TC 9-237 – Chapter 6 – Section VI


a. In this section, welding stresses and their effect on weld cracking is explained. Factors related to weldment failure include weld stresses, cracking, weld distortion, lamellar tearing, brittle fracture, fatigue cracking, weld design, and weld defects.

b. When weld metal is added to the metal being welded, it is essentially cast metal. Upon cooling, the weld metal shrinks to a greater extent than the base metal in contact with the weld, and because it is firmly fused, exerts a drawing action. This drawing action produces stresses in and about the weld which may cause warping, buckling, residual stresses, or other defects.

c. Stress relieving is a process for lowering residual stresses or decreasing their intensity. Where parts being welded are fixed too firmly to permit movement, or are not heated uniformly during the welding operation, stresses develop by the shrinking of the weld metal at the joint. Parts that cannot move to allow expansion and contraction must be heated uniformly during the welding operation. Stress must be relieved after the weld is completed. These precautions are important in welding aluminum, cast iron, high carbon steel, and other brittle metals, or metals with low strength at temperatures immediately below the melting point. Ductile materials such as bronze, brass, copper, and mild steel yield or stretch while in the plastic or soft conditions, and are less liable to crack. However, they may have undesirable stresses which tend to weaken the finished weld.

d. When stresses applied to a joint exceed the yield strength, the joint will yield in a plastic fashion so that stresses will be reduced to the yield point. This is normal in simple structures with stresses occurring in one direction on parts made of ductile materials. Shrinkage stresses due to normal heating and cooling do occur in all three dimensions. In a thin, flat plate, there will be tension stresses at right angles. As the plate becomes thicker, or in extremely thick materials, the stresses occur in three directions.

e. When simple stresses are imposed on thin, brittle materials, the material will fail in tension in a brittle manner and the fracture will exhibit little or no pliability. In such cases, there is no yield point for the material, since the yield strength and the ultimate strength are nearly the same. The failures that occur without plastic deformation are known as brittle failures. When two or more stresses occur in a ductile material, and particularly when stresses occur in three directions in a thick material, brittle fracture may occur.

f. Residual stresses also occur in castings, forgings, and hot rolled shapes. In forgings and castings, residual stresses occur as a result of the differential cooling that occurs. The outer portion of the part cools first, and the thicker and inner portion cools considerably faster. As the parts cool, they contract and pick up strength so that the portions that cool earlier go into a compressive load, and the portions that cool later go into a tensile stress mode. In complicated parts, the stresses may cause warpage.

g. Residual stresses are not always detrimental. They may have no effect or may have a beneficial effect on the service life of parts. Normally, the outer fibers of a part are subject to tensile loading and thus, with residual compression loading, there is a tendency to neutralize stress in the outer fibers of the part. An example of the use of residual stress is in the shrink fit of parts. A typical example is the cooling of sleeve bearings to insert them into machined holes, and allow them to expand to their normal dimension to retain then in the proper location. Sleeve bearings are used for heavy, Slow machinery, and are subject to compressive residual loading, keeping them within the hole. Large roller bearings are usually assembled to shafts by heating to expand them slightly so they will fit on the shaft, then allowing them to cool, to produce a tight assembly.

h. Residual stresses occur in all arc welds. The most common method of measuring stress is to produce weld specimens and then machine away specific amounts of metal, which are resisting the tensile stress in and adjacent to the weld. The movement that occurs is then measured. Another method is the use of grid marks or data points on the surface of weldments that can be measured in multiple directions. Cuts are made to reduce or release residual stresses from certain parts of the weld joint, and the measurements are taken again. The amount of the movement relates to the magnitude of the stresses. A third method utilizes extremely small strain gauges. The weldment is gradually and mechanically cut from adjoining portions to determine the change in internal stresses. With these methods, it is possible to establish patterns and actually determine amounts of stress within parts that were caused by the thermal effects of welds.

