DAC3 March 2012 PDF [PDF]

Zitiervorschau

IIW/EWF Diploma Design and Construction (Advanced) DAC3

Training & Examination Services Granta Park, Great Abington Cambridge CB21 6AL, UK Copyright © TWI Ltd

Rev 2 April 2011 Contents Copyright  TWI Ltd 2012

IIW/EWF Diploma Design and Construction (Advanced) Contents Section

Subject

Pre training briefing 1 1.1 1.2 1.3

Introduction Background Course aim - Construction design modules Course objectives – European Welding Engineer

2 2.1 2.2 2.3 2.4 2.5 2.6 2.7 2.8 2.9 2.10

Design Considerations for Aluminium Aluminium alloys Choosing the appropriate filler metal Properties of aluminium Design of weld joints Limit state design Friction stir welding (FSW) Heat affected zone softening Summary Revision questions Bibliography

3 3.1 3.2 3.3 3.4 3.5 3.6 3.7 3.8 3.9 3.10 3.11 3.12 3.13 3.14 3.15 3.16 3.17 3.18 3.19 3.20 3.21

Different Types of Loading. Different loadings causes different ways to fail Static strength Ductile failure Stress concentrations Lamellar tearing Thickness High temperature Creep Cyclic loading and fatigue failure Impact behaviour Low temperature Brittle fracture The significance of flaws Fitness for service or engineering critical assessment Fracture mechanics LEFM and EPFM Fracture mechanics testing Calculating fracture toughness Testing welds BS 7910 ECA procedure ECA case study

www.twitraining.com

Rev 2 April 2011 Contents Copyright  TWI Ltd 2012

3.22 3.23 3.24

References and further reading Summary Revision questions

4 4.1 4.2 4.3 4.4 4.5 4.6 4.7 4.8 4.9 4.10 4.11 4.12

Advanced Fatigue – Part 1. Introduction Stress concentrations Fatigue performance Joint classification Fatigue failure Improving the fatigue strength of welded joints. Techniques that reduce stress concentration Techniques that introduce compressive residual stress Typical improvement Fatigue failure case studies Summary Revision questions

5 5.1 5.2 5.3 5.4 5.5 5.6 5.7 5.8 5.9 5.10 5.11 5.12

Static Loading. Review Stress and strain revision Fillet welds under stress Forces frames and beams Bending worked examples Torsion Stresses in cylinders Reinforcing steel Welding reinforcing steel Joints between reinforcing and other steel components Standards and specifications Revision questions

6 6.1 6.2 6.3 6.4 6.5 6.6 6.7 6.8 6.9 6.10 6.11 6.12 6.13 6.14 6.15

Design of Pressure Equipment Pressure vessel components Design-by-rule and design-by-analysis Details of design Nozzle design Reinforcement or compensation Flanges Corrosion allowance Weld joint efficiency Pressure stresses Vacuum and external pressure loading Types of pressure vessels Codes and standards References Summary Revision questions

www.twitraining.com

Rev 2 April 2011 Contents Copyright  TWI Ltd 2012

7 7.1 7.2 7.3 7.4 7.5 7.6 7.7 7.8 7.9

Advanced Fatigue – Part 2 Introduction S-N curves Fracture mechanics approach Variable amplitude Cycle counting Miner’s rule Fatigue assessment Summary Revision questions

8 8.1 8.2 8.3 8.4 8.5 8.6 8.7

Stresses in Welds Why calculate weld stress? Why joints often fail Load bearing welds Strength calculations for welds Overview of weld design Summary Revision questions

9

Revision Session

10 10.1 10.2 10.3 10.4 10.5 10.6 10.7 10.8 10.9 10.10 10.11

Appendix on Welded Joint Design Introduction Welds Types of joint Fillet weld Butt welds Dilution Welding symbols Welding positions Weld joint preparation Designing welded joints Summary

www.twitraining.com

Section 1 Introduction – Designing Things

Rev 2 April 2011 Introduction - Designing Things Copyright  TWI Ltd 2012

1

Introduction - Designing Things

1.1

Background An engineering structure is one that is designed and built to withstand loads for a specified period of time. These loads may arise from a wide range of sources and include self weight (such as buildings including the pyramids), external components (eg cars traversing a bridge), internal pressure (eg pipelines and boilers), environmental loads (due to wind, waves, ice, snow etc), reaction to an acceleration (eg rotating components) and many other sources. Furthermore, engineering structures are built using materials such as steel, aluminium, fibre reinforced composites that have been specifically selected to meet the lifetime demands of the structure. These materials are used to make components which are then assembled and joined together (usually by welding but not always) to form the structure itself.

1.2

Course aim - Construction and design modules The overall aim of the Construction and Design Modules of the European Welding Engineer training course is to: 

Provide guidance on how to design engineering structures so that they operate safely to satisfy specified performance targets.

The training is provided at three levels: European Welding Specialist, European Welding Technologist and European Welding Engineer. The present course is the third of these levels and is intended to cover the scope appropriate for a European Welding Engineer. Two other courses address the scope of the other qualifications.

1.3

Course objectives: European Welding Engineer The European Welding Specialist course provided an introduction to the principles of design and strength of materials. The Technologist course described the development of the principles towards formal design principles. The present course describes design methodologies in detail, providing the theoretical background to these approaches. The objectives of the European Welding Engineer course are therefore to enable attendees to:   

Be able to design welded joints to withstand static and cyclic loads and low temperatures. Be able to design both steel and aluminium joints. Understand the principles of design for pressure vessels.

1-1

www.twitraining.com

Section 2 Design Considerations for Aluminium

Rev 2 April 2011 Design Considerations for Aluminium Copyright  TWI Ltd 2012

2

Design Considerations for Aluminium

2.1

Aluminium alloys The designation system for aluminium alloys is in accordance with BS EN 573-1. Aluminium alloys are divided into grades according to their alloy content. The terminology of each grade is defined by a series of numbers. Wrought alloys have a four-digit number and are used for plates, extrusions and forgings. Cast alloys have a three-digit number to the left of the decimal point and one digit to the right of the decimal point. The temper tells how the product was fabricated and applies to both cast and wrought products. The temper can be as-fabricated (F), annealed (O), strain-hardened (H) or heat-treated (T). T is always followed by at least one digit specifying the conditions of tempering. Only heat-treatable alloys will be seen with the T temper designation. Non heat-treatable alloys will tend to have H designation instead. Table 1 Wrought alloys designation system. Alloy

Major alloying element

Heat treatable?

1XXX

99% minimum aluminium

No

2XXX

Copper

Yes

3XXX

Manganese

No

4XXX

Silicon

No

5XXX

Magnesium

No

6XXX

Magnesium and silicon

No

7XXX

Zinc

Yes

8XXX

Other elements

Yes

Table 2 Cast alloys designation system. Alloy

Major alloying element

1XX.X

99% minimum aluminium

3XX.X

Silicon with added copper and/or magnesium

4XX.X

Silicon

5XX.X

Magnesium

7XX.X

Zinc

8XX.X

Tin

9XX.X

Other elements

Heat-treatable alloys gain strength from cold work and from precipitation hardening. Non heat-treatable alloys gain strength from cold work only.

2-1

www.twitraining.com

Rev 2 April 2011 Design Considerations for Aluminium Copyright  TWI Ltd 2012

2.1.1

1xxx series alloys (commercially pure Al) The materials in this series comprise commercially pure aluminium in a range of purities. The third and fourth digits in the reference number define the purity indicating the minimum percentage of aluminium over and above 99%. Thus 1050 refers to a material having a purity of at least 99.50%. Possible purities range from 99.00-99.99%. Pure aluminium is weak with a maximum tensile strength of about 150N/mm². It is selected when corrosion resistance is critical, as in chemical plant. The higher the purity the better the corrosion resistance, but the lower the strength. In the annealed condition, pure aluminium is relatively soft and is used when high formability is needed. The most ductile version is superpurity (99.99% pure aluminium) produced from lower purity metal by a further zone-refining process.

2.1.2

2xxx series alloys (Al-Cu alloys) These comprise a range of alloys all containing copper, together with other possible elements such as Mg, Mn and Si. The 2xxx series comprises high strength products and is largely confined to the aerospace industry, mainly supplied to special aerospace standards, with closer control of manufacture than required by general engineering ones, thus pushing up the cost. In the naturally aged (T4) condition, these alloys have mechanical properties similar to mild steel, with a typical proof stress of 250N/mm², UTS approaching 400N/mm² and good ductility. In the full strength (T6) condition, the proof and ultimate stress can reach 375 and 450N/mm² respectively, but with reduced ductility. The good mechanical properties of the 2xxx series alloys are offset by various adverse factors, such as inferior corrosion resistance, poor extrudability, unsuitability for arc welding (since they are prone to liquation and solidification cracking) and higher cost. However, the corrosion resistance of thin material can be improved by using it in the form of clad sheet. The American civil engineering structures of the 1930s were all in 2xxx type alloy, using the ductile T4 temper. After 1945 the 2xxx series was superseded by 6xxx for such use despite its lower strength. Today there is little non-aeronautical use of 2xxx alloys.

2.1.3

3xxx series alloys (Al-Mn alloys) The relatively low manganese content of these alloys, sometimes with added magnesium, makes them half as strong again as pure aluminium, while retaining very high resistance to corrosion. Tensile strengths go up to 200N/mm² or more. In construction, the main application (in fully hard temper) is for profiled sheeting, used in cladding buildings and other structures. 3xxx series strip is also used for making welded tube (irrigation pipe, for example).

2-2

www.twitraining.com

Rev 2 April 2011 Design Considerations for Aluminium Copyright  TWI Ltd 2012

2.1.4

4xxx series alloys (Al-Si alloys) At one time architectural extrusions were occasionally produced in material of this type because of the attractive dark finish that can be produced on it by natural anodising, but today these Al-Si alloys seldom appear as structural components. Their high resistance to hot cracking means they are used for castings and weld filler wire.

2.1.5

5xxx series alloys (Al-Mg alloys) These represent the major structural use of the non-heat treatable alloys. The magnesium content varies from 1 to 5%, often with manganese added, providing a range of different strengths and ductilities to suit different applications. Corrosion resistance is usually excellent, although it is possible for the stronger versions to suffer abnormal forms of corrosion when operating in hot environments (this is avoided by keeping the magnesium content low and using a weaker grade in these conditions). The 5xxx alloys appear mostly as sheet or plate. At the lower end of the range they have good formability and are the natural choice for sheet-metal fabrications. At the upper end, they are used in welded plate construction, typically in the hot-formed condition. Such plate is tough and ductile with possible tensile strengths exceeding 300N/mm². The 5xxx series is little used for extrusions, which are only available in the annealed and hot-formed conditions with a rather low proof stress. Extrudability is poor compared with 6xxx and very thin sections are impossible. An acceptable practice is to use 5xxx series plating with 6xxx extrusions as stiffeners.

2.1.6

6xxx series alloys (Al-Mg-Si alloys) These materials, mainly containing Mg and Si, have the largest tonnage use of heat-treatable alloys. They combine reasonable strength with good corrosion resistance and excellent extrudability. Adequate solution treatment of extrusions can usually be obtained by spray quenching at the die and air quenching is possible with some of the alloys. The 6xxx materials are readily welded, but with severe local softening in the HAZ. Broadly they divide into stronger and weaker types. The stronger type of 6xxx material in the T6 condition (ie solution treatment followed by artificial ageing) is sometimes described as the mild steel of aluminium because it is the natural choice for stressed members. In fact it is a weaker material than mild steel with a similar proof or yield stress (250N/mm²), but a much lower tensile strength (300N/mm²) and is less ductile.

2-3

www.twitraining.com

Rev 2 April 2011 Design Considerations for Aluminium Copyright  TWI Ltd 2012

The weaker type of 6xxx alloy is not normally offered as sheet or plate, is the extrusion alloy par excellence. It is more suitable than any other alloy for the extrusion of thin difficult sections and is a common choice for members that operate at a relatively low stress level especially when good surface finish is important. Typical examples are when design is governed by stiffness (rather than strength) as with many architectural members or when fatigue is critical, for example in the structure of railcars. 2.1.7

7xxx series alloys (Al-Zn-Mg alloys) These alloys mainly contain zinc and magnesium. Like the 6xxx series they can be either of a stronger or weaker type. The stronger type embraces the strongest of all aluminium alloys, with tensile strengths in the T6 condition (ie solution treatment followed by artificial ageing) up to 550N/mm². The application of such alloys is almost entirely in aircraft. They have inferior corrosion resistance and poor extrudability. Like the 2xxx alloys, they are unsuitable for arc welding and sheet can be supplied in clad form to prevent corrosion. The weaker type of 7xxx material is a different proposition and in the nonaeronautical field is a valid alternative to the stronger type of 6xxx material, especially for welded construction. It has superior mechanical properties to 6xxx material and HAZ softening at welds is less severe. But corrosion resistance, although much better than for the stronger kind of 7xxx alloy, is not as good as for 6xxx alloy. Also there can be a possibility of stresscorrosion. Extrudability is not quite as good as for the 6xxx series, but hollow (bridge-die) extrusions are still possible. An important factor when using 7xxx series alloy as against 6xxx alloy, is the need for greater expertise in fabrication to avoid cracking.

2.2

Choosing the appropriate filler metal The filler type to use with the different alloys is given in Table 3 below. Table 3 Filler type to use for the welding of different aluminium alloys. Parent alloy 1XXX 3XXX 5XXX 6XXX 7XXX

Filler metal

Comments

1XXX

For corrosion resistance

4XXX

For strength and crack prevention

3XXX

For corrosion resistance

4XXX

For strength and crack prevention

5XXX

Select consumable based on parent composition

4XXX

For crack prevention

5XXX

For better weld strength

5XXX

Select consumable based on parent composition

2-4

www.twitraining.com

Rev 2 April 2011 Design Considerations for Aluminium Copyright  TWI Ltd 2012

2.3

Properties of aluminium Weight Aluminium exhibits a low density (2,702g/cm3), roughly one third that of steel so the deadweight of structures is dramatically reduced. This promotes its use in applications where weight is an important factor such as the automotive or aerospace industries. Corrosion resistance Aluminium has excellent corrosion resistance due to a thin but compact selfhealing oxide layer and can normally be used unpainted. Higher strength alloys will corrode in some hostile environments and may need protection. Nevertheless, serious electrolytic corrosion of the aluminium may occur at joints with other metals unless correct precautions are taken. Aluminium is also susceptible to stress corrosion cracking (SCC) in aqueous chloride solutions and tropical marine conditions. Take care when using chemical cleaning of surface oxide prior to welding joints with backing strips. Extrusion process The standard way of producing aluminium sections, is vastly more versatile than the rolling procedures in steel and is a major feature in aluminium design. Weldability Most of the alloys can be arc welded using gas-shielded processes, but are susceptible to fusion welding defects. The main welding defects in aluminium are porosity due to the presence of hydrogen, lack of fusion due to oxide inclusions, solidification cracking due to susceptible weld metal composition and softening of the HAZ (see Section 2.7). Due to the lower melting point (ie 660ºC), welding speeds are fast compared with steel. Laser welding and friction stir welding are now used extensively for joining aluminium components with the advantage of high processing speeds, ease of automation and low heat input (low distortion). Machinability Milling can be an economic fabrication technique for aluminium because of the high metal removal rates possible, hence, U or J preparations are much easier to produce. The use of machined preparations leads to tighter tolerances and better joint fit-up, reducing the amount of weld metal required to fill the preparation and avoiding possible weld defects caused by mismatch. Adhesive bonding Adhesive bonding is well established for making structural joints in aluminium and does not produce residual stresses or other defects, which can occur during welding. Unfortunately, adhesive bonded joints have limited life as most adhesive systems degrade rapidly when the joint is both highly stressed and exposed to a hot, humid environment.

2-5

www.twitraining.com

Rev 2 April 2011 Design Considerations for Aluminium Copyright  TWI Ltd 2012

Low temperature performance Having a face centred cubic (FCC) crystalline structure, aluminium has excellent strength and toughness at low temperatures, so is eminently suitable for cryogenic applications, because it is not prone to brittle fracture at low temperature in the way that steel is. Its mechanical properties steadily improve as the temperature reduces. Electrical conductivity Aluminium has a high coefficient of electrical conductivity, which combined with a lower price per kg compared with copper, makes it the standard material for overhead transmission lines conductors (with a central steel strand which carries the weight of the cable) and welding cables. Aluminium alloys have electrical conductivity approximately 65% that of copper but because of their density can carry double the electricity Thermal conductivity Aluminium has a high coefficient of thermal conductivity (237W/m°C - about four times greater than steel). As a result, pure aluminium can be used in heat exchangers as an alternative to copper tubes. Having a high coefficient of thermal conductivity, aluminium is capable of cooling the weld pool much faster than steel. Since heat is dissipated much more quickly, a larger included angle is required to prevent lack of sidewall fusion. If the included angle for a V preparation in a structural steel weldment is approximately 60º, this may need to be increased to 90º for aluminium. A high thermal conductivity means that a larger area will be heated up by welding, thus increasing distortion and giving wide HAZs. Magnetic properties Aluminium is non-magnetic which allows its use in applications where no electromagnetic interference is allowed. These applications include base plates for Metal oxide semiconductor field effect transistors (MOSFETs) and cases for avionic devices. Unfortunately, this means that magnetic particle examination cannot be used as a NDT method to detect surface/near surface defects in an aluminium weldment. Cost The cost of aluminium (sections, sheet and plate) is typically about 1.5 times that of structural steel, volume for volume. For aircraft grade material, the differential is much more. However, fabrication costs are lower because of easier handling, use of clever extrusions, easier cutting or machining, no painting and simpler erection. In terms of total cost the effect of switching to aluminium is usually much less than one would expect and an aluminium design can even be cheaper than a steel one. The other factor on cost is aluminium’s relatively high scrap value; a significant factor in material selection for components designed to have a limited life.

2-6

www.twitraining.com

Rev 2 April 2011 Design Considerations for Aluminium Copyright  TWI Ltd 2012

Young’s modulus Aluminium exhibits a low Young’s modulus value, 0.7×105N/mm2, a third that of steel. The failure load for an aluminium component due to buckling is lower than for an equivalent steel one. Since Young’s modulus is a measure of stiffness, a lower value means less rigidity of an aluminium structure compared to an equivalent steel one. Consequently, elastic deflection becomes more of a factor than in steel and in aluminium it is often the limiting failure condition in beam design, whereas an identical structural steel component subjected to the same stress would be limited by yielding rather than buckling. Fatigue Aluminium components are more prone to failure by fatigue than steel ones. The stresses required producing the same rate of fatigue crack propagation in steels and aluminium alloys are roughly in proportion to their respective Young’s modulus, ie three to one. As a consequence, all standards have different S-N lines for aluminium welds (see BS 8118 and AWS D1.2). It may observed that the yield strength, fatigue crack propagation rate and Young’s modulus of aluminium is roughly a third those of steel. Therefore one third is often accepted as an approximate to obtain the equivalency between steel and aluminium. High temperature service Aluminium loses strength more quickly than steel with increasing temperature. Some alloys begin to lose strength when operating above 100°C. Aluminium structures have a limited upper service temperature and are not intended for creep resisting applications. Limit stress Aluminium does not present a clear yield point; instead its stress-strain curve shows continuously rising smooth behaviour once it deviates from the linear elastic region. So to define a useable limit for stress, proof stress is used (ie the stress at which the material undergoes a certain permanent strain, commonly 0.2%). Thermal expansion Aluminium expands and contracts with temperature approximately twice as much as steel – its coefficient of thermal expansion being 24 × 10-6ºC-1 compared with only 11 × 10-6ºC-1 for steel. Greater thermal expansion leads to greater distortion after welding; twice as much distortion for an aluminium structure compared with a steel one. Because of the lower Young’s modulus, thermal residual stresses in a restrained member are only twothirds those in steel, the rest being locked in distortions.

2-7

www.twitraining.com

Rev 2 April 2011 Design Considerations for Aluminium Copyright  TWI Ltd 2012

Tensile strength Pure aluminium has a modest ultimate tensile strength (UTS) (70-150N/mm2 depending on delivery condition - annealed or cold worked). For structural applications aluminium is alloyed with different elements, thus increasing its tensile strength up to 650N/mm2 (see Section 2.1). Each alloy system has a composition where is it susceptible to hot cracking. The engineer must also consider the effect of different welding processes, weld preparations and parameters to avoid the combination of dilution, consumable and parent composition making the weld at risk of hot cracks. Weld pool fluidity Molten aluminium has a high fluidity so the weld pool can spill out of the joint preparation, leading to possible overlap and burn-through problems. To avoid this, a smaller (or no) root gap is used and sometimes a backing trip is used. Affinity to oxygen Aluminium forms a tenacious oxide film, which has a melting point some three times higher than that of aluminium. Failure to remove this oxide both before and during welding results in entrapment of oxides and/or incomplete fusion, giving a joint with impaired mechanical properties. The importance of gas shielding when welding aluminium over a wider weld width requires larger diameter gas nozzles for TIG and MIG welding which leads to an increase in included angle and/or of land in the case of U preparation. If present on the weld preparation surface, the oxide layer can prevent proper melting of the parent material, leading to wetting problems and lack of fusion. To produce a sound weld the oxide layer needs to be removed by mechanical or chemical methods. Chemical cleaning must be considered from the design stages as the reagents are highly corrosive so permanent backing strips and lap joints should be assembled after chemical cleaning due to possible entrapment. Loss of strength at welds There are three considerations with respect to the strength of welded joints in aluminium alloys:   

Alloys which gain their strength by cold work are softened by welding. Alloys which are precipitation hardened are overaged by the effects of welding. The weld metal may not match the strength of the parent metal because of its as-cast structure and perhaps because of its composition which may have been selected to reduce the risk of hot tearing (for example using a crack-resistant Al-Si filler to weld Al-Cu alloys).

The issue of HAZ softening is discussed further in Section 2.6.

2-8

www.twitraining.com

Rev 2 April 2011 Design Considerations for Aluminium Copyright  TWI Ltd 2012

2.4

Design of weld joints

2.4.1

Design to overcome the limitations of aluminium BS EN ISO 9692-3 gives details of various weld joints for aluminium alloys. The higher fluidity of aluminium means that a smaller root gap is used compared to steel to avoid burn through and minimise distortion. The larger size gas nozzle required for TIG and MIG welding and the higher thermal conductivity means that a wider bevel angle is used for access and better fusion. However a wider bevel also means greater distortions when welding. It may be possible to ensure adequate access by using a longer land on a U-prep, providing adequate fusion can be ensured.

If the root gap is greater than 1.5mm, a backing bar is recommended (this can allow for a reduction of included angle and consequently a reduction of distortion). Remember that a backing bar can be a location for crevice corrosion if cleaning flux has not been removed before welding.

When welding sections of different thickness, normally a tapered transition is machined onto the thicker section to reduce stress concentration effect of the joint, particularly when the joint will be exposed to cyclic loading in service. For aluminium welds, the higher heat sink effect of the thicker member in this joint design might result in burn-though of the thinner sheet, so a modification is to machine an extended land onto the end of the taper so the weld is between materials of equal thickness. Use intermittent welds rather than a continuous run (for example, when welding stiffeners), can reduce the amount of welding and therefore distortion. For these kinds of joints, it can be possible to still obtain adequate joint strength from an intermittent weld.

2-9

www.twitraining.com

Rev 2 April 2011 Design Considerations for Aluminium Copyright  TWI Ltd 2012

To mitigate the effects of HAZ softening, the weld area cross-section can be increased locally by using an extruded section.

Weld placement is another important factor in designing aluminium structures. If a weld is positioned close to the neutral axis of a fabrication where the bending stresses are least, although the weld will have a softened HAZ, this will not affect the load carrying capacity of the structure. As a general rule, position welds in areas of low stress to cancel the effect of loss in strength in the HAZ. Examples of poor and good designs are shown below. Neutral axis

Neutral axis

Poor design

Good design

When members must be joined at locations of high stress, the welds should be parallel to the main stress in that member. Transverse welds in tension members should be avoided if possible. Welding can often be eliminated at the design stage by forming the plate or using an extruded section, as shown below.

2-10

www.twitraining.com

Rev 2 April 2011 Design Considerations for Aluminium Copyright  TWI Ltd 2012

2.4.2

Design to exploit the advantages of aluminium Special extrusions that incorporate edge preparations for welding provide savings in manufacturing costs. An integral lip can be provided on the extrusion to facilitate alignment and serve as a weld backing.

In light structures, snap-lock or snap-fit joints can be used with extruded aluminium sections in order to eliminate welded joints. Interlocking joint can be designed with a folding, locking flange which prevents counter-rotation and disassembly

2-11

www.twitraining.com

Rev 2 April 2011 Design Considerations for Aluminium Copyright  TWI Ltd 2012

2.5

Limit state design

2.5.1

Approach When checking whether a component is structurally acceptable, there are three possible limit states (or possible failure modes) to consider. The limit state approach to structural design is presented in BS 8118.

2.5.2

Static strength Ensuring that the structure has adequate static strength is usually the governing requirement for design and must always be checked. The structure must be able to resist a reasonable static overload, over and above the specified nominal loading, before catastrophic failure occurs in any of its components. It involves static calculations similar to those used for steel structures, taking into consideration the HAZ softening. More specifically, the structure is analysed under working load and stress levels must not exceed an allowable stress, obtained by dividing the material strength (usually the proof stress) by a factor of safety.

2.5.3

Deflection The limit state of elastic deflection (sometimes called serviceability) ensures the structure has adequate stiffness. It is important in beam design since the low Young’s modulus of aluminium causes it to be more of a concern than in steel. Calculations are usually performed with the structure subjected to nominal loading (ie dead weight of the structure, imposed loads, wind loads, forces due to thermal expansion/contraction, etc). This limit state is usually about the performance of members rather than joints. In elastic deflection calculations, the allowable stress is reduced to allow for buckling.

2.5.4

Fatigue Fatigue must be considered for all cases of repeated loading since fatigue in aluminium structures is more critical than for steel. The usual checking procedure is to identify potential fatigue sites and determine the number of loading cycles to cause failure. The design is acceptable if the predicted life at each site is not less than that required. The number of cycles to failure is normally obtained from an endurance curve, selected according to the local geometry and entered at a stress range based on the nominal loading. Alternatively, for a mass produced component, the fatigue life can be found by testing.

2.6

Friction stir welding (FSW)

2.6.1

FSW process Friction stir welding (FSW) was invented at TWI in 1991 and consists of using the friction of a tool between the two abutted aluminium plates to be joined. Unlike conventional welding, the metal is not melted but simply heated to a temperature which allows for plastic deformation of aluminium. Frictional heat generated between the wear resistant welding tool and the material of the work pieces causes the material to soften without reaching

2-12

www.twitraining.com

Rev 2 April 2011 Design Considerations for Aluminium Copyright  TWI Ltd 2012

the melting point and allows the tool to traverse smoothly along the weld line. The plasticised interface material is mixed by being transferred from the leading edge of the tool to the trailing edge of the tool probe and is forged by intimate contact of the tool shoulder and the pin profile. It leaves a solid phase bond between the two pieces.

The process is a solid phase keyhole welding technique since a hole to accommodate the probe is generated, then filled during the welding sequence. The use of FSW is rapidly growing in all industries and active research is producing new developments in the technique. It is used extensively for the welding of aluminium since it has a number of benefits over arc welding for this material:         

No fume or spatter. No filler or shielding gas required. Very good weld quality. Low distortion. Excellent mechanical properties (no HAZ softening). Can operate in all positions. Energy efficient. Low shrinkage. Can weld above 50mm thickness in one pass.

The main limitations of FSW are:    

Workpieces must be rigidly clamped. Backing bar required (except where self-reacting tool or directly opposed tools are used). Keyhole at the end of each weld, need run-off plates. Cannot make joints which required metal deposition (eg fillet welds).

