1.

0 FUNDAMENTALS OF FRACTURE MECHANICS

1.1 Introduction
1.2 Theoretical stress for fracture
1.3 Stresses due to notches and cracks
1.4 Energy balance approach to fracture
1.5 Stress field approach to fracture
1.6 Real materials and the effect of crack tip plasticity
1.7 Linear elastic fracture mechanics (LEFM)
1.8 Yielding fracture mechanics (PYFM)

i
 MSc SUBSEA ENGINEERING – Subsea Integrity

1.0 FUNDAMENTALS OF FRACTURE MECHANICS

1.1 Introduction

There are many discernible modes of failure of components and structures and fracture is an important
one. Rapid fracture can occur under loading that does not vary with time or which changes only slowly -
static loading. If a fracture of this type is accompanied by little plastic deformation it is termed brittle
fracture. If the load is applied very rapidly, termed impact loading, brittle fracture is more likely to
occur than for static loading.

The phenomenon of brittle fracture was vividly demonstrated with the advent of the all-welded Liberty
ships in the 1940s and gave rise to the development of the methodology known as fracture mechanics.
Failure of the Liberty ships demonstrated that brittle fracture could occur suddenly and catastrophically
at stresses below the yield stress of the material. Fracture mechanics, therefore, studies the behaviour of
cracks in materials and identifies resistance to failure by fracture in the presence of a crack by various
material properties. Resistance to brittle fracture of such materials can be described by the methodology
of linear elastic fracture mechanics (LEFM) and is measured by a material property called the fracture
toughness, designated KIC.

Generally, materials which have high yield strength have a low resistance to fracture with low values of
fracture toughness (Fig.1).

Figure 1: Effect of strength level on fracture toughness

When appreciable plastic deformation accompanies the fracture process, ductile fracture can occur,
which can often involve a gradual tearing of the crack in the material. In this case, KIC no longer
characterises resistance to fracture and the methodology of post yield or elastic - plastic fracture
mechanics (EPFM) must be invoked.
In addition, fracture can result due to a combination mechanical and chemical effects termed
environmental or stress corrosion cracking and under the influence of time dependent loads at elevated
temperature, termed creep rupture. While some of the characterising parameters are the same for these
phenomena, full treatment is beyond the scope of the present course.
 MSc SUBSEA ENGINEERING – Subsea Integrity

1.2 Theoretical stress for fracture

From consideration of the bonding between metallic atoms, a theoretical fracture stress, σ max , can be
identified as:

1/ 2
⎛ Eγ ⎞
σ max = ⎜⎜ ⎟⎟ (1)
⎝ ao ⎠

where a o is the equilibrium spacing between the atomic lattice planes under the applied force and γ is
the surface energy of the material.

However, when values of σ max are compared with actual values of fracture stress they are found to be
too high by some two orders of magnitude.

Griffith suggested that this was due to the presence of small defects in the material, the calculation of the
theoretical value of fracture stress assuming a defect free material.

1.3 Stresses due to notches and cracks

Before embarking on a discussion of the effect of cracks, or more generally crack-like defects, on
structures and components it is instructive to examine the effect of discontinuities such as notches as
stress raisers since it was observed that local discontinuities were an essential feature of brittle fracture.

1.3.1 Notches and stress concentration

A notch can be defined as a geometrical discontinuity having a definite depth and root radius. On a large
scale this may be a necessary design feature such as a thread root, change in section, weld toe etc.
However, on a smaller scale, slag inclusions, porosity and second phase particles also form stress
concentrations.

The effect of a notch on a tensile stress field can be visualised by considering a 'stress lines' analogy
(Fig.2) where concentration of supposed stress 'lines' occurs at the ends of an imposed notch lying
normal to the uniform tensile stress.

Figure 2: Effect of a notch on a tensile stress field

The sharper and smaller the notch, the more severe the bunching of the 'stress lines'.
 MSc SUBSEA ENGINEERING – Subsea Integrity

If an elliptical hole in a plate is considered, as shown in Figure 3(a), where the hole is assumed small
compared to the width of the plate, and is aligned with its major axis perpendicular to a remote, uniform
tensile stress, σnom, the stress field in the vicinity of the hole is altered as shown in Figure 3(b).

Figure 3(a) 3(b): Stress field for an elliptical hole in

an infinite plate

The major effect of the hole is on the stress σnom - this stress is raised, or concentrated, at the edge of
the hole and decays along the x-axis to the nominal value, far from the hole. The maximum value
developed is dependent on the aspect ratio and radius of the hole. Thus

σy = σnom 1+2c/d = σnom1+2 (c/r)½ (2)

The stress concentration factor, Kt, for the ellipse is defined as the ratio σmax/σnom, as

Kt = σmax/σnom = 1+2c/d = 1+2 (c/r)½ (3)

It can be noted that if the hole is circular, i.e. c = d, then the stress concentration factor reduces to the
classical value of 3 for a hole in a plate.

