Vous êtes sur la page 1sur 55



Response of a simple lattice to shear loading is shown in Figure 1. Initially the response is elastic, Figure 1(b), i.e. when the load is removed all the deformation is recovered. Beyond a limiting stress known as the yield stress, the deformation becomes elastic-plastic, Figure 1(c); when the load is removed only the elastic deformation reverses and the plastic deformation remains, Figure 1(d). This figure shows that crystallographic planes have moved against each

other. Simple equations for plastic deformation (corresponding to Hooke's law for elastic deformation) do not exist.

Theoretical estimates of the critical shear stress required to move a crystallographic plane one slip unit give crit ~ G/(2). Measurements show that the highest strength steels attain

approximately 10% of the theoretical strength. However structural steels have strengths which are two orders of magnitude less than the theoretical value. Therefore, there must be a mechanism which facilitates the slipping of crystallographic planes. This mechanism is the presence of lattice defects called dislocations. Figure 2 shows an example of a simple dislocation. During plastic deformation, a dislocation moves through the atomic lattice within the crystal, Figure 3.

A simple analogy for dislocation motion is to consider a carpet containing a ruck. By applying a small force to the ruck, it can be moved over the length of the carpet resulting in an overall displacement of the carpet by a small increment. In the absence of the ruck, a very large force would be required to overcome the friction between the complete carpet and the floor. Dislocations distort the surrounding lattice elastically. Some areas are compressed whilst others are stretched. These elastically deformed regions interact with other defects, i.e. with solute atoms, with other dislocations, with grain boundaries and precipitates. These obstacles impede dislocation movement making plastic deformation more difficult.

2.2 Stress-Strain Curve for Simple Tension Specimen

The mechanical properties of materials are determined using standardised test pieces and testing procedures. For ductile materials like most steels, tensile tests are the commonly applied testing method. Brittle materials are usually tested by bending or compression. Tensile properties are usually measured using long and narrow specimens (Figure 4) which guarantee a uniaxial stress state. Round cross-sections are preferred, although rectangular specimens are employed for plate and strip and for the determination of deep drawing properties. Important parameters that have to be controlled during tensile tests are strain rate and testing temperature. The strain rate is controlled by the cross-head speed of the testing machine; the temperature can be regulated by a furnace or a cooling chamber.

Load (F) and elongation (L) are measured during testing, from which the stress ()-strain () curve can be obtained by taking the test piece dimensions into account. The nominal or engineering stress n is the load F divided by the original cross-section area So: n = F/So (1) The nominal or engineering strain n is the ratio of the change in length (L-Lo) to the original length Lo: n = (L - Lo) / Lo = L / Lo (2) Figures 5 and 6 show the stress-strain curves obtained by Equations (1) and (2) for two types of steel.

The engineering stress-strain curve can be divided into three regions: elastic deformation, uniform plastic elongation with continuously increasing stress ( u) and non-uniform plastic elongation ( >u). In most metals there is a gradual transition from elastic to plastic behaviour and it is necessary to define a proof stress, usually Rp0,1 or Rp0,2 in order to characterise the onset of yielding. These are stresses at which a permanent elongation of 0,1 or 0,2% of the initial gauge length is obtained. Some metals and especially mild steels show a discontinuity between the elastic and plastic regimes with a marked yield point followed by a short period of non-uniform plastic strain (Figure 6). This yielding phenomenon is described by:

the upper yield stress ReH at the initiation of yielding. the lower yield stress ReL, which is the stress value during propagation of the Lders bands. the Lders strain Ld (about 1-2% for mild steels).

The reason for this behaviour is that the dislocations are pinned by interstitially dissolved Carbon and/or Nitrogen atoms. These dislocations have to be unpinned by the upper yield stress. Once the dislocations are released they can continue to move at the lower yield stress. This process does not take place homogeneously over the whole specimen. Markings, known as Lder's bands indicating plastically deformed regions, appear at 45 to the direction of loading and expand across the whole gauge length of the specimen. In the range of uniform elongation, further plastic deformation is only possible by a continuously increasing load. The dislocation density grows with increasing deformation, making further deformation more difficult due to the interaction between the dislocations. This effect is referred to as strain or work hardening. The rate of strain hardening (d/d) is high initially but decreases as plastic deformation increases. In the third region of deformation the specimen continues to become narrower, but this now takes place locally resulting in the necking shown in Figure 7.

The stress at which necking begins is the ultimate tensile stress Rm=Fmax/So, the corresponding strain is the ultimate strain u. Following the onset of necking, the load necessary to maintain elongation decreases although the local stress within the necking region continues to increase. Continued deformation leads to fracture, which is characterised by the fracture stress f and the fracture strain r. The percentage reduction of area at fracture Z = 100 (So - Su)/So is a measure of the material's ductility, where Su is the minimum cross-section area at fracture in the necked region. For the engineering stress-strain curve, both stress and stain are related to the initial specimen dimensions (So, Lo). Because the cross-section and length change continuously during

deformation, these equations do not give a true indication of the stress and strain history. For a better description of the material properties, the change of the cross-section area must be taken into account. The true stress and the true strain are defined as follows:

t =


where S is the instantaneous cross-section area.

t =

for u. (5)

Considering that plasticity takes place under conditions of constant volume, the relation between true and engineering stress and strain in the region of uniform elongation can be derived: t = n (1 + n) (6) t = ln (1 + n) (7) In the region of necking, the instantaneous cross-section of the specimen must be measured to obtain the true stress and the true strain. The true strain (Figure 8) is:

t =

for 0 r (8)

For the determination of the true stress, a correction factor km must be taken into account because of the multiaxial stress state resulting from necking. Figure 9 shows the engineering stress-strain curve (1) and the corresponding true stress-true strain curve without (2) and with (3) consideration of the multiaxial stress state. Note that the true strain becomes much larger than the engineering strain due to necking and that strain hardening is always positive.

