38

Heavy-welded Structural Fabrication J L Pratt BSc(Eng), MIEE, FWeldl formerly Research Manager, Braithwaite & Co. Engineers Ltd

Contents 38.1

Welding processes

38/3

38.2

Weld details

38/4

38.3

Weld defects

38/5

38.4

Distortion 38.4.1 Correction of distortion

38/7 38/7

38.5

Assembly

38/8

38.6

Stud welding

38/9

38.7

Testing

38/10

38.8

Significance of defects

38/10

References

This page has been reformatted by Knovel to provide easier navigation.

38/11

The vast majority of fabrications in steel are now welded and it is rare to see a new fabrication that is joined by any other method. The steel used is generally to BS 4360 Specification for weldable structural steels; the revised 1986 edition includes the weathering steels and the whole range covers steels with yield stresses ranging from 230 to 450 N/mm2.

Electrode holder

38.1 Welding processes Figure 38.1 shows the welding processes most commonly used in steel fabrications; in all cases an arc is struck between the electrode or electrode wire and the workpiece resulting in a high arc temperature which melts off the electrode and deposits it in the joint which has to be made. The manual metal arc (MMA)1 is the most common process and the electrode is deposited manually with the operator controlling the direction of the weld and the build-up of the weld metal. The flux extruded around the core wire when melted by the heat of the arc provides a gaseous shroud which protects the molten pool and arc from atmospheric contamination and controls the weld metal reactions; it can also be the vehicle for supplying certain alloying constituents to the weld metal. The fused slag around the deposited weld metal also helps to form the weld bead shape. There are several types of electrode coverings which function in different capacities and are classified in BS 639 'Covered electrodes for the manual metal-arc welding of carbon and carbon manganese steels'. Gas-shielded arcs with bare wire or cored wire, can be of the semi-automatic or automatic type. 1^3 The semi-automatic process utilizes a power source, a wire drive unit, incorporating the necessary control units, and a 'gun' which is held by the operator and manipulated manually; the wire is driven through a flexible tube to the gun and a suitable designed nozzle concentric to the gun orifice supplies the gas to the arc. The automatic process usually has a heavier 'gun' or head with the wire (also known as electrode wire or feed wire) fed directly through the gun without the intervening flexible tube; the whole apparatus travels automatically for longitudinal welds or may be stationary for circular fabrications. Higher welding currents and deposition rates are generally used with subsequent water cooling of the head being necessary. The weld metal and arc is protected from the atmosphere by the shroud but a bare wire must contain deoxidizers such as silicon, manganese and sometimes aluminium; these are necessary to prevent some oxidizing processes which occur within the arc atmosphere. A cored wire has the flux enfolded within the electrode wire as typified in the cross-section shown in Figure 38.1; it may be used in semi- or fully-automatic processes. The necessary deoxidants are carried in the flux which may also be the vehicle for additional alloys to be added to the weld; the flux allows for higher welding currents than that in solid or bare wire shielded welds, with the slag allowing better bead shapes to be obtained, and is generally more tolerant to rusty plate conditions which could otherwise lead to porosity. Shielding gases used for structural steels are carbon dioxide or argon with addition of oxygen or with carbon dioxide and oxygen, the cheapest being carbon dioxide. Another type of semi-automatic welding popular in the US and now being used in the UK is the self-shielded arc where the continuous electrode contains in its core flux ingredients which vaporize in the arc, shielding the arc and forming a thin slag around the metal droplets as they transfer across the arc gap; deoxiding materials also form part of the flux. The process requires no gas shield and is therefore better-suited for outdoor operations when windy conditions prevail. The submerged arc (SA) is a process which feeds a bare wire into the arc and the arc is covered by a granulated flux which is

Electrodewire Current contact tube

Core wire Flux

Welding current in

Electrode Parent metal \

Drive motor

Feed rolls

Gun

Welding current in Gas

Flux cored electrode Gas shroud

Slag

Thin slag

.Welded metal Gas Shielded Arc

Manual Metal Arc (MMA)

Electrode wire Drive motor Nozzle Welding current in Current contact shoes Flux Figure 38.1 Common welding processes

