD scan the basic B-scan display is adequate.
However, some applications may have multiple TOFD channels (and perhaps pulse-echo as well). All
scans displaying on the monitor at the same time may not be suitable. Using multiple displays will
result in each of the displayed B-scans having its resolution reduced in order to fit the others on the
monitor. Since one of the concerns for TOFD (and any automated or semi-automated) scan is
consistency of coupling, it is probably best to select a channel that indicates the most pronounced
lateral wave. By monitoring the amplitude of the lateral wave the operator can assess the efficacy of
coupling (at least for that probe pair). If problems are seen during data acquisition the scan can be
aborted and a re-scan done.
In some cases, part of the acquisition setup involves preparation for storing the data collected. Data
sits in volatile memory until the software is instructed to save it. In some systems, there may be an
option to save data immediately upon completion of the scan. This would require that the operator
enter the path to the save location on the computer with a specific file name. When the save function is
automatically carried out after scanning, the file name and storage directory must be entered as part of
the Acquisition setup menu.
Some software will allocate space on the computer for the file. Theru even when a scan is aborted you
may find that the file size of an incomplete scan is the same as the file size of a completed scan.
143
Chapter (10): TOFD Data Analysis
CHAPTER (10): TOFD DATA ANALYSIS
The fact that there are many manufacturers of UT equipment each with its own version of software,
has been a bit of a problem for clients of ultrasonic inspection. It is simply not possible to review data
collected on one system using data analysis software from another manufacturer. To some extent there
has been an advance in standardising the file format for images in the medical world. L'rdustrial NDT
has had some benefit from work in the medical field. DICOM (Digital Imaging and Cornmunication in
Medicine) was a protocol derived in the medical profession around the turn of the 21" century. It is an
agreed upon format for image and data transfer used in healthcare for radiographic images.
In2004 the development of DICONDE (Digital Imaging and Communication in NDE) was conceived.
This is based on the DICOM protocol, but incorporates many structures unique to NDE. Essentially,
DICONDE is a dictionary that describes all the necessary syntar; attributes and data elements to allow
users to acquire, store, archive, transmit and receive image data in a way that is universally
compatible.iat)
Unfortunately, this protocol is presently limited to "images". Data underlying ultrasonic scanning is
composed of waveforms. At this time there is still no agreement on a data format that can be read by
all ultrasonic data analysis software. As a result data analysis of TOFD scans requires the user to have
the same software as was used to collect the scans.
Examples used in this section will be taken from several versions of software. The common aspect of
all TOFD software is that only two forms of the display are required; A-scans and B-scans (B-scans
may be differentiated as either B-scan or D-scan display by some users). Mary analysis tools are
common to all manufacturers. A few of the tools found in some software will be unique (these are
usually the digital signal processing tools). Even when the analysis tools are commory the way they
are used or displayed can be significantly different from one software to another.
10.1 FILE STRUCTURE

Of some importance to both the TOFD operator and the TOFD client, is the raw data collected on a
project. The client has paid for an inspection and in addition to the report on the findings; they can
also expect to be provided the information collected on their product. Since the data is stored as a
computer file, the report package will also then include the files.
A critical aspect of good inspection policies when using computerised systems is to ensure that all files
are identified in a useful manner and organised in a way that ensures that they can be quickly and
reliably identified for subsequent use or analysis. This implies that some thought should go into the
naming of files. Some guidelines for the file organisation should be followed.
Unless there is a written document where all the file names are recorded with a description for each
file, an operator will need to have some idea as to what the file is about. Without some sort of link to
the purpose of the configuration in the file name, the file could simply be a number that provides no
745
connection to its purpose. E.g., seeing a file named 723456.datmay indicate that the file is
a scan-data
file, but there is no clue as to what project and what component is contained in the file. If the
operator
can separate files into projects, then it is possible to organise the various file types within a common
main folder
A file name should indicate some connection to the contents. Generally equipment manufacfurers
provide unique file extensions to the file naming process. These are the last 3 or 4 letters after the
"dot" in the file name. E.g., file extensions might include (where n indicates any file name)
. *.cfg (configuration
of scan parameters)
. *.lyt (for the arrangement of views
for data display)
o *.dat (for the data collected
in a scan)
o *.prc (for data files that
have been processed for various enhancements)
o *.col (for the colour
palettes used for colour displays)
Most systems used in TOFD are based on some form of "Windows" software. This organises files
using a tree structure. The location of files in the structure is critical in some cases, and it is a matter
of
convenience in others. E.g., for the instrument internal sefups and self-checks, the manufacturer
will
have organised files to be under a specific folder or "directory".The user should have no access
to
these files.
But files made by the user can be located in user-made folders or manufacturer provided folders.
This
can allow the user to keep the relevant files, for all aspects of a project, together (something
that may
be required by a client wanting a final report package that includes the raw data and the
pertinent
settings for viewing of the data).
Example:
Project - ABC
Setup and Layout Files
ABC-10MHz60TOFD.cfg
ABC-Setupl.lyt
ABC-Analysisl.lyt
Data Files
Dayl
ABC-Part1 -Weld1 -0-2050
ABC-Part1-Weld2-0-2050
ABC-Part1-Weld2 1950-4000
-
In the above example the files are named with ABC starting each file name to identify the
project they
are associated with. In the files relating to a probe setup, some attempt is made
to identify the probl
(e'g', 10MHz60TOFD indicates it was a 10 MHz probe with 60'TOFD wedge).
Similarly the layout
146
Chapter (10): TOFD Data AnalysQ
files are connected to the project name and are provided with a name to help identify the use (e.g., one
layout of views for setup of UT parameters and another for the views used for analysis of results).
Of primary concern to the client is how the scanning results will be tracked. This will allow them to
see if all the items that were supposed to be inspected, were actually inspected. Typically there will be
a component identified and if it is a weld inspectiory there may be several welds in each component. If
the welds are large, it may be necessary to scan the weld in sections so the scan distances may also
appear in the file name. For welds scanned in sections, there should be some overlap between each
scan. In the example, we indicate that the scan increment for one file is from 0-2050mm and the next
adjacent one is 1950-4050mm. These are 2.05m scans, but the overlaps are indicated in the naming.
When a scan is complete and the operator determines that it is of acceptable quality, it is then usually
saved to a file. File naming should be unique and relate to some aspect of the project. File naming
protocol is not a trivial item. In a large project, where many different comPonents and welds are to be
inspected, there must be some means of identifying the data collected. This is further complicated by
the fact that some welds may be quite long and therefore scanned in sections. Overlap between
sections is essential to ensure no regions are missed.
10.2 TOFD DATA ASSESSMENT
The ultimate goal of a TOFD scan is to generate an image that can be used to assess component
integrity.
TOFD assessments can be broadly grouped into two categories:
1) Quality Assessment
2) Data Assessment
Quality assessment means the characteristics of the data collected. We must determine if the data
collected is useful for its intended interpretation. Data is only useful if it provides clear and
unambiguous information about the structure tested.
Data assessment can onty follow if the data collected is of sufficient quality, so that indications of
potential flaws can be reliably detected. Data assessment is also a multi-stage operation. B-scan images
(or D-scan images) are reviewed and if any indications are seery their origins need to be determined.
Not all indications are flaws and not all flaws are defects. Usually there is a set of expectations for
quality level (acceptance criteria) and the flaws need to be compared to the allowed limits to assess if
the component is acceptable or not.
10.2.1. QUALITY ASSESSMENT

TOFD scanning is subject to a variety of conditions that may not always be optimum. There are some
common problems that can occur when performing TOFD data acquisition. The results may or may
not render the data un-interpretable. This will need to be determined using the data saved to a file.
(Note: most software automatically saves the file upon completion of the scan, so its usefulness needs
to be assessed before being used as the final raw data). Such concern is similar to producing a
radiograph. If, for example, the radiograph is too dark, too light or has interfering scratches or
1.47
chemical marks due to processing, it will be necessary to re-shoot the film. For TOFD the deleterious
effects may require re-scanning the section.
The most common problems experienced in TOFD scan quality are:
o Data collection gate positioning

. Gain settings (too high or too low)
r PCS settings (too large or too small)
o Electrical issues (triggering and electrical noise)
These issues are described with examples

in the following figures; however, the
first image is one of a good quality scan
and many of the subsequent scans are of
the same weld to illustrate changes that
could cause a reduction in image quality.
In Figure 10-1 the lateral wave and back-
wall amplitudes are clean and of an
acceptable amplitude. Background noise
levels are not excessive and flaw
indications are clear with a good signal
to noise ratio compared to the
background. Figure 10-1Good TOFD sgsn (+2)
In Figure 10-2, the lateral wave and some of the diffracted signals are starting to disappear. At yet
lower gain levels, some of the diffracted signals would become undetectable.
Figure 10-2 Incorrect gain - gain too lsyT (az) Figure 10-3 Incorrect gain - gain too high tazl
Figure 10-3 illustrates another incorrect gain setting. In this case the gain is set too high. Noise level
increases to obscure the diffracted signals. This can lead to reduced probability of detectiory and poor
748
sizing. High noise levels can also arise from large grains. If this is the case, the solution is to reduce the
ultrasonic frequency. In Figure 10-3 the lateral wave is saturating and the ring masks more than the
calculated minimum.
Figure L0-4 Incorrect gate start- too earTy $z) Figure 10-5 Incorrect gate start- too late (42)
Figure 10-4 illustrates poor data gating. At a minimum, the gates should encompass the lateral wave
and longitudinal wave back-wall signal. Typically, data collections would start 0.5-1ps prior to the
lateral wave. The gated data can extend to just beyond the shear wave (mode-converted) back-wall if
required. This example shows the gate set too early, the lateral wave is visible and the back-wall is not.
Any flaws on the opposite surface near the back-wall will be missed.
The example in Figure 10-5 shows the data gate set too late. The lateral wave is not visible. The first
signal is the back-wall, and the second signal is the shear wave (mode-converted) back-wall. With this
setup, all the near surface flaws will be missed.
The example in Figure 10-6 illustrates

the gate set too long. Although this is
not technically incorrect the image
shows the diffracted back-wall shear-
shear wave signal. These S-S waves may
show additional and confirmatory
information. The diffracted shear waves
show the porosity more clearly than the
diffracted longitudinal waves and there
is a strong mode-converted signal that
occurs just before the shear wave, which
could cause interpretation problems.
Figure 10-5 Incorrect gate length- too long (az)
1,49
Figure L0-7 Incorrect PCS- too long (+z)

Figure 10-8 Incorrect PCS- lss shslf (+z)
Figure 10-7 illustrates incorrect transducer separation, (too far apart). This results in the back-wall
signal becoming distorted or excessively strong, the lateral wave becomes weaker and some of the
diffracted signal closer to the near surface will have their amplitudes drop.
Figure 10-8 indicates another incorrect transducer separatiory (too close together). This results in the
lateral waves becoming stronger, and the back-wall weaker. Near surface flaw signals improve at the
expense of those closer to the far side.
Figure 10-9 Electrical noise (42)

Figure 10-10 Missing dataa27
Electrical noise spikes are seen in Figure 10-9. These can seriously impair TOFD interpretation. Noise
can come from a number of sources such as surrounding electrical equipment or ultrasonic instrument
or probe issues. Unlike noise from coarse-grain scatter, which occurs only in the material (i.e. after the
lateral wave), electrical noise appears as isolated spikes and sometimes as an interference pattern,
depending on the noise source. Once the occurrence of the electrical noise increases beyond a certain
point interpretation becomes essentially impossible.
150
\zVhen data acquisition is occurring at a rate faster than the computer can handle there will be portions
of the A-scan data not held in memory. The method of data handling may be different from one
system to another. Figure 10-10 illustrates lines of uniform grey (zero amplitude level). In this system
the computer has not had time to process and allocate the A-scan to a memory location before the
scanner moved on to the next encoded 1mm sample interval. As a result, the entire A-scan has been
missed. Some systems are able to store the portions collected. They may be able to salvage a portion of
the A-scan before moving to the next encoded interval, but the dropped data remains the same in that
there is a grey line with zero amplitude. Some missing data may be tolerated. In some TOFD codes the
amount permitted missing is regulated (typically not more than 10% of the scan, with no two adjacent
sample intervals missing).
"**!d!&.aa€'q*F*=*
Figure L0-L1" Trigger threshold level problelns (az) Figure L0-12 Coupling gap too large (az)
Figure 10-11 may at first glance look like missing data. But the lines are offset not missing. Time-base
triggering sensitivity, that is set too higlu results in a stuttering of signals if the initial pulse signal has
small variations. This is evident by the time-base shifting to the next half cycle. The effect may initially
look like missing data but the data is still seen, just not aligned.
Excessive coupling feed or using probe wear-pins that protrude can result in lifting the wedge off the
test surface by more than just 10-20microns. This results in reflection within the couplant gap and it
forms a second lateral wave signal, as seen in Figure 10-12. This will be seen as an extra ringing under
the initial lateral wave signal and can be misinterpreted as a flaw or may mask other flaws. This effect
may also be seen when there are surface irregularities that cause the couplant to increase in slightly
hollowed areas where the probe edges are on a surface that is higher than where the pulse enters the
test piece.
151
10.3 TOFD DATA ANALYSIS