2.3 Multi-axial Response

2.3.1 Poisson's ratio The previous section discussed the longitudinal response of a simple tensile specimen. If measurements of transverse strain (trans) had also been taken it would have been found that the specimen became narrower as it extended. In the elastic region trans = , where , called Poisson's Ratio, has the value of about 0,3. This implies that the volume of the elastically deformed material is increasing because = - 2trans = 0,4. In the plastic region, with much

higher strains, the volume of material does not change significantly and trans 0,5 which implies that Poisson's ratio has a value of 0,5. This constant volume condition is most clearly illustrated in the necked region as rupture approaches. The locally higher longitudinal strains, are accompanied by high transverse strains giving the local necking. 2.3.2 Multi-axial stress states and their influence on yielding In many engineering situations stresses exist in more than one direction. Examples are shown in Figure 10. The direction and magnitude of these stresses influence the onset of yielding. In the balanced triaxial compression shown in Figure 10a, the material cannot yield because it has "nowhere to go" until there is a breakdown in the atomic structure of the crystals at several orders of magnitude of stress greater than the uniaxial yield stress. In the two-dimensional pure shear case of Figure 10b the tendency to expand in the x-direction from the tensile stress 1 is "encouraged" by the y-direction compression stress 2, and vice versa.

Different mathematical expressions have been developed to describe this interactive yield, and different formulations prove to be more accurate for different materials. Figure 11a illustrates the two most common criteria for metals under a two-dimensional stress state. The Hencky-von Mises expression which is usually used for steel and for three dimensional stresses, is given by: y = (1/2) {(1 - 2)2 + (2 - 3)2 + (3 - 2)2} where 1, 2 and 3 are the three principal stresses.

2.3.3 Strain hardening under multi-axial stresses Figure 11b shows the most common model for work hardening in the presence of multi-axial stresses. In isotropic work hardening the yield surface expands uniformly, as shown.

2.4 Influence of Temperature and Strain Rate

Most engineering structures operate at ambient temperatures and are subject to loading rates that are sufficiently slow not to influence response significantly. However, elevated temperatures can be encountered in service (boiler plant) or by accident (fire). Impact and other accidental loading can lead to strain rates whose influence cannot be ignored. 2.4.1 Temperature The movement of dislocations is facilitated by increasing temperature. Both yield and ultimate strengths are reduced as a result. In structural steels there is little loss of strength up to about 300C. Thereafter there is a progressive loss of strength with increasing temperature. At 600C the strength is typically 50% of that at ambient temperature. Creep, i.e. increasing strain at constant stress, in structural steels can be discounted at ambient temperature. However at temperatures in excess of about 500C creep deformation becomes significant. The influence of elevated temperatures on both stiffness and strength is illustrated in Figure 12.

2.4.2 Strain rate Dislocation movement is a time-dependent process. It is therefore understandable that high strain rates make yielding more difficult which is reflected in increased yield stress. The most widely accepted model for strain rate effects is the Cowper-Symonds expression given below. The significance of this effect on the tensile stress/strain behaviour is illustrated in Figure 13.

/s = 1 + {/D}1/q where: = strain rate = dynamic yield stress s = static yield stress.

D and q are dimensionless curve fitting coefficients. D = 40.4 s-1 and q = 5 for mild steel

2.5 Means of Strengthening

The strength of a steel can be described as the resistance against the onset of plastic deformation under an external load. Plastic deformation occurs by the movement of dislocations through the metal lattice which enables single lattice planes to slip consecutively over one another. If this motion is hindered by lattice defects, a higher external load must be applied so that the dislocations can overcome the obstacles. It is for this reason that means of increasing the strength of steels always aim at hindering dislocation movement. Obstacles to dislocation motion can be classified according to Table 1. TABLE 1

Dimension 0 (point) 1 (linear) 2 (plane) 3 (spatial) Strengthening of Steels

Generator solute atoms dislocations grain boundaries particles

Mechanism solid solution hardening strain hardening grain refinement precipitation/dispersion hardening

Solid solution hardening

The increase of strength is produced by foreign atoms which are dissolved in the metal matrix. Since foreign atoms differ in size, physical and electrical properties compared to the metal matrix, they cause a lattice distortion that hinders the movement of dislocations. The increase of yield stress depends on the kind, amount and distribution of the foreign atoms, Figure 14. It can be seen that the increase of strength caused by C and N is especially high. This is due to the fact that these atoms dissolve interstitially between the atoms of the matrix, which results in a high lattice distortion. The other elements shown in Figure 14 dissolve substitutionally and occupy regular lattice positions producing less distortion.