2-13

www.twitraining.com

Rev 2 April 2011 Design Considerations for Aluminium Copyright  TWI Ltd 2012

FSW is now largely used for the manufacture of car body trains, cars, ship hulls and fuel tanks for aerospace applications and provides a significant advantage to the aluminium extrusion industry. However, the cost of a FSW machine is still significant and research is ongoing to design low cost machines. 2.6.2

FSW joint configurations Due to its special features, FSW can require special joint configurations.

a) b) c) d) e) f) g) h)

2.7

Square butt. Combined butt and lap. Single lap. Multiple lap. Three piece T butt. Two piece T butt. Edge butt. Possible extrusion design to enable corner fillet weld to be made.

Heat affected zone softening There tends to be a serious local drop in strength in the HAZ at welded joints in most aluminium alloys. It is not possible to recover this loss of strength by subsequent cold work in cold-worked alloys but partial recovery may be achieved in some heat-treated alloys. (If welding material in the annealed condition, weld and HAZ strength may match that of the parent metal). It is necessary for design purposes to allow for this reduction in strength, by considering two factors; the severity of the softening and the extent of the softening. BS 8118 structural use of aluminium gives guidance on what is termed the softening factor and on the width of the softened HAZs. It is usually necessary to move welds to areas of low stress or increase the thickness to compensate for the strength loss.

2-14

www.twitraining.com

Rev 2 April 2011 Design Considerations for Aluminium Copyright  TWI Ltd 2012

2.7.1

Interpass temperature control The extent or width of the softened HAZ can be affected by the control of temperature during fabrication. The interpass temperature, (To) needs to be limited. BS 8118 recognises two levels of thermal control, normal and strict, as follows: Alloy series

Normal control

Strict control

5xxx and 6xxx series alloys

To  100ºC

To  50ºC

7xxx series alloys

To  80ºC

To  40ºC

The interpass temperature To may rise if:     2.7.2

Workpiece is still hot from the welding of a nearby joint. Insufficient cooling time between passes in a multi-pass weld. Use of preheat. High ambient temperature.

Extent of the softened zone When performing calculations for welded members, the accepted practice is to use a nominal HAZ as an approximation of the true pattern of softening. A weakened region of uniform strength is assumed adjacent to the weld, beyond which a step change occurs to full parent material strength. This step change is considered to occur midway between the strength of parent material and the lowest strength in the HAZ.

2-15

www.twitraining.com

Rev 2 April 2011 Design Considerations for Aluminium Copyright  TWI Ltd 2012

To determine the width of nominal HAZ, the one inch rule was devised which states that the nominal HAZ extends 1 inch (25mm) in every direction from an appropriate reference point in the weld. For a butt weld the reference position is the centre line of the weld while for a fillet weld it is at the root. With fillet welds on thick plate, the one inch rule allows the HAZ boundary to be taken as an arc.

2.7.3

Severity of HAZ softening The severity of softening in the HAZ is defined by the softening factor kz which represents the ratio of HAZ strength to parent metal strength: kz 

HAZ proof stress Parent metal proof stress

or k z 

HAZ UTS Parent metal UTS

The corresponding values for HAZ proof stress/UTS can be approximated by hardness surveys. For precipitation hardened alloys, kz values are based on previous experience and can be found in standards as a function of thermal control (see BS 8118, Eurocode 9 and AWS D1.2). It may be assumed that the material in the HAZ of aluminium welds has the same properties as those for annealed material, so for cold worked alloys the kz factor becomes the ratio of the proof strength for the annealed and cold worked conditions. In precipitation hardened alloys, it’s the ratio of strength in the annealed and hardened conditions. kz 

Parent metal strength (annealed) Parent metal strength (hardened or cold worked)

Once the softening factor is determined, an effective plate thickness is assumed in each HAZ region instead of the true, real thickness of the component t: teffective = kz × treal

2-16

www.twitraining.com

Rev 2 April 2011 Design Considerations for Aluminium Copyright  TWI Ltd 2012

The resistance of the weld joint is based on the effective section instead of the actual one, assuming it to be entirely composed of full strength (unsoftened) material.

2.8

Summary At the end of this module you should be able to design aluminium profiles and weld joints for a given use. You should be able to discuss how to avoid common imperfections in aluminium joints and interpret softening in the HAZ. You ought to be able to explain the strength of different alloys and select alloys for given applications.

2-17

www.twitraining.com

Rev 2 April 2011 Design Considerations for Aluminium Copyright  TWI Ltd 2012

Revision Questions 1 For a structure requiring corrosion resistance and good strength what type of heat treatable aluminium alloy should be chosen?

2 What joining techniques can be used for aluminium, what are the advantages and disadvantages of each?

3 How do you determine the severity of HAZ softening? What can you assume about the extent of softening?

4 A manufacturer intends to weld aluminium plates together to make an I-beam. Sketch an improved design.

5 What causes lack of fusion defects in aluminium welds? How can they be avoided?

2-19

www.twitraining.com

Rev 2 April 2011 Design Considerations for Aluminium Copyright  TWI Ltd 2012

Bibliography Aluminum Design Manual’: by The Aluminum Association Ltd, 2000. AWS D1.2: ‘Structural Welding Code – Aluminium’. American Welding Society. BS 8118-1: ‘Structural use of aluminium. Code of practice for design’, British Standards Institution, London. BS EN 1999: ‘Eurocode 9: Design of aluminium structures. General structural rules’, British Standards Institution, London. Bull J W 1994: ‘The practical design of structural elements in aluminium’. Avebury Technical, UK. Conserva M, Donzelli G and Trippodo R, 1992: ‘Aluminium and its applications’, Edimet, Brescia, Italy. Dwight J, 1999: ‘Aluminium design and construction’. by E&FN Spon. Gene Mathers 2002: ‘The welding of aluminium and its alloys’. by Woodhead Publishing Ltd, UK.

2-20

www.twitraining.com

Section 3 Different Types of Loading

Rev 2 April 2011 Different Types of Loading Copyright  TWI Ltd 2012

3

Different Types of Loading

3.1

Different loading causes different ways to fail The various ways that a structure is loaded will critically affect the ways in which it may be expected to fail. Cyclic loading will put a structure at risk of fatigue cracking. A structure under static loading only may be designed to withstand ductile overload, but changes in the temperature can affect the strength. At very high temperatures creep can occur. At low temperatures steel is at more risk of brittle fracture. In these circumstances it is important to understand the structure, not just based on strength but also the fracture mechanics. The connection between the different types of loading and expected failure modes will be explored in this section. There are failure modes which occur in an Instant, such as buckling, fracture or overload (although some deformation or bulging might give a little warning of failure). Time-dependent failure processes occur over time and with sufficient in-service inspection, the damage might be detected before it reaches critical conditions for failure. These failure modes include fatigue (and corrosion fatigue), stress corrosion cracking and creep. The designer must be sure that the design will not suffer from an instantaneous failure mode during service.

3.2

Static strength To ensure that there is no plastic deformation in structures engineers usually design each structural member to operate at a static stress equivalent to up to two thirds of the yield stress. This leaves a safety margin, so that the structure can tolerate further loading up to 1/3 of the yield strength, without plastic deformation (some design codes limit the applied stress to only half the yield strength). In addition to service loads, other loads that can act on structures are most commonly due to adverse weather conditions such as added weight due to snow fall or additional forces due to high winds. When a structure is overloaded so that it exceeds its static strength, it fails by ductile failure.

3.3

Ductile failure Ductile failure, or plastic collapse, occurs when yielding and deformation precedes failure. It is the result of overloading. Purely ductile failures are rare since most structures are designed well within their load bearing capacity. Ductile failures are most likely to occur in service as a secondary failure mode after the section thickness has been reduced as a result of fatigue crack growth corrosion or erosion. When examining the fracture surface, a ductile failure will show evidence of gross yielding or plastic deformation. The fracture surface is rough and torn and may be highly fibrous as a result of the deformation taking place. The failure surface may show 45° shear lips or have surfaces inclined at 45° to the load direction.

3-1

www.twitraining.com

Rev 2 April 2011 Different Types of Loading Copyright  TWI Ltd 2012

Figure 3.1 Ductile rupture of a component. Figure 3.2 Ductile fracture surface.

3.4

Stress concentrations Even if a structure is designed to be stressed below, for example, two thirds of the yield strength, it is possible for the stress to become much higher at specific locations, due to the effect of stress concentrations. Sudden changes in geometry in a material section (such as holes, notches, grooves, corners, fillet welds, or defects) can act as stress raisers which concentrate the stress so that the local stress increases. The concept of stress concentration can be visualised by considering flow lines. The lines of transmission of the stress are similar to the flow lines if fluid were to enter the material from one end to the other, as in Figure 3.3. Densely packed flow lines are representative of the concentration of stress at those points.

Figure 3.3 Stress concentration around a notch.

Figure 3.4 shows the stress concentration effect of a circular hole in a large flat plate under tension. Even this simple detail increases the stress at the edge of the hole by a factor of 3. This means that close to the edges of holes (such as bolt holes), the maximum stress in the plate is about three times the nominal applied stress. Sharper notches concentrate the stress by much more and very sharp notches, such as cracks have a very high concentration of stress at their tips.

Figure 3.4 Stress concentration due to a circular hole.

3-2

www.twitraining.com

Rev 2 April 2011 Different Types of Loading Copyright  TWI Ltd 2012

Welds often lead to geometric stress concentrations at the weld toe. This feature is especially common in oversized welds as depicted in Figure 3.5a) and these welds can be ground down to reduce the severity of the stress concentration at the weld toes (Figure 3.5b).

b

a

Figure 3.5a Oversized weld with stress concentrations and 3.5b mitre weld.

3.5

Lamellar tearing When plate is stressed in the through-thickness orientation and has insufficient ductility in that direction then lamellar tearing can occur. This failure mode is specific to particular joint geometries and can occur beneath welds especially in rolled steel plate which has poor through-thickness ductility. These are plates with high concentration of elongated inclusions oriented parallel to the surface of the plate and cause the lamellar tearing to have a stepped appearance. The main distinguishing feature of lamellar tearing is that it occurs in T butt and fillet welds and it is normally observed in the parent metal parallel to the weld fusion boundary and the plate surface as shown in Figure 3.6.

Figure 3.6 Typical location of lamellar tearing in T-joints.

The cracks can appear at the toe or root of the weld but are always associated with points of high stress concentration. The surface of the fracture is fibrous and woody with long parallel sections which are indicative of low parent metal ductility in the through-thickness direction as can be seen in Figure 3.7.

3-3

www.twitraining.com

Rev 2 April 2011 Different Types of Loading Copyright  TWI Ltd 2012

Figure 3.7 Typical lamellar tearing fracture face.

It is generally recognised that there are three conditions which must be satisfied for lamellar tearing to occur. These conditions include: Transverse strain Shrinkage strains on welding must act in the short direction of the plate ie through the plate thickness. Weld orientation Fusion boundary will be roughly parallel to the plane of the inclusions. Material susceptibility Plate must have poor ductility in the through-thickness direction. The risk of lamellar tearing will be greater if the residual stresses generated from welding act in the through-thickness direction, most likely to occur in Tjoints, corner joints or cruciform joints. Lamellar tearing is more likely to occur in large welds typically when the leg length in fillet and T butt joints is greater than 20mm. The welding process, consumables and preheating are also related to and can all help reduce the risk of tearing. In butt joints, as the stresses on welding do not act through the thickness of the plate, there is little risk of lamellar tearing. Lamellar tearing is only encountered in rolled steel plate and not forgings and castings. There is no one grade of steel that is more prone to lamellar tearing but steels with a low short transverse reduction in area (STRA) will be susceptible. As a general rule, steels with STRA over 20% are essentially resistant to lamellar tearing whereas steels with below 10 to 15% STRA should only be used in lightly restrained joints. Steel suppliers can provide plate which has been through-thickness tested with a guaranteed STRA value of over 20%, this is sometimes called Z-grade steel.

3.6

Thickness The relationship between component thickness and mode of failure is somewhat counter intuitive. Thicker sections, although able to carry greater load, do not better resist crack propagation.

3-4

www.twitraining.com

Rev 2 April 2011 Different Types of Loading Copyright  TWI Ltd 2012

When a material is subject to tensional forces they elongate. However, during this elongation there are effectively also compressive forces acting perpendicular to the direction of tension at the notch due to poisson’s contraction as illustrated in Figure 3.8. A thinner material can simply contract from the two surfaces to accommodate this strain. This condition is called plane stress and is also what happens near the surfaces of a thick plate, where lateral contraction at the surface is possible. However in the very middle of a thick plate, the material cannot deform laterally when loaded under tension because it is constrained by the surrounding material. This condition is called ‘plane strain’.

Figure 3.8 Illustration of Poisson’s contraction.

Consider two rectangular beams of different thicknesses but with identical notches cut into them as in Figure 3.9. With less plastic deformation possible at the centre of a thick plate due to this constraint, a crack in the middle of a thick plate will have a smaller plastic zone ahead of the crack at the centre of the plate. A small plastic zone means the crack is more likely to propagate in a brittle manner rather than behave in a ductile fashion.

Figure 3.9 Notched specimens of various thicknesses.

Although near the plate surfaces under plane stress the plastic zone will still be large, in thick plate the proportion of plane strain conditions along the middle of the crack front can dominate the behaviour of the crack (Figure 3.10). Therefore the same material but in a thicker plate will have a lower fracture toughness than thinner plate.

3-5

www.twitraining.com

Rev 2 April 2011 Different Types of Loading Copyright  TWI Ltd 2012

Figure 3.10 Shape of the plastic zone ahead of the crack tip under plane stress and plane strain.

3.7

High temperature In general, steels exhibit a reduction in tensile strength at elevated temperatures. However, certain steels can be seen to exhibit trends showing a decrease and a subsequent increase in strength at certain temperature ranges. This characteristic is also dependant on the strain rate. The reduction of strength at high temperatures can make a structure more at risk of ductile failure at high temperature, if this has not been accounted for in design. At very high temperatures, such as those experienced in power stations, failure may also occur due to creep.

Figure 3.11 Influence of high temperature on different metallic materials.

3-6

www.twitraining.com

Rev 2 April 2011 Different Types of Loading Copyright  TWI Ltd 2012

3.8

Creep Creep is the progressive deformation of a material under constant stress. For most steels, creep is a high temperature deformation mechanism and is a function of:    

Steel composition and grain size. Stress level (pressure, residual stress etc). Metal temperature. Time at temperature.

A creep test can be carried out in a rig similar to that of a tensile testing machine. The applied load is held constant and the temperature of the sample is elevated through the use of a heated filament that surrounds the sample as illustrated in Figure 3.12. The extension (or creep deformation) is measured throughout the test. Failure occurs once the specimens has finally necked down and ruptured.

Figure 3.12 Schematic of a creep test.

Creep tests are plotted on a strain–time axis which takes the form of curve as depicted in Figure 3.13.

Figure 3.13 Typical creep curve produced from test results.

3-7

www.twitraining.com

Rev 2 April 2011 Different Types of Loading Copyright  TWI Ltd 2012

Three distinct stages exist within the creep curve. The first stage is known as primary creep and is due to dislocations orienting themselves within the microstructures. The second stage is called secondary creep or steady state creep. The creep is linear and is due to the effects of both work hardening and recovery of dislocations being balanced. Lastly, the third stage is referred to as tertiary creep and the rapid increase in strain at this stage is due to void formation, necking and general reduction in area which further accelerates the rate of strain until creep fracture occurs. Creep damage in structures is manifested by the formation and growth of creep voids or cavities within the microstructure of the material due to tertiary creep. 3.8.1

Creep resistance Coarse grained structures offer the greatest creep strength and carbide forming elements enhance resistance. Materials which are resistant to creep have additions of chromium for corrosion resistance and carbide formation and molybdenum for microstructural stability. Creep resistant chromiummolybdenum steels have varying weldability, with an increasing percentage chromium decreasing the weldability.      

C-Mn steels. 0.5Mo steel. 1.25Cr-0.5Mo steel. 2.25Cr-1Mo-(V) steel. 5Cr-0.5Mo steel. 9Cr-1Mo steel.

Another option for enhanced creep resistance is to use austenitic stainless steels; despite these have a lower yield strength than carbon steels. Alternatively, nickel alloys such as Inconel® or a selection of cobalt alloys are suitable for creep resistance.

3.9

Cyclic loading and fatigue failure Fatigue is the process by which a crack can form and then grow under repeated or fluctuating loading. Failure thus occurs by the steady progression of the crack until a final failure mode such as fracture occurs. For components which are subjected to fluctuating loading in service, the avoidance of fatigue is likely to be the factor which limits the design stresses. This is particularly true for welded components where their fatigue strengths can be much less than those of unwelded components.

3-8

www.twitraining.com

Rev 2 April 2011 Different Types of Loading Copyright  TWI Ltd 2012

Figure 3.14 Fatigue fracture surface.

Figure 3.14 shows an example of a fatigue failure. Beach marks are characteristic in a fatigue surface and are usually visible to the unaided eye. They represent the crack front at particular times and are formed when the loading conditions change eg changes in crack growth rate, or rest periods. The rate of fatigue crack growth is not dependent on material strength but only on load range and hence welded low and high strength materials can have the same fatigue life. The benefit of material strength comes in the crack initiation stage, which is effectively absent in the welded material. In welded joints, fatigue cracks readily initiate and the majority of their fatigue lives are spent propagating the crack. It should be noted that metal fatigue is not associated with any change in material properties, but rather with crack growth.

3.10

Impact behaviour

3.10.1

Effect of loading rate on material properties Tensile properties of materials are also significantly affected by the strain rate. The tensile test is designed as a quasi-static test, ie the load is applied slowly, giving plenty of time for the dislocations to slip which results in plastic deformation. If the load is applied at a fast rate, ie impact loading, this phenomenon is hindered, leading to an increase in tensile strength and a drop in ductility (see Figures 3.15 and 3.16 below).

3-9

www.twitraining.com

Rev 2 April 2011 Different Types of Loading Copyright  TWI Ltd 2012

Figure 3.15 The effect of the rate of loading on the stress-strain curve: 1) dynamic loading; 2) static loading.

Figure 3.16 The effect of the rate of loading on tensile strength and ductility.

3.10.2

Charpy impact test The Charpy impact test is the most common way to characterise impact toughness (see Figure 3.17). It has been used for many years and the large number of Charpy impact test results available make it a useful qualitative test for ensuring adequate toughness at the design and material purchase stage. The impact energy that is absorbed in a small standard specimen is measured at a given test temperature. The tests can either characterise impact behaviour at a single temperature, or by testing over a range of temperatures, be used to plot a Charpy impact transition curve.

3-10

www.twitraining.com

Rev 2 April 2011 Different Types of Loading Copyright  TWI Ltd 2012

Figure 3.17 The Charpy impact test.

3.10.3

Drop weight impact test (Pellini test) The drop-weight test (often referred to as Pellini test) was developed as a simple method to determine the nil-ductility transition temperature (NDTT). The drop-weight test was developed to simplify the determination of the NDTT, it is standardised in ASTM E208. The drop-weight test specimen consists of a rectangular coupon with a brittle weld bead deposited on one face (Figure 3.18). The weld contains a notch, from which a crack is initiated by impact loading the specimen to a fixed amount of deformation under three-point bend. Tests are carried out at a variety of temperatures, the NDTT being defined as the maximum temperature at which the brittle crack spreads completely across one or both of the tension surfaces on either side of the brittle weld bead.

Figure 3.18 Drop weight impact test specimens.

3.11

Low temperature If the temperature drops, the yield strength increases but the ultimate tensile strength remains constant. As a result of this, the yield strength approaches the tensile strength and plastic deformation is hindered. This means that if overload is about to occur, there is less warning in the form of bulging or deformation in a component before final failure. This is illustrated in Figure 3.19.

3-11

www.twitraining.com

Rev 2 April 2011 Different Types of Loading Copyright  TWI Ltd 2012

σ

σ

σ

ε

Temperature

ε

ε

Figure 3.19 The effect of test temperature on the stress-strain curve.

At low temperature, ferritic steels also undergo a ductile to brittle transition and this significant drop in fracture toughness makes them more susceptible to brittle fracture at low temperatures. A toughness transition curve can be plotted by testing Charpy (or fracture mechanics test) specimens at a range of temperatures. At the ductile to brittle transition temperature the results will show much more scatter.

Figure 3.20 Charpy impact test transition curve.

The part of the curve where the material is ductile is called the upper shelf. Where the material is brittle is called the lower shelf and there is a transition region in between. Remember that transition behaviour only occurs for ferritic steels and doesn’t occur for austenitic materials such as stainless steel or aluminium. A designer needs to ensure that the material always operates on the upper shelf in order to avoid brittle fracture.

3-12

www.twitraining.com

Rev 2 April 2011 Different Types of Loading Copyright  TWI Ltd 2012

Figure 3.21 Ductile to brittle transition curve.

3.12

Brittle fracture Brittle fracture is a fast, unstable type of fracture. The crack can propagate with the speed of sound in steel and the results can be catastrophic. Brittle fracture involves little or no plastic deformation and occurs in a fast, unstable manner. The fracture surface may show chevron marks or river lines pointing back to the fracture initiation point. With a brittle impact fracture, the surface is rough but not torn and will usually have a crystalline appearance. Some examples of brittle fracture are shown in Figure 3.22.

A

b

Figure 3.22a Brittle fracture surface showing chevron marks and historical brittle failures of 3.22b) the John Thompson pressure vessel.

For brittle fracture to occur there needs to be a critical combination of three factors; low toughness material, a flaw and a stress (Figure 3.23).

3-13

www.twitraining.com

Rev 2 April 2011 Different Types of Loading Copyright  TWI Ltd 2012

Figure 3.23 The three factors for brittle fracture.

In welded structures, brittle fracture is a particular concern due to the risk of all three occurring. Weld defects or stress concentrations act as crack starters. Poor fracture toughness may be in the parent material due to the wrong design choice, or more likely in HAZ due to too high or too low heat input. This combined with a high level of residual stress in the weld, possibly due to lack of PWHT or an incorrect design, can provide all the necessary conditions for fast fracture. As well as at low temperature for material below the ductile to brittle transition, or at welds, brittle fracture also becomes a significant consideration when fabricating structures from thick material, since the restraint effect of the surrounding material makes brittle fracture more likely (see Section 3.6). 3.12.1 Preventing brittle fracture The general approach to prevention of brittle fracture in welded structures is by constructing to an established codes or standards. These will have clauses to ensure adequate fracture toughness, usually by imposing Charpy requirements on the parent metal, weld and HAZ, but also by giving a minimum service temperature. The codes will give advice on design in order to prevent high stress and will not permit certain features that could act as stress raisers. There is usually a requirement to stress relieve thick sections and for a proof test after fabrication. The welding will need to be inspected in order to avoid defects that exceed the acceptance level. Examples of these fabrication codes are BS PD 5500 (pressure vessels), BS 5400 (steel bridges), BS 5950 (steel structures), ABS rules (fixed offshore structures), BS 4515 (pipelines). An example of design to avoid brittle fracture for a cryogenic pressure vessel will depend critically on the material selection. The materials require good impact toughness values at extremely low temperatures, whilst also having good weldability and ductility. Low thermal conductivity in order to insulate from atmospheric heat will be a benefit, but the material must be as cheap as possible to make the design economical. The materials used would be either aluminium, austenitic stainless steel, or 9% nickel steel.

3-14

www.twitraining.com

Rev 2 April 2011 Different Types of Loading Copyright  TWI Ltd 2012

3.13

The significance of flaws The integrity of plant equipment and structures is vital to ensure a continued, safe and economic operation. In the design procedure, the structure is assumed to be sound, ie free from defects. However, flaws such as cracks, welding defects and corrosion damage can occur during manufacture or service life. These flaws alter the way the load is distributed on the structure by creating a stress concentration. Within this heightened stress field the load on the material can be much greater than the design load, leading to fracture or yielding.

Figure 3.24 Two different weld flaws: Hydrogen crack and lack of sidewall fusion.

At any point in the life cycle of a structure (eg design, fabrication, operation), it may be necessary to investigate whether it is fit for its intended service conditions, or those which it will encounter in the future. For safety critical items like pipelines, pressure vessels, on- and offshore platforms, rigs, wind turbines, storage tanks, ships, bridges, aircrafts, buildings, etc, the failure of a single component due to the presence of a flaw can threaten human life, as well as presenting severe economic and environmental consequences. Other flaws may be harmless, as they will not lead to failure during the lifetime of the component. Replacement or repair of such insignificant flaws is economically wasteful. Insignificant flaws are not necessarily the smallest ones, or those which are difficult to detect. The two welds in Figure 3.25 would not be passed as acceptable when the slag inclusion or porosity had been detected from NDT, but neither of these large defects has contributed to the failure, which was due to fatigue cracking from the weld toe in both cases. The most obvious flaws were insignificant with respect to fitness-for-service in these cases.

3-15

www.twitraining.com

Rev 2 April 2011 Different Types of Loading Copyright  TWI Ltd 2012

a

b

Figure 3.25 Large but nonetheless insignificant flaws; a slag inclusion a and porosity b which has not contributed to the fatigue cracking initiating from the weld toes.

So the key questions to be asked when a flaw is detected, such as that in Figure 3.26, are:      

Does the weld contain an unacceptable flaw? Is the flaw harmful? Could repair make matters worse? What if the material does not meet specification? What if cracks occur in service? Can structure be used beyond design life?

Figure 3.26 Crack detected underneath a repair weld, but is the structure still fitfor-service?

Fitness-for-service assessment, or engineering critical assessment, is used to quantitatively answer these questions.

3.14

Fitness-for-service or engineering critical assessment For safety critical items like pipelines, pressure vessels, on- and offshore platforms, rigs, wind turbines, storage tanks, ships, bridges, aircrafts, buildings, etc. the failure of a single component due to the presence of a flaw can threaten human life, as well as presenting severe economic and environmental consequences. Other flaws may be harmless, as they will not lead to failure during the lifetime of the component. Replacement or repair of such insignificant flaws is economically wasteful and repair welding can actually introduce more harmful flaws.

3-16

www.twitraining.com

Rev 2 April 2011 Different Types of Loading Copyright  TWI Ltd 2012

A fitness-for-service (FFS) procedure using fracture mechanics principles allows flaws to be evaluated consistently and objectively and thus enables to make informed and confident decisions on the best measures to be taken. The object of fracture mechanics is to provide quantitative answers to specific problems concerning cracks in structures and to enable determination of whether the structure can be considered fit-for-service, or whether repair, downrating or replacement is needed. This procedure is described as an engineering critical assessment, or ECA. The background to the development of FFS procedures was in the 1950’s and 60’s when research on the integrity of welded structures due to fatigue and fracture was carried out at the British Welding Research Association (which later became TWI). Flaws and other characteristics of welds were found to influence the fracture resistance, but some flaws, especially those found by radiography, were insignificant. A method was needed to evaluate the significance of flaws. A range of techniques is available in different Standards such as BS 7910, FITNET, API 579-1/ASME FFS-1, SINTAP, R6. These techniques are often very time consuming and require a high degree of expertise. However, specialist tools such as the CrackwiseTM software provide a great help when carrying out the ECA of a component. Many offer several levels of assessment starting with the most simple, but conservative and moving to those with more accuracy and analytical complexity but more cost. The basic principle of the assessment can be thought of as a triangle, assessing the critical combination of a flaw of certain geometry, under stress, for given material mechanical properties. Failure occurs when critical values for the three requirements for fracture are all met. It is possible to determine the critical value for one parameter if all the other inputs are known (eg find the critical flaw size for a given geometry, stress and toughness).