Once satisfied, that the data collected is good (i.e. contains none of the issues above), the operator can
save the file.
Having saved the file of the acquired TOFD scan to a specified location on the computer, the operator
can then open the file with the analysis software.
As with data Acquisition displays, data Analysis displays are also configured to meet the needs of the
application and operator's preferences.
There are only limited options for the data displays. Views will be composed of either B-scans or D-
scans with or without A-scans. Usually both are used in analysis. Operators may choose which goes
where on the monitor and the orientation (rotation) of each.
Examples of some TOFD displays are showrr in Figure 10-13. Although the A-scan on its own is not
useful, all TOFD analysis software display should have the ability to extract the specific A-scan from
the B-scan so as to improve the phase and time analysis features. Figure 10-13 indicates that all
manufacturers have a very similar selection of TOFD display options.
-:ffi
echnology Design Sonotron
l: : l.:' .: l!4f,. i: a.
US Ultratek ONDT
Figure 10-13 TOFD analysis displays
752
In addition to the basic A-scan and B-scan presentations, TOFD analysis software needs several
feafures to make it useful as a sizing tool. Some features are essential and some features might be
considered advanced enhancements. Those features considered advanced or simply enhancements are
not found on all software packages, so may be considered non-essential (but handy).
Essential features of the analysis software would include:
o Reference position cursor
. Measurement position cursor
. Reference time cursor
o Measurement time cursor
. Depth linearisation algorithms
Advanced features or enhancements would include such features as:
. Scaling or zoom (in and out)

o Lateral wave straightening
o Lateral wave removal (or backwall removal)
' HyPerbolic cursors

. Synthetic Aperture Focusing Technique (SAFT)
. Soft gain
o Palette adjustment
r Lateral wave amplitude equalisation
o Notation addition
o Geometry corrections (cylindrical, Tee, other)
Scaling of the image, lateral wave straightening, lateral wave removal and hyperbolic cursors are now
often considered essential enhancements by most users and found on most software.
10.3.1 ESSENTIAL TOFD ANALYSIS FEATURES

L0.3.1.1 CURSORS
TOFD B-scans are composed of the grey-scale amplitude images with time on one axis and distance
travelled on the other. Both axes contain information that allows us to determine position of flaws
within the volume of the test piece. Along the scan axis the position is given as a distance travelled
from an origin. Here lies the advantage (need) for encoded position. Without encoding of distance via
a calibrated encoder, we cannot use the analysis display to indicate the length of a flaw. We can
identify the start and end of a flaw on the display but unless the scan axis is calibrated in distance
units there is no length assessment possible.
When the scan is encoded, scan axis cursors are linked to the encoded position and a scan coordinate
is indicated at the cursor or somewhere on the display. It is usual to have a reference cursor and a
measurement cursor. TypicalLy, the reference cursor is also the location of the A-scan that is displayed
(when A-scan display is configured as part of the data analysis display).
153
Figure 10-14 illustrates the basic display with two

scan axis cursors. The reference cursor (left) is also
the location along the scan axis from which the A-
scan (right) was extracted. The secondary cursor
(dotted markers in the image) is used to measure
the distance to the primary or reference cursor.
The vertical scale in Figure 10-14 is the time axis

(the units illustrated are in millimetres). Distance
relates to depth and the A-scan extracted can be
used to precisely identify the phase peak of the
waveforms. To ensure that the measurements for
sound-travel are made at well-defined points on
the waveform (e.9. peak positive, peak negative or
the zero crossing points) a second set of cursors is
required on the time sca1e. These should be Figure 10-14 TOFD primary (reference) and
available to the A-scan as well. secondary (measurement) cursors (as)
Figure 10-14 also illustrates the A-scan image extracted by the reference cursor and displays it on the
right side of the image. The horizontal lines on the B-scan are seen extended onto the A-scan and
allow identification of the upper and lower tip signals.
The vertical lines on the A-scan have no relation to the vertical cursors on the B-scan. Instead, the
vertical lines on the A-scan indicate amplitude displacement. Some software has a feature that allows
the amplitude cursor to track the amplitude of the signal at the point in time that the time-cursor
crosses the waveform. This link between time and amplitude can be an effective aid to precisely locate
the time that the peak amplitude occurs.
10.3.1,2 LINEARISATION ALGORITHMS

The other essential requirement for analysis software is the linearisation of the data displayed. A flaw
seen at 17.4y,s is useless to the operator and the client. Only when the depth and height of a flaw can
be provided in real units of distance are they meaningful. The process of translating the time along the
A-scan into real units of distance is called linearisation.
We have previously noted that depth to an indication can be derived from the equation
_-.2
(;) x(t-2to)2-s2 (10.1)
This equation is effectively solving the Pythagorean equation for depth. Right angle triangles are
made by the three points; the exit point from the probe, the point above the flaw and the depth to the
flaw. The triangles' legs are formed by half the PCS (indicated as S in Figure 10-15), the hypotenuse
made by the sound path ct* where c is the acoustic velocity and t* the time of travel to the specified
depth and the depths are indicated by d.. This indicates that the depth is determined by taking the
square root of the hypotenuse squared minus half the PCS squared. Flaws need not be present and
154
since the PCS defines the base of the triangle, then it is possible to calculate a scale along the Y axis
(depth) based on the distance travelled for different sound paths, to the line at distance S on the X axis.
Figure 10-15 Non-linear relationship of depth with time
This process can be applied to either the time-base on the A-scans collected or the A-scans can be
unaltered and the cursor readouts corrected for depth.
When the cursors are linearised, for image
analysis, the image is unaltered.
Therefore, when an image has the lateral
wave reasonably well straightened and
the reference time (or depth) cursor is
placed on the lateral wave representing
zero depth, smaller depth changes for the
same, as the cursor is moved farther from
the lateral wave. This is illustrated in
Figure 1,0-L6 TOFD Cursor Linearisation
Figure 10-16.
There we see the depth scale on the left of the B-scan. For the 25mm thick plate the cursors (and depth-
corrected scale) indicate that approximately a quarter of the way between the lateral wave (0mm
depth) and back-wall signal (25mm depth) the actual depth would be half the thickness. i.e., the depth
scale is not linear with respect to the time scale.
The other option provides a linear scale and stretches the A-scans as required. Figure 70-77 is a B-scan
image using linearised data. The display shows the depth along the vertical scale for a 20mm thick
plate. The lateral wave positive peak is set to Omm and noticeably stretched. \Atrhile the back-wall
region still retains the approximate same spacing of black and white signals. The spacing along the
vertical scale is now linear with respect to depth, so that the uppermost edge of the image is at 0mm,
the lowermost signal from the back-wall is at 20mm and the 10mm depth is exactly midway between
those two points.
155
Figure 10-17 TOFD data linearisation
Both methods require input of reference values. In all cases the PCS will be entered. In some cases
there will be a reference signal that can be used. This may be the lateral wave, in which case its depth
is known and for a plate is considered 0mm. Under some conditions there will be no lateral wave and
it may be the back-wall that is seen. This too can be entered as a known depth.
Under some conditions ri/lÊt do yEu want rompute? doyou want
(e.9., multi-zone TOFD ,a i' Compute ?elocity and wedge delay
Compute 'relocity end wedq3e delay
where no lateral wave and
' Compute oedge deloy
no back-wall signals are Scan eris Sran aril

seen) the PCS, the Farallel to beam , Parallel lo beam
estimated acoustic velocity . FerpPndicuiEr lo beem 'i Perpendicul+ io beam
Lll-U pr mary value

and the wedge delay (total I
o DoD i-. mm . !.OOO . *,

time of the sound in the
Probe sEoarat on: 61 000 , .. Probe seperation: Eo.ooo -l-, -*
transmitting and receiving
Soundvelocityr : _ m/t bSl.lrl
wedges) will be required
Sound velocatli m/s
-:.:
Weoge delay : --t

to linearise the TOFD data !s Wedge delay: ps
or cursors. These steps are

usually reduced to a :{:.r.j-.."'tt ir lq
-4.:
mm
simple data entry window,

such as seen in Figure I
10-18. Once values are I UK Cencel
entered, pressing "OK" has

the computer calculate the :r:iBlenr;ri.,:tlilere!r::.ie hl !:.!!!y.
parameters to linearise the ja.i'e .r.:-'rFr': , ;' .t-.-,,.
data or cursors. -" rrt.;,.:r ::.= '. ,.
'r I .r..
I-lr--l i;n::r .r,:i!
Figure 10-18 TOFD Linearisation Data Entry Windows
156
10.3.2 ENHANCED TOFD ANALYSIS FEATURES

If all the TOFD indications in the field were as obvious as the illustrations given in the
encountered
text books there would be little need for further enhancements to TOFD. Unfortunately TOFD is
subject to many aspects that make the results less than ideal. Improvements to the analysis process
have been made using computer enhancements. A1l the items discussed are done after the data has
been collected. Raw A-scans collected are unaltered and al1 the enhancements can be redone without
corrupting the original data.
Most of the enhancements might be considered "post-processing". Data processing was touched upon
in Chapter (4): Section 4.3.6. In addition to those already covered, some enhance'ments that are
typically found in most TOFD software include palette adjustmenf image zoorr9 simple text labelling
and data merging.
10.3.2.1 PALETTE ADIUSTMENT

Palette adjustment can be one of two varieties.
Most common is to change the amplitude level to
the maximum that the greyscale goes to. This has
the advantage of enhancing the contrast between
small amplitude signals. Alternatively, the pallet
may be changed to other colours; however, this
has limitations in that some of the detail can be
lost if the palette is too colourful.
Examples of these two palette options are

provided in Figure 10-19. On the upper left is a
symmetric even adjustment of greyscale with the
+100% and -100% signals corresponding to the
maximum white and black respectively. Zero
voltage is indicated by the middle grey level. The
two greyscale examples, to the right of the
symmetrical example, show that the maximum
values to achieve white and black are reduced to
*60"/" and then t40%. In the next row is a
selection of alternative colour options for the
palette.
10.3.2,2 ZOOM (IN AND OUT)

Zooming or magnification of the TOFD image is useful when scans are long and the full detail cannot
be perceived from the B-scans. Scans two or three metres long are sometimes made along a weld. At
one sample per millimetre, this represents a total of 2000 to 3000 data points. These scans must then be
displayed on a computer monitor. Monitor resolution is measured in pixels and there are typically
7280xL024 or less pixels on a monitor. Even if the display could present the 3000mm as the full width
of the monitor (1280 pixels) it would require that over 2.3mm of scan be crammed into 1 pixel width.
157
Computer algorithms are used to decide which item is displayed since it is not possible to put 2.3mm
of scan into one pixel width. A similar concem would exist for high digitisation rates on long A-scans.
A 50mm thick plate may require a 15ps window. At 100MHz sampling, this would be 1500 samples
along the time axis. Even if all 1024 pixels in the vertical plane of the monitor could be used there
would be more than 1 time sample per pixel. For accurate sizing of length or depth the ability to
magnify the image is necessary
In Figure 10-20 the
details of the notches
are poorly seen in the
fuII view, but the
individual time and
scan steps can be seen
on the zoomed image.
It is good practice to
adjust the resolution to
one sample per
horizontal line and one
A-scan per vertical line
if that feature is
available. That way, no
data is compressed and
180-290mm full scale zoomed to225-245mm
flaws are less likely to
be missed. Figure L0-20 TOFD image zoom
1.0,3,2,3 ADDING TEXT TO IMAGES

