Fracture

h:\data/a8/wp/training1/fracture Rev.2

!"#$!"#$!"#$!"#$Materials & NDE Foundation Course

Introduction toFracture

Page 2 of 35

FRACTURE LECTURE NOTES

Index

1. Introduction1.1 Overview............................................................................................... 31.2 Control of Fracture............................................................................... 6

2. Mechanisms of Fracture2.1 Stress Concentrations........................................................................... 72.2 Fracture Mechanisms

2.2.1 Brittle Fracture........................................................................82.2.2 Ductile Failure........................................................................ 10

2.3 Ductile to Brittle Transition................................................................. 112.4 Structural Thickness and Constraint..................................................12

3. Toughness Testing3.1 The Charpy Test

3.1.1 Introduction............................................................................ 133.1.2 Test Procedure....................................................................... 15

3.2 Fracture Toughness Testing................................................................ 173.3 Crack Arrest Tests.................................................................................19

4. Material Effects4.1 Definition of the Ductile to Brittle Transition Temperature........... 214.2 Chemical Composition.........................................................................224.3 Microstructure Effects.......................................................................... 234.4 Strain Rate.............................................................................................. 244.5 Service Embrittlement.......................................................................... 24

5. Toughness Of LR Materials5.1 Normal Strength Ship Steels............................................................... 255.2 Higher Strength Ship Steels.................................................................25

6. Material Selection For Fracture Control6.1 General LR Rule Requirements for Ships..........................................276.2 Ice-breaking Ships.................................................................................286.3 Liquefied Gas Ships.............................................................................. 29

7. Fracture Mechanics7.1 Introduction........................................................................................... 317.2 Design against Failure..........................................................................33

8. Summary......................................................................................................... 35

Page 3 of 35

1. Introduction

1.1 Overview

Fracture is defined as the separation of a solid body into two or more partsunder the action of stress. This process consists of two parts: crack formation(initiation) and crack growth. Obviously for marine applications, it isimportant to design against fracture as even a localised structural failure canhave catastrophic consequences for the integrity of a vessel. Fracture canoccur rapidly under constant load, but slow crack growth can occur under theinfluences of alternating load (fatigue) or chemical attack (stress corrosioncracking). In this paper, we shall consider the rapid fracture of metals underconstant load.

There are two general types (modes) of fracture: brittle and ductile.

Brittle fracture involves hardly any deformation, and often occurs in materialssubjected to applied stresses far below the yield strength. The crack growthcan be very rapid (up to 1km/s) and unstable, so failure is often catastrophicand without warning. Ferritic steel structures are particularly susceptible tobrittle fracture at low temperatures.

Brittle fracture has only become a major problem this century with theintroduction of fusion-welded structures. With old-style rivetedconstructions, a crack in a plate would usually be arrested at the plate edge.However, with an all-welded fabrication, the crack can propagate through theentire structure. In a ship for example, this can result in the vessel breaking intwo. The classic examples of this are the Liberty ships, built in the USA duringWorld War II (Figure 1). These failures and others led to the introduction oftoughness requirements for ship steels into LR Rules (1958).

Figure 1 - Brittle fracture of the Liberty Ship USS Schenectady (1943)

Page 4 of 35

However, there have been more recent examples. In 1979, the bulk carrier MVKurdistan broke in two (Figure 2). Here the crack was shown to have startedat a defective butt weld in the ground bar attaching the bilge keel to the sideshell. The crack then grew into the bilge keel itself, through the side shell, andpropagated round the entire vessel. Since then LR Rules have been introducedprohibiting the bilge keel from being directly welded to the side shell andintroducing rat holes into bilge keel butt welds, to prevent any cracks frompropagating into the main structure (LR Rules Part 3 Chapter 10 Section 5.6).

Figure 2 - Fracture initiation of Kurdistan failure

crack propagatedvia intermittentweld into bulb bar

crack initiated at various sitesin ground barweld

ground barbutt weld

crack propagatedvia ground bar toshell weld intobilge strake andshell plate

Page 5 of 35

Catastrophic brittle fractures have also occurred in bridges, pipelines,pressure vessels (Figure 3) and electrical turbo generators.

Figure 3 - Brittle fracture of a pressure vessel, initiated from a defective weld (1958)

Ductile fracture is always accompanied by a large amount of (plastic)deformation. Failure, which occurs by slow and stable crack growth, will onlyoccur when the yield stress of the material is exceeded locally. Under normalcircumstances, structural metals are designed to fail by ductile mode, but thiswill only happen if the structure is overloaded (Figure 4).

Figure 4 - Ductile fracture following explosion of a fuel pipeline (1997)

Page 6 of 35

1.2 Control Of Fracture

The common feature of all of these failures is the total loss of the structure,which inevitably has significant safety and economic consequences.

There are generally common features between incidents of brittle fracture,typified by the Liberty ship failures:

• cracks or notches in the structure (e.g. weld defects)• high local tensile stress (e.g. hatch cover corners)• low ambient temperature (e.g. Arctic waters)

Therefore, to combat the incidence of brittle fracture, the followingprecautionary measures can be taken:

• avoid excessive tensile stress by attention to the construction details• use qualified welding procedures and carry out NDE on welds to

ensure that they do not contain significant defects.• select a steel that will not behave in a brittle manner at the service

temperature.

It was the need for adequate resistance to cracking, defined as the toughnessof steel, that was not properly understood at the time of construction of theLiberty ships. As a direct result of these failures, a number of toughness testswere developed to ensure that ship steels had adequate resistance to fracture.It is important to note that as cracks in service developed from cracks ornotches, the laboratory test testpieces had to contain similar discontinuities.

