August 1998 646/1998

7311.01/98/975.03
1995-97 CRP Programme 9402-5
APRIL 1998

INTERPRETATION OF ULTRASONIC TIME-OF-

FLIGHT DIFFRACTION DATA –
A GUIDE

M.R. Kirby and B.W. Kenzie

Copyright © TWI 1998

INTERPRETATION OF ULTRASONIC TIME-OF-FLIGHT DIFFRACTION DATA – A GUIDE

TECHNOLOGY BRIEFING

This report from the Core Research Programme (CRP) contains a compilation of TWI
staff knowledge and experience of the interpretation of ultrasonic time-of-flight diffraction
(TOFD) data.

Background

The ultrasonic TOFD technique was developed for the UK nuclear industry during the
1970s to provide a more accurate ultrasonic method for measuring the through-wall height
of planar flaws. Over the last decade a general increase in the understanding of the
technique has resulted in its use for weld inspection in a wider range of industries,
especially oil, gas and petrochemical. Further, the use of TOFD for initial flaw detection as
well as accurate flaw sizing has become more commonplace due to the rapidity of data
acquisition. This has resulted in the publishing of UK and provisional European standards
for the application of the TOFD technique1,2.

TWI have extensive experience of the TOFD technique for inspection of a wide range of
materials and applications. However, the majority of the work has concerned examination
of welds in ferritic steels including critical flaw sizing to provide data for fitness-for-
purpose assessments. This report is intended for inspection personnel who have a basic
understanding of the technique and provides guidance on the interpretation of TOFD data
acquired primarily from plane geometry welds in ferritic steels. The report contains many
useful hints and tips learned through TWI's experience of interpretation of TOFD data.

Objective

To develop guidelines for accurate interpretation of images obtained from the ultrasonic
TOFD technique.

Scope and Discussion

The TOFD technique normally uses a pair of compression wave probes (typically
between 2 to 10MHz in frequency) arranged in a transmit-receive configuration. The
technique relies on the interaction of ultrasonic waves with the extremities, or tips, of flaws.
When planar flaws are examined by TOFD, diffracted waves are produced from their
upper and lower tips and it is the relative time delay of these responses which allows the
through-wall position and height of the flaw to be accurately determined. This method of
flaw sizing is inherently more accurate than conventional ultrasonic sizing techniques, which
rely on the measurement of echo amplitude, or change in echo amplitude as the probe is
moved relative to a flaw. These echo amplitudes can be very variable, which leads to large
uncertainties in the estimated flaw size.

PRAD No: 7311.01/98/975.03 Page i

Copyright © TWI 1998
INTERPRETATION OF ULTRASONIC TIME-OF-FLIGHT DIFFRACTION DATA – A GUIDE

Despite its high accuracy for measuring flaw height, the advantage of using TOFD for
initial flaw detection is not so certain. TOFD relies heavily on the correct interpretation of
data images, which can sometimes be difficult. For example, due to the low amplitude of
diffracted signals, high equipment gain settings are generally used. In some circumstances
this can result in numerous indications from features such as small gas pores and plate
inclusions which can make the identification of more serious flaws difficult. To help with
the correct interpretation of TOFD images, this report draws on the experience of the
authors and on the recommended procedures given in standards.

The report contains information on the basic principles of the TOFD technique and data
collection, together with the method of construction of the two types of data image that are
produced, D-scan and B-scan. Guidance is provided on the interpretation of these
images, which includes determination of flaw type, or character, and accurate
measurement of flaw size and position. Advice on the recognition and sizing of planar,
threadlike, volumetric and point flaws is given, and a selection of schematic and real
TOFD images are included.

Concluding Remarks

The following concluding remarks are made with regard to the interpretation of TOFD
data. In addition, it is TWI's opinion that TOFD should not be used in isolation for the
detection and assessment of flaws. It is recommended that complementary NDT
techniques are also employed where there is concern over the initial detection of particular
flaws (e.g. surface or near-surface flaws), or where doubt exists over the interpretation of
particular indications seen on TOFD images. As a consequence, where it is appropriate,
TWI's preferred approach is to use TOFD to complement pulse-echo ultrasonic
inspection.

• It is recommended that companies develop their own specific procedures for the
interpretation of TOFD data based on the guidelines provided in this document.

• Where TOFD inspection is used exclusively, a full evaluation of flaws is only possible
if both D-scan and B-scan images are available.

• Where a weld cap causes an obstruction to B-scans, consideration should be given to

local removal of the cap. Note that this is only possible in situations where removal of
the weld cap does not cause strength matching problems.

References

1 BS 7706:1993: 'Guide to calibration and setting-up of the ultrasonic time-of-flight

diffraction (TOFD) technique for the detection, location and sizing of flaws'

Page ii PRAD No: 7311.01/98/975.03

Copyright © TWI 1998
INTERPRETATION OF ULTRASONIC TIME-OF-FLIGHT DIFFRACTION DATA – A GUIDE

2 prEN 583-6 'NDT - Ultrasonic examination - time-of-flight diffraction technique as

a method for defect detection and sizing'.