When preparing a report, it is often *;:=:::RE*S€=
useful to add text to identify features of e-$ !:w

;E
the TOFD images. It is possible that the ** l{:i=::i*--:
=:€i:iEt=.8*+*s€
:]:*al;allll':i
::H.=':.=- as*+
text could be added in the word- P= :*.-- ;:r-:
:sl+:v,tE
processor being used for the report;
H;l;:::L:s=Ei::i*
cs:5:isKi.*s s**
#*{ia.:}.-ii+*:i*=*
however, it is sometimes more ffi*i',!i:::-irii:€l =*n-s=ls
Ba!.=-:+
convenient if the TOFD software is

{ffi.-::::<::::
i
::
t E*i,: . :=-_i
E#;.::sr=:+=
%:-:i€t:E+1
used to more precisely identify the d €*:i:€.::;:-:t:
I
features and add any text. A label N
=-:,=:l
combined with the cursor positions is

9+r.xEQ
an effective way of identifying a flaw r^-n
for a repair report (as in Figure 70-21).
Figure 10-21TOFD image labelling
158
1.0.3.2,4 DATAMERGING
Already mentioned was the need to zoom in on regions in a long scan, as a result of the loss of
resolution, when too much data is being presented on the monitor. Flowever, there can also be a
situation when a weld scan is divided into many short scan sections. Large strucfures, such as a girth
weld on a large diameter tower, may be scanned in several 2 metre scan sections, each with an overlap
to the next section. If the scans are configured such that the encoder is set to read the correct
circumferential position at the start of each short scan, some software allows merging of the data
based on encoded position. Data merging will take the encoded position in each scan segment and
arange the data sequentially based on the encoded position. For areas of scan overlap the software
would compare the amplitudes for each position and use the larger value. \rVhen 5 or 10 scan segments
are merged at the start of a weld analysis, the operator can then focus on analysis of the entire weld
using the zoom feature to limit the scan length being reviewed on the monitor to just 250-350mm. The
process creates a separate large file. If the data was not merged the operator would need to open each
file individually (time consuming in itself) and then deal with the assessment of indications that cross
from one file to the next.
10.3.3 DATA ASSESSMENT

Having corrected any issues with image quality, the next step after data acquisition is to assess the
TOFD images for potential flaw indications. This would involve preparing the data appropriately
using the analysis tools such as the cursors and linearisation features in the software.
In addition to actual "flaw" identificatioru TOFD is also useful in helping to identify the geometric
condition of mismatch (or high-low), which is not generally considered a "flaw" in ultrasonic testing.
Flaw characterisation with TOFD is not as detailed as pulse-echo methods. Earlier, it was noted that
flaw characterisation is one of the limitations of TOFD. This has reduced the characterisation process
for TOFD to a simple, but very useful scheme, grouping flaw indications into one of three types:
o Point-like
. Thread-like
o Planar
Point-like flaws have neither length nor height (typically these are isolated pores or very small
inclusions). Thread-like flaws have lengttr, but a vertical extent cannot be assessed from the TOFD
display. This means that the lower tip signal cannot be discerned. Planar flaws have both vertical
extent and length (meaning a lower tip signal is discernible). In spite of these limitations, knowledge
of the weld layout can still be used to indicate possible characteristics of the known flaws that might
occur in the welding process being tested. (E.9" a length of indication occurring at the mid-wall of a
double V weld, with a fairly constant depttr, might reasonably be speculated as originating from
incomplete cross penetration).
Many of the following flaw examples are from a book on phased array technology (Introduction to
Phased Array Ultrasonic Technology Applications by published by RD Tech(44). A similar set of images
1,59
has appeared earlier in B3-7706 and more recently in the ASME Boiler and Pressure Vessel Code
Section V.
Images in this section provide examples of TOFD, along with the interpretation/explanation. Caution
is advised here, in that there can be significant variations amongst flaws and TOFD setups, and
displays. As a result the following images should be used as a guide only.
10.3.4 FLAW IMAGES

10,3,4,1 PO/NT DIF FRACTORS
Point flaws, such as porosity, appear as single multi-cycle
points between the lateral and back-wall signals. Point flaws
typically display a single pulse-cycle signal, since this type
of flaw has a height smaller than the ring-down of the pulse.
Point flaws usually show parabolic "tails" where the signal
drops off towards the back-wall with no associated "length"
between the two tails. This is illustrated in Figure 10-22
where the hyperbolic cursor has been fit to the arc shape of
the indication. It can be seen that there is a reasonably good
fit of the calculated curvature to the shape of the indication.
When the peak of the curve matches the peak of the
indication, the tails fit both sides of the indication (i.e. the
second cursor is not required as there is no length and no
lower tip is discernible).
Figure 10-22 P oint diffractor (pore)
1.0.3.4,2 I}/S/DE (ID) FAR SURFACE-BREAKING FLAWS

Figure 10-24 shows no interruption of the lateral wave, a signal near the back-wall and a related
interruption or break of the back-wall (depending on flaw size). The shape and intensity of the
indication will depend on the flaw type. Illustrated is a concave root and incomplete penetration.
Surface breaking flaws do not always obscure the back-wall and may be associated with "tails"
extending past the back-wall signal (seen in the incomplete fusion example, in Figure 10-24;but not in
the concave root example). Phase reversal is expected on the upper tip, as it is simply an earlier
introduction of the back-wall condition. No lower tip signal will be present.
160
Figure 1:}-Z3Far side flaw - concave root or burn-through (+a)
@ililil[-
Figure l0-24Far side flaws - incomplete penetration single V weld (44)
10.3.4.3 OUTSIDE (OD) NEAR- SURFACE BREAKING FLAWS

The effect of a near-surface breaking flaw is to show a perturbation in the lateral wave (see Figure
10-25). The flaw interrupts the lateral wave, allowing TOFD to be used in determining if the flaw is
surface breaking or not. Since only a lower tip signal is present, that signal is used to measure the
depth of the flaw. However, if the flaw is not surface breaking, i.e. just subsurface, the lateral wave
will not be seen to shift to a position later in time. If the flaw is near-subsurface and shallow (that is,
less than the ringing time of the lateral wave, or a few millimetres deep), then the flaw will likely be
invisible to the unprocessed TOFD image and lateral wave removal will be required to expose the
indication (see Figure 70-27).
761
Figure L0-25 Near side flaws (aa)
It is important that TOFD data files are reviewed in the raw, "un-straightened" format prior to lateral
wave straightening being performed. If the file shown in Figure 10-25 is straightened along the lateral
wave, including the area of the near-surface breaking flaw, the resultant image will show the flaw at
the back-wall, since all of the data below the flaw would be lifted up by the amount that the flaw is
actually below the lateral wave position along the time base.
10,3.4.4 MID_WALL FLAWS
Mid-wall flaws show complete lateral and back-wall signals, plus diffraction signals. For a planar flaw
the presence of both the upper and lower tip diffracted signals on the B-scan image will depend on the
depth and vertical extent of the flaw and the PCS and ring-time of the TOFD setup. Flaw tip echoes (if
observable) can provide a good profile of the actual flaw (Figure 70-26).
Figure L0-26 Incomplete penetration in double V (44)
If a mid-wall flaw is shallow, i.e. less than the transducer pulse ring-time, the top and bottom tip
signals cannot be separated. This is considered a thread-like flaw. Under these circumstances, it is not
possible to differentiate the top from the bottom of the flaw, so the evaluator can only say that the flaw
is less than the ring-time depth equivalence (which deperils on transducer frequency and damping,
etc.).
762
M[n, Figure 10-27 Lack of side wall fusion (aa)
Speculation as to the origin of the flaw requires further assessment. Figure 10-26 illustrates incomplete
penetration in a double V weld. To assess the flaw, as such, requires knowledge of the weld bevel
preparation and weld process. Further assessment can be made using a parallel scan. If the depth
corresponds to the root position and the parallel scan indicates that the flaw is located on the
centreline, there is good evidence to call this incomplete penetration.
If the flaw is side wall lack of fusiory the depth probably will not coincide with the root and the
parallel scan will place the shortest time travel, with the probes centred at some point off the weld
centreline.
Lack of side wall fusion may often occur near the test surface. This can result in the upper tip
occurring in the dead zone, i.e. partly buried in the lateral wave. This may increase the amplitude of
the lateral wave and serve as a clue to recognising the presence of the upper tip, but it will not be
possible to identify the upper tip without lateral wave subtraction (Figure 10-27). Some equipment
allows for the removal of the lateral wave, revealing flaws partially hidden within the lateral wave,
although this feature must be used with care.
10.3.4,5 ?RANSYERSE FLAWS
Transverse flaws, when detected using a nonparallel scan, are similar to a point diffractor. The TOFD
scan displays a typical hyperbolic arc. Normally, it is not possible to differentiate transverse flaws
from near surface pores using TOFD. Further inspection using other techniques is recommended (e.g.,
pulse-echo UT or MPI or LPI) to better establish the nature of the flaw causing the indication. Figure
10-28 indicates a signal seen from a parallel scan past a transverse crack.
763
t-
Figure 10-28 Transverse flaw (aa)
Some techniques use a supplemental scan with the TOFD probe pair skewed 45" to the weld axis. This
is intended to increase the diffracting area and thereby increase the signal amplitude to enhance the
probability of detection. Ability to use this option depends on the weld cap size.
10,3,4,6 /NTERPASS NON-FUS/ON
The only difference (from a UT perspective) between interpass (inter-run) non-fusion and lack of side-
wall fusion is the orientation of the unfused metal. Interpass non-fusion presents a large area for the
beam to interact with. Thus, the beam is reflected rather than diffracted, resulting in a very large
amplitude signal. A macro example of the conditior; along with the principles and signals are shown
in Figure 10-29.
--F*ii-
--*
Figure 10-29 Lack of interpass fusion (a)
764
10.3.4.7 POROSITY
Porosity appears as a series of hyperbolic curves of varying amplitudes, depths and lateral positions,
as illustrated in Figure 10-30. B-scan images of cluster porosity superimpose the hyperbolic curves,
due to the individual pores being closely spaced. Since the scattering nature of pores can provide
multiple paths, after interactions with several adjacent pores, there is no way to use the depth of the
lower arcs to estimate the lower bound of the pore cluster. TOFD can be a useful method to detect the
porosity, but other methods should be used to improve sizing.
Figure 10-30 Porosity t++)
10.3.4.8 CRACKING
Cracking has a variety of causes and shapes.
Therefore, no single description is appropriate to
describe the TOFD signals associated with cracks.
The B-scan image in Figure 10-31 illustrates the
signal that results from copper pickup in a gas metal
arc weld (GMAW) process. The copper introduced
into the melt, embrittles the metal; and a dendritic
crack forms as the molten metal solidifies. As
subsequent weld passes are made, the re-melting of
the lower metal puts the copper back into solution
and draws it up to higher and higher levels. Figure L0-31 Copper Cracking
The indication is generally short (the illustrated crack is only about 20-30mm long), but extends from
the region near the root up to the penultimate weld pass closest to the test surface. Due to its dendritic
nature, this flaw is poorly (if ever) detected by pulse-echo UT. Cracks in tension tend to ensure the
facets are open to air. However, under compression, the facets may be pressed hard against one
another and the acoustic impedance at the boundary will be closely matched. This can make some
cracks "transparent".
165
10.3,4,9 MISMATCH (HIGH-LOW)