Page 7 of 35

2. Mechanisms Of Fracture

In order to understand how the fracture of a structure can be prevented, it isuseful to briefly consider the mechanisms of the fracture process.

2.1 Stress Concentrations

A critical factor in determining whether or not fracture will occur is the levelof tensile stress within the structure. However, the stress level is not constantthroughout a structure. Geometrical discontinuities, such as holes andnotches, produce a local stress concentration that can far exceed the generalstress level applied. This concentration tends to increase with decreasingradius of the discontinuity. Thus a volumetric defect, such as a gas pore in aweld, will only provide a slight stress concentration. However a planar defect,such as a crack, can produce a stress of yield magnitude with only a smallapplied stress (Figure 5).

1

1.5

2

2.5

3

0 0.2 0.4 0.6 0.8 1discontinuity radius (mm)

stre

ss c

once

ntra

tion

gas

pore

crac

k tip

Figure 5 - Variation of stress concentration with radius of discontinuity(note that this is only an example - the actual level of stress concentration varies according to discontinuity orientation and structural geometry)

Welds are a common source of discontinuities which, depending on theirorientation, may initiate brittle fracture. If the weld profile itself is not smooth(e.g. undercut), a severe stress concentration can arise. On a larger scale, othercommon stress raisers in ships include hatch cover openings and ratholes atthe heel of stiffener brackets. Note that fracture is most likely to initiate in themidships of a vessel, where the applied tensile stresses are at a maximum.Therefore, higher toughness steels tend to be used in this area.

Page 8 of 35

It must be remembered that in addition to the applied stress, residual stresseswill be present within a fabricated structure, which will reinforce the appliedstress. Contraction of the weld metal during cooling will set up significantstresses (of up to yield strength in magnitude) in the weld and adjacent parentmaterial. These can be decreased by a suitable post weld heat treatment,which will, therefore, reduce the susceptibility of the structure to fracture.

2.2 Fracture Mechanisms

2.2.1 Brittle Fracture

When a material fails by brittle fracture, very little deformation occurs withinthe structure. Instead the material literally cleaves in two. Failure is producedby tensile stresses, thus the fracture surface is very flat and perpendicular tothe tensile axis. In plate materials, brittle fractures show a characteristicchevron pattern (Figure 6). These chevron markings have a useful property, inthat they show the direction of crack propagation by pointing back to theorigin of fracture. Therefore, the possible cause of the brittle crack initiationmay be traced.

Figure 6 - Chevron markings on brittle fracture surface of grade A plate

Brittle fracture is initiated by a microcrack, which can form at a brittleintermetallic inclusion or other microstructural discontinuity. Once amicrocrack has formed, the large stress concentration ahead of the crack tipallows the crack to rapidly propagate through the structure, cleaving grains inits path. The initiation of a microcrack requires only a small amount ofenergy, and, as there is little plastic deformation, there is only a smallresistance to crack growth. Thus a brittle material has a low toughness. Asthere is little ductile deformation at the crack tip, significant crack tip bluntingdoes not occur, thus a brittle crack may grow right through a structure withlittle chance of crack arrest (providing that the load is maintained).

A brittle fracture surface has a brittle, shiny appearance. Under an electronmicroscope, the fracture surface can be seen to be composed of a number ofangular flat facets, corresponding to the fracture faces of the cleaved grains(Figure 7).

direction ofcrack growth

Page 9 of 35

Figure 7 - Brittle (cleavage) fracture of mild steel (x250)

The general brittle fracture mechanism described above is that oftransgranular (across the grains) fracture. However, brittle fracture may alsooccur by an intergranular (between the grains) mechanism (Figure 8). Anexample of this is a result of temper embrittlement, which may occur in lowalloy steels if they are heat treated at an incorrect temperature. Residualelements segregate along the grain boundaries resulting in local embrittlment.This provides a path along which brittle fracture can occur, even though thebulk of the material retains a high toughness.

Figure 8 - Intergranular brittle fracture in low carbon steel steel x50

intergranularcrack

Page 10 of 35

2.2.2 Ductile Failure

Ductile fracture (or shear) occurs as a result of extensive deformation of themicrostructure, and is promoted by shear stresses within the material. Thesestresses reach a maximum at an angle of 45° to the tensile axis, and ductilecrack growth will tend to take place along this plane. This results in thecharacteristic angular appearance of a ductile fracture surface (Figure 9).

Figure 9 - Ductile fracture surface showing shear at 45° to tensile axis in grade A plate

When the material is locally stressed above the yield point, shear (plastic)deformation occurs. As the stress in increased, the deformation becomes moreextensive until small holes (microvoids) become nucleated at microstructuraldiscontinuities (such as inclusions). These microvoids then grow together(coalesce), resulting in fracture (Figure 10). The important thing to note aboutductile fracture is that because a large amount of plastic deformation occurs, agreat deal of energy is required to produce failure. Thus there is a largeresistance to crack growth, so the toughness of a ductile material is high. Oncecrack growth has initiated, ductile deformation tends to blunt the crack tip,thus reducing the local stress concentration. Therefore, ductile fractures oftenstop (arrest) with relatively little crack growth.

Figure 10 - Schematic of ductile fracture mechanism

ductile fracture surface

nucleation of largevoid at inclusion

linkage of crack withvoid

plastic deformationahead of crack tip

linkage of crack withother voids

increasingstress

Page 11 of 35

At low magnifications, a ductile fracture surface appears dull and fibrous.However, under a microscope the characteristic fracture surface of ductilemicrovoid coalescence can be clearly seen (Figure 11).

