Dramatic innovations and developments have occurred in civil and structural engineering in recent years. Difficulties of analysis which appeared insurmountable only twenty years ago have largely disappeared with the advent of the mainframe computer and the finite elements method; new generation microcomputers now increasingly provide such analyses with great convenience and economy. The engineer today has more time to devise new forms of construction, to improve design details, and to allow for phenomena and data which were previously overlooked or approximated. Much of this new expertise has been used to improve the design of ships and aircraft, offshore platforms, subway systems, high-rise towers and buildings, and many other forms of construction previously designed by rules-of-thumb and simple codes of practice. There is now much more internationalism in engineering too, with design methods and codes becoming more standardized, and large computers providing technical literature and patent information from all over the world. There is a need for these advances to be presented to an international audience by leading engineers of international repute; this is the purpose of the new Civil Engineering Series by Chapman and Hall ‘The first of the new series is by Professors G. Ballio and F. Mazzoleni of Milan and Naples respectively. Both authors have imposing lists of publications in this area, both are seasoned in theoretical and experimental research, both are eminent authorities in Italy, in Europe and on the international scene. They have the problems of the steelwork designer at heart and are dedicated to helping him by explaining and improving current practice. Their approach is fundamental and general, with instances of many varied examples of systems, joints and components. Complete design examples, however, are not included and should be sought elsewhere in professional journals and periodicals. Such examples, at their best, demonstrate conceptual thinking and ingenious application, but all of them ced to be sound and economical in their detailed anatomy, a feature which this book sets out to establish and improve. I believe that every structural steel designer with pride in his own competence ought to find time to read and absorb this substantial contribition to his understanding and expertise. E, Lightfoot Oxford FORTHCOMING TITLES Probabilistic Methods in Structural Engineering G, August, A. Baratta and F, Casciati ‘Underground Structures P.S. Bulson, THEORY AND DESIGN OF STEEL STRUCTURES Giulio Ballio Dipartimento di Ingegneria Strutturale Politecnico di Milano, Italy Federico M. Mazzolani Istituto di Teenica delle Costruzioni Bate belibiceiaceeied 6e.01L/BALA AH => ait ’ ‘ .= Sti ae := on DRG. OAT. = ieee = IN. TWIGS adr efvsly de (wie LONDON NEWYORK Chapman and Hall FEUP DEP_ENG MECCHAPTER FIVE welds 5.1 TECHNOLOGY OF WELDS ‘the practice of connecting steel plates ty welding started at the beginning of this century when it was recognized that inconveniences associated with the use of rivets could be avoided if the parts were monolithically Joined after local melting. ‘The most obvious ifficulty lay in the fact that heating had to be Limited to prevent modification of the micrographic structure of the plates, thus making them unfit for use "The electric are waa therefore adopted, providing a concentrated source 1 a remarkable quantity of heat, At first two carbon Glectrodes were used set very near to the pieces to be joined (Zrener). Next, the arc vas formed between one carbon electrode and the piece to be welded (Sernados), and finally the caxbon electrode vas Feplaced by a ‘metal wire coated with inowlating material (Kielberg, 1908), this new connecting technique rapidly spread throughout. the USA and then also in Europe. Laboratory tests have alvays confirmed its complete reliability. Subsequently, technology has led to the Gevelopaent of many other welding procedures xelated to the nature of the pieces to be joined and to the use for which they are intended. Currently, about forty different welding systems are available to melt different constructional requirements |1,2|. 5.1.1 Plane cutting steel plates before welding are normally cut by means of an ony-acetylene or oxy-propane torch, which provides the heat source heeded to prodice a concentrated high temperature and a jot of oxygen, Waich generates with the metal a strong isothermic reaction capable Of melting the resultant oxide. Tf the torch advances automatically at a regular specd and if conygen purity is sufficient, the oxy-cut surface is regular enough to enable welding to be carried out without further preparation. Using special shapes to guide the torch, curved cuts of any desirea form can also be obtained, The torch can be inclined when necessary to obtain oblique cuts to form the grooves required between the edges 153‘THEORY AND DESIGN OF STEEL STRUCTURES to be welded. Cutting with ionized gasos (plasma cutting), due to the extrenely nigh temperatures it can reach, obtains good results also in cases where oxy-cutting is inadequate. 