If the dimensions of the elliptical hole are now reduced such that the radius and semi-minor axis become
vanishingly small, the resulting slit can be considered as an idealised crack and, from the preceding
discussion, it is clear that the associated stress concentration will be severe. Indeed, if the crack is
considered as ideally sharp, the elevated stress at the crack tip becomes infinitely large and can be shown
to be given by
 MSc SUBSEA ENGINEERING – Subsea Integrity

1/ 2
⎛ Eγ ⎞
σ max =⎜ ⎟ (4)
⎝ 4a ⎠

where, a , is the half-crack length.

Thus, the local stresses in the vicinity tip will certainly exceed the material's yield stress, even for
modest applied loads. However, because the stress gradient ahead of the concentrating feature is steep,
plasticity is limited to a small region thus avoiding plastic collapse. Clearly, an infinite stress cannot
exist in a real material and the notion of a stress concentration for a crack breaks down since it cannot
distinguish between various crack lengths and applied stress levels.

1.4 The energy balance approach to fracture

Early work on the fracture of brittle materials by A A Griffith focused on energy based analytical
methods and, interestingly, current thinking has concentrated on fracture characterising parameters
related to this philosophy.

The basic idea was that a system under load will move to a state of lower free energy so that a crack
would only grow if there was a decrease in the free energy of the system comprising the cracked body
and the loading mechanism.
 MSc SUBSEA ENGINEERING – Subsea Integrity

1.4.1 Cracks and strain energy release rate (G)

If the cracked member shown in Figure 4 is loaded under a remote tensile load, P, the load/displacement
behaviour will be linear assuming linear-elastic behaviour of the material, Figure 5.

Figure 4: Cracked body under tensile load Figure 5: Load - displacement

curves

As the load P is applied, the loading points will move apart by a distance, u, and elastic stored energy,
UE, will be stored in the body given by

UE = 1/2Pu (5)

Note that the load and displacement are related by the compliance, C, as

U = CP (6)

It is now possible to examine the changes in energy which occur as the crack is extended by an
infinitesimal amount from length, a , to length, (a + δa ) . As new crack surfaces are produced, there will
be a surface energy associated with these. The other energy terms have to be considered at two bounding
conditions, constant displacement and constant load.

Constant displacement

An increase in crack length will result in a decrease in stored elastic energy of magnitude, 1/2(P1-P2)u1.

Constant load

In this case, the strain energy is greater (1/2P1u2) after crack extension but the load must move through
a distance (u2-u1) resulting in work being done on the system. The total change in potential energy is,
therefore, a decrease of magnitude, P1(u2-u1) - 1/2P1(u2-u1) which equals, 1/2P1(u2-u1).

For infinitesimally small increases in crack length, the compliance, C, remains essentially constant, so
that
 MSc SUBSEA ENGINEERING – Subsea Integrity

δu = CδP (7)

Also, we can define (P1-P2) as, δP , and (u2-u1) as, δu .

Thus, using these definitions, the strain energy release under constant displacement is given by
− 1 / 2uδP and the potential energy release under constant load is given by, − 1 / 2 Pδu , for crack
extension from length, a , to (a + δa ) .

Now, substituting for the compliance in the respective energy equations, the strain energy release and the
potential energy release are identical as, − 1 / 2CPδP .

Thus, the driving force for crack extension is the difference between the energy which could be released
if the crack were extended and that needed to create new surfaces.

Now, this change in energy can be expressed as a function of increase in crack area and is termed the
strain energy release rate, G. Thus,

G = −1 / B.dU / da (8)

G, therefore, characterises the energy per unit area necessary to extend the crack and, as such, can be
thought of as a parameter characterising the behaviour of the crack.

Developing this approach, Griffith was able to show that a fracture stress, σ c , could be derived as

1/ 2
⎛ 2 Eγ ⎞
σc = ⎜ ⎟ (9)
⎝ πa ⎠

The concept was developed originally for brittle materials such as glass where essentially all of the
potential energy released on crack growth was assumed available for creation of the new crack surfaces.
While this holds true, approximately, for materials showing little local plastic deformation in the
presence of cracks, it does not apply to most engineering materials which are more ductile. For these
cases, involving metals, significant amounts of the released potential energy will be absorbed in causing
localised crack tip plastic deformation. However, the general concept has been shown to be applicable to
metals provided local crack tip plastic deformation is small.