Mismatch is not a welding flaw. It is a geometric condition whereby one side of a butt weld is higher
or lower than the other. When the joined plates are not of equal thickness, but different by only a
small amount, there may not be a need for a taper and the sides are mismatched. The more common
condition occurs when equal thicknesses are being joined. Typically in two sections of pipe, the pipe
may not be exactly rounded, so the diameters do not match all the way around the circumfererlce.
Although not a flaw condition, the fact that the joint surfaces are not at the same elevation is of
concerry because it can be the source of real welding flaws. \zVhen the welder (manual or automatic)
tries to adjust the arc to compensate for the unequal spacing between the two bevel faces it can result
in a failure to fuse in the root area. A common defect that can occur as a result of mismatch or high-
low in a single V weld is incomplete root penetration. This is problematic for TOFD in that the most
likely place the non-fusion occurs in the root is on the low side of the condition. In the TOFD
configuration the low side is lost in the back-wall ringing and so the flaw may go undetected.
Figure 10-32 illustrates the
conditions leading to the
TOFD signals and Figure
10-33 illustrates the typical
"splitting" that is seen on the
back-wall signal where high-
low occurs. The image in
Figure 70-32 uses a 2mm
difference between the low
side (right) and high side
(left).
Figure 10-32 High-Low soundpaths
If we imagine the receiver on the right side, the soundpath from the transmitter to the receiver is
nominally defined by the locus of equal time for the depth equal to the wall thickness. When the high-
low occurs, the total soundpath is different for the beam components that define the locus of equal
time. Both the high and the low side back-wall signals are detected, but the high side now occurs
earlier than the arrival for the condition when both were at the same height; and the low side occurs
later than when both sides were at the same height.
If the elevation difference results in adequate separation of the back-wall signals, the deviation can be
measured. Figure 10-33 is a mismatch region with the lateral wave straightened. If the negative phase
(black) of the lateral wave is used as the reference, the back-wall in the "level" condition is seen to be
approximately 32mm in the area around 1000mm scan distance (on the left). At approximately
1550mm scan distance the earlier (high side) indication of the back-wall arrives at approximately
31mm and the later arrival (low side) is at approximately 34mm. We, thus estimate the condition to be
a difference of 3mm.
166
Figure 10-33 High-Low TOFD B-scan
This is not complete coverage of all possible defects and conditions that could be detected and
assessed with TOFD. As noted, the sort of characterisation of flaws typically allowed by manual
pulse-echo (or radiography) is not possible when using TOFD, so the methodology of classification as
noted in B,57706 (poinf thread-like and planar) is recommended.
10.3.5 DOUBLE IMAGING AND SHEAR MODE

Caution must be taken to avoid overcalling indications that might be a result of mode conversion to
shear in the interval between the lateral wave and back-wall. Although this region is primarily
associated with just the compression mode there are conditions that can result in a doubled indication.
The opposite mode conversion is possible too, i.e. shear to compressiory and the same result may
occur.
We can consider these mode conversion effects with a closer

review of the photoelastic image of the TOFD pulse. Figure 10 34
illustrates a close-up of the diffraction patterns that occur wher
the compression mode and shear head wave passes a flaw
(notch).
The compression mode (labelled arc #1) is generating the shear
head wave (labelled line #2) at the part surface. The
compression wave velocity ensures that it is the first arrival at
1. lniti.rl compression
the notch and forms the upper and lower diffracted 2. Shear head n'rve
compression waves (arcs #3 and #4). But this diffraction of the 3. Upper tipr conlprl'g55i6n
compression mode also generates a mode-converted shear (arc -1. Lon'er tip compressiorr
#5). Then, as the shear head wave hits the notch, it too diffracts 5. \lotle-concerted slrear. otf upper tip
(.. \ lotle-eottverterl rom Ppg-.ic|11.
a shear mode but also mode-converts to form a compression
off shear he.tr1 u'.:r'e
wave (arc #6). Since it is nearly double the velocity, the mode-
converted compression wave (arc #6) will arrive at the receiver Figure 1G34 Mode-conversion at flaws
before the mode-converted shear wave (arc #5).
When the flaw is close to the receiver, the travel time for the mode-converted compression off the
shear head wave can arrive prior to the compression reflected back-wall signal.
167
Depending on the depth and the relative signal strengths, it may be possible to see both the direct
compression mode diffracted signals (arc #3 and 4) and the diffracted mode-converted shear signals.
This is common enough for, signals seen after the back-wall, verifying the presence of a flaw seen
(sometimes poorly) from the longitudinal diffractions.
Conditions for both signals to arrive prior to the back-wall must satisfy certain timing requirements.
These are indicated in Figure 10-35.
Direct ramp.ession mcde
M*de*Conv€rted €omprn$Eitt!
mode korR shsar h€ad $rave
Siear head $ave Ssm lateral
{r$rnpt*ssio.} wave
Sack-wall e*mpre$sion mode
-
T# indicates iime of the ray paltr
io. the indicaied mode
B-:ca!l t tll imalje tha irr'oCc-conveflea iias sillna1 p.ro. !o lis 53i{ ,r3ii sle ..ri i: t5}Ti>T3+ I.l
Figure 10-35 Mode-converted arrival times
Similarly, the mode-converted shear from the direct incident compression wave may also provide a
signal that can arrive prior to the back-wall.
When the full time range is

displayed, the indication can be Flaw from direct ompression mode
seen multiple times. This is

indicated in Figure 10-36, for a Iiaw frem shear head lvave mode-
flaw 5mm below the surface with a onverted to mmpression wave
vertical extent of Smm. The upper

tip indication is seen again at
22mm depth and the multiples are Flaw from shear head wave
seen in the mode-converted region diffracted as :hear mode and frour
direct eeprssion modeconverted
after the back-wall. to shea tcave
Figure 10-36 Double image of the same indication
1.68
10.3.6 FLA\{, VERII]ICATION I'ROCEDURES

Details of the TOFD analysis process usually require that multiple quantities and judgements be
derived and reported. These would typically include:
. Defect depth (improved using Offset scans and locus curves)
. Defect length
o Defect height
o Proximity to nearest surface (ligament)
o Practical defect characterisation (point, thread, planar...others)
. Assessment with respect to acceptance criteria
Report generation and inspection details after the test and analysis would normally be compiled using
this information.
Comments and examples of the process complete this section.
Defect depth, length, height and ligament are best determined using the software analysis tools and
by having some knowledge about the welding process (if it is a TOFD inspection of a weld).
As an example of the process, a 25mm double V weld was scanned with a 60mm PCS and a SMHz 60"
TOFD probe pair. A flaw was noted at about 85mm along the scan path. The lateral wave was
straightened and the cursors linearised. The resulting image is shown in Figure 10-37.
Figure 10-37 TOFD Analysis of flaw in 25mm double V weld
Position along the scan and length of the flaw are determined using the vertical cursors. The
hyperbolic arcs can be added to the cursors to identify the ends by the curvature shape. This is
illustrated in Figure 10-38 where the B-scan image around the indication of interest has been magnified
with the zoom feature, the lateral wave removed and 6dB of soft gain added to highlight the flaw
features.
169
It should be noted that in the lateral wave

removal process, the fainter lateral wave in the
original scary just to the right and left of the
indicatiory was not totally removed, due to an
overcompensation of the removal signal. This
results when the amplitude of the reference A-
scan lateral wave, being used for the removal of
the other lateral wave signals, is larger than the
signal being removed.
With the flaw now more visible, we locate the

arc curvatures to fit the diffraction patterns at
the ends of the indication. From the cursor
readout we assess the flaw as starting at 63mm
and ending at 98mm. This provides us a length Figure 10-38 TOFD analysis of flaw
of 35mm.
By placing the reference depth cursor at the upper tip and the measure cursor at the lower tip, an
estimate of the flaw height can be made. Since the lateral wave phase, used to calibrate the cursor
linearisation, utilised the black portion of the wave, the upper tip is identified by the opposite phase at
4.2rnrn and the lower tip is located by the lowest point on the indication where the phase of the
suspected tip signal is black.
The peak positions of the waveform

can be derived using the A-scan. The
peak negative signal is seen to occur
at 13.8mm depth. This provides the
flaw height as 9.6mm and the nearest
surface is the test surface, so the
ligament is 4.2mm. Details of the
analysis can be placed on the image
and stored as part of the file after
processing. The sizing and
characterisaLion notation are indicated
in Figure 10-39.
Figure t0-39 TOFD Analysis for flaw height and ligament
The noise level presented by the grain structure in the scary imaged in Figure 10-37, is not particularly
poor. If the noise level is higher, the upper or lower tip indications and the hyperbolic arc formations
may not be as easy to discern. A greater error band, for all aspects of TOFD sizing, should be expected
for coarse-grained materials.
770
True depth of the flaw may not be correct if the flaw is not at the midpoint of the PCS. Ideally, in
order to determine the true depth of the flaw, a parallel scan is made at the flaw position. A practical
problem usually exists when attempting to carry out the parallel scansi the weld cap interferes with
the probe movement over the flaw. More often than not, the client does not want to add the extra cost
of weld cap removal.
A practical solution to the extra parallel scan is to carry out a tandem scan using two transmitters or
two receivers on one side of the weld. The purpose of this is to take advantage of the locus of equal
time. Such a setup is illustrated in Figure 10-40.
In Figure 70-40, a single probe on the left side of the weld and two probes on the right bide of the weld
can be seen. The operator then collects 2 separate B-scans. These can be identified by some convention
(e.g., inner and outer, or by the PCS values). When a flaw is detected, the operator sees the same
indication at the same scan position. Upon depth assessmen! two depths are obtained. By using the
locus of equal time crossing points the true depttr, as well as the lateral offset of the flaw, can be
determined. In the illustration in Figure 10-40, one probe pair is shor.tm as indicating the flaw was
measured at 25.5mm depth. The other probe pair indicates the flaw depth as 23mm. When the loci of
equal time are used, the crossing point indicates the true depth as 22.78rnrn The plot also suggests
that the flaw is about 7mm from the weld bevel. This might suggest that the indication is associated
with a weld repair or that the flaw is in the heat affected zone.
depth locus
{7lzs.s"',rn
/
\-----__
,--_ --_ 1-:"d123mm depth locus
Ii!:iurc ilenerJted uring trSBearu l ool
Figure 10-40 Tandem TOFD for true depth assessment
10.,1 AUTOMATION OF THE TOFD PROCESS

Because it is computerised and generally position encoded, many researchers have examined the
options to automate the various aspects of TOFD.
Automation of TOFD is an attempt to eliminate the subjective nature of humans and make each of the
steps in the process of TOFD inspection consistently repeatable. In effect, automation of TOFD is an
effort to eliminate the operator. In a motorised TOFD data acquisition system the acquisition operator
is essentially eliminated. Although there may still be a requirement for a scanner operator to place the
771
probes correctly at the scan datum position, once the couplant flow has been checked to be adequate,
the motorised motion is usually slow enough that the coupling and acquisition throughput of the
computer ensures relatively uniform signal amplitude and no missing data points.
Once the data is acquired, computerisation can be applied for initial processing. Zahran( '\ describes a
process of sequences that involves computer algorithms to suppress noise, correct for A-scan drift
from the zero-voltage, A-scan alignment (lateral wave straightening) and computer estimations of the
lateral wave and back-wall signal locations. This pre-processing provides a relatively uniform
background (zero-voltage level), with the image coordinates identified, so depth assessment can be
made.
After pre-processing, another set of algorithms is used to automatically identify flaw features. This
requires background removal using statistical in{ormation about the image. The end-result is similar
to lateral wave and back-wall removal; however, the process also removes some of the noise patterns
present in the rest of the image. Once the surrounding background is made "cleant", a form of
thresholding is applied to the image intensity to highlight potential flaws. Highlighting areas of
interest can then be carried out (automatically). The highlighting of flaws is termed segmentation.
Having highlighted potential flaws, the initial processing step of identifying the lateral wave and
back-wall allows for the image depth and vertical extent to be computed. Since the data is position
encoded, the length can also be calculated.
The final hurdle to the automation process is classification of the flaw type. The BS Standard 7706 tses
five categories of classification of TOFD flaw indications:
1) Planar flaws. Flawsin this category include cracks, lack of fusion, etc. Ultrasonic techniques are
not expected to distinguish perfectly between the flaws in this class.
2) Volumetric flaws. Internal volumetric flaws with depth comprise a general category which
includes volumetric flaws such as lack of penetratiory larger slag lines, etc.
3) Thread-like flaws. Flaws in this category include flaws with significant length but little through-
wall extent, typically <3 mm. In a TOFD search the same type of echo may be observed from
long, narrow lamellar flaws and from near horizontal areas of lack of fusion.
4) Point flaws. Flaws in this category include pores, small pieces of slag, etc. They are usually the
most common feature and are not usually reportable.
5) Uncategorised flaws. It may be found that some flaws are not neatly categorised by the
examination. These may, for instance, be due to cracks with jagged profiles, internal contact
points or complex form. As far as the categorisation is concerned the reflectors are likely to be of
two types:
i. Reflectors which are not interpretable by eye
ii. Reflectors which are provisionally entered in a more serious category, such as planar for
volumetric, continuous for broken, etc.
The automatic classification of defects is perhaps the most challenging aspect of TOFD automation.
Techniques for this process are constantly being evaluated. One of the techniques used is called
Artificial Neural Networks. ANN technology is, effectively, an attempt to use features of a situation in
order to teach a computer to recognise a pattern or situation. That is, it is a method of trying to get the
772
: TOFD Data Anal
computer to use the same evaluation process that a human would use. Flowever, the advantage of the
computer doing the analysis is that it does it consistently and quickly.
An example of the end-result of an automatic TOFD analysis is seen in Figure 10-41. This illustrates
the automatic detection and segmentation of flaws in an early version of software by Lawson.
Figure L0-41 Automatic flaw identification of TOFD flaw indicatisls