Increasing strength by solid solution hardening leads to a decrease in toughness. (Toughness is the ability of a material to avoid brittle fracture - see Lecture 2.3.2.) Therefore, the potential level of strength attainable by solid solution hardening is usually not sought because of the possibility of brittle fracture.

Strain hardening

Linear lattice defects are the dislocations themselves. The lattice distortion surrounding the dislocation disturbs the movement of other dislocations. This interaction increases with increasing dislocation density. When cold forming steel, e.g. cold rolling, dislocations must continuously be produced because they permanently block each other. The dislocation density rises and increases the strength of the

steel by strain hardening (Figure 15). Such hardening is accompanied by a pronounced reduction in toughness.

Grain refinement

Grain refinement is the most important strengthening mechanism in structural steels because it is the only method of strengthening which is accompanied by an increase in resistance to brittle fracture. The grain boundaries are barriers to dislocation motion. Consequently as the grain size is decreased, the number of barriers increases and this is reflected in increased yield strength. The effect of grain size on yield stress is described by the Hall-Petch equation: ReL = iy + ky . d-1/2

where ReL is the minimum yield strength iy is the friction stress; the stress at which yielding begins in a material with very large grains ky is the grain boundary resistance d is the mean grain diameter.

Dispersion and precipitation hardening

When a moving dislocation encounters a particle it can pass by only two mechanisms: cutting through the particle, or bowing between and around particles, leaving a section of the dislocation as a ring around the particle. In steels, the dominant particle hardening mechanism is dislocation bowing. Strength is then inversely related to the interparticle spacing. Consequently strength increases as the particle size decreases and as the volume fraction of precipitates increases. Strengthening particles are obtained most usually by the precipitation process in the matrix during heat treatment. In steel this is most familiar during tempering of quench hardened medium carbon steels, but it also occurs in structural steels during cooling after controlled rolling or during subsequent tempering treatments.

2.6 Hardness
"Hardness" is a measure of resistance to deformation when a loaded indenter is forced to penetrate the surface of the metal under test. The penetration of the indenter into the specimen leads to a local deformation which is both elastic and plastic. Testing methods can be either static or dynamic. Both elastic and plastic deformation are considered for the evaluation of dynamic hardness, expressed in terms of the absorbed energy. Dynamic testing methods include impact hardness testing and Shore hardness testing. Their advantage lies in the possibility of quick testing which can be performed at any location. Compared to static hardness testing methods, the dynamic methods are less precise. The common static testing methods of indentation hardness differ in the type of indenter forced into the metal. The Brinell test uses a hardened steel ball (EN 3), the Vickers test (EN 5) a square-based diamond pyramid (included angle = 136), and the Rockwell test (EN 10004) is performed with a diamond cone indenter (included angle = 120).

The indenter is slowly (almost statically) pressed into the specimen. After removing the load, the size of the indentation is measured. Compared to dynamic testing, only the plastic deformation is taken into account. Static testing methods are favoured in industry and research because of the consistency of test results. The different hardness measurements correlate quite closely, especially at lower values. The correlation of hardness values with other measures of resistance to deformation, such as tensile properties, is more complicated but a useful engineering rule of thumb is that the tensile strength in units of Nmm-2 is approximately 3 times the Vickers hardness.


Most engineering properties of metals are structure sensitive. The principal quantities defining the mechanical properties of metals under non-repeating loading are:

Young's modulus. Poisson's ratio. Yield stress, or proof stress for metals without a defined yield point. Ultimate strength. Hardness.

These properties may be influenced by:

temperature. strain rate. multi-axiality. geometry.

The optimal combination of strength and toughness may be achieved by controlling:

alloying content. level of impurities. physical treatment.

1. Leslie, W.L., The Physical Metallurgy of Steels, Hemisphere Publishing Corporation, Washington, New York, London, 1981. 2. Dahl, W., Steel - Handbook of Materials Research and Engineering Volume 1, Springer-Verlag, Dsseldorf 1990.

Typical mechanical property data for weldable structural steels

Steel (Grade according to EN10025) Grade Fe430B (0,25% carbon) Grade Fe430D (0,19% carbon) Grade Fe510B (0,23% carbon) Grade Fe510D (0,18% carbon, 0,10% vanadium)


Yield Stress (N/mm2) 275 275 355 355

Ultimate Tensile Strength (N/mm2) 470 470 550 550

% Elongation to Fracture (Lo = 5,65So) 22 22 20 20

hot-rolled normalised hot-rolled normalised

Typical mechanical property data for some common non-ferrous metals and alloys 0,2% Proof Stress (N/mm2) Aluminium (annealed) Aluminium (cold worked) Duralumin (annealed) Duralumin (age-hardened) Cooper (annealed) Copper (cold worked) 70-30 brass (annealed) 70-30 brass (cold worked) 34 94 123 278 54 285 85 378 Ultimate Tensile Strength (N/mm2) 77 115 231 432 223 316 320 463 % Elongation to fracture (Lo - 5,65So) 47 13 15 15 56 13 65 20