5.1.2 Welding processes Autogenous fusion welding processes vary substantially according to ‘the applied heat source and to how the molten pool {a protected against the atmosphere. (a) Oxy-aeetylene welding. the heat source consists of the oxy-acetylene flame produced by the combustion of acetylene (C2H2) with oxygen. These two gases are fed to the torch separately, mix together inside it and energe at its tip, where the combustion takes place. ‘The flame forming at high temperature (about 3100°C) algo produces reducing gases (CO and Hz) which protect the pool. This process ~ the first to be adopted industrially ~ is now much less common than in the past. (b) Shielded metat aro process. (Pig. 5.1a). ‘he heat source is the electric are which struck betwoen the electrode and the parent metal and the heat developped causes the quick melting of both. ‘The melting of the coating of the cylindrical electrode (Length: 350 to 450 mm.) produces sone gases which protect the area aurrounding the arc and the weld pool. Depending on the composition of the coating, the most comsion types of electrode can be basic (excellent mechanical and metallurgical characteristics), acid (good mechanical characteristics), or cellulose (special uses, sich as fixings). This process is applicable to a wide range of structural applications: (c) Sulmerged are procese. (Pig. 5.1b). The heat source is again the fare struck between the electrode and the parent metal. ‘The electrode consists of a continuous wire wound on a drum or spool. An automatic Feeding device makes it advance at the sane speed at wich it is ‘crested zone | Fig. 5.1 154 weLos arc area protection is provided by a granulated flux distriputed fon the joint to form an accumulation inside which the arc strikes, thus being subrerges and invisible, This process is highly productive. (@) Ga ektelded metal ave process with consumable electrode (MIG and MAG). This too ie continuous wire process in which arc protection is provided by an inert gas (e.g. argon: MIG), or by a chemically active gas (e.g. carbon dioxide or a mix of the latter with argon: mac) . ‘Fool dimensions are smaller than in the previous case and productivity ie lower. The HAG process is used to weld mild steel and lowealloy. steel. (©) Gao shielded metal are process with non-consumable electrode (1G). ‘The heat source still consists of the arc. It strikes between a tungsten element and the parent wetal. Arc area protection is Provided using argon. ‘he only purpose of the electrode is to allow fre formation, ‘The veld metal mist be introduced separately in the form of rods, ao in the oxy-acetylene process. This process is used co weld materiale where the weld pool is easily oxidizable, such as stainless steel and aluminium alloys. (f) Electrostag welding. The parts to be welded are mounted vertically, with a given distance between them, Copper sliding blocks are set actos the gap so that a kind of vertical ‘melting-pot' (having @ rectangular section) is thus obtained. In its lover part a molten flux is provided with particular electrical characteristics, nto which the continuous wire electrode sinks. Both the electrode land the parent metal are connected to a current generator. According to the Joule effect, the current passing through the molten flux keeps it at a sufficiently high temperature to welt the edges of the parent material and the wire. The copper sliding blocks are prevented from fuelting by an internal water circulation aystem. ‘his process offers very high productivity and is used for vertical welding of very thick steel parts, with one only weld pass. 5.1.3 Consequences of metallungie phenomena ‘ere are, essentially, two metallurgical phenosena |1,2|+ poliditication of the material that has nelted during he various weld passes and heat trestnent of the parent aetal around the weld Welding is characterized by snall masses of metal, rapidly melted and rapidly cooled due to heat absorption by the surrounding aetal. Te therefore consists of thermal cycles with a high rate of cooling which produces zones of high hardness, expecially in the material Close to the weld. ‘hese hard cones in the parent aetal near the weld can lead to so-called cold cracking (Fig. 5.28) The origin of such cracks 4e actibed to the hydrogen absorbed by weld material in the molten state end by the edjacent area of parent Material Brought to high temperature. This hydrogen generally cones from the arc. atnoaphere of the coated electrodes, except in the case of electrodes with a basic costing |31-‘THEORY AND DESIGN OF STEEL STRUCTURES ” Pig. 5.2 To prevent this fault, the thermal cycle should be moderated by appropriately pre-heating the part to be welded and using basic electrodes. ‘These precautions are strongly recommended whenever steel having a high elastic Limit is being welded, as very hard areas could, be easily formed. Another dangerous phenowenon is that of hot evacking which can ‘occur in the molten area whenever the content of parent material 1s high (Pig. 5.25). Three cracks fora during solidification because of impurity segragations thickening in preferential zones of the molten area and solidifying at a lover temperature than steel, entailing material Aienociation dve to shrinkage stresses, ‘The netallurgical structure of molten areas is strictly related to its mechanical characteristics, particularly toughness. hen welding is performed by means of a few large weld passes, recrystallization is coarse and toughness low. It can possibly be corrected ising special reactive powders. hen, however, welding is performed in many weld passes of Limited section, a high toughness results, partly due also to the useful: normalizing action which each pass produces on the proceding one. 5.1.4 onsequances of thermal phenomena Welding processes are always accompanied by a development of a great. quantity of heat, waich causes considerable therm-plastic deformations in tho connected steel elements |1,4], due to the lowering at high teaperatures of yield limit and of elastic modulus (see 4.4.4) . ‘the physical nature of such phenoene can he demonstrated by reference to the model of a bar of length L having in the middie a welded section J, (Fig. 5.3). Consider the limit of deformation due to transverse coftraction and the internal stresses induced by totally preventing contraction. 