Volume 08 - Mechanical Testing and Evaluation Part 14 pdf

All peel tests have the common characteristic that failure propagates from an initially debonded area. They also
generally involve large displacements/deformations. For these and other reasons, linear elastic stress analysis is
often not well suited to peel tests. The stresses and strains in the peel configuration are complex and seldom
well understood. Test results are generally not given in terms of stress but rather as force per unit length
required to peel the specimen. It is, therefore, generally difficult to compare the results from a peel test with
those from other testing methods.
Because of the large deformations involved in peel tests, the analysis of such geometries is very difficult except
under certain simplifying assumptions (Ref 3, 4, 6, 9, and 10). Some very interesting and informative
observations can be made on the basis of simplifying assumptions and approximations. Indeed, considerable
useful work has been completed using peel tests. The informative work of Gardon (Ref 10) and Kaelble (Ref
11) is noteworthy. The polymer research group at The University of Akron, under the direction of Professor A.
Gent, has been particularly adroit in applying peel techniques and the concepts of fracture mechanics (see the
section “Adhesive Fracture Mechanics Tests” in this article) to obtain critical information and insight into the
behavior of adhesive joints (Ref 12, 13). The peel specimen is, in principle, a very versatile geometry for
obtaining adhesive fracture energy because various combinations of mode I and mode II loadings can be
applied by varying the peel angle (Ref 3). The stress analyses of Adams and Crocombe (Ref 14) have provided
additional insight into the peeling mechanisms. They examined the stress distributions in peel specimens using
elastic large-displacement, finite-element analysis techniques.
References cited in this section
3. G.P. Anderson, S.J. Bennett, and K.L. DeVries, Analysis and Testing of Adhesive Bonds, Academic
Press, 1977
4. A.J. Kinlock, Adhesion and Adhesives, Chapman and Hall, 1987
6. K.L. Mittal, Adhesive Joints, Plenum Press, 1984
7. Adhesives, Annual Book of ASTM Standards, Vol 15.06, ASTM (updated annually)
9. G.P. Anderson and K.L. DeVries, Predicting Strength of Adhesive Joints from Test Results, Int. J.
Fract., Vol 39, 1989, p191–200
10. J.L. Gardon, Peel Adhesion, I. Some Phenomenological Aspects of the Test, J. Appl. Polym. Sci., Vol 7,
1963, p 654
11. D.H. Kaelble, Theory and Analysis of Peel Adhesion: Mechanisms and Mechanics, Trans. Soc. Rheol.,
Vol 3, 1959, p 161
12. A.N. Gent and G.R. Hamed, Peel Mechanics, J. Adhes., Vol 7, 1975, p 91

13. A.N. Gent and G.R. Hamed, J. Appl. Polym. Sci., Vol 21, 1977, p 2817
14. R.D. Adams and A. Crocombe, J. Adhes., Vol 12, 1981, p 127

Testing of Adhesive Joints
K.L. DeVries and Paul Borgmeier, University of Utah

Lap Shear Tests
The most popular test geometry for testing adhesive joints is the lap shear specimen. Its appeal is probably
based on the fact that it closely duplicates the geometry used in many practical joints. These lap joints are
popular for several reasons:
• They facilitate use of larger contact areas than, for example, a butt joint.
• They are easier to make and align than butt joints.
• The adhesive is not exposed to “direct” tensile stresses. Direct tensile stresses are known to have
deleterious effects on adhesives.
Typical lap shear test specimens for which ASTM standards have been written are presented in Fig. 3 (Ref 7).
The specimens shown in this figure conform most closely to ASTM Standards D 1002, D 3163, D 3164, D
3165, and D 3528 for testing adhesives used to bond metals, plastics, and laminates. These represent only a
small sampling of the more than two dozen standards in the Annual Book of ASTM Standards, Volume 15.06,
that relate to shear testing. These other standards range from descriptions of block-type sample configurations
for testing lumber and wood bonding in shear by compression loading, through descriptions of devices to
simultaneously expose samples to lap shear stresses and extremes in temperature. Still others describe apparatus
for exposing lap joints to sustained loads (using springs) to measure long-term creep or time to failure.

Fig. 3 Typical lap shear geometries. (a) ASTM D 1002, D 3163, and D 3164. (b) ASTM D 3165. (c) ASTM
D 3528. Source: Ref 7
The results from lap shear tests are generally reported as the force at failure divided by the bonded area (overlap
area). Such values are listed in a number of reference books and manufacturers' literature for a wide variety of
adhesives. The reference book on types of adhesives (Ref 1) lists typical lap shear strength values for literally
thousands of commercial adhesives. Such tables of “shear strength” values are without doubt of considerable
utility for comparison and other purposes. However, their use also can lead to faulty expectations and

conceptions. Otherwise knowledgeable designers might logically assume from the tabulations that these
average stress values could, in a straightforward manner, be used to design an adhesive joint.
For example, the tabulated shear stress value for a given adhesive from an ASTM D 1002 test might be given as
3000 psi. It might be assumed that this adhesive is to be used to bond two 25 mm (1 in.) wide by 3 mm (0.12
in.) thick 7075-T6 aluminum pieces together to carry a tensile load of 3200 lb with a safety factor of two. First,
the designer must ascertain whether the aluminum pieces can carry such a load. Typically, 7075-T6 aluminum
has a yield strength slightly in excess of 65 ksi for an allowable stress of 32.5 ksi. The pieces in question would
have an allowable load of 4000 lb, which is more than the 3200 lb required in the design. The “straightforward”
method to design the joint would be to assume that the allowable shear strength for the adhesive used in the
joint would be 3000/2 = 1500 psi, suggesting that an overlap of 3200/1500 = 2.13 in. would be sufficient to
support the load. This is, in fact, the approach taught by a variety of otherwise very good texts on material
science and mechanical design. However, doubling the length of a lap joint almost never doubles its load-
carrying capacity, and the increased joint strength is usually much less than doubled. The length of overlap
recommended in ASTM D 1002 is 12.7 mm (0.50 in.). Typically, quadrupling the amount of overlap does not
increase the load at failure by anywhere near a factor of four. For reasons given in the next few paragraphs, it is
likely that it is not even the value of the maximum shear stress that determines the failure of the “lap shear
joint.” As this article reveals, joint failure is more likely determined by the value of secondary induced cleavage
stresses.
The stresses along the bond line of lap specimens are not constant. The bond stress distribution is highly
dependent on the thickness of the adherends and the adhesive as well as the length of overlap. As a
consequence, the load to initiate failure also varies markedly with both the adherend(s) and adhesive-bond
thicknesses. The failure load increases very nearly linearly with width of the overlap but increases in a very
nonlinear manner with length of the overlap. As the load is increased in a lap shear test, the debonding
generally initiates at or near one of the bond terminations. Elastic stress analysis generally indicates that the
stresses are singular at these termination points. Debond initiation in lap shear specimens can perhaps,
therefore, be best characterized in terms of fracture mechanics parameters, which are discussed in the section
“Adhesive Fracture Mechanics Tests” in this article. In addition, it has been demonstrated that for debonds after
initiation, crack propagation is dominated by crack- opening mode displacements (mode I). For this reason and
reasons given in the next couple of paragraphs, the word shear in the test titles and generally reported in test
results may, therefore, be a misnomer.

It has been known for many years that the shear stresses in the bond line of lap specimens are accompanied by
tensile stresses. Many analyses have been completed for lap shear geometries, almost all of which have clearly
demonstrated the presence of induced tensile stresses in so-called lap shear specimens under load. In 1938,
Volkersen (Ref 15) obtained expressions for the stresses in a lap shear joint by considering the differential
displacements of the adherends and neglecting bending. This study was followed in 1944 by the now classical
treatment of Goland and Reissner (Ref 16) who used standard beam theory and strength of materials concepts
to obtain expressions for the joint stresses. Plantema (Ref 17) combined the results of these two earlier
investigations to include shear effects in the system.
Because the stress state of the lap shear joint is so complex and does not lend itself to closed-form solutions, it
is only logical that as numerical methods became available, researchers would apply them to analyze adhesive
joints. Wooley and Carver (Ref 18), for example, used finite-element methods to calculate the joint stresses.
They compared their results with the results obtained by Goland and Reissner and reported very good
agreement. Adams and Peppiatt (Ref 19) used a two- dimensional finite-element code to analyze the stresses in
a standard lap shear joint and also reported good agreement with Goland and Reissner. These authors also
investigated the effect of a spew (triangular adhesive fillet) on the calculated stresses. A nonlinear finite-
element analysis of the single lap joint was completed by Cooper and Sawyer (Ref 20) in 1979.
Anderson and DeVries conducted a linear elastic stress analysis of a typical single lap joint (Ref 21) making use
of plane-strain finite- element computer programs using elements as small as 0.00025 cm (0.0001 in.). They
considered steel (modulus of elasticity, 207 GPa; Poisson's ratio, 0.30) adherends of various thicknesses bonded
with a 0.25 mm (0.01 in.) thick epoxy (modulus of elasticity, 2.76 GPa; Poisson's ratio, 0.34). The overlap
region was taken as 13 mm (0.5 in.) long. The results of these analyses are shown in Fig. 4. Note that both the
shear and tensile stresses are distributed very nonlinearly over the length of the bond region. Reference 21
reports stresses resulting from other adherend thicknesses. As the bond termini is approached, both shear and
normal stresses appear to become singular. Careful analysis in this region suggests that the local mode I stresses
(tensile or crack opening) are significantly higher than mode II stresses (shear). Perhaps even more importantly,
the mode I energy release rate is greater than that for mode II. From these results, it might be concluded that lap
shear specimens fail by mode I crack growth. Therefore, the failure of lap shear specimens is usually governed
by tensile stress rather than shear stresses. This is true for double lap joints as well as single lap joints (Ref 22,
23).