Metal or alloy

Typical values of Young's modulus for metals

Metal kN/mm2

Al 70

Au 79

Ti 120

Cu 130

Fe 210

Ni 200

Metals often show quite acceptable properties when small smooth bar specimens are tested in tension at ambient temperature and at slow loading rates. However they fail in a brittle manner when large components are loaded or when the loading is performed at low temperatures or applied rapidly. Susceptibility to brittle fracture is enhanced if notches or other defects are present. Resistance to brittle fracture is commonly referred to as toughness. Metals with a body-centred cubic lattice, e.g. pure iron and ferritic steels have the unfortunate characteristic that their fracture mechanism undergoes a dramatic transition with decreasing temperature from a tough ductile mode in the higher temperature region to a brittle cleavage mode at lower temperatures. Face-centred cubic metals, e.g. copper, aluminium and austenitic steels, do not fail by cleavage under all loading conditions and at all temperatures.

1.1 Types of Fracture

Ductile fracture involves the formation, growth and coalescence of voids. A simple analogy is the fracture of plasticene or putty containing particles of sand. The voids form around precipitates or non-metallic inclusions, Figure 1. The ductility or toughness of the material is basically dependent on the volume fraction of the void nucleating particles, i.e. the proportion of sand in the previous analogy. The amount of deformation prior to rupture and thus the toughness of the material increases with its purity.

The macroscopic orientation of a ductile fracture surface may vary from 90 to 45 to the direction of the applied stress. In thick sections most of the fracture surface tends to be oriented at 90 to the direction of the applied tensile stress. However, ductile fractures commonly have a "shear-tip" near a free boundary as the transverse stresses reduce to zero causing the plane of maximum shear to be at 45 to the direction of the applied stress. Cleavage fracture occurs in body-centred cubic metals when the maximum principal stress exceeds a critical value, the so-called microscopic cleavage fracture stress f. Certain crystallographic planes of atoms are separated when the stress is sufficiently high to break atomic bonds. Crystallographic planes with low packing densities are preferred as cleavage planes. In steels the preferred change planes are the bee cube planes. The fracture surface lies perpendicular to the maximum principal stress and appears macroscopically flat and crystalline. When viewed by eye a cleavage fracture usually displays characteristic chevron markings which point back to the origin of the fracture. When brittle fracture occurs in a large structure, such markings can be invaluable in identifying the site of crack initiation. When viewed in the microscope, cleavage cracks can be seen to pass through the grains along preferred crystallographic planes (transgranular cleavage). If grain boundaries are weakened by precipitates or by the enrichment of foreign atoms, cleavage cracks can also propagate along grain boundaries (intergranular cleavage).

1.2 Influence of Temperature, Loading Rate, Multi-axiality and Geometry

Temperature influences fracture behaviour mainly due to its effect on yield strength and the transition from ductile to cleavage fracture. Figure 2 shows schematically the yield strength and the microscopic cleavage fracture stress as a function of temperature for a ferritic steel. The yield strength falls with increasing temperature, whereas the cleavage fracture stress is hardly influenced. The transition temperature is defined by the intersection between the yield strength and cleavage fracture strength curves. At lower temperatures specimens fail without previous plastic deformation (brittle fracture). Somewhat above the transition temperature, cleavage fracture can still occur due to the effect of deformation induced work hardening. At higher temperatures cleavage is not possible and the fracture becomes fully ductile.

The yield strength rises with increasing loading rate (marked with dashed line in Figure 2) whereas the microscopic cleavage fracture stress shows almost no strain rate dependence. This rise causes the ductile-brittle transition temperature to move to higher values at higher rates of loading. Thus, an increase of loading rate and a reduction of temperature have the same adverse effect on toughness.

A multi-axial stress state has an important influence on the transition from ductile to cleavage fracture. A triaxial state of stress, in which the three principal stresses 1, 2 and 3 are all positive (but not equal), inhibits or constrains the onset of yielding. Under these conditions, yielding occurs at a higher stress than that observed in a uniaxial or biaxial state of stress. This situation is illustrated in Figure 3 where it can be seen that the transition temperature arising from the intersection of the cleavage and yield strength curves is shifted to a higher temperature, i.e. the metal has become more brittle.

The most familiar situation in which multi-axial states of stress are encountered in steel structures is in association with notches or cracks in thick sections. The stress concentration at the root of the notch gives rise a local region of triaxial stresses even through the applied loading may be uni-directional (Figure 4).

1.3 Notched Impact Bend Test

The notched impact bend test is the most common test for the assessment of susceptibility to brittle fracture because it is inexpensive and quickly performed. 10mm square bars with a machined notch, (ISO-V or Charpy specimens), are struck by a calibrated pendulum. The energy absorbed from the swinging pendulum during deformation and fracture of the test specimen is used as a measure of the impact energy. The notch impact energy consists of elastic and plastic deformation work, fracture energy and kinetic energy of the broken pieces. Figures 5 and 6 show the notch impact energy as a function of testing temperature. At low temperatures the failure of ferritic steels occurs by cleavage fracture giving a lustrous crystalline appearance to the fracture surface. At high temperatures failure occurs by ductile fracture after plastic deformation. In the transition range small amounts of ductile fracture are found close to the notch but, due to the elevated stresses near the crack tip, the fracture mechanism changes to cleavage. Throughout the transition range the amount of cleavage fracture becomes less and the notch impact energy rises as the testing temperature increases.