Early attempts to account for the local plastic deformation at the crack tip modified Equation (9) with a
'plastic work' term as

⎛ 2 E [γ + P ] ⎞
1/ 2

σc = ⎜ ⎟ (10)
⎝ πa ⎠

Determination of either γ or P is not practicable and Irwin showed that the [γ + P ] term was
equivalent to G, the strain energy release rate, so that

1/ 2
⎛ EG ⎞
σc = ⎜ ⎟ (11)
⎝ πa ⎠
He also showed that G could be measured for a cracked body by means of compliance calibrations and
that a value, Gc, could be identified at the onset of fracture. This approach, therefore, provided the basis
of a practical fracture methodology.
 MSc SUBSEA ENGINEERING – Subsea Integrity

1.5 The stress field approach to fracture

To develop the energy release concept completely, it is necessary to compute the actual rate of energy
release for the cracked body which would occur on crack growth for comparison with Gc. Thus, it is
necessary to compute the distribution of stress and strain around the crack.

1.5.1 The stress intensity factor (K)

While the energy associated with the initial propagation of a crack is of some value in understanding
why cracks grow, energy itself is not a parameter readily conducive to design and, for general use,
stresses around a crack are of more intuitive use. In 1957 Irwin developed a series of linear elastic crack
stress field solutions assuming a crack having a boundary defined by a simple curve or straight line, and
crack extension in the crack plane. It was shown that the stress field at the tip of a crack was
characterised by a stress singularity which decreases in proportion to the inverse square root of the
distance from the crack tip.

Now, any crack can be loaded in any one of or a combination of ways, as shown in Figure 6.

Figure 6: Loading modes for cracked bodies

A co-ordinate system for describing the stresses in the vicinity of a crack is shown in Figure 7.
 MSc SUBSEA ENGINEERING – Subsea Integrity

Figure 7: Coordinate system for crack tip stresses

The polar co-ordinates r and θ lie in the x-y plane, which is normal to the plane of the crack. The z-
direction is parallel to the front edge of the crack.

The stresses near the crack tip, for Mode I loading depend on r and θ -

σ (a) 1/2
σx = cosθ / 2[1- sinθ / 2 sin3θ / 2] (12)
( 2r ) 1 2
σ (a) 1/2
σy = cosθ / 2[1- sinθ / 2 sin3θ / 2]
( 2r ) 1 2

σ (a) 1/2
τ xy = sinθ / 2 cos θ / 2 cos3θ / 2 (13)
( 2r) 1 2
σz = 0 - plane stress

σz = ν(σx + σy) - plane strain; ez = 0

τyz = τzx = 0

These equations are based on the theory of linear elasticity and describe the stress field near the crack
tip. It will be noticed that the higher order terms are omitted as they do not markedly affect the
magnitude of the stresses near the crack tip. Several points can be observed from the crack tip stress
equations:
(i) all the local stress components are singular, i.e. increase to infinite values as the crack tip is
approached,
(ii) the spatial distribution is a function of θ only,
(iii) The magnitude of the stresses at any given distance from the crack tip is scaled by the term
σ (πa )1 / 2 , i.e. a simple function of the remotely applied stress and the crack length.

Thus, a stress intensity factor can be defined as,

K = σ (πa )
1/ 2
(14)

and the crack tip stresses can be rewritten as;

KI
σx = cosθ / 2 [1 - sinθ / 2 sin3θ / 2] (15)
( 2πr ) 1 2

KI
σy = cosθ / 2 [1 - sinθ / 2 sin3θ / 2]
( 2πr ) 1 2

KI
τ xy = sinθ / 2 cos θ / 2 cos3θ / 2 (16)
( 2πr ) 1 2

where θ < r << a , and the subscript I refers to Mode I loading. The magnitude of the stress field near
to the crack tip can be characterised by giving the value of the factor KI.
 MSc SUBSEA ENGINEERING – Subsea Integrity

On this basis, KI is a measure of the severity of the crack and it can be defined, formally, as,

KI = limr,θ->0[σy(2πr)1/2] (17)

Similar equations exist for cracks loaded in Mode II and Mode III with the respective values of stress
intensity described by KII and KIII.

The stress intensity is generally expressed, conveniently, as:

K I = Yσ (πa )
1/ 2
(18)

where the factor Y accounts for different geometries.

For the class of fractures commonly referred to as brittle it is usually assumed that Mode I is the
predominant mode and that KII = KIII = 0.

The advantage of the stress intensity approach is that (along the x-axis, θ = 0), the crack tip region
knows only the stress field and it is immaterial whether this has been produced by an increase in σ or in
crack length - the K factor conveniently embraces both. It follows that if two bodies containing cracks
have identical values of K then the stress fields in the crack tip region will be identical.