i. Figure 6-53 shows residual stresses in an edge weld. The metal close to the weld tends to expand in all directions when heated by the welding arc. This metal is restrained by adjacent cold metal and is slightly upset, or its thickness slightly increased, during this heating period. When the weld metal starts to cool, the upset area attempts to contract, but is again restrained by cooler metal. This results in the heated zone becoming stressed in tension. When the weld has cooled to room temperature, the weld metal and the adjacent base metal are under tensile stresses close to the yield strength. Therefore, there is a portion that is compressive, and beyond this, another tensile stress area. The two edges are in tensile residual stress with the center in compressive residual stress, as illustrated.

edge welded joint - residual stress pattern

j. The residual stresses in a butt weld joint made of relatively thin plate are more difficult to analyze. This is because the stresses occur in the longitudinal direction of the weld and perpendicular to the axis of the weld. The residual stresses within the weld are tensile in the longitudinal direction of the weld and the magnitude is at the yield strength of the metal. The base metal adjacent to the weld is also at yield stress, parallel to the weld and along most of the length of the weld. When moving away from the weld into the base metal, the residual stresses quickly fall to zero, and in order to maintain balance, change to compression. This is shown in figure 6-54. The residual stresses in the weld at right angles to the axis of the weld are tensile at the center of the plate and compressive at the ends. For thicker materials when the welds are made with multipasses, the relationship is different because of the many passes of the heat source. Except for single-pass, simple joint designs, the compressive and tensile residual stresses can only be estimated.

butt welded joint - residual stress pattern

k. As each weld is made, it will contract as it solidifies and gain strength as the metal cools. As it contracts, it tends to pull, and this creates tensile stresses at and adjacent to the weld. Further from the weld or bead, the metal must remain in equilibrium, and therefore compressive stresses occur. In heavier weldments when restraint is involved, movement is not possible, and residual stresses are of a higher magnitude. In a multipass single-groove weld, the first weld or root pass originally creates a tensile stress. The second, third, and fourth passes contract and cause a compressive load in the root pass. As passes are made until the weld is finished, the top passes will be in tensile load, the center of the plate in compression, and the root pass will have tensile residual stress.

l. Residual stresses can be decreased in several ways, as described below:

(1) If the weld is stressed by a load beyond its yield, strength plastic deformation will occur and the stresses will be more uniform, but are still located at the yield point of the metal. This will not eliminate residual stresses, but will create a more uniform stress pattern. Another way to reduce high or peak residual stresses is by means of loading or stretching the weld by heating adjacent areas, causing them to expand. The heat reduces the yield strength of the weld metal and the expansion will tend to reduce peak residual stresses within the weld. This method also makes the stress pattern at the weld area more uniform.

(2) High residual stresses can be reduced by stress relief heat treatment. With heat treatment, the weldment is uniformly heated to an elevated temperature, at which the yield strength of the metal is greatly reduced. The weldment is then allowed to cool slowly and uniformly so that the temperature differential between parts is minor. The cooling will be uniform and a uniform low stress pattern will develop within the weldment.

(3) High-temperature preheating can also reduce residual stress, since the entire weldment is at a relatively high temperature, and will cool more or less uniformly from that temperature and so reduce peak residual stresses.

m. Residual stresses also contribute to weld cracking. Weld cracking sometimes occurs during the manufacture of the weldment or shortly after the weldment is completed. Cracking occurs due to many reasons and may occur years after the weldment is completed. Cracks are the most serious defects that occur in welds or weld joints in weldments. Cracks are not permitted in most weldments, particularly those subject to low-temperature when the failure of the weldment will endanger life.

n. Weld cracking that occurs during or shortly after the fabrication of the weldment can be classified as hot cracking or cold cracking. In addition, weld may crack in the weld metal or in the base metal adjacent to welds metal, usually in the heat-affected zone. Welds crack for many reasons, including the following:

(1) Insufficient weld metal cross section to sustain the loads involved.

(2) Insufficient ductility of weld metal to yield under stresses involved.