156 eins es0’c Cz aor WA el As a consequence of cooling from the temperature after welding (650 °C) to anbient temperature (20 °C), the weld metal undergoes a transverse contraction AL. According to some experiments on steel plates 14 mm thick transverse contraction (or shrinkage) is of the order of about 208 of the initial length, s.e. AL = 0.18 L. If from the beginning the bar is prevented from shortening, an axial force J arises, It can be evaluated according to the following compatibility condition: ot where £ is a mean value of the elastic modulus for the consideyed temperature range (650 to 20 °c). If one assumes £ = 2.1 x 10? N mi’ and Eq = 0.75 £ then: zy 6 2 Ho 0 = Bx 0.18 x 0.75 x 2.1 x 10° = 280 x 107-2 (vw Such a stress could reach the yield Limit when p,/1 is of the order of magnitude 1/100. Sunsing up, this schematic example shows that, as a thermal consequence of welding, either relatively inportant defornations take place or internal (residual) stress of considerable intensity is induced. Both these deformations and stresses are perpendicular to the seam and, if the deformations are not prevented, transverse shrinkage Bhenomena take place, producing angular deformation of the welded parts (Pig. 5.40) - Deformation also takes place along the bean but this longitudinal shrinkage cannot take place entirely freely, because it is prevented by the elasticity of the surrounding cold parts and therefore Longitudinal stresses are always present (Fig. 5.40) - The reador is referred to 4.2.2 for a study of longitudinal residual stress in welded olenents, here only mention is made of most common Systems applied in building practice to keep deformations and residual Stresses due to welding within reasonable limits, In the preventive phase, they are: init{al pre-bending, clamping the parts in the seam ) 157TTREORY AND DESIGN OF STEEL, STRUCTURES \ steal plate a »» 2 Fig. 5.4 deposition phase, pre-heating, and a rational study of welding sequences. ‘Once the welding has been carried out, concentrated heating can he applied to the distorted parts, or the whole fabrication can be stress relieved by subjecting it to low heating in a furnace to a uniform temperature overall followed by very slow cooling. 5.1.5 Welding defects and controle |5,6,7,8| The worst defect which can affect a weld is the presence of cracks (see 5.1.3). Other defects include blow holes which are metallurgical defects due to unforeseen reactions vhich arise in the molten pool because of operational imperfection and slag inclusions, which are generally caused by inproper shape or sequence of weld pass. A further possible defect is lack of penetration due to the failure of the weld metal to penotrate and fill the root of the weld at the vertex of the grooves (Fig. 5.5a) or at the centre (Fig. 3.5b). ‘The vertex of @ V-shaped or x-shaped or any other setting is always that part at which the edges to be welded are closest and it is not so easy to obtain a regular and complete melting of the shaped edge by the first weld pass and faulty edge preparation cen only uake watters worse. the defect can be eliminated if it is possible to back gouge and apply a sealing pass. a. rUhUD a ») Fig. 5.5 158 | Another defect ie sticking, i.e. molten filler material being deporited on the parent uetal before it has melted, so that an oxide coating is interposed between the edge and the melted area. This takes place in welding processes having low or poorly concentrated thermal input . ‘The most comon way of detecting internal welding defects is by X-ray examination, sing the X-rays produced by special radiographic ‘equipment, or by Y-raye spontaneously released by some radioactive Internal defects, such as porosity, blow holes, slag inclusion, lack of penetration, cracks, appear as darker stains on the film. ‘They ace interpreted by comparing them to sample defects, correspond~ ing to radiographic standards set out in codes or proviously agreed uupan as a condition for welding approval. : Controls can be performed also by means of @ thin ray of ultrasounds produced by a probe containing a quartz crystal, which is vibrated at its fundamental frequency by means of a variable high frequency electic field. ‘The probe i# laid on the surface of the part. to be examine’, in order to transmit a series of ultrasonic impulses to it. Those impulses can be reflected from obstacles consisting of other surfaces of the specimen or defects. By moving the probe along the specimen, if there 2 no defect, clearly distinct Luminous vertical Lins appear on the screen of a cathode ray oscillograph. These Lines correspond to the pulse at the top surface and the echo fron the bottom. if instead the beam meets a defect, intermediate Lines ‘appear on the screen, from which the location and size of the defect can be assessed. vention should also be made of some simpler methods, such as magnetoscopic (magnaflux) examination and examination by penetrants. ‘The former consists of two wire-bearing poles which when put in contact with the weldment create a magnetic field which, by means of magnetic poxders, allows detection of cracks near the surface. Liquid penetrance, because of their low surface tension can penetrate even into very narrow superficial cracks, which are practically invisible to the naked eye. Their presence 1s detected’ by leans of a thin coat of a revealing Liquid applied over the area after the surplus penetrating liguid has been resoved from the surface. 