(46)
Curve-fitting and pattern recognition algorithms to identify the arcs associated with flaws in TOFD
scans have been the topic of research for several years. Neural network and co-occurrence matrix
work has been carried out by several groups including Corneloup and Moysan in 1988 ar.d 1992(47)(48).
A system based on an algorithm locating indications in the TOFD data, that fit the parabolic curve, has
also been investigated.(+o)
As of ye! none of the commercial systems sold are equipped with automatic detection and
characterisation algorithms. Therefore operators of TOFD systems are still required to make the
complicated judgements involved in TOFD analyses.
773
Chapter (11): Digital Signal Processing
CHAPTER (1"1): DIGITAL SIGNAL PROCESSING
Digital Signal Processing (DSP) uses techniques developed to enhance pertinent information and to
suppress the detracting effects of noise. Digital signal processing can generally be grouped into three
categories: one dimensional, two dimensional and three dimensional. One dimensional processing is
applied to the captured waveform and can be either filtering or spectrum analysis. Two dimensional
processing is concemed with enhancing spatial structures of the image. \AIhen data is collected in a
raster scan patterry such that separate beams (A-scan) pass through each unit of volume of the tested
componen! a three dimensional tomographic image of the component can be reconstructed. The
volume units in a 3D image are called voxels. DSP processes, similar to the two dimensional concepts,
can be applied to three dimensions as well.
Some engineers group the processes into 4 categories based on domains. They define time, space,
frequency and wavelets as the domains for digital signal processing. For our purposes we will
simplify the process and group wavelet and frequency domains in our one dimensional category of
DSP.
Since this section is merely an overview of the options, we will consider the simple division of DSP
into either one dimensional or two dimensional.
One dimensional DSP deals with individual waveform signals. Examples of one dimensional DSP
include:
. Amplitude analysis
o Fast Fourier transforms
. Signal averaging
r Filtering
Two dimensional DSP concerns itself with the spatial relationship of samples in a single plane.
Examples of two dimensional DSP include:
r Image spatial averaging
e Image smoothing
o Contrast adjustments (e.9., photo images)
A general grouping of DSP options is indicated in Figure 11-1.
Digital Signal Processing (DSP)
Figure 11-l Digital signal processing options
175
r (11):
11j, ONE DIMENSIONAL SIGNAL PROCESSING

11.1.1 SOFT GAIN
One of the simplest processes that can be applied to a waveform signal is "soft gain". This simply
applies an amplification multiplier to every point along the A-scan. In Figure 71,-2 a TOFD scan with
an A-scan, selected from the scan that contains a lateral wave signal of about 50"/, at.bfer"n""
sensitivity, is illustrated. The slider-bar, on the left of the image, indicates the "soft gain" added in the
lower image of the same scan. With 6dB added in software gain (also referred to as soft gain) the
signal is seen to saturate (exceed 100%) in the lower image. Similarly, signals cair be reduced in
amplitude using a negative software gain. However, any signal that is 100% at reference will not be
reduced by soft gain (i.e. removing 6dB soft gain from a saturating signal would still have the signal at
100%).
Gain Information
*l=] loo
l-l
*J=l o
ffiH
Soft sain[dE]: ! g
,,,
Hard gain [dB]: 0.0
Global gain [rJB]: 8.0
Figure 11-2 Software gain added to TOFD signals
11.1.2 FAST FOURIER TRANSFORMS

A so-called "F-scan" can be presented by narrowing a gated region to a signal of interest and
analysing for its frequency content. This process is shown in Figure 11-3. The green parallel lines, in
the upper left portion of the image, indicate the gated region where the signal is selected for
processing. The gated region is seen to be limited to just the lateral wave. To the upper right is the B-
scan image of the modelled scan, while the lower right indicates the probe setup and the artificial flaw
placement in the scan model. The F-scan display in the lower left of the image indicates the frequency
content of the signal in the gated region. Here we see that the centre frequency is about 4.5MHz. This
is an interesting feature because the nominal frequency of the probe used in the model is 10MHz. The
reason for the significant drop in frequency is due to the divergence effect at the probe. The lateral
wave is made by the off-axis portion of the pulse. From equation (1.22) for the beam spread, we can
see how lower frequencies will have more beam spread and higher frequencies will have lower beam
spread. Therefore, in the band of frequencies produced in a TOFD probe pulse, the lower frequencies
will dominate in the lateral wave and the higher frequencies will be stronger along the beam axis. This
176
is seen when we do an F-scan on the lateral wave and back-wall signals, at the same point as
illustrated in Figure 11-4.
;=t- ,
..'.
- .:*
Figure 11-3 Fast Fourier transform (F-Scan)
Figure LL-4 Comparing Fast Fourier transform (F-Scan) on lateral wave and back-wall
71.1,,3 SIGNAL FILTERING

A transducer produces a band of frequencies. However, due to the material properties of the part
being inspected, some of these frequencies may be more attenuated than the fundamental frequency
and will primarily produce noise. \zVhen noise is known to originate at a higher or lower frequency
than the pertinent UT signal, a bandpass filtering process can be applied. This selectively removes
spurious components from the A-scan. Figure 11-5 illustrates such a process.
777
Figure 11-5 Bandpass filtering by DSP
The weighting of the process in bandpass filtering can be affected by adjusting the shape between the
limiting frequencies. In Figure 1'L-5,2 of several forms of bandpass filters are illustrated; rectangular
and Hanning. The original signal is seen on the upper A-scan trace and a gate is placed on either side
of the signal of interest (in this case the back-wall signal). This defines the frequency of the signal of
interest via the F-scan process. The filter limits and shape are selected based on the frequency content.
The selected filter parameters are then applied to the entire A-scan. The resultant output signal of the
processed data is seen on the lower A-scan in each case. Even though the same frequency window is
used for both the rectangular and Hanning windows, i.e. 4-72MHz, it can be seen that the Hanning
window provides a weaker signal. In the filter representations, shown by the red lines inFigure 4-12,
only the frequencies under the red line are passed for processing. Filtering does not always improve
results. In Figure 4-12 the input signal indicates a flaw was detected just prior to the back-wall signal.
After bandpass filtering the signal is not as well defined.
1,1,,2 TWO DIMENSIONAL FILTERING PROCESSES

Two dimensional Digital Signal Processing (DSP) techniques are used to enhance spatial information.
As such, two dimensional DSP is applied to B-scan and C-scan images. Note that B-scans and C-scans
contain no more information than the A-scans used to generate them. However, they provide spatial
relationships, which are not possible to determine from isolated A-scans.
In TOFD several two dimensional DPS options are useful. These include:
o Lateral wave (or back-wall) straightening
r Lateral wave equalisation
o Lateral wave subtraction
. SAFT
. Hysteresis correction (also called backlash compensation)
178
71.2.1. LATERALWAVE (BACK-WALL)STRAIGHTENING

Mechanical movement of the TOFD probe pair results in slight variations in the arrival time of the
lateral wave. This can be as a result of drag effects from the scanner or even weld cap interference
pushing the probes apart. The result is an irregular arrival time of the lateral wave as seen in Figure
11-6. This is clearly not a result of flaws, as it can be seen that the back-wall is moving with the same
pattern of change in time of arrival. hr order to make sensible depth measurements the operator must
establish the zero depth at the lateral wave, but if it arrives at a slightly different time at every point
along the scan then separate corrections must be made at each scan point. To avoid this problem,
software is used to align the peaks of the lateral wave. This feature is called lateral wave straightening.
Once straightening has been achieved the lateral wave is much more uniform in its arrival time and
can be used as a reference for depth sizing over the entire straightened distance. The same principles
are used to straighten the image using the back-wall as a reference. This may be preferable to the
lateral wave, when there is suspicion that a surface breaking flaw exists on the near surface.
fnsEff€{ffit
oo r@o so
!6"S
!?jl{D
,?,rbo
Figure 1,L-6 Lateral wave straightening (50)
1L.2.2 LATERAL WAVE (BACK-WALL) REMOVAL

Throughout the literature on TOFD, the problems of dead zones are explained. These are a result of
the pulse ring-time interfering with signals that could occur in the same time frame. This problem is
primarily seen at the lateral wave, but can also be experienced on the back-wall. Straightening the
lateral wave (or back-wall) and using it as the reference signal aids in depth measurements. However,
there is still a signal present near the surface that prevents the identification of flaws that occur within
the ring-time. By identifying the shape of the dominant reference signal, the reference signal can be
mathematically removed. This processing is usually a vector addition of the inverse of the wave
shape. When the processing is completed the lateral (or back-wall) signal is "nulled" and any signal
shapes that were not the same as the lateral or back-wall can be exposed. This allows detection of the
flaws in regions often considered to be dead zones. Figure 11-7 illustrates the two step procedure. On
the left the raw data is prepared by locating the region of concem (between the blue cursors) and the
reference waveform is selected between the green and red cursors. After subtractiory the image on the
179
Chapter (1 I ): Digital Signal Processing
right is shown with the upper tip of the flaw, that is not seen in the raw data, now clearly defined and
available for sizing. (A small section is similarly treated for back-wall signal removal and is also
indicated on the right side image in Figure 11-7).
Figure 1l-7 Lateral wave removal (50)
11.2.3 LATERAL WAVE EOUALISATION

An assumption made when carrying out lateral wave subtraction is that the entire length of interest
has the same shape of response from the lateral wave where no flaws exist. This assumption also
applies to the amplitude. If there are variations in lateral wave amplitude along the region of concern
due to poor coupling, then the subtraction of the reference signal will overcompensate and leave a
remnant subtraction signal. Some software allows for an extra step between lateral wave straightening
and lateral wave subtraction; i.e. lateral wave equalisation. This process compares the amplifude of
the lateral wave from the reference signal with the lateral waves in all of the subsequent A-scans being
processed. The software gain required to equalise the other lateral wave amplitudes to the reference is
calculated and the entire A-scan at each point is re-drawry applying the software gain correction that
brings the lateral wave equal to the amplitude of the reference lateral wave.
Figure 11-8 illustrates the effect of lateral wave equalisation. The lateral wave amplitude in the raw
data frame is seen to be weaker between about Omm to about 50mm and again between about 150mm
and 200mm. On the equalised image a uniform lateral wave ceu:r be seen.
180
Figure 11-8 Lateral wave equalisatiory Weaker signals (right) made uniform amplitude (left) rsol
1,1.2.4 SYNTHETIC APERTURE FOCUSSING TECHNIQUE (SAFT)

Another type of B-scan DSP enhancement used in some TOFD systems is the method called, synthetic
aperture focussing technique (SAFT). Transit-time for the ultrasonic beam to travel to and from a
point is a hyperbolic function of the probe position and target depth. When the equation of this
hyperbola is known, A-scan signals can be shifted in time and added together. \Alhen a defect is
presen! constructive interference of the waveforms will form a large signal. When no defect is
presen! the interference is destructive and the signal is small. This SAFT processing may be
performed in either two dimensions or three dimensions; however, three-dimensional SAFT requires
considerable processing time.
An example of the improved signal-to-noise ratio and lateral resolution of SAFT processing is shown
in Figure 11-9, in which three pairs of 6mm diameter side-drilled holes were scanned using a parallel
scan on a 30mm thick steel plate. The image on the left is the raw data as it would appear in a B-scan
and the image on the right is the result of SAFT corrections to improve resolution.
181
Before SAFT processing After SAFT processing