Thus, the value of the stress intensity factor 'characterises' the crack and can be considered as a
material parameter which defines the redistribution of stress in a body due to the presence of a
crack.

Note: The stress intensity factor K has no connection with the elastic stress concentration factor,
Kt.

1.5.2 Effect of finite boundaries

The expression for the stress distribution ahead of a sharp crack in Equation (14) holds only for an
infinite body.

If finite surfaces are present, the infinite plate solution is modified by an algebraic, trigonometric or
polynomial function, chosen to make the surface forces zero.

Thus, the resultant value of K, as the dominant term in a series expansion, is then not equal to
σ (πa )1 / 2 but is amended to take account of the width, W, of the test piece or structure.

The simplest geometry factor is that for an edge crack of length, a, in a semi-infinite plate: the increase
in ability of the crack to open causes the stress intensity factor to increase by some 12% to become:
K I = 1.12σ (πa )
1/ 2
(19)

Examples of such K solutions are shown in Figure 8 for some common loading geometries:
 MSc SUBSEA ENGINEERING – Subsea Integrity

Figure 8: K solutions for a range of loading geometries

1.5.3 Relationship between K and G

We have shown that the elastic energy release rate, G, for infinitesimal crack extension has the
magnitude equal to dU/da for constant load and constant displacement. Thus, G can be obtained from
Equation (11) for a crack of half length, a:
 MSc SUBSEA ENGINEERING – Subsea Integrity

For plane stress:

πσ 2a
dU / da = G = (20)
E

For plane strain:

(1- ν2 ) πσ 2a
dU / da = G = (21)
E

By substituting K I = σ (πa )
1/ 2
in Equations (20) and (21), the relationship between G and K can be
obtained:

For plane stress:

K I2
G= (22)
E

and for plane strain:

( )
G = K I2 1 − ν 2 / E (23)

Thus, for linear elastic conditions, the prediction of crack growth and fracture is identical for both the
energy balance and stress field approaches.
 MSc SUBSEA ENGINEERING – Subsea Integrity

1.6 Real materials and the effect of crack tip plasticity

Since structural materials deform plastically (permanently) above the yield stress, there will in reality be
a plastic zone surrounding the crack tip and so an elastic stress solution is not unconditionally applicable.

Along the x-axis, θ = 0, and the expression for σy in Equations (15) gives:

σ (πa )1 / 2 KI
σ yy = = (24)
(2πr )1/ 2
(2πr )1 / 2
By substituting for the material yield stress, σy, for σyy in Equation (22), an estimate can be obtained of
the distance ry over which the material plastically deforms ahead of the crack tip.

2
1 ⎛ KI ⎞
ry = ⎜⎜ ⎟⎟ (25)
2π ⎝ σy ⎠
As a first estimation, ry can be taken as the diameter of a circular plastic zone and the distribution of
stress σyy ahead of the crack are as shown in Figure 9.

Figure 9: First estimate of plastic zone size

Since part of the stress distribution is truncated at the value of the yield stress, this is clearly inaccurate.

Indeed, finite element analysis has also shown that the plastic zone is not circular in shape.

One further difficulty in describing the plastic zone size and shape is that the state of stress i.e. plane
stress or plane strain, will affect it. It is known that under plane strain conditions yielding need not occur
until the applied stress is much higher than σy i.e. the plastic zone size may be smaller.

A second approximation, due to Irwin, of the plastic zone size can be made by attempting to account for
the fact that the stress distribution cannot simply be cut off above σy. To simplify the analysis several
restrictions are required.

(i) plastic zone is circular

(ii) only the case of the x-axis i.e. θ = 0, is analysed
 MSc SUBSEA ENGINEERING – Subsea Integrity

(iii) the material is assumed to be elastic-perfectly plastic i.e. stress cannot exceed σy.

It was argued that the occurrence of plastic deformation made the crack behave as if it were longer - i.e.
the displacements are longer and the stiffness lower, than in the elastic case.

So an effective crack size, a eff = a∆a n can be identified, where ∆a n accounts for the redistribution of
stresses that were above σy i.e. the stress distribution is not cut off at σy as in the first approximation
(Figure 10).

Figure 10: Second estimate of plastic zone size

Thus, plastic zone size is twice that obtained from the first estimation.

1.6.1 Plane stress and plane strain

For a simple through-thickness crack in a plate, at least, plane stress (biaxial) conditions exist, Equation
(15).

A section through the plate in the plane of the crack is shown in Figure (11).