5.2 STRENGTH OF WELDED JOINTS Strength verification mainly refers to two different kinds of joint (Fig, 5.6): butt wolds and fillet welds. 5.2.1 Bubt velda In a butt weld which is free from internal defects and axially loaded perpendicularly to ite own axis, the stress state can be considered equal to that of a homogeneous piece (Fig. 5.7). Consequently, the criterion generally adopted by modern codes is to assume its resistant cross section #8 having an area given by the weld length tines the connected thickness. ‘The corresponding strength fcan be considered equal to or lower than that of the parent material, 159‘THEORY AND DESIGN OF STEEL STRUCTURES BUTT WELDS FILLET WELDS. 9 tty — 3" Jo : : o= < z Fig. 5.6 according to the value assumed for the efficiency coefficient, defined fas the ratio between the design strength of the weld f, and'that of ‘the parent metal f,. One therefore assuse: faye "wa Such a! criterion derives from the allowable stress method, as it was first defined as the ratio between the allowable stress of the weld o,, aq and that of the parent material o_,_. the" #8ffous codes assune a value varying £288'0.6 to 1 for 160 WELDS coefficient n,. This is related to the technological parameters characterizing weld joint perforsance, including the degree of control, welding execution (in the shop or on site), the type of joint (eith of without back gouging) . Briefly, the efficiency coefficient n, reduces allowable stresses in welds in order to compensate for lack“of control and as an indirect precaution against presumed but unknown defects. rt is therefore obvious that statistical meaning can be given to it. Tt can be considered only as an empirical coefficient justified by experience. For this reason the IIW (International Institute of Welding) , and later the ECCS (European Convention for Constructional steelwork), have recently suggested the adoption of an efficiency cooffictent ‘equal to one, at the same tine recommending suitable tests for the qualification of elders and of the welding process before the actual, work 1s begun, together with the adoption of adequate inspection and control methods, in order to have a reasonable degree of assurance that welded joint quality be sufficient to restore at least original strength of the parent metal. If an efficiency coefficient of 1. is assumed, it Ss obvious that. strength verifier cion for a butt wold isunneceseary as st ie superseded by parent metal control in general, the calculation method for a butt welded joint can be based on the following assumptions. ‘The resistant cross section of a weld is the longitudinal one (enroat section) having as its length the entire welding length and as its height the smaller of the two connected thicknesses, measured In proximity to the welding in the case of butt wolés (Pig. 5.6a), oF else the thickness of the completely penetrated element in the case of ‘Teshaped or cross-shaped joints (Fig. 5.6b). ‘The following stresses axe considered to act on the throat section (Pig 5.60,8): ©, iS the tensile or compressive stress acting perpendicularly to the throat section T te the shear stress acting parallel to the throat section axis G, 1s the tensile or compressive stress acting parallel to the sean akis, referred to the cross eectionaf the weld Joint (1.8. parent etal and weld) perpendicular to such axis on these bases, atrength verifications are carried out adopting the sane yield criterion used for the parent material, i.e. the Hencky-Von Mises criterion: Boia went Mya VO + oh - 89 + SOT 5.2.2 Millet welde Many studies have examined the problem of checking the strength of a fillet weld once the stress atate created in it by external loads is known ‘The various calculation methods are all based on the simplifying hypothesis that stresses are uniformly distributed within the sean throat section. ost of then consider the throat section overturned fon one side of the sean, 161i ‘THEORY AND DESIGN OF STEEL STRUCTURES ” » a A ° 0 Fig. 5.8 In 1961 180 approved a calculation method proposed by IIW which faseuned the throat section in ste actual position ae the resisting section. ‘When this method first appeared, it was considered appropriate progressively to replace all other methods as the IS0 recommendation sould be adopted by the various constituent Countries. However, Jt seemed to be too complicated for practical use and some important codes, suchas that of the AISC (USA, 1969) and DIN 4100 (Germany, 1966) Zolloved simpler calculation methods considering the throat section as caverturned on one of the two sides of the sean, 5.2.2.1 Definition Agsune the resistant cross section of a fillet weld to be the throat. Section, given by the production of the throat depth times the seam effective length. ‘Throat depth 2 is the smallest length of the triangle inscribed in the sean cross section (Pig. 5.8). The effective length Lcoincides with the overall length of the sou, provided that, obviously, faulty ends due to tailing off are eliminated. Furthermore, the following stress components are taken into account: (a) stresses referred to the throat section, overturned on the plane of either side of the seam (Fig. 5.98) n, is the tensile or compressive stress acting perpendicularly to the plane containing one side of the seam, referred to the area of the throat section, overturned on the same’plane 