Figure 11 - Ductile fracture surface of C-Mn steel (x250)

2.3 Ductile To Brittle Transition

As the ambient temperature is lowered, it becomes progressively moredifficult for ductile fracture to occur, as plastic deformation requires thermalenergy to operate. Eventually a point will be reached at which there isinsufficient thermal energy for ductile failure to occur, and brittle fracture willresult. This is known as the ductile to brittle transition temperature (DBTT).Plain carbon manganese steels typically have a DBTT of between -60 and+20°C. Note that the fracture mode does not instantly switch at the transitiontemperature, but changes gradually over a temperature range of 20-30°C.Thus fracture surfaces are often of mixed-mode appearance, and contain bothductile and brittle regions (Figure 12).

Figure 12 - Mixed mode fracture surface of C-Mn steel (x500)

brittlefacet

ductiletear

Page 12 of 35

The ductile to brittle transition will usually only occur in materials with abody-centred cubic (bcc) structure, such as ferritic steels. Materials with aface-centred cubic (fcc) structure, such as aluminium and copper alloys andsome stainless steels are much more ductile, and will remain so even at verylow temperatures. However that these materials can become embrittled ifbrittle compounds are precipitated within the structure, such as by anincorrect heat treatment.

2.4 Structural Thickness And Constraint

The fracture mode can also be influenced by the thickness of a structure. Thestress condition within a material is not uniform throughout the depth of asection (Figure 13). In the central region, the stress is distributed in threedimensions (triaxial), a condition known as plane strain. Here deformation ofthe microstructure is restricted by the triaxial stress state, and it is difficult forductile fracture to occur (i.e. stresses above the uniaxial yield stress arerequired to cause yield). This is known as a region of high constraint.

Towards the surface of a material, the constraint is reduced, as it is notpossible for a triaxial stress state to exist. Here the stress state is twodimensional or biaxial (known as plane stress) and it is much easier for ductilefailure to occur. This accounts for the characteristic shear lips which can befound at the edge of (even brittle) fracture surfaces.

Figure 13 - Change in stress state through depth of section

It should be noted that constraint is not the only reason that a thick section isparticularly susceptible to brittle fracture. During processing of the material, itis not possible to produce a uniform microstructure throughout a thicksection, thus the central region often has poorer mechanical properties thanthe surface.

stress, σ

σ

xyz

xz

plane stress(biaxial stress x,z)

plane strain(triaxial stresses x, y, z)

Page 13 of 35

3. Toughness Testing

In order to avoid brittle fracture in a structural material, it is necessary toperform tests to measure the material’s resistance to fracture, i.e. thetoughness. There is no single method of measuring the toughness of amaterial. Tests can be grouped into three main categories, which measuredifferent aspects of the material toughness:

• Charpy tests• fracture toughness testing• crack arrest tests

It is important to note that as cracks in service developed from cracks ornotches, the laboratory test testpieces must contain similar discontinuities.

3.1 The Charpy Test

3.1.1 Introduction

Figure 14 - The Charpy test

By far the most widely used test within industry for assessing the toughnessof metals is the Charpy V-notch impact test (Figure 14). This measures theenergy required to fracture a standard testpiece. The testpiece is loaded by aweighted pendulum at a very high strain rate (approximately 103/sec) andthe energy required to produce total fracture (crack initiation andpropagation) is measured. Note that similar tests with other shaped notches(such as the U-notch test) are sometimes used, but the V-notch is by far themost common and is the test required by LR Rules.

IMPACT LOAD

10mm

10mm 55mm

standard specimen

2mm deepV-notch

Page 14 of 35

Test machines normally measure the energy absorbed during fracture inJoules (J) over a range of 0-300J. If the energy absorbed during impact isrelatively high, a large amount of plastic (ductile) deformation will haveoccurred within the testpiece, thus the fracture mode is ductile. However, ifthe impact energy is relatively low, little ductile failure has occurred and thefracture mode will be mostly brittle. Standards (including LR Rules) usuallyquote a requirement for a minimum impact energy at a set temperature toguard against brittle fracture.

As well as impact energy, the fracture mode can be deduced from theappearance of the testpiece (Figure 15). A ductile testpiece will have a fibrous(dull) fracture surface, and will be heavily deformed, thus the lateralexpansion of the testpiece is high. A brittle testpiece will have a crystalline(shiny) fracture surface, and the lateral expansion will be low. Minimumlateral expansion and maximum percentage fracture surface crystallinityvalues are sometimes required by materials standards, but not by LR Rules.

189J impact energy 83J impact energy 21J impact energy0% crystalline 50% crystalline 90% crystalline2.1mm lateral expansion 1.3mm lateral expansion 0.2mm lateral expansion

Figure 15 - Fracture surface appearance according to failure mode (grade A material)

The overwhelming advantage of a Charpy test is that it is a quick, simple andcheap test to conduct. It is also easy to perform tests at a range of sub-ambienttemperatures. Therefore, the test is widely used for the batch testing ofmaterials for quality control and acceptance purposes.

However, a problem with the Charpy test is the small size of the testpiece.The testpiece thickness may be much smaller than the structure for whichmaterial is being used, thus the constraint of the testpiece is lower. Therefore,a more ductile fracture tends to occur in the testpiece than would happen inpractice. This makes it impossible to predict the behaviour of a crackedstructure, and thus design against failure. Also, the high rate of loading of thetestpiece is not usually representative of that in a real structure, and amachined notch does not truly reflect a crack in a real structure.