To flux hopper Feed rolls

Submerged Arc

also automatically fed; some of the flux is melted to cover the weld pool as slag and to provide the arc with a gaseous shield. Again alloys can be added to the weld either via the arc or the flux; very high currents can be used in this process1 and very smooth bead-contour shapes can be obtained. The arc is completely shrouded by the flux and thus it cannot be seen; this gives a total absence of arc glare but, correspondingly, guiding is that much more difficult. This process is more susceptible to rusty or dirty plate conditions than MMA but less susceptible than metal arc inert gas (MIG) and for very heavy weld metal depositions on thick plate, multiple electrodes may be used in the same weld. A semi-automatic form using a small diameter wire can also be obtained. Figure 38.2 shows schematically two other processes, electroslag and electro-gas welding; both are completely automatic. In electro-slag welding, the plates to be welded are mounted vertically with the edges of the plate square or unprepared; watercooled copper shoes are mounted either side of the weld seam to contain the molten metal. An arc is struck on the starting block with a little granulated flux added to the weld pool; as the wire or electrode burns off, the temperature of the slag bath increases and the slag becomes electrically conducting; from then on the electrode protrudes into the bath, the arc extinguishes, and the wire metals off due to the I2R heating of the current. It is thus not an arc process but a continuous cast process used on plate thicknesses over 25 mm and certainly up to 450 mm; for the greater thicknesses three electrodes are fed

Copper shoe support arm

Copper shoe support arm Feed motor

Feed motor

Feed rolls

Feed Electrode Electrode rolls wire wire Direction of Direction of Parent motion of motion of metal' cooling shoes cooling shoes and feed head and feed head ,mechanism mechanism Copper Copper cooling cooling shoes shoes Water Water Water Water Molten Molten' Molten Welding metal slag metal arc bath Weld Weld Electrogas Electroslag Figure 38.2 Electro-gas and electro-slag processes simultaneously into the slag bath with, in one application, the whole assembly oscillating across the width of the joint. There are two methods of applying this process known as electrode or consumable guide (see Figure 38.3); in the electrode method the feed head is at the side of the plate being welded and moves up with the copper codling shoes as the weld is made. In the consumable guide method the feed head is stationary at the

Drive motor

Drive rollers 'Electrode wires

Direction of oscillation

top of the joint to be welded and the wire(s) are fed down to the slag bath by a consumable guide which is insulated from the workpiece by fusible spacers. As the wire(s) melt so does the bottom of the consumable guide and the copper shoes can be stationary of a length equal to the length of the welded joint; this method requires less sophisticated machinery than the former and is therefore, where it can be applied, cheaper. It cannot obviously be oscillated across the width of the joint. The electro-gas process is similar to that of electro-slag welding in that the weld metal is contained by watercooled shoes but different in that the weld metal is deposited by a true arc with a thin slag from the flux in the cored electrode; the weld metal and arc is protected by a stream of CO2. Since both processes can evolve large heat inputs to the heataffected zone of the weld and to the weld itself with resultant large grain microstructure, poor fracture toughness may result. Some improvement may be obtained by postweld normalizing treatment, but where fracture toughness may be a problem, expert opinion is advisable.

38.2 Weld details The two main types of welds used in fabrication are the fillet and butt welds. Fillet welds are shown in Figure 38.4; BS 54004 and 4495 lay down that the allowable stress in a fillet weld is based on the throat thickness, t, or '0.7O x L" where L is the leg length, because t for a stated leg length L will vary according to the included angle, /, between the fusion faces. Table 38.1 gives the values of t for varying angle y.

Table 38.1 Parent metal Copper cooling shoe (both sides of joint) Slagbath Weld pool

Direction of motion of copper shoe

Angle between fusion faces

60 to 90°

91 to 100°

101 to 106°

107 to 113°

114 to 120°

Factor by which fillet size is multiplied to give throat thickness

0.70

0.65

0.60

0.55

0.50

•Solidified weld metal

Starting block Drive rollers Electrode wire: Drive motor

Consumable guide tubes

Parent metal

Stationary copper cooling shoes (both sides of joint)