Fig. 4 Bond line tensile and shear stresses in lap shear specimen (adherend thickness = 1.6 mm, or 0.06
in.)
As noted, the end(s) of the overlap on bond termini on lap shear specimens are points of stress concentration
and of large induced tensile stresses. While this closely simulates many practical situations, some have
suggested that for determination of intrinsic adhesive properties, it would be useful if these termini could be
eliminated. ASTM E 229 “Standard Test Method for Shear Strength and Shear Modulus of Structural
Adhesives” is a test designed specifically for this purpose. In this test, the adhesive is applied in the form of a
thin annulus ring bonded between two relatively rigid adherends in circular disc form. Torsion shear forces are
applied to the adhesive through this circular specimen, which produces a peripherally uniform stress
distribution. The maximum stress in the adhesive at failure is taken to represent the shear strength of the
adhesive. By measuring the angle of twist experienced by the adhesive and having knowledge of sample
geometry, it is possible to calculate the strain. A stress- strain curve can then be established from which the
adhesive's effective shear modulus can be determined.
References cited in this section
1. Adhesives, Edition 6, D.A.T.A. Digest International Plastics Selector, 1991
7. Adhesives, Annual Book of ASTM Standards, Vol 15.06, ASTM (updated annually)
15. O. Volkersen, Die Nietraftverteilung in Zugbeanspruchten Nietverblendugen mit Knastaten
Laschenquerschntlen, Luftfahrt forsch., Vol 15, 1938, p 41
16. M. Goland and E. Reissner, The Stresses in Cemented Joints, J. Appl. Mech., Vol 11, 1944, p 17
17. J.J. Plantema, “De Schuifspanning in eme Limjnaad,” Rep. M1181, Nat. Luchtvaart-laboratorium,
Amsterdam, 1949
18. G.R. Wooley and D.R. Carver, J. Aircr., Vol 8 (No. 19), 1971, p 817
19. R.D. Adams and N.A. Peppiatt, Stress Analysis of Adhesive-Bonded Lap Joints, J. Strain Anal., Vol 9
(No. 3), 1974, p 185
20. P.S. Cooper and J.W. Sawyer, “A Critical Examination of Stresses in an Elastic Single Lap Joint,”
NASA Tech. Rep. 1507, NASA Scientific and Technical Information Branch, 1979
21. G.P. Anderson and K.L. DeVries, Analysis of Standard Bond-Strength Tests, Treatise on Adhesion and
Adhesives, Vol 6, R.L. Patrick, K.L. DeVries, and G.P. Andersen, Ed., Marcel Dekker, 1989
22. J.K. Strozier, K.J. Ninow, K.L. DeVries, and G.P. Anderson, Adhes. Sci. Rev., Vol 1, 1987, p 121
23. G.P. Anderson, D.H. Brinton, K.J. Ninow, and K.L. DeVries, A Fracture Mechanics Approach to

Predicting Bond Strength, Advances in Adhesively Bonded Joints, Proceedings of a Conference at the
Winter Annual Meeting of ASME, 27 Nov-2 Dec 1988 (Chicago), S. Mall, K.M. Liechti, and J.K.
Vinson, Eds., ASME, 1988, p 98–101
Testing of Adhesive Joints
K.L. DeVries and Paul Borgmeier, University of Utah

Tensile Tests
Generally, the idea of mechanical failure produces a vision of an object being pulled apart by tensile force. As
noted previously, most practical adhesive joints are designed to avoid (or at least reduce) direct tensile forces
across the bond line. Examples of such joints are lap joints and scarf joints. It was also pointed out that for
many joints, where it appears that the primary loading is shear, failure might be initiated by the induced
secondary tensile stresses. There are, therefore, reasons why an adhesive's or adhesive joint's tensile strength
might be of interest. Accordingly, the third most common type of adhesive joint strength test is the tensile test.
ASTM has also formalized this type of test.
The geometries of several tensile tests for which there are specific ASTM test procedures are shown in Fig. 5
(Ref 7). Some of these test geometries seem relatively simple; however, it has been demonstrated that the
stresses along the bond line have a rather complex dependence on geometric factors and adhesive and adherent
properties (adhesive thickness and its variation across the bonded surface, modulus, Poisson's ratio, and so on)
(Ref 21).

Fig. 5 Typical specimen geometries for testing the tensile strength of adhesive joints. Source: Ref 7
It is almost always difficult to load tensile adhesion specimens in an axisymmetric manner, even if the sample
itself is axisymmetric. Nonaxisymmetric loads have been shown to reduce the bond failure load capability and
to cause large scatter in the resulting failure data. Superficially, the geometry for standard tensile adhesion tests
is deceptively simple. The result of the tensile adhesion test, as normally reported by experimentalists, is simply
the failure load divided by the cross-sectional area of the adhesive (Ref 22). Such average stress at failure can
be very misleading. Because of the differences in mechanical properties of the adhesive and adherend, the
stresses may become singular at the bond edges when analyzed using linear elastic analysis (Ref 21, 23). Even
if the edge singularity is neglected, the stress field in the adhesive is very complex and nonuniform, with
maximum values differing markedly from the average value (Ref 21, 23).

Some sense of the complex nature of the stresses can be obtained by visualizing a butt joint of a low modulus
polymer (e.g., a rubber) between two steel cylinders. As these are pulled apart, the rubber elongates much more
readily than the steel. Poisson's effect will cause a tendency for the rubber to contract laterally. However, if it is
tightly bound to the metal, it is restrained from contracting, and shear stresses are induced at the bond line.
Reference 9 provides the results of a finite element analysis that demonstrates how these stresses vary across
the sample. As noted, for an elastic analysis, both the shear and tensile stresses are singular (tending to infinity)
at the outer periphery.
For the tensile specimen configurations considered to this point, the applied loading is intended to be
axisymmetric. There is another class of specimen in which the dominant stress is deliberately tensile but in
which the loading is obviously “off center.” At least four ASTM standards describe so-called cleavage
specimens and tests. These tests are a logical preface to the next section in this article, “Adhesive Fracture
Mechanics Tests”. The reader familiar with cohesive fracture mechanics will see a similarity between the test
specimen in ASTM D 1062 (Fig. 6) and the compact tensile specimen commonly used in fracture mechanics
testing. ASTM D 1062 specifies reporting the test results as force required, per unit width, to initiate failure in
the specimen, while in fracture mechanics, the results are given as G
c
with units of J/m
2
, which might be
interpreted as the energy required to create a unit surface. A knowledgeable and enterprising reader may want
to adapt the D 1062 specimen for obtaining fracture mechanics parameters. ASTM D 3807, “Standard Test
Method for Strength Properties of Adhesives in Cleavage Peel by Tension Loading,” uses a different geometry
to measure the cleavage strength. In this case, two 25.4 mm (1 in.) wide by 6.35 mm (0.25 in.) thick plastic
strips 177 mm (7 in.) long are bonded over a length of 76 mm (3 in.) on one end, leaving the other ends free and
separated by the thickness of the adhesive. Approximately 25 mm (1 in.) from the end of each of these free
segments, a “gripping wire” is attached as shown in Fig. 7. During testing, these wires are clamped in the jaws
of a universal testing machine and the sample pulled to failure. The results are reported as load per unit width
(kg/m or lb/in.). Again, it would be possible to analyze this sample in terms of fracture mechanics, but it is
unnecessary because, as the next section explains, this analysis is done in ASTM D 3433 for a very similar
beam geometry.