DUCTILE BRITTLEMIXED-MODE

Page 15 of 35

3.1.2 Test Procedure

Charpy testing should be performed to BSEN 10045-1 (Charpy Test forMetallic Materials, 1990) or an equivalent National Standard. Note that thetest is particularly sensitive to the size and shape of the machined notch, sothis should be checked before testing, using a suitable gauge. Details of thetest procedure requirements are contained within the LR Rules Chapter 2Section 3.

There is a degree of scatter in Charpy test data, particularly within the ductileto brittle transition temperature region. Therefore, it is standard practice toconduct the test upon three testpieces and calculate the average impactenergy of the set. This average must meet the minimum impact energyrequirement of the specification. However, according to LR Rules, oneindividual value less than this level can be accepted, provided that it is no lessthan 70% of the required energy. Further explanation of the acceptance andre-testing procedures for Charpy testing are contained within the SurveyProcedures Manual Part B Chapter 1.

If the material being tested is less than 10mm thick, it is not possible toproduce a full-sized Charpy testpiece. In this case a sub-sized testpiece(7.5x10x55mm or 5x10x55mm) can be used for testing. Obviously theconstraint of these thin testpieces is even less than that of a standard testpiece,thus the minimum impact energy requirement is not directly reducedaccordingly to fracture surface area. LR Rules specify the following minimumenergy requirement:

• testpiece 5x10x55mm 2/3 of tabulated energy• testpiece 7.5x10x55mm 5/6 of tabulated energy

For rolled materials, such as plates and sections, the microstructure isdirectionally deformed (anisotropic) and so the fracture properties varyaccording to notch orientation (Figure 16).

0

25

50

75

100

125

-60 -40 -20 0 20test temperature (°C)

Cha

rpy

impa

ct e

nerg

y (J

) longitudinaltransverse

Figure 16 - Variation of Charpy impact energy with testpiece orientation for mild steel(note - this is an example only and should not be used as a direct comparison)

rolling direction

longitudinal

transverse

Page 16 of 35

A longitudinal testpiece (notch perpendicular to the rolling direction) willgenerally have a higher Charpy impact energy than a transverse (notchparallel to rolling direction) testpiece. LR Rules now contain different impactenergy requirements according to the testpiece orientation. Note that for across-rolled plate, the direction of the final pass is taken as the principalrolling direction. Here the anisotropy in the Charpy values is much reduced.

When witnessing an impact test, it is important to note the type of Charpymachine used. The pendulum hammer radius (2mm) of BSEN/ISO standardmachines is smaller than that of ASTM type machines (8mm), which affectsthe loading geometry at the notch. This can give different test results foridentical material, particularly if the impact energy is large, as shown inFigure 17. LR Rule requirements are for BSEN/ISO machines.

0

50

100

150

200

250

300

350

-80 -60 -40 -20 0 20 40test temperature (°C)

Cha

rpy

impa

ct e

nerg

y (J

)

ISOASTM

Figure 17 - Variation in measured Charpy impact energy for ASTM and ISO standard impacttest machines for a structural steel (note - this is an example only and should not be used forenergy conversions between the two tests)

To counter the limited constraint of a Charpy testpiece, plates over 50mmthick, which are particularly susceptible to brittle fracture, usually have anincreased Charpy energy for acceptance purposes.

Page 17 of 35

3.2 Fracture Toughness Testing

The fracture toughness of a material is a measure of its resistance to crackinitiation. Unlike Charpy toughness, which is limited by the constraint of atestpiece, the fracture toughness of a material may be a material property.Therefore, the fracture toughness of a material may be used for designpurposes to predict the failure of a cracked structure (design against failure).The theory behind toughness and structural integrity (known as fracturemechanics) will be discussed later in the paper.There are several different parameters which can be used to assess thefracture toughness of a material, which include:

• KIC - plane strain fracture toughness• JIC - J-integral fracture toughness test• CTOD - crack tip opening displacement

The test procedures for these are all described in BS 7448-1 (FractureMechanics Toughness Testing, 1991). Generally, tests are conducted usingvery slow (quasi-static) loading rates, although high loading rate testing isoccasionally required for high strain rate applications. KIC and JIC are absolutematerial properties independent of thickness. However, this is only true forKIC above a minimum thickness, where the testpiece is of high constraint (i.e.a plane strain state predominates), as shown in Figure 18.

0

50

100

150

200

250

1 10 100specimen thickness (mm)

Figure 18 - Typical variation of KC with testpiece thickness

As well as testpiece thickness, the constraint of a testpiece is affected by itsductility. Thus for structural steels, which are normally partially ductile atambient temperature, the minimum testpiece thickness at which plane strainconditions are predominant (which can be calculated from the test standard)can be in excess of 1 metre.

KIC

AB

C A - plane stressB - transitionC - plane strain

KC

(MPa√√√√m)

Page 18 of 35

This thick section size is impractical for testing and not relevant to realstructures, thus CTOD testing is generally used for ductile materials (Figure19). This is the only fracture toughness test that is mentioned within LRrequirements, such as for the approval of a stud link chain cable manufacturer(LR MQPS Book H Procedure 10-1 Section 6), weldability tests (LR MQPSBook A Procedure 0-3 Section 3), and approval of controlled rollingprocedures (LR MQPS Book C Procedure 3-3 Section 5).

Figure 19 - A CTOD fracture toughness test

The CTOD test is described in the LR SPM Part B Chapter 2 Section 6. Agraphical record of force versus clip gauge extension is measured. The pointof crack initiation can be determined from a discontinuity in the force-extension curve (Figure 20), and the amount by which the crack tip hasopened (δ, measured in mm) is calculated. The resistance to crack initiation(i.e. fracture toughness) increases with increasing δ.