PRAD No: 7311.01/98/975.03 Page iii

Copyright © TWI 1998
INTERPRETATION OF ULTRASONIC TIME-OF-FLIGHT DIFFRACTION DATA – A GUIDE

1. INTRODUCTION

The time-of-flight diffraction (TOFD) technique1 was developed for the UK nuclear
industry during the 1970s to provide a more accurate ultrasonic method for measuring the
through-wall height of planar flaws. This development was timely since the nuclear industry
had a need to reduce the error that was assumed in the flaw height for the application of
fracture mechanics assessments. Over the last decade, the technique has found application
in other industries such as oil, gas and petrochemical. Further, an increase in the general
understanding of the technique and of the resultant images has led to its use as a means of
flaw detection, benefiting from its rapidity, as well as sizing capability. This has resulted in
the publishing of UK and provisional European standards for the application of the TOFD
technique2,3. Further, a document proposing flaw acceptance standards for TOFD has
been put forward as a possible work item for a working group of the CEN TC 121
committee4.

TWI has been involved in TOFD developments since the late 1970s5,6 using a variety of
equipment. The TOFD technique has been used by TWI for an extensive range of
applications, ranging from laboratory examination of welds in polyethylene utility pipes to
on-site inspection of thick walled hydrocrackers at elevated temperature. However, the
majority of applications have concerned the examination of welds in ferritic steels including
critical flaw sizing to provide data for fitness-for-purpose assessments.

TWI's preferred approach is to use the TOFD technique to complement pulse-echo

ultrasonic inspection. In part, this is because current flaw acceptance standards have been
developed with the latter in mind. Hence, TWI's opinion is that the most effective
ultrasonic test procedure is one which involves automated pulse-echo inspection for initial
flaw detection and length sizing, with TOFD for determining the through-wall position and
height of flaws. However, where cost is an overriding factor in specifying an inspection,
often only one ultrasonic technique can economically be employed. TOFD, despite a high
cost for this method of NDT, is sometimes selected in preference to pulse-echo ultrasonic
inspection, since it can rapidly inspect a given length of weld.

This report provides guidance for inspection personnel on the interpretation of ultrasonic
TOFD data acquired primarily from plane geometry welds in ferritic steels. However, the
effects of curved geometries are also discussed. It is intended to supplement information
provided in existing TOFD standards governing application of the technique and includes
many useful hints and tips learned through TWI's extensive experience of interpretation of
TOFD data.

2. OBJECTIVE

To develop guidelines for accurate interpretation of images obtained from the ultrasonic
TOFD technique.

PRAD No: 7311.01/98/975.03 Page 1

Copyright © TWI 1998
INTERPRETATION OF ULTRASONIC TIME-OF-FLIGHT DIFFRACTION DATA – A GUIDE

3. BASIC PRINCIPLES OF THE TOFD TECHNIQUE

The TOFD technique relies on the interaction of ultrasonic waves with the tips of flaws.
When an ultrasonic wave interacts with a planar flaw, it results in the production of
diffracted waves from the extremities, or tips, of the flaw in addition to the normal
reflected wave. This diffracted energy is emitted over a wide angular range and is usually
very weak (i.e. low amplitude).

The TOFD technique normally uses ultrasonic compression waves in the frequency range
2 to 10MHz. Compression waves are used as, (i) these arrive first at the receiving probe,
and (ii) they diffract favourably. Their use simplifies the received waveform and therefore
the interpretation of the resultant images. However, the commonly used compression wave
probe angles (45, 60 and 70°) also generate shear waves which travel at a slower
velocity. These shear waves provide additional diffracted signals which usually arrive at a
later time than the compression waves. The basic configuration for the TOFD technique
consists of a transmitter and receiver probe pair held in a jig at a fixed separation
straddling the volume of material to be inspected, e.g. a weld (Fig.1).

Highly damped probes having a wide beam width are usually used, since the diffraction of
ultrasonic waves from flaw tips is less dependent on the orientation of the flaw than for the
case of reflection from a flaw face. Figure 2 shows an idealised A-scan signal received
from an embedded planar flaw.

The first signal to arrive at the receiver probe is the lateral wave (LW) which travels just
beneath the upper (scan) surface of the specimen. In the absence of any flaws, the second
signal to arrive is that reflected from the far surface, commonly referred to as the backwall
echo (BWE).

Where a planar flaw exists, diffracted energy from its upper tip will be received before that
from its lower tip. These two compression wave signals from the flaw tips will occur
between the lateral wave and backwall echo responses. Therefore it is possible to
determine not only the through-wall height of the flaw, but also its location within the
thickness of the specimen. Note that Fig.2 shows a phase reversal between the lateral
wave and backwall echo, and between the upper and lower flaw tip signals. This
characteristic effect can assist interpretation of TOFD data and is consequently why
TOFD A-scans are acquired unrectified, i.e. as RF waveforms. The time separation of the
diffracted signals from the flaw tips is directly related to the through-wall height of the flaw.
All measurements of flaw height and through-wall position are made from the time-of-flight
times of the received signals, and as such, signal amplitude (which is notoriously variable)
is not used.

Recognition of these diffracted signals makes it possible to establish the presence of flaws.
The technique may be used for initial flaw detection purposes using a single probe pair, as
signals may be received from flaws of different orientation.

Page 2 PRAD No: 7311.01/98/975.03

Copyright © TWI 1998
INTERPRETATION OF ULTRASONIC TIME-OF-FLIGHT DIFFRACTION DATA – A GUIDE

4. TOFD IMAGES

4.1. GENERAL

There are two types of TOFD images that are generally used together. These are
commonly referred to as B-scan and D-scan images. They differ only in the way in which
the probes are manipulated with respect to the scan axes during data collection.