Fig. 6 Specimen for testing the cleavage strength of metal-to-metal adhesive bonds (ASTM D 1062)

Fig. 7 Specimen for testing cleavage peel (by tension loading) (ASTM D 3807)
ASTM D 5041 also makes use of a sample composed of two thin sheets bonded together over part of their
length. In this case, forcing a wedge (45° angle) between the unbonded portion of the sheets facilitates the
separation. The results are typically given as “failure initiation energy” or “failure propagation energy” (i.e.,
areas under the load deformation curve).
This latter test is similar to another test, formalized as ASTM D 3762, that has been found very useful for
studying time-environmental effects on adhesive bonds. This test is called by various names, but the authors
prefer the name “Boeing Wedge Test” (Ref 24, 25). The test has been used by personnel at this and other
aerospace companies to screen various adhesives, surface treatment, and so on for long-term loading at high
temperatures and humidities. For testing, two long, slender strips of candidate structural materials are first
treated with the prescribed surface treatment(s) and bonded over part of their length with a candidate adhesive
(Fig. 8). As in the test described in the previous paragraph, the free ends are forced apart by a wedge. The
amount of separation by the wedge (determined by wedge thickness and depth of insertion) determines the
value of the stresses in the adhesive. These stresses can, of course, be adjusted and the values calculated from
mechanics of material concepts. When the wedge is in place, the sample is placed in an environmental chamber.
At periodic time intervals, the length of the crack is measured, and a plot of crack length versus time is
constructed. The more satisfactory adhesives and/or surface treatments are those for which the crack is arrested
or grows very slowly. While the environmental chamber typically contains hot, humid air, there is no reason
why other environmental agents cannot be studied by the same method, including immersion in liquids.

Fig. 8 Boeing wedge test (ASTM D 3762) (a) Test specimen. (b) Typical crack propagation behavior at 49
°C (120 °F) and 100% relative humidity. a, distance from load point to initial crack tip; Δa, growth
during exposure. Source: Ref 49
References cited in this section
7. Adhesives, Annual Book of ASTM Standards, Vol 15.06, ASTM (updated annually)
9. G.P. Anderson and K.L. DeVries, Predicting Strength of Adhesive Joints from Test Results, Int. J.
Fract., Vol 39, 1989, p191–200

21. G.P. Anderson and K.L. DeVries, Analysis of Standard Bond-Strength Tests, Treatise on Adhesion and
Adhesives, Vol 6, R.L. Patrick, K.L. DeVries, and G.P. Andersen, Ed., Marcel Dekker, 1989
22. J.K. Strozier, K.J. Ninow, K.L. DeVries, and G.P. Anderson, Adhes. Sci. Rev., Vol 1, 1987, p 121
23. G.P. Anderson, D.H. Brinton, K.J. Ninow, and K.L. DeVries, A Fracture Mechanics Approach to
Predicting Bond Strength, Advances in Adhesively Bonded Joints, Proceedings of a Conference at the
Winter Annual Meeting of ASME, 27 Nov-2 Dec 1988 (Chicago), S. Mall, K.M. Liechti, and J.K.
Vinson, Eds., ASME, 1988, p 98–101
24. V.L. Hein and F. Erodogan, Stress Singularities in a Two-Material Wedge, Int. J. Fract.,Vol 7, 1971, p
317
25. J.A. Marceau, Y. Moji, and J.C. McMillan, A Wedge Test for Evaluating Adhesive Bonded Surface
Durability, 21st SAMPE Symposium, Vol 21, 6–8 April 1976
49. J.C. McMillan, Developments in Adhesives in Engineering, 2nd ed., Applied Science, London, 1981, p
243
Testing of Adhesive Joints
K.L. DeVries and Paul Borgmeier, University of Utah

Adhesive Fracture Mechanics Tests
Fracture mechanics originated with the pioneering efforts of A.A. Griffith in the early 1920s. The field
remained relatively dormant until the late 1940s when it was developed into a very effective and valuable
design tool to describe and predict “cohesive” crack growth. Interested readers are referred to a number of
excellent texts on fracture mechanics (e.g., Ref 26 and Fatigue and Fracture, Volume 19 of the ASM
Handbook).
In the 1960s and 1970s, researchers began exploring the use of the concepts of fracture mechanics in adhesive
joint analysis as reviewed in Ref 3. These methods have the potential to use the results from a test joint to
predict the strength of other joints with different geometries.
In a common fracture mechanics approach (including Griffith's papers), the conditions for failure are calculated
by equating the energy lost from the strain field as a “crack” grows to the energy consumed in creating the new
crack surface. This energy per unit area, G
c
, determined from standard tests, is called by various names,

including the Griffith fracture energy, the specific fracture energy, the fracture toughness, or the energy release
rate.
In 1975, ASTM Committee D-14 adopted a test configuration and testing method with fracture mechanics
ramifications based on the pioneering efforts of Mostovy and Ripling (Ref 27, 28). The method is described in
ASTM D 3433 “Standard Test Method for Fracture Strength in Cleavage of Adhesives in Bonded Joints.”
Figure 9 shows the shape and dimensions for one specimen type recommended for use in this standard. The
specimen is composed of two “beams” adhesively bonded over much of their length as shown. Testing is
accomplished by pulling the specimen apart by means of pins passing through the holes shown near the
sample's left end. This adhesive sample configuration and loading to failure gives rise to the sample's nickname,
“split-cantilever beam.” Another recommended geometry in ASTM D 3433 is similar except the adherends are
not tapered.

Fig. 9 Specimen for the contoured double-cantilever-beam test (ASTM D 3433)
It should now be clear that the stress distribution in adhesive joints is generally complex. Furthermore, the
details of this distribution are highly dependent on specific details of the joint system. The maximum stresses in
the bond almost always differ markedly from the average value, and elastic analyses often exhibit mathematical
singularities at geometric or material discontinuities. From these observations, it should be clear that the use of
the conventionally reported results from most tests (i.e., values of the average stress at failure) would be of little
use in designing joints that differ in any significant detail from the sample test configuration.
For the resolution of this problem, the concepts of fracture mechanics have much to offer. One of the more
popular and graphically appealing approaches to fracture mechanics views the joint as a system in which failure
(often considered as the growth of a crack) of a material (or joint) requires the stresses at the crack tip to be
sufficient to break bonds and an energy balance. It is hypothesized that even if the stresses are very large (often
theoretically infinite), a crack can grow only if sufficient energy is released from the stress field to account for
the energy required to create the new crack (or adhesive debond) surface as the fractured region enlarges. The
specific value of this energy (J/m
2
, or in. · lbf/in.
2
, of crack area) for the adhesive bonding problem uses the

same basic titles as given previously but prefaced with the term adhesive. Hence, adhesive fracture toughness
might be used to distinguish adhesive failure from tests of cohesive fracture. The word adhesion is dropped
from the comparable term when cohesive failure is being considered. The cohesive and adhesive embodiments
of fracture mechanics both involve a stress-strain analysis and an energy balance.
The analytical methods of fracture mechanics (both cohesive and adhesive) are described in Ref 3 and 25.
These are not repeated here other than a few comments on the concepts and a brief outline of a numerical
approach that can be applied where analytical solutions are tedious or impossible. Inherent in fracture
mechanics is the concept that natural cracks or other stress risers exist in materials and that final failure of an
object often initiates at such points. For a crack (or region of debond) situated in an adhesive layer, modern
computation techniques are available (most notably, finite element methods) that facilitate the computation of
stresses and strains throughout a body, even if analytical solutions may not be possible. The stresses and strains
are calculated throughout the entire adhesive system (adhesive and all adherends), including the effects of a
crack in the bond. These can then be used to calculate the strain energy, U
1
, stored in the body for the particular
crack size, A
1
. Next, the hypothetical crack is allowed to grow to a slightly larger area, A
2
, and the preceding
process is repeated to determine the strain energy, U
2
. This approach to fracture mechanics assumes that at
critical crack growth conditions, the energy loss from the stress-strain field goes into the formation of the new
fracture energy. The quantity ΔU/ΔA is called the energy release rate, where ΔU = U
2
- U
1
and ΔA = A
2