Figure 20 - Typical force-displacement curve for CTOD test of a structural steel

A

B

C

A - elastic strain

B - stable (ductile) crack growth

C - unstable crack growth

force

extension

critical δδδδ

Page 19 of 35

Unlike KIC and JIC, the CTOD is not a true material property, thus the testshould always be conducted on a full thickness testpiece with regard to thestructural application. Testpieces are usually of an SENB (single edge notchbend) or CT (compact tension) geometry (Figure 21). Prior to testing, a fatiguepre-crack is grown from the machined notch under alternating load, in orderto provide a sharp notch (thus high stress concentration) to initiate fracture.

Figure 21 - Typical fracture toughness testpieces

3.3 Crack Arrest Tests

Fracture toughness tests are used for predicting the susceptibility of a crackedstructure to fracture. However, they only relate to the initiation of a crack,thus do not predict whether a growing crack will arrest. If a brittle fractureinitiates in a material, it is important to try to ensure that it does not growthrough the entire structure. The crack arrest test can be used to determinethis will occur.

There are several crack arrest tests, the most common being the Drop WeightTear or Pellini test which is described in ASTM E208 (Standard Method forDrop Weight Test of Ferritic Steels, 1995). There are more sophisticatedfracture mechanics crack arrest tests, but these are beyond the scope of thispaper.

σσ

σσ

σ

σ

SENB specimen

CT specimen

fatigue precrack

Page 20 of 35

The Pellini test was developed by the US Navy Research Establishmentfollowing the Liberty ship failures. In order to initiate a brittle crack, adeliberately brittle weld bead is deposited upon a plate of full thickness andsubsequently notched. A weight with a known kinetic energy is then droppedonto the opposite face under controlled conditions (Figure 22), and the testplate is deformed by a fixed displacement.

Figure 22 - The drop weight tear (Pellini) test

Following the test, two criteria for acceptance/rejection exist, subject to visualexamination of the propagated crack (Figure 23). The first is the ‘break’condition. Here the crack must reach at least one edge of the plate. The secondis the ‘no-break’ condition, in which the crack has been found to reach neitheredge.

Figure 23 - Schematic of break and no-break condition Pellini testpieces

Tests are conducted upon at least six successive specimens to gauge thecritical break/no break test temperature. An iterative process is to increase ordecrease the test temperature of each subsequent test (according to a Nationalstandard procedure), in order to hone in upon the critical temperature. Thehighest temperature at which a break result occurs is defined as the NilDuctility Transition Temperature (NDTT).

Provided that the design temperature does not fall below some margin abovethe NDTT, it may be assumed that a brittle crack will not propagate through astructure. Determination of the NDTT is required for the approval of certainsteelworks, (e.g. LR MQPS Book C Procedure 3-1 Section 5),

no-break break

weld beadmachined

notch

Pellini testplate

Page 21 of 35

4. Material Effects

It has already been shown that structural steels are susceptible toembrittlement below a certain temperature. This ductile to brittle transitiontemperature (DBTT) is affected by a number of material parameters including:

• chemical composition• grain size and microstructure• strain rate

The DBTT must be low enough to ensure an adequately small risk of brittlefracture at normal operating temperatures. Firstly, it is important to definewhat the DBTT is.

4.1 Definition Of The Ductile To Brittle Transition Temperature

There are various methods of defining the ductile to brittle transitiontemperature. The most common method used in industrial specifications is tospecify the transition as occurring at some set Charpy impact energy value.For structural steels, this is usually taken as the temperature at which animpact energy of 27J is achieved (Figure 24), and this is widely used within LRRules. This implies that structural steels above this energy level will fracturein a largely ductile mode, and hence be resistant to brittle fracture at the testtemperature.

Figure 24 - Typical Charpy impact energy transition temperature for a structural steel

brittleDBTT

27J

test temperature

Charpyimpactenergy

ductile

Page 22 of 35

An alternative method of estimating the DBTT is by plotting a transitioncurve in terms of percentage crystallinity (Figure 25). Here the transitiontemperature (known as the fracture appearance transition temperature,FATT) is defined as that at which a fracture surface of 50% ductile, 50%crystalline appearance is achieved. Above this temperature, ductile failurepredominates and hence the material will be more resistant to brittle failure

Figure 25 - Transition temperature for a structural steel, based upon crystallinity

4.2 Chemical Composition

The DBTT of a material can be altered by varying the chemical composition ofa material. Figure 26 summarises the effects of various alloying elementsupon the transition temperature of a mild steel.

Figure 26 - Variation of Charpy transition curve of a mild steel with alloying additions

Carbon, which is added to strengthen the metal, and certain other elementsraise the transition temperature and can, therefore, embrittle the material.Elements that lower the transition temperature such as manganese improvethe toughness. Carbon raises the DBTT by 15°C for every 0.1wt% added,

brittle

DBTT

DBTT

50%

test temperature

test temperature

Charpyimpactenergy(CV)

Charpyimpactenergy

ductile

fracturesurfacecrystallinity (CX)

brittle

ductile

CPSiMoO

MnNiAl

CV

CX

Page 23 of 35

whereas manganese lowers it by 5°C for every 0.1wt.% added. Therefore, tocounter the effect of increased carbon content manganese should be added ina ratio of 3 to 1. LR Rules for Grade A rolled steel plates specify a manganeseto carbon ratio of at least 2.5 to 1 in order to achieve adequate toughness (LRRules Part 2 Chapter 3 Section 2.2).