Figure 11: Section through crack plane

The material which deforms plastically ought to be able to contract in the thickness direction but the
adjacent, elastic material constrains this plastic material as it cannot contract to the same extent. The
result is tensile stresses at the plastic zone boundary and in the thro-thickness direction. Thus, a triaxial
stress state is set up. Clearly, at the plate surface, there are no stresses in the thickness direction and no
plane stress exists here. As the centre of the plate is approached the position changes to plane strain and
the plastic zone size alters accordingly (Figure 12).
 MSc SUBSEA ENGINEERING – Subsea Integrity

Figure 12: Variation of plastic zone size

1.6.2 Effects of stress state on fracture behaviour

The typical behaviour of a cracked component or sample is shown in Figure 13.

Figure 13: Behaviour of a cracked component

The crack extension is initially flat but is quickly accompanied by small amounts of shearing at the
edges which widen to cover the entire surface. Final fracture takes place by slanted or double shear
behaviour.

The initial flat fracture is attributable to plane strain loading, due to a "hinge-type" mechanism (Figure
14(a)). On the other hand, the slanted, shear fracture (plane stress) takes place by a different mechanism
(Figure 14(b)).
 MSc SUBSEA ENGINEERING – Subsea Integrity

Figure 14 (a): Flat fracture mechanism Figure 14 (b): Shear fracture

mechanism

Such information can be very useful in investigation of structural failures.

 MSc SUBSEA ENGINEERING – Subsea Integrity

1.7 LINEAR ELASTIC FRACTURE MECHANICS (LEFM)

1.7.1 Introduction
A fracture criterion can be based on functions of the elastic stress components near the crack tips. No
matter what function is assumed, it is implied that K reaches some critical value since each stress
component is uniquely determined by K. In other words, the crack will become unstable when K reaches
a critical value, KIC, which can be used to classify and compare materials which fail by brittle fracture.

1.7.2 Concept of fracture toughness

Because of the thickness effect influencing deformation behaviour, the critical stress intensity KC will
depend on specimen thickness.

Under plane strain conditions, the value of KC tends to a limiting, constant value, KIC, for a given
material (Figure 15).

Figure 15: Effect of thickness on critical stress intensity

 MSc SUBSEA ENGINEERING – Subsea Integrity

1.7.3 Measurement of fracture toughness

From a great deal of testing and examination of test variables during the 1960s, a standardised test
procedure resulted. The purpose of the fracture toughness test is to make a valid measurement of KIC
using material in the same condition as will be used in service. This will be particularly important if heat
treatment is carried out since KIC is dependent upon metallurgical condition. The main standards
specifying the KIC test method are

• BS 7448: Part 1: 1991

• ASTM E399: 1978

both of which allow the testing of a fatigue, pre-cracked single edge notched bend or compact tension
specimen (Figure 16).

Figure 16: Standard samples for KIC testing

The recommended test procedures have shown that these samples enable KIC values to be reproducible
within about 15% by different laboratories.

The accuracy with which KIC describes the fracture behaviour depends on the degree to which the stress
intensity characterises the state of stress and strain ahead of the crack tip since it is in this region that
fracture will initiate. Thus, certain geometrical restrictions are placed on the major specimen dimensions
so that they will be large with respect to the plastic zone size. These dimensions are:

2
⎛ K IC ⎞
• crack length a ≥ 2.5 ⎜⎜ ⎟⎟
⎝ σy ⎠
 MSc SUBSEA ENGINEERING – Subsea Integrity

2
⎛ K IC ⎞
• specimen thickness B ≥ 2.5 ⎜⎜ ⎟⎟ (26)
⎝ σy ⎠

2
⎛ K IC ⎞
• specimen width W ≥ 2.5 ⎜⎜ ⎟⎟
⎝ σy ⎠
Due to the nature of these expressions, provisional values of toughness, KQ, are determined prior to
assuring the validity of the test and confirming that KIC has been measured.

1.7.4 Factors affecting KIC

Effect of material type

For engineering materials, values of KIC are generally in the range of 20 to 200 MPam1/2 and for a
given class of material, fracture toughness generally decreases with increasing strength along with
tensile ductility. Values of KIC for a range of materials are shown in Table 1.
 MSc SUBSEA ENGINEERING – Subsea Integrity

Material Fracture toughness (MPam1/2)

Pure ductile metals (e.g.Cu, Ni) 100 - 350
Rotor steels (A433, discalloy) 204 - 214
Pressure vessel steels (HY100) 170
High strength steels 50 - 140
Mild steels 140
Titanium alloys 55 - 115
Aluminium alloys 23 - 45
Medium carbon steels 50
Steel reinforced concrete 10 - 14
Cast iron 6 - 20
Cobalt/tungsten carbide cermets 14 -16
Silicon nitride 4–5
Unreinforced cement/concrete 0.2
Nylon 3
CFRPs 32 - 45
GFRPs 20 - 60
Fibreglass (glassfibre epoxy) 42 - 60

Table 1: KIC data for a range of materials

Effects of temperature and loading rate

Fracture toughness of materials generally increases with increasing temperature, an abrupt change
occurring for ferritic steels. This temperature transition behaviour is similar to that observed in the
Charpy impact test in these steels. The shift in toughness is due to a difference in fracture mechanism
with temperature, being predominantly cleavage fracture at low temperature and dimpled rupture above
the transition temperature.