In order to characterise the transition behaviour, a transition temperature is defined as the temperature at which:

a defined value of the notch impact energy is reached (eg. T 27J, T40J), half of the maximum impact energy value is reached (T 50%), or 50% ductile fracture is observed on the fracture surface (FATT 50: Fracture Appearance Transition Temperature, 50% ductile fracture).

The impact energy values obtained show a high amount of scatter in the transition area because here the results depend on the local situation ahead of the crack tip. Beyond this area, scatter becomes less because there is no change of fracture mechanism. The notched impact bend test gives only a relative measure of toughness. This measure is adequate for defining different grades of toughness in structural steels and for specifying steels for well established conditions of service. For the assessment of known defects and for service situations where there is little experience of brittle fracture susceptibility, a quantitative measure of toughness which can be used by design engineers is provided by fracture mechanics.

1.4 Fracture Toughness

Fracture mechanics provides a quantitative description of the resistance of a material to fracture. The fracture toughness is a material property which can be used to predict the behaviour of components containing cracks or sharp notches. The fracture toughness properties are obtained by tests on specimens containing deliberately introduced cracks or notches and subjected to prescribed loading conditions. Depending on the strength of the material and the thickness of the section, either linear-elastic (LEFM) or elastic-plastic fracture mechanics (EPFM) concepts are applied. The Linear-Elastic Fracture Mechanics Approach The stress intensity factor KI describes the intensity of the elastic crack tip stress field in a thick, deeply cracked specimen loaded perpendicular to the crack plane. KI = Y where is the nominal stress a is the crack depth Y is the correction function dependent on the crack and test piece geometry The critical value of the stress intensity factor for the onset of crack growth is the fracture toughness KIC. Another material property obtained from linear-elastic fracture mechanics is the energy release rate GI. It indicates how much elastic strain energy becomes free during crack propagation. It is determined according to Equation (2): GI = Y2 2 a / E = K12 / E (2) where E is the Young's modulus Analogous to the stress intensity factor, crack growth occurs when GI reaches a critical value GIc. The fracture toughness properties KIc and GIc are determined with fracture mechanics specimens, generally as shown in Figures 7 and 8. (1)

The great value of the fracture toughness parameters KIc and GIc is that once they have been measured for a particular material, Equations (1) and (2) can be used to make quantitative predictions of the size of defect necessary to cause a brittle fracture for a given stress, or the stress which will precipitate a brittle fracture for a defect of known size. As the designation implies, linear elastic fracture mechanics is applicable to materials which fracture under elastic conditions of loading. The fracture phenomena in high strength quenched and tempered steels are of this type. In lower strength structural steels, extensive plasticity develops at the notch root before failure occurs. This behaviour invalidates many of the assumptions of linear elastic fracture mechanics and makes testing difficult or not meaningful. In such cases elastic-plastic fracture mechanics must be applied. There are two alternative techniques of elastic-plastic fracture mechanics:
1. Crack Tip Opening Displacement (CTOD) 2. J Integral

Their essential features are summarised below. The Elastic-Plastic Fracture Mechanics Approach A consequence of plasticity developing at the tip of a previously sharp crack is that the crack will blunt and there will be an opening displacement at the position of the original crack tip. This is the crack tip opening displacement (CTOD). As loading continues, the CTOD value increases until eventually a critical value c is attained at which crack growth occurs. The critical crack tip opening displacement is a measure of the resistance of the material to fracture, i.e. it is an alternative measurement of fracture toughness. For materials which exhibit little plasticity prior to failure, the critical CTOD, c, can be related to the linear elastic fracture toughness parameters KIc and GIc as follows: KIc2 = E.Gk / (1 - 2) = where E is Youngs modulus y is the uniaxial yield strength is Poissons ratio m is a constraint factor having a value between 1 and 3 depending on the state of stress at the crack tip. m.E.y.c / (1 - 2)

Another way of taking account of crack tip plasticity is the determination of the J-integral. J is defined as a path-independent line-integral through the material surrounding the crack tip. It is given by:



U is the potential energy B is the specimen thickness a is the crack length

U= F is the load


Vg is the total displacement Since the determination of J is difficult, approximate solutions are used in practice.

J= where b=w-a


= 2 (for SENB-specimens) = 2 + 0,522 b/w (for CT-specimens) The critical value of J is a material characteristic and is denoted JIc. For the linear elastic case, JIc is equal to GIc.

1.5 Fitness for Purpose

Conventional assessment of components is based on a comparison of design resistance with applied actions. Toughness criteria are generally satisfied by the appropriate selection of material

quality, as discussed in Lecture 2.5. However there are situations where a more fundamental assessment has to be carried out because of:

onerous service conditions. defects during manufacture. defects, e.g. fatigue cracks, developing during previous service life.

