Вы находитесь на странице: 1из 9

Forensic Engineering Volume 164 Issue FE1 Cracks in steel structures Mann

Proceedings of the Institution of Civil Engineers Forensic Engineering 164 February 2011 Issue FE1 Pages 1523 doi: 10.1680/feng.2011.164.1.15 Paper 1000003 Received 22/06/2010 Accepted 20/10/2010 Keywords: Failures/fatigue/steel structures

ice | proceedings

ICE Publishing: All rights reserved

Cracks in steel structures


Allan Mann CEng, FREng, FIStructE, MICE Senior Consultant, Jacobs, Manchester, UK

Under overload, the mode of failure in any steel structure should ideally be ductile not brittle; the possibility of cracking in steelwork should thus be actively guarded against. Historically, steel cracking has caused some spectacular failures. Steel can fracture rapidly at low temperature and can crack during welding, in fatigue or as a result of stress corrosion. Certain fabrication processes such as welding, flame cutting and punching exacerbate the risks, especially if the steel is thick. Galvanising can also be a risk factor. Design engineers should understand potential cracking mechanisms and risk factors to minimise the likelihood of in-service failure. This paper discusses the historical background of cracking, explains causes and suggests avoidance measures.

1.

History

The Tsar bell, situated at the Kremlin in Moscow illustrates that cracking in metals has long been a problem. During a fire in 1737, a massive portion (weighing some 11 t) of the bell cracked while it was in the casting pit (Figure 1). It might be argued that the modern steel industry owes much to a fear of cracking: in the nineteenth century, when cast iron was the most widely available structural metal, tensile performance was an unpredictable quality and safety was sought by deploying large safety margins. However, the cracking risk was not controllable and several examples of bad structural failure occurred. All failures are intolerable, but failure by fracture is especially intolerable since it violates the key safety attribute of ensuring a visible collapse warning by means of prior distortion. The danger of cracking/brittle fracture of cast iron has long been recognised, at least empirically, as most early bridges such as that at Ironbridge (1779) had arches as their structural form (although the bridge still has cracks within it). Longer bridges supported by chains (e.g. the Menai suspension bridge (1826)) used wrought iron (at least for the chains) for better reliability. Nevertheless, cast iron continued to be used for beams for a time. In 1847, Robert Stephensons cast iron girder bridge over the River Dee near Chester collapsed as a train crossed over; the collapse may have been precipitated by fracture of the girders bottom flange. In 1891, a cast iron bridge at Norwood Junction collapsed due to fracture emanating from a large blowhole in the flange. This was a key event in the decline of cast iron. After an inquiry into the Norwood crash, the Board of Trade recommended that all cast iron bridges be replaced. Steel became the preferred material, as used for construction of the Forth railway bridge in 1890. This use of steel continued more or less until the Second World War. Up to then, most structures and their connections were

made by riveting fairly thin plates of steel together. This had two benefits the steel was thin so the materials tensile reliability was better than that of thicker steel and failure of one part in a multiple riveted assembly did not necessarily spell disaster for the whole unit. The 1920s1930s, however, heralded the start of a new era when the Journal of the Institution of Structural Engineers published a series of papers on welding. In 1932 the journal published a report by the Steel Structures Research Committee that highlighted some welding tests with the comment A discussion and description of tests on the welding of steel structure, a subject which is of the greatest interest as offering perhaps one method whereby the intolerable noise associated with riveting may be avoided. The demand for rapid production throughout the Second World War gave welding technology a strong impetus. However, the new technology brought new problems, the most spectacular of which were full fractures of the all-welded American Liberty ships (Boyd, 1970) (Figure 2). At the time, this was a mystery: the failures were extensive and could occur apparently spontaneously while the ships were in calm waters. A new phenomenon clearly existed and required urgent investigation. To some extent, cracking and crack propagation had always been a design issue even with the steels used in riveted boats, but at least in those structures the various overlapping plates and riveted-on stiffeners offered some crack arrest features. Gordon (1978) describes some significant crack incidences in transatlantic liners. After the war, the Liberty ship fractures generated intense research into the cause of failure. It had, of course, been known for a long time that steel possessed properties of both strength and toughness, with the two being separate qualities. Toughness had been formally investigated in 1905 by
15