Data on the effect of temperature on KIC is essential in selecting materials for a particular service
application since it is generally important to avoid the use of a material under high service loading at a
temperature where it has low toughness. Additionally, the transition region should also be avoided as
there is a tendency for scatter in KIC data here.

High loading rates tend to lower the fracture toughness, the effect being similar to a reduction in
temperature. It is interesting to note that notch impact tests tend to give a higher transition temperature
than KIC tests carried out at normal loading rates (Figure 17).
 MSc SUBSEA ENGINEERING – Subsea Integrity

Figure 17: Transition temperatures from Charpy and KIC tests

Effect of material microstructure

Apparently small changes in chemical composition or processing route can change the measured value
of fracture toughness. A good example is the effect of sulphur content in steels where the existence of
sulphide inclusions causes a decrease in toughness (Figure 18).

Figure 18: Effect of sulphur content on KIC

Fracture toughness is also generally more sensitive to orientation effects in worked material such as
rolled plate, forgings and extrusions. It is thus important to select or measure toughness values in the
appropriate direction for service loading.
 MSc SUBSEA ENGINEERING – Subsea Integrity

1.7.5 Application of linear elastic fracture mechanics

In practice there is no such thing as a 'defect free' material. Cracks or crack-like defects are inherent in
every component. The important point to ensure is that these cracks remain harmless and do not grow in
service to reach a critical size.

The adoption of LEFM is particularly useful, since it enables evaluation of the reliability of a structure
or component in service. Its use in design can assist in materials selection and to determine whether
defects are likely to be of a size to cause catastrophic fracture.

For the component or structure, from a knowledge of KIC, it should be possible to determine the applied
stress to cause failure for a given defect size or to predict the size of the defect necessary to cause failure
for a given applied stress.

In practice, the compliance function for a particular component or structure may not be known so that a
different approach for calculating critical defect sizes is required. Since many service flaws are surface
breaking and adopt or can be bounded by a semi-elliptical shape, the following analysis for an edge-
cracked plate is particularly useful.

For a surface defect, the critical size is:

a cr =
2
K IC[φ 2 − 0.212(σ / σ y ) 2
] (27)
1.21πσ 2

where φ is a double elliptical integral.

Let

⎡ 2 ⎛σ ⎞ ⎤
2

Q = ⎢φ − 0.212⎜ ⎟ ⎥ (28)
⎣⎢ ⎝ σ y ⎠ ⎦⎥

where Q is considered as a flaw shape parameter, allowing for the geometry of the flaw.

Thus, for a surface flaw:

{a / Q}cr = K IC
2
/ 1.21πσ 2 (29)

and for an embedded flaw:

{a / Q}cr = K IC
2
/ πσ 2 (30)

The relationship between Q and the elliptical aspect ratio, a / 2c , is shown in Figure 19.
 MSc SUBSEA ENGINEERING – Subsea Integrity

Figure 19: Flaw shape parameter curves

Thus, for defect analysis, it is necessary to know two of the following three parameters relative to the
component or structure being analysed:

* the material's fracture toughness, K IC

* the applied stress, σ

* the flaw aspect ratio, a / 2c

1.8 YIELDING FRACTURE MECHANICS

1.8.1 Introduction

LEFM was developed to determine the behaviour of cracked bodies under essentially elastic loading
conditions and can be applied only in situations where the stress field is dominated by the stress intensity
factor. However, there are many important structural materials which have low to medium strength and
high toughness which cannot be described adequately by LEFM.

e.g. Consider the nuclear pressure vessel steel ASTM A533B.

Typical values for this steel are:

σy = 500 MPa
KIC = 180 MPa m1/2

Using the requirements for specimen size, a test sample would have to be greater than 324mm thick in
order to measure a valid KIC. This is, clearly, outlandish but, more importantly, if the steel is not used
in service at these sizes, then KIC would not be a relevant characterising parameter.