Figure 3.27 Basic principle of ECA in terms of input requirements and analytical output.

3-17

www.twitraining.com

Rev 2 April 2011 Different Types of Loading Copyright  TWI Ltd 2012

Fracture mechanics principles are described here with a focus on BS 7910 procedures (Guide to methods for assessing the acceptability of flaws in metallic structures). They can be used throughout the lifetime of a component, including:     

Design Determine material properties requirements or stress levels. Fabrication Determine flaw acceptance criteria and inspection strategy or to justify the avoidance of post-weld heat treatment or proof test. Operation Avoid unnecessary repair if flaws are detected and to define appropriate maintenance and inspection strategies. Investigate the causes of failure and therefore to avoid future similar cases. Support cases for life extension and change of service.

To carry out an ECA it is necessary to know the full details of each corner of the assessment triangle, in terms of the material properties, geometry and stresses, as detailed in the following table. Material properties

Geometry

Fracture toughness at minimum temperature

Flaw type (surface or embedded)

Tensile properties, yield, UTS

Flaw size and position

Stresses

Primary stresses Residual and thermal loading

Fatigue crack growth law and threshold (if assessing fatigue)

3.14.1

Fluctuating stress history Joint configuration (butt or fillet weld)

Stress concentrations (due to weld toes etc)

General procedure The general procedure of an ECA is to determine whether a given flaw will fail either by plastic collapse or by brittle fracture. When assessing the design for strength to avoid plastic collapse there is a balance between the driving force for failure, which is the applied stress and the material resistance which is its strength. The critical condition for safety comes when the driving force equals or exceeds the material resistance. A similar approach is also used when assessing the design against fracture. This time the driving force is the stress intensity produced by a certain crack under an applied stress and the material resistance is its fracture toughness.

3-18

www.twitraining.com

Rev 2 April 2011 Different Types of Loading Copyright  TWI Ltd 2012

a

b

Figure 3.28 Design to prevent a plastic collapse and b brittle fracture, by balancing the ‘driving force’ against the ‘material resistance’.

The method for carrying out the ECA is to first identify the type of flaw (whether it is planar, non-planar, or a shape imperfection), then gather all the essential data (from inspection records, mechanical test certificates, published data, the operating conditions). The procedure then assesses the significance of flaw. A second stage may be to iterate the size of the flaw to find the limiting flaw size for failure avoidance. If fatigue is likely then any sub-critical crack growth must be considered. Most procedures are validated to be conservative, but if an additional safety margin is required this may be added to the applied stress or fatigue cycles, or removed from the given tolerable flaw size. The benefits of using an ECA are on one hand to ensure safer structures, due to the need for fewer weld repairs (weld repair can introduce more severe defects) and the focus on design and materials selection. On the other hand, this is driven by the economic benefits of reduced amounts of repair welding, less rigorous, or better focused NDT demands and being able to justify repair deferral, possibly indefinitely, or at least until it is convenient to do so.

3.15

Fracture mechanics Understanding how a crack behaves in material is called Fracture Mechanics and this is how the ECA can determine whether a flaw is tolerable or not.

3-19

www.twitraining.com

Rev 2 April 2011 Different Types of Loading Copyright  TWI Ltd 2012

Figure 3.29 Fracture mechanics is about calculating the crack driving force; the ECA compares this to the material resistance.

In fracture mechanics, the highly stress concentrating effect of the crack tip is accounted for in a parameter called the stress intensity factor, K. The stress intensity factor is a single parameter which characterises the stress field near the crack tip and the fracture event. It can be calculated in its simplest terms by the following expression. ___ K = Y √ a

[3-1]

Y is a geometrical factor, depending on the type of flaw, its location in a structure and the crack opening mode. The units of K are N/mm3/2 or MPa√m (in America ksi√in in used). These units can be converted using the following conversion factors: 1 N/mm 3/2 = 0.0316 MPa√m 1 N/mm 3/2 = 0.029 ksi√in

Figure 3.30 Stress at the tip of a crack.

3-20

www.twitraining.com

Rev 2 April 2011 Different Types of Loading Copyright  TWI Ltd 2012

Solutions for Y can be calculated for simple components such as pipelines or sheet panels but they can also be found in a tabulated form in the Standards, eg BS 7910. For more complex shapes, finite element modelling can also be used. Figure 30 shows different values of Y for a component containing a centre-crack, an edge crack and a penny-shaped embedded crack. The Y factor is larger when the plate is cut in half and the crack breaks the surface; the stress intensity is higher. This is because it is easier to propagate a crack that is open to the surface. When the crack is surrounded by material on all sides (the buried penny crack) then the Y factor is lower and it has a lower stress intensity. It requires more driving force to propagate this crack.

Small throughthickness crack in large plate

Small throughthickness edge crack in large plate

Small penny-shaped embedded crack in large body

Y=1

Y = 1.12

Y = 0.637

Figure 3.31 Geometrical factors for centre crack, edge crack and penny-shaped crack.

Y depends not only on the geometry under consideration, but also the loading mode. There are three possible loading modes, Modes I, II and III, which relate to crack behaviour (crack opening, in-plane shear and out-ofplane shear). Mode I tends to be the most severe loading type and gives the highest stress intensity. Most fracture mechanics analyses are based on K solutions for Mode I loading, which is identified by a subscript I (ie KI).

Mode I (crack opening)

Mode II ( in-plane shear)

Mode III (out of plane shear)

Figure 3.32 Three possible crack loading modes.

3-21

www.twitraining.com

Rev 2 April 2011 Different Types of Loading Copyright  TWI Ltd 2012

3.16

LEFM and EPFM Linear Elastic Fracture Mechanics (LEFM) is the simplest approach to fracture mechanics and is a good model of the behaviour of cracks for brittle materials. However LEFM does not properly account for the plasticity effects at the crack tip where the stresses are limited by the material’s yield strength. Also, when testing ductile materials, it was realised that the amount of energy to add to initiate crack propagation was significantly higher than predicted by LEFM. This was explained by the role of the plastic zone in which the crack propagates. For ductile materials where there is a large plastic zone at the crack tip which affects the fracture mechanics of the crack, instead of LEFM, Elastic-Plastic fracture mechanics (EPFM) is used. EPFM considers the elastic behaviour of the material and the local plastic deformation at the crack tip. This plastic zone absorbs energy from the crack driving force as deformation occurs and makes the material tougher. The equations governing EPFM are more complex than those for LEFM (such as Equation 3-1) and beyond the scope of this course.

Figure 3.33 The plastic zone at the crack tip limiting the crack tip stress in EPFM, compared the purely elastic assumption in LEFM.

3.16.1 The plastic zone In instances of brittle failure the plastic zone size is small and the plastic contribution to failure is negligible. But depending on the properties of the material, the magnitude of the applied stress, the size of the crack and the thickness of the component, the plastic zone can grow to have a substantial effect on the failure mode. If the plastic zone can extend across the entire ligament of the un-cracked body in a test specimen, this can give a completely plastic net section of material. The failure mode in this case is by plastic collapse. Large amounts of energy are absorbed as the material work hardens and the plastic hinge is formed. It is common for failure of both test specimens and structures to occur via a combination of brittle and ductile failure modes, the latter being recognised by the formation of shear lips on the fracture surface.

3-22

www.twitraining.com

Rev 2 April 2011 Different Types of Loading Copyright  TWI Ltd 2012

Figure 3.34 A large plastic zone at the crack tip of a fracture mechanics test specimen resulting in a plastic hinge across the remaining ligament.

3.17

Fracture mechanics testing Fracture toughness in terms of KIC (in N/mm3/2 or MPa√m) is used to characterise brittle materials. It is also possible to express the material’s fracture toughness in terms of the driving force required to drive the crack, as measured from the energy absorbed within the plastic zone. This is called the J integral or simply J and is measured in kJ/m2 and applies for ductile materials. Finally the fracture toughness can be measured from the strain or opening displacement to propagate the crack, called Crack Tip Opening Displacement, CTOD or , measured in mm and used for both brittle and ductile materials.

Figure 3.35 Fracture mechanics specimen illustrating how J and CTOD are measured.

Fracture mechanics testing is carried out to an established standard. In the UK BS 7448 is used for testing metallic materials for measuring K, J and CTOD. It comes is several parts. Part 1 covers testing of parent material; while Part 2 covers weldment testing (this part is now superseded by BS EN ISO 15653). Part 3 gives details of fracture mechanics testing under high strain rate and Part 4 covers testing specimens to allow for the ductile tearing. There are also American standards such as ASTM 1820 which covers KIc, CTOD and J testing.

3-23

www.twitraining.com

Rev 2 April 2011 Different Types of Loading Copyright  TWI Ltd 2012

Fracture mechanics testing usually uses a specimen under bending, with a machined notch on one edge. This is called the Single Edge Notched Bend specimen, or SENB. An alternative type of specimen is the compact tension or CT specimen. The CT specimen has the advantage that it requires less material, but is more expensive to machine and more complex to test compared to the SENB specimen. Although the CT specimen is loaded in tension, the crack tip conditions are predominantly bending (high constraint). High constraint specimens tend to measure lower fracture toughness than an equivalent thickness higher constraint specimen.

Figure 3.36 Fracture mechanics test specimens; a single edge notched bend (SENB specimen and a compact tension (CT) specimen.

Full thickness specimens are used so that the thickness effect on fracture toughness can be accounted for in the test. Specimens have a sharp cracklike notch. The specimens are loaded under representative service conditions. Usually three identical specimens are tested at a given temperature. The principle of fracture mechanics testing is to load the specimen and measure the displacement at the crack mouth using a clip gauge (or sometimes a pair of clip gauges). SENB specimens can be used to determine a value of K, J or CTOD depending on how the results of the test are interpreted.

Figure 3.37 Typical test arrangement for a fracture mechanics test.

3-24

www.twitraining.com

Rev 2 April 2011 Different Types of Loading Copyright  TWI Ltd 2012

The machined notch is sharpened to crack sharpness using fatigue precracking, by applying cyclic loading to the specimen. Specialised high frequency resonance hydraulic machines are often used for this process. A fatigue crack is the most severe type of crack in terms of notch tip sharpness, so using a fatigue pre-cracking means that the fracture mechanics at the crack tip of the specimen can be accurately related to the fracture mechanics of cracks in the structure. The maximum fatigue load is limited so that the failure plastic zone of the specimens is much larger than the fatigue pre-crack plastic zone.

Figure 3.38 Fatigue pre-crack.

3.18

Calculating fracture toughness During the fracture mechanics test a load versus crack mouth displacement curve is produced. Using this data and measurements taken from the specimen itself, including measurements of the fatigue pre-crack after the failed specimen has been broken open; it is possible to calculate the fracture toughness. Depending on whether it is K, J or CTOD that is required, different formulae are used, but the principle of the test is the same for each.

Figure 3.39 Dimensions from an SENB specimen and a load-displacement trace used to determine fracture toughness.

Although these formulae are beyond the scope of this course, the three methods for calculating the fracture toughness are given below. For CTOD and J the overall toughness value is comprised of both an elastic and plastic component.

3-25

www.twitraining.com

Rev 2 April 2011 Different Types of Loading Copyright  TWI Ltd 2012

Stress intensity factor, K:

a

Y

K

 

 

Crack tip opening displacement, CTOD or : l p











z a



a



W rp / Vp a

Y

 ︶ 

W rp

2

  ︵ 

E

2 /

1 K



l 2 e

  





J-integral:

 ︵

︶ o



b B / p U



2



E /



l



Jp 1 l 2 Je K

J



Figure 3.40 illustrates the various shapes of load-displacement curve that may be produced from a fracture mechanics test at different temperatures (and therefore different positions along the toughness transition curve). A brittle result is characterised by toughness in terms of K and is explained using linear elastic fracture mechanics. A more ductile result will instead be characterised by toughness in terms of CTOD or J and is described using elastic plastic fracture mechanics. Where the test piece has fractured in a brittle manner with little or no plastic deformation the fracture result is described with a subscript c, eg KIc. Where the specimen starts to tear in a ductile manner and at least 0.2mm of tearing is measured, but the specimen doesn’t reach maximum load before failure, the result is given a subscript u, eg u. When the load-displacement curve shows completely plastic behaviour and exceeds maximum load, the result is given a subscript m, eg Jm.

Figure 3.40 Typical load-displacement behaviours obtained by fracture mechanics tests.

3-26

www.twitraining.com

Rev 2 April 2011 Different Types of Loading Copyright  TWI Ltd 2012

The fracture mechanics test standards include many checks to ensure that results are valid. These include restrictions on the fatigue crack size, position and shape, together with limitations on the maximum allowable fatigue load. The final fatigue pre-crack shape and length is not determined until after the test and the specimen is broken open. If the crack front it not straight enough then the test may not be fully qualified to the standard. In fracture mechanics testing it is not unusual to get results that are not fully qualified because parameters like the fatigue pre-crack cannot be checked until after test and are affected by weld residual stresses which could be present. If some validity criteria are not met, the results may still be useable, but expert advice is needed to justify this.

3.19

Testing welds Welds are critical in terms of fracture. They are under high residual stress plus can contain brittle microstructures in addition to weld flaws, all of which combined can lead to fracture. Welds also have complex geometries, complex microstructures and complex residual stress distributions which can add extra challenges when fracture toughness testing. The orientation of the specimen becomes important. The specimen thickness is the same as the section thickness (B), but if the test is intended to model an actual crack, then the notch is likely to be a surface notch in a BxB specimen, located into a specific microstructure (SM). SM testing requires post test metallography to confirm whether the notch sampled the required region (microstructure). If the test is to measure a lower-bound toughness for procedure qualification or for setting flaw acceptance criteria, then a through-thickness notch and a Bx2B specimen is used and the notch is called weld positional (WP) and is positioned relative to the weld centreline or mid-thickness HAZ.

Figure 3.41 Choosing the specimen and notch orientation when testing welds.

3-27

www.twitraining.com

Rev 2 April 2011 Different Types of Loading Copyright  TWI Ltd 2012

The location of the notch in the weld HAZ or parent metal is important as an incorrectly positioned fatigue crack will not sample the required area, making the test invalid. To be certain that the crack tip is in the correct region, polishing and etching followed by a metallurgical examination are often carried out prior to machining the notch and fatigue cracking. This enables the notch to be positioned very accurately. This examination may be carried out after testing as further confirmation of the validity of the test results. The testing of HAZs presents particular difficulties, owing to the narrow width of the HAZ, the irregular shape of the HAZ and local variations in microstructure. Positioning the fatigue crack tip to sample a particular target microstructure can be difficult, owing to the somewhat uncertain nature of the fatigue pre-cracking process. For a through-thickness notch it is easiest to test the HAZ when a vertical fusion line can be produced, such as from a half K or K-prep weld. A surface notched specimen will need a notch of a precise depth to target the microstructure at a particular location in the weld.

Figure 3.42 Surface notching and through-thickness notching into the HAZ to target a specific microstructure.

3.20

BS 7910 ECA procedure BS 7910:2005 is entitled Guide to methods for assessing the acceptability of flaws in metallic structures. It gives detailed guidance on assessing flaws under fracture or fatigue (and both). It also has advice about assessing flaws under creep and other modes of failure (Figure 3.43).

3-28

www.twitraining.com

Rev 2 April 2011 Different Types of Loading Copyright  TWI Ltd 2012

Figure 3.43 Failure modes for which BS7910 gives guidance for assessing flaws.

Other Annexes in the standard provide advice on safety factors, Charpytoughness correlations, the assessment of pop-ins, misalignment, stress intensity factor solutions and reference stress solutions. 3.20.1

Input data The procedure models the actual flaw size and dimensions into an idealised flaw geometry for the assessment. The flaw is either a semi-elliptical surface flaw, an elliptical embedded flaw, or a through-thickness flaw (Figure 3.44). The flaw dimensions are chosen so that the flaw in the assessment fully encompasses the real flaw.

a

b

Figure 3.44 a Flaw types and dimensions in BS 7910 to assess a real flaw b.

Residual stresses also need to be taken into account in the BS7910 procedure. Residual stresses can impose significant driving forces on welds flaws (Figure 3.45). For as-welded components, or if no information on residual stresses is known then they are assumed to reach yield magnitude. Sometimes it is possible to use a known value, or a given residual stress

3-29

www.twitraining.com

Rev 2 April 2011 Different Types of Loading Copyright  TWI Ltd 2012

distribution. The residual stress in the assessment is relaxed if the structure undergoes mechanical loading (such as a proof test, or plastic straining in service), or if it has been post weld heat treated.

Figure 3.45 Residual stresses as a result of welding.

The fracture toughness input value is the lowest result taken from three similar specimens. Additional fracture mechanics testing is recommended if the results from the set of three are too scattered; if the lowest value of CTOD is less than half the average, or if the highest value is more than twice the average. For values of K then more testing is recommended if the minimum is below 0.7 times the average, or if the maximum is above 1.4 times the average. 3.20.2

Failure assessment diagram BS 7910 calculates both the proximity to failure by plastic collapse and the proximity to brittle fracture and plots each of these on the axes of a Failure Assessment Diagram (or FAD). The proximity to brittle fracture is calculated by dividing the stress intensity factor at the tip of the flaw (KI) by the fracture toughness of the material (KIc). This is plotted on the Y-axis of the FAD and is termed Kr, the fracture ratio. The proximity to plastic collapse is calculated by dividing the net section stress on the cracked ligament (also called the reference stress) by the material yield stress or flow stress (average of yield and UTS). This is plotted on the X-axis of the FAD and is termed Lr, the stress ratio.

3-30

www.twitraining.com

Rev 2 April 2011 Different Types of Loading Copyright  TWI Ltd 2012

The failure assessment line, plotted from a point on the X-axis to a point on the Y-axis shows the limit of safety for failure from either mode. If the assessment point lies inside the curve, the flaw is acceptable and if it is outside the curve it is unacceptable (Figure 3.46). A critical calculation can be made by making iterations until the assessment point lies on the failure assessment line.

Figure 3.46 Failure Assessment Diagram (FAD), showing an acceptable and unacceptable assessment point on a Level 2 FAD.

Figure 3.47 BS 7910 Level 1 and Level 2 FADs.

BS 7910 provides different levels of assessment which increase in their analytical effort and accuracy. Level 1 uses fairly simple formulae for making the calculations and the Level 1 FAD has the most conservatism (ie the smallest area inside the FAD). Level 2 requires more analysis and is usually carried out using specific software and the Level 2 FAD has a reduced level of conservatism. It is possible to apply Level 3 methods which require using

3-31

www.twitraining.com

Rev 2 April 2011 Different Types of Loading Copyright  TWI Ltd 2012

FEA modelling, but it is possible to reduce the conservatism even more, however the more analysis that is needed, the more costly the ECA. All of the assessment levels in BS 7910 have been validated against a large number of test cases, including wide plate tests and full-scale tests, the assessment points for some of which are shown in Figure 3.48.

Figure 3.48 Validation of the BS 7910 procedure with full-scale tests and wide plate tests.

The FAD is a useful visual tool for assessing a single flaw size and determining whether is can be considered safe or not. However, when a structure is experiencing fatigue, the crack grows from a sub-critical size until it is sufficiently large to cause failure. In the BS 7910 procedures the rate the fatigue crack grows can be estimated and an initial tolerable flaw size can be derived if the lifetime is known and hence the final (critical) crack size calculated. To carry out a fatigue ECA it is necessary to measure the fatigue crack growth rate in the parent material, weld metal or HAZ. The crack growth rates are affected by the environment eg whether in seawater with cathodic protection or in air. The fatigue threshold is then determined. In addition it is also necessary to know the flaw type and flaw geometry and the applied stresses in terms of a stress range and number of cycles at that range. The basis of the fatigue crack growth calculations is the relationship between K, the change in stress intensity due to the cyclic stress and the rate of crack growth, da/dN, for the material under consideration (Figure 3.49).

3-32

www.twitraining.com

Rev 2 April 2011 Different Types of Loading Copyright  TWI Ltd 2012

Crack growth relationship

Flaw type & geometry

Applied stresses

Figure 3.49 Additional input information required for a fatigue assessment.

Figure 3.50 Recommended fatigue crack growth law.

3.21

ECA case study Lack of fusion was detected in a propane storage sphere (Figure 50) in Saudi Arabia during outage. Before the vessel could go back into service, its defect tolerance had to be assessed to show that the sphere would not fail by brittle fracture or plastic collapse. A fracture mechanics assessment showed that the storage sphere had adequate defect tolerance despite the lack of fusion flaw and the vessel could return to service with a valid safety case.

Figure 3.51 Storage sphere.

3-33

www.twitraining.com

Rev 2 April 2011 Different Types of Loading Copyright  TWI Ltd 2012

3.22

References and further reading Anderson T L (2005): ‘Fracture mechanics: fundamentals and applications’, CRC Press, USA. API 579-1/ASME FFS-1 (2007): ‘Fitness-for-service’, American Petroleum Institute, Washington DC, USA. ASTM E1820: ‘Standard Test Method for Measurement of Fracture Toughness’, American Society for Testing and Materials. ASTM E139: ‘Conducting Creep, Creep Rupture and Stress Rupture Tests of Metallic Materials’ BS 3500: ‘Methods for Creep and Rupture testing of Metals’ BS EN 10291: ‘Metallic Materials - Uniaxial Creep Testing in Tension’ BS EN ISO 899: ‘Plastics - Determination of Creep Behaviour’ BS EN 761: ‘Creep Factor Determination of Glass Reinforced Thermosetting Plastics Dry Conditions’ BS EN 1225: ‘Creep Factor Determination Thermosetting Plastics Wet Conditions’

of

Glass

Reinforced

Broek D (1987): ‘Elementary engineering fracture mechanics’, Fourth revised edition, Martinus Nijhoff Publishers, Dordrecht. BS 6729: ‘Determination of the Dynamic Fracture Toughness of Metallic Materials’, British Standards Institution, London. BS 7448 Parts 1-4: ‘Fracture Mechanics Toughness Tests’, British Standards Institution, London. BS 7910 (1999): ‘Guide on methods for assessing the acceptability of flaws in metallic structures’, British Standards Institution, London. CrackwiseTM (2010): ‘Automation of fracture and fatigue assessment procedures (BS 7910) for engineering critical assessment’. More information at: http://www.twisoftware.com/crackwise FITNET Fitness-For-Service Network (2006), Project G1RT-CT-200105071, Procedure. R6 (2007): ‘Assessment of the integrity of structures containing defects, British Energy Generation, Revision 4. SINTAP (2000): ‘Structural integrity assessment procedures for European industry, Final Procedure’, European Union Brite-Euram Programme. Project No. BE95-1426, Contract No. BRPR-CT95-0024, Rotherham. Wells A A (1961): ’Unstable crack propagation in metals, cleavage and fast fracture’, The Crack Propagation Symposium, pp.210-230, Cranfield. Wells A A (1963): ‘Application of fracture mechanics at and beyond general yield’, British Welding Research Association Report M13/63.

3-34

www.twitraining.com

Rev 2 April 2011 Different Types of Loading Copyright  TWI Ltd 2012

3.23

Summary At the end of this module you should be able to describe the risks for brittle fracture, ductile failure, creep and fatigue etc based on the different types of loading experienced in service and select suitable materials to be resistant to these failure modes. You should know the difference between Charpy testing and fracture mechanics testing of welds and describe them. You should understand the basics of fracture mechanics in terms of the behaviour of cracks in materials, LEFM and EPFM. You should know how to carry out an engineering critical assessment (ECA), what input data is required and how the assessment point is plotted on an FAD for different assessment levels. You should be able to explain the benefits of fitness-forservice assessment

3-35

www.twitraining.com

Rev 2 April 2011 Different Types of Loading Copyright  TWI Ltd 2012

Revision Questions 1 What do LEFM and EPFM stand for and what are the differences between them?

2 Sketch a failure assessment diagram and describe how you would determine where to plot the failure assessment point.

3 What are the advantages and disadvantages of using very thick steel in a structure?

4 Sketch a ductile to brittle transition curve for ferritic steel and austenitic stainless steel. How would you determine the ductile to brittle transition temperature for the ferritic steel?

3-37

www.twitraining.com

Section 4 Advanced Fatigue – Part 1

Rev 2 April 2011 Advanced Fatigue – Part 1 Copyright  TWI Ltd 2012

4

Advanced Fatigue - Part 1

4.1

Introduction In order to cover the subject of fatigue in sufficient depth for Engineer level of this diploma, it has been divided into two parts. In this, the first part, we’ll start with understanding stress concentration locations, before moving on to fatigue joint classifications. This part will also describe post weld fatigue life improvement techniques and review case studies of fatigue failures.

4.2

Stress concentrations Fatigue cracks initiate from sharp features, changes in section, or most likely in welded structures, from weld toes. These are all examples of local stress concentrations. The fatigue performance of any structure depends entirely on the effective stress concentrations present. Weld toes and weld roots are the most critical areas in respect to stress concentration in welded structures (Figure 4.1).

Figure 4.1 Stress concentration locations in a butt weld.

Production arc welding processes lead to the formation of non-metallic intrusions at the weld toe, which is typically 0.1-0.4mm in depth (Figure 4.2). Fatigue life is governed by the growth of this pre-existing flaw with little or no initiation stage. Also, the factors which affect crack initiation can be quite different to those that affect crack growth.

Figure 4.2 Fatigue crack initiating from a weld toe intrusion.

4-1

www.twitraining.com

Rev 2 April 2011 Advanced Fatigue – Part 1 Copyright  TWI Ltd 2012

Fatigue performance Behaviour of a structure under cyclic loads is determined by the severity of stress concentration. Different welded joint designs are grouped together according to their stress concentration effect. Each group is given a fatigue joint classification, based on its fatigue resistance. Fatigue resistance data in all design rules are in the form of S-N curves. Fatigue design rules for welded structures have developed extensively over the last 30 years, following a major review and analysis of the published data by The Welding Institute (TWI). A key feature of the design curves is that they refer to particular welded joints, directions of loading and failure sites. Fatigue resistance is quantified in terms of constant amplitude S-N curves SmN = A (where m and A are constants). Curves for different design details are based on statistical analysis of test data and current fatigue design curves are given in BS 7608. It is well established that the fatigue performance of welded joints is vastly inferior to the performance of parent material, Figure 4.3. Furthermore, the fatigue strength of a welded steel joint is independent of parent material strength, Figure 4.4. 400

Stress range, range N/mm2

300 200 100 50

Steel 350 N/mm2 yield

10 105

106

107

108

Cycles

Figure 4.3 Fatigue performance of welded joints compared with unwelded material. Stress range for life of 106 cycles, N/mm2

4.3

500 400 300 200

100 400 500 600 700 800 900

Ultimate tensile strength of steel, N/mm2

Figure 4.4 Fatigue strength of welded joints is independent of material strength.

4-2

www.twitraining.com

Rev 2 April 2011 Advanced Fatigue – Part 1 Copyright  TWI Ltd 2012

This means that for a structure under fatigue loading, using a higher grade of steel will not extend its lifetime at all. In fact using thinner sections under higher stress will reduce the fatigue performance for higher strength steels. Evaluation of a great deal of experimental work has led to the development of design S-N curves that are relevant to individual joint geometries, as shown in Figure 4.5. Stress range, N/mm 2

300

static design limit

200

100

30

10 5

10 6

Endurance, cycles

10 7

Figure 5 Design curves for different joint geometries.

4.4

Joint classification In the British Standards for steel structures, welded joints are organised into different classes according to the loading direction and geometry of the joint. Typical fatigue classes vary from A-G although Classes S, T and W are also found. Although Class A is mentioned in the Standards, it is actually not used for welds but represents the fatigue curve of plain steel (in the Eurocodes, weld classes are defined according to their fatigue strength at 2 million cycles, See Eurocode 3 for steel and Eurocode 9 for aluminium). For a given applied stress range the expected fatigue life decreases the further along the alphabet the joint class, so Class A has the highest fatigue strength and Class G the lowest (as shown in Figure 4.6).