Figure 38.3 Electro-slag welding process

Some typical butt welds are shown in Figure 38.5; these are generally for manual welding.4 The root-run is usually backgrooved (except where a backing strip is used) so that clean weld metal from the previous root is obtained (Figure 38.6); this ensures homogeneity of weld metal at the root area. These same preparations may be used for semi-automatic welding with no root gaps where root gaps are shown, or with a root run of manual weld to seal the root before applying any semi-automatic process for the rest of the weld. The root run on the second side does not generally require back-grooving since the penetration is enough to ensure weld metal homogeneity. Submerged arc welding is a high-deposition welding process with deep-penetration characteristics although with a direct current electrode negative power source the burn-off or deposition rate increases with a large diminution in penetration; multiple wires or electrodes may be used in the same weld with

Asymmetrical Concave Symmetrical Figure 38.4 Fillet welds. Note: minimum length of both legs to be measured for L, for concave weld f^0.7 L Single V butt

Backing strip

the electrodes sharing, in parallel, the same power source or each electrode connected to a separate power source. Weld preparations for such a process are infinite and reference should be made to the suppliers of electrodes and fluxes for their advice; for notch ductile materials, basic fluxes and an increase in the number of runs may be necessary.

Double V or double U butt 38.3 Weld defects Some typical weld defects are shown in Figure 38.7.

Single U or single J butt

Undercut Undercut

Single or double bevel butt

Undercut

Inclusion Figure 38.5 Typical butt welds. Note, angles and dimensions of root gaps and root faces may be altered to suit welding technique and position of weld, the above being suitable for flat-position welding. Welding is carried out from both sides of all preparations except where a backing strip is employed. To achieve complete penetration, back-gouging (back-grooving) may be employed

Slag inclusions

Side wall Root (penetration) Lack of fusion

Side wall Root Linear porosity

First root run Lack of fusion Porosity (linear) Figure 38.7 Some weld defects Sound weld metal Back groove into interface after first (root) run grooving Figure 38.6 Butt weld back groove

'Undercut.' A groove melted into the parent metal at the toe of a fillet weld or root of a butt-weld - produced by the arc but left unfilled by the filler metal. Undercut may be due to incorrect angle between the electrode and the workpiece, too high an arc voltage or travel speed or scaled parent material.

'Porosity.' Due to dirty or rusty parent material surface, damp consumables, arc instability (as evidenced by the stop or starting of the arc when using MMA), gas entrapment from air due to inefficient shielding gas, grease on filler wires. When linear can denote 'lack of penetration'. 'Lack of penetration' for butt welds. Inclusive angle of prepared faces too small to allow the electrode to get at the root, insufficient current to penetrate the landing or landing too small for the set arc parameters or root gap too small to allow more penetration. 'Lack of fusion.' Incorrect manipulation and angle of electrode to ensure side-wall fusion, or root fusion.

affected by the heat input of the weld and whose microstructure and physical properties might be affected by that heat. This zone is rapidly cooled by the mass of the surrounding parent metal and if this cooling rate is high enough a hardened (martensitic) microstructure may be formed. Cracking may develop in this hardened structure (see Figure 38.9) owing to: (1) the alloy content of the parent material increasing; (2) high cooling rate; (3) restraint and therefore higher residual stresses resulting from the weld contraction; (4) stresses within the microstructure due to the transformation to a hardened structure; (5) the presence of absorbed hydrogen from the weld diffusing into the HAZ when that weld cools and contributing to the creation of micro fissures; and (6) for fillet welds, where the fit-up is bad with root gaps.