4, 1s the shear stress acting perpendicularly to seam axis, lying on a'plane containing one side of the seam, referred ta the area of the throat section, overturned on the seam plane 4) 46 the shear stress acting parallel to the seam axis, lying on a 162 weLDs Fig. 5.9 Plane containing one side of the seam and referred to the area of the throat section, overturned on the sane plane () stresses referred to the throat section in its actual position (Pig. 5.95) 6, is the tensile or conpressive stress acting perpendicularly to the ‘throat section, considered in its actual position 4, is the shear stress acting perpendicularly to the seam axis, lying ih the throat section, considered in its actual position 3, is the shear stress acting parallel to the seam axis, lying in the tikoat section, considered in its actual position (c) Stress referred to the cross section of the seam (Pig. 5.9e): Gg, 18 the tensile or compressive stress acting parallel to the sean tilts on its cross section 5.2.2.2 Strese state in a fillet weld Whereas stresses in a butt weld are very close to those in a continuous piece, ina fillet weld the actual stress distribution in the plane of the cross section of the veld is considerably more complicated. The stress state changes fron one point to another and - especially at sean vertex and edge ~ considerable stress peaks are Prosont, due to the sensible daviation of the stress lines passing From one piece to the other (Fig. 5.10). However, as for butt welds, Stresses in fillet welds ere usually considered uniformly distributed in the throat section, Substantially, this assumption is based on satisfactory ductility and toughness of material, which are checked 163‘THEORY AND DESIGN OF STEEL STRUCTURES Fig. 5.10 by means of appropriately related control of welding material and qualification tests of the welding process ‘The authors consider that stress dy has no influence on the behaviour of welded joints because it 1s not present in the weld - the weld itself just connects the various parts together - but it is due to the external action undergone both by the weld and by the connected elements, In support of this, consider a fillet welded I-beam, The feaas are loaded by shear stresses (Ty) which act against the sliding between flanges and web and therefore represent the stresses due to the connecting function. Stresses due to bending under external loading act on the overall I-beam cross section and in the weld, they are stresses of the type 0) « opinions have been divided as to whether it is or is not necessary to take the stresses 0) into account in the strength calculation of a fillet welded joint and the problem vas strongly discussed some time ago in both ITW and 150 Comittess, sone countries wore in favour of taking 0, into account because it was on the safe side and because it led to @ calculation method which was uniform with that for the parent metal; other countries were against it. The latter prevailed because experinental date justified thelr view and oj was ignored. Further tests have shown that in fact even high values of , 40 90 reduce the load bearing capacity of welded Joints in the case Of static losde and of ductile and tough materials. ‘The most important codes, ©.g. DIN 4100 (Germany, 1968), ATSC (USA, 1969) and BS 153 (Great Britain, 1966), as well as Prench, Italian, Dutch, Hungarian and Russian recommendations ignored o, in the calculation of the strength of fillet welds. 164 was O(N mari?) for point Vi timate ’ : ® specimen 5 ( TP Nnat i _ Vandepere t specimen for eal 8 peroid : suas ‘Van Der E> ‘Specimen fexperimental range Fig. 5.1 5.2.2.3 Recommendations and calcalationmethods |1,9,10,11 Biperinontal bases. Many tests were carried out to anlyse fillet weld strength ac stresses varied, in order to trace their ultimate surface in the space of co-ordinates 0,, 1,, ty, referred to the throat section in its actual position: Mond the earliest tests, mention should be made of those on Fillet wolde only perpendicularly and tangentially loaded (Vandeperre, Belgiua) and those on fillet welds undergoing internal actions in any Girection in the plane perpendicular to the seam axis (Van dor Eb, Holland, 1952-53), ‘The ultimate surface corresponding to Van der Eb's feats was called ‘peroid’ (Pig. 5-11) 165‘THEORY AND DESIGN OF STEEL STRUCTURES A few yoars later (1968), an international series of tests on sillet welds was prompted by the 15th Commission of Ii, with the participation of France, Belgium, Holland, Ttaly, Sweden, Canada, Japan and the USA, A study of the corresponding results influenced the choice of the various calculation methods. 180 caleutation method. ‘This method is based on the principle of considering an ultimate domain for the fillet velds which is as close fas possible to the rupture "perold',but able to be expressed in an equation. ‘The proposed equation (7, , being rupture stress referred to the sean throat section) is: | °" a 2 Fw, representing an ellipse having its major semi-axis coincident with tthe upper semi-axis of the ‘peroid’ section in the plane 0, t, and the "peroid’ semi-axie as its minor semi-axis. by rotating such an ellipse around its major axis (C,), an ellipsoidal surface is obtained, as the ultimate domain, which conveniently replaces the "peroid', if the mall part in compression is neglected, where the ellipsoid is more conservative. ‘the equation of the ellipsoid of revolution is therefore: 2 2 2 Fw, gw, 9 sts eqatisyenabies one" to doternin yhother a plary~nial stress ate Shececterised by generic yates of Ope dey aa