4-3

www.twitraining.com

Rev 2 April 2011 Advanced Fatigue – Part 1 Copyright  TWI Ltd 2012

Figure 4.6 Weld joint fatigue classes.

For a design assessment against fatigue, the stress range needs to be predicted at the likely failure location, ignoring residual stresses and stress concentration factors. A few of the BS 7608 weld classes are summarised below. Class A Plain steel with all surfaces machined and polished with a uniform or smoothly varying cross section (no guidance given in the standard). Class B As-rolled steel with no flame cut edges. Full penetration butt welds free of defects. The weld must be parallel to the direction of stress with smooth dressed faces.

Figure 4.7 Class B welded joint: Longitudinal, butt weld with smooth surface (ground flush cap).

4-4

www.twitraining.com

Rev 2 April 2011 Advanced Fatigue – Part 1 Copyright  TWI Ltd 2012

Class C Full or partial penetration butt or fillet welds which are parallel to the direction of stress. These must have been made by mechanised welding and have no start/stop locations. This class also includes butt welds made from both sides transverse to the direction of stress, which are proved defect free by inspection and machined flush.

a

b

c Figure 4.8 Class C: a and c Longitudinal fillet welds and b Transverse, ground flush butt welds.

Class D Full or partial penetration butt or fillet welds which are parallel to the direction of stress with a stop/start point. Full penetration butt welds made from both sides transverse to direction of stress with a smooth weld-plate transition.

A

b

Figure 4.9 Class D: (a) Longitudinal butt weld and (b) Transverse butt weld.

4-5

www.twitraining.com

Rev 2 April 2011 Advanced Fatigue – Part 1 Copyright  TWI Ltd 2012

Class E Full penetration butt welds transverse to the direction of stress with an abrupt weld-plate transition. Such welds are usually found in overhead or vertical butt welds. This class also includes intermittent welds parallel to the stress direction.

Figure 4.10 Class E joint: Transverse butt weld with large toe angle.

Class F Butt welds on backing strip, transverse to the stress direction. Member carrying fillet or butt welded attachments clear of edge. Cruciform or T-joints made with full penetration butt welds.

a

b

c Figure 4.11 Class F welded joints: a butt weld with backing strip; b longitudinal and transverse fillet welded attachments; c Full penetration transverse fillet welds.

4-6

www.twitraining.com

Rev 2 April 2011 Advanced Fatigue – Part 1 Copyright  TWI Ltd 2012

Class F2 Cruciform or T-joints made with partial penetration butt or fillet welds. Butt welds between two rolled or built-up sections.

a

b

Figure 4.12 Class F2: a Cruciform joint with fillet welds and b butt weld between two built-up sections.

Class G Members with attachments welded to a free edge or closer than 10mm to the edges.

Figure 4.13 Class G: Weld attachment to the edge.

Class W Weld metal in load carrying fillet or partial penetration butt welds, regardless of direction of stressing. Stress used in calculating the fatigue life is the nominal stress on the weld throat.

A

b

Figure 4.14 Class W: a Weld root cracking in a fillet weld, b weld root cracking in a partial penetration butt weld.

4-7

www.twitraining.com

Rev 2 April 2011 Advanced Fatigue – Part 1 Copyright  TWI Ltd 2012

Here is an exercise to label the fatigue classifications of the welds shown in the feature shown in Figure 9 (answers in the revision session).

Figure 4.15 What are the fatigue classifications of the welds in this detail?

4.5

Fatigue failures In welded structures, fatigue failure nearly always initiates from the weld toes, so it is in these locations that should be inspected most frequently for fatigue cracks (Figure 4.16)

Figure 4.16 Fatigue cracking at weld toes.

When examined macroscopically, fatigue fracture surfaces are relatively flat and featureless, with beach marks occasionally identifying the crack front at stages in the crack extension, Figure 4.17. The beach marks are centred on the initiation point and provide a record of the crack front location at particular time. They are associated with differential corrosion, sudden change in crack growth rate, rest periods etc. The final failure mode will be either brittle or ductile as the remaining ligament (reduced in size by the fatigue crack) is no longer able to withstand the applied load.

4-8

www.twitraining.com

Rev 2 April 2011 Advanced Fatigue – Part 1 Copyright  TWI Ltd 2012

Figure 4.17 Low magnification view of a fatigue fracture surface showing beachmarks.

At high magnification however, the fracture surface may exhibit striations, Figure 4.18. These marks are interpreted as representing the position of the crack front after individual cycles. Therefore, each fatigue cycle produces a small but measurable amount of crack extension.

Figure 4.18 High magnification view of a fatigue fracture surface.

Fatigue failure in components depends on the loading direction and the geometry. The fatigue performance of large structures can be tested using representative smaller-scale specimens under controlled fatigue testing conditions (Figure 4.19).

Figure 4.19 Small scale fatigue test specimen as representative of full scale structures.

4-9

www.twitraining.com

Rev 2 April 2011 Advanced Fatigue – Part 1 Copyright  TWI Ltd 2012

4.6

Improving the fatigue strength of welded joints There are a number of general ways to improve fatigue strength.  

By design. By a postweld treatment.

At the design stage, improvements to the fatigue life can be gained by avoiding geometric stress raisers and achieving a smooth transition between the weld and plate and by using smooth shapes and transitions. Minimising the stress range and using an appropriate choice of joint geometry (and hence joint class) will help achieve the required fatigue life. Grind flush excess weld metal, making sure that grinding marks should be parallel to stress direction. Local machining to remove weld toes and edges can be specified at the design stage. Considerable work has been carried out on postweld fatigue improvement techniques. Because fatigue strength is usually limited by joints with low fatigue classifications, eg Class F (transverse and longitudinal stiffeners attached to the surface of a stress plate), improving the performance of these joints can have a substantial benefit on the fatigue life of the structure. Broadly speaking, there are two main methods for postweld fatigue improvement:   

4.7

Reduce stress concentration at the weld toe using techniques such as weld toe grinding, or TIG or plasma re-melting. Reduce residual stresses by applying post weld stress relief heat treatment Introduce compressive residual stresses at the weld toe using hammer, needle or shot peening, ultrasonic impact treatment or proof loading.

Techniques that reduce stress concentration The first way to reduce the stress concentration effect of the weld toe is to grind the weld flush so that there is no longer a change in section at the weld toe. The improvement that this can have on the fatigue performance is shown in Figure 4.20.

4-10

www.twitraining.com

Rev 2 April 2011 Advanced Fatigue – Part 1 Copyright  TWI Ltd 2012

Figure 4.20 Fatigue performance improvement as a result of weld flush grinding.

The most widely used technique is weld toe grinding, using either a burr grinder or disc grinder. The fatigue lives of welded joints which are likely to fail from the weld toe can be improved by weld toe grinding. The aim is to remove the sharp discontinuities (intrusions) due to welding and to blend the region to reduce the stress concentration factor. Burr grinding is the most effective method, but disc grinding (with great care to avoid deep notching and the introduction of deep scratches) can be effective. It is essential that material is removed at the weld toe, Figure 4.21, to ensure a smooth transition. Provided the grinding is carried out in a controlled and approved manner, an increase in fatigue life of a factor of 2 can be achieved (an improvement of 30% on stress or 2.2 on life). Slow and careful application required and the groove must not penetrate below the recommended depth for any undercut or flaw. Figure 4.22 shows a macro cross section of a burr ground joint. Figure 4.23 shows typical equipment used for toe grinding, Figure 4.24 shows the method of weld toe grinding using a burr grinder running along each weld toe, Figure 4.25 shows similar equipment but using a disc grinder instead of a burr grinder. Disc grinding is quicker than burr machining, but there is less improvement (of only one fatigue Class) because of the direction of the grinding marks. All toe grinding should be a controlled procedure and inspection is required. Other techniques such as remelting the weld toe with a TIG torch or plasma torch are also available but are more difficult to control and are less widely used.

4-11

www.twitraining.com

Rev 2 April 2011 Advanced Fatigue – Part 1 Copyright  TWI Ltd 2012

Figure 4.21 Removal of material when weld toe grinding.

Figure 4.22 A weld with toes that have been burr ground.

Figure 4.23 Equipment for weld toe grinding.

4-12

www.twitraining.com

Rev 2 April 2011 Advanced Fatigue – Part 1 Copyright  TWI Ltd 2012

Figure 4.24 Method for weld toe burr grinding.

Figure 4.25 Method for weld toe disc grinding.

In order to maximise the benefit of weld toe grinding, it needs to go deeply enough to remove the weld toe imperfections (eg intrusions in arc welded steel) from which fatigue cracks propagate. This will usually necessitate grinding to a depth of ~0.8mm. In addition, the resulting groove geometry needs to be controlled to minimise its effect as a stress concentration. Fatigue tests have shown that grinding with a small diameter burr (to achieve good access to the toe of a poor profile fillet weld) can leave such a sharp groove that the fatigue life of the joint is not increased. Experience indicates that grooves which meet the criteria shown in Figure 4.26 are acceptable. As seen, correct treatment can increase the fatigue strength of a fillet welded joint close to that of the unwelded steel plate. The improvement in the fatigue performance that can be achieved by weld toe dressing is shown in Figure 4.26.

4-13

www.twitraining.com

Rev 2 April 2011 Advanced Fatigue – Part 1 Copyright  TWI Ltd 2012

Figure 4.26 Fatigue performance improvement as a result of toe grinding.

A similar result to grinding can be achieved by re-melting the weld toe region with a TIG or plasma welding gun. It seems that the non-metallic material at the weld toe (including the weld toe intrusions) is removed and the resulting weld profile is improved locally, as shown in Figures 4.27 and 4.28. It is probably significant that weld toe intrusions do not seem to be produced with these two welding processes (probably because of the inert nature of each). These can be very slow techniques that require separate qualified procedures and a significant skill element. Applying these fatigue improvement techniques will delay subsequent weld inspection

Figure 4.27 TIG re-melting of weld toes.

4-14

www.twitraining.com

Rev 2 April 2011 Advanced Fatigue – Part 1 Copyright  TWI Ltd 2012

Figure 4.28 A fillet weld with plasma dressed toes.

4.8

Techniques that introduce compressive residual stresses These techniques generally involve severely peening the weld toe so that the ensuing plastic deformation gives rise to a compressive residual stress at the weld toe. The advantage to this is that by the principle of superposition, any subsequent cyclic load, even tensile loading, will not be experienced as so damaging at the weld toe. An example of this compressive stress superposition on an otherwise tensile cyclic load is shown in Figure 4.29. One common technique is hammer peening using a solid tool and a pneumatic hammer, Figure 4.30.

Figure 4.29 Effect of imposing a compressive residual stress on a tensile cyclic stress. The effective cycles are partly compressive and the tensile stress that might propagate a fatigue crack is greatly reduced.

4-15

www.twitraining.com

Rev 2 April 2011 Advanced Fatigue – Part 1 Copyright  TWI Ltd 2012

Figure 4.30 Weld toe hammer peening.

Other techniques include the use of a needle tool peening (Figure 4.31) and ultrasonic impact treatment (Figure 4.32).

Figure 4.31 Needle peening.

Figure 4.32 Ultrasonic impact treatment for weld toes.

4-16

www.twitraining.com

Rev 2 April 2011 Advanced Fatigue – Part 1 Copyright  TWI Ltd 2012

The general characteristics of the fatigue behaviour of welded joints with peened (eg hammer, needle, shot) toes are shown in Figure 4.33. These reflect the influence of applied mean stress (tensile mean stress will reduce the benefit of a compressive residual stress) and the level of compressive residual stress induced (material strength dependence). The potentially large effect of fatigue loading on the performance of peened welds needs to be considered very carefully before using this type of improvement method. On the positive side, there is some indication that high-strength materials gain more benefit than low strength. This is probably because the magnitude of residual stress induced depends on yield strength since it has been shown that peening does not re-introduce a crack initiation period but just slows down the rate of crack growth.

Figure 4.33 Improvement of fatigue performance as a result of weld toe peening.

Although the peening residual stress improvement techniques can be very effective, a major drawback is that they are vulnerable to factors which reduce the beneficial residual stress level in service. The main problem is the nature of the fatigue loading. It is known that occasional very high applied stress in a service stress history can modify residual stress, but even operation at a high tensile mean stress can eliminate their benefit. Essentially, the fatigue performance of welded joints containing compressive residual stress depends on both the range and the maximum value of the applied stress. Thus, there is no benefit from a residual stress-based improvement method when the maximum stress reaches yield.

4-17

www.twitraining.com

Rev 2 April 2011 Advanced Fatigue – Part 1 Copyright  TWI Ltd 2012

4.9

Typical improvement Figure 4.34 summarises fatigue improvement in terms of S-N curves for weld toes modified using a range of techniques. The greatest benefit is generated by hammer peening, but this technique gives rise to a plastically deformed weld toe that may make subsequent inspection difficult to interpret. Weld toe grinding is the next best approach. The aim is to remove the crack-like discontinuities at the weld toe and to reduce the stress concentration effect of the weld shape by smoothly blending the weld surface into the plate. If successful, such a treatment should re-introduce a significant crack initiation period into the fatigue life and hence raise it to a level approaching that of the un-welded material. The graph shows examples of the benefit from various improvement techniques. Other data show different curves, depending on the method of application of the technique. Optimisation of the techniques suggests that although hammer peening is the most effective, needle and shot peening can be equally effective, while toe grinding or re-melting can give about the same level of improvement. The IIW is currently developing guidelines for the application of weld toe improvement techniques. One additional benefit of weld toe improvement methods is that they tend to be more effective when applied to welds in high strength material. Thus, they offer the possibility of utilising high strength materials in fatigue-loaded welded structures (Figure 4.35). However, the reduced severity of the weld toe as a source of fatigue cracking means that the weld is less tolerant to flaws than it was in the aswelded condition. Thus, although the fatigue life of the improved weld detail will be increased, it is still unlikely to reach that of the un-welded material due to the inevitable presence of flaws. 400 Mild steel R=0 300

Hammer peened

Stress range 200 N/mm2 150

100 5 10

Burr ground Shot peened As welded

Plasma dressed

6

10

7

Endurance, cycles

10

7

5x10

Figure 4.34 S-N curves for postweld treated joints.

4-18

www.twitraining.com

Rev 2 April 2011 Advanced Fatigue – Part 1 Copyright  TWI Ltd 2012

Figure 4.35 Effect of fatigue improvement techniques over a range of steel yield strength.

Postweld fatigue improvement techniques are particularly useful when applied to a weld repair of a location that has experienced fatigue cracking. Summarising this topic, significant improvements can be made in fatigue life for joints at risk of failure from the weld toe. Typical improvement by a factor of 2-3 on fatigue life can be achieved. A well-defined method statement and operator training is required when applying a fatigue improvement technique and then post-treatment inspection is necessary to ensure complete coverage. Detailed recommendations are published by IIW and in BS 7608

4.10

Fatigue failure case studies

4.10.1 Alexander Kielland Platform The Alexander L. Kielland was a pentagone type of semi-submersible rig, converted for use as an offshore accommodation platform anchored in the North Sea. On the evening of 27 March 1980, one of the five columns (buoyancy elements) of the platform anchored broke off. The platform immediately heeled over to an angle of about 30° and then continued to heel and sink slowly. Twenty minutes after the loss of column D, the platform capsized, with the loss of 123 lives out of 212 men on board. Of the seven lifeboats on board only two were launched successfully. The failure initiation site was a fillet weld between the hydrophone support and a brace (Figure 4.36).

4-19

www.twitraining.com

Rev 2 April 2011 Advanced Fatigue – Part 1 Copyright  TWI Ltd 2012

Figure 4.36 Failure of the Alexander Kielland platform.

Fatigue crack growth in brace D6 initiated from pre-existing cracks in the fillet welds between a hydrophone support and the brace (Figure 4.37). A brittle fracture initiated from this fatigue crack, leading to total failure of brace D6. The pentagone design was non-redundant and the loss of one brace resulted in overloading of other braces supporting column D. Compensating measures did not function with the result that the platform could not remain afloat with only four columns. Examination of the hydrophone to brace fillet welds revealed poor weld penetration and an unsatisfactory weld bead shape. There was evidence of fabrication cracks in this region; although the Charpy impact energy levels were adequate, the through-thickness ductility of the hydrophone material was poor. This failure emphasises the importance of welded attachments to critical structural components. A crack that initiates at an attachment weld can run into the main elements of the structure.

Figure 4.37 Fatigue crack propagation of the Alexander Kielland platform.

4-20

www.twitraining.com

Rev 2 April 2011 Advanced Fatigue – Part 1 Copyright  TWI Ltd 2012

4.10.2

Mooring buoy failure A second fatigue-related failure was a mooring buoy, which separated within the tower structure below the flotation chamber. The top section was recovered for a failure investigation, whereas the larger part of the structure then fell to the bottom of the sea.

Figure 4.38 Failure of a floating mooring chamber.

Figure 4.39 The recovered top section of the mooring failure (now upside down).

4-21

www.twitraining.com

Rev 2 April 2011 Advanced Fatigue – Part 1 Copyright  TWI Ltd 2012

Some of the tubulars showed classic brittle fracture surfaces (Figure 4.40). Others were found to show fatigue cracking (Figure 4.41). Final ductile overload was also observed. From these failure mechanisms the engineers could formulate a sequence of failure (Figure 4.42).

Figure 4.40 Brittle fractures observed in parts of the failure.

Figure 4.41 Fatigue cracks observed in other parts of the failure.

Figure 4.42 Failure modes in different parts of the failed structure.

4-22

www.twitraining.com

Rev 2 April 2011 Advanced Fatigue – Part 1 Copyright  TWI Ltd 2012

A reasonable hypothesis for the cause of the fatigue crack initiation was that the weld was significantly misaligned, which would act as a stress concentration (Figure 4.43). However, much later on when the other half of the failed structure was recovered from the sea bed, the other half of the weld could be placed next to the initial half, which disclosed the reason why that weld had initiated a fatigue failure. Although no misalignment would have been measured on the outer surface, a significant part of the wall thickness had been removed to allow this fit-up (Figure 4.44). This would result in much higher stresses in the thinner wall, increasing the risk of fatigue cracking.

Figure 4.43 Reasonable theory for the initiation of fatigue in this weld.

Figure 4.44 Both recovered halves of the failed weld showing actually that there was significantly reduced wall thickness, causing the fatigue to initiate.

4.10.3 Tacoma Narrows Bridge The original Tacoma Narrow Bridge, Washington State USA opened on July 1, 1940. It received its nickname Galloping Gertie due to the vertical movement of the deck observed by construction workers during windy conditions. The bridge collapsed into Puget Sound the morning of November 7, 1940, under high wind conditions.

4-23

www.twitraining.com

Rev 2 April 2011 Advanced Fatigue – Part 1 Copyright  TWI Ltd 2012

Figure 4.45 Tacoma Narrows Bridge collapse.

The collapse of the bridge was due to a physical phenomenon known as aero elastic flutter caused by a 67 kilometer per hour (42 mph) wind. The cause of the collapse was due to forced resonance with the wind providing an external periodic frequency that matched the natural structural frequency. No human life was lost in the collapse of the bridge. The design of the bridge was decisive in it collapse. Previous bridges were constructed with thick deck sections which meant that wind loading would not produce resonance loading. For the Tacoma Narrows Bridge the designers opted for a slim deck section which made the bridge look ‘better’, but this meant that the bridge was in danger of entering resonance loading due to wind (this was not known at the time of the design). The bridge was rebuilt in 1950 with an improved design due to the lessons learned from the first bridge and is now nicknamed Sturdy Gertie. Modern suspension bridges use a box section for the deck, with a cross section specially designed to avoid this form of resonant excitation.

4.11

Summary At the end of this section you should understand the practical aspects of fatigue cracking, being able to explain its initiation from weld toe intrusions and its appearance on a fracture face. You should be able to describe a number of possible post-weld fatigue improvement techniques and explain how they improve the fatigue life.

4-24

www.twitraining.com

Rev 2 April 2011 Advanced Fatigue – Part 1 Copyright  TWI Ltd 2012

Revision Questions 1 What are the weld classes shown in this joint?

Answer:

2 Draw a fillet weld under cyclic loading and sketch fatigue cracks at the locations they will initiate.

3 You need to carry out postweld fatigue improvement on a stiffened ship structure. Which method would you choose and why?

4-25

www.twitraining.com

Section 5 Static Loading

Rev 2 April 2011 Static Loading Copyright  TWI Ltd 2012

5

Static Loading

5.1

Review This section of the course shows how to develop principles of design for static loading. When a metal is loaded under increasing force there is a period of elastic deformation, where the material resumes its original dimensions once the load has been removed. The limit of elastic behaviour is the material’s yield point, after which plastic deformation and ultimately necking and tensile rupture occurs. These features are seen on a stressstrain curve (Figure 5.1).

Figure 5.1 Features of a stress-strain curve.

The Elastic Design Method ensures that the stresses in structure do not exceed the yield stress (ie that only elastic deformation occurs, not plastic deformation). However, it is not usually possible to design a structure to be loaded up to yield stress safely due to factors including material defects, joint/weld mismatches, unforeseen loads (weather conditions etc), or degradation. For structures that have to withstand principally static loading, the maximum allowable stress is limited to a proportion of the material yield strength. The maximum design stress is expressed as a fraction of the yield strength of the parent material. For critical structures such as pressure vessels this was once set at 1/4 UTS, but later changed to 2/3 yield stress. Other relevant codes dictate design stresses through other formulae, but half to 2/3 of yield strength is usual. Imposing a limit to the design stress is in effect a ‘safety factor’ to account for some material degradation and unforeseen loads. The ratio of yield stress (or UTS) to design stress is known as factor of safety (FoS). The factor of safety depends on the material and its utilisation.

5-1

www.twitraining.com

Rev 2 April 2011 Static Loading Copyright  TWI Ltd 2012

In the more general case, the maximum allowable stress is restricted such that all criteria for failure are satisfactorily avoided, with an appropriate margin of safety. This permits a very limited degree of plastic deformation in some cases. Additionally, it considers other potential ‘failure modes’; for example the allowable elastic deformation may correspond to a stress less than ⅔ yield strength. This is called Limit State Design.

5.2

Stress and strain - revision A component such as a bar or a plate may be subjected to a simple tension load, giving rise to a normal stress, Figure 5.2.



N A

A N

N

where:  = axial stress (N/mm2 or MPa) N = load or axial force (N) A = cross section area (mm2)

Figure 5.2 Normal stress.

A second form of loading may cause the material to slide over itself; this is known as shear loading, which gives rise to a shear stress, Figure 5.3.



T

T A

A

where:  = shear stress (N/mm2 or MPa) T = transverse force (N) A = cross section area (mm2)

T

Figure 5.3 Shear stress.

In the general case, a two dimension square may be subject to two normal stresses at right angles and a shear stress. This may be extended to three dimensions.

5-2

www.twitraining.com

Rev 2 April 2011 Static Loading Copyright  TWI Ltd 2012

5.3

Fillet welds under stress When considering the load carrying capacity of a weld, it is assumed that the weld metal strength overmatches that of the parent metal. The parent strength therefore defines load carrying capacity of the weld. Remember that this may not be the case for high strength low alloy steels where the weld metal sometimes undermatches the parent metal strength and for welded joints in aluminium where the weld can be softened by the heat of welding. Fillet welds are often used because they are the simplest and cheapest type of welded design. Fillet welds require less joint preparation and may be made in flat or horizontal positions by semi-skilled operators. Welds can be completed with any number of passes There can be possible problems with penetration into the parent metal and they are difficult to examine by NDT methods for subsurface defects. Once the required size of fillet weld has been calculated it is important to stick to it, the volume or weight of weld metal increases as the square of leg length and it is surprisingly easy to over-weld fillet welds, with associated cost and time implications. When calculating fillet weld sizes, not only is it assumed that the weld metal matches (or overmatches) the strength of parent metal, but it is also assumed that the weld will fail across the throat. The weld throat dimension is the shortest distance between the root and the chord between the toes or profile face (whichever is less), see Figure 5.4. Excess weld metal is neglected and for calculation use the design throat instead actual throat. Adequate weld quality is assumed, ie the weld is defect-free and stress concentrations due to bead shape and residual stresses are neglected.

A

b

Figure 5.4 The throat dimension in a fillet weld a and for a single bevel Tee butt weld b.

The design stress is half the yield strength when it is not explicitly given in the applicable standard. This is lower than a butt weld because fillet welds tend to experience high shear forces rather than axial forces. The effective yield under shear is lower than for axial loading, while fillet welds, being difficult to inspect for root fusion defects, tend to be designed with a more

5-3

www.twitraining.com

Rev 2 April 2011 Static Loading Copyright  TWI Ltd 2012

conservative approach to reflect the probability of defects. When designing a fillet welded joint, welds with faces meeting at more than 120° or less than 60° should NOT be used for load carrying parts. Some codes specify minimum fillet weld size depending on thickness of plate to be welded (Table 1). This has more to do with prevention of cold cracking than for strength. The maximum throat size for fillet welds made from one pass is 6mm. If plates are 6 mm thick or more then the leg length should be not less than 5mm. Do not overweld as this gives a risk of lamellar tearing (Figure 5).

Figure 5.5 Lamellar tearing. Table 5.1 Fillet weld sizes given by standards. Recommended by BS 449 Thickness of thicker part (mm)

Minimum leg size (mm)

10-18

4

18-30

5

Over 30

6

Recommended by AWS D1.1

5.4

Thickness of thicker part (mm)

Minimum leg size (mm)

Up to 6

3

6-12

5

12-20

6

Over 20

8

Forces, frames and beams As well as forces acting axially in tension and compression and forces acting offset in shear, forces can act to impose a bending moment on a beam (Figure 5.6).

5-4

www.twitraining.com

Rev 2 April 2011 Static Loading Copyright  TWI Ltd 2012

Figure 5.6 Different types of loading.

In a simple frame, resolving of forces can be used to determine the forces in each member. If each member can rotate fully above its ends, then the members are subjected only to axial forces and determining stress is simply evaluated using force/area. If, however, a member is not allowed to rotate fully, eg it is fixed in a support and then a bending moment is experienced by the beam, eg Figure 5.7. The bending moment, M, is equal to the force, F, multiplied by the perpendicular distance that the load is applied, d. The force applied at the end of the cantilever beam also imposes reaction forces at the attachment point, as well as the bending moment.

Force, F

Fy M

Fx d

Figure 5.7 A cantilever beam.

The bending moment is not constant along the length of the beam; it is highest at the fixed end of the beam and drops to zero at the point of loading. The shear force experienced, however, is constant across the length of the beam. These can be illustrated by constructing a bending moment diagram, Figure 5.8.

5-5

www.twitraining.com

Rev 2 April 2011 Static Loading Copyright  TWI Ltd 2012

P

x

L

BM

+PL

SF

+P

Figure 5.8 A typical bending moment (BM) diagram for a cantilever beam of length L under an applied force, P, also showing the shear force (SF).

Applying a bending moment causes the beam to defect which imposes a tensile stress on the outer fibre of the beam and a compressive stress on the inner fibre of the beam (Figures 5.9 and 5.10). The plane of the beam that exhibits zero stress is known as the neutral axis.

Figure 5.9 A beam in bending.

Figure 5.10 The stress in a beam in bending.