Delivered by ICEVirtualLibrary.com to: IP: 130.88.85.64 On: Wed, 28 Sep 2011 22:58:11

Forensic Engineering Volume 164 Issue FE1

Cracks in steel structures Mann

Cleavage fracture

Fibrous fracture

Energy absorption

Fracture transition

Brittle failure

Shear failure

Figure 1. Tsar bell

Easy crack initiation and propagation zone

Crack propagates but initiation difficult

Crack arrest zone

Charpy, a French scientist who gave his name to the standard test that defines toughness by a measure of the energy a particular steel specimen can absorb while fracturing: the test is still used today. The test can be used to demonstrate change in steel toughness with temperature, the toughness magnitude dropping off dramatically below a value known as the transition temperature (Figure 3). Although the Charpy test demonstrated toughness variation, it did not in itself explain the fast fractures observed in ships. This was investigated by navy engineers and eventually led to greater understanding of cracking and fracture mechanics (Boyd, 1970).

Temperature

Figure 3. Ductile/brittle transition curve

Underlying all this, a puzzling problem in the strength of materials had been identified in about 1914 when Inglis looked at stress concentrations and showed that, theoretically, the localised stress that must exist linked to flaws/notches could be many times the known tensile capacity of steel (Gordon, 1978). Yet patently, despite these localised stress concentrations, most steel structures were able to cope with such stresses in service and not only from flaws introduced by design (such as sharp re-entrant corners) but also from flaws that were intrinsic to the basic manufactured material. The plausible explanation relied on the ability of steels ductility to yield locally without the high stress causing fracture. A milestone in the understanding of cracking came about in the 1920s when Griffiths (Gordon, 1978) pondered the phenomenon, speculated on cause and deduced that there was such a thing as a tolerable crack length that could exist in a tensile stress field. This led to the definition of a critical crack length marking the transition point between stability on the one hand and rapid instability via crack growth on the other. Griffiths thought in terms of energy rather than stress. This was the birth of modern fracture mechanics, which is a complex subject seeking to address questions of stability for cracks of a given length in a given material in a given stress field.

Figure 2. Dramatic brittle fracture of a Liberty ship

Another cracking phenomenon was first investigated by Wohler around the 1850s. Wohler was a German railway

16

Delivered by ICEVirtualLibrary.com to: IP: 130.88.85.64 On: Wed, 28 Sep 2011 22:58:11

Forensic Engineering Volume 164 Issue FE1

Cracks in steel structures Mann

engineer charged with investigating the continuing failures of axles on rolling stock. It was Wohler who first produced SN curves and showed that metals had a finite life brought about by crack propagation when subjected to repeated loading that generated stress levels very much below known yield values. This was the birth of our understanding of fatigue.

pass/fail criteria (20 ft lbs) developed when investigating ship failures. Moreover, the designation of a Charpy value at a defined low temperature does not mean that the steel is only suitable for use at that temperature. The phenomenon of lowtemperature brittle fracture is more complicated, and other risk factors are material thickness (Burdekin, 1999) and whether or not the steel has been welded (TWI, 1971). A second, and separate, property that can be defined and selected by a designer is to use steel with guaranteed throughthickness properties to avoid lamellar tearing. Such steel is designated as having Z quality ductility and steels are available with various amounts of ductility. Codes such as BS EN 19931-10: 2005(E) (BSI, 2005) give guidance on requisite risk factors and grade selection but a key factor is the application of high stress, or strain, in a direction normal to the steel rolling direction. Lamellar tearing is characterised by a plate of steel splitting along its length. Application of high stress might be due to in-service loading but is more commonly associated with high shrinkage strains imposed during welding (i.e. when the cooling shrinkage is restrained). Risk factors include the use of thick steel. Other risk factors include the presence of a large number of non-metallic inclusions within the steel; Z quality steels are clean steels with very low sulphur content. In manufacture, steel toughness is achieved by a number of means, two of which are control of hardness and control of microstructure. A hard steel is one that has low resistance to crack growth and so hardness needs to be carefully controlled. In everyday engineering, hardness can be increased by the application of high heat followed by rapid cooling, such as may occur in uncontrolled welding or flame cutting. Overall, a designers first responsibility is the selection of a steel that is appropriate to the service conditions and the method of fabrication. In normal design this is straightforward, but selection becomes specialised when casting is involved. In those circumstances more care is required. This follows partly because the steel is thicker and partly because there is a more direct responsibility on the parties involved to assign appropriate ductility and toughness values. The dangers are real, as illustrated by press reports on the failure of certain cast cable anchorages on the Glasgow Clyde arc bridge in 2008. The Standing Committee on Structural Safety has issued a topic paper on castings (Scoss, 2010).