Other elements, such as nickel, can be used to produce further reductions inthe DBTT. Note, however, that care should be taken with such highly alloyedsteels. An incorrect heat treatment can embrittle the microstructure anddrastically increase the DBTT. An example of this is temper embrittlement,which can occur in ferritic steels containing chromium, molybdenum ornickel, during slow cooling between 600 and 300°C. If impurity elements(especially antimony, phosphorous, tin or arsenic) are present, these segregateto the grain boundaries and cause intergranular fracture. Tests are sometimesrequired (LR Rules Part 2 Chapter 2 Section 5.1) to ensure that a material isnot susceptible to temper embrittlement.

4.3 Microstructure Effects

The processing method of a rolled material, which can alter themicrostructure, can also affect the DBTT. Materials with a fine grain sizegenerally have a lower DBTT and therefore improved toughness. Decreasingthe grain size of a mild steel from ASTM 5 to ASTM 10 can lower the DBTT byup to 60°C (Figure 27). Note that an additional advantage of decreasing thegrain size is that the yield strength will be increased.

-50

-40

-30

-20

-10

0

5 6 7 8 9 10ASTM grain size

DBT

T (°

C)

150

200

250

300

350

400

yiel

d st

reng

th (M

Pa)

DBTT

yield strength

Figure 27 - Typical variation of yield strength and toughness of a mild steel with grain size(note - this is an example only and will vary according to the material)

Page 24 of 35

The factors that can be used to decrease the grain size of steels include:• adding grain refining alloy additions (e.g. Al, Nb, V)• lowering the finishing temperature in as-rolled steels• normalising a finished product• using a thermo-mechanical controlled process (TMCP)

This can be a problem in the heat affected zone (HAZ) of a weld, where graingrowth may occur (depending on the heat input of the process), which willthus raise the transition temperature.

4.4 Strain Rate

The toughness of a material will also be affected by the rate at which it isloaded. For a material to fracture in a ductile manner, plastic deformation isrequired. This takes a finite amount of time, whereas brittle fracture(cleavage) can occur almost instantly. Therefore, if a ductile material is shockloaded (such as the wave slamming experienced against a ship hull in astorm), there may be insufficient time for it to undergo ductile deformation,and it may show a tendency towards brittle fracture.

It is not possible to alter the loading rate of a Charpy test, although this is ahigh loading rate test, and thus will provide a conservative measure of thetransition temperature. However, for fracture toughness testing, the DBTThas been shown to increase by up to 40°C by changing the loading rate fromslow (quasi-static) to shock loading.

4.5 Service Embrittlement

It is important to realise that the toughness of a material can degrade duringservice. An example of this is neutron embrittlement, which can beexperienced in nuclear reactor vessels. However, the most important serviceembrittlement mechanism is strain ageing.

Strain ageing may occur in ferritic steels with a high free nitrogen content. Ifthe material is cold strained (such as during fabrication processes such as coldforming or line heating), nitrogen atoms diffuse within the structure to thesestrained areas and produce an embrittling effect. This can increase the DBTTof the material by as much as 60°C. To counter this, LR recommends that thenitrogen level of the ladle analysis should not exceed 90ppm or 0.009wt.%(SPM Part B Chapter 3 Section 1.2), and that the cold strain applied to steelsections should not exceed 5% without some form of stress relieving heattreatment (LR Rules Part 2 Chapter 13 Section 7.3). In addition, strain ageembrittlement tests (LR Rules Part 2 Chapter 2 Section 5.2) are requiredduring the works approval of steel plates and sections, to ensure the materialsare not susceptible to strain ageing.

Page 25 of 35

5. Toughness Of LR Materials

This section will deal with the LR Rule requirements for the toughness(Charpy) testing of materials. As discussed in earlier sections, compositionalrequirements have been introduced into LR Rules to ensure adequate materialtoughness. Minimum ductility levels of tensile test specimens are alsorequired to guard against brittle fracture.

5.1 Normal Strength Ship Steels

There are four toughness grades for LR normal strength ship steels, which areclassified according to achieving a minimum Charpy impact energy at aspecified test temperature (LR Rules Part 2 Chapter 3 Section 2.4):

• Grade A tested at +20°C• Grade B tested at 0°C• Grade D tested at -20°C• Grade E tested at -40°C

It is important to realise that this is a test temperature only, and doesnecessarily reflect the minimum temperature to which a structure can beexposed. This is dealt with in the next chapter (materials selection). For steelswith a thickness of less than 50mm, an impact energy of 27J is consideredsufficient to avoid brittle fracture at the specified test temperature. However,if the section thickness of a steel exceeds 50mm, the required minimumCharpy energy is increased accordingly, to order counter the effects ofincreased constraint in the structural application.

Although only longitudinal oriented Charpy testpieces are usually required,transverse Charpy requirements (which are sometimes specified in a design)have a lower specified minimum energy to reflect the anisotropic toughnessproperties of a rolled material.

Note that impact tests are not directly required for grade A material less than50mm thick. However, regular in-house checks are required every 250t ofmaterial, to ensure that the material meets the required toughness.

5.2 Higher Strength Ship Steels

As the strength (UTS) of a steel is increased, in general its ductility andresistance to brittle fracture (toughness) are decreased. LR Rules sometimeslimit the UTS, or yield to UTS ratio, in order to ensure that high strengthmaterials have adequate ductility (LR Rules Part 2 Chapter 3 Section 3.6).

Page 26 of 35

Higher strength steels also generally have a higher minimum Charpy energyrequirement, which can be approximated (in J) as the minimum yield strength(in MPa or N/mm2) divided by ten:

• AH 32 grade (minimum yield strength 315MPa) - energy requirement 31J• AH 36 grade (minimum yield strength 355MPa) - energy requirement 34J• AH 40 grade (minimum yield strength 390MPa) - energy requirement 41J

These steels are similarly graded to normal strength steel according to testtemperature. However, there is no BH grade and an additional FH gradeexists.