It is possible to calculate the tensile stress on the outer fibre of the beam in bending using the engineers bending formula. The stress  at a distance y from the neutral axis is given by:

5-6

www.twitraining.com

Rev 2 April 2011 Static Loading Copyright  TWI Ltd 2012

y M I



where: M = bending moment. I = second moment of area, or moment of inertia. Maximum values for bending stress are obtained on the extreme fibres of the beam. For example, an I beam has flanges to resist bending while the web resists shear. The moment of inertia depends on the geometry of the beam. For a beam of width b and depth d, 3 d 2 b 1

I 

Stresses have a continuous linear variation throughout entire transverse section. The neutral axis (where σ=0) goes through the centre of gravity of the transverse section. For many geometries the moment of inertia can be calculated from first principles. However, values are given in standard text books. The section modulus (Z) is a geometric property only which relates stress and internal moment during elastic bending. The bigger the value of section modulus for a particular cross-section the bigger the bending moment which it can withstand for a given maximum stress.

where ymax is the distance from the neutral axis to the extreme fibres of the beam. Using the section modulus makes the engineers bending formula simplified into  = M/z. Figure 5.11 shows improvements to beam cross sections to increase the section modulus. The intention is to move material further from the neutral axis and centre of gravity.

Figure 5.11 Ways to improve the stiffness of beams, by increasing the section modulus.

5-7

www.twitraining.com

Rev 2 April 2011 Static Loading Copyright  TWI Ltd 2012

5.5

Bending worked examples a For this cantilever beam show how to calculate the bending moment; the moment of inertia; the maximum distance to outer fibre; and the bending stress.

The bending moment, M, is force multiplied by perpendicular distance, i.e PxL. The moment of inertia, I, is breadth times by height cubed, divided by twelve for a rectangular beam, ie bh3/12 The maximum distance to the outer fibre, y, is the distance from the neutral axis (middle of the beam) to the top surface in tension, or half the height, ie h/2 The bending stress is determined using the engineers bending fornmula, stress is the bending moment multiplied by the maximum distance, divided by the moment of inertia ie Mxy/I b Calculate the maximum stress in this beam subjected to 6kNm bending.

Using the engineers bending formula, stress equals My/I. The bending moment, M is 6,000Nm. y is half the height, which equals 150mm. The second moment of area, I is calculated by:

5-8

www.twitraining.com

Rev 2 April 2011 Static Loading Copyright  TWI Ltd 2012

The bending stress is therefore:

This is the maximum stress on the outer fibre of the beam under bending. This bending stress decays to zero at the middle of the beam along its neutral axis, where it becomes negative or compressive stress towards the inner fibre of the beam. This is illustrated below.

c A bending moment of 9x105Nmm is applied to the end of a beam 1.5m long. The beam is 50mm deep and 20mm wide. Calculate the leg length needed for the fillet welds shown below if the allowable stress in the fillets is 250MPa. How would this change if the beam length was 0.75m?

The applied moment equals the moment reacted in welds = 50 x F, where F is the force in the fillet welds. F = 9x105/50 = 1.8x104N The stress in each fillet weld is the force divided by the cross section area = F/(t.w), where t is the throat thickness and w is beam width (20mm). This stress must equal the allowable stress in the welds, 250MPa. Remember that the throat, t = L/√2, where L is the leg length Hence allowable stress 250 is related to leg length as: So,

250 L

= = = =

(F√2)/(L.w) (1.8x104 √2)/(250.20) (2.54x104)/(5000) 5.1mm

5-9

www.twitraining.com

Rev 2 April 2011 Static Loading Copyright  TWI Ltd 2012

It would be rare to specify a leg length to a fraction of a millimetre, so the welding procedure would round up to the next millimetre and state a leg length of 6mm. In answer to the second part pf the question, the beam length makes no difference to required weld size because the applied loading is a bending moment. That is, F is independent of beam length.

5.6

Torsion Torsion is similar to bending, but acts on round tubes and bars. The torsional stress, , calculation for the stress on the outer surface of a beam under torsion (which is valid only for round shape transverse section such as beams, bars or tubes) is given by: τ

Mt Zp

Where Mt is the torsion moment and Zp is the polar section modulus of the transverse section (equivalent to second moment of area and can be looked up for given cross section shapes). The largest values of stress are obtained on the extreme fibres of the beams while in centre of section, stress is zero. The stress has a continuous linear variation throughout entire transverse section.

Figure 5.12 Torsion in a round bar.

The fillet weld in Figure 5.13 is subject to torsion loading. The weld area is given by  x D x t, where D is the diameter and t is throat thickness.

5-10

www.twitraining.com

Rev 2 April 2011 Static Loading Copyright  TWI Ltd 2012

Figure 5.13 Torsional stress around a cylinder.

The torsion moment can be replaced by a force F and perpendicular distance and then the force can be calculated:

The stress is the force over the cross section area, ie the torsional force over the weld area. The overall calculation is therefore given by:

5.7

Stresses in cylinders Another kind of static loading is pressure loading inside a cylinder. This is covered in detail in the section on pressure vessels, but is included here for revision. For a hollow cylinder (or pipe) of radius r and wall thickness t containing a pressure p, the longitudinal stress is given by: t r 2 p

Longitudinal stress =

The hoop stress, ie the stress acting at right angles to the longitudinal stress is given by: r t p

Hoop stress =

The hoop stress in a sphere is given by: t r 2 p

Sphere hoop stress =

5-11

www.twitraining.com

Rev 2 April 2011 Static Loading Copyright  TWI Ltd 2012

5.8

Reinforcing steel

5.8.1

Purpose Reinforcing steel describes the use of steel to reinforce materials, most often concrete. Concrete is a brittle material which is strong in compression but weak in tension which limits its use in construction and makes it unsuitable for use in many structural members. A beam composed only of concrete has little or no bending strength since cracking occurs in the extreme tension fibres in the early stages of loading. To improve the tensional capability of concrete steel bars called reinforcing bars, or rebar, are embedded in the concrete. In concrete beams the steel bars are embedded in the tension fibres of the beam as shown in Figure 5.14.

Figure 5.14 Steel reinforced concrete I-beam.

One advantage of using steel for this purpose is that the coefficients of thermal expansion of steel and concrete are very nearly equal so, any change in temperature will result in the steel and concrete expanding or contracting at the same rate, minimising deformation stresses. 5.8.2

Bar profile If concrete is cast around a steel bar, on setting the concrete shrinks and grips the steel bar. To aid this bond the rebar is rolled with a ribbed die when it is manufactured to produce a ribbed surface, Figure 5.15.

Small ribs to ensure a tight grip between the concrete and the reinforcing bars

All reinforcing bars are deformed bars; ribs are rolled on the surface of the bar Figure 5.15 Reinforcing-steel bar profile.

5-12

www.twitraining.com

Rev 2 April 2011 Static Loading Copyright  TWI Ltd 2012

Ribbed bars are characterised by the dimensions and the number and configuration of transverse and longitudinal ribs which influences the bond with the concrete. 5.8.3

Joints Reinforcing bar is available from 6mm - 50mm diameter. Depending on the size of the concrete casting a whole assembly of reinforcing bars will usually be used. Three methods are used to join reinforcing bars together:   

Welded joint. Wire joint. Rebar coupler.

The most common types of joint are welded and wire. Wire joints are simply connected by metal wire wrapped around the bars and tightened and are used on non-load bearing joints. Rebar couplers are mechanical fixings used to connect bars together, often using threaded ends on the bars. This course looks at welded joints. Welding can be used for both load and non-load bearing joints. A non-load bearing joint normally only keeps the reinforcing components in their correct place during fabrication, transportation and concreting and are often referred to as tack welds. Welds which may be classified as non-load bearing in terms of their action in the design of the structure may be subject to significant loads during handling and transport of the assembly to site, in such cases the welds should be treated as load bearing welded joints. 5.8.4

Properties Reinforcing bars are available for a wide range of chemical compositions and mechanical properties. The type of bar to be used will be detailed in the relevant construction code for the structure being built. Not all reinforcing bars are weldable as weldability is determined by the carbon equivalent value and the limitations on the content of certain elements. BS 4449 and BS EN 10080 give the maximum values for these elements.

5-13

www.twitraining.com

Rev 2 April 2011 Static Loading Copyright  TWI Ltd 2012

5.9

Welding reinforcing steel The following welding processes in accordance with ISO 4063 may be used to perform the welding of reinforcing steel joints. Table 5.2 Welding processes.

5.9.1

Welding process

Description

Load bearing

Non-Load bearing

111

Manual metal arc welding (MMA)

●

●

114

Self-shielded tubular cored arc welding

●

●

135

Metal active gas welding (MAG)

●

●

136

Tubular cored metal arc welding with active gas shield

●

●

21

Resistance spot welding

●

●

23

Projection welding

●

●

24

Flash welding

●

25

Resistance butt welding

●

42

Friction welding

●

47

Oxy-fuel gas pressure welding

●

Types of Joints The designation of joints discussed are in accordance with BS EN ISO 17660:2006. Other standards such as AWS D1.4 use different joint classifications. All of the joints discussed below, with the exception of cross joints, are designed to give full load bearing capacity of the bar. Note: the joints shown are applicable for welding processes 111, 114, 135 and 136. Butt welds are used to join two bars together, end to end, only necessary for load bearing joints and a range of groove designs are available to use. Single V groove

Double V groove

Single bevel groove

Double bevel groove

Figure 5.16 Groove designs for butt welding of reinforcing bar.

5-14

www.twitraining.com

Rev 2 April 2011 Static Loading Copyright  TWI Ltd 2012

Backing strips may be used when making a butt weld. Figure 5.17 shows a single V groove butt weld with a backing strip (applicable only to bars over 12mm). The backing strip should be tack welded to the bar on the inside.

Figure 5.17 Full penetration butt weld with permanent backing strip.

Lap joints join two bars together by staggering the ends and then welding them together and can be used for both load and non-load bearing joints. For load bearing joints a double sided weld is also possible with a minimum weld length of 2.5d. The weld dimensions given in Figure 18 are valid for load bearing joints. The minimum throat thickness, a, should be ≥ 0.3d.

d ≥4d

≥2d

≥4d

Figure 5.18 Lap joint.

Strap joints use short pieces of reinforcing bar as straps to join two pieces of bar together. Strapping provides a strong joint so is only necessary for load bearing joints. It requires no groove preparation unlike a butt joint but requires extra material (straps) and more welding is required. A double sided weld is possible with a minimum weld length of 2.5d. Where the mechanical properties of the strap and bar are equal, the combined area of the two straps shall be equal to or greater than the cross-sectional area of the bars to be joined. The minimum throat thickness, a, should be ≥ 0.3d.

5-15

www.twitraining.com

Rev 2 April 2011 Static Loading Copyright  TWI Ltd 2012

Straps d

≥4d

≥2d

≥4d

Figure 5.19 Strap joint.

A variation on the strap joint makes use of a splice plate or angle as a replacement for the strap bars. The cross or cruciform joint is used when two bars are joined perpendicular to each other, one on top of the other and are suitable for both load and non-load bearing joints. Non-load bearing joints are usually single sided but for load bearing joints double sided welds should be used when possible. To avoid cracks the minimum throat thickness, a, should be ≥ 0.3dmin and the length of the weld, l, should be ≥ 0.5dmin. If more than one transverse bar is used on the same side of the longitudinal bar, the spacing of the transverse bars should be at least three times the nominal diameter of the transverse bar. When welding bars of different diameters, dmin/dmax should be ≥ 0.4.

a Figure 5.20 Cross joint - single sided weld.

5.10

Joints between reinforcing bar and other steel components For load bearing functions, steel reinforcing bars are often joined to other steel components in a structure such as plates or sections. The type of joint used includes side lap weld and transverse end plate. Specifications of these joints can be found in the relevant standard.

5-16

www.twitraining.com

Rev 2 April 2011 Static Loading Copyright  TWI Ltd 2012

Table 5.3 Common range of bar diameters for non-load bearing welded joints depending on the welding process. Welding process

Type of joint

Bar diameters, mm

21 23

Cross joint

4-20

24 25

5-25

Butt joint

6-50

Joint to other steel component

6-50

Butt joint

6 to 50

Butt joint without backing

≥16

111

Butt joint with permanent backing

≥12

114

Lap joint

6 to 32

135

Strap joint

6 to 50

136

Cross joint

6 to 50

Joint to other steel component

6 to 50

42 47

5.10.1

5-50

Butt joint

Reinforcing bar joint loading Direct and indirect loading describe the way forces are transmitted across a joint. To classify the loading the forces in each bar are treated as acting through the centroid of the cross-section. If the two force lines in a joint match up then the forces are transmitted directly, ie concentrically. A butt joint is classed as a direct loading joint as the bars are axially aligned; lap and strap joints are indirect loading joints as the forces are transmitted with eccentricity. Eccentric loading causes the joint to flex hence bending stresses are set up. Attention needs to be paid to the steel-concrete bond strength at these joints to ensure the concrete does not split when the joint flexes.

5.10.2 Preheating Preheating is the process applied to raise the temperature of the parent steel before welding, used to slow the cooling rate of the weld and the base material, resulting in softer weld metal and HAZ microstructures with a greater resistance to fabrication hydrogen cracking. The slower cooling rate encourages hydrogen diffusion from the weld area by extending the time over which the temperature is elevated to where hydrogen diffusion rates are significantly higher than at ambient temperature. The reduction in hydrogen reduces the risk of cracking.

5-17

www.twitraining.com

Rev 2 April 2011 Static Loading Copyright  TWI Ltd 2012

The minimum preheat temperatures for reinforcing steel welded joints are dependent on the type of joint, the carbon equivalent value of the steel, the diameter of the bar or combined diameter of the joint, the hydrogen content of the weld metal and for some joints the arc energy. The following data is taken from BS 7123:1989. Table 5.4 Minimum preheat temperatures for butt and cross joints. Nominal bar size, mm

Carbon equivalent, %

≤ 25

> 25-40

> 40

Hydrogen controlled consumable

Nonhydrogen controlled consumable

Hydrogen controlled consumable

Nonhydrogen controlled consumable

Hydrogen controlled consumable

Nonhydrogen controlled consumable

0.42 or less

0°C

50°C

0°C

75°C

50°C

100°C

> 0.42 to 0.51

50°C

100°C

75°C

*

100°C

*

* Hydrogen controlled consumables to be used

5-18

www.twitraining.com

Rev 2 April 2011 Static Loading Copyright  TWI Ltd 2012

5.11

Standards and specifications BS EN ISO 17660: Welding - Welding of reinforcing steel. AWS D1.4: Structural Welding Code - Reinforcing Steel. Eurocode 2: ‘Design of Concrete Structures’. BS EN 10080: ‘Steel for the reinforcement of concrete - Weldable reinforcing steel – General’. BS 4449: ‘Steel for the reinforcement of concrete - Weldable reinforcing steel - Bar, coil and decoiled product – Specification’. BS 7123: ‘Metal arc welding of steel for concrete reinforcement’. AWS D1.1: ‘Structural Welding Code’ BS 5950: ‘Structural use of steelwork in building’ BS 8118: ‘Structural use of aluminium’ Eurocode 3 (BS EN 1993): ‘Design of steel structures’ Eurocode 9 (BS EN 1999): ‘Design of aluminium structures’ BS 7608: ‘Code of practice for fatigue design and assessment of steel structures’ BS 7910: ‘Guide to methods for assessing the acceptability of flaws in metallic structures’ BS EN 22553: ‘Welded brazed and soldered joints. Symbolic representation on drawings’

5.12

Summary At the end of this section you should be able to calculate the static stress in beams under axial loading, bending moments and torsion. You are expected to explain the way that reinforcing steel bars improve the load bearing ability of concrete and give joint designs for welding reinforcing bars together.

5-19

www.twitraining.com

Rev 2 April 2011 Static Loading Copyright  TWI Ltd 2012

Revision questions 1 What formula do you use to calculate the second moment of area for a rectangular section beam?

2 What is the section modulus?

3 Show designs with improved section modulus compared to solid round or square bars.

4 Why are reinforcing bars used in concrete beams and structures?

5 What precautions are needed when welding reinforcing bars?

5-21

www.twitraining.com

Section 6 Design of Pressure Equipment

Rev 2 April 2011 Design of Pressure Equipment Copyright  TWI Ltd 2012

6

Design of Pressure Equipment Pressure vessels often contain a combination of high pressures together with high temperatures and in some cases flammable fluids and highly radioactive materials. Because of such hazards it is imperative that the design be such that no leakage can occur. In addition, the failure of a pressure vessel has a potential to cause extensive physical injury and property damage. Plant safety and integrity are of fundamental concern in pressure vessel design and these of course depend on the adequacy of the design codes. The main standards for the design construction of boilers and pressure vessels in the UK and in Europe are PD 5500 and the BS EN 13445 series. BS EN 13445 is harmonised with the Pressure Equipment Directive (PED), so that compliance with BS EN 13445 can be used to demonstrate compliance with the PED. In other parts of the world, the ASME Boiler and Pressure Vessel Code (ASME= American Society of Mechanical Engineers) is most widely used. These design codes specify maximum allowable stress, the minimum design temperature (MDT) and list permitted materials, mechanical properties and design features that can be used in the design of pressure vessels. These are all to ensure that the pressure vessel designs conservatively meet the service requirements and ensure the safety of the vessel over its lifetime. A range of materials are available to use in the construction of pressure vessels depending on the service pressure, service temperature and the fluid being contained. Generally, the base requirements for a material to be used in a pressure vessel are good mechanical properties (particularly strength and toughness) and adequate corrosion resistance for the purpose. The majority of vessels are made from steel. The corrosion resistance is inherent in the shell material for thin-walled pressure vessels which may be made from stainless steel, aluminium or polymer, or it could be in the form of an internal cladding or liner made from ceramic, corrosion-resistant alloy or polymer. Some pressure vessels are even made from composite materials, such as wound Kevlar or carbon fibre held in place with a polymer. These are used for firefighter’s breathing tanks. Pressure vessels can be lined with a variety of materials such as polymers, ceramics and other metals in order to improve corrosion resistance or to help carry a portion of the applied load. The relevant standard will contain a list of all the approved materials which can be used.

6.1

Pressure vessel components A simple pressure vessel comprises of the following parts: Shell(s) Main body of the vessel, it is most often cylindrical but some pressure vessels might use conical or spherical shells.

6-1

www.twitraining.com

Rev 2 April 2011 Design of Pressure Equipment Copyright  TWI Ltd 2012

Heads A head is present at each end to complete the basic shape and produce a closed container. They are most often dished but can also be flat. Nozzles A number of openings for filling, inspection or drainage. Saddle supports Saddle supports hold the pressure vessel in place. Nameplate Nameplates indicate the main working parameters of the pressure vessel including work pressure and temperature. Details also included may be the manufacturing company, year of manufacture, the relevant code according which the pressure vessel has been designed and manufactured and the inspection body stamp.

Figure 6.1 Cylindrical pressure vessels.

6.2

Design-by-rule and design-by-analysis Virtually all national pressure vessel design codes are based on the concept of design-by-rule (also called design-by-formula in EN 13445). Basically, this involves the use of relatively simple equations to determine the required thicknesses of the components of a pressure vessel based on a standardised design stress. Such equations have become more and more complicated with time due to the use of computers. The rules also incorporate various specific requirements and limitations. The rules are usually based on experience. The main advantages of the design-by-rule approach are simplicity and consistency. In theory, if the rules are correctly applied, the results should be consistent. In practice, however, the rules are sometime open to interpretation. The main limitations of the design-by-rule approach become apparent when dealing with loadings and geometries which are not covered by the standards. The design-by-analysis approach involves the use of stress analysis of a component under loading and compares the stresses against

6-2

www.twitraining.com

Rev 2 April 2011 Design of Pressure Equipment Copyright  TWI Ltd 2012

specific criteria. A similar limit analysis approach determines the load to cause failure by gross plastic deformation. Safety factors are applied to the limit loads to give an allowable load that ensures a constant level of safety is present in all parts of the vessel. The design-by-analysis method is now incorporated to some extent in several pressure vessel standards, including EN 13445, PD 5500 and ASME VII-Division 2.

6.3

Details of design Shell Cylindrical shells (also called strakes) are usually made up of one or more of curved plates welded longitudinally (seam welds) to obtain a cylindrical section. The number of plates used to make a shell depends on the diameter of the shell itself. The shells are then welded together and welded to the heads to form the pressure vessel (circumferential welds). According to the vessel length, one or more shells can be used (Figure 8). Shell/Strake

Figure 6.2 Schematic representation of a pressure vessel.

The shell of a pressure vessel can range in thickness from only a few millimetres for an LPG gas bottle to several hundred millimetres for industrial pressure vessels. The minimum design thickness is dependent on the shape of vessel, the internal pressure, the diameter of the vessel and the material strength. A spherical shell requires a much smaller wall thickness than a cylindrical shell for the same diameter, internal pressure and construction material. If a pressure vessel is to contain corrosive fluids there may be an additional corrosion allowance given on the required wall thickness, although the design stresses are calculated without the corrosion allowance. The hoop stress within a pressurised cylinder is twice that of the axial stress, so there is a higher potential that the failure of a pressure vessel will occur along the longitudinal welds. In order to prevent cracks propagating along the entire length of a vessel, longitudinal welds are normally offset. In the ASME Boiler and Vessel Code, Section VIII, Division 1, vessels made of two or more strakes shall have the centres of the welded longitudinal joints of adjacent courses staggered or separated by a distance of at least five times the thickness of the thicker plate. In PD 5500, the longitudinal seams

6-3

www.twitraining.com

Rev 2 April 2011 Design of Pressure Equipment Copyright  TWI Ltd 2012

of adjacent courses shall be staggered by four times the thickness of the plate or 100mm, whichever is greater, measured from the toe of the welds. 6.3.1

Dished heads Dished heads are produced by pressing or forming a blank in a die using a former. If the size of this blank is too large to be cut from a single plate, then it is fabricated by welding a number of petals around a central crown.

Figure 6.3 Schematic of a fabricated dished head.

It is a common practice to subject all these welds to a 100% UT examination to ensure they are defect free and will carry the required pressure. It must be noted that a dished head is fabricated out of an odd number of petals in order to offset the welds and avoid the propagation of a longitudinal defect from one weld across the entire dished head.

a

b

c

Figure 6.4 Types dished head; a hemispherical, b semi-elliptical, c torispherical.

Hemispherical dished head A hemispherical dished head requires the smallest thickness among all types of dished heads, since the stress in a sphere is half the hoop stress in a cylinder. However, they are the most difficult to fabricate.

6-4

www.twitraining.com

Rev 2 April 2011 Design of Pressure Equipment Copyright  TWI Ltd 2012

Semi-elliptical dished head The required thickness for a semi-elliptical dished head is equal to that of the cylindrical shell, so there is no need to supply different plate thickness to manufacture a pressure vessel with semi-elliptical dished heads. Unfortunately, an ellipse is difficult to generate (the die-former set requires special machining, usually on a CNC machine with elliptic interpolation programme). Torispherical dished head A torisphere consists of a spherical central portion or crown with a large radius and a toroidal knuckle of smaller radius. The ratio between the large and small radii is approximately twelve and the portion with the large radius is about 95% of the cylinder diameter. The junction of the torus with a cylinder gives rise to bending stresses., but torispherical dished heads are often preferred to semi-elliptical ones since the depth of drawing is less and so they are cheaper to manufacture. Also the small axial dimension is of advantage when the longitudinal size of the pressure vessel is a critical factor. However, the thickness required for a torispherical dished head is larger than the thickness of the corresponding cylindrical shell. 6.3.2

Dished head-to-shell joints In order to accommodate heads and shells of different thicknesses, taper transitions are used, with the gradient of the taper being at least 1 in 4. The weld preparation used at the shell to head joint can be designed to account for any misalignment which may occur between the head and the shell (Figure 6.5a). The simplest of these is a ’K’ bevel as seen in Figure 6.5b).

a

b

Figure 6.5 Weld preparations to accommodate misalignment.

The main advantage of this type of preparation is that only one side of the joint needs to be machined to a specific geometry. The disadvantage however is that access to both the internal and external surfaces is required in order to complete the weld. Also, the misalignment can cause local bending stresses in the shell. Alternative joint designs suitable for singlesided access for shells to heads welds are shown in Figure 6.6.

6-5

www.twitraining.com

Rev 2 April 2011 Design of Pressure Equipment Copyright  TWI Ltd 2012

a

b

Figure 6.6 Alternative self-jigging single-sided access joint designs for shell to head welds.

Both these designs shown in Figure 6.6 have the advantage of requiring no jigs during fit-up since both are self-jigging. They both also allow for use of high heat input welding processes to deposit the weld root as backing is provided, require access from one side only and have no danger of burn through. The disadvantages of such weld preparations are that they are difficult to inspect and have poor resistance to fatigue. In addition, Figure 6.6a) results in a high level of residual stresses and is hard to apply to thin walled components. Figure 6.6b) on the other hand can only be applied to thin walled components.

6.4

Nozzle design Nozzles are used to connect pressure vessels to other components (eg pressure piping), as well as providing access for inspection and sampling. The presence of the nozzle results in a hole being drilled through the shell and thus the designer needs to consider whether it is necessary to compensate for the loss in strength in the design. The two types of nozzle that are used in pressure vessel manufacturing are set-on nozzles and setthrough nozzles. Each nozzle design is characterised by its own set of advantages and disadvantages.

Figure 6.7 Set–on nozzle.

Set-on nozzles have the benefits of being cheaper to forge since they are shorter, requiring less weld metal leading to less distortion; groove preparation is easy and can be used with full or partial penetration welds. Disadvantages of set-on nozzles however include the limitation for access from one side only and welding in the 2G/2B position and thus requirement for a skilled welder. With set-on nozzles, the through thickness stress from welding can lead to lamellar tearing and it can be difficult to apply UT on

6-6

www.twitraining.com

Rev 2 April 2011 Design of Pressure Equipment Copyright  TWI Ltd 2012

small diameters. Set-on nozzles are generally used when the nozzle diameter is small (typically under two inches) and/or when the vessel has a large diameter or thick wall. When using set-on nozzles, it is often necessary to use compensating plates or other reinforcement.

Figure 6.8 Set–through nozzle.

Set-through nozzles on the other hand can be welded in the 1G/PA positions thus having potential for less defects and the groove preparation can be flame cut cheaply. There is no danger of lamellar tearing and easy access to back side of root allows full penetration to be achieved more easily and provides access for easier inspection, although it can be hard to UT smaller diameter nozzles. The disadvantages of set-through nozzle are the requirement for the nozzle body to be longer, meaning a greater weld volume and potentially larger weld distortion. Smaller diameter set-through nozzles generally do not require reinforcement. Set-through nozzles are the preferred design especially for larger diameter nozzles or for thinner walled or smaller diameter pressure vessels. It is often difficult to weld a very small diameter nozzle to a shell effectively and obtaining full penetration. In this case a solid round bar is welded to the shell and then drilled through to produce the opening (Figure 6.9).

Figure 6.9 Drilled out round bar to aid penetration of small diameter nozzles.

6-7

www.twitraining.com

Rev 2 April 2011 Design of Pressure Equipment Copyright  TWI Ltd 2012

6.5

Reinforcement or compensation A hole or opening in the shell of a pressure vessel for nozzles can have a detrimental effect on the structural integrity of the shell, due to the stress concentration effect of the hole. In order to compensate for this local reduction in strength, reinforcement is used. The required thickness of the shell is a function of the operating stresses, so if the shell experiences higher stress in a region due to stress concentration the shell thickness needs to be greater in this region to withstand the stresses. Reinforcement can be used either on the shell, on the nozzle or both. Figure 6.10 shows two types of reinforcement.

A

b

Figure 6.10 Nozzle reinforced by a) compensating ring and b) extended nozzle neck.

The approach to determining the amount of reinforcement required is shown below.

Parent metal.

6-8

www.twitraining.com

Rev 2 April 2011 Design of Pressure Equipment Copyright  TWI Ltd 2012

Hole drilled.

Compensation around the nozzle. Figure 6.11 Illustration of the increase in thickness needed to accommodate the loss of strength around a nozzle hole.