'Slag inclusion.' Nonmetallic solid material entrapped between runs of weld metal or between weld metal and parent metal. Due to inefficient clearing of the slag between each run which in turn may be due to wrong weld parameters giving the wrong-shaped interpass bead and positional requirements. 'Spatter.' Small metallic particles ejected from the weld area and forming on the parent material adjacent to the weld. Spatter varies with the arc process and within that particular process may increase or decrease with differing arc parameters. 'Hot-cracking' (solidification cracking). Cracks appearing in the central region of the weld (Figure 38.8) where segregation of sulphur and phosphorus, the lowest melting-point constituents of the weld metal, occurs; thus at temperatures in the region of the solidus thin films of the liquid segregates occur along the grain boundaries (intergranular). The weld metal may thus become susceptible to cracking because of the high shrinkage stresses generated during the cooling of the weld metal. The effect of sulphur may be reduced by obtaining a higher manganese : sulphur ratio in the weld metal whilst at the same sulphur and phosphorus levels an increased carbon content may cause cracking; likewise silicon. Weld metal on its own is low in all the above elements but high 'pick-up' or dilution may be derived from the parent metal; thus any deep-penetration welding process could lead to hot-cracking. A weld bead or nugget whose depth is greater than its width can, in such deeppenetration processes such as the submerged arc or gas-shielded arcs, promote such cracks when the above metallurgical conditions are marginally operative. In this case a wider preparation or the use of more than one run of weld with lower current values would reduce the dilution factor and the depth:width ratio, to decrease the risk of cracking. Guidance is given in BS 1535 on such cracking. 'Cold-cracking' (underbead or HAZ or hydrogen-induced cracking). The heat-affected zone (HAZ) of a weld is that (generally) narrow zone in the parent metal adjacent to the weld bead

Hot crack Hot crack

Figure 38.8 Hot-cracking in SA welds

Combined thickness, t = a + Ib for joint shown

Crack

Weld 'nugget'

HAZ

Figure 38.9 HAZ crack

In (1) the presence of alloys in increasing amounts increases the hardenability in the HAZ and their effect can be related to that of carbon by the following carbon equivalent formula: Carbon equivalent % = c% + Mn % + 0 ± Mo ± V % +Ni + Cu %

O

J

IJ

Thus any increase of carbon equivalent due to the increase in any of the above alloys will increase the hardenability of the steel. This formula only applies to those steels in BS 4360. In (2) the cooling rate is assessed partly by the combined thickness t of the joint being welded (Figure 38.9) and partly by the heat input from the weld and any given preheat. The total heat input from any arc may be expressed as: __ . , . a r c voltage x current (amps) XT H (joules/mm) = —— where T is the time in seconds to deposit L mm of weld. Lower / and higher H lead to a lower cooling rate in the HAZ with a less hardenable microstructure. In (3) the restraint increases with the stiffness of the components making the joint. In (5) the hydrogen content can be reduced by using a low-hydrogen process, e.g. hydrogen-controlled MMA electrodes. The MMA electrode may have to be baked to reduce its hydrogen content to the lowest level possible; in SA welding the flux would have to be dry and

preferably of the agglomerate rather than fused flux. With all automatic wires or electrodes no wire drawing compound contaminates should be present; gas-shielded arc processes with solid wire could prove to give weld metals with the lowest hydrogen content. Preheat curves necessary for combined thicknesses and size of weld deposit (and thus heat input) are given in BS 5135. Preheat, when applied, reduces the rate of the cooling of the weld and allows more hydrogen to be evolved by the weld metal to the surrounding atmosphere; therefore to be effective it must be applied to the correct temperature and over a sufficient width of the plate. British Standard 5135 indicates that the width of the preheat zone on each side of the weld should be at least 75 mm in any direction from the joint preparation. In practice the temperature is measured by using thermo indicating crayons or paints, the former melting and the latter changing colour when the correct temperature is achieved, and to make certain that the heat has penetrated the full thickness it is customary to heat the far side of the plate with the temperature indicator on the near side, or by heating the near side until the required temperature is indicated on the same side for 2min for each 25 mm of steel thickness after the heat source has been removed. Although the heats applied are generally low (on the average 10O0C), the wide area over which they are used can lead to more distortion than that of the weld itself; it is therefore preferable to use a higher heat input weld source to reduce the preheat required. It is also more economical.

38.4 Distortion

Fillet weld

Contraction of fillet welds Depth of bow due to distortion

Bending of web Contraction across weld.