4.1.1. The D-Scan

A D-scan is an image that shows the RF A-scans collected when scanning the probe pair
normal to the direction of the beam along the length of a weld or flaw (Fig.3). The D-scan
is the usual type of scan performed during the flaw detection stage of an inspection. D-
scans are also known as longitudinal or non-parallel3 scans.

4.1.2. The B-Scan

A B-scan is an image that shows the RF A-scans collected when scanning the probe pair
in the direction of the beam transversely across a weld or flaw (Fig.4). B-scans are usually
acquired to enable more accurate location and sizing of flaws following the detection scans
(usually D-scans). B-scans are also known as transverse or parallel3 scans. The latter
means that the probe movement is parallel to the ultrasonic beam direction.

4.2. M ETHOD OF CONSTRUCTION

During scanning, as the probe pair is moved, an RF A-scan is acquired at specified

intervals of probe movement, typically every 1mm. Each A-scan is digitised, typically using
an 8-bit analogue to digital converter. Such a system results in the signal being represented
by a series of values from 0 to 255. To form an image each of these values is assigned a
shade of grey. For example, 255 may represent black, (+100% full screen height), and 0
may represent white (-100% full screen height). Therefore, the mid range value of 128
(i.e. zero signal amplitude) will lie between black and white and be assigned a shade of
mid grey. Note that depending on the software the black and white could be reversed.

Each grey-scale converted A-scan is then 'stacked' beside the next, thus forming an image
showing the lateral wave (i.e. scan surface) and backwall echo (i.e. far surface). The gap
between the lateral wave and backwall echo responses represents the material thickness,
and any flaw signals are displayed between these two. It should be noted that the
relationship between time and depth is not linear on as-acquired TOFD data.

PRAD No: 7311.01/98/975.03 Page 3

Copyright © TWI 1998
INTERPRETATION OF ULTRASONIC TIME-OF-FLIGHT DIFFRACTION DATA – A GUIDE

This unique presentation of the A-scan data allows the interpreter to discern the phase of
the lateral wave, backwall echo, and of any flaw signals detected.

Irrespective of whether a B-scan or D-scan has been acquired, the A-scans from which it
has been generated can usually be displayed alongside the grey scale image to aid data
interpretation.

4.3. LIMITATIONS

4.3.1. D-Scans

It is important to realise that a D-scan provides an image of a volume of material. For

example, two separate indications seen on a D-scan image could be either interpreted as
two edges of a planar flaw or two unassociated thread-like flaws (Fig.5). In this situation
the availability of a B-scan image through the indications will help to determine the flaw
type(s). In addition, attention should also be paid to the phase of the two indications,
although in TWI's experience planar flaws do not always provide clear evidence of phase
reversal between their upper and lower tip indications. Where there is doubt, the
indications should be reported as potentially a planar flaw, subject to a cross-check using
pulse-echo ultrasonic inspection.

Further, an indication on a D-scan image which appears to be threadlike, may arise from a
planar flaw oriented parallel to the scanned surface e.g. a lamination. The availability of a
B-scan image will again help to confirm the flaw type, i.e. whether it has measurable width
as well as length.

It should be realised that not all of the flaws present may be apparent on a D-scan image,
i.e. the presence of one large flaw may obscure another at a similar depth but at a different
'Y' (i.e. lateral) location (Fig.6).

Lastly, when using the probe separation required for initial flaw detection scans, near scan
surface flaws may be difficult to identify owing to the time duration of the lateral wave
response. Where available, data acquired using a more appropriate procedure (i.e. probe
type and/or separation) should be evaluated.

4.3.2. B-Scans

These are usually performed to size flaws that lie parallel to the longitudinal 'X' axis of the
weld. Where a weld cap exists, the options for B-scans are limited, often resulting in the
probe separation being too great to resolve near scan surface flaws satisfactorily from the
lateral wave response.

If a weld cap was present when the data were acquired, this may result in a less than
optimum image of a flaw. For example, during the B-scan, the probe movement may have

Page 4 PRAD No: 7311.01/98/975.03

Copyright © TWI 1998
INTERPRETATION OF ULTRASONIC TIME-OF-FLIGHT DIFFRACTION DATA – A GUIDE

been insufficient to produce the peak (or minimum transit time) of the diffraction arcs
required to determine the true flaw ligament and/or height. Where this is the case, to fully
characterise a flaw, it is recommended that the weld cap is locally dressed flush and
further B-scans acquired.

5. INTERPRETATION OF DATA

5.1. GENERAL

Ideally, prior to data interpretation, reporting and acceptance criteria should be

established between the relevant parties in relation to the particular application.

5.2. DATA QUALITY CHECKS

Where possible, the first step in any interpretation is to read the procedure used for
acquisition of the data. The data should then be examined looking for signs of spurious
data and/or areas of poor coverage. These may occur due to lack of couplant, poor
surface condition, electronic interference etc.

The RF A-scans should be examined in order to assess the suitability of the probe pulse-
length (a function of probe frequency and damping) and the test sensitivity used to acquire
the data. Guidance on the suitability of the probe frequency used to acquire the data may
be found in Tables 1 and 2 of reference 3. Where an adequate procedure has been
followed the scan parameters and operator's comments should contain valuable
information for the interpreter. Important parameters, such as the ultrasonic velocity,
probe centre separation (PCS) and sample thickness, should be checked for accuracy.
The method of setting the reference sensitivity should also be verified at this stage.