The methods of yielding fracture mechanics extend the description of fracture behaviour beyond the
elastic region, up to the position where plastic collapse becomes the dominant failure mode, (Figure 20).
 MSc SUBSEA ENGINEERING – Subsea Integrity

Figure 20: Regimes for fracture behaviour

1.8.2 Plasticity correction factors

One approach which was used to compensate for the effects of extensive yielding was to treat the infinite
body in plane stress for situations with the plastic zone preceding catastrophic fracture being considered
as an additional contribution to the crack length.

Thus, for a plastic zone size of 2ry, where ry is given by:

ry = a / 2.(σ f / σ y )
2
(31)

( )
it was argued that the distribution of stress was equivalent to that for an elastic crack of length a + ry .

Basically, this showed that, if the amount of plasticity is significant, the value of K developed for a given
applied stress, s, is higher than the linear elastic relationship indicates.

1.8.3 Fracture characterising parameters under extensive plasticity

For such cases, a formal scheme of general yielding or elastic plastic fracture mechanics follows that of
LEFM in attempting to define a single parameter that measures the intensity of the deformations applied
to a sharp crack but admitting the presence of plasticity. The response of the material is measured
notionally by some critical value of the chosen parameter. The two most widely used characterising
parameters are the crack tip opening displacement (CTOD) and a term called the J-integral. These two
parameters are closely related, but have a different physical basis. The CTOD concept takes the
stretching or blunting at the crack tip as the measure of crack deformation whereas the J-integral uses an
averaged intensity of stress or strain around the crack tip.
 MSc SUBSEA ENGINEERING – Subsea Integrity

Crack tip opening displacement (CTOD)

In simple terms, the CTOD is the opening of the crack at a position corresponding to the original, sharp
crack tip (Figure 21) and a specific value of CTOD, δ i , characterises the onset of crack extension.

Figure 21: Physical definition of CTOD

It must be appreciated that total instability may not coincide with the attainment of δ i but may occur at
a higher value, δ c . This higher value will be associated with a finite amount of crack growth.
In plane stress, for values of applied stress less than the yield stress i.e. below general yield, δ can be
related to the applied stress, σ app by:

8σ y ⎡ ⎛ πσ app ⎞⎤
δ= a ln ⎢sec⎜ ⎟⎥ (32)
πE ⎢⎣ ⎜⎝ 2σ y ⎟⎥
⎠⎦

For low values of σ / σ y , the 'ln sec' term can be expanded to the first order as:

2
⎡ ⎛ πσ app ⎞⎤ ⎛ πσ ⎞
ln ⎢sec⎜ ⎟⎥ = ln 1 ⎜ app ⎟
⎢⎣ ⎜⎝ 2σ y ⎟⎥
⎠⎦ 2 ⎜⎝ 2σ y ⎟
⎠

π 2σ app
2

= ln (33)
8σ y2

Hence, for conditions of small scale yielding, Equation 33 becomes:

 MSc SUBSEA ENGINEERING – Subsea Integrity

σ app πa
δ=
Eσ y

K2
= (34)
Eσ y

G
or δ= (35)
σy

On the assumption that the stress σ remains constant at σ y in the plane stress yielded region and that
the crack tip displacement is constant at δ c , then:

G = σ yδ c (36)

The principle of CTOD testing is that a critical value, δ c , can be measured in a sample which has
developed extensive yielding and that this value of CTOD can be used to determine the toughness of a
structure where σ app / σ y is small, so that the relationship of Equation 36 can be utilised.

Measurement of CTOD

The measurement of CTOD is specified in the following standards:

• BS 7448: Part 1: 1991

• ASTM 1290: 1989

The purpose of the test procedure is to obtain an unambiguous estimate of the displacement at the tip of
the fatigue crack in a bend specimen at the onset of unstable fracture. The preferred test piece has a
thickness equal to that of the material to be tested.

To make the test practical, δ is determined using a formula employing the plastic displacement, Vp,
measured at or a distance, z above the crack mouth as shown in Figure 22.
 MSc SUBSEA ENGINEERING – Subsea Integrity

Figure 22: Measurement of Vp in a CTOD test

In addition, it is assumed that there is a plastic rotational factor of 0.4, i.e. that the bend sample hinges
around a fixed point 0.4 of the distance between the crack tip and the back face of the sample. The
elastic contribution to the CTOD, δ e , is estimated using the equivalent K value at the failure load,
assuming plane strain conditions. Thus,

K 2 / E ' = 2σ y δ e (37)

where (
E ' = E / 1 −ν 2 )
Thus,

δ=
(
K 2 1 −ν 2
+
)
0.4(W − a )V p
(38)
2σ y E (0.4W + 0.6a + z )
where W - is the specimen width
σ y - is the material yield strength
z - is the knife edge thickness
a - is the original crack length
ν - Poisson's ratio
It must be noted that the values of CTOD obtained do not refer, necessarily, to the physical opening
displacement at a particular point along the crack front, since the parameters Vp and K in Equation (38)
are influenced by the entire specimen behaviour and the true plastic rotational factor which can vary
between 0.3 and 0.5. However, these discrepancies are usually small and are offset by the advantage of
defining CTOD in terms of a relatively simple formula. The concept of defining the opening of the crack
at the crack tip makes the CTOD a conceptually simple fracture characterising parameter.