(3) Under-bead cracking due to hydrogen pickup in a hardenable type of base material.

o. Restraint and residual stresses are the main causes of weld cracking during the fabrication of a weldment. Weld restraint can come from several factors, including the stiffness or rigidity of the weldment itself. Weld metal shrinks as it cools, and if the parts being welded cannot move with respect to one another and the weld metal has insufficient ductility, a crack will result. Movement of welds may impose high loads on other welds and cause them to crack during fabrication. A more ductile filler material should be used, or the weld should be made with sufficient cross-sectional area so that as it cools, it will have enough strength to withstand cracking tendencies. Typical weld cracks occur in the root pass when the parts are unable to move.

p. Rapid cooling of the weld deposit is also responsible for weld cracking. If the base metal being joined is cold and the weld is small, it will cool quickly. Shrinkage will also occur quickly, and cracking can occur. If the parts being joined are preheated even slightly, the cooling rate will be lower and cracking can be eliminated.

q. Alloy or carbon content of base material can also affect cracking. When a weld is made with higher-carbon or higher-alloy base material, a certain amount of the base material is melted and mixed with the electrode to produce the weld metal. The resulting weld metal has higher carbon and alloy content. It may have higher strength, but it has less ductility. As it shrinks, it may not have enough ductility to cause plastic deformation, and cracking may occur.

r. Hydrogen pickup in the weld metal and in the heat-affected zone can also cause cracking. When using cellulose-covered electrodes or when hydrogen is present because of damp gas, damp flux, or hydrocarbon surface materials, the hydrogen in the arc atmosphere will be absorbed in the molten weld metal and in adjoining high-temperature base metal. As the metal cools, it will reject the hydrogen, and if there is enough restraint, cracking can occur. This type of cracking can be reduced by increasing preheat, reducing restraint, and eliminating hydrogen from the arc atmosphere.

s. When cracking is in the heat-affected zone or if cracking is delayed, the cause is usually hydrogen pickup in the weld metal and the heat-affected zone of the base metal. The presence of higher-carbon materials or high alloy in the base metal can also be a cause. When welding high-alloy or high-carbon steels, the buttering technique can be used to prevent cracking. This involves surf acing the weld face of the joint with a weld metal that is much lower in carbon or alloy content than the base metal. The weld is then made between the deposited surfacing material and avoids the carbon and alloy pickup in the weld metal, so a more ductile weld deposit is made. Total joint strength must still be great enough to meet design requirements. Underbead cracking can be reduced by the use of low-hydrogen processes and filler metals. The use of preheat reduces the rate of cooling, which tends to decrease the possibility of cracking.

t. Stress Relieving Methods.

(1) Stress relieving in steel welds may be accomplished by preheating between 800 and 1450°F (427 and 788°C), depending on the material, and then slowly cooling. Cooling under some conditions may take 10 to 12 hours. Small pieces, such as butt welded high speed tool tips, may be annealed by putting them in a box of fire resistant material and cooling for 24 hours. In stress relieving mild steel, heating the completed weld for 1 hour per 1.00 in. (2.54 cm) of thickness is common practice. On this basis, steel 1/4 in. (0.64 cm) thick should be preheated for 15 minutes at the stress relieving temperature.

(2) Peening is another method of relieving stress on a finished weld, usually with compressed air and a roughing or peening tool. However, excessive peening may cause brittleness or hardening of the finished weld and may actually cause cracking.

(3) Preheating facilitates welding in many cases. It prevents cracking in the heat affected zone, particularly on the first passes of the weld metal. If proper preheating times and temperatures are used, the cooling rate is slowed sufficiently to prevent the formation of hard martensite, which causes cracking. Table 6-1 lists preheating temperatures of specific metals.

preheating temperatures

(4) The need for preheating steels and other metals is increased under the following conditions:

(a) When the temperature of the part or surrounding atmosphere is at or below freezing.

(b) When the diameter of the welding rod is small in comparison to thickness of the metal being joined.