Any other relevant documents should be examined for useful information. For example,
drawings of the weld detail will normally show if the weld is double-sided, thereby giving
an indication of where root flaws are likely to occur on the TOFD images.

5.3. IDENTIFICATION OF FEATURES OF INTEREST

To identify features of interest that require further evaluation, all TOFD images should be
viewed individually and preferably printed out as a set of hard copies. In the majority of
cases, this will involve viewing one or more D-scan images covering the total length of
testpiece scanned. Features of interest are identified as indications which are not due to
testpiece geometry and which exceed the background noise level by a significant amount

PRAD No: 7311.01/98/975.03 Page 5

Copyright © TWI 1998
INTERPRETATION OF ULTRASONIC TIME-OF-FLIGHT DIFFRACTION DATA – A GUIDE

(for example, a signal to noise ratio of 6dB might be appropriate). Such features should be
recorded and assigned a unique indication number.

6. FLAW CHARACTERISATION

6.1. GENERAL

Owing to the relatively high gain required to show the low amplitude diffracted signals
arising from planar flaws, it is often the case that TOFD images contain numerous
indications from less serious flaws, such as small gas pores and slag inclusions (especially
where the weld or parent plate is 'dirty'). However, it is possible to distinguish between
these less serious flaws and more serious ones, such as cracks, etc. Small slag inclusions
and gas pores are generally referred to as point reflectors and they produce characteristic
arc-shaped indications on TOFD images. Planar flaws (cracks, lack of fusion etc.) will
generally have measurable length and through-wall height.

The reason for these diffraction arcs is simple: when a flaw is symmetrically located
between the probes, the time-of-flight of the flaw signal is at a minimum. As the probes
move away from this position the flaw will still be detected (due to beam divergence) but
the time-of-flight of the flaw signal will be increased. Hence, if the probes are scanned
from one side of the symmetrical position to the other, the time-of-flight of the diffracted
signal will reduce to a minimum and then increase again, forming an arc shaped indication
on the TOFD image. Semi-arcs are often seen at each end of planar and linear flaws as
the beam moves on to and off the flaw.

It is not usually necessary to report indications characterised as point reflectors, but in

some cases high concentrations of such flaws may be reportable. Most advanced TOFD
flaw detectors have a data processing aid in the form of a hyperbolic cursor (or 'point
reflector overlay') which simulates the response from a point reflector thereby helping the
interpreter to identify such flaws correctly. In addition, this cursor usually provides a direct
readout of through-wall position.

6.2. CHARACTERISATION REQUIREMENTS

Flaws should be characterised by the following:

• ‘X’ position and length: from D-scans.

• ‘Y’ position: from B-scans.

• Ligament (distance from scan surface to the nearest extremity of the flaw): from B-
scans wherever possible.

• Through-wall height (if measurable): from B-scans wherever possible.

Page 6 PRAD No: 7311.01/98/975.03

Copyright © TWI 1998
INTERPRETATION OF ULTRASONIC TIME-OF-FLIGHT DIFFRACTION DATA – A GUIDE

• Type, e.g. point reflector, planar, etc.

In addition, to characterise an indication on a TOFD image fully, the phase of the signal
associated with the flaw should be determined. In general, the following rules apply:

• A signal with the same apparent phase as the lateral wave shall be considered to
originate from the lower tip of a flaw.

• A signal with the same apparent phase as the backwall echo shall be considered to
originate from either the upper tip of a flaw, or from a flaw with no measurable height.

6.3. FLAW CATEGORIES

The following categories of flaws are generally recognised:

a) Planar flaws: e.g. cracks, lack of fusion. These will have measurable through-wall
height and length (Fig.7).

b) Volumetric flaws: e.g. large slag lines. These flaws generally have measurable
through-wall height, length and width (Fig.8).

c) Thread-like flaws: e.g. fine slag lines. Flaws in this category will have measurable
length but no measurable through-wall height, i.e. ≤3mm. Similar responses may
arise from lamellar flaws and lack of inter-run fusion flaws (Fig.9).

d) Point flaws (or point reflectors): e.g. gas pores and small pieces of slag. These flaws
will have no measurable through-wall height, length or width, i.e. all ≤3mm (Fig.10).

e) Uncategorised flaws: for those that do not fall within categories (a) to (d) above.
Uncategorised flaws should be considered as planar until further data are available.

The following is a guide to identifying flaws that fall within categories (a) to (d).

6.3.1. Planar Flaws

This type of flaw will have a different effect on the image depending on its through-wall
position.

Planar flaws that are open to the scan surface will only give an indication from the bottom
edge of the flaw. Depending on the extent of the flaw, this indication may weaken the
lateral wave or replace it completely. It is also important to observe the behaviour of the
backwall echo in such areas. Variations in coupling can result in significant time shifts of
the lateral wave which may be interpreted as a scan surface breaking planar flaw.