Contraction across welds

Transverse ,bowing of flange Welded T sections 'Cusping' due to distortion of Maximum contraction across transverse weld on long plate widest width of weld

distortion Transverse butt weld Stiffeners Welds Dish Plate

Boxweb flange Panel distortion

Figure 38.10

Distortion due to welding is dependent on the heat input from the weld; such heat is concentrated in a narrow zone around the weld area. The subsequent contraction of the heated weld metal and parent metal produce undue stresses in the fabricated part; if unrestrained the fabrication will distort and, if restrained against distortion, residual stresses up to the yield point of the material may occur. The parts being welded may in themselves have residual stresses due to their shape and size and thus their manufacture; these stresses or some of these stresses may be relieved or increased by the local welding heat and thus their distortion due to welding may be difficult to predict. Metals with differing expansion coefficients, thermal conductivities and physical properties will produce different distortion levels with the same weld heat input. Figure 38.10 shows distortions for typical welds. Figure 38.11 shows joint preparations, welding procedures and some typical plate presetting to compensate for weld distortions. For the heavier type of fabrication it is generally better to fabricate all subsections prior to incorporating them into the main structure but to control the increased distortions for thin-walled constructions it may be preferable to assemble and tack the whole assembly to give a much stiffer structure more able to withstand distortion. 38.4.1 Correction of distortion For a dished plate (e.g. a dish resulting from an area of plate welded all round the periphery of that area on one side of the plate only) the amount of dishing resulting from such a weld depends on the heat input of the weld and the thickness of the plate. To flatten such an area, spot heat from a heating torch can be applied in several places within the dished area on the outside (convex side) of the bulge; this will increase the amount of dishing on heating but on contracting that side will shrink and reduce the dish. Heat can be up to red heat (600 to 65O0C) but does depend on the thickness of the plate; for very thin plate the applied heat may heat both sides to an equal temperature

Flange

Packs Prebending

Presetting

Prebending Small preparation welded first

Presetting

Asymmetrical preparation

Figure 38.11 Methods to reduce distortion resulting in equal contractions on both sides of the plate with no decrease in the dish. Triangular heating on the web and bar heating on the flange of a plate girder or section (Figure 38.12) will increase the camber and can also be applied, within certain limits, to box sections. It is important to note that heating the flange and not

Resultant camber from heating

Weld

Weld

Weld

Weld

Bar heat on flange Triangular heat on web Figure 38.12 Correction of camber the web may shrink the flange and increase the camber of the girder but the web, not being heated, cannot shrink to accommodate the increased camber and may therefore buckle. Angular bending of a flange plate due to the two fillet welds attaching the web to the flange may be corrected by heating in a straight line (Figure 38.13). The effect of introducing heat to shrink areas and to introduce distortions to reduce others, must of a necessity induce stresses into the fabrication; the effect of these stresses and the subsequent increased load on some welds must be watched carefully and if necessary those welds increased in size to accommodate the increased load. All welds or any other form of localized high heats give high residual tensile stresses local to that heat; these stresses in turn generate compressive stresses outside those tensile stress areas. Stress relieving (at about 600 to 65O0C) may relieve the structure of any induced stress but in turn must lead to increased or different distortions to accommodate the subsequent movement of the structure.

Figure 38.14 Plate girder welds stage. For thin flanges it may be found necessary to prebend the plates as shown in Figure 38.11. For crane girders it may be necessary to make full penetration welds (Figure 38.15) for the web-to-top-flange welds. When using automatic welds such as SA, care must be taken that hotcracking does not occur; this can happen when trying to achieve penetration and the dilution of the weld metal by the parent metal is high. To reduce dilution back-grooving may be used (but is difficult in this situation) or the web preparation made wider; the increase in the ratio p:w in Figure 38.15 indicates a bigger dilution of the weld metal by the patent metal and the possibility of hot-cracking increases. Hot-cracking invariably occurs on the second side of the joint to be welded since the first weld has made the web-flange assembly rigid or constrained, and it should be noted that hot-cracking may be contained below the surface of the weld and thus not be visible (Figure 38.8).

1st weld 1

Heat applied along length of welds Figure 38.13 Correction for transverse flange distortion

38.5 Assembly 'Plate girders.' These may be welded as in Figure 38.14(a) or (b) by MMA or automatic welding. Tack welding to hold the assembly together must conform to the requirements of BS 5135, with minimum root gaps; large root gaps may lead to HAZ cracks as described previously or 'burn through' when using high current density automatic welds. For girders with top and bottom flanges of differing thicknesses or with top-flangeto-web and bottom-flange-to-web welds of different sizes, different shrinkages may occur in each flange and hence alter the camber of the girder; in such cases it may be necessary to induce triangular heating as described above or to increase deliberately the camber if the web plate is cut to camber in the preparatory