- A
1
.
The so-called critical energy release rate (ΔU/ΔA)
crit
is that value of the energy release rate that will cause the
crack to grow. Loads that result in energy release rates lower than this critical value will not cause failure to
proceed from the given crack, while loads that produce energy release rates greater than this value will cause it
to accelerate. This critical energy release rate value is equivalent to the adhesive fracture energy, or work of
adhesion, previously noted. While the model just described is conceptually useful, computer engineers have
devised other convenient ways of computing the energy required to “create” the new surface, such as the crack
closure method (Ref 29, 30).
It is hoped that this simple model of fracture mechanics will help the reader who is unfamiliar with fracture
mechanics to visualize the concepts of fracture mechanics. The molecular mechanisms responsible for the
fracture energy or fracture toughness are not completely understood. They generally involve more than simply
the energy required to rupture a plane of molecular bonds. In fact, for most practical adhesives, the energy to
rupture these bonds is a small but essential fraction of the total energy. The total energy includes energy that is
lost because of viscous, plastic, and other dissipation mechanisms at the tip of the crack. As a result, linear
elastic stress analyses are inexact.
While fracture mechanics has found extensive use in cohesive failure considerations, its use for analyzing
failure of adhesive systems is more recent. There has, however, been a significant amount of research and
development in the adhesive fracture mechanics area. To review it all, even superficially, would take more
space than is allocated for this article. A small sampling of publications in this extensive and rich area of
research is listed as Ref 12, 13, 26, 27, 28, and 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47,
48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58. Not only is this listing incomplete, but also many of the researchers
listed have scores of other publications. It is hoped that the one or two listed for each investigator will provide
the reader with a starting point from which more details can be found from reference cross listings, searching of
citation indexes, abstracting services, and so on. These investigators have treated such subjects as theory; mode
dependence, effects of shape, thickness, and other geometric dependence; plasticity and other nonlinearities;
numerical methods; testing techniques; different adhesive types; rate and temperature effects; fatigue; and

failure of composites, as well as a wide variety of other factors and considerations in adhesion.
Modern finite element or other numerical methods have no difficulty in treating nonlinear behavior. Physical
understanding of material behavior at such levels is lacking, and effective use of the capabilities of such
computer codes depends, to a large extent, on the experimental determination of these properties. For many
problems, it has become conventional to lump all dissipative effects together into the fracture energy and not be
overly concerned with separating this quantity into its individual energy-absorbing components. Another
fracture mechanics approach, called the J-integral, has some advantages in treating nonlinear as well as elastic
behavior (Ref 51, 52, 59, and 60).
It was noted previously that most adhesive systems are not linearly elastic up to the failure point. Nevertheless,
researchers have shown that elastic analyses of many systems can be very informative and useful. Several
adhesive systems are sufficiently linear so that it is possible to lump the plastic deformation and other energy
dissipative mechanisms at the crack tip into the adhesive fracture energy (critical energy release rate) term.
There has recently been some significant success in explaining many aspects of adhesive performance and
predicting the strength of a bond from tests on other, quite different, joints by using linear elastic fracture
mechanics.
As noted, in principle, fracture mechanics lends itself to using test results from one test in the design of other
joints that have significantly different geometries. A number of adhesive geometries have been proposed to
measure fracture toughness in addition to the split-cantilever beam, but to date, it is the only one formalized by
ASTM (Ref 3, 4, 6, 12, 36, 47, and 48). A recent paper by the authors (Ref 61) has demonstrated how such
factors as end rotation (at the cantilever point assumed rigidly fixed in the original ASTM analysis) shear, and
presence of the adhesive and its thickness (also neglected in the original analysis) affect the energy release rate.
It is shown that inclusion of these effects can dramatically affect the results and greatly reduce test scatter.
Furthermore, this paper demonstrates how fracture mechanics may be used to predict the locus of adhesive
crack growth. To accomplish this, various crack paths were assumed, and using finite element methods, the
energy release rate calculated for each path.
References cited in this section
3. G.P. Anderson, S.J. Bennett, and K.L. DeVries, Analysis and Testing of Adhesive Bonds, Academic
Press, 1977
4. A.J. Kinlock, Adhesion and Adhesives, Chapman and Hall, 1987
6. K.L. Mittal, Adhesive Joints, Plenum Press, 1984

12. A.N. Gent and G.R. Hamed, Peel Mechanics, J. Adhes., Vol 7, 1975, p 91
13. A.N. Gent and G.R. Hamed, J. Appl. Polym. Sci., Vol 21, 1977, p 2817
25. J.A. Marceau, Y. Moji, and J.C. McMillan, A Wedge Test for Evaluating Adhesive Bonded Surface
Durability, 21st SAMPE Symposium, Vol 21, 6–8 April 1976
26. D. Broek, The Practical Use of Fracture Mechanics, Kluver Acad. Press, Dordrect, NL, 1989
27. S. Mostovoy and E.J. Ripling, J. Appl. Polym. Sci., Vol 15, 1971, p 661
28. S. Mostovoy and E.J. Ripling, J. Adhes. Sci. Technol., Vol 9B, 1975, p 513
29. E.F. Rybicki and M.F. Kanninen, Eng.Fract. Mech., Vol 9, 1974, p 921
30. G.P. Anderson and K.L. DeVries, J. Adhes., Vol 23, 1987, p 289
31. E.H. Andrews, T.A. Khan, and H.A. Majid, J. Mater. Sci., Vol 20, 1985, p 3621
32. E.H. Andrews, H.A. Majid, and N.A. Lockington, J. Mater. Sci., Vol 19, 1984, p 73
33. D.W. Aubrey and M. Sherriff, J. Polym. Sci. Polym. Chem. Ed., Vol 18, 1980, p 2597
34. W.D. Bascom and J. Oroshnik, J. Mater. Sci., Vol 10, 1978, p 1411
35. W.D. Bascom and D.L. Hunston, Fracture of Epoxy and Elastomer-Modified Epoxy Polymers, Treatise
on Adhesion and Adhesives, Vol 6, R. L. Patrick, K.L. DeVries, and G.P. Anderson, Ed., Marcel
Dekker, 1988, p 123
36. H.F. Brinson, J.P. Wightman, and T.C. Ward, Adhesives Science Review 1, VPI Press, 1987
37. J.D. Burton, W.B. Jones, and M.L. Williams, Trans. Soc. Rheol, Vol 15, 1971, p 39
38. G. Danneberg, J. Appl. Polym. Sci., Vol 5, 1961, p 125
39. F. Erdogan, Eng. Fract. Mech., Vol 4, 1972, p 811
40. T.R. Guess, R.E. Allred, and F.P. Gerstle, J. Test. Eval., Vol 5 (No. 2), 1977, p 84
41. G.R. Hamed, Energy Conservation During Peel Testing, Treatise on Adhesion and Adhesives, Vol 6, R.
L. Patrick, K. L. DeVries, and G. P. Anderson, Eds., Marcel Dekker, 1988, p 233
42. R.W. Hertzberg and J.A. Manson, Fatigue of Engineering Plastics, Academy, 1980
43. G.R. Irwin, Fracture Mechanics Applied to Adhesive Systems, Treatise on Adhesion and Adhesives, Vol
1, R. L. Patrick, Ed., Marcel Dekker, 1966, p 233
44. W.S. Johnson, J. Test. Eval., Vol 15 (No. 6), 1987, p 303
45. D.H. Kaelble, Physical Chemistry of Adhesion, Wiley-Interscience, 1971
46. H.H. Kaush, Polymer Fracture, Springer-Verlag, Berlin, 1978
47. W.G. Knauss and K.M. Liechti, Interfacial Crack Growth and Its Relation to Crack Front Profiles, ACS

Organic Coatings and Applied Polymer Science Proceedings, Vol 47, American Chemical Society,
Washington, DC, 1982, p 481
48. K.M. Liechti and C. Lin, Structural Adhesives in Engineering, Institute of Mechanical Engineering,
London, 1986, p 83
49. J.C. McMillan, Developments in Adhesives in Engineering, 2nd ed., Applied Science, London, 1981, p
243
50. D.R. Mulville, D.L. Hunston, and P.W. Mast, J. Eng. Mater. Technol., Vol 100, 1978, p 25
51. J.R. Rice and G.C. Sih, J. Appl. Mech., Vol 32, 1965, p 418
52. E.F. Rybicki and M.F. Kanninen, Eng. Fract. Mech., Vol 9, 1974, p 921
53. G.B. Sinclair, Int. J. Fract., Vol 16, 1980, p 111
54. J.D. Venables, D.K. McNamara, J.M. Chen, T.S. Sun, and R.L. Hopping, Appl. Surf. Sci., Vol 3, 1979, p
88
55. S.S. Wang, J.F. Mandell, and F.J. McGarry, Int. J. Fract., Vol 14, 1978, p 39
56. J.G. Williams, Fracture Mechanics of Polymers, Ellis Horwood, Chichester, 1984
57. M.L. Williams, J. Adhes., Vol 5, 1973, p 81
58. R.J. Young, in Structural Adhesives: Developments in Resins and Primers, A.J. Kinlock, Ed., Applied
Science, London, 1986, p 163
59. J.R. Rice, A Path Independent Integral and the Approximate Analysis of Strain Concentration by
Notches and Cracks, J. Appl. Mach., 1979, p 379–386h
60. J.W. Hutchinson and P.C. Paris, Stability of J-Controlled Crack Growth, STP 668, ASTM, 1979, p 37–
64
61. K.L. DeVries and P.R. Borgmeier, Fracture Mechanics Analyses of the Behavior of Adhesion Test
Specimens, Mittal Festschrift, W.J. Van Ooij and H.R. Anderson, Jr., Ed., 1998, p 615–640