PRAD No: 7311.01/98/975.03 Page 7

Copyright © TWI 1998
INTERPRETATION OF ULTRASONIC TIME-OF-FLIGHT DIFFRACTION DATA – A GUIDE

However, in this case, the backwall signal will also be seen to mirror the shift in the lateral
wave. All suspected scan surface breaking flaws should be cross-checked against the
results of any available surface NDT methods, e.g. magnetic particle inspection. The phase
of such signals is the same as the lateral wave.

Planar flaws that are open to the far surface (i.e. backwall) will only give an indication
from the top edge of the flaw and typically have a phase that is opposite (i.e. 180°
different) to the lateral wave, but similar to the backwall echo. In addition, the backwall
echo may be weakened or replaced. The main causes of false identification of this type of
flaw are due to the responses from large pores or slag lines lying close to the backwall.
Where there is a doubt, a cross-check with pulse-echo ultrasonics is recommended, to
determine if a corner echo signal can be detected.

Embedded planar flaws (with the exception of laminar flaws) will appear on a D-scan
image as two tip diffracted indications lying one above the other in the through-wall
direction. These indications will usually show a distinct phase change between the upper
(similar phase to backwall-echo) and lower (similar phase to lateral wave) tips. If, from a
D-scan, this type of flaw does not lie close to the centre of the probe separation, there will
be some error in the estimation of through-wall height and position within the material
thickness. For this reason, it is recommended that B-scans are performed, such that the
flaw lies symmetrically between the probes. Common causes of incorrect interpretation of
D-scan images arise from (i) volumetric flaws, which in certain circumstances can appear
planar in nature, and (ii) chance associations of fine slag lines. Volumetric flaws may be
identified by the relative sizes of the responses from the upper and lower extremities of the
flaw. Parallel fine slag lines at different depths within the material do not show the evidence
of phase reversals that are often observed from the tip responses of an embedded planar
flaw. Where doubt exists, it is recommended that pulse-echo ultrasonics be used to look
for a strong spectrally reflected signal from the implied planar flaw.

Depending on the agreed inspection code or procedure, embedded planar flaws that lie
close to the scan or far (backwall) surfaces may be treated as if they are surface breaking,
requiring only the lower (for near scan surface flaws) or upper (for near far surface flaws)
extremities to be determined.

6.3.2. Volumetric Flaws

As previously mentioned, these flaw types can be identified by the differences in the
echoes from the upper and lower extremities of the flaw. Although the phase change
associated with embedded planar flaws may be evident, the response from the upper
extremity of the flaw will be considerably greater in amplitude than that from the lower
one.

If in doubt such indications should be initially considered as planar until further data are
available.

Page 8 PRAD No: 7311.01/98/975.03

Copyright © TWI 1998
INTERPRETATION OF ULTRASONIC TIME-OF-FLIGHT DIFFRACTION DATA – A GUIDE

6.3.3. Thread-Like Flaws

These flaws appear on D-scan images as an apparent upper tip signal (phase
approximately the same as the backwall echo) with no corresponding lower tip signal.
Broken lines of slag are often evident as an intermittent thread-like indication on a D-scan
image.

6.3.4. Point Flaws

These flaws are also commonly referred to as point reflectors. They show similar
characteristics to thread-like flaws, but have no measurable length (i.e. less than half of the
probe diameter). Concentrations (or clusters) of these flaws are often referred to as
'multiple' flaws, and are usually treated as a single 'boxed area' for interpretation purposes.

6.4. OTHER FEATURES SEEN ON TOFD IMAGES

In addition to the above flaw categories, the backwall echo signal can provide valuable
information to the interpreter. On a D-scan image, for example, where misalignment
occurs, the backwall echo will usually split into two components. In this case the TOFD
technique can provide a measure of the degree of offset around a weld. It is important that
features such as a split backwall echo signal are not misinterpreted as being due to flaws
breaking the far surface. Therefore the following should be considered:

i. Is it as a result of the component design?

ii. Is it a result of the manufacturing process?

iii. Are both signals the same width and of similar strength? This is unlikely if a crack tip
is causing one of the echoes.

When a compression wave interacts with a flaw or the far surface of the specimen some
mode conversion occurs. If the timebase of the RF A-scan extends past the backwall
echo response, it is often possible to detect these mode converted or direct shear wave
responses from large embedded or far surface flaws. Such responses can provide a useful
pointer for the interpreter, particularly when examining D-scan images. However, the
interpreter should also be aware that if the PCS used was in excess of 2t (where t is the
thickness of the sample), shear wave responses from flaws can occur before the
compression wave backwall echo. Such shear wave responses are usually very similar in
shape and size to the compression wave response, but are of considerably lower intensity
on the image, i.e. the signals are lower in amplitude. It is obviously important that these
shear wave responses are recognised and not reported as 'real' flaws.

PRAD No: 7311.01/98/975.03 Page 9

Copyright © TWI 1998
INTERPRETATION OF ULTRASONIC TIME-OF-FLIGHT DIFFRACTION DATA – A GUIDE

Appendix A contains some examples of grey scale TOFD D-scan images showing some
real responses from point, threadlike and planar flaws (near surface and embedded).

7. MEASUREMENT OF FLAW SIZE AND POSITION

7.1. LOCATION AND LENGTH

In general, it is sufficiently accurate to assume that the flaw is located mid-way between
the two probes. The time-of-flight of signals from a flaw can be used to estimate its
position. However, information from both B and D-scans is required to locate it in all three
planes accurately.