2.

Failures

Fatigue failures remain common in certain industries and the failures can be serious. A series of rail cracking problems in the 1990s led to some dreadful accidents (Mann, 2008). Several major oil rig collapses have been put down to fatigue, notably the Sea Gem in 1965 and the Alexander Kielland in 1980. In both cases, cracks rapidly propagated through the rig legs, leading to total leg loss and subsequent capsize. In both cases, the initiation sites for crack propagation were linked to welding and associated defects, possibly exacerbated by lamellar tearing in the base material. Generally, faulty fabrication/ welding processes can increase the risk of cracking and the performance of welded joints under high strains imposed during seismic action has highlighted this. Cracking failures in both the Kobe and Northridge earthquakes (Maranian, 1997; 1998) revealed the sensitivity of a whole frames integrity to the quality of the welding deployed at joints. Certain cracking problems occur in certain environments. One of these is stress corrosion cracking that, while not normally presenting a problem for structural engineers working in carbon steels, has led to building failures when stainless steel has been used (Mann, 1993). Fortunately, failures are comparatively rare. However, rarity should not lead to complacency because the very nature of rarity brings its own problems: as failure occurrence will be outside the experience of most engineers, the chance that risk factors may be overlooked is increased. Given that cracking failures can be instant and catastrophic it is incumbent on all design engineers to understand the importance of recognising potentially hazardous operating conditions, to be knowledgeable in the selection of appropriate materials and to be skilled in controlling fabrication quality so as to minimise the risk of cracking developing in service.

3.

Basic metallurgy

Steels can be purchased in a variety of strength grades and a variety of sub-grades that offer enhanced toughness values such that steel can be used in low temperatures. The first protection against low-temperature fracture (very rapid crack propagation) is to specify a steel with the appropriate toughness properties. Design codes and other advice offer guidance as to the grade required (Swann, 2005). The standard definition of toughness is a designated Charpy value (usually 27 J) achieved at a designated temperature. It should be noted that 27 J is just an arbitrary value (it derives historically from a

4.

The role of thickness

The risk of cracking in thick steel is higher than in thin steel for a variety of reasons (Burdekin, 1999). In the late 1980s there were some spectacular fracture problems in the USA (ENR, 1986; 1987a; 1987b) when welded jumbo sections were used in tension (flanges 125 mm, webs 75 mm). In one truss, the tension boom spontaneously cracked right through while in
17

Delivered by ICEVirtualLibrary.com to: IP: 130.88.85.64 On: Wed, 28 Sep 2011 22:58:11

Forensic Engineering Volume 164 Issue FE1

Cracks in steel structures Mann

service. Forensic examination proved the material to have low toughness and the investigation concluded there were inherent problems with the way the jumbo sections were produced leading to excessive grain size and particular weakness in the centre of the thick sections (where the cooling was too slow) and at the flange/web junction (where grain refinement by rolling was weakest). The initiation point (certainly for one fracture) was the hardened edge of a flame-cut mousehole utilised for completing the butt weld across the flange. These failures led to recommendations to not weld jumbo sections and for controlling the cooling rate when flame cutting.