2nd weld

Note wider web preparation, increased number of runs and therefore reduced dilution of weld metal by parent metal Figure 38.15 Full penetration butt weld, submerged arc 'Box girders.' These are invariably assembled on one flange as the base fabrication plate and must lie perfectly flat on the assembly stallage or a twist in the box may result; all diaphragms are then placed in position after being subassembled and the two webs then tack-welded to diaphragms and flange. As much internal welding as possible is then made before the fourth closing flange plate is placed in position and tack-welded; the four longitudinal web-to-flange welds are then made. The same comments about differing flange thicknesses or web-to-flange welds in plate girders can apply to box girders.

Damper

The choice of flange-to-web longitudinal weld detail may be dictated by the camber and or curvature required in the box. For a large box where it may be difficult to rotate during fabrication (a) in Figure 38.16 may be preferable; where there is camber, (b) is easier to assemble with the flanges outside the webs than (a) where the box-closing flange coming inside the webs can only sit on the diaphragms unless backing strips on the webs are installed between the diaphragms to maintain the closing flange profile.

Bottom flange

Solenoid

Spring

Pistol grip Current cable

Shaft Chuck Stud Ferrule CUD

Web

Control cable

Stand Foot

Figure 38.18 Stud-welding gun

Figure 38.16 Box-girder assembly Where boxes are to be jointed on site the open ends of adjacent boxes should be stiffened if there are no diaphragms close to the open ends to hold those ends to the required square or trapezoidal profile; for all such ends they would tend, without such stiffening, to have inward bows on all flanges and webs although the four corners are dimensionally correct (Figure 38.17). In boxes all stiffeners are subassembled on the webs and flanges before the main assembly is completed; to keep all resultant distortion, shrinkage, etc to a minimum it is better that intermittent welding be used on such items if the design requirements can be met with such welds. Stiffeners may be of the bulb, angle or T-type; the latter two may present difficulties for blast or other type of cleaning after welding and also for the subsequent painting.

Open end of box Elevation on X-X Figure 38.17 Box open-end distortion

38.6 Stud welding Stud welding shear connectors on the top flange for bridge girders for composite concrete decks is now a widespread practice; the diameters of the studs usually range from 12 to 25mm and generally vary from 100 to 150 mm in length although 250 mm studs have been welded. The form of the stud is shown in Figure 38.18 and the head of the stud fixes into a gripping chuck in the operator's gun which in turn is placed vertical over the spot to be welded and which rests firmly on a three-point support on the steel surface. When the trigger is pulled an electronic timing device controls the following sequence: The chuck is lifted about 3 mm by an electromagnet and

a pilot arc is formed about the tapered point (Figure 38.19) which then develops into the main arc, the main arc current being drawn usually from a drooping characteristic transformer-rectifier power source. This arc melts the end of the stud with a resultant melted area on the workpiece and after a preset time the solenoid is de-energized and the stud is plunged by a return spring on to the workpiece while the arc current is still flowing.

Ceramic ferrule Set-up Pilot arc Main arc Figure 38.19 Sheer-connector stud

Welded stud

The stud when correctly welded should be of a correct length after welding with a formed upset fillet around it with no undercut; such undercut may be due to incorrect welding parameters or arc blow and when present can lead to easy fracture of the stud from the plate surface. Arc blow because of the high, though transient, currents used may be prevalent when the studs are placed near to the edge of the plate; in such cases an edge plate to extend the magnetic field of the current in the main plate may be utilized (see Figure 38.20). Studs greater than 22 mm (p are difficult to weld, leading to erratic arcs and sometimes unsound welds; the plate surface on which the studs are being welded should be free of all oily

Stud to be welded

Earth cable

Earth cable Figure 38.20 Magnetic field edge plate

Edge plate (held or clamped against workpiece)

contaminants, millscale and deep rust. A light surface grinding in the stud area is recommended. The tip of the stud is either sprayed with aluminium or holds a 'slug' of aluminium which acts as a deoxidant when vaporized in the arc; it is important that this deoxidant is not damaged. One test sometimes employed to ensure the soundness of the steel weld is to bend some to an angle of 30° and to hit or 'ring' the others with a hammer.