• Grade AH tested at 0°C• Grade DH tested at -20°C• Grade EH tested at -40°C• Grade FH tested at -60°C

Impact energy requirements are also varied according to section thicknessand Charpy testpiece orientation.

Page 27 of 35

6. Material Selection For Fracture Control

Material selection for fracture control is the task of the naval architectdesigning a vessel, thus the requirements are contained within Part 3 of theLR Rules. Although this section only details materials selection according totoughness, it is important to note that other material properties (such asstrength and corrosion resistance) must also be considered.

6.1 General LR Rule Requirements For Ships

Under the general LR Rule requirements for materials selection regardingfracture control, each region of a vessel is assigned one of five differentmaterial classes (LR Rules Part 3 Chapter 2 Section 2.1). For each hullmember, this class reflects the:

• level of stress, for example the midships of a vessel tend to experiencehigher stresses than the bow or stern

• exposure to weather, as exposed material will experience lowertemperatures and may corrode at a higher rate than within a sheltered region,thus may eventually experience a higher stress and be more susceptible tobrittle fracture

• need to arrest cracks, as a brittle crack could be catastrophic for theintegrity of the vessel in some hull members, e.g. a sheerstrake, but low risk inothers

This class is then related to a toughness grade of steel, a higher class requiringa higher grade. The required grade also increases with section thickness,reflecting the increase in constraint for thick materials. Higher toughnessgrade materials are also required (for certain members) if a ship operates for aextended periods in low air temperatures or contains refrigerated spaces.

Page 28 of 35

6.2 Ice-Breaking Ships

Due to the ductile to brittle transition of structural steels at lowertemperatures, particular care has to be taken for ice-breaking ships. The LRrequirements for ice-class vessels (LR Rules Part 3 Chapter 2 Section 2.3)allow for specification of air temperatures of between 0 and -45°C. Materialselection for lower air temperatures must be given special consideration.

As for the normal LR Rule requirements, the hull members are classedaccording to location. The forward part of the vessel (which will experiencethe most severe ice conditions) has the most stringent material requirements.As well as stress however, the expected operating temperatures arespecifically considered. These are calculated from the minimum design airtemperature, making allowance for the degree of exposure to the elements.Thus the operating temperature of sheltered sections may be higher than theminimum design air temperature.

The grade of material is then determined from one of two graphs (Figure 28)according to minimum air temperature and structural thickness.

-45

-40

-35

-30

-25

-20

-15

-10

-5

0

0 10 20 30 40 50 60material thickness (mm)

oper

atin

g te

mpe

ratu

re (°

C) D

DH

E, EH

FH

Boundary lines formpart of the lower grade

Figure 28 - Selection of steel grade for exposed regions of ice-breaking ships

Page 29 of 35

6.3 Liquefied Gas Ships

A similar situation to that of ice-breakers exists in the design of liquefied gasships, where even lower temperatures are encountered in the containmentand transport of liquid fuel gases. Examples of temperatures for commonliquid gases are:

• propane (LPG) -45°C• ethylene (LEG) -104°C• methane (LNG) -163°C

The material requirements for vessels required to transport such gases aregiven in the LR Rules for Ships for Liquefied Gases Chapter 6 Section 2. Theproblem of using C-Mn structural steels is that even with grade FH material,the lower temperature is limited to -60°C. Therefore, special low temperaturematerials need to be employed. Figure 29 details the Charpy impact energy ofvarious candidate materials at low temperatures.

0

5

10

15

20

25

30

35

40

-250 -200 -150 -100 -50 0temperature (°C)

Cha

rpy

impa

cy e

nerg

y

stainless steel

Invar

9%Ni 5%Ni 3.5%NiLT60 C-Mn

LT40 C-Mn

LT20 C-Mn

Al alloys

Figure 29 - Typical variation of Charpy impact energy with temperature for various materials

Page 30 of 35

It can be seen that high nickel alloys (such as Invar, Fe-35%Ni), stainless steelsand aluminium alloys remain ductile even at very low temperatures.However, these materials are very costly and may exhibit other problemssuch as poor weldability or corrosion susceptibility.

A cheaper solution is to use ferritic steels with a low to medium nickel content(sufficient to provide a transition temperature suitable for liquid gas cargo)which are detailed in LR Rules Part 2 Chapter 3 Section 6. If an operatingtemperature of between 0 and -55°C is required, special (LT designated) highstrength carbon-manganese grades can be used. However, these are subject tospecial processing and compositional limits, and more extensive impacttesting than standard high strength materials. If the required operatingtemperature is between -60 and -165°C, nickel grade ferritic steels can be used,with a nickel content of between 1.5 to 9.0wt.%. The required material isselected according to design temperature and section thickness (Figure 30).

-200

-160

-120

-80

-40

0

0 5 10 15 20 25 30 35 40material thickness (mm)

min

imum

des

ign

tem

pera

ture

(°C

)

1.5Ni

3.5Ni5Ni

9Ni

Figure 30 - Minimum design temperature for nickel grade steels

Page 31 of 35

7. Fracture Mechanics

It has been shown that Charpy tests can be used for acceptance purposes toensure adequate toughness in service. However, such tests cannot be used toanalyse to assess the integrity of a cracked structure.