non-destructive examination (NDE) to a degree proportionate with the welds service function. It is these same factors that explain the failure of the Liberty ships: the basic material had poor toughness; the welds probably had defects; the driving energy for the crack (in apparently low external stress conditions) was the internal residual stress left over from welding. The continuity provided by the all-welded structure allowed the crack to propagate a considerable distance without arrest. These risk factors can now be controlled by appropriate steel selection, application of proven weld procedures and post-weld NDE. In unusual circumstances, post-weld heat treatment can additionally be used to enhance fracture toughness (Berenbak et al., 2001). In the aftermath of the Northridge earthquake of 1994, it was observed that a great many welded joints of the steel rigid frames used in the region had cracked (Maranian, 1997; 1998). This was put down partly to poor material and partly to faulty welding procedures whereby large welds had been laid down but allowed to cool too quickly, thus creating a local hardened microstructure. While the welds might have possessed adequate strength, they lacked ductility. Thus, in the distortion imposed under seismic conditions, the welds cracked (and the cracks propagated into parent material).

5.

Effects of fabrication

The failure example of the jumbo sections shows how faulty fabrication can exacerbate the risk of in-service cracking. Given that stress concentrations and brittleness are risk factors, anything that promotes either feature is best avoided. However, in practical structures, eliminating these features is not possible. It is certainly not possible to eliminate all stress concentrations since they exist at every re-entrant corner and around every bolt hole. In many cases, observed failures under service conditions have been associated with welding, which is clearly one of the most common joining technologies. While most welds perform perfectly well, it still remains important to comprehend what problems welding may introduce. Welding can be a root cause of failure for several reasons. (a) Many welds contain a defect of some sort (albeit they might be at microscopic level and have no effect on static strength) and those defects might be sharp: given the right orientation in the stress field they may thus generate extremely high stresses at their tips. In effect, such defects might be the seed for future crack development. Welding requires the application of great heat and subsequent cooling. A hard and therefore brittle microstructure results if cooling is too rapid. Conversely, if cooling is too slow, the grain size might be too large and this lowers the inherent toughness. Thus, even with sound parent metal, an inappropriate welding procedure may degrade the basic material properties and this may coincide with a crack-like feature. Differential cooling post-welding leaves a residual stress field, which will be of yield value, in the material.

6.

Punching

(b)

Punching is commonly employed as a cheap means of making holes. However, cold working increases both the strength and hardness of steel and if the working is too great (e.g. in trying to punch through thick steels) micro-cracks can be generated around the hole edge. Such cracks have been known to propagate explosively in zones of plastic stress and hence seismic codes forbid punching in regions of anticipated ductility demand. Likewise, trying to bend steel through too tight a radius can create cracking; crack avoidance thus requires the specification of minimum radii and/or application of heat.

7.

Fabrication control: welding

(c)

In short, welding can simultaneously add the three risk factors most feared: a crack, a susceptible microstructure and a builtin high tensile stress (it is those same residual stresses that explain welding distortion or the lower buckling capacity of welded columns). It is thus vitally important to make all welds to a proper qualified procedure and to carry out post-weld
18

Properly controlled welding procedures should prevent the Northridge effect. Given a particular steel chemistry, its carbon equivalent value (CEV) a measure of the steels ability to harden under a heating/cooling cycle can be evaluated. For steel with CEV . 0?41, weldability is decreased, meaning that the risk of cracking in the heat-affected zone (HAZ) increases. HAZ cracking can occur if the hardness exceeds about 350 HV and if sufficient hydrogen is present. Welding procedures (for example BS EN 1011-1 (BSI, 2009)) take account of the CEV and the amount of heat applied during the welding process. They also account for the thickness of the steel (which governs heat flow away from the weld) collectively to assure that the weld and its local HAZ have appropriate engineering properties.

Delivered by ICEVirtualLibrary.com to: IP: 130.88.85.64 On: Wed, 28 Sep 2011 22:58:11