Testing of Adhesive Joints
K.L. DeVries and Paul Borgmeier, University of Utah

Conclusions
The adhesive researcher or technologist has many standard test methods from which to choose. These
techniques are designed with various goals and objectives in mind. Many of these methods are useful for the

purposes of comparing different adhesives and substrates, investigating the effects of different loading,
investigating chemical or physical attacks on adhesives, exploring aging phenomena, determining the effects of
radiation and moisture combined with sustained loading on adhesive properties, and so on. On the other hand,
care should always be exercised not to use the test results for purposes for which they are not well suited.
Results from many of the adhesive strength tests are conventionally reported as the failure force divided by the
bond area. Such average stress at failure results cannot, in general, be consistently and reliably used to predict
failure of other joints that differ even slightly from the test geometry. Fracture mechanics approaches, on the
other hand, show promise and have been used to predict the strength of joints that differ considerably from the
reference joint. ASTM D 3433 and Ref 27 and 28 describe a standard adhesive fracture mechanics joint in the
form of a tapered double-cantilever beam. The specimen dimensions are shown in Fig. 9. It is important to note,
however, that fracture mechanics is not limited to this or any other specific testing geometry. In principle, any
geometry for which the described energy balance (or alternatively, calculation of the stress intensity factor, J-
integral, and so on) can be accomplished might be used as an adhesive test.
Sometimes, circumstances dictate the use of a nonstandard test geometry. For example, a few years ago, the
authors were given the problem of measuring the quality of natural barnacle adhesive. The barnacle dictated the
exact form of the joint between the barnacle's shell and the plastic sheets that were placed in the ocean. This
form did not lend itself to tensile, lap shear, or split-cantilever testing. It was, however, possible to predrill
holes in the plate and to fill these holes with dental waxes that were solid and hard at the ocean temperatures
near San Francisco, CA, where the barnacle growth experiments were conducted. The wax was later easily
removed at a moderately elevated temperature. The base of the barnacle covering this hole was thereby exposed
and could be tested by application of fluid pressure, thus forming a blister. Measurement of the pressurization at
failure allowed the determination of the adhesive fracture energy (Ref 3, 62).
Once the adhesive fracture energy is determined by testing, fracture mechanics points the way that it, along
with a knowledge of the flaw size and a stress-strain analysis of the joint, can be used to predict the
performance of other joints. Modern computational techniques greatly facilitate the application of these
methods.
Finally, it is noted that the stresses, strains, fracture energy, and other such parameters used in the adhesive
analysis depend on loading rate, mode of stress at the crack tip, temperature, environment, and other factors.
Development of means for incorporating these parameters into joint design has been, and continues to be, an
area of active research. Such concepts and methodology can be found in the references cited previously in this

section.
References cited in this section
3. G.P. Anderson, S.J. Bennett, and K.L. DeVries, Analysis and Testing of Adhesive Bonds, Academic
Press, 1977
27. S. Mostovoy and E.J. Ripling, J. Appl. Polym. Sci., Vol 15, 1971, p 661
28. S. Mostovoy and E.J. Ripling, J. Adhes. Sci. Technol., Vol 9B, 1975, p 513
62. R.R. Despain, R.D. Luntz, K.L. DeVries, and M.L. Williams, J. Dental Res., Vol 52, 1973, p 742

Testing of Adhesive Joints
K.L. DeVries and Paul Borgmeier, University of Utah

Acknowledgments
Major portions of the authors' research has been supported by The National Science Foundation, most recently
under grant No. CMS-9522743.

Testing of Adhesive Joints
K.L. DeVries and Paul Borgmeier, University of Utah

References
1. Adhesives, Edition 6, D.A.T.A. Digest International Plastics Selector, 1991
2. R.L. Patrick, Ed., Treatise on Adhesion and Adhesives, Vol 1–6, Marcel Dekker, 1966–1988
3. G.P. Anderson, S.J. Bennett, and K.L. DeVries, Analysis and Testing of Adhesive Bonds, Academic
Press, 1977
4. A.J. Kinlock, Adhesion and Adhesives, Chapman and Hall, 1987
5. A. Pizzi and K.L. Mittal, Ed., Handbook of Adhesive Technology, Marcel Dekker, 1994
6. K.L. Mittal, Adhesive Joints, Plenum Press, 1984
7. Adhesives, Annual Book of ASTM Standards, Vol 15.06, ASTM (updated annually)
8. E.P. Plueddemann, Silane Coupling Agents, Plenum Press, 1982
9. G.P. Anderson and K.L. DeVries, Predicting Strength of Adhesive Joints from Test Results, Int. J.
Fract., Vol 39, 1989, p191–200

10. J.L. Gardon, Peel Adhesion, I. Some Phenomenological Aspects of the Test, J. Appl. Polym. Sci., Vol 7,
1963, p 654
11. D.H. Kaelble, Theory and Analysis of Peel Adhesion: Mechanisms and Mechanics, Trans. Soc. Rheol.,
Vol 3, 1959, p 161
12. A.N. Gent and G.R. Hamed, Peel Mechanics, J. Adhes., Vol 7, 1975, p 91
13. A.N. Gent and G.R. Hamed, J. Appl. Polym. Sci., Vol 21, 1977, p 2817
14. R.D. Adams and A. Crocombe, J. Adhes., Vol 12, 1981, p 127
15. O. Volkersen, Die Nietraftverteilung in Zugbeanspruchten Nietverblendugen mit Knastaten
Laschenquerschntlen, Luftfahrt forsch., Vol 15, 1938, p 41
16. M. Goland and E. Reissner, The Stresses in Cemented Joints, J. Appl. Mech., Vol 11, 1944, p 17
17. J.J. Plantema, “De Schuifspanning in eme Limjnaad,” Rep. M1181, Nat. Luchtvaart-laboratorium,
Amsterdam, 1949
18. G.R. Wooley and D.R. Carver, J. Aircr., Vol 8 (No. 19), 1971, p 817
19. R.D. Adams and N.A. Peppiatt, Stress Analysis of Adhesive-Bonded Lap Joints, J. Strain Anal., Vol 9
(No. 3), 1974, p 185
20. P.S. Cooper and J.W. Sawyer, “A Critical Examination of Stresses in an Elastic Single Lap Joint,”
NASA Tech. Rep. 1507, NASA Scientific and Technical Information Branch, 1979
21. G.P. Anderson and K.L. DeVries, Analysis of Standard Bond-Strength Tests, Treatise on Adhesion and
Adhesives, Vol 6, R.L. Patrick, K.L. DeVries, and G.P. Andersen, Ed., Marcel Dekker, 1989
22. J.K. Strozier, K.J. Ninow, K.L. DeVries, and G.P. Anderson, Adhes. Sci. Rev., Vol 1, 1987, p 121
23. G.P. Anderson, D.H. Brinton, K.J. Ninow, and K.L. DeVries, A Fracture Mechanics Approach to
Predicting Bond Strength, Advances in Adhesively Bonded Joints, Proceedings of a Conference at the
Winter Annual Meeting of ASME, 27 Nov-2 Dec 1988 (Chicago), S. Mall, K.M. Liechti, and J.K.
Vinson, Eds., ASME, 1988, p 98–101
24. V.L. Hein and F. Erodogan, Stress Singularities in a Two-Material Wedge, Int. J. Fract.,Vol 7, 1971, p
317
25. J.A. Marceau, Y. Moji, and J.C. McMillan, A Wedge Test for Evaluating Adhesive Bonded Surface
Durability, 21st SAMPE Symposium, Vol 21, 6–8 April 1976
26. D. Broek, The Practical Use of Fracture Mechanics, Kluver Acad. Press, Dordrect, NL, 1989
27. S. Mostovoy and E.J. Ripling, J. Appl. Polym. Sci., Vol 15, 1971, p 661