(c) When welding speed is high

(d) When the shape and design of the parts being welded are complicated.

(e) When there is a great difference in mass of the parts being welded.

(f) When welding steels with a high carbon, low manganese, or other alloy content.

(g) When steel being welded tends to harden when cooled in air from the welding temperature.

u. The following general procedures can be used to relieve stress and to reduce cracking:

(1) Use ductile weld metal.

(2) Avoid extremely high restraint or residual stresses.

(3) Revise welding procedures to reduce restraint.

(4) Utilize low-alloy and low-carbon materials.

(5) Reduce the cooling rate by use of preheat.

(6) Utilize low-hydrogen welding processes and filler metals.

(7) When welds are too small for the service intended, they will probably crack. The welder should ensure that the size of the welds are not smaller than the minimum weld size designated for different thicknesses of steel sections.


Weldments must be designed and built to perform adequately in service. The risk of failure of a weldment is relatively small, but failure can occur in structures such as bridges, pressure vessels, storage tanks, ships, and penstocks. Welding has sometimes been blamed for the failure of large engineering structures, but it should be noted that failures have occurred in riveted and bolted structures and in castings, forgings, hot rolled plate and shapes, as well as other types of construction. Failures of these types of structures occurred before welding was widely used and still occur in unwelded structures today. However, it is still important to make weldments and welded structures as safe against premature failure of any type as possible. There are four specific types of failures, including brittle fracture, fatigue fracture, lamellar tearing, and stress corrosion cracking.

a. Brittle Fracture. Fracture can be classified into two general categories, ductile and brittle.

(1) Ductile fracture occurs by deformation of the crystals and slip relative to each other. There is a definite stretching or yielding and a reduction of cross-sectional area at the fracture (fig. 6-55).

ductile fracture surface

(2) Brittle fracture occurs by cleavage across individual crystals. The fracture exposes the granular structure, and there is little or no stretching or yielding. There is no reduction of area at the fracture (fig. 6-56).

brittle fracture surface

(3) It is possible that a broken surface will display both ductile and brittle fracture over different areas of the surface. This means that the fracture which propagated across the section changed its mode of fracture.

(4) There are four factors that should be reviewed when analyzing a fractured surface. They are growth marking, fracture mode, fracture surface texture and appearance, and amount of yielding or plastic deformation at the fracture surface.

(5) Growth markings are one way to identify the type of failure. Fatigue failures are characterized by a fine texture surface with distinct markings produced by erratic growth of the crack as it progresses. The chevron or herringbone pattern occurs with brittle or impact failures. The apex of the chevron appearing on the fractured surface always points toward the origin of the fracture and is an indicator of the direction of crack propagation.

(6) Fracture mode is the second factor. Ductile fractures have a shear mode of crystalline failure. The surface texture is silky or fibrous in appearance. Ductile fractures often appear to have failed in shear as evidenced by all parts of the fracture surface assuming an angle of approximately 45 degrees with respect to the axis of the load stress.

(7) The third factor is fracture surface and texture. Brittle or cleavage fractures have either a granular or a crystalline appearance. Brittle fractures usually have a point of origin. The chevron pattern will help locate this point.

(8) An indication of the amount of plastic deformation is the necking down of the surface. There is little or no deformation for a brittle fracture, and usually a considerable necked down area in the case of a ductile fracture.

(9) One characteristic of brittle fracture is that the steel breaks quickly and without warning. The fractures increase at very high speeds, and the steels fracture at stresses below the normal yield strength for steel. Mild steels, which show a normal degree of ductility when tested in tension as a normal test bar, may fail in a brittle manner. In fact, mild steel may exhibit good toughness characteristics at roan temperature. Brittle fracture is therefore more similar to the fracture of glass than fracture of normal ductile materials. A combination of conditions must be present at the same time for brittle fracture to occur. Some of these factors can be eliminated and thus reduce the possibility of brittle fracture. The following conditions must be present for brittle fracture to occur: low temperature, a notch or defect, a relatively high rate of loading, and triaxial stresses normally due to thickness of residual stresses. The microstructure of the metal also has an effect.