6.6

Flanges Flanges connected to the nozzles attach other pieces of plant to the pressure vessel via a network of piping. The flange faces are held together using bolts, which enables them to be disconnected and reconnected when needed. A gasket, made of softer material sits between the flange faces and deforms under load to produce a tight seal in the joint. Most flanges are designed to a particular code (such as ASME B31.3) and will have standard dimensions and a given pressure rating. This is to utilise their interchageability and flexibility to attach to other pieces of pipework and equipment.

6-9

www.twitraining.com

Rev 2 April 2011 Design of Pressure Equipment Copyright  TWI Ltd 2012

Figure 6.12 Flanges, showing the design principle and a flange attached to a pressure vessel nozzle.

6.7

Corrosion allowance Corrosion occurring over the life of a vessel is catered for by a corrosion allowance on the design wall thickness. The design value of corrosion allowance depends upon the vessel duty and the corrosiveness of its contents. As an example, a design criterion of 1mm corrosion allowance is typical for air receivers in which condensation of air moisture is expected. Most vessel codes use the principle that, when dimensions in any formula refer to a corrodible surface, then the thickness dimensions inserted into the stress calculation formula are those at the end of the vessel life, ie when all the corrosion allowance has been used up.

6.8

Weld joint efficiency Most vessel codes assume that welded joints are not as strong as the parent material, unless they are exhaustively inspected during manufacture and repaired if defects are found. This strength reduction is characterised by the welded joint efficiency. The welded joint efficiency is defined as the ratio of joint strength to parent material strength. It varies from 100%-45% according to the standard and to the level to which the integrity is assured, such as the extent of NDT (whether 100%, spot inspection or none), joint design (whether full penetration, using backing bar, lap joint, etc). The efficiency factor is often applied to the design stress to reduce the level to which the welded joints will be loaded if there is any uncertainty in their integrity. The weld joint efficiency factors employed by certain pressure vessel codes are given below.

6-10

www.twitraining.com

Rev 2 April 2011 Design of Pressure Equipment Copyright  TWI Ltd 2012

Code

Weld and/or Level of inspection

Weld efficiency factor

EN 13445

100% NDT

100%

No NDT

70%

Full penetrated butt-joint, full RT

100%

Butt joint with backing strip, full RT

90%

Single full filet lap joint, no RT

50%

ASME VII Div 1

PD 5500

6.9

not applied

Pressure stresses The main loading which most pressure equipment will be subject to is internal pressure. Some simple equations for the calculation of the resulting generated stresses are given in the following sections. The equations given by standards such as EN 13445, PD 5500 and ASME VIII are derived from those given below and take into consideration design stresses (also called nominal design stress or allowable stress in different standards) and weld joint efficiency. When a closed ended cylinder is subjected to internal pressure, the tendency of the process fluid to expand in both the radial and axial directions in the cylinder generates stress in the shell in both circumferential and axial directions. These two stress components are called hoop stress and axial stress, respectively. They are the two major stress components that need to be considered in pressure vessel design, as they represent the main driving force for the propagation of longitudinal defects in a seam weld (ie defect parallel to the cylinder axis) or in a circumferential weld (ie defect transverse to the cylinder axis), respectively.

6.9.1

Hoop stress To calculate the hoop stress, a section of the cylinder is cut across a diameter and the force due to internal pressure acting radially outwards (Fy) must be in equilibrium with the Force at the cross section F’y due to hoop stress, σy.

L

Figure 6.13 Equilibrium section of cylinder.

6-11

www.twitraining.com

Rev 2 April 2011 Design of Pressure Equipment Copyright  TWI Ltd 2012

Considering a section of length, L, the total vertical force due to the internal pressure acting on half the cylinder Fy equal to: Fy'  2 P r L [6-1] The internal force in the vessel wall needed to balance this force for equilibrium, F’y, which is supplied by the resulting hoop stress, is equal to: Fx'  2 L t  y

[6-2]

Thus we can set these to forces equal to one another to find a value for the hoop stress σy.

L r P 2

t L 2

y

 

[6-3]

Therefore, the hoop stress is given by:

r P t

y

 

. [6-4]

This is also the same as the membrane stress in a spherical pressure vessel. 6.9.2

Axial stress The section of a thin walled cylinder shown in Figure 6.14 is in axial equilibrium simply due to the balance of the axial force and the axial stress established within the material.

Figure 6.14 Equilibrium section of cylinder.

The axial force exerted on the shell, Fx, as a results of the internal pressure, p, can be calculated by the following equation.

Fx  p  π r 2

[6-5]

6-12

www.twitraining.com

Rev 2 April 2011 Design of Pressure Equipment Copyright  TWI Ltd 2012

where r is the radius of the pressure vessel. The cross sectional area of the vessel shell wall, A, is approximately equal to: A  2  r t [6-6]

Thus the resulting opposing axial force within the shell, F’x, as a result of the axial stress, x, is equal to:

F' x   x  2  r t

[6-7]

Where t is the thickness of the cylindrical shell. Since the section is in equilibrium, the two axial forces equal to one another and thus the axial stress, σx, can be evaluated in a pressurised cylinder by equating Equation [6-5] with Equation [6-7], which leads to:

 x  2rt  P  r 2

[6-8]

The axial stress in a cylinder is therefore given by:

x  6.9.3

Pr 2t

[6-9]

Stress in a thin-walled sphere The stress in a spherical dished head can be found by considering a sectional hemisphere across any diameter and considering the equilibrium forces as before. Assume the sphere has a radius, r, a wall thickness, t and is under a pressure, p.

Figure 6.15 Hemispherical section of equilibrium dished head.

6-13

www.twitraining.com

Rev 2 April 2011 Design of Pressure Equipment Copyright  TWI Ltd 2012

The force F acting over the cross section due to the internal pressure is equal to: F  p  r 2 , [6-10]

The cross sectional area A of the vessel wall is equal to the circumference multiplied by the wall thickness: A  2  r t , [6-11]

And the reaction force due to the internal membrane stress  is equal to: F  2  r t .

[6-12]

Setting these two forces equal to one another allows us to evaluate the stress resulting from the internal pressure p:

2  r t   p  r 2 [6-13] The axial stress in a spheroidal pressure vessel is therefore given by:



pr . 2t

[6-14]

A sphere thus provides ideal closure to a cylinder since the stresses are less than those in other shapes which would have an associated hoop stress. In practice however, the high degree of mechanical forming necessary together with the necessity to change the set of die-former for each different diameter makes it impractical except for very high-pressure vessels.

6.10

Vacuum and external pressure loading The majority of pressure vessels are subject to internal pressure. However, it is important to understand what is specifically meant by the terms pressure and vacuum for vessels which are not internally pressurised: A vessel is subject to external pressure when the pressure acting on the outside surface exceeds the pressure acting on the inside surface. A vacuum condition is a specific case where the external pressure is normal atmospheric pressure and the internal pressure is less than atmospheric pressure. Examples include vessels for subsea operation, jacketed vessels and the tubes in heat exchangers.

6-14

www.twitraining.com

Rev 2 April 2011 Design of Pressure Equipment Copyright  TWI Ltd 2012

The problems associated with the design of vessels subject to external pressure are quite different from those arising from internal pressure design. The main failure mode for vessels subject to external pressure or vacuum is buckling. These vessels are generally thicker than vessels subject to internal pressure and may incorporate stiffeners inside the vessel. Shape imperfections are also much more important as they tend to increase under external pressure loading. Therefore, acceptance criteria on ovality are much tighter. In addition, internal stiffeners are used and a minimum safety factor of 1.5 is applied.

6.11

Types of pressure vessels

Figure 6.16 Pressure vessels in an oil refinery.

Pressure vessels are widely used in various industry sectors such as Oil and Gas, Petrochemical, Power Generation, etc. Generally speaking, a pressure vessel is a container designed to hold liquids or gases and store them at a different pressure to that surrounding the structure without changing volume. Fired pressure vessels are vessels that include power boilers and other vessels that are designed to accept heat. This category of component would include coal/gas and oil fired boilers as well as boilers that are heated by a waste heat gas stream or a hot oil system. Electrically heated steam generators are also considered to be fired. Steam boilers are used to generate steam by boiling water; often the steam is required for petrochemical process plant.

6-15

www.twitraining.com

Rev 2 April 2011 Design of Pressure Equipment Copyright  TWI Ltd 2012

Figure 6.17 Piping and pressure vessel columns in a chemical plant.

Fractional distillation columns are used to separate crude oil into its many components and can be as tall as 60 meters or more. The heating within the vessels is incremental, operating at about 400°C at the bottom, decreasing the as you move up the vessel. This means that the weight of the vessel rests upon the area most susceptible to creep, hence making it prone to deformation. This is exacerbated by the pressure exerted by the vessels contents. In order to provide the best resistance to creep possible, the vessels are usually made from high strength, low alloy steel such as CrMo steel. Pressure vessels for high temperature application are normally operated at temperatures up to approximately 600°C (and beyond in specific cases). Other examples of high temperature pressure vessels are nuclear reactors, steam headers and boilers.

Figure 6.18 Nuclear reactor.

Unfired pressure vessels are not in direct contact with a heating flame. These would include equipment such as pressurized tanks storing air, nitrogen ammonia or natural gas.

6-16

www.twitraining.com

Rev 2 April 2011 Design of Pressure Equipment Copyright  TWI Ltd 2012

Figure 6.19 Liquid petroleum gas (LPG) storage tank.

Double walled pressure vessels consist of an outer envelope containing a pressurised inner container. These are usually required for insulation purposes and applications include dewar vessels and cryogenic storage tanks. Separator vessels used on offshore stations which have a minimum design temperature of -50°C are an example of low temperature service applications, which operates at temperatures below ambient, but not below 75°C. Below -75°C the operation is usually classed as cryogenic. Other low temperature applications nearer to home are aerosol canisters, fire extinguishers and gas bottles. For cryogenic applications, special construction materials must be used, as many carbon steel materials become brittle at temperatures below -30°C. Generally, carbon and carbon-manganese grades are applicable down to approximately -45°C. These often have deliberate alloy additions and/or heat treatments to enhance fracture toughness at low service temperatures. In addition, fine grained melting practice improves low temperature service. For minimum service temperatures under -45°C, it is more common to select Ni-alloyed steels. Typically, 1.5% Ni steels are used down to -60°, 3% Ni steels down to approximately 100°C whilst 9% Ni steels can be used at 196°C, ie the temperature of liquid nitrogen. Sometimes stainless steel and aluminium are used as alternative materials for cryogenic pressure vessels. Reasons for failure in pressure vessels The possible reasons why failures may occur in pressure vessels can be dived into four categories. In reality it can be a combination of one or more reasons for a failure to occur. Materials Improper selection of materials for given service loading and temperature may result in overload or fracture. If there are defects in the materials such as inclusions, these could cause lamellar tearing after welding.

6-17

www.twitraining.com

Rev 2 April 2011 Design of Pressure Equipment Copyright  TWI Ltd 2012

Design Incorrect design data or inaccurate or incorrect design methods will not result in the correct design to meet the required service conditions. This may not have been identified in time is there was inadequate shop testing. Fabrication Poor quality control during fabrication, or improper or insufficient welder and welding procedure qualification can result in defective welds being present, giving a risk of fatigue, or fracture. Inadequate post weld heat treatment can leave high residual stresses or excessive hardness in the welds. If forming methods are inadequate, then the fit-up during welding can be poor and additional defects, distortion or bending stresses can result. Service It can be common for failures to occur as a result of a change of service condition by the user without proper revalidation. Inexperience operations and maintenance personnel can mean that safety devices can be overlooked and failures be more likely. Sometimes the upset conditions during operation exceed those anticipated during design, which can cause failures. A classic example of a pressure vessel failure is the John Thomposn pressure vessel which failed during hydrotest in the winter of 1965. Brittle fracture initiated from small hydrogen cracks in the HAZ of the shell to flange weld. The cause was poor toughness in the higher alloy flange steel and residual stresses due to inadequate post weld heat treatment. The failure could have been avoided by using better control during welding to avoid hydrogen cracks occurring and better control during post weld heat treatment to ensure all of the material had reached the required temperature.

Figure 6.20 The John Thompson pressure vessel after failure during hydrotest.

6-18

www.twitraining.com

Rev 2 April 2011 Design of Pressure Equipment Copyright  TWI Ltd 2012

6.12

Codes and standards Most of the international pressure vessel codes have been developed to the point where there is a high degree of technical similarity between them. Code areas such as vessel classes, design criteria and requirements for independent inspection are based on similar guiding principles. The main standards in use for pressure vessels are listed below and the design principles of some of them are summarised below. BS EN 13445 (2009): Specification for unfired fusion welded pressure vessels, BSI London. BS PD 5500 (2009): Specification for unfired fusion welded pressure vessels, BSI London. ASME (2007): Boiler and Pressure Vessel Code, Section VIII. The basic premise behind a pressure vessel code is that equipment should meet its requirements in all respects, ie full code compliance. This is not always strictly adhered to, with the result that some equipment may be specified as being code intent. This means it may comply with the code in some areas, eg design stresses, but not in others, such as the requirements for NDT and defect acceptance criteria. Equipment built to such code intent cannot be officially code stamped though.

6-19

www.twitraining.com

Rev 2 April 2011 Design of Pressure Equipment Copyright  TWI Ltd 2012

6.12.1

6.12.2

EN 13445 Latest Edition

2009

Design covered by

Part 3

Corrosion

There is some clear guidance, including a minimum corrosion allowance of 1mm (Section 5)

Design methods

Design-by-rule (Section5) Design-by-analysis (Annexes B and C) Experimental techniques (very little guidance)

Weld joint coefficient

Depend on the extent of NDE, similar to ASME VIII Div 1

Notes

EN 13445 was developed as a harmonised standard for use with the Pressure Equipment Directive (PED) and is intended to cover the essential safety requirements (ESRs) of the PED. Use of this standard is not mandatory in the PED, but vessels, designed, manufactured and tested in accordance with this standard have an automatic presumption of conformity with the ESRs of the PED.

PD 5500 Latest Edition

2009

Design covered by

Section 3

Construction categories

Three construction categories (Section 3.4)

Joint efficiencies

-

Joint types

Type A and B with different NDT requirements (Section 5.6.4)

Design features

Figures E.1 to E.6 show typical (rather than mandatory) weld details

Notes

The technical content claims to satisfy the essential safety requirements of the PED

6.12.3 ASME VIII Latest Edition

2007

Design covered by

Div 1 (most vessels) and Div 2

Construction categories

Div 2 is for specialised vessels

Joint efficiencies

UW-12

Joint types

Part UW-3 defines category A, B, C, D welded joints, which have different NDT requirements

Design features

Figures UW-13.1, 13.2 and 16.1 show typical acceptable (and some unacceptable) welded joints

Design pressure

UG-19, UG-21

Design temperature

UG-19, UG-20

Notes

Div 1 coves normal vessels, Div 2 covers alternative rules for pressure vessel and is used for various types of special applications, including nuclear.

6-20

www.twitraining.com

Rev 2 April 2011 Design of Pressure Equipment Copyright  TWI Ltd 2012

6.13

References The following references are recommended for further reading: A L Kholan, Pressure Vessel Systems C Matthews, Engineers Guide to Pressure Equipment S Earland at alk, Guide to European Presure Equipment

6.14

Summary At the end of this module you should be confident to explain the design of different weld details for welding pressure vessels and describe their particular advantages. You should be able to interpret different pressure vessel codes and calculate axial and hoop stresses in a cylindrical pressure vessel. You should be able to design structural details such as nozzles, dished heads or flanges for a pressure vessel and explain the advantages of your chosen design.

6-21

www.twitraining.com

Rev 2 April 2011 Design of Pressure Equipment Copyright  TWI Ltd 2012

Revision Question 1 You are designing a 3 metre diameter cylindrical pressure vessel which will be supported horizontally. The vessel requires a manway access as well as connections to 2inch diameter pipework for both the incoming and outflowing process fluids. The vessel shell is made from 30mm thick steel plate. Sketch your pressure vessel, showing the structural details and the weld joint preparations you intend to use, explaining the reasons for your choices.

6-23

www.twitraining.com

Section 7 Advanced Fatigue – Part 2

Rev 2 April 2011 Advanced Fatigue – Part 2 Copyright  TWI Ltd 2012

7

Advanced Fatigue - Part 2

7.1

Introduction In Part 1 of this session on fatigue, we covered the practical issued of fatigue. In this second part, we take a close look at the theory behind fatigue assessment. Fatigue occurs by the steady, progressive growth of a crack through a cyclically loaded member and at right angles to the direction of applied stress. Typical fatigue cyclic loading is shown in Figure 1. Fatigue design using S-N curves acknowledges this, but does not provide any information whatsoever about the crack growth process. This session will cover construction of S-N curves for parent metals and welds and also describe the rate of crack extension per cycle in fracture mechanics terms. It will also show techniques for calculating how many cycles are required to cause a fatigue crack to extend by a given amount and how to count real-life fatigue cycles using the rain flow method.

Figure 7.1 Cyclic stress cycle, showing terms and definitions.

7.2

S-N curves If applied stress range, σ, is plotted against number of applications of load required for failure (N) we obtain the S-N curve (Figure 7.2). Increasing the stress for a given number of cycles, or increasing the number of cycles at a fixed stress range plots a point above the S-N curve, which predicts fatigue failure (Figure 7.3). There is a stress range below which fatigue will not occur, called the endurance limit. Typical S-N curves are plotted on logarithmic scales, which give a linear relationship in the high cycle regime (Figure 4). S-N curves are based on statistical analysis of fatigue test data obtained from tests on welded joints, to produce the best fit mean S-N curve by the method of least squares. A straight line relationship between log S and log N was assumed, where S = stress range and N = fatigue life. The resulting curves have the form S m N = A, where m is the slope and A is a constant. The final design curves are set below the mean S-N curves, usually two standard deviations of log N to give approximately 97.5% probability of survival.

7-1

www.twitraining.com

Rev 2 April 2011 Advanced Fatigue – Part 2 Copyright  TWI Ltd 2012

Figure 7.2 An S-N curve plotted on normal axes.

Figure 7.3 Effect of higher stress or larger number of cycles on the predicted fatigue failure (above the S-N curve).

Figure 7.4 S-N curve plotted on logarithmic axes, showing a linear relation in the high cycle regime.

The position of the S-N curve is affected by stress concentrations and other features. A simple notch will reduce the S-N curve (Figure 7.5). Welded structures have intrinsic sharp stress concentrations at the intrusions at weld toes and show much lower fatigue resistance than plain or notched material (Figure 7.6). Welded structures effectively do not have any initiation period for fatigue and will propagate fatigue from the weld toes if exposed to

7-2

www.twitraining.com

Rev 2 April 2011 Advanced Fatigue – Part 2 Copyright  TWI Ltd 2012

susceptible loading conditions. For components which are subjected to fluctuating loading in service, the avoidance of fatigue is likely to be the factor which limits the design stresses. This is particularly true for welded components where their fatigue strengths can be much less than those of unwelded components.

Figure 7.5 The effect of a notch on the fatigue S-N curve.

Figure 7.6 The effect of welding on reducing the S-N curve.

One of the most serious consequences of the fact that the fatigue lives of welded joints are dominated by crack growth concerns the influence of material strength. Although the fatigue strength of un-notched material usually increases with tensile strength, the level of increase decreases if the material contains a notch until there is no increase at all for welded material (Figure 7.7). This is because rate of fatigue crack growth is not dependent on material strength and hence welded low and high strength materials give the same fatigue life. The benefit of material strength comes in the crack initiation stage, which is effectively absent in the welded material. For a structure under fatigue loading, using higher strength material, whereby a thinner section can be used under higher stress, is a real mistake. Not only does the higher stress level result in fewer cycles to failure, the thinner

7-3

www.twitraining.com

Rev 2 April 2011 Advanced Fatigue – Part 2 Copyright  TWI Ltd 2012

section takes less time to fatigue crack though. A thicker member under lower stress will have a better fatigue life.

Figure 7.7 The change in S-N curve with steel strength for parent material, notched metal and welds.

7.3

Fracture mechanics approach to fatigue When trying to determine the rate that a fatigue crack will grow under applied cyclic stress conditions, it is necessary to use a fracture mechanics approach, to understand the behaviour of the fatigue crack in the structure. Fracture mechanics is concerned with cracks and how they behave under load. In the context of fatigue, fracture mechanics is used mainly to describe the rate at which a crack will grow under repeated loading and for determining the extent of fatigue cracking which can be tolerated before failure occurs. For a crack of length a loaded under a static stress , the stress intensity factor K is given by:

a

Y

K

  

Where Y is a geometric term.

Figure 7.8 Stress distribution close to a crack tip.

7-4

www.twitraining.com

Rev 2 April 2011 Advanced Fatigue – Part 2 Copyright  TWI Ltd 2012

K defines the magnitude of the crack tip singularity (stresses near tip are proportional to 1/√r and K) and it defines the crack tip conditions under linear elastic conditions: if K is known, stresses, strains and displacements can be determined. If a material fails locally at some critical combination of stresses and strains, this corresponds to a critical value of K, which is a measure of fracture toughness. Fracture mechanics principles are applicable to fatigue when a crack or a crack-like flaw is present eg fatigue cracks, small inherent crack-like flaws at the toes of welds, or planar flaws eg lack of fusion, incomplete penetrations, hydrogen cracking. For cyclic loading where the stress range is , the stress intensity factor range is given by K

a

Y

K

   

The stress intensity factor varies with time in the same way that the stress varies with time, Figure 7.9.

Figure 7.9 Variation of stress intensity factor with time.

A graph is then drawn of rate of crack extension per cycle (da/dN) as a function of K, Figure 7.10. The stress intensity factor range, K is Kmax – Kmin. The stress ratio, R is Kmin/Kmax. Fracture

Relationship between da/dN and K determined experimentally

increasing R

Log (da/dN) m 1

da Paris law :

= A (K)m

dN increasing R threshold, Ko

Log (K)

Figure 7.10 Relationship between crack extension per cycle and stress intensity factor ranges.

7-5

www.twitraining.com

Rev 2 April 2011 Advanced Fatigue – Part 2 Copyright  TWI Ltd 2012

When drawn on logarithmic axes, there is a linear relationship between crack extension per cycle and the stress intensity factor range, over a wide interval of stress intensity. This relationship can be expressed as:

 

m

K

A

a N d d



Where A and m are material constants. This is known as the Paris equation. There is no influence of mean stress on crack growth rate in this central region. At low values of K, the rate of crack extension per cycle reduces rapidly and effectively becomes zero. There is therefore a threshold stress intensity factor range (K0) below which fatigue crack growth does not occur. This threshold occurs at lower values of K as the mean stress is increased. The Paris equation can be used to establish the rate at which a crack of a given length will grow when subjected to a specific stress range. Simple mathematical techniques based on integrating the Paris equation are available to determine the number of cycles required for a crack to grow from a given size to a greater limiting size. This approach forms the basis for fitness-for-purpose assessment of cracks or planar weld flaws subjected to fatigue loading.

The basis of fatigue life calculations by fracture mechanics is integration of the crack growth law. This leads to the S-N curve for the flawed component under consideration, as a function of initial flaw and final crack sizes (ai and af), geometry (Y) and material (A and m). For a given situation, all the terms on the left hand side of the equation are approximately constant, so that the integration predicts a linear S-N curve of the form ∆σmN = constant, where the slope m is the same as that in the Paris law. This partly explains why most of the S-N curves for welded joints are parallel with a slope corresponding to m = 3, the value which most fatigue crack growth data approximately conform to. The flexibility of the fracture mechanics approach is such that any of the four main variables, ai, amax, ∆s and N can be calculated if the other three are known. For example, if N is unknown it is possible to calculate the life of a known flaw under specified loading or inspection periods based on detectable crack amax. If flaw size ai is unknown you can estimate the tolerable flaw size to satisfy the required S-N curve; if amax is unknown you can estimate the required toughness to avoid unstable fracture from fatigue

7-6

www.twitraining.com

Rev 2 April 2011 Advanced Fatigue – Part 2 Copyright  TWI Ltd 2012

cracking; and finally if ∆s unknown it’s possible to calculate the allowable stress for known life and flaw size. In most metals, fatigue crack propagation progresses in regular steps to produce striations on the fracture surface (as shown for an aluminium alloy in Figure 7.11a). In such cases, da/dN depends mainly on K. However, this striation formation mechanism may be accompanied by other fracture mechanisms in some microstructures. Figure 7.11b shows intergranular fracture in a hard HAZ in steel. Similarly, striations may be accompanied by cleavage or microvoid coalescence. In such cases, da/dN is higher and it is also dependant on Kmax; the threshold value of K may also be reduced. In general, microstructures which give rise to these static failure modes are rare.

a

b

Figure 7.11 Fatigue fractures as seen under the scanning electron microscope; a fatigue striations in an aluminium alloy; b intergranular fracture in a hard HAZ in steel.

a

b

Figure 7.12 Experimental validation of fracture mechanics fatigue calculations.

7-7

www.twitraining.com

Rev 2 April 2011 Advanced Fatigue – Part 2 Copyright  TWI Ltd 2012

Limited experimental results are available to validate the fracture mechanics approach to fatigue life prediction. Figure 7.12a shows an example for an aluminium alloy plate with a fillet welded attachment. The actual progress of the weld toe fatigue crack was determined from beach marks on the fracture surface after the test and the prediction made use of crack growth data obtained from the actual alloy. Less precision must be expected when use is made of the recommended material properties in BS 7910. Figure 7.12b refers to fatigue life predictions for fillet weld toe cracks in steel which have propagated from either hydrogen cracks or small fatigue cracks of known dimensions. As will be seen, use of the crack growth law for steel recommended in PD6493 (the forerunner to BS 7910) seriously underestimated the fatigue lives of the flawed welded joints. This reflects the fact that the recommended laws are based on upper-bounds to published crack growth data, for safety. In contrast, use of a crack growth law obtained by testing the actual steel from which the welded joints were made results in much more accurate estimates of fatigue life. Clearly, if it is expected that a new structure will require some evaluation of the significance of welding flaws or fatigue cracks in service, it is prudent to reserve some material to establish the actual fatigue crack growth characteristics.

Figure 7.13 Fatigue crack growth laws in BS 7910.

The graphs in Figure 7.13 summarise the mean + 2SD 2-stage crack growth laws recommended in BS 7910, all for R ≥ 0.5. Other curves are provided for R < 0.5. In addition, conservative single Paris laws are provided for convenience. Special attention is paid to environmental influences and the

7-8

www.twitraining.com

Rev 2 April 2011 Advanced Fatigue – Part 2 Copyright  TWI Ltd 2012

recommendations cover marine corrosion with and without cathodic protection (as shown) and fatigue crack growth at elevated temperature. Although the main focus of attention is steel, guidance is also given on fatigue crack growth in other metals. Free corrosion produces around 3-fold increase in da/dN compared with air; use of cathodic protection (correct or over-protection) restores in-air behaviour up to K = 144 N/mm3/2 (315 N/mm3/2 for R < 0.5), but da/dN increases at higher values. Cathodic over-protection (-1100 mV) is particularly detrimental at high K and K max due to the generation of hydrogen. The majority of the raw data were generated to reproduce North Sea conditions, approx. 6-10oC and a cycling frequency of 1/6 Hz. Crack growth rates could be higher at higher temperatures and lower frequencies and vice versa.

7.4

Variable amplitude loading Real structures in service do not experience the kind of constant amplitude stress cycling that is shown in Figure 7.1, they may experience a continually fluctuating variable amplitude stress history, more similar to that shown in Figure 7.14. Variable amplitude stresses in a structure can arise because the loads to which the structure is subjected vary, or due to the fact that the structure is subjected to several different forms of loading and/or loading in different planes. For example a bridge may experience traffic which includes trucks of various sizes, buses and small cars all resulting in different loads. Offshore structures are subjected to a wide range of waves and wave heights with wind of different intensities and from different directions resulting in a variable loading spectrum. Figure 7.14 presents strain gauge records from two different structures subjected to variable amplitude loading.