38.7 Testing 'Methods.' These may include nondestructive testing methods such as X or gamma radiography, ultrasonics, dye penetrant or magnetic particle testing. Radiography is used almost exclusively on butt welds and ultrasonics on butt and some fillet welds; site welds are invariably tested by ultrasonics and/or gamma radiography employing in general iridium as the source of gamma rays. The standard and scope of testing required is usually determined by the customer and should be ascertained at the enquiry stage. Before most contracts are commenced some welding procedures may have to be approved by the customer, i.e. a weld joint simulating the thickness, preparation, etc. of an actual weld configuration used in the fabrication must be welded to prove that the proposed welding consumable and method of welding is satisfactory. Such welds may be subsequently tested by nondestructive methods and then physically tested by means of transverse tensile and bend tests, Charpy impact tests, nickbreak tests (for fillet welds) and cross-section macrostructures (see BS 709). The skill of the welders may be approved by the customer's own specific test or by BS 4872 'Approval testing of welders when welding procedure approval is not required' or any other subsequent standard with any appropriate nondestructive test requirements. 'Laminations.' Where a plate is laminated or where a plate must be tested for laminations before being incorporated in a fabrication, ultrasonics is the only method by which any such lamination may be detected and the extent measured. Material may be supplied to a standard of lamination testing by the steelmakers; the details of such standards and the appropriate costs may be obtained from the steel supplier. The effect of any lamination on the stability of the structure must be referred to the designer, e.g. the effect of a lamination in a compression member is generally more severe than one in a tension member. It is probably true to say that most structures can tolerate a fairly large degree of lamination in a member before repair is required; the repair of such a lamination is shown in Figure 38.21.

'Lamellar tearing.' This is a result of nonmetallic inclusions in the steel in the plane of rolling merging into a tear due to the stress imparted by the weld (Figure 38.22) or other derived stresses normal to the plane of inclusions. Because these inclusions are very small and scattered throughout the thickness of the material they are not detected radiographically, and, up to now, although detectable by ultrasonic inspection, they cannot be quantified to assess potential cracking. The tear, when it occurs, is of a ductile nature, fibrous and stepped or ragged as shown schematically in Figure 38.22, the steps resulting from the inclusions in different planes being joined to form the tear; the presence of such inclusions decreases the ability of the material to withstand extension under a load applied across the thickness of the plate. One destructive method of assessing the material for tearing is to machine a small transverse tensile test piece and to measure its reduction in area at failure of applied tensile load.

Contraction across weld

Direction of rolling of plate

Inclusions in plane of rolling

Lamellar tear

Lamellar tear surface (showing stepped nature)

Figure 38.22 Lamellar tearing

Joint details that can promote tearing are shown in Figure 38.23; generally the welds must be large to cause contraction across the welds on cooling to create sufficient tensile stresses across the plate to produce a tear in a susceptible material. For fillet welds to generate such a tear the weld in almost every occasion would have to be greater than 12 mm, although many materials can tolerate much bigger welds, and any details which have both large weld and imposed load stresses across their thickness should be avoided; some preferential details are shown in Figure 38.23. On some suspect material it may be helpful to reduce the risk of tearing by buttering the weld fusion face with MMA or SA welds as also shown.

38.8 Significance of defects Edge of lamination qouqed out and welded

Figure 38.21

Repair to lamination

Plugwelds

Sudden and catastrophic failures in some steel fabrications has led to the further development of Griffiths'6 classical work on linear elastic fracture mechanics (LEFM) by Wells7 and Cottrell,8 who independently proposed the crack-opening displacement test for determining the fracture toughness of engineering materials where crack propagation ahead of a crack was accompanied by a large plastic deformation at the tip of the crack. Fracture toughness gives a measure of the material's resistance to failure by cracking due to the stress intensity around the