Forensic Engineering Volume 164 Issue FE1

Cracks in steel structures Mann

Depending on the various factors, it may be a requirement to apply pre-heat or post-weld heating to control the rate of weld cooling and so assure adequate weld/HAZ ductility. A further measure of protection is to specify post-welding NDE and the amount required is typically covered by specifications such as the National Structural Steelwork Specification for Building Construction (NSSS) (BCSA/SCI, 2010). All welds should be examined visually and to an extent determined by the designer but it is normally not commercially viable to examine every weld by NDE and it is not possible to examine fillet welds internally at all. Hence, control of the welding processes is the first line of defence. In assessing NDE scope and allied acceptance criteria, the design engineer has to have regard to the service conditions of the weld, especially if these in any way involve fatigue loading. The NSSS standards for weld inspection are inadequate for cases where fatigue loading operates since much higher standards are required to guard against the existence of a flaw that can grow in length under alternating stresses (fatigue). In all circumstances, the objective is to ensure that no weld goes into service having a built-in crack-like defect. There are certain types of cracking that can occur during the welding process itself; again, these ought to be eliminated by an appropriate procedure. A root cause of weld cracking during welding is hydrogen entrapment. To avoid this, certain steels require the use of low-hydrogen electrodes and the elimination of all moisture (by baking the electrodes) that might be a source of hydrogen. Because hydrogen cracking can be delayed, specifications such as the NSSS (BCSA/SCI, 2010), define cooling times in hours before any NDE can be carried out. Overall, although potential risk factors promote cracking, this is not a real issue in the vast majority of structural applications provided that the basic steel is of adequate quality and proper weld procedures are applied and appropriate NDE is specified as a quality control measure. Nevertheless, more caution is required for thicker steels, especially if there is any form of fatigue loading.

Figure 4. Example of fracture at a re-entrant corner

stress concentrations, and the internal edge can be hard and notched giving the classic conditions for fracture. It is for these reasons that the NSSS requires control of cutting procedures. Codes such as BS EN 1090 2: 2008 defined edge hardness standards to be achieved (BSI, 2008). Modifications to edge hardness are possible by altering the cutting speed (slower speeds are beneficial) and possibly by pre-heating (to slow the cooling rate or cut down thermal gradients (mandatory in rail cutting)) or the application of post-weld heat.

9.

Cracking in fatigue

8.

Fabrication control cutting

Another process that requires heat is thermal (flame) cutting. If the cutting is uncontrolled then, just as in welding, the local microstructure next to the cut can be hardened and so embrittled (as in the failure of the jumbo sections noted in Section 4). Worst still, the cuts will not be uniform and can leave notches. Figure 4 shows a crack propagating from a standard connection re-entrant corner of a beam end. The loading conditions and asymmetrical shape of the T section at the end of the long stalk provide a high tensile stress at the reentrant corner. This would be further enhanced locally by

Perhaps the most likely form of cracking to be encountered by engineers is that from fatigue. For a crack to progress there has to be an initiation site and it can be assumed one will always exist, certainly where there has been welding or thermal cutting. There also has to be an alternating tensile stress, preferably with some kind of stress concentration, to provide the frequent input of energy required for crack progression. Figure 5, which is not unlike Figure 4, shows a fatigue crack growing from a very badly cut edge. The fracture surface is also shown in Figure 6. Note the part of the crack starting from the steel edge has the characteristic beach marks of a fatigue progression then, after a certain length, the crack has propagated rapidly; this is shown by the change to cleavage in the fracture surface appearance. It is not difficult to show the effect of crack length on rates of progression: adding a nick into the side of a sheet of paper will provide ample demonstration. In the example shown in Figures 5 and 6, the alternating stress was linked to a supported structure enduring wind oscillation. A problem with fatigue is that the worst failures often occur when the designer has not spotted that a fatigue condition
19

Delivered by ICEVirtualLibrary.com to: IP: 130.88.85.64 On: Wed, 28 Sep 2011 22:58:11