Figure 11-9 DSP by SAFT on a Parallel scan over side-drilled hslgs (st)
11.2.5 HYSTERESIS OR BACKLASH COMPENSATION

In a mechanical system backlash is the maximum distance through which one part of something can
be moved, without moving a connected part. When a mechanised scanning apparatus moves a probe
back and forth, this is the play or clearance between mating components that results in lost motion
due to clearance or slackness when movement is reversed. The effect is also sometimes termed
hysteresis, which is a delayed response by an object to changes in the forces acting on it.
Ideally, the backlash should be zero, but in practice some backlash must be allowed to prevent
jamming. It is unavoidable for nearly all reversing mechanical couplings, although its effects can be
negated. In an ultrasonic scanning system, when moving probes in an X-Y patterry the reversal of
direction at the ends of the scan can cause the encoded positions to be offset, since encoder motion
encodes the position when no movement is occurring, as the slack is taken up.
If the scanner has significant backlastr, the scan image will have scan lines that are offset in alternating
rows. There are two ways to correct this: use unidirectional data collection, so data is collected in orLly
one direction; or the data can be post-processed using a backlash removal amount. Every other row
will be offset by the amount assigned in the software to re-align the data in the image.
182
Chaoter (I1): Dieital Sienal Processing
71,.3 OTHER DSP

11.3.1 DATA COMPRESSION
'l
Data compression is a post-processing function that t: l
eliminates some of the amplitude data points along

the time base of each waveform. Software attempts
ii
to identify the "highlights" of the signal so that it
can be reconstructed from a smaller number of
points without losing the main features, including
position, along the time-base and amplitude. The
before and after of such a Process is illustrated in
Figure 11-10.
Figure 11-10 Data compression with 4:1

reduction of data in an A-scan
1-1..3.2 SPLITSPECTRUMPROCESSING
Split spectrum DSP is a method of noise reduction that can be used in coarse-grain structures.
Electrical noise is random, so reasonably filtered with simple averaging. Grain-structure noise is
patterned and therefore not removed by averaging.
"Split Spectrum Processing"(52) is equivalent to filtering the signal by sectioning the spectrum. This is
because the signal spectrum is multipliedby a weighting window whose width is narrower than the
signal pass band. This leads to the first sub signal. By reiterating the operation and shifting the
weighting window central frequenry, a series of sub signals is obtained. The final result is produced at
any time by comparison of these sub signals. The aim of such processing is to improve the signal-to-
noise ratio and it is usually used in cases where the noise is due to the grain structure of the inspected
material. This procedure assumes that the useful signals are coherent and the noise is not. This type of
processing is designed for use with highly degraded signals with signal-to-noise ratios of the order of
1. Detection is then no longer based on an amplitude criterion, but instead on frequency coherence.
In split spectrum DSP, the signals are collected and then converted into their frequency spectrum via
the Fast Fourier process as noted above.
The spectrum is then "filtered" in the frequency domairy by selecting regions of the frequency content
to suppress. The process is illustrated in Figure 11-11 where the upper trace shows the input signal,
the upper middle section indicates the FFT of the entire signal with nine frequency filters placed along
the spectrum. The resultant output from each of the filtered components is indicated in the lower
middle frame. The combined output is then seen in the lower frame.
183
iI
Figure 11'11 Split spectrum filtering process
Because the split spectrum process is not based on an amplitude criterion, but instead on frequency
coherence, the process could be used for improved phase discrimination for separating closely spaced
upper and lower tip signals.
11.4 DSP ILLUSTRATIONS

Although DSP options are numerous, most TOFD systems incorporate only the basic DSP functions.
InspectionWare has perhaps the largest collection of DSP processors available in an NDT software
package. Some examples are shown below:
184
. l.{r}rlai: i3 thi:r il
. liiarrP.a:1 ::{an , ?ha;a ii Pt3l:
.:ari1i.'alua
*ini*ii F:;':gt [rP;*-r:lr
r Pr!n(iFil :ati**aant -titr:,
H aie,rtii Ul ,:3tr
. 1.,:i!laflatri
i |]ri1,:, i
lalar:naa it:!l*3!ii4
ir!
lattl{n{: },t':i3n
aiF. f:ilit I i i.ie.entt 5!t]!iJ*raa
' i:A U*:t*1Pi.r [;at+it ai*!i] i:t arj:at
F,! t,!^d,-1n
iin'l;a al,Perr*
' iit :ina ;"'3:a P.inr';e Ci illt.:
firp iaer:i Un*atrfl:J P':i*!: ii
F.a+iinai ]a*irn lritir ' 1.:ln'..
f req.iina-,' raax:ir :!'a jrntrnq R:liie t::',','
[u*n!rE !airdel::if,4
U;ali
':i'i ti-. aai.
ji :nalq, ) its
:-itl l',',i,-tr::
i ale: r':'3:*a
, !:3rle Pe;nt
l*t. E 13l
SF3tril ea:nilPirn!
u Lelei atf;Jin,l
!!1!n. 1C
1p;it -ip:{trtr
' !tlitii:{a :3
,1,";r:
Dimensional Processors
185
* jr Hi:lagr*rn i*
t.3 H+ri:.atit *.af *r*t:ea iubt{a{ticn
?nilr air:;aiir.3fi:*
- . '... irr )r.., rn:i -.rn.n .nani' . Piaii!l$ral:1 :fa:r :a*ra*
Jfi[ iiu-ri;r:r31;11s " P.:n._:t: lI

:lu:i.i :.r:11-,5 ii,'i] Qe{1 F.- -. . --î.,.^ , -r.... .'n
-.. r:, (:îi^,.,^^
. Rim::.,-.lnJif:*ai For:rt: i*
':a*i;.:i Firj*{trrn
.:rau.i;:ii, IttSt.anai i*.fa{tr*n
'_..* E:ir-.:
l _trt,.1â! l-l>:i > ,n'r
:al*a i!{in:ai}e:1
, Srn!i; ?,:rnt lC
:oh.rt.a:: 9.1!na aan:atr!:x .:S
" -rF+r,'.:,Jih Fiil,:,

, :ttn.tii.J i.:!:i:'3n .:S
, !titi"iiri::3
: rn -.-: :")
' !,,.:anp:-irit*i:n
I i.]:r: tri:r*. ifaali
, ';':a'i:a.::::3na{
. :,':ti:i:: FrJ*r al]i* t**iar!'lr.].'l
' ,:ir*lt i*
aai::Ji ltiltr.in
Plane and Volume (2D and 3D) Processors
186
Clearly, not all of the DSP options available are applicable to TOFD data. Users often find that even
applicable processors will work better on some data than on others.
DSP becomes a matter of trial and error. \Mhen in a production inspection environmenf
experimenting with 20 or 30 options is not practical, so a limited, fixed set of processors would be
used. For a special application where data is difficult to analyse due to grairy geometry or other issues,
the operator might gain some benefit from having several options to choose from.
Even using the standard processors, it is usually seen that there is a preferred order of processing.
Therefore the first step in post process DSP is lateral wave straightening, to correct the small
variations of lateral wave arrival time due to mechanical irregularities. After that, it is'most beneficial
to equalise the response from the lateral wave signals, so that when the removal of the lateral wave is
done, the magnitude of the subtraction is equal for each A-scan. Subsequently, any remaining flaw
indication signals exposed can be analysed or further processed to see if more information or clearer
representations can be made.
To illustrate some of the DSP options not commonly available, a scan of a standard welded flaw
sample was made. This is a plate 200mm long, 200mm wide and 12.5mm thick. Three flaws were
seeded in the single V weld as follows:
Flaw Flaw type Distance from edge Leneth Height

1 Lack of Fusion 25 8 J
2 Porosity 100 8 J
J Root crack 150 10 J
The tabulated values are intended locations. Actual locations, as seen in a scan, may not correspond to
these values exactly.
The porosity is of a very light and dispersed nature, so does not characterise well and in fact is not
well defined as separate from other small indications in the area; however, the root crack and lack of
side wall fusion are relatively easy to see. Of particular interest for processing is the lack of fusion
indications. This is seen to be just at the edge of the lateral wave and its position relative to the exit
points makes the identification of separate upper and lower tip signals difficult. The scan was made
using a TOFD pair of probes 3mm diameter 10 MHz, on 70" refracting wedges and a 45mm PCS (see
Figure 1,1,-12).
-_, fr*Er
!b
,
!
,
Figure ll-1.2Civa modelled image of setup with flaws embedded
787
Chapter (11): Disital Signal Processing
Results of the data acquisition of the plate are seen in Figure 11-13'
rofd 1.sdt
Figure 11-13 Initial TOFD scan (raw data)
Setting an amplitude threshold of +1.0"/o (selecting the positive going portion of the RF signal) and
instructing the software to place the reconstructed image 1ps after the start of the A-scan display, the
lateral wave straightening is set as in Figure 11-14.
:J l'1se arsFl€te $an
trh;iltst+i;;,;
.....' -l
--- -,--- i a{
Align A-scan
Figure L1-1,4 A-scans aligned to lateral wave amplitude at +10% threshold
188
Having aligned the lateral wave signals, the next typical step is to equalise the amplitudes. This could
be done using a reference signal, however, the software allows the user to not only equalise the
amplitudes, but also set the amplitude that the signals will be equalised to. An extraction feature
provides the user with a view of all the amplitudes of the lateral wave signals (Figure 11-15).
i 9E r-l6il -- = 3l itlt ""
Figure 11-15 Variation of all lateral wave amplitudes along scan
The lateral wave amplitude variation allows the operator to select an equalisation level. For this
illustration 50% of positive displacement was used to obtain the equalisation level, as seen in Figure
11-16.
Lat*al Eqmli:e
r1 5:tlus. 1!i il*"rt =-l l:li
Figure 11-16 B-scan with lateral wave signals equalised to 50% positive displacement
The next step would normally be lateral wave subtraction. Instead of simply locating a typical A-scan
away from any flaws and selecting the lateral wave portiory we can use a different DSP functior;
mean or average of a series of A-scans.
A reference A-scan is made by taking the average of all the waveforms from every A-scan in the
straightened data set. This process averages out any irregularities caused by flaws in the mid-wall and
also considers the back-wall and mode-converted signals. The mean (or average) A-scan obtained by
this process is displayed in Figure 11-17.
1.89
's";""'"
Figure 11-17 Average A-scan from straighter-red dut
\Alhen the mean A-scan is then used as the reference, to subtract from all the A-scans collected, the
effect is seen in all parts of the scan, not just the lateral wave area. The result of the full waveform
subtraction is seen in Figure 11-18.
btract mean A-scan
Figure 11-18 TOFD B-scan after subtraction of the mean A-scan
\rVhen the lateral wave is removed in the region of interest (e.g., in the region where the lack of fusion
has entered the lateral wave ring-time), we can see evidence of an upper tip signal. Since the reference
used has the positive phase of the RF signal as white, the upper tip indication to look for, will be black.
The B-scan has been normalised such that the test surface is Ops (i.e. Ops = 0mm). The weak negative
signal at 0.101prs is most likely the upper tip of the lack of fusion indication. This corresponds to a
corrected depth of 3.6mm. Because of the small angle between the upper and lower tips at that depth,
the lower tip signal seems to be lost in the ringing that would be associated with the spatial resolution
limit. The theoretical lateral wave dead zone for this setup is 4.5mm. The flaw is estimated to occur
inside that limit at 3.6mm.
190
The spatial resolution of a flaw at 3.6mm is calculated to be at least 2.2mrn, but that assumes the
lateral wave wavelength to be the same as calculated for the nominal probe frequency. Because it is
off-axis, the actual pulse frequency in the lateral wave region is something less than the nominal. As a
result, the ringing from the upper tip is probably more than the equivalent 2.2rnm calculated. Looking
at the signal in Figure 11-19, this appears to be the case. If asked to estimate the lower tip location, an
operator might use the large positive signal at 0.36prs. This is equivalent to 7.0mm depth. This
indicates a flaw height of approximately 3.4mm (close to the intended flaw height made by the
manufacturer).
i a:
i-:r:ri: I
Figure 11-19 Analysis with lateral wave removed
By processing the d.ata, to provide the actual phase of the RF signals, we can attempt to identify the
points where the phase is 180" different. Using a 10% noise filter the effect is seen in Figure 11-20.
Figure 11-20 Analysis of just lack of fusion indication at 32mm scan path using phase analysis
191
Using the positions in time, for the points where phase is +90" and -90", for the signals selected in the
RF display, depths for the upper and lower tips of 4.3mm and 7.3mm are obtained.
In an effort to reduce the ring-time and associated noise, a two step process is used. First a high-pass
filter is used on the data (passing frequencies over TMHz). Then the resultant signals are averaged
with their two nearest neighbours to obtain the image in Figure 11-21.
qlrraging Flher 1D
{-S.480!s, 22 500nrn)l = 3.12506
Figure ll-2lLackof fusion indication at 32mm scan path using HP frf,"r'r"a'r1^raging
Using the positions in time, for the points where phase is peaked on the extracted A-scan, depths for
the upper and lower tips of 3.43mm and 6.80mm for a flaw height of 3.37 are obtained.
Selection of the best combination of DSP processors will usually be a compromise based on the
accuracy required and the time available. At some point, there will be a diminished level of return on
effort. DSP can help with some aspects of signal analysis; however, a significant factor in useful data
for analysis will still be a well planned technique and good quality data collection.
792
Chapter (12): Codes, Standards, Procedures, Techniques, And Written Instructions
CHAPTER (12): CODES, STANDARDS, PROCEDURES,