aecepeae or not. Now 140,752 = 1.8 and if this is adopted, the equation (1) Sines ine weet f1ction caution ve Prom a formal point of view, it is convenient to express the verification condition in an analogous manner to that adopted for the parent metal. For this purpose, the first tem of the above relation can be used as equivalent 'ideal'stress: aan 3 Og “Wo? + s.0t? + 1) 6.3) 6.0) 3 +e e th Styy (5.2) The second term of (2) indicate the ultimate tensile strength of the fillet weld. ‘This depends on electrode quality, which must be chosen in relation to the type of steel of waich the elements to be welded are made. In general, thexefore, the ultimate tensile strength Of the weld f, (conventionally assuning , = 1) can be expressed a5 2 tunction of tie Gosign strength fa of the*parent metal io eet es wherein fi, is the efficiontly coefficient of the fillet weld. The general virification, therefore, becomes: 166 = wens o,,\or+ e+e) < ya Ver + a eS BS, 6.) 150 has recommended this formula in the more general fora: aes ON rhe ee) 6.6) where paraneter k, has not been fixed, so that the codes of the various countries are fret to adopt different values of K,, according to the chosen safety factor in shear. Sone codes adopt KY'= 3, in order to maintain, also for fillet welds, a combined street formula similar to that of the Muber-Henky-von wisde criterion already used both for the parent metal and for butt welds. tn this case, the formula becomes 2am cape 2 og hosed <8 ty 6.0 representing an ellipsoid having semi-axis ty = = 0.580. Further tests have shown that the ellipsoid of ultimate strength cannot be considered as 2 solid of revolution, semi-axis 1), being longer than T. A generalization of the ISO formula (5.6) vas thezefore proposed as foes ace 5. Sq "VO Re * Ry 2th $ Bfa ee wnere K, /i, » 18 about 3/2. Pormla (7) therefore becomes: ecrertrecea Og Ms a BS 6.9) representing an ellipsoid having semi-axes ty = 0.760, and tT, = 0.580, ‘This fornula is closer than the ISO one to experimental results, but it still requires considerable calculation time to determine stresses 0,, T,, Tj, Which refer to the throat section in its actual position (Fig. 5.95). Formilae referred to the overtumed throat section. combination formilae are less complicated to apply when they are in terms of stresses n,, t, and ty vhich refer to the throat section overturned on either of’ ené tho eldes of the seam (Fig. 5.98). Tt is obviously possible to express design formalae (5.7) and (5.9) also in ters of f,, %, and ty by rotating the reference axes, bat his would entail considerable formal complications. The relations between the components d,, t,, Ty and the components m,, t,, ty, in the assumption that the Sean has'an isosceles triangle cross suction, are the following: ete 5 ms tL - I/2 emer te ty eile wet 1 tet, nt Fu net Formulae (5.7) ,(5.9) therefore become: at + ogg “Wain? 2 22) + 302 - ant, < Bg 6.7) 167‘THEORY AND DESIGN OF STEBL, STRUCTURES Dl Sega ag Na? + t+ 208 — One, < 8 6.0 ‘The fastest way to simplify calculations consists in interpreting the ultiate limit surface by means of a solid the equation of which does not vary if the reference oxes rotate through 45°, wich 42 equivalent to overturning the throat section on one side. ‘The solid having this property is the sphere and ite equation can be written in the following form in terms of O., Ty Ty: -1 (5.10) , ¢ or else, using n,, t, ond ty components: 2 mG ttt (5.10%) Caw? OP? OF) Value yf, ,, Indicates the sphere radius and the value of ultimate strength tn“tHe most favourable direction for the material i.e. for Ty. These formulae obvicusly do not take into account the different behaviour of the weld in respect of axes T, and 1). Equations (5.10), (5.10"), according to (5.4) , give the following verification conditions: yg "$V ee eR wan) Seamer Aetna cat, ja Although with different radii spherical strength domains have been adopted by the following codes: British (BS-153, 1966), assuming X = 0.58 american (ATSC, 1969), assuming x = 0.61 German (DIN 4100, 1968), assuming x = 0.70 Out ophere. when the CAR-UNI 10011-67 Recommendations vere revised [|p the faeieuto Italiano della Saldatura rigorously studied the probien of simplifying fillet eld calculation, starting from a Becength domain of spherical type, in order to determine stresses acting on the throat section nore quickly. One, however, sought to maintain the interpretation of te different. behaviour of shear Stresces T, and ty, but remaining on the safe side. This was done by using aout 9phere’ (Big.5.12a), le. a sphere having the sane radius a2. the German sphere (DIN 4100, 1968), but cut by two pairs of planes perpendicular to ates 0, and T, and passing by points 4, = 0.56f, and ©, = 0.587, +.” he’ ‘cut sphere’ section with plane 3,, 1, ts Allow in'rig. 5.128!" ror the equivalent stress 0, the lover 8f the following vaites most be assumed: 168 weLDs 6.12) 6.13) comparison among the vavioue formilae. Fig. 5.13 shows the sections in the two planes (n, ~ t,) and (nm, ~ ¢)) of the above strength donains and, in particular, those defined by 150 formula (5.7"), the generalised elipsoid (5.9'},the spherical formula (5.11") with X= 0.58 (Gp), x = 0.61 (USA), x= 0.70 (DBR), and the cut sphere (5.12), (5.13). ‘The British sphere te internal to’ the ISO ellipsoid (5.7") and (5.9") and it is everywhere safe. ‘The German sphere is tangential to the generalised ellipsoid (5.9") in the plane n, , £, but intersects sein the plane 4, , {, at the bisector Lines of the four squares. consequently, the Gérman sphere