Professional Documents
Culture Documents
TWI 581-CTOD Testing Update (1996)
TWI 581-CTOD Testing Update (1996)
By H G Pisarski
No embargo
Electronic copyright
in this document as follows:
Copyright 2000, TWI Ltd
8 TWI 1996
UPDATE ON FRACTURE TOUGHNESS TEST METHODS FOR
WELDED JOINTS
Bv: H G Pisarski
EXECUTIVE SUMMARY
Background
Current standard fracture toughness test methods are based on testing specimens
taken from plain, homogeneous materials. They do not consider the distortion
curvature and residual stresses often present when testing specimens taken from
welded product forms. Tests on weld metals and heat affected zones require special
procedures to evaluate fracture toughness. Furthermore, the differences in tensile
properties between the notched region and adjacent material on the deformation
behaviour of the specimen are ignored by standard CLOD and J estimation
procedures. These factors can affect the result and the outcome of defect assessment
analyses and, therefore, need to be recognised when designing the appropriate test
to measure fracture toughness.
Main Conclusions
The review has identified the major factors that need to be considered when
measuring the fracture toughness of weldments. Some of the limitations of current
standard methods are described. These include the lack of standard testing
procedures to cater for specimens containing shallow cracks, the uncertainty in the
CLOD and J estimation procedures where yield strength mismatch is present and
their applicability to specimens notched into the HAZ, the difficulties in meeting
fatigue crack front straightness requirements, and the lack of guidance on the
assessment of pop-in which may occur in weldments.
Recommendations
By: H G Pisarski
1. INTRODUCTION
In this report, some test methods used to measure fracture toughness in welded steel
materials are reviewed. These are based on fracture mechanics principles, since this
information is a direct input to an engineering critical assessment (ECA) which
would be used to assess the significance of the results with respect to defect
tolerance and structural integrity. For convenience, the fracture mechanics test
procedures described relate to the Crack Tip Opening Displacement or CTOD test.
However, the procedures are also generally applicable to Iz, and J integral tests. It
may be noted that the design of the test specimens for I(I,, J and CTOD determina-
tion are essentially the same. The determination of Iz, requires failure under linear
elastic conditions. Iz, quantifies the critical amplitude of the stress singularity ahead
of the crack tip at fracture. When fracture occurs under general yielding or elastic
plastic conditions it is appropriate to measure CTOD or J integral. CTOD is a
measure of the local strain at the crack tip, whilst J is a line integral which
quantifies the amplitude of the stress and strain at the crack tip. Fracture may occur
at a critical value of CTOD or J. For small scale yielding and brittle fracture, all
three fracture parameters are related through the following expression:
Jc = Ki(l-v2)/E = ma*6
where
Jc - critical value of J
6 - critical value of CTOD
Y - Poissons ratio
The paper begins with test requirements generally applicable to fracture toughness
measurement on both weld metal and heat affected zones (HAZ). The specific
requirements for testing weld metals are described next. HA2 testing is more
involved and more controversial than weld metal testing, so this is dealt with
afterwards. However, it should be pointed out that some of the following sections
provide a reiteration and update on earlier TWI recommendations. Furthermore,
ASTM and BSI committees are currently in the process of drafting weldment
fracture toughness standards and have used previous TWI recommendations as a
basis. The report also highlights some of the unresolved problems with testing of
weldments.
An essential requirement for tests on welded joints is that the welds should be fully
representative of those in the service structure of interest. This is important when
the results are to be used in an ECA, or, where the specification requires a target
fracture toughness which has been derived from an ECA. Experience indicates that
the fracture toughness of weld metals and I-IAZs may be critically dependent on
factors such as:
a) welding process(es)/consumable(s),
b) base metal composition,
Cl joint thickness,
4 preheat/inter-passtemperature,
4 heat input,
f) welding position and detailed welding procedure (e.g. weave, arc length),
joint configuration,
i; restraint against lateral and rotational distortion,
9 post weld heat treatments (PWHT) including those for hydrogen release,
3 other post weld treatments (e.g. plastic deformation, irradiation),
k) the time between welding and testing,
1) environment.
The presence of diffusible hydrogen in the weld can reduce fracture toughness of
the weld metal and HAZ, especially when the rate of straining is low. Hydrogen
release heat treatments are sometimes carried out on welded panels prior to testing
to ensure that they contain similar levels of hydrogen to the structure of interest.