(10) Temperature is an important factor which must be considered in conjunction with microstructure of the material and the presence of a notch. Impact testing of steels using a standard notched bar specimen at different temperatures shows a transition from a ductile type failure to a brittle type failure based on a lowered temperature, which is known as the transition temperature.

(11) The notch that can result from faulty workmanship or from improper design produces an extremely high stress concentration which prohibits yielding. A crack will not carry stress across it, and the load is transmitted to the end of the crack. It is concentrated at this point and little or no yielding will occur. Metal adjacent to the end of the crack which does not carry load will not undergo a reduction of area since it is not stressed. It is, in effect, a restraint which helps set up triaxial stresses at the base of the notch or the end of the crack. Stress levels much higher than normal occur at this point and contribute to starting the fracture.

(12) The rate of loading is the time versus strain rate. The high rate of strain, which is a result of impact or shock loading, does not allow sufficient time for the normal slip process to occur. The material under load behaves elastically, allowing a stress level beyond the normal yield point. When the rate of loading, from impact or shock stresses, occurs near a notch in heavy thick material, the material at the base of the notch is subjected very suddenly to very high stresses. The effect of this is often complete and rapid failure of a structure and is what makes brittle fracture so dangerous.

(13) Triaxial stresses are more likely to occur in thicker material than in thin material. The z direction acts as a restraint at the base of the notch, and for thicker material, the degree of restraint in the through direction is higher. This is why brittle fracture is more likely to occur in thick plates or complex sections than in thinner materials. Thicker plates also usually have less mechanical working in their manufacture than thinner plates and are more susceptible to lower ductility in the z axis. The microstructure and chemistry of the material in the center of thicker plates have poorer properties than the thinner material, which receives more mechanical working.

(14) The microstructure of the material is of major importance to the fracture behavior and transition temperature range. Microstructure of a steel depends on the chemical composition and production processes used in manufacturing it. A steel in the as-rolled condition will have a higher transition temperature or liner toughness than the same steel in a normalized condition. Normalizing, or heating to the proper temperature and cooling slowly, produces a grain refinement which provides for higher toughness. Unfortunately, fabrication operations on steel, such as hot and cold forming, punching, and flame cutting, affect the original microstructure. This raises the transition temperature of the steel.

(15) Welding tends to accentuate Some of the undesirable characteristics that contribute to brittle fracture. The thermal treatment resulting from welding tends to reduce the toughness of the steel or to raise its transition temperature in the heat– affected zone. The monolithic structure of a weldment means that more energy is locked up and there is the possibility of residual stresses which may be at yield point levels. The monolithic structure also causes stresses and strains to be transmitted throughout the entire weldment, and defects in weld joints can be the nucleus for the notch or crack that will initiate fracture.

(16) Brittle fractures can be reduced in weldments by selecting steels that have sufficient toughness at the service temperatures. The transition temperature should be below the service temperature to which the weldment will be subjected. Heat treatment, normalizing, or any method of reducing locked-up stresses will reduce the triaxial yield strength stresses within the weldment. Design notches must be eliminated and notches resulting from poor workmanship must not occur. Internal cracks within the welds and unfused root areas must be eliminated.

b. Fatigue Failure. Structures sometimes fail at nominal stresses considerably below the tensile strength of the materials involved. The materials involved were ductile in the normal tensile tests, but the failures generally exhibited little or no ductility. Most of these failures development after the structure had been subjected to a large number of cycles of loading. This type of failure is called a fatigue failure.

(1) Fatigue failure is the formation and development of a crack by repeated or fluctuating loading. When sudden failure occurs, it is because the crack has increased enough to reduce the load-carrying capacity of the part. Fatigue cracks may exist in some weldments, but they will not fail until the load-carrying area is sufficiently reduced. Repeated loading causes progressive enlargement of the fatigue cracks through the material. The rate at which the fatigue crack increases depends upon the type and intensity of stress as well as other factors involving the design, the rate of loading, and type of material.