28. S. Mostovoy and E.J. Ripling, J. Adhes. Sci. Technol., Vol 9B, 1975, p 513
29. E.F. Rybicki and M.F. Kanninen, Eng.Fract. Mech., Vol 9, 1974, p 921
30. G.P. Anderson and K.L. DeVries, J. Adhes., Vol 23, 1987, p 289
31. E.H. Andrews, T.A. Khan, and H.A. Majid, J. Mater. Sci., Vol 20, 1985, p 3621
32. E.H. Andrews, H.A. Majid, and N.A. Lockington, J. Mater. Sci., Vol 19, 1984, p 73
33. D.W. Aubrey and M. Sherriff, J. Polym. Sci. Polym. Chem. Ed., Vol 18, 1980, p 2597
34. W.D. Bascom and J. Oroshnik, J. Mater. Sci., Vol 10, 1978, p 1411
35. W.D. Bascom and D.L. Hunston, Fracture of Epoxy and Elastomer-Modified Epoxy Polymers, Treatise
on Adhesion and Adhesives, Vol 6, R. L. Patrick, K.L. DeVries, and G.P. Anderson, Ed., Marcel
Dekker, 1988, p 123
36. H.F. Brinson, J.P. Wightman, and T.C. Ward, Adhesives Science Review 1, VPI Press, 1987
37. J.D. Burton, W.B. Jones, and M.L. Williams, Trans. Soc. Rheol, Vol 15, 1971, p 39
38. G. Danneberg, J. Appl. Polym. Sci., Vol 5, 1961, p 125
39. F. Erdogan, Eng. Fract. Mech., Vol 4, 1972, p 811
40. T.R. Guess, R.E. Allred, and F.P. Gerstle, J. Test. Eval., Vol 5 (No. 2), 1977, p 84
41. G.R. Hamed, Energy Conservation During Peel Testing, Treatise on Adhesion and Adhesives, Vol 6, R.
L. Patrick, K. L. DeVries, and G. P. Anderson, Eds., Marcel Dekker, 1988, p 233
42. R.W. Hertzberg and J.A. Manson, Fatigue of Engineering Plastics, Academy, 1980
43. G.R. Irwin, Fracture Mechanics Applied to Adhesive Systems, Treatise on Adhesion and Adhesives, Vol
1, R. L. Patrick, Ed., Marcel Dekker, 1966, p 233
44. W.S. Johnson, J. Test. Eval., Vol 15 (No. 6), 1987, p 303
45. D.H. Kaelble, Physical Chemistry of Adhesion, Wiley-Interscience, 1971
46. H.H. Kaush, Polymer Fracture, Springer-Verlag, Berlin, 1978
47. W.G. Knauss and K.M. Liechti, Interfacial Crack Growth and Its Relation to Crack Front Profiles, ACS
Organic Coatings and Applied Polymer Science Proceedings, Vol 47, American Chemical Society,
Washington, DC, 1982, p 481
48. K.M. Liechti and C. Lin, Structural Adhesives in Engineering, Institute of Mechanical Engineering,
London, 1986, p 83
49. J.C. McMillan, Developments in Adhesives in Engineering, 2nd ed., Applied Science, London, 1981, p
243

50. D.R. Mulville, D.L. Hunston, and P.W. Mast, J. Eng. Mater. Technol., Vol 100, 1978, p 25
51. J.R. Rice and G.C. Sih, J. Appl. Mech., Vol 32, 1965, p 418
52. E.F. Rybicki and M.F. Kanninen, Eng. Fract. Mech., Vol 9, 1974, p 921
53. G.B. Sinclair, Int. J. Fract., Vol 16, 1980, p 111
54. J.D. Venables, D.K. McNamara, J.M. Chen, T.S. Sun, and R.L. Hopping, Appl. Surf. Sci., Vol 3, 1979, p
88
55. S.S. Wang, J.F. Mandell, and F.J. McGarry, Int. J. Fract., Vol 14, 1978, p 39
56. J.G. Williams, Fracture Mechanics of Polymers, Ellis Horwood, Chichester, 1984
57. M.L. Williams, J. Adhes., Vol 5, 1973, p 81
58. R.J. Young, in Structural Adhesives: Developments in Resins and Primers, A.J. Kinlock, Ed., Applied
Science, London, 1986, p 163
59. J.R. Rice, A Path Independent Integral and the Approximate Analysis of Strain Concentration by
Notches and Cracks, J. Appl. Mach., 1979, p 379–386h
60. J.W. Hutchinson and P.C. Paris, Stability of J-Controlled Crack Growth, STP 668, ASTM, 1979, p 37–
64
61. K.L. DeVries and P.R. Borgmeier, Fracture Mechanics Analyses of the Behavior of Adhesion Test
Specimens, Mittal Festschrift, W.J. Van Ooij and H.R. Anderson, Jr., Ed., 1998, p 615–640
62. R.R. Despain, R.D. Luntz, K.L. DeVries, and M.L. Williams, J. Dental Res., Vol 52, 1973, p 742

Mechanical Testing of Welded Joints
William Mohr, Edison Welding Institute

Introduction
IN WELDED STRUCTURES, the welds typically have a mechanical purpose. Loads must be carried across the
weld joint. Standard mechanical tests have been devised to demonstrate that not only the base metals but also
the entire welded joint can fulfill this mechanical purpose (Ref 1, 2). This article primarily discusses standard
test methods that can be applied to many types of welds. These include tension, bending, impact, and toughness
testing.
Residual stress measurement techniques and weldability testing also are discussed. Residual stress can be
imposed by the welding itself, as well as by cutting and forming processes. The presence of high-tension or

compression residual stresses can affect the ability of the welded structure to carry the mechanical loading.
Cracking due to welding can also affect the load-carrying capacity of welded joints, and weldability testing
techniques that combine welding and mechanical loading to test the resistance to cracking are available. Many
other testing techniques can be applied to weld joints and welded structures. Fatigue and creep are both
important areas where mechanical tests on welded joints have indicated properties different from those of the
base metal. Testing of welded structure properties can also be done on structures that more closely model the
service structure than the standard specimens described below.
References cited in this section
1. “Standard Welding Terms and Definitions,” AWS 3.0, American Welding Society
2. L.P. Connor, Ed., Welding Handbook, 8th ed., Vol 1, American Welding Society, 1991





Mechanical Testing of Welded Joints
William Mohr, Edison Welding Institute

Reasons to Measure Properties of Welds
The mechanical properties of welded joints, the properties related to stress and strain, are most often measured
to show that such a weld and other similar welds will serve their purpose under loading. More rarely, several
welds are compared to see which welding techniques, processes, or chemistries provide the best combination of
mechanical properties.
Four stages in the qualification process for the weld joint can use standard mechanical test methods. The weld
metal can be chosen based on the mechanical properties from standard tests. The base metal, in some situations,
may also need to be qualified to demonstrate that its mechanical properties will not be substantially degraded
by welding. Once base metal and weld metal are chosen, the other weld process parameters, such as weld shape
or heat input, can be qualified by standard mechanical testing. Finally, after the weld is made, it may require
qualification by mechanical testing.
Each of the four qualification stages requires different types of tests and different approaches. These approaches

are described in the next four sections.
Weld Material Qualification
Weld metals are qualified by making welds that pass mechanical property tests on the weld metal itself. Such
tests can be used to qualify filler materials, such as welding wire or electrodes. Mechanical tests for such
qualification are described in the individual specifications of the filler metals, such as those in American
Welding Society (AWS) Specification A5.1 (Ref 3) and others of the AWS A5 series.
Mechanical property tests applied to weld qualification are designed to determine a small number of standard
values to check whether the weld metal passes or fails. This approach will not reveal the entire range of
properties that the weld metal can achieve. For instance, weld metal toughness in AWS A5.1 is measured by a
Charpy test specimen taken from the weld centerline. Other locations, which may have different toughnesses,
are not checked.
Base Material Qualification for Welded Service
The heat of welding will modify the structure and properties of the region of the base metal adjacent to the weld
in the heat-affected zone (HAZ). To prevent this modification in properties from causing service failures, some
standards require that a sample of the base material be tested after undergoing a representative heat treatment.
American Petroleum Institute standard (API) RP2Z is an example of a standard that requires base-material
qualification for welded service (Ref 4). Multipass welding provides the heat treatment. The welding
parameters are chosen to represent the most severe heat treatment of the base material that may occur during
fabrication. The parameter of interest is the fracture toughness, commonly measured by crack tip opening
displacement (CTOD). Particular regions of the HAZ are the most likely to show low toughness (the local
brittle zones, or LBZs), so there is also a requirement that the crack sample the required portion of that kind of
microstructure.
Weld Procedure Qualification
Weldment properties are dependent not only on the materials used to make the joint but also on the other
parameters of the welding process. Weld metal properties may be modified by the admixture of base material
melted by the heat of welding. The region where the base metal was only partially melted, at the fusion line,
may have local mechanical properties differing from those of the neighboring weld metal and HAZ regions. In
addition, the welding processes and procedures may induce specific imperfections, such as slag inclusions,
blowholes, or cracks.
Because the issues described cannot be resolved by either weld metal or base material tests alone, test

procedures that use specimens containing weld metal, base metal, and HAZ are used to determine mechanical
properties of welded joints in weldment procedure qualifications.
Weld procedure qualification tests may be less quantitative than weld metal qualification tests and often
provide only a “yes-or-no” answer. They are often capable of being completed in a shop floor environment
rather than a testing laboratory with calibrated equipment.
While many of the weld procedure qualification tests in wide use are discussed subsequently in this article,
several are not, because they are not properly mechanical tests; that is, neither a loading parameter, such as
stress, or a displacement parameter, such as strain, is measured. This group includes visual examination for
surface flaws and the breaking of fillet welds to examine the weld root.
Weld Service Assessment
Assessment of existing welds to determine if they meet the needs of future service may require that material
properties be obtained from representative weldments. In some cases, new weldments can be made with the
same materials and process parameters, so that sections from these weldments can be tested to find
representative properties for the existing welds. Generally, neither the original materials nor full information is
available to allow replication of existing weldments. Instead, a sample must be taken from the existing
weldments.
Taking a sample requires trading the advantage of obtaining mechanical property data for the disadvantage of
damaging the existing structure. Choices often are made that limit the damage to the existing structure by
limiting the amount of material to be tested. Tests such as macrohardness or microhardness, which damage a
small surface volume, may be appropriate. Smaller-scale test specimens may also be used, for instance, subsize
Charpy specimens. Alternatively, specimens can be taken from regions where subsequent repair is easiest or
from an area that is being removed as part of a modification.
References cited in this section
3. “Specification for Carbon Steel Electrodes for Shielded Metal Arc Welding,” ANSI/AWS A5.1-91,
American National Standards Institute/American Welding Society, Miami, 1991
4. “Recommended Practice for Preproduction Qualification for Steel Plates for Offshore Structures,” API
2Z, 2nd ed. (includes November 1998 addenda), American Petroleum Institute, Washington DC, 1998