Appropriate heat treatment temperatures and times can be found using the diffusion
curves in Ref.1. Where the time between the completion of welding and the
component entering service is short (e.g. during pipe lay operations), heat treatment
would not be appropriate.
Current fracture testing standards (2,3) permit use of full section thickness single
edge notch bend specimens of rectangular and square section designs, Fig.1 and
Fig.2. However, these standards are specifically designed for testing plain
homogeneous material, not weldments. Choice of specimen geometry and notch
location can critically affect the outcome of the test, and, consequently have to be
chosen with care. The choice of specimen type and notch position is dependent on
the objective of the test. This can be generally categorised ‘general assessment’ or
‘specific assessment’ of fracture toughness (4).
Tests for general assessmentsare used mainly for weld selection and weld procedure
qualification, where nothing may be known about the existing or assumed cracks in
a structure. For these tests, the specimens are designed so that the notch size,
position and orientation, and overall specimen geometry are most likely to give a
lower bound value of fracture toughness. Two aspects need to be considered,
maximizing triaxial constraint and maximizing the volume of the target
microstructure located at the crack tip. These requirements can generally be met by
testing a specimen having a thickness equal to the joint thickness, which is deeply
cracked (a/W > 0.45) and a rectangular or compact design. The target microstructure
is often selected to be that suspected to have the least toughness. Generally, these
will be the unrefined regions of the weld metal and HAZ. As will be discussed later,
sometimes special welding procedures and weld preparations may be necessary to
test target microstructures, especially when assessing HAZ fracture toughness.
Tests for specific assessments of fracture toughness are used when it is necessary
to measure the fracture toughness associated with a specific crack in a structure (e.g.
fatigue cracks where the crack path can be predicted, analysis of failures). In these
instances the crack in the specimen should simulate the size, position and orientation
of the crack in the structure, and the specimen thickness should be equal to the
thickness of the structure. Often the structural cracks are best represented by surface
breaking cracks in square section geometry specimens.
An international research project has recently been completed to help remedy this
situation (5). This specifically addressed testing requirements for shallow cracked
specimens with a/W down to 0.05. Although the research project was on parent
material, the techniques have been extended by TWI and EWI to testing weld metals
and HAZs for specific applications. These techniques have been especially useful
when assessing the significance of shallow weld root defects in single sided welds
and defects located in the HAZ at weld toes. It is expected that the research work
on shallow cracked specimens will be recognized by code making bodies so that
standards may be changed soon. However, it should be noted that procedures for
testing shallow cracked specimens are based on measuring crack mouth opening
displacement (CMOD). Unless integral knife edges are machined into the notch,
which is impractical for very shallow notches, a special double clip gauge
arrangement needs to be used to infer CMOD. Further discussion on this subject is
made in Section 8.3.
Currently, there are no national standards that describe fracture toughness testing of
weldments. However, ASTM and BSI committees are drafting standards at present.
TWI recommends that geometries in the present standards, which are for parent
material, are employed but with relaxed dimensional tolerances, as shown in Fig.1
and 2. These tolerances are based upon practical experience of testing bend
specimens, and a tacit acceptance of the effect they have on CTOD measurements
(4).
When large distortions or curvature are involved (see Fig.3), bend specimens may
be straightened by bending at points remote from the crack. This is sometimes
referred to as ‘gull winging’ and is illustrated in Fig.4. It is often applied to
specimens taken from small diameter pipe when a through pipe wall thickness notch
is aligned parallel to the longitudinal axis of the pipe. A straightness requirement of
2S%W is recommended for the specimen sides to permit curvature of the specimen.
It is not recommended that these tolerances are applied to compact specimens, since
there is relatively little experience in testing full thickness welded product forms
with these types of specimen. If compact specimens are required from welds with
Local machining, grinding and etching may be necessary to facilitate locating the
crack in the target microstructure. Also, local machining of the specimensides ahead
of the machined notch is often necessary to achieve a flat surface for local
compression (described in the next section) prior to fatigue precracking. These
necessary treatments may cause small local variations in specimen dimensions and
must be allowed for when calculating the fracture toughness parameter.