Most TOFD equipment has provision to allow the interpreter to identify features such as
point flaws by use of a mathematically generated hyperbolic cursor that models the
response from a point-like flaw. Since the shape of the diffraction arcs seen on TOFD
images depends on the through-wall position of the flaw, probe centre separation and
direction of probe movement, it is pre-determined for any given depth. The software
models the shape of the arc and provides a cursor which may be overlaid on TOFD
images and matched to arc-shaped indications to aid in both their characterisation and
sizing. For example, the cursor may be matched with the two semi-arcs (or 'tails') at each
end of the response from a thread-like flaw on a D-scan image, effectively allowing the
interpreter to remove the effects of beam divergence and obtain an estimate of the flaw's
length. In addition, the use of this type of cursor usually provides a direct through-wall
position readout from non-linear TOFD images.

It should be noted that the use of the hyperbolic cursor is best when the length profile of a
flaw is relatively constant. Where a flaw has a rounded or sloping profile, particularly at its
ends, the use of the hyperbolic cursor may underestimate its length.

7.2. LIGAMENT AND THROUGH-WALL HEIGHT

The positions within the material of the top and bottom indications from an embedded flaw
directly define its through-wall height. Signals arising from the top of a flaw are usually
defined as 'reflected' signals and will have a phase similar to that of the backwall echo.
Signals arising from the bottom of flaws are defined as 'transmitted' signals and will have a
phase similar to that of the lateral wave.

In estimating the through-wall extent of a flaw, this phase change should be taken into
account. It is normal practice to use a point on the lateral wave response as a reference
for time difference measurements.

The actual choice of reference point on a pulse is not so important, as long as it is used
consistently. For example, if the first positive peak in the reflected pulse is chosen for
measurement purposes, the first negative peak should be used on the transmitted pulse. As

Page 10 PRAD No: 7311.01/98/975.03

Copyright © TWI 1998
INTERPRETATION OF ULTRASONIC TIME-OF-FLIGHT DIFFRACTION DATA – A GUIDE

the backwall echo response is usually much larger than other responses on a TOFD
image, it may be difficult to use the first peak for measurement purposes (since it often
exceeds full screen height). It is therefore recommended that, where possible, use is made
of the zero-crossing point at which pulses initially break the timebase. In addition, use of
zero-crossing points eliminates a possible source of error due to the often slightly lower
frequency (and distorted) lateral wave response.

The through-wall height of a scan surface breaking flaw is given by the distance between
the scan surface (i.e. lateral wave) and lower flaw extremity. The ligament in this case is
zero.

The through-wall height of a far surface breaking flaw is given by the distance between the
upper flaw extremity and far surface (i.e. backwall echo).

To minimise errors, where measurements are made from original data, the zero time
reference should be taken from the lateral wave directly above the flaw of interest.

On most TOFD systems the ligament and through-wall height of a flaw can be estimated in
at least three ways:

i. By use of a mathematically generated hyperbolic cursor that models the response

from a point-like flaw at different depths within the material (see Section 6.1).

ii. By time-to-depth linearisation of the raw data. When this is done, the B or D-scan
image is re-plotted so that the depth axis is linear, i.e. in mm rather than µsec.

iii. By calculation from timing measurements made from the original RF A-scan signals.
In this case, use is made of the following equation:

d = 0.5 TD2 C 2 + 2TD CS

Where:

d = depth below scan surface

TD = time difference between the lateral wave and flaw response

C = velocity of compression waves for material under test

S = probe centre separation (or PCS).

7.3. FLAW M EASUREMENT ERRORS

PRAD No: 7311.01/98/975.03 Page 11

Copyright © TWI 1998
INTERPRETATION OF ULTRASONIC TIME-OF-FLIGHT DIFFRACTION DATA – A GUIDE

Most of the errors associated with the TOFD technique can be minimised by applying
carefully prepared calibration and data acquisition procedures. The potential error due to
the lateral uncertainty (i.e. in the 'Y' plane) in the position of a flaw seen on a D-scan
image limits the precision of this type of scan. The additional data available from a B-scan
through a flaw can be used both to eliminate this source of error and to provide valuable
information on the possible association of indications seen on a D-scan image. TOFD B-
scans should always be acquired across flaws, to ensure maximum positioning and sizing
accuracy, whether they have been detected as a result of TOFD detection scans or a
pulse-echo examination. B-scan data can also provide the following:

i. The ability to measure flaw ligament and lateral 'Y' position accurately at the point
where the probes are positioned symmetrically about the flaw.

ii. The ability to measure the extent of far wall features such as misalignment or
counterbores.

iii. The ability to determine whether two apparent flaw tip signals are associated with
one another, and, if they are, the orientation of the flaw. For example, the tip
responses may be displaced in the 'Y' plane, thereby suggesting a lack of sidewall
fusion flaw.

iv. The ability to determine if two flaws are present that have similar 'X' position but
different 'Y' positions. On a D-scan image this situation is impossible to resolve.

v. The ability to determine if a flaw has width (in the 'Y' plane) or not. Planar
embedded flaws will tend to show a pair of sharp arcs from the tips, whereas a
volumetric flaw may often show evidence of a plateau between the tails of the arcs,
indicating that the flaw has some width.

vi. The ability to resolve near scan surface flaws from the lateral wave response.