Forensic Engineering Volume 164 Issue FE1

Cracks in steel structures Mann

of an issue in recent years simply because floor spans are getting longer and these lengthened spans have moved floors into the realms where there can be dynamic response to imposed human loading. Cases to be concerned about are crowd loading from dancing or aerobics, or perhaps stadia response. Cases are known of where cracking has developed under these conditions from beam re-entrant corners. Technical issues that increase the risks from fatigue loading (and thence cracking) are as follows. (a) There is a low stress limit below which fatigue cracking will not occur; however, at higher stresses, fatigue life is finite. Life is determined not so much by the maximum stress but by the stress range acting on a component. Moreover, life is inversely proportional to the cube of the applied stress range. Thus, any enlargement in stress range has a dramatic effect on life, shortening it considerably. In fatigue-designed structures, the stress range has to be kept low. It should be obvious that if secondary stresses or stress concentrations are neglected, a structure can easily be subjected to a high stress range when under fatigue loading. The consequences then are that structural lives might be measured in hundreds of cycles rather than thousands or millions. It is unfortunate that secondary stresses, which can be mostly ignored in statically loaded structures, really count in fatigue-loaded structures; such stresses are almost impossible to compute with accuracy yet fatigue cracks respond to the true stress state. Cracks can thus develop in all sorts of surprising circumstances. There has to be a tensile stress state. But if there is a residual tensile stress in the structure yet that part is in the compressive range of external loading, then an applied compressive stress will merely cycle the tensile stress from high to low rather than the other way around. Thus fatigue failures can occur in apparently compressive zones. An added complication occurs when corrosion and fatigue combine. Where there is corrosion (or fatigue in a corrosive environment) lives can be shortened even more and there may be no endurance limit even at low stress.

Figure 5. Fracture at a re-entrant corner

exists. In the case of the oil rig failures mentioned earlier, the varying loads were linked to varying sea states. Oscillation from wind-induced vibration might be quite hard to anticipate but a number of failures have occurred in this way. It is not the frequent application of wind that counts so much as the possibility of a large number of cycles from harmonic oscillations such as may be generated by vortex shedding or other wind dynamic response phenomena. Historically, repeated loading from human footfall has rarely presented a structural problem. However, it has become more

(b)

(c)

(d)

Figure 6. Fracture surfaces

The uncertainty of stress conditions leading to cracking was illustrated by Senior (1963; 1964), who reported on fatigue cracks developing in crane girders due to imperfect fit between flanges and webs or between rails and their supports. Likewise, fatigue cracks have developed on the underside of bridge decks when stiffeners have not been forced into full contact prior to welding (Mann and Morris, 1981). Fretting fatigue is a phenomenon of fatigue crack growth generated simply by two parts rubbing together. Mann (2008) described a complex mechanism that caused fatigue crack generation and cracking (with subsequent train derailment) in a railway track.
Delivered by ICEVirtualLibrary.com to: IP: 130.88.85.64 On: Wed, 28 Sep 2011 22:58:11

20

Forensic Engineering Volume 164 Issue FE1

Cracks in steel structures Mann

Checking for fatigue is an obvious condition on structures carrying traffic. This is very pronounced in tracks carrying rollercoasters where down forces can be linked to accelerations up to 4g and where there are multiple cycles of load applied through the many train wheels and with trains passing every few minutes all day long. The condition is so onerous that track design is dominated by the need to consider fatigue loading and such tracks have a defined life as a consequence (Figure 7).

10. Avoidance of fatigue damage


Experience suggests that prediction of fatigue damage is far from certain: the stress loading conditions are just not predictable with sufficient accuracy because secondary stresses and stress concentrations play such a significant role. Any defined life in terms of cycles that may be endured should therefore be treated as an estimate. The first defence is to be aware that a fatigue loading regime exists and to define joints to suitable codes. There must also be appropriate NDE before joints are put into service. Finally, an inspection monitoring regime with the objective of detecting cracks before they progress far enough to become dangerous must be carried out throughout the service life. In advanced structures such as aircraft components, certain parts are time limited in use because of the dangers of fatigue crack propagation.