TECHNIQUES, AND WRITTEN
INSTRUCTIONS
12.1, DOCUMENTATION HIERARCHY

TOFD is like any other NDT application of ultrasonic testing. The concepts of good practice are laid
out in the company's procedural documentation and these procedures generally conform to
regulatory documents. The regulatory documents may or may not include acceptance criteria.
The documents in NDT can be considered as having something of a hierarchy. Codes and standards
are at the top of the chair; while specification and procedures are below them.
A standard is a written document assembled by recognised experts, with the purpose of

recommending actions to achieve certain objectives.
A standard is usually enforced or given authority by an organisation or agency (typically professional
societies or national institutions). When a set of standards is incorporated into law and thereby
enforceable legally, it is considered a code.
When a user or purchaser of a specific product requires assurance of quality level they will often
arrange their ornm document describing specific test parameters and acceptance criteria. Such a
product-specific standard is considered a specification. Although it may reference other codes and
standards, it can require more stringent limits than the more general standards.
When an AUT inspection company bids on a project they will be expected to provide a written
procedure describing how they will meet the requirements of the contract (Standard, Code or project
Specification). The requirement for a written procedure is almost universally required in all NDT
methods.
8.g., ASME Section V Art. 4 on ultrasonic testing of welds states:

" T -422 Writt en P ro cedur e Requir ements
T-422.1 Requirements. Ultrqsonic examination shall be performed in accordance with a written procedure..."
E.9., DNV OS F101 Submarine Pipeline Systems:
"Appendix D - Non-destructioe Testing - B 102 Non-destructioe testing shall be performed in accordance Toith
written procedures. . ."
793
The procedure is generally submitted to a client to indicate how the inspection company proposes to
implement its standard practices for the various aspects of the test method, in order to meet the
requirements of the Standard or Specification being used on the project.
In summary, there is a hierarchy of documentation in NDT:
1) Standards and codes, which provide general guidelines and limits.
2) Specification, which is a product specific document usually assembled by the customer.'
3) Procedures, which are usually an inspection company's statement of their standard practices or
a document designed to address the requirements of a specification used by tlts company they
are providing a service to.
4) A technique is often part of a procedure and provides a brief description of test application to a
specific part (also termed a Written Instruction).
For many years, lack of related Codes and Standards were a significant detractor to the use of TOFD
in NDT applications. Even with the introduction of the well-reputed document on the use of TOFD,
BS 7706, there was still reluctance to use it, because it was generally not possible to assess the
acceptability of a weld to TOFD. This was due to the fact that in the existing regulatory documents
(Codes and Standards) the acceptance criteria were based on length of a flaw exceeding some
reference amplitude.
A few companies derived workable specifications, but these were in-house documents and not widely
accepted. It has only been since about the year 2000 that TOFD has actually become widely accepted,
due to the change in the way flaws are assessed for acceptance. In 2000, an ASME Code Case (Code
Case 2235) was published, that permitted the use of ultrasonic test methods in lieu of radiography on
all ASME Section VIII welds with one half inch (12.5mm) thickness and greater. Provided the
ultrasonic technique met a list of prerequisites, any ultrasonic technique could be used. The list
included many factors, such as the need to use computer-based data acquisition and the need to
determine both the length and height of flaws and to demonstrate detection capabilities. Code Case
2235 has been revised several times over the years, but the requirements unique to automated UT
have remained.
The applicability of TOFD may not be obvious when first glancing at Code Case 2235, as it is not
specifically mentioned. However, there are two clues: reference to non-amplitude-based techniques
and the requirement to determine acceptable flaw size from the tables in the Code Case. These require
the height of a flaw be determined and from that height, an allowed length is established. This puts
two requirements on the UT: detection and sizing.
The detection is common enough to most NDT methods including pulse-echo UT. However, sizing
places an entirely new requirement on UT and it is a requirement that TOFD is well-suited to provide.
Although other methods for sizing can be used (e.g., tip echo backscatter methods in pulse-echo), only
TOFD can easily be used as both the detection and sizing tool.
Although the European Standards B3-7706 and EN-583-6 relating to the use of TOFD predated the
ASME Code Case 2235, they had no acceptance criteria. As a result TOFD remained a good idea that
was not widely used. The concepts, on which the ASME Code Case 2235 is based, are called "fitness-
794
for-purpose". These establish the tolerance to a flaw, based on the toughness of the material and its
thickness and service conditions.
Subsequent to the ASME Code Case several other Codes have incorporated similar acceptance criteria
based on the height of detected flaws. The list of useful Codes that allow TOFD and incorporate
acceptance criteria is growing and includes:
o ASME Code Case 2235 -Use of Ultrasonic Examination in Lieu of Radiography

o API 620 (American Petroleum -Welded Storage Tanks)
o API 650 (American Petroleum - Welded Stee1 tanks for Oil Storage)
. EN 75617-2009 - Acceptance Criteria for the Time of Flight Diffraction Inspection Technique
In all cases, the acceptance criteria found in these standards are in the form of tables that require the
assessment of a flaw length, for a measured flaw height and for the material thickness tested.
Not all Codes and Standards provide acceptance criteria. Some provide just the test requirements and
leave it to construction standards to provide the acceptance criteria.
Examples of standards regulating or providing guidance on the implementation of the TOFD test
method include:
. BS 7706 (7993) Guide to Calibration and setting-up of the ultrasonic time-of-flight diffraction
(TOFD) technique for detectioru location and sizing of flaws. British Standards Institute 1993.
o CEN EN 583-6 Time of Flight Diffraction Technique as a method for defect detection and sizing
o ASTM E 2373 Standard Practice for Use of the Ultrasonic Time of Flight Diffraction (TOFD)
Technique
o ASME Section V Article 4 Appendix III and Appendix N
. ISO L0863, Technical Specification Welding - Use of Time-of-Flight-Diffraction Technique
(TOFD) for Testing of Welds (Previously identified as EN TS EN 1,4751)
Standards, such as those listed above, are the attempt of industry to ensure a minimum level of
functionality or quality. When a new method arrives on the scene, industry often lags the research and
development associated with that new method. This was the case with TOFD where it took over 72
years, from its introduction in 1976, to establish the first TOFD Guidance Standards (draft document
BS 7706). It took another 12 years to then establish the first acceptance criteria tools for the use of
TOFD in day-to-day applications (ASME Code Case 2235).
An important aspect in the practical application of TOFD, as it relates to the standards, is the need for
accompanying documentation. General aspects of a company's standard practices for an NDT test
method are contained in a document called a Procedure. ISO 10863 is a convenient option to avoid
writing a separate document in some applications. In ISO 10863 is stated, concerning a "Written
Inskuction":
6.3 Written test instruction
For examination leaels A qnd B, this internatioruil standard satisfies the need for a written test procedure.
795
Cha r (12): Codet, Stuodur@ ues, And Written Instruitions
When required to draft such a document (i.e. a TOFD Procedure), it is recommended to review the
contents and layout of iSO 10863'
12.2 PROCEDURE GUIDELINES

A procedure should follow suitable guidelines for the title page and overall layout' A title page would
inciude:
. The company name o Date of issue
. A document identification number r Signatures
. Originator (writer, qualification) o Provision for Revisions recording
o Authorising/approvingPerson r
(Note: All Pages need to be numbered)
After the title page, headings should be used to group the information presented'
72.7.1 SCOI',E
. Introductory statement describing the project and material tested
r Details of the material, component type, thickness range, test area, etc.
72.2.2 STANDARDS AND REFERENCES

o A list of relevant stand ards and reference documents, including specifications
12.2.3 DEFINITlONS
o A statement on the use of standard definitions, together with definitions of non-standard terms
used; which, if not defined, could lead to ambiguit/, and a list of abbreviations used in the text
of the procedure
72,2.4 I'ERSONNEL
. The minimum requirements for training, certification and experience of AUT operators and
supervisors
12"2.5 IDENTIFICATION AND DATUM POINTS

r Identification of the material, weld or component to be inspected (e.g., weld numbers)
. The means of marking datum points to ensure repeatability of test results or the future location
of defects
o A clear system for measuring from datum points to defect indications, e.g., 'clockwise from
datum looking in the direction of flow'
a A statement that repairs etc. must be identified separately from original work
12.2.6 SURFACE PREPARATION
a The standard or level of cleanliness, surface finish and/or flatness required, together with
details of the precise extent of the area to be prepared
796
12,2,7 EXTENT OF lNSPECTION

. Specify the precise extent of the test zones, e.g., "Test weld and HAZ", "Test all surfaces", etc.
12.2.8 EQUIP\{ENTSPECIFICATION
o Should include type(s) of equipment to be used, details of any specifications or requirements to
be met, as well as ancillary equipment to be used during the test, e.9., probes, calibration-pieces
12.2.9 EQUiP\4ENT CALIBRATION

o Method of and requirements for, calibration of system and system elements, including records
o Required frequencies of calibration or recalibration
.12.2.10 EQLiIPL,{ENT CHECKS

. What functional checks are to be performed andl,vhery records to keep
o The action to be taken in the event that the equipment fails a functional check
12.2.11 ASSESS\,{E\T OF TEST RESULTS - ACCEPTANCE CRITERIA

. The procedure must specify the minimum size, or type of indication to be recorded, and the
defect classification and sizing method to be used
12"2,12 REPORTINC OF TEST RESULTS

Either hardcopy or "software" reports are made, but both versions must be capable of identifying
aspects of the test, including but not limited to:
. Procedure reference
o Proiectname/location
. Weld number
. Equipment (system name/software version)
o Location of defects
. Status of defects, e.g., accept, reject
. Operator name and qualification
r Date of test
12.2.]3 ATTACHi\{ENTS
The attachments generally include a statement describing the sizing policies used and the specific
details of the test setup, including the calibration block details and specifics of ultrasonic probes used.
Instead of a separate document, prepared by a Level 2, to identify the scanning details as instructions
for a Level 1 operator, Written Test Instructions (techniques) can be incorporated into a procedure as
appendices.
797
Chapter (13): Industrial Applications Of TOFD
CHAPTER (1.3): INDUSTRIAL APPLICATIONS OF TOFD
TOFD has long been associated with weld inspections. Many of the examples used in this text draw on
weld profiles to explain aspects of setups and analyses. F{owever, TOFD has seen opportunities in
several different industrial applications. Many examples exist where users have found the advantages
of TOFD to be superior to that of conventional manual UT. Examples are:
. Weld inspection
. Corrosion surveys (root and area attack)
. Cladding assessment
o Crack detection (fatigue and stress corrosion)
o High frequency shear mode TOFD
13.1 HEAVY-WALL PRESSURE VESSEL CONSTRUCTION
Construction of pressure-vessels, such as the butane spheres in Figure 13-1, is an ideal candidate for
TOFD. 1,00% of the welds on these spheres in Brazil were inspected. This provided more coverage
than radiography (radiography would only have been required to do a spot check of about 10% of the
weld lengths) and it was done without the interruption to production that radiation hazards would
have involved. It was decided that the increased speed of TOFD could be used to increase the amount
of inspection, thereby increasing the probability of detecting any serious flaws. The wall thickness
(nearly 60mm) would have made it necessary to use Cobalt 60, thus compounding the safety issues
and reducing the detection capability.
Figure 13-1 Butane sphere TOFD inspection (ss)
1.99
Chapter (13): Industrial Applications Of Tqry
13.2 WELD ROOT EROSION