Specimens taken from joints in the as welded or partially stress relieved condition
contain residual stresses which may result in uneven fatigue crack growth and
invalid crack shapes. This is commonly found with through thickness cracked
specimens taken from multipass welds. To achieve uniform fatigue crack shapes,
which meet the requirements of the test standard, it is recommended that the
residual stresses are mechanically stress relieved prior to fatigue precracking using
the technique called ‘local compression’. With local compression a plastic strain of
approximately 1% of the specimen thickness, B, is applied across the ligament in
front of the machined notch, as illustrated in Fig.5 Often 0.5% strain is applied to
each side of the specimen, and multiple indents may be necessary when the ligament
is large. In these cases experience indicates more favourable crack front shapes are
obtained if the indent covering the machined notched tip is made last.
Alternative methods for obtaining straight fatigue cracks have been proposed, such
as reverse bending and high stress ratio techniques. TWI experience is that they are
less reliable than local compression. In addition, use of high R ratio techniques may
be contrary to the requirements of the testing standard. Although the technique may
produce a straight fatigue crack it will not result in a reduction of residual stresses
ahead of the crack tip (6). The presence of the residual stresses may also influence
specimen compliance and the test result (6).
It is, of course, possible that local compression may reduce the fracture toughness
of the material, especially if it is susceptible to strain ageing embrittlement.
However, limited experimental evidence indicates that it has little effect. It may be
noted that conventional tests to assess strain ageing use Charpy specimens which
have been strained 5%; five times the strain used for local compression. In general,
it is considered preferable to accept the (small) risk of measuring a reduced fracture
toughness as a result of local compression, rather than accept the certainty of
obtaining an invalid crack shape and invalid test if local compression is not carried
out. A grossly invalid crack shape may cause fracture toughness to be overestimated.
When possible, the use of the shortest fatigue crack length allowable by the
standards is recommended. This is because with less fatigue crack growth, there is
a smaller chance of obtaining a non-uniform crack shape (e.g. bowing), and the
problem of the crack deviating out of the target microstructure is reduced.
A specimen notched parallel to the weld length and perpendicular to the plate
surfaces is generally recommended, as shown in Fig.6. This through thickness notch
plane is widely used for tests on weld metals because it coincides with both the
weakest planes of coarse columnar solidification structures and the greatest
concentration of weak grain boundary segregates.
In narrow gap multipass welds it is important to locate the notch in the centre of the
pass that has the greatest through thickness depth of unrefined microstructure. In a
two per layer narrow gap weld, the notch should be positioned as shown in Fig.7,
and not through the centre of the weld.
With joints containing asymmetric double sided welds or welds made from one side
only, the region of suspected lowest fracture toughness may be located towards the
edge of the specimen. (A region of locally higher hardness from a hardness survey
may be an indication of lower toughness). Since this may not be located in the
region of highest constraint in the fracture toughness specimen, it may be necessary
to use a surface notch in order to maximize the material sampled by the crack. In
this case a specific assessment is required and, in addition, it will be necessary to
use a non-standard shallow crack specimen if a/W < 0.45. With surface notches, the
fatigue crack tip should be located just inside the columnar microstructure of an
unrefined weld bead. To confirm that the target microstructure is indeed tested by
the fatigue crack tip, it may be necessary to carry out post-test sectioning and
metallurgical analysis. The method for doing this is similar in principle to that used
for HA2 tests and is described in Section 9.
The relatively narrow width of the HA2 compared with that of the weld and
specimen thickness, can make accurate placing of the crack difficult. For example,
if surface notches are used, the fatigue crack could grow through the I-IA& so that
the tip is in a different microstructure. This is less of a problem with through
thickness notched specimens, since some part of the crack front will be positioned
in the HAZ. However, in a multipass weld this may not be the region of lowest
Although weld grooves containing a vertical edge are often used to assess the
fracture toughness of the grain coarsened HA2 adjacent to the fusion boundary, they
can be also used to assess the toughness of other HA2 regions that could have low
toughness, for example, the intercritical/subcritical HAL
In choosing the weld metal for HA2 tests, consideration should be given to the
degree of yield strength mismatch between the parent material and weld metal.
Ideally, the weld metal used should be the same as that in the structural weld being
assessed. If this is not specified, significant yield strength overmatch should be
avoided, as there is a risk that HAZ toughness will be reduced due to higher stress
triaxiality being generated in the lower strength material (13). An example of the
effect that weld metal yield strength over matching can have on HA2 fracture
toughness is illustrated in Fig.10.