Stainless steels used for structural purposes (generally austenitic grades) have strengths comparable to normal structural steels. Beneficially, they are ductile and extremely tough even at low temperatures. In fact, the toughness can be so high that certain grades can be used for cryogenic applications. Nonetheless, stainless steel is not immune to problems if the wrong grade is used. One reported problem is stress corrosion this occurs if an inappropriate grade is used in the presence of warmth and chlorides. High stress is required, but this can be just from residual stress or from stress concentrations. Failures have been reported from structures above swimming pools. Problems have been known in the UK (in wire hangers). Page and Anchor (1988) report on a failure (with fatalities) in Switzerland in 1985 when, after 13 years of service, a heavy concrete ceiling fell over a swimming pool. The collapse resulted from the failure of austenitic stainless steel ceiling supports. In 2005, Scoss issued a fresh alert about this type of failure (Scoss, 2005). Avoidance involves the identification of risk environments and the use of an appropriate stainless steel grade. Problems associated with galvanising have also been reported. The spontaneous fracture of four galvanised Macalloy bars in a post-tensioned bridge in 1968 caused a major alert (NCE, 1990), with the problem being put down to hydrogen embrittlement. More recently, a worldwide investigation has been carried out on Liquid Metal Assisted Cracking (Figure 8). This is an apparently rare phenomenon (Moore, 2005), or at least one not yet widely recognised, where for some reason steelwork cracking occurs while a part is in a galvanising bath. A significant worry is that the cracks may become filled with zinc and thus not be apparent until load is applied in service

11. Materials
The risk of low-temperature brittle fracture is controlled by specification of a steel with adequate toughness. The risk of weld-promoted cracking is partly controlled by reference to the steels chemistry. The risk of fatigue cracking is not related to the material per se; indeed, the risk is the same for all steels since the rate of crack propagation is not much influenced by steel type.

Figure 7. Fatigue failure on a rollercoaster

Figure 8. Liquid metal assisted cracking

Delivered by ICEVirtualLibrary.com to: IP: 130.88.85.64 On: Wed, 28 Sep 2011 22:58:11

21

Forensic Engineering Volume 164 Issue FE1

Cracks in steel structures Mann

and the cracks open up. Scoss has issued two advisory notes about this danger (Scoss, 2004; 2006) and the British Constructional Steelwork Association, in conjunction with the Galvanisers Association, has issued guidelines for inspection and identification of risk factors (BCSA/GA, 2005). Further guidance is available from the Steel Construction Institute (SCI, 2006).

Steelwork: An Approach to the Management of Liquid Metal Assisted Cracking. BCSA, London, Publication 40/ 05.
BCSA/SCI (British Constructional Steelwork Association and Steel Construction Institute) (2010) National Structural Steelwork

12. Summary
For the vast majority of structures, the risk of steel cracking is very low. But if it does occur, the consequences can be catastrophic. To protect structures, the first line of defence is always to use steel with the right ductility and toughness values. This might be overlooked since these attributes do not appear in the calculations. Other risk factors are welding and fatigue loading. Thick steel is an added risk factor, as is the case if a structure needs to operate in cold conditions. Where a combination of these risk factors exists, a prudent designer will also consider the consequences of component failure and try to build in a measure of structural robustness. In everyday structural work, designers are faced with two main lines of work design of new structures and appraisal of existing structures. For new structures, in the absence of fatigue loading, selection of a grade of steel appropriate to the operating conditions will suffice as an adequate precaution provided that control of fabrication procedures is maintained (welding, cutting and punching) so as not to degrade material properties or introduce cracks. For existing structures (both relatively new and ageing structures) the biggest risk is one of degradation through fatigue crack propagation, the danger being that cracks progress slowly to a critical length and then propagate extremely rapidly with little warning. Even if structures are designed against fatigue loading, experience shows that there is considerable uncertainty in life prediction. Safety therefore depends on in-service inspection with this perhaps focusing on locations identified by the designer as potential crack initiation points and with NDE techniques appropriate to the location and type of cracking that might develop. Old structures may well not have known Charpy values or indeed may contain material with no specified Charpy value. Assessing their capability in low-temperature environments is highly specialised and appropriate expert advice will need to be sought. All assessments should be guided by knowledge of what forms of failure are possible and this paper has described the major ones along with influencing factors.
REFERENCES

BCSA/GA (British Constructional Steelwork Association and Galvanisers Association (2005) Galvanising Structural