The theory behind the mathematical analysis of fracture is known as fracturemechanics. This enables a designer to conduct engineering criticalassessments, to demonstrate the fitness for purpose or safety of a structurewith crack-like defects. It is important to realise that a cracked structure willnot necessarily fail, and it is sometimes safer to leave a crack within astructure than attempting a repair, which could embrittle the microstructureor introduce residual stresses, for instance. Fracture mechanics is a highlycomplex subject, so this section is only intended as a brief introduction to thesubject. Further information can be obtained from the LRTA paper:Engineering Fracture Mechanics, A.Cameron and V.Pomeroy, Paper no.2, session 1985-6

7.1 Introduction

When a body containing a crack (of length a) is subjected to an applied stress(σ), a stress concentration (σ’) is produced ahead of the crack tip, which varieswith distance from the crack tip (x). Elasticity theory suggests that as the cracktip is approached (i.e. x→0), the stress concentration approaches infinity(σ→∞), as shown in Figure 31.

Figure 31 - Theoretical variation of stress concentration with distance from a crack tip

However, in practice no material can withstand an infinite stress, and materialclose to the crack tip yields by ductile deformation, resulting in the formationof a plastic zone (Figure 32). Note that the local stress at which crack tipyielding occurs may be greater than the uniaxial (tensile) yield stress, andvaries according to the specimen constraint. The yield stress is greater for a

x

σ’

σ

σ

a

σ

Page 32 of 35

triaxial stress state (plane strain) than a biaxial one (plane stress). Note thateven in a brittle material, some ductility always occurs at the crack tip

Figure 32 - Actual variation of stress concentration with distance from the crack tip

The stress concentration at the crack tip is quantified by a value K, the stressintensity factor. This parameter can be used to scale the stress distributionahead of the crack tip according to the applied stress, crack length andgeometry of the crack and surrounding structure (geometrical factor Y):

K Y a= σ πAs an increasing stress is applied to a cracked structure, the stress intensityfactor increases, until one of three possible failure mechanisms occurs:

Brittle failureIf the stress intensity factor reaches a critical value, which is known as KIC, thestress ahead of the crack tip is high enough to nucleate a brittle crack. Thiscrack propagates in a rapid, unstable manner through the structure, causing itto fail catastrophically. The critical stress intensity factor to produce brittlefracture is called the fracture toughness of a material. KIC is a materialproperty which can be measured in the laboratory, and can be used to predictthe occurrence of brittle fracture in engineering structures. This theory isknown as linear elastic fracture mechanics (LEFM), and generally holds trueas long as the size of the plastic zone is small compared to the crack, and thematerial is thick enough to experience plain strain conditions.

Stable (ductile) crack growthFor a ductile material, such as a structural steel at ambient temperature, theplastic zone size is much larger than that of a brittle material and the stressdistribution ahead of the crack tip is distorted. The size of the plastic zone isso large, that before sufficient stress is generated ahead of the crack tip to

x

σ’

σ

σ

plasticzone

σ

σf

Page 33 of 35

produce brittle fracture, local ductile failure occurs at the crack tip. Crackgrowth tends to be slow and stable, controlled by the plastic strain at thecrack tip region. This local strain is measured by the CTOD test, and hencecan be used to quantify the resistance to crack extension (fracture toughness).This theory is known as elastic plastic fracture mechanics (EPFM) and can beapplied to ductile or brittle materials, as local ductility always occurs at thecrack tip.

plastic collapseIn a very ductile material, the size of the plastic zone is so large that it reachesthe edge of a structure without any local ductile crack growth happening. Inthis event, the whole cross section of the structure fails uniformly by necking(reduction of area) or shear, a mechanism known as plastic collapse.

7.2 Design Against Failure

It is relatively straightforward to design against brittle fracture in a structure.This can be achieved by ensuring that the stress intensity factor (K) from themaximum design stress is below that of the fracture toughness (KIC).Similarly, a structure will only fail by plastic collapse if the design stressexceeds the flow stress of the material.

In practice, however, the failure mode of a structure will be between thesetwo extremes. One way of assessing the integrity of a structure against theseintermediate failure modes is to use a failure assessment diagram (FAD) of amaterial, such as shown in Figure 33.

Figure 33 - Typical failure assessment diagram

The two extremes of behaviour are linked by a curve separating a SAFEregion from a FAIL region. The shape of the curve can be derived from thestress/strain curve of the material. Both KR and SR can then be calculated for aparticular defect in a loaded structure, and plotted on the FAD. Inside the

KR = stress intensity factorfracture toughness

SR = load on structureplastic collapse load

KR

SR

SAFE

FAIL1.0

0.5

01.00.50

Page 34 of 35

SAFE region, the structure will be safe, but in the FAIL region there will be arisk of failure.

Failure assessment diagrams and fitness for purpose applications are outlinedwithin BS7910 (British Standard for Assessing the Acceptability of Flaws inMetallic Structures, 1999)

Further information on the application of fracture mechanics to engineeringstructures can be obtained from the LRTA paper:Fracture Mechanics Applications, D.Howarth and P.Pumphrey, Paper no.6, Session 1997-8

Page 35 of 35

8. Summary

This paper has introduced the basic aspects of fracture in metallic structures.

In summary:

• there are two basic fracture modes: ductile and brittle. Brittle fracturecan be particularly catastrophic for a structure, as it propagates in a rapid,unstable manner.

• carbon-manganese steel structures are subject to brittle fracture atlow temperatures. Therefore for low temperature service (<-55°C), specialhigh nickel grade steels and non-ferrous materials should be used.

• the susceptibility to brittle fracture increases with section thickness.

• Charpy testing is used as a quality control check to ensure theadequate toughness of structural materials.

• fracture toughness testing can be used to predict the integrity of acracked structure.