8. TEST PROCEDURE
If the fatigue crack tip is located in the fusion boundary, or in the transformed I-IA&
the choice of yield strength is more problematic. For most welds it is not possible
to measure HAZ yield strength directly. A lower bound estimate of fracture
toughness is obtained if the higher of the parent material or weld metal yield
strength is used to calculate the elastic component of CLOD. However, this may be
unnecessarily conservative if extensive yielding precedes failure, since the
development of CTOD will be controlled by the lower strength material. As a
compromise it has been suggested that the average of the yield strength of the parent
material and weld metal are used for the calculation of CTOD (4). It may be noted
that when conducting an assessment of the significance of flaws in the HA&, it is
recommended (14) to use the yield strength of the lower strength material adjacent
to the HAZ.
Historically, it has been usual practice to calculate J and CTOD for specimens
notched into weldments using the formulae intended for plain, homogeneous
materials such as given in Ref.2 and 3. The weldment specimen has been assumed
to be homogeneous and of the same tensile properties as the region into which it is
notched. The possible effects of yield strength mismatch between the notched and
adjacent regions on specimen deformation behaviour are ignored. The accuracy of
the J and CTOD estimation procedures can therefore be called into question.
(Despite this, defect assessmentprocedures based on BS PD6493 (14) have not been
shown to be unsafe 1516). Nevertheless, recent finite element analysis studies have
been able to investigate the accuracy of previous assumptions. Analyses conducted
on specimens notched into the weld metal centreline have shown that the plain
material CLOD and J estimation procedures are accurate (to within 10%) provided
that (17,lS):
iii) the degree of weld metal mismatch (a,&,,) does not exceed LSO%, and
iv) the weld width (h) to untracked ligament (W-a) ratio is x0.2 or rl
Examples of the accuracy of the standard CLOD and J estimation procedures for
centrally located weld metal cracks (17) are shown in Fig.11 and 12. However, there
is still uncertainty in the validity of the standard CTOD and J estimation procedures
when the notch is offset from the weld centreline, and where notches are located in
the HAZ. Nevertheless, it may be guessed that since validation of flaw assessment
procedures for HA2 cracks using wide plate specimens have not highlighted any
particular problems (l&16) (the assessment procedures are acceptably conservative
when using CTOD data generated from HAZ specimens), the current estimation
procedures (SKID) are reasonable. In such studies standard (deeply notched) CTOD
specimens were used and the degree of yield strength mismatch was approximately
0 to +50% overmatch. In addition, comparisons of CLOD estimated using standard
estimation procedures (to BS 5762) with a more direct measure of CIYOD using the
a5 procedure (CTOD measured near the crack tip over a Smm gauge length) indicate
excellent agreement when deeply notched (a/W = 0.5) bend specimens are used to
test HAZ and weld metal (19). Consequently, evidence indicates that for deeply
notched specimens testing HAZ and weld metal, the standard CTOD (and J)
estimation procedures are satisfactory.
Further studies on the subject are currently underway (20). However, for bend
specimens, J estimation procedures based on measurement of crack mouth opening
displacement (CMOD) rather than load line displacement are likely to be recom-
mended, since they produce more reliable results (17,21).
Current standards (2,3) do not permit the testing of specimens containing shallow
cracks, i.e. a/W < 0.45. For specific assessments of fracture toughness this
restriction can present problems, since the target microstructure may be only present
near to the surface of the specimens (e.g. the weld root region of a single sided
weld). Furthermore, the application of standard J and CLOD estimation procedures
to shallow notched specimen can lead to significant errors (17,21).
A,,, is the area under the plastic component of CMOD (measured using a special
double clip gauge arrangement mounted above the notch) versus applied force curve.
Preliminary work indicates that this equation leads to less than 10% error in J for
yield strength mismatch of up to *25%, and weld width to ligament ratios greater
than 1 (17).
8.4. Pop-in
Pop-in is a term used to describe a discontinuity in the applied force versus clip
gauge displacement curve which is characterized by a sudden increase in displace-
ment, and generally, a decrease in force. Subsequently, the force and displacement
increase smoothly to above their respective values at pop-in. Often, pop-in can be
attributed to an arrested brittle crack. (Apparent pop-ins can be caused by splits,
delaminations, linking up of weld defects or multiplanar fatigue cracks, breaking of
ice particles around rollers during sub-zero temperature tests, or electrical
interference. In these instances, the pop-in has no significance with respect to
characterizing the fracture toughness of the material).