is less penalising than the British Pig. 5.13 169‘THEORY AND DESIGN OF STEEL STRUCTURES fone with reference to joints loaded according to axis G, (i.e. perpendicular to the throat section) and it correctly interprets the strength of joints loaded in the T, direction, but is is unsafe in the { direction. The ‘cut sphere’ is tangent to the generalised ellipsoid (5.9") in the plane n,, ¢ and practically always internal to it in the plane ©, + thus, while Less penalising than the British sphere, it is always safe; it is however stricter than the 150 formulae and the German sphere along the 0, axis, 5.3 EFFECTS OF INTERNAL ACTIONS AND STRENGTH VERIFICATIONS 5.3.1 Calewlation methods ‘The foregoing section shows that the various calculation methods for welded joints can be classified substantially, in two groupe [10]: those using an ultimate strength domain having the shape of an ellipsoid and those usinga spherical ona. ‘the out sphere’ method |11] can be considered as a variant of the latter class. ECCS Recommendations suggest the adoption of the 8, values given sn the table of Fig. 5.14 as a function of yield trols 7, of the material of which the velded elements are made. ‘the sane Recommendations allow the following strength domains to be adopted: Ellipsoid 150 with k, = 3 and thus formlae (5.7) or (5.7") Sphere having its radius oqual to fy,y/¥3 = 0.58f, ,, and thus formlae (Sel) oF (S-11") with 20.58 5.3.2 Jointe under tension In the case of side welds (Pig. 5.15), the throat section of each seam can be considered overturned on the seam side lying in the contact plane of the two welded parts. ‘The tvo throat sections of each pair of seams therefore 1ie on the sane plane and undergo shear stress which can be considered as uniformly distributed in the resistant section, according to the basic assumption. Therefore, considering the overturned throat sections, one By fy 143 5265 as Loss Le Ss 1.00 es ig. 5.14 170 weLos Fig. 5.15 hast : Pe th" Ta in this case: Bla = ata, In the case of front welds (Fig. 5.16a), the throat section can bbe considered overturned on the plane of either side of the sans. Depending on whether the throat section is overturned on x or y axis (Pig. 5.18), one has: FP wy axis) ny fe Gani) nO tage Por the calculation of joints with inclined welds (Fig. 5.17), a (good approximation is obtained if only bi-axial stresses are taken into account in the throat section. ‘The throat section is overturned on the seam side which is set transversely to the direction of losding and the load F is resolved Ante components = F sin 8 and V= 8 cos 0, Then Zeer V__Vcos 8 0" Ba” ater Pig. 5.16 am‘THEORY AND DESIGN OF STEEL, STRUCTURES Fig. 5.17 5.3.3 Joint under combined bending and shear ‘The Joint of Fig. 5.16 is undergoing a shear force and a bending ver 3) MER the resistant section of weld lies in the vertical plane and consists of two rectangular sections corresponding to the overturned throat sections having dimensions a,x hyand therefore it has an area A= 2ah and section modulus W = 2ah7/6. Maximum stress due to the bending moment is: Me 3 thaw be Ay hae 9 _ | VA e “e—— 3 a Ba 4 a ' a | B 4 Fig. 18 Pig.19 im ELS his result is correct on the basic assumption that the neutral axis of the joint bisects the sean length. ‘the shear force is generally considered uniformly distributed on the throat section. ‘This assumption gives: 2 °° Dak If there are horizontal welds (Fig.5.19), first consider the effect of bending monent N= FL alone. Bending stresses act on the resistant section obtained by overturning the throat sections of the two seams fon the side lying in the plane of the joint. In this case, 1f ah is the section modulus of the overturned throat sections and this can be considered ogual to that of two concentrated masses, i = bah. Bending stress is therefore: M_ OW” bak The Joint is also subject to a shear force V = F producing tangential stresses that can be considered uniformly distributed over ‘the throat eection, Hence: F > be A tipical example of a combination of vertical and horizontal welde is the bean-to-coluan connection shown in Fig. 5.202. calewlation can be carried out as for the preceding types of joint, provided that the various parts of the joint are of more or lese the same stiffness and fillet weld dimensions are adequate for flange and web thicknesses Of the I-beam to be connected. The resistant section is comprised of the various throat sections of all the seans, overturned on the joint plane, and it must bear a shear force Y= F and a bending monent: M = FL. bending shear joa b) 2 Fig. 5.20 173{THEORY AND DESGIGN OP STEEL STRUCTURES Conventionally it supposes that the shear force is carried by the web seams alone, the tangential stresses being uniformly distributed over their throat sections and having a value (Fig. 5.20c) by = FPagly Bending stresses can be evaluated by considering the overall ares of the throat section corresponding to seam ‘A’ 'B' and 'C! so that ‘the maximum stress value is: "max = MUP Strength checks mst be made for the most highly stressed parts of the joint. They are the external flange seams ‘A’ and the ends of ‘the web seams 'C'. ‘Another simplified calculation method separates the effects of bending and shear, by assuning that the web seams alone resist the shear forces and the Flange seams resist only the bending mouents. 