Note that particularly long flaws may need more than one B-scan to determine the
minimum ligament accurately and to verify the maximum through-wall extent. Performing
several B-scans has the added advantage in that any variation in the flaw's lateral 'Y'
position along its length can be ascertained.

8. TOFD DATA PROCESSING AIDS

8.1. GENERAL

The following sections describe additional data processing aids that may be available to
the interpreter. In TWI's experience, where a hyperbolic cursor is available, the use of
these aids is generally unnecessary.

Page 12 PRAD No: 7311.01/98/975.03

Copyright © TWI 1998
INTERPRETATION OF ULTRASONIC TIME-OF-FLIGHT DIFFRACTION DATA – A GUIDE

8.2. LINEARISATION

An alternative to using the hyperbolic cursor for depth measurements on non-linear original
data is to re-display the image with a linear depth axis. Linearised images are
characterised by a very wide (i.e. stretched) lateral wave response. Depending on the
software, the hyperbolic cursor is often not selectable for use with linearised data.

8.3. LATERAL WAVE STRAIGHTENING/R EMOVAL

Some advanced TOFD systems offer the possibility of straightening out the lateral wave
response. Owing to coupling variations, the lateral wave on typical TOFD images is rarely
perfectly straight. Each RF A-scan can be time shifted to align the lateral wave responses,
thereby cosmetically improving the appearance of the image. The advantage of
straightening the lateral wave is that, when making through-wall measurements the zero
time reference is constant for the whole image. However, such processing usually only
works satisfactorily on original data which are reasonably straight. Another related
processing option is lateral wave removal which, when it works well, can aid with the
identification of near scan surface flaws.

8.4. SYNTHETIC APERTURE FOCUSING TECHNIQUE (SAFT)

The conventional 6dB drop method for determining the length of flaws that are larger than
the beam width, relies on locating the positions where the beam is 50% off the ends of the
flaw. This is effectively what the interpreter does when using the hyperbolic cursor on
TOFD data.

However, for smaller flaws and for flaws at longer ranges (where the beam width is
significant due to divergence), the cursor approach may not give sufficient accuracy for
some applications. For example, length measurements made on flaws that are smaller than
the beam divergence at the appropriate range will be consistently overestimated.

The SAFT technique1 is a data processing method that can be used to improve the length
sizing of flaws. The principle of SAFT is that the RF data from a number of probe
positions are combined with the correct phases to model the effects of a single probe
having a large aperture which can be focused accurately at all depths. Ultrasonic data
acquired from line scans are combined to give a theoretical lateral resolution of half the
transducer width. The apparent effect of SAFT processing of a particular flaw depends on
the ratio of R/N, where R is the range (or distance) of the flaw from the probe and N is the
nearfield distance for the probe in use.

For example, for a small flaw where R/N ≈ 2 there will be very little difference between
the raw and processed flaw lengths. However, when R/N ≥ 4, the SAFT process will
provide an image from which a significantly more accurate length measurement can be
made.

PRAD No: 7311.01/98/975.03 Page 13

Copyright © TWI 1998
INTERPRETATION OF ULTRASONIC TIME-OF-FLIGHT DIFFRACTION DATA – A GUIDE

9. EFFECTS OF SPECIMEN GEOMETRY

Where the geometry is not plane (i.e. flat), the same basic approach to data interpretation
applies. However, the interpreter should be aware of the effects of the geometry on
measurement accuracy.

For example, where scans have been performed on a convex surface (e.g. a scan of a
longitudinal seam weld in a pipe) the apparent material thickness on a TOFD image will be
less than the actual thickness. Furthermore, for a given probe separation, the difference
will depend on the radius of the convex surface: the smaller the radius the greater the
difference in apparent thickness. This effect is due to the fact that the lateral wave does not
travel around the circumference of the pipe but by the shortest path between the probes,
whilst the backwall echo does originate from the true location (i.e. the reflection from the
bore of the pipe).

Scans performed on concave surfaces (e.g. a scan of a seam weld along the internal bore
of a pipe) may show an apparent material thickness on the TOFD image which is greater
than the actual wall thickness.

It is therefore important that the interpreter be aware of these effects and, if necessary,
apply a correction to through-wall measurements made from such data. Figures 11 and 12
show the ultrasonic signal paths for convex and concave geometries.

10. REPORTING OF INTERPRETATION RESULTS

Ideally, separate flaw reporting and acceptance criteria should have been specified prior
to inspection. Note that where this is the case, not all reported flaws are necessarily
unacceptable. However, in some cases, the acceptance criteria may also be the basis for
reporting flaws.

For example, reporting criteria such as the following could be specified, with additional
flaw acceptance criteria such as proposed in reference 4.

i. Report all linear (i.e. thread-like) flaws exceeding Xmm in length which have no
measurable through-wall height (i.e. ≤3mm).

ii. Report all flaws having a through-wall height of ≥Ymm, regardless of their length.

iii. Other than point flaws, report any flaws that lie within Zmm of either surface.

iv. Report all clusters of point flaws that exceed n flaws within an area of mmm2 of
projected (D-scan) image.

Page 14 PRAD No: 7311.01/98/975.03

Copyright © TWI 1998
INTERPRETATION OF ULTRASONIC TIME-OF-FLIGHT DIFFRACTION DATA – A GUIDE

X,Y,Z,n,m are dependent on the application and should be agreed between the relevant
parties.