Tear

Preferred detail Load

Tear

Load

Preferred detail

Tear

Preferred detail

Buttering MMA

Buttering

Buttering SA

Figure 38.23 Lamellar tearing. Preferred details and buttering

tip of the crack in the presence of an applied and/or residual stress in the material. The measurement of fracture toughness has led to the publication of the BS document PD 64939 which gives guidance on the methods for determining acceptance levels of weld defects. Fracture mechanics can also be applied to the growth of a fatigue crack under load-cycle conditions and the prediction of fatigue failure is largely due to Maddox10 and Gurney." Factors which affect fracture toughness of steels are material thickness, service temperature, residual stress due to welding, applied stress, material type and strain rate. Lower service temperatures, thicker material, higher-yield material will adversely affect the fracture toughness; strain rate increase will not permit the normal mechanism of fracturing by slip along the atomic planes in time, and the material behaves elastically. The total stress across a weld defect may be the sum of both applied and residual, the latter due to the stresses set up by the contraction of the weld when cold; it is not unknown for brittle fracture to occur under residual stress .only. Once the material has been selected for welding, the correct electrode must be used for yield, ultimate and required impact values; ultimately the level of defects in the weld will govern the service performance of the structure. Crack-like defects normal to principal stresses are most at risk, such as heat-affected zone cracks and lack of penetration; cracks in the outer quarters of a weld are more significant than those in the middle half of the weld. Rounded defects such as porosity and slag inclusions can be ignored in

many cases but they themselves may, if profuse or long, indicate possible cracking of some form; such profuse defects may shadow more serious defects when ultrasonic or radiographic testing is applied. However, they indicate bad welding practice and a limit should be applied to the extent of their occurrence. The fitness for purpose concept, i.e. defects are acceptable in structures if their size, orientation and type do not affect the integrity of the structure, indicate that where stress and design criteria require it an acceptable level of defects must be tabled. A typical statement that no weld should contain defects is inadmissible since all welds contain some defects, however small; with the advent of ultrasonic techniques with improved discriminating powers many small, structurally insignificant defects can be found. Repairs of such defects are not necessary and are expensive, with the possibility of reintroducing more significant defects upon repairing. Burdekin12 has proposed acceptance criteria for welded joints in crane girders and proposals are in hand for bridge structures to BS 5400. Some limited guidance is given in BS5135 related to welders' tests but to relate these tests to a structure's service requirements would be unrealistic. Design detail concepts to minimize the risk of brittle fracture include the obvious step of avoiding welded joints in areas of high stress concentration or high tensile stress and, where this is impossible, to lay down acceptable welddefect levels. Keep all weld sizes to a minimum commensurate with design criteria to reduce the incidence of weld defects and to reduce weld residual stress. Recognize that a higher-yield material will reduce the acceptable weld defect size and will not be superior to lower strength steel in fatigue life. To select the correct notch-ductile material related to material thickness and service temperature. A welded joint that has to be tested should be fully accessible to the testing method chosen and to the arc process used to make that joint; poor welding access will inevitably lead to more significant defects. To achieve defect size, location, orientation and type, the use of ultrasonics is mandatory; it lacks a permanent record (compared to radiography) and hence the skill and integrity of the operator is paramount. Generally, building frames in steel are at little risk to brittle fracture but bridges, pressure vessels and oil platforms all have their areas where care in design, correct steel selection, proven fabrication skill and erection procedures will reduce the possibility of such failures.

References 1 2 3 4 5 6 7 8 9 10 11 12

Gourd, L. M. (1980) Principles of welding technology. Edward Arnold, London. Houldcroft, P. T. (1967) Welding processes. Cambridge University Press, Cambridge. A. W.S. Welding handbook, VoI 2. American Welding Society. British Standards Institution (1978-1980) Steel, concrete and composite bridges. BS 5400. BSI, London. British Standards Institution (1969-1970) The use of structural steel in building. BS 449. BSI, London. Griffiths, A. A. (1920) Phil, trans. Roy. Soc. Series A, 221, 163. Wells, A. A. (1963) 'Application of fracture mechanics at and beyond general yielding.' British Welding Journal 573/570 (Nov.) Cottrell, A. H. (1961) International Standards Institute, Report No. 69, pp. 281-291. British Standards Institution (1980) Guidance on some methods for the derivation of acceptance levels for defects in fusion-welded joints. PD 6493. BSI, London. Maddox, S. J. (1972) Welding Institute Report No. E49. Gurney, T. R. (1979) Welding Institute Report No. 91. Burdekin, F. M. (1981) 'Practical aspects of fracture mechanics in engineering design.' Proc. Inst. Mech. Engrs., 195, 12.