Such assessments can be performed by different methods. If the component is small, it may be possible to test it. For large or unique structures, such as bridges or offshore platforms, this method of producing the most realistic data has to be excluded. Tests on representative details of a component may be performed, if the simulation of the real structure is done carefully, e.g. accounting for specific service conditions including the geometry of the structure and discontinuities, loading rate, service temperature and environmental conditions. A typical example of such a test method is the wide plate test, which is discussed below. Fracture mechanics concepts have been developed to assess the safety of components containing cracks. Depending on the overall behaviour of the component (linear-elastic or elastic-plastic) different methods can be used for failure assessment. 1.5.1 Wide plate testing During the last 20 years, large flat tensile specimens, so-called wide plates, have been used to simulate a relatively simple detail of a tension loaded large structure. A main objective of wide plate testing is the evaluation of the deformation and fracture behaviour of a specimen under service conditions. The second reason for this kind of test is the application of test results for the development and checking of failure assessment methods, e.g. fracture mechanics methods. Wide plate tests require testing facilities with high loading capacities due to the fact that such tests are usually carried out at full thickness. The maximum dimensions of wide plates tested on large test rigs with a load capacity of up to 100MN are as follows:

specimen width W 3000mm specimen thickness to 300mm specimen length l 5000mm

Figure 9 shows different types of specimen containing discontinuities for tests on the base metal or welded joints. The discontinuities may be through-thickness or surface notches or cracks. The configuration of the plate is usually chosen according to the specific structural situation to be assessed.

Stress or strain criteria can be used as safety criteria which must be fulfilled to assure the safety of a specific structural element. The production of a given amount of overall strain is in some cases used as the failure criterion. The gross-section-yielding concept requires that gross-sectionyielding (GSY) occurs prior to fracture. Based on this concept, wide plates with different crack lengths are tested under similar loading conditions to determine a critical crack length just fulfilling the GSY-criterion. Figure 10 shows the ratio of the maximum gross-section stress in the structure to ultimate tensile strength as a function of the crack length ratio 2a/W of centrenotched wide plates. The upper limit line describes the theoretical maximum stress, if the ultimate tensile strength is reached in the cross-section containing the discontinuity. All test results show lower values than are implied by the theoretical line, resulting from the important influence of toughness in the presence of discontinuities. Only in the case of infinite toughness can the theoretical line be reached. The intersection of the experimentally determined curve and the yield strength line marks the critical crack length ratio 2ac/W. As long as the 2a/W ratio is smaller than the critical ratio, the GSY-criterion is fulfilled. Unfortunately, the critical 2ac/W ratio depends strongly on the dimensions of the crack and the plate, so that different types of cracked components always require a series of specific wide plate tests. This concept is therefore only used if other concepts cannot be applied.

1.5.2 Fracture mechanics concepts The basis of a fracture mechanics safety analysis is the comparison between the crack driving force in a structure and the fracture toughness of the material evaluated in small scale tests. The application of one of the concepts depends on the overall behaviour of the structure which may be linear-elastic (K-concept) or elastic-plastic (CTOD- or J-Integral-concepts). For a safe structure the crack driving force must be less than the fracture toughness. In general the toughness values of the material are evaluated according to existing standards. The crack driving force can be calculated on the basis of analytical solutions (K-concept), empirical or semiempirical approaches (CTOD-Design-Curve approach, CEGB-R6-procedures) or using numerical solutions (indirectly: EPRI-handbook, directly: finite-element calculations). The different methods are explained briefly below:


The K-concept can be applied in the case of linear-elastic component behaviour. The crack driving force, the so-called stress intensity factor KI, defined in Section 1.4, has been evaluated for a large range of situations and calculation formulae are for example given in the stressanalysis-of-cracks handbook. Usually the critical fracture toughness KIc of the material is evaluated according to the ASTM standard E399 or the British Standard BS5447. Brittle failure can be excluded as long as: KI < KIc For a given fracture toughness the critical crack length or stress level can be calculated from:

ac =

c =

CTOD-Design-Curve approach

A critical crack length or stress level can be determined using the limit curve of the CTODDesign-Curve approach for the driving force assessment together with measured values of CTODcrit for the material. The limit curve has been adopted by standards, e.g. the British Standard BS-PD 6493. The latest version of the limit curve is shown in Figure 11 and can be used for: 2a/W 0,5 and net YS.

Analysis can only be performed under global elastic conditions (net YS) although local plastic deformation may occur in front of a crack tip which is accounted for in the CTOD-value of the material.


The CEGB-R6-routines can be used to assess the safety of structures for brittle and ductile component behaviour. The transition from linear-elastic to elastic-plastic behaviour is described by a limit curve in a failure analysis diagram (Figure 12). The ordinate value Kr can be regarded

as any of three equivalent ratios of applied crack driving force to material fracture toughness as follows:

Kr =

Other methods

Other methods are emerging. The Electrical Power Research Institute (EPRI) in New York has used a detailed analysis by finite elements to determine limiting J contour values for standard geometries. Alternatively the J contour values may be obtained by direct finite element analysis of the particular situation.