Process & transport piping systems can be subject to chemical attack causing structural breakdown.
Areas of the weld adjacent to the parent metal are particularly sensitive. If allowed to progress, the
affected material can be more readily eroded by product flow.
A technique used for assessing this problem uses a TOFD profiling of the back-wall, because
conventional (zerc degree compression wave) wall thickness measurement techniques are
inappropriate, due to poor surface access from the weld cap and poor sensitivity for accurately
determining the mis-oriented surfaces of the irregular corrosion surface. An example'of the condition
and associated TOFD signal is seen in Figure 13-2.
:i:irl ir:!i:{} o: f,alJ ioal Lis'! ix. :f,'rrl
Figure 13-2 Weld 1s6f slesisn (5a)
1a a
-tJ.J CLADDiI{G INITERFACE INSPECTTON
Refinery vessels are often constructed with a corrosion resistant liner, clad to the interior surface of the
vessels. This liner material is deposited by explosive or fusion welding. As a result of "impingement
erosion" it is possible to break down regions of the bonded protective cladding and the vessel may
suffer attack on the ferritic steel. Figure 13-3 illustrates the TOFD corrosion mapping on a clad reactor
vessel in a refinery. Areas of cladding breakdown were identified and depth assessed. This allowed
the affected areas to be monitored and repaired.
Figure 13-3 Cladding interface inspection (ss)
200
1,3.4 STRESS CORROSION CRACKING

Turbine disks present a challenge for inspections due to their size, mass and complex shapes. TOFD
has been found useful in several applications on turbine disks. Disks of low-pressure turbine rotors
are highly stressed components, subject to aging, due to the environment in which they operate.
Mechanical stresses, especially in the area of the shrink fit may result in stress corrosion cracking
exacerbated by the temperatures during turbine operation. The dendritic inter-crystalline stress
corrosion cracks occur preferentially on the hub inside surfaces and on the axial anti-rotation devices
(keyways) of the disks. They exhibit axial-radial orientation, due to the tangential direction of the
main stress. As a resul! unallowable crack growth, or in extreme cases disk rupture, may occur after
long periods in service.
Turbine disks are constructed with a series of disks mounted on a shaft using a shrink fit process.
Inspection is carried out by several ultrasonic techniques, including TOFD, using a long arm on which
the probes are mounted. The gantry, from which the probes are moved, is moveable from disk to disk
and the scans carried out by rotating the turbine unit so the entire volume of concern is scaru:red.
Figure 13-4 illustrates the turbine with the cover removed and the probe gantry positioned to access
one of the disks. The image at the right in Figure 13-4, illustrates the scans made. These include a
small angle shear pulse-echo relying on the corner effect, a high angle shear pulse-echo relying on the
tangential effect, a zero degree in the radial direction to assess for corrosion and TOFD to size the
flaws detected by the pulse-echo techniques.
.E
*a
G*
s€ =
-I.-ffi
Figure 13-4 Turbine disk inspection (inspection arm (left), techniques used, (right)) tse)
13.5 FATIGUE CRACKING

Vibrations or cyclic loading can set up conditions that risk fatigue on components. When potential
areas are identified for this failure mechanism, they can be monitored. Although subsurface flaws can
be a site for the initiation of fatigue cracks, surface irregularities are perhaps more often the initiation
source. A sharp transition between excess root beads, or a sharp rise of a weld cap from the parent
metal, may be all that is needed to concentrate the stresses and get the cracking process started.
Because they tend to double the stress concentration, surface flaws are generally of greater concern.
201.
Moving parts are commonly plagued with fatigue issues. One such component is the turbine rotor
shaft in large electricity generating facilities. These shafts are machined to hold the disks and couple to
the generator. The transition from one diameter to another presents a particular concern for thermal
fatigue cracks. Space is limited and the elevation differences, between the probe access surfaces,
means that the TOFD techniques for some of these geometries provides no lateral or back-wall signal.
Figure 13-5 illustrates a turbine rotor shaft on a rail car, ready to be delivered to the location where the
turbine disks will be shrunk fit.
Figure 13-5 Turbine rotor shaft (s7)
A paper by Nathtssl describes a TOFD technique for inspecting a critical groove region subject to
thermal fatigue cracking. This geometry had the probes either side of the groove and both were placed
well above the radius surface. Sizing of surface breaking flaw depth was the goal of the technique and
it proved to be effective down to the 5mm deep slits used (see Figure 13-6).
{s}
-l
Figure 13-5 Reference targets for rotor inspection by TOFD €8)
202
1,3.6 SHEAR MODE TOFD

Great debate continues, in some circles, on the ability to size flaws by ultrasonic means. A wide
variety of performance demonstrations and round-robin trials have been carried out on weld flaws.
These have typically provided a precision (using standard deviation of error in sizing) of just over
lmm.
A notable difference is seen when the material and geometry are more uniform than butt welds. On
nuclear materials using much shorter wavelengths and perfect geometry, repeatable accuracies of t 0.1
mm have been achieved using shear mode TOFD. Lindenschmidt(se) used an imr:nersion TOFD
configuratioru with 20MHz probes and a 45' shear mode, to size delayed hydrogen cracks in
Zirconium/Niobium tubing having a wall thickness of only 4mm. Origin of the cracks and the TOFD
setup is seen in Figure L3-7.
Contact between the pressure tube and calandria tube in the CANDU reactor is normally maintained
via small annular spacers (called garter springs). When the springs are out of positior; it is possible for
the hot pressure tube surface to contact the cooler calandria tube and develop a hydride blister at the
contact point. The localised hydrogen concentration embrittles the Zr-Nb alloy and cracking can form.
Crack depths were determined to be on the order of 0.4mm and subsequent sectioning established
that the sizing errors were less than 0.1mm.
20/v/Hz
€laEdria lube
?0\ss13$s
ll isier ard caaak

formed at corlacl ccrnt
Figure L3-7 Immersion shear mode TOFD to size hydride blister cracks in 4mm thick Zr-Nb pressure
tubes (contacting tubes - 1eft, detail of TOFD - right)
13.7 HDPE FUSION BUTT \,VELDS

High Density Polyethylene (HDPE) has become a popular material for pipeline applications. With a
density of 0.96grams per cubic centimetre and an acoustic velocity (compression mode) of 2430mls
(and no measurable transverse velocity component), the normal wedges used for refraction result in
negative refraction angles.
203
Chapter (1Q: Indgstrial Applications Of TOFD
Welds are typically made by preparing the end faces of the

pipe to be joined with perfectly squared and smoothed
surfaces. These surfaces are cleaned with a liquid cleaner
and then heated simultaneously.
When the welding temperature has been achieved, the

heater plate is removed and the pipe sections pushed
against each other with sufficient pressure to allow the hot
surfaces to fuse. Since the polymer is heated it deforms and
results in a displacement of material on both the inside and
Figure 13-8 HDPE fusion butt weld
outside surface of the joint as seen in Figure 13-8.
cross section (60)
Although it is possible that there is the occasional inclusion in the HDPE material, by far, the most
common flaw encountered is non-fusion. Because of its vertical orientation TOFD has proven to be
one of the most effective test methods to detect non-fusion in the HDPE butt joints.
An effective option for HDPE fusion weld inspections uses a specially designed wheel probe that
contains two sets of TOFD probes and is covered with a specialised polymer. This scanner allows
inspections throughout the welding process to detect and recognise defects early on. It is available as a
magnetic scanner (for steel welding) or non-magnetic scanner for HDPE welds. The tyre material for
the wheel covering has an acoustic velocity of 1600m/s and a density of about 0.96grams per cubic
centimetre. This provides a very close match of acoustic impedance to the HDPE. The positive
refracting angle ensures that the beam entry points can assure that the weld cap material does not
limit the approach for the beam entry point. Positioning and beam paths are illustrated in Figure 13-9
and Figure 13-10.
Figure 13-9 TOFD wheel probe mounted on HDPE Figure 13-10 Schematic representation of dual
weld samPle(to) TOFD beam paths in TOFD wheel probe(16)
204
Chapter (13): Industrial APPlications Of TOFD
Non_fusion flaws can be seen in the HD'E scans with a similar characteristic to non-fusion in steel
fusion welds. See Figure 13-11'
EZ9lleZl
o3u3drurl :
g
sLs??oerl6B
lllllttllrr
uTsld (oo)
Figure 13-11Non-fusion flaws detected in HDPE
205

Collection: Software

Uploaded by

Copyright:

Available Formats

Collection: Software

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Collection: Software

Uploaded by

Copyright:

Available Formats

Chapter (9): TOFD Data Collection Software

CHAPTER (9): TOFD DATA COLLECTION SOFTWARE

9.1 ULTRASONIC SETUP OPT]ONS

Figure 9-1UT setup window for A-scan display tao)

Table 9-1 UT setup parameters

9.2 ACQUISITrON OPTTONS

Menu displays for other features controlled include:

Figure 9-2 TOFD acquisition lngnss (40)

Figure 9-3 TOFD probe and part entry menu (ao)

Figure 9-4 TOFD Scanner entry menu (ao)

CHAPTER (10): TOFD DATA ANALYSIS

10.1 FILE STRUCTURE

10.2 TOFD DATA ASSESSMENT

10.2.1. QUALITY ASSESSMENT

The most common problems experienced in TOFD scan quality are:

o Data collection gate positioning

These issues are described with examples

The example in Figure 10-6 illustrates

Figure 10-5 Incorrect gate length- too long (az)

Figure L0-7 Incorrect PCS- too long (+z)

Figure 10-9 Electrical noise (42)

10.3 TOFD DATA ANALYSIS

echnology Design Sonotron

Figure 10-13 TOFD analysis displays

. Scaling or zoom (in and out)

' HyPerbolic cursors

10.3.1 ESSENTIAL TOFD ANALYSIS FEATURES

Figure 10-14 illustrates the basic display with two

The vertical scale in Figure 10-14 is the time axis

10.3.1,2 LINEARISATION ALGORITHMS

Figure 10-15 Non-linear relationship of depth with time

Figure 10-17 TOFD data linearisation

no back-wall signals are Scan eris Sran aril

estimated acoustic velocity . FerpPndicuiEr lo beem 'i Perpendicul+ io beam

Lll-U pr mary value

o DoD i-. mm . !.OOO . *,

Weoge delay : --t

or cursors. These steps are

simple data entry window,

10-18. Once values are I UK Cencel

entered, pressing "OK" has

I-lr--l i;n::r .r,:i!

Figure 10-18 TOFD Linearisation Data Entry Windows

10.3.2 ENHANCED TOFD ANALYSIS FEATURES

10.3.2.1 PALETTE ADIUSTMENT

Examples of these two palette options are

10.3.2,2 ZOOM (IN AND OUT)

1.0,3,2,3 ADDING TEXT TO IMAGES

useful to add text to identify features of e-$ !:w

convenient if the TOFD software is

combined with the cursor positions is

10.3.3 DATA ASSESSMENT

10.3.4 FLAW IMAGES

Figure 10-22 P oint diffractor (pore)

1.0.3.4,2 I}/S/DE (ID) FAR SURFACE-BREAKING FLAWS

Figure 1:}-Z3Far side flaw - concave root or burn-through (+a)

10.3.4.3 OUTSIDE (OD) NEAR- SURFACE BREAKING FLAWS

Figure L0-25 Near side flaws (aa)

firp iaer:i Unatrfl:J P':i!: ii