Mechanical Testing of Welded Joints
William Mohr, Edison Welding Institute

Mechanical Testing for Weldment Properties
Tensile Strength and Ductility of Weldments
Testing for mechanical properties of strength and ductility for welded joints is somewhat more complicated
than it is for base metal, because these properties vary across the weld metal, the adjacent HAZ, and the base
metal. Several different tests may be used or combined to assess the strength of the overall welded joints.
Tensile testing is widely used to measure the strength and ductility of the weld metal alone. Tensile testing of
welds in place, with weld metal, HAZ, and base metal, allows an overall strength to be determined but usually
cannot provide the strengths of the individual parts of the weldment.
Tensile tests of welds can also measure elastic modulus. However, except in rare cases of dissimilar metal
joining, the elastic modulus is not sensitive to the differences between weld, HAZ, and base metal. So,
measurement during weld tensile tests is not usually required. Also, most tensile testing procedures for weld
joints cannot be relied upon to provide accurate values of elastic modulus. The specific procedures for testing of
elastic modulus distributed by ASTM should be used if required (Ref 5).
Testing of Weld Material. Deposited weld metal can be tested for the mechanical properties of strength and
ductility using the same test methods used for base metals (Ref 6, 7). However, a sufficient volume of deposited
weld metal is required to remove a test specimen made entirely of weld metal. Often, arc welds are long only in
one direction (the longitudinal direction), while the through-thickness and cross-weld directions are much
smaller. This encourages all-weld-metal tensile test specimens to be removed with the long direction of the
specimen corresponding to the longitudinal direction of the weld. Such longitudinal tensile test specimens are
standard for all-weld-metal tests.
All-weld-metal tests are most commonly done on specimens with round cross section. The diameter of the
specimen may need to be reduced from that used for base metal so that the specimen can be taken entirely from
weld metal. Rectangular cross-section specimens also are used occasionally.
Ultimate tensile strength, yield strength (usually based either on yield point or a specified offset), elongation,
and reduction of area are all commonly recorded.
While the specimen surface should be smooth, without deep machining marks, imperfections within the gage
length due to welding should not be removed. This requirement may increase the variability of results within a
group of similar specimens.
If the data required are for a class of weld material such as an electrode lot, the material can be taken from

specimens that reduce the possibility of dilution of base metal into the weld, such as a built-up weld pad. If the
data required are for a particular weldment, the geometry as well as the welding process and procedure should
model those of the weldment as closely as possible. Some modifications of the weldment may be allowed, such
as increasing the root opening by 6 mm ( in.) or buttering the groove faces with the weld metal to be tested.
The surface of the tested section, in the gage length, is recommended to be 3 mm ( in.) or more from the
fusion line.
Testing of Welds in Place. When the weld metal extends over only part of the tested gage length, tensile tests
can be performed similar to those performed on the round and rectangular specimen tests of weld metal. The
nonuniformity of deformation or stresses of the weld, HAZ, and base metal combination limits the information
normally recorded.
For transverse tests, ultimate strength and the location of fracture are the only commonly recorded parameters,
because strength, elongation, and reduction in area will all be affected by the constraint of the adjacent differing
materials. If the weld is undermatched, the yield strength tends to be higher than it is for an all-weld-metal
specimen, while the elongation over the gage length and reduction in area are smaller. If the weld yield strength
exceeds that of the base material, that is, it is overmatched, the failure tends to occur not in the adjacent HAZ,
but in the base material closer to the end of the gage length, because of the constraint provided by the high-
strength weld metal.
Local strain measurements, such as those made by strain gages, can add useful information to the results of
transverse testing. The local strain information can be correlated to the load and displacement information to
allow local strengths to be determined.
For longitudinal tests, the strain will be nearly uniform across the weld metal, HAZ, and base metal.
Differences in response to the applied strain may result in stresses varying across the cross section. Only
ultimate strength is commonly measured.
Testing standards may need to be varied for some specific geometries. For instance, girth welded tubes of less
than 75 mm (3 in.) diameter are commonly tested in the form of tubes with central plugs at the grips. The weld
is placed at the center of the gage length between the grips. The additional constraint induced by the hoop
direction continuity tends to increase the measured strengths and decrease the measured ductilities for tube
welds tested in this manner compared to a similar joint between flat sheets.
Shear Testing of Fillet Welds
Shear strength tests for fillet welds are described in AWS B4.0 for two orientations of fillet welds (transverse to

the tension loading and longitudinal to the tension loading) (Ref 8, 9). The transverse specimen is a double lap
specimen with loaded fillet welds, as shown in Fig. 1. The longitudinal specimen is a combination of two
lapped shear plates that are tack welded back to back, as shown in Fig. 2. These geometries are chosen to avoid
rotation during loading. The longitudinal specimen requires machining of grooves after the fillet welds are
made. The base plate is cut under the center of the lapped plate. The lapped plate is cut near each end so that
each length of weld connecting the base plate to the lapped plate across the gap in the base plate is 38 mm (1
in.).

Fig. 1 Transverse fillet weld shear test specimen. Source: Ref 9

Fig. 2 Longitudinal fillet weld shear test specimen. Source: Ref 8
Fillet-weld strength tests are sensitive to surface contour of the welds and to the condition of the weld root.
Excessive gaps between the lapped plates should be avoided, because these tend to magnify stresses at the weld
root. The specimens are also sensitive to underbead cracking and undercut.
Fillet size is most accurately measured after failure in the test. The stress is calculated based on assuming
uniform stress across the entire weld throat.
Bending Strength and Ductility
Bend tests are commonly used to evaluate the acceptability of weld procedures for providing sound welds (Ref
10). They allow rapid determination of strength and ductility on a specimen substantially simpler than the
standard tensile specimen. Bend tests tend to provide vivid demonstrations of difference between welds with
surface or near-surface flaws and welds without flaws adjacent to the convex surface of the bend. Bend tests are
further described in the article “Bend Testing” in this Volume.
Bending ductility can be calculated by determining the radius of the outer surface of the bend specimen at the
completion of the test. This ductility is usually smaller than that measured in a uniaxial tensile test. The bend
ductility is localized at the outer surface of the specimen, and the constraint is more severe because of the shear
stresses generated through the thickness of the bend specimen.
The thickness of the specimens and the size of the plunger or mandrel determine the outer surface ductility
requirement. Table 1 provides a summary of the radii of plungers or mandrels and the maximum test specimen
thicknesses for several groups of materials as required by the American Society of Mechanical Engineers
(ASME) Boiler and Pressure Vessel Code (Ref 11).

Table 1 Bend test geometry for testing based on material thickness and material type
Radius of
plunger or

mandrel
Maximum test

thickness
Materials
2t
in.
With >20% elongation
(3 + )t
in.
Alloy steels with <20% elongation
High strength Al alloys
4t
in.
Ti or Ti alloy with strength <65 ksi
5t
in.
Ti or Ti alloy with strength ≥65 ksi
Zr or Zr alloy
(8 + )t in.
4000 series Al alloy
Al alloy welded with 4000 series electrodes
Cu base alloys with Al and <20% elongation

(a)

(a)
Other alloys with <20% elongation
t, material thickness.
(a) Radius of plunger or mandrel and maximum test thickness chosen to achieve the required minimum
elongation from a base metal tensile test at the outer fiber of the convex surface of the bent specimen
Increasing specimen width can increase the constraint and reduce the likelihood of achieving the required bend
test radius without cracking. Face bends for welds in plates of greater than 38 mm (1 in.) thickness may
require multiple specimens.
Root, Side, and Face Bends. The primary geometries for bend test specimens place the butt weld so that the
bending stress is transverse to the weld axis. Different areas of the welds reach the highest bending stress in
transverse root bends, transverse face bends, and transverse side bends. Root bends put the weld-root side of the
tested butt weld on the convex side of the bend specimen. Face bends put the weld-cap side of the tested butt
weld on the convex side of the bend specimen. Side bends put a cross section of the weld on the convex side of
the bend specimen.
Longitudinal bend tests may sometimes be used to replace transverse tests, particularly when the strengths of
the regions within the specimen differ greatly. However, longitudinal side bends are not possible since the weld
cross section does not include the longitudinal direction. Longitudinal root bends, with the convex side of the
bent specimen on the weld-root side, and longitudinal face bends, with the convex side of the bent specimen on
the weld-cap side, can both be made and tested.
Longitudinal tests are less likely than transverse tests to fail from flaws that are long in the same direction as the
weld.
Wrap-Around Bend Testing. While the plunger-type bend fixtures are by far the most widely used for guided
bend testing, some circumstances require a fixture that creates a different distribution of strain. The most
common is a wrap-around testing fixture. Both the plunger type and the wrap-around type force the material
into a specified radius. However, the wrap-around fixture moves the points of bending load application around
a mandrel rather than using a fixed location for the central force, as shown in Fig. 3.