7-9

www.twitraining.com

Rev 2 April 2011 Advanced Fatigue – Part 2 Copyright  TWI Ltd 2012

Figure 7.14 Variable amplitude stress loading.

7.5

Cycle counting A stress distribution such as this needs to be separated into blocks of cycles with the same amplitude in order for its damaging effect with respect to fatigue to be calculated. The process of defining the service stress spectrum involves establishing the service load history (eg as specified, measured, estimated, etc), calculating corresponding stresses (eg by FEA, direct measurement, etc) and then converting the stress sequence to blocks of identifiable stress range cycles using a cycle counting method (Figure 7.15).

Figure 7.15 Objective of cycle counting to convert a variable amplitude stress cycle into blocks of common stress range.

7-10

www.twitraining.com

Rev 2 April 2011 Advanced Fatigue – Part 2 Copyright  TWI Ltd 2012

For variable amplitude loading it is not straightforward to decide what constitutes a cycle. Figure 7.16 shows a constant and a variable loading sequence. For the constant amplitude sequence there are two cycles ABCDE (cycle one) and EFGHI (cycle two). For the variable amplitude there are a number of options:  

ABCDEFGHI (cycle one), IJKLM (cycle two), MNOPQRS (cycle three). ABCD (cycle one), DEFGH (cycle two), HIJKLM (cycle three), MNOP (cycle four), PQRS (cycle five).

Figure 7.16 Sections of constant and variable amplitude loading illustrating the problem of cycle counting.

For the variable amplitude sequence there are many other options to divide the cycles and each different cycle count if used in a design calculation will give a different result. There is no ‘correct’ way of counting cycles in variable sequences but there are several standard methods which can be used. Two methods (rainflow and reservoir) are extensively used as they give similar results and produce the most severe count from the fatigue point of view. The rainflow method gets its name from the analogy with the flow of raindrops down a pagoda roof. The first step is to redraw the stress as a sawtooth pattern with the magnitudes of the peaks and troughs unaltered. To simplify the peaks and troughs have been labelled with letters. The stress sequence is reproduced with time as the vertical axis and the starting point at the top, Figure 7.17. As a convention the trace will begin and end at the highest peak in two successive occurrences. The waveform is broken into a series of paths which raindrops can follow, the beginning and end of each path defining the extremities of a half cycle. Raindrops can flow both from right to left and from left to right and a path or a combination of paths must cover the top of every roof.

7-11

www.twitraining.com

Rev 2 April 2011 Advanced Fatigue – Part 2 Copyright  TWI Ltd 2012

The count will extend from the first occurrence of the highest peak J to the corresponding peak J’ the second occurrence of the same event. The raindrop starts at J and is allowed to run down until a peak or trough is reached, when the rain falls vertically until it reaches another roof, the process is repeated until one of the two conditions is satisfied: 1 For rain falling down a roof, the path cannot cross paths with rain which is falling from above (examples in Figure 7.17 LM, E’F’ and G’H’) 2 For a raindrop starting at a trough and falling vertically from a peak the flow stops if the drop passes opposite a trough which is more negative than that at the start of the path under consideration (drops falling from L for which trough N is more negative than K). Similarly for a drop starting from a peak and falling vertically from a trough the flow stops if if the drop passes opposite a peak which is more positive than the one at the start (drops falling from G’, for which peak J’ is more positive than C’). Each complete flow path (JKMN, KL, LM, C’D’F’G’ Figure 7.14) is considered a half cycle and half cycles of equal stress range are combined to give complete cycles. Small interruptions in the flow path such as KLM result in complete cycles, ie KL combined with LM but a cut-off value in the magnitude of the stress may be imposed for small interruptions.

Figure 7.16 Rainflow method for cycle counting.

7-12

www.twitraining.com

Rev 2 April 2011 Advanced Fatigue – Part 2 Copyright  TWI Ltd 2012

Reservoir counting involves counting whole cycles rather than half cycles for the rainflow method, but produces the same result. The stress history is taken to represent the cross section of a reservoir, which is drained successively from each low point (trough) counting one cycle for each draining operation. The magnitude of the corresponding cycle is equivalent to the depth of the water drained. Figure 17 represents the same stress sequence as in Figure 16, but counted using the reservoir method.

Figure 7.17 Reservoir method for cycle counting.

The method follows: 1 The reservoir is bounded by the two highest peaks (JJ’) and a line is drawn representing the surface level. 2 Drain the reservoir from the lowest trough (N) and note the height of the original water level (to N) which represents the range (260). This results in water being trapped behind other peaks L, O, U and C’ 3 Step two is repeated for the remaining bodies of water until all water is drained, listing each cycle with each drainage operation 4 Compile a list of the number of cycles of each magnitude in descending order which represents the summary of the loading spectrum. If there are two or more equal lowest points (such as P and R) it makes no difference which one is used as the initial drainage point, the final answer will not depend of the choice but will be the same whichever is chosen first. Once the fatigue spectrum can be separated into sets of cycles with the same stress range, Miners rule of linear cumulative damage is then used to assess the contribution of each set to the total fatigue damage.

7-13

www.twitraining.com

Rev 2 April 2011 Advanced Fatigue – Part 2 Copyright  TWI Ltd 2012

7.6

Miner’s rule Miner’s rule is the established method of relating the fatigue behaviour under constant and variable amplitude loading. The basic assumption that the fatigue damage caused by a particular stress range applied n times is directly proportional to the ratio n/N, where N is the fatigue life expected at the same stress range (from the S-N curve, see Figure 7.18). Then, the sum of the contributions due to all the stress ranges in a stress history will be unity at failure. The Mimer’s rule method takes into account the damage the cycles at various amplitudes have on the overall life of a component. Putting it another way, the design criterion is:

3

2

1



n 3N

n 2N

n 1N



 < 1 for design

where: n1 = number of cycles applied at stress range 1 N1 = number of cycles allowed at stress range 1 This is sometimes also expressed as: Σn/N_(i )

1 at failure

where i = 1,2,3 etc, should be less than 1 at the end of the service life. Often a safety factor is applied such that the allowable fatigue damage is limited to less than 1. Miner’s rule was originally developed for considering fatigue of unwelded parts, where fatigue crack initiation dominated the life. In such circumstances, the implicit assumption that stresses below the fatigue endurance limit are non-damaging (see Figure 7.18) is reasonable. However, fatigue cracks initiate early in the lives of welded joints under stresses in the stress history which exceed the fatigue limit. The result is that the fatigue limit starts to decrease and more and more stresses below the original limit becoming damaging as the life progresses.

Figure 7.18 Miner’s rule for fatigue damage summation.

7-14

www.twitraining.com

Rev 2 April 2011 Advanced Fatigue – Part 2 Copyright  TWI Ltd 2012

7.7

Fatigue assessment The procedures in BS 7910 can be used to calculate the tolerable fatigue life of a known flaw, or the critical flaw size for a given fatigue life. To carry out this type of fatigue assessment there are three sets of input parameters: The fatigue stress spectrum The flaw type and The fatigue crack growth collated into blocks of same geometry relationship stress range.

The flaw type and the applied stress range  contribute to the determination of the stress intensity range K. This, in conjunction with the fatigue crack growth relationship allows integration of the crack growth cycle by cycle to determine final flaw size. This final flaw size can then be evaluated in a fracture assessment. The type and geometry of the flaw will have a significant effect on its fatigue performance. The parameters to be considered include whether the flaw is is a surface flaw or a buried defect, the flaw height, length and ligament height (Figure 7.19). The plate thickness also needs to be known. In a fatigue assessment, all flaws are assumed to be planar, which can be overconservative for volumetric defects such as slag inclusions. If a flaw is not aligned perpendicular to the principal applied stress, then the flaw dimension is taken as the resolved flaw dimension perpendicular to the applied stress.

Figure 7.19 Fatigue flaw dimensions to use as inputs into a fatigue assessment.

BS 7910 includes procedures for dealing with groups of flaws. If flaws are close to each other, they may interact (ie the stress intensity factor for the flaw is increased by close proximity of other flaws). BS 7910 gives flaw interaction rules, where pairs of flaws are considered. However, it is now recognised that these are not applicable in a fatigue assessment. Multiple fatigue cracks behave independently until they actually coalesce. Therefore, the flaw interaction rules required in a fracture assessment are not applied in a fatigue assessment.

7-15

www.twitraining.com

Rev 2 April 2011 Advanced Fatigue – Part 2 Copyright  TWI Ltd 2012

Figure 7.20 Flaw interaction rules for fracture are not applicable to fatigue assessments.

The stresses used as input to a fatigue assessment are the primary and secondary (excluding residual) stress ranges due to fluctuating components of applied load, allowing for stress concentration factor kt due to any gross structural discontinuity, if necessary. kt (P + Q) The through-thickness stress distribution is linearised (but do not underestimate surface stresses) to deduce m and b, the membrane and bending stress ranges. All stresses input to BS 7910 are based on stress analysis of the structure in the absence of the flaw. The presence of the flaw is accounted for by the stress intensity factor solution. Stress input can be obtained in various ways, including analytical or handbook solutions, finite element analysis, or experimentally (eg strain gauges).

Figure 7.21 stress linearization of the actual stress distribution.

The applied stress intensity range is proportional to the stress range and the size of the flaw, ΔK YΔσ √πa. There is also a significant influence of geometrical factors which makes the calculation more complex:

7-16

www.twitraining.com

Rev 2 April 2011 Advanced Fatigue – Part 2 Copyright  TWI Ltd 2012

The general procedure in BS 7910 involves estimation of the rate of growth of a flaw under fatigue loading, in the limit considering the increment of growth per applied stress cycle. Use is made of a relationship between the rate of crack growth (da/dN) and the fracture mechanics stress intensity factor parameter, expressed in terms of its range, K. Guidance is given in BS 7910 on this relationship, with special attention to fatigue crack growth in welded structures.

Figure 7.22 Incremental crack growth assessed using the Paris Law.

It is usually necessary to use software to perform the cycle by cycle integration calculations, which may need to be combined with a fracture assessment if there is a risk of failure by unstable fracture or yielding. Alternative failure criteria include leakage (ie through-thickness cracking) or the attainment of a detectable crack. The stages in a fracture mechanics fatigue analysis are shown in Figure 7.23 for an embedded flaw. This will grow under fatigue loading provided the applied stress intensity factor range is above the threshold value. Growth will proceed until fracture or some other ‘failure’ criterion is reached or the embedded crack breaks through to become a surface crack. The assessment then re-characterises the flaw and this surface crack will grow in turn until the critical size is reached, or it becomes a through-thickness crack. This crack will grow in turn until the conditions are such that failure occurs, so the instability conditions for fracture must be checked. For pressure containing equipment the development of a through-thickness crack may be unacceptable, due to leakage.

7-17

www.twitraining.com

Rev 2 April 2011 Advanced Fatigue – Part 2 Copyright  TWI Ltd 2012

If the conditions are such that fracture is a possibility before the fatigue crack breaks through to the surface, then additional allowances must be included on flaw size to account for dynamic snap-through effects. Provided failure conditions are not reached during the required service life of the component, the crack can be considered to be acceptable. The calculations can also be used to define appropriate inspection strategies for detecting and monitoring fatigue crack growth.

Figure 7.23 The re-characterisation of an embedded flaw as it grows under fatigue to become surface breaking and again when it growth to become throughthickness.

7.8

Summary At the end of this module you should understand how to interpret an S-N diagram and how the S-N curves are affected by factors such as joint design or tensile strength. You should be able to explain how to carry out a fatigue assessment on a variable amplitude stress spectrum, including methods for cycle counting and Miner’s rule. You should also understand how fatigue cracking is assessed using fracture mechanics methods to determine the crack growth over a given number of cycles.

7-18

www.twitraining.com

Rev 2 April 2011 Advanced Fatigue – Part 2 Copyright  TWI Ltd 2012

Revision questions 1 Sketch a graph of S-N curves for plain material, notched material and welded material.

2 What are cycle counting methods for? Name two methods.

3 What is Miner’s rule?

4 What input data is needed for a fracture mechanics fatigue assessment, in addition to the inputs required for a fracture assessment?

5 What might be the failure criteria for an embedded flaw growing by fatigue?

7-19

www.twitraining.com

Section 8 Stresses in Welds

Rev 2 April 2011 Stresses in Welds Copyright  TWI Ltd 2012

8

Stresses in Welds

8.1

Why calculate weld stress? Most products are made from more than one piece of material; most products, therefore, need joints to hold the pieces together. The cost of assembling and making joints is frequently the major part of the total production cost. More products fail from inadequate joints than any other cause. Many products contain hundreds or thousands of joints. The cost of disassembly and repair of one failed joint can often exceed the cost of an entire new replacement product. Inadequate joint design costs the world economy hundreds of billions of $/£ each year.

8.2

Why joints often fail Many joints are designed on the basis of experience of previous service performance and this can mean that design improvements are often made by trial and error. But trial and error is becoming less acceptable as customers' reliability expectations increase. It is necessary to predict the performance of the joint, using calculations, prototypes, testing and modelling. The variety of materials, product forms, joining processes and service conditions in today's hi-tech world is vast. The choice of possible design situations is astronomical. Scientific testing and design rules only cover a small fraction of the possible design situations. If a designer is to consider solutions outside his/her immediate range of experience, it is necessary to be aware of the general principles of joint design and manufacturing quality problems The main causes of design failures are therefore:    

Lack of awareness of one or more important factors which need to be considered in the design process. Lack of appropriate design data. Lack of resources to conduct the necessary testing. Lack of consultation with production personnel.

8.3

Load-bearing welds

8.3.1

Approach to calculating stresses When a calculation of the stress in a weld needs to be made, there are four steps to be taken. 1 Draw a diagram! Make sure you have a large sketch of the joint design and the weld dimensions. Label the information you have, such as component size, leg length or throat size, plate thickness, applied load etc.

8-1

www.twitraining.com

Rev 2 April 2011 Stresses in Welds Copyright  TWI Ltd 2012

2 Calculate the weld cross section area. Is the load carried by just one weld, or are there two welds (eg in a lap joint or double sided fillet weld). What is the design throat (not the actual throat)? 3 Convert the applied forces into a stress. What are the units of the forces or loads you have been given? What units will you need for the calculation? Is it an axial stresses or shear stress? Does more than one stress need to be combined into an effective stress? 4 Compare the stress to the allowable value. What is the design stress? Do you need to determine it from the parent metal yield strength? 8.3.2

Static strength The joint must be designed to withstand the highest service force likely to be experienced in the design life of the product. If the parent materials are stressed close to their own mechanical limits, the joint will need to be full strength, ie at least as strong as the weakest parent member being joined. The static stress limit is never the material’s yield strength, but is often ⅔ of the yield strength, so that there is a sufficient margin in the design. Conversely, if the members at the joint position are always stressed well below their potential static strength, the joint can be designed to be partial strength. A joint that is designed to be able to withstand ⅔ of the parent metal yield strength, but which only ever experiences, say, a quarter of the yield stress is over-engineered and there may be a case to use less welding, less consumable and save unnecessary costs. The most onerous mode of stressing on a joint is, generally, tension. This will arise whenever the members joined are subjected to axial tension or bending. The most common modes of failure are shown in Figures 8.1-8.3. It is good practice to ensure that welded components fail in the parent material rather than in weld metal, by using an over-matching weld metal.

Figure 8.1 Possible locations of failure in fillet weld under static loads.

8-2

www.twitraining.com

Rev 2 April 2011 Stresses in Welds Copyright  TWI Ltd 2012

In-line butt weld – full penetration either: Weld throat

or: HAZ or parent material

In-line butt weld – partial penetration

Weld throat

Figure 8.2 Possible locations of failure in butt weld under static loads. It is good practice to ensure that welded components fail in the parent material rather than in weld metal. Single fillet weld – lap joint

Weld throat

Double fillet weld – lap joint either:

Weld throat or:

HAZ or parent material

Figure 8.3 Possible locations of failure in lap welds under static loads.

8-3

www.twitraining.com

Rev 2 April 2011 Stresses in Welds Copyright  TWI Ltd 2012

8.3.3

Fatigue Fatigue will only occur if there are a sufficient number of stress fluctuations of a sufficient magnitude during the design life. Fatigue is covered in more detail in the lectures on the dynamic loading of structures. Fatigue involves the initiation of cracks at points of severe stress concentration, which then propagate under cyclic stressing until there is insufficient cross-section left to carry the static forces on the joint. The art of fatigue design is to detail the joint so as to avoid severe stress concentrations. Figures 8.12-8.15 shows fatigue initiation sites and directions of crack propagation for commonly-used joint types. It also shows how the fatigue and static strengths of joints compare. This comparison is a clear reminder that design should not only be based upon the material strength under static conditions of loading. Full penetration butt weld (no flaws)

Toe crack

Figure 8.4 Fatigue crack propagation from weld toes under cyclic loads.

8.4

Strength calculations for welds

8.4.1

Fillet welds The static strength of butt welds is generally straightforward, since a full penetration butt weld has the same load bearing capacity as the parent plate. The strength of fillet welds is more complicated because of the combination of both shear and axial stresses in the weld.

8-4

www.twitraining.com

Rev 2 April 2011 Stresses in Welds Copyright  TWI Ltd 2012

Fillet weld (unpenetrated) Toe

Nominal throat `a´ Z2 Toe Root Z1 leg length

Figure 8.5 Geometry of a fillet weld.

The relative magnitude of shear and axial stresses in the weld is determined by both the geometry of the component and the direction and position of loading. Figure 6 shows the orientations of the axial and shear stresses in a fillet weld. It is possible that two types of shear might exist in a fillet weld as shown in Figure 8.6. The shear stresses and the axial stress are all shown acting on the weld throat.

Figure 8.6 Definition of stresses on the weld throat of typical fillet welds.

The overall strength of a fillet weld is usually determined by the calculation of the average shear stress () and axial stress () in the weld throat and the conversion of these stresses into an effective stress. The effective stress is then compared to the weld strength to determine the safety of the joint. The effective stress in a weld is a complex function of the way in which the weld is loaded. Failure is assumed to occur when the effective stress in the weld

8-5

www.twitraining.com

Rev 2 April 2011 Stresses in Welds Copyright  TWI Ltd 2012

reaches a critical value. There are a number of methods to combine all the stress components to calculate the effective stress. A simple, but conservative, formula for when there is one axial stress and one shear stress is given below. 2

2

8.4.2

  

e v i t c e f f e



IIW method This method assumes that, as a result of a load, three different stresses occur in a fillet weld:   

Tensile stress acting perpendicular to the throat . Shear stress along the throat of the weld . Shear stress along the longitudinal axis of the weld II.

Figure 8.7 Stresses occurring in a fillet weld under load according to the IIW formula.

The value of these stresses can be calculated by resolving the load P onto these three axes. The stresses are calculated by dividing these load components to the weld’s cross section area. Let’s consider the load component acting perpendicular to the throat as Ptension, the load component acting along the throat as P and the load component acting along the longitudinal axis of the weld as PII. If the weld throat is t and the weld length is L, the stresses produced by these load components are:

 

Ptension Lt

 

P Lt

II 

PII Lt

8-6

www.twitraining.com

Rev 2 April 2011 Stresses in Welds Copyright  TWI Ltd 2012

These stresses are further combined to give an equivalent stress using the following formula:

 eq    2  3( II2   2 )   all In the above formula,  is a coefficient that takes into account the type of material. It has a value of 0.7 for carbon steels and 0.85 for high strength low alloy steels. In order to avoid failure, this equivalent stress must not exceed the allowable tensile stress all. Since this method is more refined than the previous simple one and a higher applied load can be tolerated, sometimes this method gives smaller weld sizes. The method for calculating stresses in fillet welds is also applicable to fillet welds in lap joints too. 8.4.3

Over-welding fillet welds Once the size of the fillet weld throat has been calculated it is important not to then over-weld in practise. This weld metal is not needed to carry the load and even a small increase in leg length can have a large effect on the amount extra (unnecessary) weld metal deposited and therefore the additional unnecessary cost. Take this example of a fillet weld deposited using the correct leg length of 4mm (Figure 8a) and where the weld has been over-welded and a 6mm leg length has been deposited (Figure 8.8b). What is the difference between the cross section area (CSA) and hence the relative volume of weld metal, for these two welds?

Figure8. 8 Over-welded fillet weld b containing more weld metal than necessary.

The welds are assumed to be mitre fillet welds, with no excess weld metal. Weld a CSA = 4 x 4 = 8 mm2 Weld b CSA = 6 x 6 = 18 mm2 2 2 The cross section area of weld b is over double that of weld a. Twice as much weld metal has been deposited than is needed, with only a 2mm increase in leg length.

8-7

www.twitraining.com

Rev 2 April 2011 Stresses in Welds Copyright  TWI Ltd 2012

8.4.4

Butt welds Calculating stresses in butt welds is much simpler than fillet welds since they do not generally experience shear stresses. Calculating the axial stress in a butt weld requires only that the weld cross section area is known.

Figure 8.5 Full penetration butt weld dimensions. The excess weld metal at the cap and root is ignored. For partial penetration welds, the cross section area is reduced by the unfused ligament and is calculated using the nominal throat of the partial penetration weld, or the depth of penetration below the plate surface.

Figure 8.6 Partial penetration butt weld dimensions.

8-8

www.twitraining.com

Rev 2 April 2011 Stresses in Welds Copyright  TWI Ltd 2012

8.5

Overview of weld design The most important dimension for the weld designer to specify is the nominal throat: S for butt welds and a for fillet welds. This is defined as the shortest distance between the root and a straight line drawn between the toes. For a butt weld, the throat is equal to the penetration depth, either'full or partial. The root is the position of furthest penetration into the joint. The position of the root, eg which side of the joint and whether backing material is to be used may affect the static and fatigue performance of the joint. For a fillet weld, the throat may be specified by the dimension a or by specifying the leg length (dimension Z (or Z 1 and Z 2 if different), see Figures 8.7 and 8.8. The preparation is the shape of the joint faces needed to ensure that the necessary access for weld metal deposition is available. The fit-up is the dimensional relationship between the two sides of the joint (eg root gap, alignment, etc). The selection of preparation and fit-up dimensions, together with their tolerances, will depend on the welding process and procedure being used and have a considerable influence on the economy, quality and distortion of the joint. The responsibilities (in chronological sequence) should be:   

8.6

Designer to determine weld type, penetration, size, material strengths and inspection criteria. Welding Engineer to determine preparation, fit-up, tolerances and processes. Designer to incorporate all above information on the fabrication drawings.

Summary At the end of this module you should be confident to determine weld cross section areas. You should understand the difference between axial and shear stresses and between equivalent and design stresses. You should be able to combine stresses (such as axial and shear stress) in calculations of weld stresses.

8-9

www.twitraining.com

Rev 2 April 2011 Stresses in Welds Copyright  TWI Ltd 2012

Revision Questions Remember to draw a sketch and check your units! 1 What is the cross section area of a full penetration butt weld in a 4m wide plate of 12mm thick steel? 2 The plate has a yield strength of 320MPa. What will you assume the design strength is? 3 Therefore, what is the maximum load that this weld can carry? 4 A single fillet weld attaches a plate of length 200mm onto a base plate. The single fillet weld has a throat thickness of 6mm. If the designer wishes to change the design to have two fillet welds, one on either side of the attachment plate, what size leg length will they specify? 5 A 50kN load is carried by a 300mm long fillet weld in shear. If the steel has a yield strength of 240MPa, what size does the fillet weld need to be? Answer: 1) 48000mm2 2) 213MPa 3) 10,240kN 4) 4.5mm 5) 1.4mm throat.

8-11

www.twitraining.com

Section 9 Revision Session

Rev 2 April 2011 Revision Session Copyright  TWI Ltd 2012

9

Revision Session IWE Long questions 1 What can be done during design to minimise distortion during welding? 2 Describe four joints that could be used to weld 75mm thick steel. 3 Explain the basic procedure for carrying out an Engineering Critical Assessment. 4 Describe the main causes of in-service failure in relation to welded products. How would you recognise each type of failure when assessing fracture faces? 5 Briefly describe eight joint configurations for thin sheet ( 2 ave

Kmin < 0.7 Kave

or

Kmax > 1.4 Kave

residual stress

compression Copyright © TWI Ltd 2012

Copyright © TWI Ltd 2012

3-9

Failure Assessment Diagram Axes

Failure Assessment Diagram Axes • Kr is defined as:

• The Y-axis of the FAD is termed Kr, the fracture ratio.

KI K IC where KI is the stress intensity or crack driving force and KIC is the material toughness Kr 

– This axis represents proximity to failure of the flawed component by fracture

• The X-axis of the FAD is termed Lr, the stress ratio.

• Lr is defined as:  Lr  ref y σref is the reference stress or the net section stress due to a flaw and σy is the yield stress (or flow stress)

– This represents the proximity to failure of the component by plastic collapse

• Failure assessment line shows the limit of safety for failure from either mode. Copyright © TWI Ltd 2012

Copyright © TWI Ltd 2012

Failure Assessment Diagram

Failure Assessment Diagram

• If an assessment point falls between the inside the failure assessment line, the flaw is considered to be deemed acceptable and is unlikely to lead to failure.

x 1

Assessment point outside: Flaw unacceptable

Fracture axis:

• If the assessment point falls outside the failure assessment line, the law is deemed unacceptable and could lead to failure.

x

Kr

represents proximity to failure by fracture

or

Assessment point inside: Flaw acceptable

r 0

Lr (or Sr) Collapse axis: represents proximity to failure by collapse (or yielding)

Copyright © TWI Ltd 2012

Copyright © TWI Ltd 2012

Level 1 vs Level 2 Method

Validation of FFS Procedures

• Level 1 uses simpler formulae.

4

• Level 1 FAD has most conservatism.

3 1.0

• Level 2 needs more analysis (use software).

• Level 3 method using FEA has even less conservatism.

Fracture

• Level 2 FAD reduced conservatism.

BS7910 level 2a

0.8

r

Level 1

0.6 0.4

2

1

0.2 0.0 0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

Plastic collapse

1.6

0 0.0

0.5

1.0

1.5

2.0

Sr

• More analysis more costly. Copyright © TWI Ltd 2012

Copyright © TWI Ltd 2012

3-10

BS 7910 Fatigue Assessment Procedures

• Fatigue grows a crack from a sub-critical size until it is critical to cause failure. • The rate the fatigue crack grows can be estimated, and an initial tolerable flaw sizes can be derived if the lifetime is known. • For an ECA measure the fatigue crack growth rate in parent material, weld metal or HAZ. – Crack growth rates affected by environment e.g. seawater with CP compared to air

• Determine the fatigue threshold.

• The parameters required to carry out a general fatigue assessment are:

Flaw Type and Geometry Stress range

n1 cycles at 1 n2 cycles at 2 n3 cycles at 3 etc...

Time

Applied Stresses

Log (da/dN)

Assessment of Fatigue

Log ( K) Crack Growth Relationship

Copyright © TWI Ltd 2012

Copyright © TWI Ltd 2012

BS 7910 Fatigue Assessment Procedures

ECA Case Study

• The basis of the fatigue crack growth calculations is the relationship between K and the rate of crack growth da/dN for the material under consideration.

• Lack of fusion was detected in a propane storage sphere in Saudi Arabia during outage.