Specification for Building Construction, 5th edn. BCSA, London, Publication 52/10. Berenbak J, Lanser A and Mann AP (2001) The British Airways London Eye Part 2: Structure. The Structural Engineer 79(2): 1928. Boyd GM (1970) Brittle Fracture in Steel Structures. Butterworths, London. BSI (British Standards Institution) (2005) BS EN 1993-110:2005(E): Eurocode 3. Design of steel structures. Material toughness and through-thickness properties. BSI, London. BSI (2008) BS EN 1090-2: 2008: Execution of steel structures and aluminium structures. Technical requirements for the execution of steel structures. BSI, London. BSI (2009) BS EN 1011-1: 2009: Welding. Recommendations for welding of metallic materials. General guidance for arc welding. BSI, London. Burdekin FM (1999) Gold medal address: Size matters for structural engineers. The Structural Engineer 77(21): 2329. ENR (Engineering News Record) (1986) Cracks, fractures spur study. Engineering News Record, 21 August, pp. 1011. ENR (1987a) John W Fisher. Pressing for research to advance technology and prevent failures. Engineering News Record, 19 February, pp. 4050. ENR (1987b) Engineer-fabricator cooperation a must. Engineering News Record, 7 May, pp. 1011. Gordon JE (1978) Structures or Why Things Dont Fall Down. Pelican, London. Mann AP (1993) The structural use of stainless steel. The Structural Engineer 71(4): 6068. Mann AP (2008) Learning from failures at the interface. Proceedings of the Institution of Civil Engineers, Civil Engineering 161(2): 815. Mann AP and Morris LJ (1981) Lack of Fit in Steel Structures. Construction Industry Research and Information Association, London, Report no. 87. Maranian P (1997) Vulnerability of existing steel framed buildings following the 1994 Northridge (California USA) earthquake: considerations for their repair and strengthening. The Structural Engineer 75(10): 165172. Maranian P (1998) Correspondence. The Structural Engineer 76(11): 227228. NCE (New Civil Engineer) (1990) Stressing success. New Civil Engineer, 26 April, pp. 18. Moore D (2005) Technical note: rare problem can be easily solved. New Steel Construction 13(8): 2829. Page CL and Anchor RD (1988) Stress corrosion cracking of

22

Delivered by ICEVirtualLibrary.com to: IP: 130.88.85.64 On: Wed, 28 Sep 2011 22:58:11

Forensic Engineering Volume 164 Issue FE1

Cracks in steel structures Mann

stainless steel in swimming pools. The Structural Engineer 66(24): 416. SCOSS (Standing Committee on Structural Safety) (2004) Liquid Metal Assisted Cracking of Galvanised Steelwork: A Rare but Important Issue. SCOSS, London, SC/T/04/02. SCOSS (2005) A Reminder of the Risk of Failure due to Stress Corrosion Cracking in Swimming Pools. SCOSS, London, SC/05/69C. SCOSS (2006) Liquid Metal Assisted Cracking. SCOSS, London, SC/06/59. SCOSS (2010) Major Cast Metal Components. SCOSS, London, SC/10/029. Senior AG (1963) The design and service life of the upper part

of welded crane girders. The Structural Engineer 41(10): 301312. Senior AG (1964) Discussion of The design and service life of the upper part of welded crane girders. The Structural Engineer 42(3): 8793. SCI (Steel Construction Institute) Steel Industry Guidance Notes. SN10 11/2006: Galvanising Structural Steelwork Guidance for Engineers on how to reduce the Risk of Liquid Metal Assisted Cracking. SCI, Berkshire. Swann W (2005) Is your steel tough enough? (specifying steel to BS 5950-1: 2000). The Structural Engineer 83(21): 2427. TWI (The Welding Institute) (1971) Brittle Fracture of Welded Structures. TWI, Cambridge.

WHAT DO YOU THINK?

To discuss this paper, please email up to 500 words to the editor at journals@ice.org.uk. Your contribution will be forwarded to the author(s) for a reply and, if considered appropriate by the editorial panel, will be published as discussion in a future issue of the journal. Proceedings journals rely entirely on contributions sent in by civil engineering professionals, academics and students. Papers should be 20005000 words long (briefing papers should be 10002000 words long), with adequate illustrations and references. You can submit your paper online via www.icevirtuallibrary.com/content/journals, where you will also find detailed author guidelines.
Delivered by ICEVirtualLibrary.com to: IP: 130.88.85.64 On: Wed, 28 Sep 2011 22:58:11

23

Вам также может понравиться