The standards simply specify the methods by which the critical values of CTOD, δ c , may be obtained.
For cleavage fracture, where fast fracture coincides with instability, this is fairly straightforward but in
other cases, it may be necessary to measure the initiation value, δ i , or a value, δ u , corresponding to a
 MSc SUBSEA ENGINEERING – Subsea Integrity

fixed amount of ductile crack growth. If no fracture instability occurs prior to the attainment of the
maximum load in the test piece, a value of δ m is recorded. It should be noted that while fracture
instability in a test piece can perhaps be related to fracture instability in a structure in service, it is
dangerous to suppose that the onset of plastic instability in the test piece ( δ m ) necessarily relates to any
critical event in a structure.

1.8.4 Factors affecting CTOD

Specimen geometry

Both the width and the thickness of the specimen may affect the measured value of CTOD as shown in
Figure 23.

Figure 23: Effect of geometry on CTOD

As the width of the specimen ligament is increased, fracture becomes more probable than failure by
plastic collapse and for temperatures at or below the transition temperature, increasing the specimen
thickness increases the probability of brittle behaviour. It is therefore recommended that the size of the
specimen should reflect the size of the section in service.

Temperature and strain rate

For ferritic steels, CTOD behaviour is similar to that of the Charpy transition curve with a sudden
ductile-to-brittle curve being observed, Figure 24.
 MSc SUBSEA ENGINEERING – Subsea Integrity

Figure 24: Effect of temperature and strain rate on CTOD

The effect of increased strain rate is equivalent to decreased temperature and so an upward shift in the
transition temperature is observed, Figure 24. These effects could be important for structures subjected
to dynamic loading.

1.8.5 CTOD and fracture toughness

It is possible, in principle, to determine the relationship between fracture stress and defect size in a
structure using the critical value of CTOD, which for plane stress is:

K c2 = EGc = Eσ y δ c (39)

Under conditions of plane strain no full analytical dependence of CTOD on applied stress has been
produced but under 'small scale' yielding the relationship:

δ = βK / σ y E (40)

has been proposed, with β in the range 0.45 to 0.7.

1.8.6 J - integral

The J-integral was derived from the analysis of non-linear elastic behaviour such as is observed in
rubbers where the stress-strain curve is non-linear. These materials still behave elastically; however, in
that loading, unloading and reloading follow the same curve.

The energy release associated with the notional crack extension can be characterised by a parameter, J,
which is the non-linear equivalent of the potential energy release rate, G, per unit thickness for linear
elastic materials. J is defined in a similar way as:

J = -dUtot/BdA (41)

where Utot is the total potential energy within the system. J is the non-linear equivalent to G and in a
linear elastic material, J is identical to G.
 MSc SUBSEA ENGINEERING – Subsea Integrity

Although the J-integral was developed strictly for materials which behave in a non-linear elastic manner
it has been applied to materials which behave in an elastic/plastic manner. The significant difference is
that the unloading curve will not follow the loading curve and so for an elastic/plastic material, J is a
characterising parameter which does not bear the full implication of energy release. Thus, J is defined as
the work input rate during the loading of a test piece containing a crack.

Measurement of J
Specifications for the measurement of J are contained in:

• BS 7448: Part 1: 1991

• ASTM E 813: 1991

Both standards allow testing of either SEN bend or CT test pieces with J being determined from:

J=
(
K 2 1 −ν 2
+
)
2U p
(42)
E B(W − a )

where Up is the area under the load/load line displacement curve derived from the test.

It should be noted that the ASTM Standard allows the evaluation of J approximating to the onset of
stable tearing, JIC, whereas, the British Standard identifies J definitions analogous to those for CTOD.

1.8.7 Relationship between CTOD and J

It can be shown that:

J = σ yδ t
and it was shown that for σ / σ y less than unity that:
δ t = K I2 / Eσ y or δ t = G / σ y

So
J = G = δ tσ y (43)

under LEFM conditions.

For elastic/plastic loading where appreciable strain hardening takes place, the relationship is modified to:

J = Mδ t σ y (44)

with M having a value between 1.15 and 2.95.

MSc SUBSEA ENGINEERING – Subsea Integrity

Page 1