Log (da/dN)

fracture

m

1

threshold,  Ko

Log (  K)

• Before the vessel could go back into service, its defect tolerance had to be assessed to show that the sphere would not fail by brittle fracture or plastic collapse. • A fracture mechanics assessment showed that the storage sphere had adequate defect tolerance despite the lack of fusion flaw.

da  A  K  m dN Copyright © TWI Ltd 2012

Copyright © TWI Ltd 2012

Summary • Risks for brittle, ductile, creep, fatigue etc. • Charpy testing & fracture mechanics testing of welds. • Engineering critical assessment (ECA). – – – –

Equipment dimensions and flaw size Materials properties: strength, toughness Loading: applied & residual stress Failure assessment diagram to determine whether tolerable

• Different levels of assessment. • Benefits of fitness-for-service assessment. Copyright © TWI Ltd 2012

3-11

Objectives • Understand stress concentration locations. • Fatigue joint classifications. • Describe post weld fatigue life improvement techniques. • Review case studies of fatigue failures.

Advanced Fatigue – Part 1 TWI Training & Examinations Services (EWE/IWE Diploma)

Copyright © TWI Ltd 2012

Copyright © TWI Ltd 2012

Influence of Welding

Fatigue Cracking From a Weld Toe

Welds introduce stress concentrations from which fatigue can propagate

•

weld toe

weld toe

weld root

Weld toes and weld roots are the most critical areas in respect to stress concentration

Production arc welding processes lead to the formation of non-metallic intrusions at the weld toe.

•

Typically 0.1-0.4mm in depth.

•

Fatigue life governed by the growth of this pre-existing flaw.

•

Little or no initiation stage.

•

Factors which affect crack initiation can be quite different to those that affect crack growth.

Preexisting sharp flaw A

Fatigue crack ~ 50m

Copyright © TWI Ltd 2012

Fatigue Resistance Data - S-N Curves •

• •

•

Resistance quantified in terms of constant amplitude S-N curves. SmN = A (m and A are constants). Curves for different design details based on statistical analysis of test data. BS 7608.

Classification of Welded Joints Behaviour of a structure under cyclic loads is determined by the severity of stress concentration UK design rules: welds are grouped into Classes giving similar fatigue strength (similar stress concentration effect)

Stress range, N/mm 2 300

Copyright © TWI Ltd 2012

static design limit

200

Class of joint

100

B C D E F F2 G W 30

10 5 10 7

10 6

Endurance, cycles Copyright © TWI Ltd 2012

Expected life increase Copyright © TWI Ltd 2012

4-1

Classification of Welded Joints

Classification of Welded Joints Fatigue weld joint classification (BS 5400)

Fatigue weld joint classification (BS 5400) • Class A: plain steel, all surfaces machined and polished. Uniform or smoothly varying crosssection. • Class B: as-rolled steel, no flame cut edges. Full penetration butt welds free of defects, parallel to the direction of stress and with faces dressed smooth.

• Class C: full or partial penetration butt or fillet welds, parallel to the direction of stress, made by mechanised welding, with no stop/start. Butt welds transverse to stress direction, machined flush and free from defects.

Copyright © TWI Ltd 2012

Classification of Welded Joints

Copyright © TWI Ltd 2012

Classification of Welded Joints

Fatigue weld joint classification (BS 5400) • Class D: full or partial penetration butt or fillet welds, parallel to the direction of stress, with stop/start. Butt welds transverse to stress direction, with smooth transition to plate.

Fatigue weld joint classification (BS 5400) • Class E: full penetration butt welds, transverse to stress direction, with abrupt transition to parent metal.

Copyright © TWI Ltd 2012

Classification of Welded Joints

Copyright © TWI Ltd 2012

Classification of Welded Joints

Fatigue weld joint classification (BS 5400)

Fatigue weld joint classification (BS 5400)

• Class F: butt welds on backing strip, transverse to stress direction. Member carrying fillet or butt welded attachments clear of edge. Cruciform or Tjoints made with full penetration butt welds.

• Class F2: cruciform or T-joints made with partial penetration butt or fillet welds. Butt welds between two rolled or built-up sections.

Copyright © TWI Ltd 2012

Copyright © TWI Ltd 2012

4-2

Classification of Welded Joints

Classification of Welded Joints Fatigue weld joint classification (BS 5400) • Class G: members with attachments welded to their edges or close to the edges.

Fatigue weld joint classification (BS 5400) • Class W: weld metal in load carrying fillet or partial penetration butt welds, regardless of direction of stressing. Stress used in calculating the fatigue life is the stress on the weld throat.

Copyright © TWI Ltd 2012

Copyright © TWI Ltd 2012

Worked Example: Joint Class

Worked Example What are the weld classes shown in this joint?

Longitudinal fillet weld: Class C but Class F at the end (Start/Stop)

Transverse, vertical butt weld: Class E

Transverse butt weld: Class D (Class E if overhead or vertical)

Note: Class W – undersized welds or weld defects!

Copyright © TWI Ltd 2012

Weld Toe Failure

Copyright © TWI Ltd 2012

Fatigue Fracture Surface

Copyright © TWI Ltd 2012

Copyright © TWI Ltd 2012

4-3

Fatigue Fracture Surface

Fatigue Failure in Components Depends on: • Load direction • Geometry

Beach marks • Provide record of crack front location at particular time • Associated with differential corrosion, sudden change in crack growth rate, rest periods…

Striations

x1800

• Usually represent crack increment per cycle

Large structures can be tested using representative smaller-scale specimens

Copyright © TWI Ltd 2012

Fatigue Improvement

Copyright © TWI Ltd 2012

Fatigue life improvement techniques

Improve by Design

Reduce stress concentration and eliminate sharp discontinuites at weld toe: • Weld toe grinding. • TIG or plasma re-melting.

• Achieve a smooth transition between weld and plate. • Use smooth shapes and transitions. • Grind flush excess weld metal, grinding marks should be parallel to stress direction. • Local machining, remove weld toes and edges.

Reduce residual stresses: • Stress relief. Introduce compressive residual stress: • Hammer, needle or shot peening. • Ultrasonic impact treatment. • Proof loading.

Copyright © TWI Ltd 2012

Effect of Flush-Grinding Girth Welds

Copyright © TWI Ltd 2012

Recommended Material Removal

500

t = plate thickness Stress range 2

N/mm

r = groove radius

300

d = groove depth

r/t >0.25 r/d >4 to minimise notch effect

200 150

100

As-welded Flush-ground weld Flush ground, failed from flaw  Unbroken 50 4

10

5

10

6

10

7

10

8

10

Endurance, cycles Copyright © TWI Ltd 2012

Copyright © TWI Ltd 2012

4-4

Fatigue Improvement Techniques

Burr Machined Weld Toe

Toe burr machining appearance

Copyright © TWI Ltd 2012

Fatigue improvement techniques Weld toe burr grinding (machining)

Copyright © TWI Ltd 2012

Fatigue Improvement Techniques Toe burr machining - method

45º

45º

• improves fatigue life by 2 classes (30% on stress or 2.2 on life). • slow careful application required. • must penetrate recommended depth below any undercut or flaw. Copyright © TWI Ltd 2012

Fatigue Improvement Techniques

Minimum depth 0,8 mm

Direction of travel

Copyright © TWI Ltd 2012

Burr Machining and Disc Grinding

Toe disc grinding

30-45º

Direction of travel • Quicker than burr machining. • Less improvement (1 Class) – direction of grinding marks. • Controlled procedure and inspection required. Copyright © TWI Ltd 2012

Copyright © TWI Ltd 2012

4-5

Weld Toe Burr Grinding

TIG or Plasma Dressing H/course/burr

600 500

Weld toe burr ground with r/T > 0.25 & r/d > 4  Unbroken

400

Depth of grinding d = 0.5mm below undercut r

Stress range

300

2

N/mm d Stressed plate

Butt welds

200 150

T=13mm Unwelded steel

Proposed requirement: r/d > 4 to minimise notch effect

100 80 4

10

5

10

6

10

7

10

Endurance, Cycles

• Very slow technique. • Requires separate qualified procedure and significant skill element. • Delays subsequent weld inspection.

Copyright © TWI Ltd 2012

Plasma Dressed Fillet Weld

Copyright © TWI Ltd 2012

Effect of Compressive Residual Stress

Copyright © TWI Ltd 2012

Weld Toe Hammer Peening

Copyright © TWI Ltd 2012

Weld Toe Needle Peening

Copyright © TWI Ltd 2012

Copyright © TWI Ltd 2012

4-6

Fatigue Performance of Welded Joints Improved by Peening

Ultrasonic Impact Treatment

log (stress range) Stress at cross-over increases with tensile strength

Peened

Decrease (to negative) in R Increase in tensile strength Increase in compressive residual stress

As- welded Increase in fatigue limit

log (fatigue life) Copyright © TWI Ltd 2012

Comparison of Improvement Methods 400

Copyright © TWI Ltd 2012

Effect of UTS on Benefit 600

Mild steel R=0

500

300

400

150

Fatigue strength 300 at 2x106 cycles, 200 MPa

Hammer peened

Stress range 200 N/mm2

Burr ground Shot peened As welded

Plasma dressed

100 100 5 10

6

10

7

Endurance, cycles

10

7

5x10

0 300

TIG dressed weld Toe ground weld 500

700

1000

1200

Ultimate tensile strength of steel, MPa Copyright © TWI Ltd 2012

Concluding Remarks

Copyright © TWI Ltd 2012

Alexander L. Kielland platform

• Significant improvements in fatigue life for joints at risk of failure from the weld toe. • Typical improvement by a factor of 2-3 on fatigue life. • Well defined method statement and operator training required. • Post-treatment inspection to ensure complete coverage. • Detailed recommendations published by IIW and in BS 7608. Copyright © TWI Ltd 2012

• Pentagone type semi-submersible rig in North Sea. • One of five columns broke off, and platform capsized. • Initiation site was fillet weld between hydrophone support and brace.

Copyright © TWI Ltd 2012

4-7

Fatigue cracking in brace D6 of Alexander L. Kielland Platform • Incomplete penetration and poor weld bead shape. • Fatigue crack propagated into brac.e • No redundancy (loss of one brace led to overloading of others). • Importance of welded attachments.

Mooring Buoy Failure Flotation Chamber

Plane of Failure 100m

Universal Joint Copyright © TWI Ltd 2012

Mooring Buoy Failure

Copyright © TWI Ltd 2012

Mooring Buoy Failure

Copyright © TWI Ltd 2012

Copyright © TWI Ltd 2012

Mooring Buoy Failure

Mooring Buoy Failure

Copyright © TWI Ltd 2012

Copyright © TWI Ltd 2012

4-8

Mooring Buoy Failure

Mooring Buoy Failure

Copyright © TWI Ltd 2012

Mooring Buoy Failure

Copyright © TWI Ltd 2012

Mooring Buoy, Sequence/Cause of Failure Fatigue Ductile overload

Multiple stage cleavage/arrest

Multiple stage cleavage/arrest

Copyright © TWI Ltd 2012

Copyright © TWI Ltd 2012

Missing Material

Missing Material

Reasonable theory

Actual situation Copyright © TWI Ltd 2012

Copyright © TWI Ltd 2012

4-9

Tacoma Narrows Bridge, 1940

Tacoma Narrows Bridge, 1940

Copyright © TWI Ltd 2012

Copyright © TWI Ltd 2012

4-10

Objective • Develop principles of design for static loading.

Static Loading TWI Training & Examinations Services (EWS/IWS Diploma)

Copyright © TWI Ltd 2012

Copyright © TWI Ltd 2012

Elastic Design Method

Tensile Test Results

• Elastic Design Method. – Ensure that stresses in structure do not exceed yield stress (ie elastic deformation)

• However we cannot design up to yield stress safely due to: – – – –

Material defects Joint/weld mismatches Unforeseen loads Degradation

Copyright © TWI Ltd 2012

Copyright © TWI Ltd 2012

Elastic Design Method

Factor of Safety

• Use design stress which is a fraction of the yield strength of the parent material. • For critical structures such as pressure vessels this was once set at 1/4 UTS, but later changed to 2/3 yield stress. • Relevant codes dictate design stresses.

Copyright © TWI Ltd 2012

• Ratio of yield stress (or UTS) to design stress is known as factor of safety (FoS). FoS 

Yield Stress 1 Design Stress

• FoS depends on: – Material – Utilisation Copyright © TWI Ltd 2012

5-1

Material Load Carrying Capacity • Weld metal overmatches parent metal. – Parent strength defines load carrying capacity

• High strength low alloy steels. – Weld metal sometimes undermatches parent metal

• Welded joints in aluminium. – The static strength may be reduced by the heat of welding

Fillet Weld Features • The simplest and cheapest type of welded design • requires less preparation. • May be made in flat or horizontal positions by semiskilled operators. • Possible problems with penetration into the parent metal. • Welds can be made with any number of passes • Difficult to examine by NDT methods for subsurface defects. • Once you calculate the required size of fillet weld stick to it - the volume or weight of weld metal increases as the square of leg length.

Copyright © TWI Ltd 2012

Copyright © TWI Ltd 2012

Fillet Welds - Simplifying Assumptions

Fillet Weld Throat

• The weld metal must match the strength of parent metal. • Weld will fail across the throat. • Excess weld metal is neglected (for calculation use design throat instead actual throat). • Assume adequate weld quality. • Neglect stress concentrations due to bead shape. • The design stress is YS/2 (when it is not given in the applicable standard). • Ignore residual stress.

Throat dimension: - the shortest distance between root and the chord between the toes or profile face (whichever is less). throat

throat 60º

Welds with faces meeting at more than 120° or less than 60° should NOT be used for load carrying parts.

Copyright © TWI Ltd 2012

Copyright © TWI Ltd 2012

Minimum fillet weld sizes

Fillet weld design

(recommended by BS 449)

• some codes specify minimum fillet weld size depending on thickness of plate to be welded. This has more to do with prevention of cold cracking than for strength • maximum throat size for fillet welds made from one pass is 6mm. • if plates are 6mm thick or more leg length should be not less than 5mm. • do not overweld - risk of lamellar tearing.

Thickness of thicker part (mm) 10-18 18-30 Over 30

Minimum leg size (mm) 4 5 6

(recommended by AWS D1.1.) Thickness of thicker part (mm) Up to 6 6-12 12-20 Over 20 Copyright © TWI Ltd 2012

Minimum leg size (mm) 3 5 6 8 Copyright © TWI Ltd 2012

5-2

Types of Forces

Bending Moment • A cantilever beam:

Compression

– Compression

M

– Tension

Force, F

Fy Fx

Tension

d

– Shear

Shear

– Bending Bending

Bending moment:

Reaction forces:

M=Fxd

Fx

Fy

M

Copyright © TWI Ltd 2012

Copyright © TWI Ltd 2012

Bending Stresses

Bending Stresses Stress calculation - engineering bending formula

• Consider a beam subjected to pure bending (no shear).

M

M

Tension (+) M

M After bending Compression (-)

Neutral Axis Longitudinal stresses are zero

• Maximal values for stress are obtained on extreme fibres. M • I beam: flanges resist bending while web resist shear. • Stresses have a continuous linear variation throughout entire transverse section.



+σ where: M = bending moment Iz = moment of inertia towards axis Z ymax = maximum distance between cross section’s CG and extreme fibre

Copyright © TWI Ltd 2012

Bending



x a m



x a m

Before bending

y Iz M

-σ

Copyright © TWI Ltd 2012

Section Modulus

M

Compression -σ

+σ

• The section modulus (Z) is a geometric property only which relates stress and internal moment during elastic bending. • The bigger the value of section modulus for a particular cross-section the bigger the bending moment which it can withstand for a given maximum stress. Z

Tension

Neutral axis (where σ=0) goes through the centre of gravity of the transverse section Copyright © TWI Ltd 2012

Iz y max

• ymax is the distance from the neutral axis to the extreme fibres of the beam. Copyright © TWI Ltd 2012

5-3

Ways to Improve Resistance to Bending

Ways to Improve Resistance to Bending

Push the material as far as you can from the neutral axis Copyright © TWI Ltd 2012

Bending - Application

Copyright © TWI Ltd 2012

Bending - Application

Determine the bending stress:

Determine the bending stress:

P

P h

h

L

L b

b Bending moment M = Force x distance = P x L Moment of inertia I = (b x h3)/12 Maximum distance ymax= h/2 Bending stress σ = M x ymax/I

Bending moment? Moment of inertia? Maximum distance? Bending stress? Copyright © TWI Ltd 2012

Bending Stresses

Copyright © TWI Ltd 2012

Worked Example

• Calculate the maximum stress in a beam subjected to 6kNm bending 50mm

Using the simple theory of bending:  max  I

4MPa Compression

My max I

100

bh 3 100  300 3   2.25  10 8 mm 4 12 12

 max 

6  10 6  150 2.25  10 8 I

 4MPa

300

Distribution of longitudinal stresses 4MPa Tension Copyright © TWI Ltd 2012

Moment

A bending moment of 9x105Nmm is applied to the end of a beam 1.5m long. The beam is 50mm deep and 20mm wide. 1. Calculate the leg length needed for the fillet welds shown above if the allowable stress in the fillets is 250MPa. 2. How would this change if the beam length was 0.75m?

Copyright © TWI Ltd 2012

5-4

ANSWER 1

ANSWER 1 F 50mm

Moment

Throat, t = L/2 L, leg length Hence allowable stress 250 is related to leg length as:

F

Applied moment = moment reacted in welds = 50 x F F = 9x105/50 = 1.8x104N

So,

250 L

Stress in fillet = F/(t.w)

= = = =

(F2)/(L.w) (1.8x104 2)/(250.20) (2.54x104)/(5000) 5.1mm

Where: t is the throat thickness and w is beam width (20mm) Copyright © TWI Ltd 2012

Copyright © TWI Ltd 2012

ANSWER 2

Torsion Stress calculation (valid only for round shape transverse section beams-bars or tubes):



M tZ p

Beam length makes no difference to required weld size because the applied loading is a bending moment. That is, F is independent of beam length.

τ

Mt - torsion moment Zp - polar section modulus of the transverse section

τ Mt

Note that:

• Largest values of stress are obtained on the extreme fibres while in centre of section, stress is 0. • the stress have a continuous linear variation throughout entire transverse section. Copyright © TWI Ltd 2012

Copyright © TWI Ltd 2012

Design for Static Loading

Design for Static Loading

M

 D

Copyright © TWI Ltd 2012

Load F  CSA Weld area 

F D



 



t M 2 D 2 π

F

Fillet weld subjected to torsion t

t M DD 2 π

Fillet weld subjected to torsion t Weld area = π ·D ·t M The torsion moment can be replaced by a D 2 M M  F  F  force F: 2 D







Copyright © TWI Ltd 2012

5-5

Reinforcing–Steel Welded Joints • • • • • • • • •

Purpose • The term reinforcing-steel is used to describe the use of steel to reinforce materials, most often concrete.

Purpose. Profile. Joints. Properties. Welding processes. Joint types. Joint loading. Preheating. Standards and specifications.

• Concrete is a brittle material which is strong in compression but weak in tension. • This limits the use of concrete in construction and makes it unsuitable for use in many structural members. Copyright © TWI Ltd 2012

Purpose

Copyright © TWI Ltd 2012

Purpose

• A beam made only of concrete has little or no bending strength since cracking occurs in the extreme tension fibres in the early stages of loading.

• In concrete beams the steel bars are embedded in the tension fibres of the beam.

• To improve the tensional capability of concrete we can embed steel bars called reinforcing bars, or rebar, in the concrete.

Copyright © TWI Ltd 2012

Copyright © TWI Ltd 2012

Joints

Bar Profile • Reinforcing bar is rolled with a ribbed die when manufactured to produce a ribbed surface. • Concrete shrinks around the bar when it sets to grip the reinforcement.

• Reinforcing bar is available in sizes ranging from 6mm up to 50mm diameter. • A whole assembly of reinforcing bars will usually be used. • To join bars together there are several methods: – Welded joint – Wire joint - wire wrapped around bars and tightened – Rebar coupler – mechanical fixing

Copyright © TWI Ltd 2012

Copyright © TWI Ltd 2012

5-6

Joints

Properties

• Welding can be used for both load bearing joints and non load bearing joints.

• Reinforcing bars available for a wide range of chemical compositions and mechanical properties.

• Non load bearing joints used to keep components in place during fabrication, transportation and concreting - tack welds. • Joints may be subject to significant loads during handling and transportation – should be treated as load bearing.

• Not all reinforcing bars are weldable weldability is determined by the carbon equivalent value and the limitations on the content of certain elements.

Copyright © TWI Ltd 2012

Copyright © TWI Ltd 2012

Welding Processes

Types of Joint

• The following welding processes, in accordance with ISO 4063, may be used to perform the welding of reinforcing-steel joints. Description manual metal arc welding (MMA) self-shielded tubular cored arc welding metal active gas welding (MAG) tubular cored metal arc welding with active gas shield

Load Bearing ● ● ● ●

• Butt Joint: – Load bearing joints only – Needs preparation – Backing strip may be used

Non Load Bearing ● ● ● ●

Copyright © TWI Ltd 2012

Copyright © TWI Ltd 2012

Types of Joint

Types of Joint • Strap Joint:

• Lap Joint:

– – – – –

– Load bearing and non load bearing – Double sided weld possible – Minimum throat thickness, a ≥ 0.3d

Load bearing joint only Uses strapping bars Double sided weld possible Minimum throat thickness, a ≥ 0.3d Variation uses splice plate or angle

Straps

d ≥4d

≥2d

≥4d ≥4d Copyright © TWI Ltd 2012

≥2d

≥4d Copyright © TWI Ltd 2012

5-7

Types of Joint

Types of Joint

• Cross Joint: – – – – –

Load bearing and non load bearing Double sided weld used for load bearing Minimum throat thickness, a ≥ 0.3dmin Minimum weld length, l ≥ 0.5dmin When welding different diameters: dmin/dmax ≥ 0.4

• Joints between reinforcing bar and other steel components: – For load bearing functions, steel reinforcing bars are often joined to other steel components in a structure such as plates or sections. – The types of joint used include side lap weld joints and transverse end plate joints.

a Copyright © TWI Ltd 2012

Types of Joint

Joint Loading

• Common ranges of bar diameters for load bearing welded joints. Welding Process 21 23 24 25 42 47 111 114 135 136

Type of Joint

Bar Diameters (mm)

cross joint

4 to 20

butt joint butt joint joint to other steel component butt joint butt joint without backing butt joint with permanent backing lap joint strap joint cross joint joint to other steel component

Copyright © TWI Ltd 2012

• Direct and indirect loading: – Describes the way forces are transmitted across a joint – A butt joint is classed as a direct loading joint as the bars are axially aligned

5 to 50 5 to 25 6 to 50 6 to 50 6 to 50 ≥16 ≥12 6 to 32 6 to 50 6 to 50 6 to 50

– Lap and strap joints are classed as indirect loading joints as the forces are transmitted with eccentricity. Eccentric loading causes the joint to flex and hence bending stresses are set up Copyright © TWI Ltd 2012

Copyright © TWI Ltd 2012

Preheating

Standards and Specifications

• The minimum preheat temperatures for reinforcing steel welded joints are dependant on:

• BS EN ISO 17660: Welding - Welding of reinforcing steel. • AWS D1.4: Structural Welding Code - Reinforcing Steel. • Eurocode 2: Design of Concrete Structures. • BS EN 10080: Steel for the reinforcement of concrete - Weldable reinforcing steel – General. • BS 4449: Steel for the reinforcement of concrete Weldable reinforcing steel - Bar, coil and decoiled product - Specification. • BS 7123: Metal arc welding of steel for concrete reinforcement.

– Type of joint – Carbon equivalent value of the steel – Diameter of the bar or combined diameter of the joint – Hydrogen content of the weld metal – Arc energy

Copyright © TWI Ltd 2012

Copyright © TWI Ltd 2012

5-8

Standards and Specifications • AWS D1.1: Structural Welding Code. • BS 5950: Structural use of steelwork in building. • BS 8118: Structural use of aluminium. • Eurocode 3 (BS EN 1993): Design of steel structures. • Eurocode 9 (BS EN 1999): Design of aluminium structures. • BS 7608: Code of practice for fatigue design and assessment of steel structures. • BS 7910: Guide to methods for assessing the acceptability of flaws in metallic structures. • BS EN 22553: Welded, brazed and soldered joints. Symbolic representation on drawings. Copyright © TWI Ltd 2012

5-9

Design Codes • Design and build pressure vessels to an established design code. • In EU must be in accordance with PED. • Codes will define:

Design of Pressure Equipment

– – – –

TWI Training & Examinations Services EWE/IWE Diploma

Maximum allowable stress Minimum design temperature Materials and mechanical properties Permitted design features

• PD 5500, EN 13445, ASME Boiler & PV codes. Copyright © TWI Ltd 2012

Copyright © TWI Ltd 2012

Construction of Pressure Vessels

Material Choice • Mechanical properties:

• Shell

– Often made from types of steels for strength – Low temperature use stainless steel, aluminium – Lightweight vessels use composites

–

Main body, usually cylindrical, sometimes sphere

• Head –

Used to close the cylindrical shell, usually dished

• Nozzle

• Corrosion resistance:

–

– Use aluminium, polymers, ceramic, stainless steel – Thin wall vessels made from corrsion-resistant material – Thick wall, use a CR liner in a steel vessel

Opening for filling, drainage or inspection

• Saddle Supports –

Used to hold vessel in place

• Nameplate –

Contains important information about the vessel

Copyright © TWI Ltd 2012

Copyright © TWI Ltd 2012

Design of Pressure Vessels

Design Approach • Design-by-rule:

• Design-by-analysis:

– Virtually all codes use this approach – Relatively simple equations

– Use stress analysis to compare the stresses against specific criteria – Use limit analysis to determine the load to cause failure by gross plastic deformation – Safety factors applied to ensure constant level of safety

• Advantages: – Simple and consistent

• Disadvantages: – The rules are sometime open to interpretation. – Difficult to apply when dealing with loadings and geometries which are not covered by the standards

Copyright © TWI Ltd 2012

Copyright © TWI Ltd 2012

6-1

Dished Head Design

Shell Design • The hoop stress is twice the axial stress. • longitudinal welds designed with an offset between strakes. • Shell thickness from few mm to few 100mm.

Knuckle Petals

offset

Crown

Copyright © TWI Ltd 2012

Dished Heads

Copyright © TWI Ltd 2012

Shell to Head Welds Joint design to allow for misalignment: misalignment

Hemispherical • Thinnest material. • Difficult to make.

Semi-elliptical • Thickness equal to shell. • Difficult to make.

Torispherical • Smallest axial dimension. • Easy to make. • Thicker than the shell.

Advantage: • Double bevel prep allows for misalignment. • Only one item needs to be prepared.

Disadvantage: • Requires access both sides. • Misalignment can set up local bending stress.

Copyright © TWI Ltd 2012

Shell to Head Welds Used for thick-walls

Copyright © TWI Ltd 2012

Nozzles Nozzles connect a pressure vessel with other components

Used for thin-walls

Type of nozzle depends on: 1.5 x t

• Diameter/thickness ratio of the shell.

Advantages:

Disadvantages:

• Diameter/thickness ratio of the nozzle.

• Self-jigging. • Allow for use of high heat input welding process. • No danger of burn-trough. • Weld only on outside.

• High level of residual stress and distortions. • Difficult to inspec.t • Poor resistance on fatigue stress.

• Access (one side only or both sides).

Copyright © TWI Ltd 2012

• Type of joint required (partial/full pen). • Groove preparation methods available. Copyright © TWI Ltd 2012

6-2

Set-on nozzle

Set-Through Nozzle • • • •

• Shorter nozzle is cheaper. • Easy to make groove for full or partial penetration. • Single side welding in 2G/PB position means high welder skill is required. • Through thickness stress means danger of lamellar tearing. • Can be difficult to UT especially on smaller diameters. • Mainly used for small (