Preceding sections have described the influence of the micro structure on strength and toughness using metallurgical mechanisms. Chemical and physical metallurgy can change microstructural characteristics so that optimum strength and toughness requirements may be obtained. By combining the various treatments it is possible to achieve a wide range of steel properties (Figure 13):

Chemical metallurgy treatments

Variation of the chemical composition of a steel by adding alloying elements aims to increase strength and/or increase resistance to brittle fracture. Solid solution hardening generally lowers toughness and is not widely employed. Precipitation hardening also increases strength and decreases toughness. The addition of manganese and nickel produces a small increase in strength due to solution hardening but a more significant reduction is impact transition temperature due to grain refinement (Figure 14). Alloying with the micro-alloying elements Niobium, (Nb) Vanadium (V) and Titanium (Ti) producing carbides and nitrides simultaneously raises strength

by precipitation hardening and toughness by grain refinement. Decreasing the content of elements such as S and P improves the degree of purity, which has positive effects on toughness and weldability.

Physical metallurgy treatments

The microstructure of a steel can be greatly affected by heat treatment or forming. Correctly chosen temperature, degree of deformation, time between deformation steps and cooling rate can reduce the grain size and control the state of precipitation, thus raising toughness and strength (Figure 15).

This combination of heat treatment and forming known as thermo-mechanical treatment leads to even better results if micro-alloying elements such as V or Nb are added, causing additional grain refinement with improved toughness and strength properties.

When considering the response of metallic materials to cyclic loading, it is essential to distinguish between components such as machined parts, which are initially free of defects, and those such as castings and welded structures, which inevitably contain pre-existing defects. The fatigue behaviour of these two types of component is quite different. In the former case, the major part of the fatigue life is spent in initiating a crack; such fatigue is 'initiation-controlled'. In the second type of component, cracks are already present and all of the fatigue life is spent in crack propagation; such fatigue is 'propagation-controlled'.

For a given material, the fatigue strength is quite different depending on whether the application is initiation- or propagation-controlled. Also the most appropriate material solution may be quite different depending on the application. For example with initiation-controlled fatigue, the fatigue strength increases with tensile strength and hence it is usually beneficial to utilise high strength materials. On the other hand, with propagation-controlled fatigue, the fatigue resistance may actually decrease if a higher strength material is employed.

3.1 Initiation-Controlled Fatigue

3.1.1 Testing The fundamental diagram in fatigue testing is the Whler or S-N-diagram (Figure 16). Specimens are exposed to cyclic loading with a constant amplitude and the number of cycles to fracture is recorded. This parameter is plotted against the corresponding stress amplitude with a double- or semi-logarithmic scale. The diagram is divided into two parts. In the first part, life time increases with decreasing alternating stress amplitude. In the second part for most-ferritic steels the curve becomes horizontal and defines a 'fatigue limit' stress below which failure can never occur. The transition or 'knee' between the two parts of the curve lies between 3 and 10 x 106 cycles, depending on the material. For other alloys, e.g. fcc-metals, which do not show a fatigue limit, an 'endurance limit' is defined as the stress amplitude corresponding to a life of 107 cycles.

One characteristic feature of fatigue properties is the wide scatter of results under constant testing conditions. Therefore 6-10 experiments must be performed for each stress amplitude. The analysis is done by means of statistical evaluation leading to different S-N curves for various life time probabilities (10%, 50%, 90% curves). 3.1.2 Fatigue damage

Crack-free stage

During the first 104 stress cycles, although the loading is nominally elastic, dislocation activity occurs in localised areas and leads to the formation of bands of localised plastic deformation known as "persistent slip bands" (PSB).

Crack initiation

Crack initiation generally takes place within the persistent slip bands. In the case of pure metals, crack initiation usually occurs at the surface. In commercial quality materials, crack initiation usually occurs at non-metallic inclusions or other impurities which act as microscopic sites of strain concentration.

Crack propagation

Once initiated the crack propagates through the first few grains in the direction of maximum shear stress, i.e. at 45 to the normal stress. When the crack has attained a length of a few grain diameters, continued propagation is controlled by the cyclic stress intensity field at the crack tip and the crack path becomes oriented at 90 to the maximum principal stress direction. Although the major part of the fatigue life is spent in crack initiation, this is not apparent from examination of the fracture surface where only the final propagation stage can be seen. 3.1.3 Influences of various parameters The relationships between initiation-controlled fatigue strength and other parameters are complex and sometimes only known qualitatively. Nevertheless they are of great importance for material selection and dimensioning of structural parts. Therefore a number of different parameters are discussed below with respect to their influence on fatigue properties.

Loading: Different loading conditions include cyclic tension and compression, cyclic torsion, cyclic bending and any possible combination of these. As discussed in the context of yielding in Section 2.3 of Lecture 2.3.1, such complex stresses can be combined by means of the Hencky-von Mises expression to generate an equivalent stress which can be compared with the fatigue strength obtained from uniaxial loading. Mean stress: Fatigue strength is reduced by tensile mean stress and increased by compressive mean stress. Frequency: For most materials no influence is observed over a wide range. Some alloys show a smaller life time for lower frequencies because corrosion effects interfere. Microstructure: The influence of microstructural modification on fatigue strength is similar to that on tensile strength. In general fatigue strength increases in proportion to tensile strength. For example, for a wide range of wrought steels, the fatigue strength is between 40% and 50% of the tensile strength. Improved purity raises fatigue strength. Residual stresses: As with mean stress effects, compressive residual stress improves fatigue strength, whereas internal tensile stress has the opposite effect. To optimise fatigue strength, surface compressive residual stress is generated by techniques such as shot peening, and surface rolling. Surface: Surface finish has a large influence on fatigue; the smoother the surface the better the fatigue strength. The treatment of surfaces during manufacturing often causes strain hardening and compressive residual

stresses which both increase fatigue strength. The influence of notches is described under "Geometry".