5.3.4 Jointe under combined toreion, bending and shear ‘Me various methods for calculating torsionally Loaded welded joints are gonornily based on simplifying assumptions. "The two Host Gomoniy epplied ones re the ‘polar nosent" method and the "=> forces" nethod. ‘he ‘polar sogent! method extends to the case of fillet welde the principles of material strength theory, which are strictly valid only for circular sections. Ail the throat sections are overtuened on the aide lying in the plane of the connection. ‘the centrold of the overturned sections 4s assumed as the contro of rotation and the polar movent of inertia 1. of the overturned throst sections is evaluated with reference to ‘Bat pointe (rig. 5.21). he maximus ioad 5,., ts thus supposed to act at the farthest sean point srom Gy in a aiBBeion perpenaicaier to radius 7.) Joining Geo G. The value of S,., therefore, is: te Snax ~ —F {T being the torsional moment. By resolving $. in two directions, one parallel to and the other normal to the aiz¥8Eion of welding values for t, and ¥, ,can be Gbtained and used in strength checks. On the overturned throat Section one haa: tS, 0088 1 ty =H sin ‘The ‘to forces’ method considers the torsional moment T as replaced by a couple given by two opposing forces # acting in the parallel seams, so that, for rotational oguilibrium Hh ta) =P In the seams these forces produce only a tangential stress: ty = 0, = lao 7 Because of its simplicity , this method ie used to ilustrate the following few significant oxamplos. In the case of side welds (Fig.5.22), torsional moment P= Fa is replaced bya couple of force # acting at the middle of the seam sections. Since ~as is usual in practice - @ «hy H = Fe/h. ‘The two forces # produce tangential stresses in the seams expressed by A. Fe 7 La” ila ‘The value of the stresses due to shear Y= F, also considered uniformly distributed in the seam, is: ee ‘0 ia In the case of front welds (Pig. 5.23), the torsional moment Fe is replaced by the couple formed by the two opposite forces V acting con the seams, so that V = Fe/z. Fig. 5.22TTHEORY AND DESIGN OF STEEL STRUCTURES Fig. 5.23 "These two forces produce tangential stresses in the seams, equal tee hs te hs ta? ae ‘The tangential stresses due to shear force P can also be considered uniformly distributed in the welds, their value being. nek Ua In the case of the combination of two aide welds and one front weld (Pig. 5.24), the torsional moment Fe produces tangential stresses in the two side Seams of pe Pig. 5.24 176 weLos ar fata As side wolds are less fit to bear shear forces, the direct shear action from F can be entrusted only to the front weld. Its value is: tr ee ee snore: A is the area defined by the sogments axially bisecting the sean ‘throat sections, overturned on the connection plane @ is seam throat depth. IE the section is not completely welded on its perimeter, the connection can be considered as a pair of vertical welds and a pair of horizontal welds. The torsional moment can than be distributed proportionally to naximn moments 2, and, which each pair of welds can bear: Tras + ys tagkok as T=, + 7), one can assune r te "Ha + Fla, eae Fig. 5.25 7‘THEORY AND DESIGN OF STEEL STRUCTURES Arce Pig. 5.26 in all the seams. (Suffixes 1 and 2 refer to vertical and horizontal welds respectively) . TE the above is extended to the case of Fig.5.26, a good ‘approximation of the effect of shear / = F is obtained by considering stresses ¢, ue to F in the horizontal velds as nil, the shear being entirely absorbed by vertical seams. Hence, for the vertical seams: : Fe PR. ST anEs aie * ab, ‘and for the horizontal seams; Fe ty = WORE ale ahh + ab Tt is thus possible to balance the external forces by means of stresses alone. t bs ® CATT 14, b) °) Fig. 5.27 176 wets Cross sections such as angles, tees and cruciform shapes (Fig. 5.27), in which the shear centre Ls at the intersection of the contre Lines’of the eleneats forming then, have a low degree of torsional stittness, Nonent of inertia for uniform torsion (see 8.2.5) for the welding resistant section, ag in the case of open profile section formed by thin-walled elenents, is expressed by. es era thee here the sean is extended to the various parte of the overturned throat section having dimensions a, xf, Tangential stresses due to uniférm téesion vary linearly with the thickness a of each sean of the resistant section and they have & Girection parallel to the seam ario. Meir maxim value ier rd ‘umn * Tone land it occurs in the seam having the thickest throat. section. [REFERENCES 4, Costa, G., Dada, 1. and Mazzolant, F. M. (1977) Colleganenté saldait, ¢xsia, milan 2. Pratt, J. L. (1979) Introduotion to the Velding of Structural Stestsork, Constrado, Croydon 3. Granjon, H. (1972) La formazione ai cricche 4 freddo nella salda~ tura degli acciai, Aivdeta Italiana della Saldatura, 3, 159-172 4. Daddi, T, and Mazsolani, P.M. (1972) Deverminazione sperinentale delle imperfesioni strutturali nei profilati ai acciaio, Costrustont Netaliiche, 5, 374-394 Costa, G. (1976) Principali difetet dei giunti saldati, La Netaliungia Italiana, atti notizie, 101-107 6. McMaster, R. C. (1959) Non-Destructive Testing Handbook, the Ronald Press Company, New York 7. Welding Tnspection (1968) Anerican Welding Society, New York 8. Xarren MeGonnagie, J. (1963) Nou-Destiwotive Testing , Gordon and Breach Science Publishers, New York Blodgett, 0. w. (1966) Design of Welded Structures , The Janes F. Lincoln Are Welding Foundation, Cleveland, Chic 10. 12H = Commission xV (1976) Design rules for arc-elded connections in steel submitted to static load, Welding in the World, 4, a. 5/6 Al, Guerrera, U. (1971) Boperimenté alla base delle madifiche alle Horne CWR-UNT 10011-87, ALlegato III al Doo. CUS-UNIMET (GL), 6 9 179