(2) The fracture surface of a fatigue failure is generally smooth and frequently shins concentric rings or areas spreading from the point where the crack initiated. These rings show the propagation of the crack, which might be related to periods of high stress followed by periods of inactivity. The fracture surface also tends to become rougher as the rate of propagation of the crack increases. Figure 6-57 shows the characteristic fatigue failure surface.

fatigue fracture surface

(3) Many structures are designed to a permissible static stress based on the yield point of the material in use and the safety factor that has been selected. This is based on statically loaded structures, the stress of which remains relatively constant. Many structures, however, are subject to other than static loads in service. These changes may range from simple cyclic fluctuations to completely random variations. In this type of loading, the structure must be designed for dynamic loading and considered with respect to fatigue stresses.

(4) The varying loads involved with fatigue stresses can be categorized in different manners. These can be alternating cycles from tension to compression, or pulsating loads with pulses from zero load to a maximum tensile load, or from a zero load to a compressive load, or loads can be high and rise higher, either tensile or compressive. In addition to the loadings, it is important to consider the number of times the weldment is subjected to the cyclic loading. For practical purposes, loading is considers in millions of cycles. Fatigue is a cumulative process and its effect is in no way healed during periods of inactivity. Testing machines are available for loading metal specimens to millions of cycles. The results are plotted on stress vs. cycle curves, which show the relationship between the stress range and the number of cycles for the particular stress used. Fatigue test specimens are machined and polished, and the results obtained on such a specimen may not correlate with actual service life of a weldment. It is therefore important to determine those factors which adversely affect the fatigue life of a weldment.

(5) The possibility of a fatigue failure depends on four factors: the material used, the number of loading cycles, the stress level and nature of stress variations, and total design and design details. The last factor is controllable in the design and manufacture of the weldment. Weld joints can be designed for uniform stress distribution utilizing a full-penetration weld, but in other cases, joints may not have full penetration because of an unfused root. This prohibits uniform stress distribution. Even with a full-penetration weld, if the reinforcement is excessive, a portion of the stress will flow through the reinforced area and will not be uniformly distributed. Welds designed for full penetration might not have complete penetration because of workmanship factors such as cracks, slag inclusions, and incomplete penetration, and therefore contain a stress concentration. One reason fatigue failures in welded structures occur is because the welded design can introduce more severe stress concentrations than other types of design. The weld defects mentioned previously, including excessive reinforcement, undercut, or negative reinforcement, will contribute to the stress concentration factor. A weld also forms an integral part of the structure, and when parts are attached by welding, they may produce sudden changes of section which contribute to stress concentrations under normal types of loading. Anything that can be done to smooth out the stress flow in the weldment will reduce stress concentrations and make the weldment less subject to fatigue failure. Total design with this in mind and careful workmanship will help to eliminate this type of problem.

c. Lamellar Tearing. Lamellar tearing is a cracking which occurs beneath welds, and is found in rolled steel plate weldments. The tearing always lies within the base metal, usually outside the heat-affected zone and generally parallel to the weld fusion boundary. This type of cracking has been found in corner joints where the shrinkage across the weld tended to open up in a manner similar to lamination of plate steel. In these cases, the lamination type crack is removed and replaced with weld metal. Before the advent of ultrasonic testing, this type of failure was probably occurring and was not found. It is only when welds subjected the base metal to tensile loads in the z, or through, direction of the rolled steel that the problem is encountered. For many years, the lower strength of rolled steel in the through direction was recognized and the structural code prohibited z-directional tensile loads on steel spacer plates. Figure 6-58 shows how lamellar tearing will come to the surface of the metal. Figure 6-59, shining a tee joint, is a more common type of lamellar tearing, which is much more difficult to find. In this case, the crack does not cane to the surface and is under the weld. This type of crack can only be found with ultrasonic testing or if failure occurs, the section can actually come out and separate from the main piece of metal.

corner joint
tee joint

(1) Three conditions must occur to cause lamellar tearing. These are strains in the through direction of the plate caused by weld metal shrinkage in the joint and increased by residual stresses and by loading; stress through the joint across the plate thickness or in the z direction due to weld orientation in which the fusion line beneath the weld is roughly parallel to the lamellar separation; and poor ductility of the material in the z, or through, direction.