The results of the interpretation should be recorded on an appropriate form, an example of

which is below in Appendix B. To document the interpretation fully, some information on
the equipment and procedure used will usually be required. Most computerised TOFD
systems have provision for all relevant inspection details to be read from the data file. This
information should include:

Date of inspection
Operators name(s)
Procedure reference
Probe types and serial numbers
Probe frequency and shoe angle
Probe centre separation
Reference target used
Filter settings
Material velocity
Thickness of material inspected
Location of scan datum and scan direction
The A-scan collection interval
Digitiser settings used, including number of A-scans averaged

11. CONCLUDING REMARKS

The following concluding remarks are made with regard to the interpretation of TOFD
data. In addition, it is TWI's opinion that TOFD should not be used in isolation for the
detection and assessment of flaws. It is recommended that complementary NDT
techniques are also employed where there is concern over the initial detection of particular
flaws (e.g. surface or near-surface flaws), or where doubt exists over the interpretation of
particular indications seen on TOFD images. As a consequence, where it is appropriate,
TWI's preferred approach is to use TOFD to complement pulse-echo ultrasonic
inspection.

• It is recommended that companies develop their own specific procedures for the
interpretation of TOFD data based on the guidelines provided in this document.

• Where TOFD inspection is used exclusively, a full evaluation of flaws is only possible
if both D-scan and B-scan images are available.

• Where a weld cap causes an obstruction to B-scans, consideration should be given to

local removal of the cap. Note that this is only possible in situations where removal of
the weld cap does not cause strength matching problems.

PRAD No: 7311.01/98/975.03 Page 15

Copyright © TWI 1998
INTERPRETATION OF ULTRASONIC TIME-OF-FLIGHT DIFFRACTION DATA – A GUIDE

12. ACKNOWLEDGEMENT

The work was carried out within the Core Research Programme of TWI which is funded
by the Industrial Members of TWI.

13. REFERENCES

1 Charlesworth J P and Temple J A G: 'Engineering applications of ultrasonic time-of-

flight diffraction'. Research Studies Press Ltd, 1989.

2 BS 7706:1993: 'Guide to calibration and setting-up of the ultrasonic time-of-flight

diffraction (TOFD) technique for the detection, location and sizing of flaws'.

3 prEN 583-6 'NDT - Ultrasonic examination - time-of-flight diffraction technique as

a method for defect detection and sizing'.

4 CEN TC 121/SC 5B/WG 2 N99: 'Acceptance criteria for the time-of-flight

diffraction methods'. November 1997.

5 Jessop T J, Mudge P J, Charlesworth J P, Silk M G, Clare A B, Short K A and

Bowker K J: 'Size measurement and characterisation of weld defects by ultrasonic testing
– Part 1 Non-planar defects in ferritic steels'. Publ The Welding Institute, Abington,
Cambridge, 1979.

6 Jessop T J, Mudge P J, Harrison J D, Charlesworth J P, Aldridge E E, Silk MG,

Coffey J M, Bowker K J, Wrigley J M and Denby D: 'Size measurement and
characterisation of weld defects by ultrasonic testing – Part 2 Planar defects in ferritic
steels'. Publ The Welding Institute, Abington, Cambridge, 1982.

Page 16 PRAD No: 7311.01/98/975.03

Copyright © TWI 1998
 Transmitter Scan Surface Receiver

Back-wall Echo

Far Surface

Fig. 1 Basic two probe test configuration

Tip Diffracted Start of Back-wall

Signals Echo

/ \ \ ^ n #

----+ T/me

/ ’ Flaw ‘Bottom’
Lateral Wav/e Flaw ‘Ton’

Fig. 2 TypIcal RF A- Scan showing responses from an embedded planar flaw

 b T+
Direction of LW
Time
movement _ _ Rx
(depth)
Y+
A Weld
-+X

v
Y-
- TX
Probes BWE

Image produced looking in the direction of arrow A

Fig. 3 D-scan: Probe movement and image orientation

Direction of Weld +
LW
movement Time
(depth)

oAL3
Tl+I;

BWE

Image produced looking in direction of arrow A

4
A
Fig. 4 B-scan: Probe movement and image orientation
 a) Two thread-like flaws, well separated in the Y dlrection

b) Two thread-like flaws, in close proximity In the Y direction

c) One planar flaw

Fig. 5 Weld cross sections illustrating situations where D-scan images may show similar flaw responses

Fig. 6 Weld cross sectton illustrating situation where a D-scan image may only show an apparent
single flaw response due to obscuration
 LW LW

BWE
BWE

Fig. 7 Planar flaw (measurable length and height)

LW

BWE

D-scan B-scan

Fig. 8 Volumetric flaw (measurable length, height and width)

 LW

BWE BWE

Fig. 9 Threadlike flaw (measurable length only)

LW
LW

BWE
BWE
B-scan

Fig. 10 Point flaw (no measurable length, height or width)

 Transmitter Receiver

Diffracted Energy

Fig. II Ultrasonic signal paths for convex geometries

Diffracted Energy

Back-wall Echo

Fig. 12 Ultrasonic signal paths for concave geometries

 APPENDIX A

Example TOFD Images

__I_ .................. .....--- I._..................................-......-
-
{.?-f;::;Qi
MM
; 1:i.!.Kl 1%.
. . O:j 163;kj