Fig. 3 Wrap-around bend testing. Source: Ref 9
The wrap-around fixture is most commonly used for welds that have significant mismatch between base metal
strengths, between base metal and weld metal strengths, or where the HAZ strength differs greatly from the

weld or base metal. If such welds are tested in a plunger-type fixture, the strain can be concentrated into the
lower-strength material, leaving a sharper bend in that material and much less bending in the higher-strength
material. The wrap-around fixture forces a more uniform strain into the materials because the loading point in
the center of the specimen moves across the weld.
Wrap-around test fixtures are commonly used for aluminum alloys where the strength of the weld metal and
HAZ may be chosen to be lower than the base metal. For instance, 6061-T6 aluminum base metal that is
welded with 4043 electrodes has a minimum yield strength of 240 MPa (35 ksi) in the base metal but only 100
MPa (15 ksi) for cross-weld tensile specimens (Ref 12). Wrap-around testing can limit the localization of the
strain at the weld.
Hardness
Hardness testing of welded joints is widely used as a rapid measurement of mechanical properties across the
varying microstructures of the welded region. It allows local regions and individual microstructures to be
compared for strength, because strength is correlated to hardness. A further discussion of hardness testing can
be found in the Section “Hardness Testing” in this Volume.
Hardness has been primarily related to the tensile strength rather than to the yield strength or the ductility.
Standard conversion charts are available for conversion of one hardness measurement to another and from
hardness to tensile strength measurement. Such converted information should be used with caution, because the
variation of weld microstructure may cause the average hardness to correspond to values that cannot be
obtained in larger scale specimens.
Macrohardness testing of welds requires preparation of a small region of the surface. The major techniques are
Brinell testing, which uses a spherical indenter, and Rockwell testing, which uses a diamond penetrator or a
sphere. The Brinell indentation is typically 2 to 6 mm in diameter while the Rockwell indentation is much
smaller but still is visible, unaided. Rockwell methods use several different loads for different hardness scales,
so it is possible for a weld to require different hardness scales for different regions.
Macrohardness testing results can be limited by the microstructural gradients around the welds. A result of 240
HB may represent a hardness for one uniform microstructure or an average over the regions deformed by the
indenter. Welds and HAZs often have gradients of microstructure and chemistry that can cause variations in
hardness across the indentation. Interpretation of the hardness from the impression may be made more difficult
if there is a large gradient in the hardness of the material under the indenter. This can result in noncircular
Brinell impressions and Rockwell tests with the deepest point not under the deepest point of the indenter.

Microindentation hardness testing using an indenter requires an even smaller region of the surface to be used
than macrohardness testing, but the surface preparation requirements are more stringent. Thus the Knoop and
Vickers microindentation hardness tests are primarily applied to ground and polished cross sections or to
ground, polished, and etched cross sections. Microindentation hardness traverses are often used to determine
the variation of hardness within the weld, across the fusion line, and across the HAZ.
Impact Toughness
Several methods are available for measuring the material resistance to starting and growing cracks that can be
applied to welded joints. This section discusses test methods that cause a crack to grow from a notch under the
rapid load of an impact. Methods that use sharp crack tips and thus can apply the loading more slowly are
discussed in the next section on fracture toughness.
Charpy. The Charpy V-notch impact test is the most common measurement method for fracture toughness of
welded joints. Specifications for the test are given in ASTM E 23 (Ref 13) and AWS B4.0. The test uses a
pendulum hammer to rapidly fracture a notched bar with dimensions of 55 mm by 10 mm by 10 mm (2.165 in.
by 0.394 in. by 0.394 in.).
Several measures of toughness can be obtained from a Charpy test. Absorbed energy, measured in ft · lbf or
joules, is the most commonly reported, but the percent shear fracture and the lateral expansion in inches or
millimeters are also sometimes reported. Greater toughness material will have higher values of each of these
three parameters. Occasionally, percent fibrous fracture, which is 100% minus the percent shear fracture, is
reported.
Many metals, including carbon and alloy steels, have toughnesses that vary strongly with temperature. So tests
on welded joints are often conducted at several temperatures, and the absorbed energy or other parameter is
plotted as a function of temperature. Material specifications and weld qualifications that include Charpy V-
notch testing normally require a minimum absorbed energy at a particular temperature. In this case, testing is
routinely conducted only at the temperature of interest.
The choice of minimum absorbed energy and test temperature are often varied between standards or within a
standard, based on service conditions. For instance, welded joints on bridges to be used in cold climates are
qualified to lower temperatures than those used in warm climates.
The absorbed energy in a Charpy V-notch test includes both the energy to start the crack from the 0.25 mm
(0.010 in.) radius notch and the energy to propagate the crack across the Charpy specimen. For many cases,
including constructional steels, these two parts are of comparable magnitude. In fact, the popularity of the

Charpy V-notch test was originally based on its ability to predict both crack initiation and crack arrest in ship
steel plates. This means that both the metal microstructure at the notch tip and through the specimen thickness
contribute to the reported toughness. For welded joints with heterogeneous microstructures, the position of the
notch tip will be important in determining the measured absorbed energy. The absorbed energy, however, will
also depend on the microstructures through which the fracture passes.
The dependence of Charpy impact test results on microstructure for many metals causes weld joints with
heterogeneous microstructures to have a range of Charpy values depending on specimen orientation in the weld
and notch position. Often weld centerline values are reported or compared with standards. Sometimes the HAZ
is tested at a particular location, such as 1 mm from the fusion line. These tests cannot determine a toughness
appropriate to all microstructures in the weld or HAZ. Additional tests of a greater variety of specimens may
reveal zones of lower toughness, such as unrefined columnar weld metal or coarse-grained HAZ, or zones or
higher toughness, such as reheated weld metal or fine-grained HAZ.
Subsize Charpy specimens are sometimes taken from thin material or areas where the geometry prevents a full-
size specimen. Only one dimension is reduced, the distance from the notched face to the unnotched surface
opposite. Reductions of this dimension can be from 10 mm (0.394 in.) to 7.5 mm (0.296 in.), called three-
quarter size; to 5 mm (0.197 in.), called half-size; and to 2.5 mm (0.099 in.), called quarter-size. These are the
most common reductions. Reduced thickness Charpy tests can be used to test the toughness of the root or cap
regions of fillet welds.
Charpy toughness test specimens can be taken from welded joints in several orientations. These orientations can
be given two-letter designations to show the orientation. The first letter is the direction normal to the crack
plane (the long direction of the Charpy specimen), while the second letter is the direction in which the crack
will propagate. The letter designations are L, longitudinal direction; T, long transverse direction (the weld width
direction); and S, short transverse direction (the through thickness direction). The letter designations are shown
for compact tension specimens in Fig. 4. Care should be taken that the orientation letters describe the weld area,
because different combinations of these letters may apply to the same orientation of specimen in base metal.
For instance, in a girth weld in a pipe, the long direction of the weld is the hoop direction of the pipe, not the
longitudinal or axial direction of the pipe.

Fig. 4 Orientations of toughness specimens in relation to welds. L, longitudinal direction; T, long
transverse direction (weld width direction); S, short transverse direction (weld thickness direction). In

the two-letter code for specimen designation, the first letter designates the direction normal to the crack
plane, and the second letter designates the expected direction of the crack plane. Source: Ref 9
Nil-Ductility Temperature. Drop weight tests use a notched weld bead as the starting point for a crack. The test
determines the ability of the base metal to arrest the crack running from an overlay of brittle weld metal. The
test results do not describe the properties of the overlay weld metal, so the overlay weld metal is standardized.
A brittle hard-facing alloy with good surface adhesion is used as the crack starter.
ASTM E 208 describes the test procedure (Ref 14). A standard weight is dropped onto test specimens at
different temperatures. The lowest temperature without full-section fracture is determined as the nil-ductility
temperature (NDT).
Weld metal can be tested for crack arrest by placing the notched weld overlay across a machined butt weld of
the weld metal of interest. The notch is typically placed so its long direction is above the longitudinal direction
of the butt weld.
Fracture Toughness
Fracture toughness testing of welded joints introduces several complications to standard fracture toughness
measurement as described in the Section “Impact Toughness Testing and Fracture Mechanics” in this Volume.
The weld and adjacent HAZ will have heterogeneous microstructures that can have widely varying strength and
toughness. In addition, welding residual stresses may be retained.