(2) Lamellar tearing can occur during flame-cutting operations and also in cold-shearing operations. It is primarily the low strength of the material in the z, or through, direction that contributes to the problem. A stress placed in the z direction triggers the tearing. The thermal heating and stresses resulting from weld shrinkage create the fracture. Lamellar tearing is not associated with the under-bead hydrogen cracking problem. It can occur soon after the weld has been made, but on occasion will occur at a period months later. Also, the tears are under the heat-affected zone, and are more apt to occur in thicker materials and in higher-strength materials.

(3) Only a very small percentage of steel plates are susceptible to lamellar tearing. There are only certain plates where the concentration of inclusions are coupled with the unfavorable shape and type that present the risk of tearing. These conditions rarely occur with the other two factors mentioned previously. In general, three situations must occur in combination: structural restraint, joint design, and the condition of the steel.

(4) Joint details can be changed to avoid the possibility of lamellar tearing. In tee joints, double-fillet weld joints are less susceptible than full penetration welds. Balanced welds on both sides of the joint present less risk of lamellar tearing than large single-sided welds. corner joints are common in box columns. Lamellar tearing at the corner joints is readily detected on the exposed edge of the plate. Lamellar tearing can be overcome in corner joints by placing the bevel for the joint on the edge of the plate that would exhibit the tearing rather than on the other plate. This is shown by figure 6-60. Butt joints rarely are a problem with respect to lamellar tearing since the shrinkage of the weld does not set up a tensile stress in the thickness direction of the plates.

redesigned corner joint to avoid lamellar tearing

(5) Arc welding processes having higher heat input are less likely to create lamellar tearing. This may be because of the fewer number of applications of heat and the lesser number of shrinkage cycles involved in making a weld. Deposited filler metal with lower yield strength and high ductility also reduces the possibility of lamellar tearing. Preheat and stress relief heat treatment are not specifically advantageous with respect to lamellar tearing. The buttering technique of laying one or more layers of low strength, high-ductility weld metal deposit on the surface of the plate stressed in the z direction will reduce the possibility of lamellar tearing. This is an extreme solution and should only be used as a last resort. By observing the design factors mentioned above, the lamellar tearing problem is reduced.

d. Stress Corrosion Cracking. Stress corrosion cracking and delayed cracking due to hydrogen embrittlement can both occur when the weldment is subjected to the type of environment that accentuates this problem.

(1) Delayed cracking is caused by hydrogen absorbed in the base metal or weld metal at high temperatures. Liquid or molten steel will absorb large quantities of hydrogen. As the metal solidifies, it cannot retain all of the hydrogen and is forced out of solution. The hydrogen coming out of the solution sets up high stresses, and if enough hydrogen is present, it will cause cracking in the weld or the heat-affected zone. These cracks develop over a period of time after the weld is completed. The concentration of hydrogen and the stresses resulting from it when coupled with residual stresses promote cracking. Cracking will be accelerated if the weldment is subjected to thermal stresses due to repeated heating and cooling.

(2) Stress corrosion cracking in steels is sometimes called caustic embrittlement. This type of cracking takes place when hot concentrated caustic solutions are in contact with steel that is stressed in tension to a relatively high level. The high level of tension stresses can be created by loading or by high residual stresses. Stress corrosion cracking will occur if the concentration of the caustic solution in contact with the steel is sufficiently high and if the stress level in the weldment is sufficiently high. This situation can be reduced by reducing the stress level and the concentration of the caustic solution. Various inhibitors can be added to the solution to reduce the concentration. Close inspection must be maintained on highly stressed areas.