Geometry: Notches and changes of section act as sites of stress concentration and hence have a considerable influence on fatigue properties. For large smooth notches, the stress concentration must be evaluated and incorporated in the fatigue analysis. Sharp notches behave as crack-like defects and cause the fatigue behaviour to be propagation-controlled. Welding: Welding inevitably generates small crack-like defects which greatly lower the fatigue strength and cause the fatigue to be propagationcontrolled. Corrosion: Exposure to a corrosive environment facilitates both crack initiation and propagation. Consequently the fatigue strength is reduced. The fatigue limit in steels may be eliminated in a corrosive element.

3.1.4 Fatigue limit under actual service conditions The S-N diagram characterises material behaviour under single-amplitude loading. For weightsaving constructions exposed to complex stresses, the parameters determined by such tests are not sufficient. For testing under realistic conditions, an analysis of the actual stresses has to be obtained. For that purpose the sequence and duration of different stress levels, as well as their rise or fall, are recorded. This stress-time function is either reproduced under laboratory conditions, or special testing programmes are calculated from these data and used in experiments. Results obtained by this method cannot be transferred to different materials and loading conditions. 3.1.5 Prediction of cumulative damage The fundamental method of life time cumulative damage prediction was formulated by Miner. The damage from each cycle at a certain stress level is defined as the reciprocal value of the number of cycles to fracture (1/Ni). Fracture occurs when the sum of cycles at each level (ni) related to the number of cycles to failure (Ni) is equal to unity. The mathematical expression is:

Since this is a very simple equation, results are widely scattered. In reality the values form a Gaussion distribution with a maximum around 1. To guarantee safe construction, calculations are made with factors smaller than 1 and stresses below their maximum values. Furthermore it is possible to take the effects of different loading levels into account with respect to their number, maximum stress and sequence.

3.2 Propagation-Controlled Fatigue

Steel castings, rough forgings and welded structures invariably contain surface imperfections which behave as minute crack-like defects which effectively eliminate the crack-initiation stage in fatigue. Consequently the whole of the fatigue life is concerned with crack propagation. The rate of crack advance is determined by the cyclic stress intensity Kr which is the cyclic equivalent of the stress intensity factor KI defined in Section 1.5. KI = Y where is the cyclic stress range a is the crack depth Y is the correction function dependent on the crack and test piece geometry. The rate of crack propagation is then given by the following relationship which is known as Paris' Law:

= C KIm N = Number of cycles C is a material constant which is inversely proportional to Young's modulus E. The power m has a value of about 3 for most metallic materials. The advantage of the fracture mechanics description of crack propagation is that the rate equation can be integrated to determine the number of cycles required for a crack to propagate from some initial length ai to same final length af. Thus for m = 3; Nf = 2 (1/ai - 1/af ) / (CY333/2) ai may be a known crack size or an NDT limit, af may be a critical defect size for unstable fracture or a component dimension such as the wall thickness of a vessel. In the above equation for the fatigue life, the constant C is dependent on the type of material but is not sensitive to variations in microstructure or strength level. Consequently, for a given cyclic stress range, , the fatigue life is independant of the strength of the material. If, however, the stress range increases in proportion to the material yield strength, then the fatigue life will be less for the higher strength material. For example, a two-fold increase in stress range produces almost a ten-fold reduction in fatigue life. This is a major constraint on the utilisation of higher strength structural steels for fatigue dominated applications.

The fatigue behaviour of welded joints is propagation-controlled. However it is impracticable to apply a fracture mechanics analysis because the initial defect size cannot be evaluated and the cyclic stress range is amplified by local stress concentration effects associated with the weld profile. Instead the fatigue strength is determined experimentally for the range of weld types and welding processes which are commonly employed in welded structures. This data is presented as a series of S-N curves for different weld classifications as shown in Figure 16. The fatigue strength of welded joints is not sensitive to the strength of the parent plate. Consequently, as explained previously, it is difficult to take full advantage of higher strength steels in welded structures where there is significant exposure to cyclic loading.


Steels may fail by unacceptable brittle fracture. Satisfactory ductility has generally to be achieved by ensuring ductile rather than cleavage fracture. The tendency for brittle fracture increases in:

Reducing temperature Increasing strain rate Multi-axial tension Geometric discontinuities causing stress concentrations.

Fracture mechanics is a valuable means of quantifying the resistance of a material to fracture. The notched impact bend test (Charpy test) is a cost effective means of qualitatively monitoring toughness. More accurate methods of monitoring toughness, e.g. CTOD testing, have developed from the understanding of fracture mechanics. The optimal balance of strength and toughness can be achieved by a combination of chemical and physical metallurgical treatments. Structures under repeated loading may fail by fatigue. Resistance to fatigue is influenced by stress range, number cycles, mean stress, geometry, residual stresses and defects, especially those associated in welding.