In the BSI and ASTM testing standards (2,3), criteria are defined permitting small
pop-ins to be ignored if they cause an increase in specimen notch opening
compliance of less than a 5%. In the ASTM standard, there is an additional criterion
that the maximum increase in crack length is less than 4% of the original uncracked
ligament. These criteria have been set with plain, homogeneous materials in mind.
When testing weld metal and HAZ, the size of the arrested crack associated with the
pop-in is often related to the size of a local brittle zone (LBZ) sampled by the crack
front. For example, when surface notching a HAZ which contains low toughness
grain coarsened material (often termed a LBZ), a pop-in can be caused by initiation
in the LBZ followed by arrest in tougher, grain refined IIAZ. The size of the pop-in
and arrested brittle crack can be altered by small changes in crack length. Clearly,
it would be incorrect to ignore a pop-in if this was caused by a fortuitous
positioning of the crack tip. Only by careful sectioning of the specimen at the
fracture initiation site and metallurgical examination, possibly followed by further
testing using a more appropriate crack tip location, can it be established whether the
toughness measured at pop-in is relevant or not. Consequently, it is recommended
that when testing weldments, pop-ins classed as ‘not significant’ according to Ref.2
are assessedby metallurgical examination to decide their actual significance.
It has been suggested that pop-in events can be used to estimate the arrest
toughness of the material. The procedure requires a series of specimens with pop-
9. POST-TESTSECTIONING
A number of codes require mapping of the microstructures present along the fatigue
crack front in through thickness notched HAZ specimens. A typical mapping
procedure could involve recording the lengths of weld metal, HAZ with grain sizes
>50pm, 20-50pm and < 2Opm present along the fatigue crack front. Codes such as
API RP 22 (10) also require that, the crack front should sample at least 15% grain
coarsened HAZ for a valid test. The reasoning behind this requirement is that if
15%, or more, grain coarsened HAZ is sampled, then the chances of low toughness
microstructure (LBZ) being found are significantly increased and the lower bound
toughness is relatively independent of further increases in the amount of LBZ
sampled, see Fig.9.
Sometimes it is necessary to measure the distance from the crack plane to the lowest
toughness microstructure, if this exceeds a specified distance (typically 0.5mm) it
is unlikely that the microstructure has been adequately sampled and the lower bound
toughness may not have been measured (12). In this situation a retest may be
required.
Fracture mechanics tests on welds and HA2Ls are more involved and difficult to
carry out than tests on plain material. This is partly due to the heterogeneity inherent
in welds. It is also due to the desire to test the actual product form and the
compromises that have to be made to specimen and weld design in order to carry
out a practical test. Furthermore, often difficult choices have to be made in selecting
the weld and specimen designs that best represent the defects that are present or are
postulated to occur in the structural component of interest. Some of these difficulties
are minimised if a distinction can be made between general and specific assessments
of fracture toughness and by ensuring that fracture mechanics testing is closely
related to an ECA. Another benefit of such an approach is that target toughness
requirements are no longer arbitrary but are based on fitness for purpose principles.
This report has proposed methods of selecting the weld and specimen designs and
test procedures which best represent the requirements of a structural integrity
assessment. In addition the report has highlighted some of the limitations in current
weldment testing procedures. These are currently being worked on. They include the
lack of standard testing procedures to cater for specimens containing shallow cracks,
the uncertainty in the CTOD and J estimation procedures where yield strength
mismatch is present and their applicability to specimens notched into the HAZ, the
difficulties in meeting fatigue crack front straightness requirements and the lack of
guidance on the assessment of pop-in in weldments.
11. ACKNOWLEDGEMENTS
This review was carried out within the core research programme of TWI which is
jointly funded by the Information and Manufacturing Technology Division of the
Department of Trade and Industry, and the Industrial Members of TWI.
12. REFERENCES
2 BS 7448: Part 1: 1991: ‘Method for determination of I(I,, critical CLOD and
critical J values of metallic materials’. BSI, London.
18 Joch J, Ainsworth R A and Hyde T H: ‘Limit load and J-estimates for idealised
problems of deeply cracked welded joints in plane-strain bending and tension’.
Fatigue Fract. Engng. Mater. Struct. Vo1.16, No.10, pp.1061-1079, 1993.
19 Ko@ M, Yao S, Schwalbe K-H and Walter F: ‘Effect of crack depth (a/W) on
weld metal fracture toughness’. Welding ‘90, Geesthacht, Germany, October 1990,
iitt Int.