(3) Graphitization is another type of cracking, caused by long service life exposed to thermal cycling or repeated heating and cooling. This may cause a breakdown of carbides in the steel into small areas of graphite and iron. This formation of graphite in the edge of the heat-affected area exposed to the thermal cycling causes cracking. It will often occur in carbon steels deoxidized with aluminum. The addition of molybdenum to the steel tends to restrict graphitization, and for this reason, carbon molybdenum steels are normally used in high-temperature power plant service. These steels must be welded with filler metals of the same composition.

6-31. ARC BLOW

a. General. Arc blow is the deflection of an electric arc from its normal path due to magnetic forces. It is mainly encountered with dc welding of magnetic materials, such as steel, iron, and nickel, but can also be encountered when welding nonmagnetic materials. It will usually adversely affect appearance of the weld, cause excessive spatter, and can also impair the quality of the weld. It is often encountered when using the shielded metal arc welding process with covered electrodes. It is also a factor in semiautomatic and fully automatic arc welding processes. Direct current, flowing through the electrode and the base metal, sets up magnetic fields around the electrode, which deflect the arc from its intended path. The welding arc is usually deflected forward or backward of the direction of travel; however, it may be deflected from one side to the other. Back blow is encountered when welding toward the ground near the end of a joint or into a corner. Forward blow is encountered when welding away from the ground at the start of a joint. Arc blow can become so severe that it is impossible to make a satisfactory weld. Figure 6-61 shows the effect of ground location on magnetic arc blow.

effect of ground location on magnetic arc blow

b. When an electric current passes through an electrical conductor, it produces a magnetic flux in circles around the conductor in planes perpendicular to the conductor and with their centers in the conductor. The right-hand rule is used to determine the direction of the magnetic flux. It states that when the thumb of the right hand points in the direction in which the current flows (conventional flow) in the conductor, the fingers point in the direction of the flux. The direction of the magnetic flux produces polarity in the magnetic field, the same as the north and south poles of a permanent magnet. This magnetic field is the same as that produced by an electromagnet. The rules of magnetism, which state that like poles repel and opposite poles attract, apply in this situation. Welding current is much higher than the electrical current normally encountered. Likewise, the magnetic fields are also much stronger.

c. The welding arc is an electrical conductor and the magnetic flux is set up surrounding it in accordance with the right-hand rule. The magnetic field in the vicinity of the welding arc is the field produced by the welding current which passes through it from the electrode and to the base metal or work. This is a self-induced circular magnetic field which surrounds the arc and exerts a force on it from all sides according to the electrical-magnetic rule. As long as the magnetic field is symmetrical, there is no unbalanced magnetic force and no arc deflection. Under these conditions, the arc is parallel or in line with the centerline of the electrode and takes the shortest path to the base plate. If the symmetry of this magnetic field is disturbed, the forces on the arc are no longer equal and the arc is deflected by the strongest force.

d. The electrical-magnetic relationship is used in welding applications for magnetically moving, or oscillating, the welding arc. The gas tungsten arc is deflected by means of magnetic flux. It can be oscillated by transverse magnetic fields or be made to deflect in the direction of travel. Moving the flux field surrounding the arc and introducing an external-like polarity field roves the arc magnetically. Oscillation is obtained by reversing the external transverse field to cause it to attract the field surrounding the arc. As the self-induced field around the arc is attracted and repelled, it tends to move the arc column, which tries to maintain symmetry within its own self-induced magnetic field. Magnetic oscillation of the gas tungsten welding arc is used to widen the deposition. Arcs can also be made to rotate around the periphery of abutting pipes by means of rotating magnetic fields. Longer arcs are moved more easily than short arcs. The amount of magnetic flux to create the movement must be of the same order as the flux field surrounding the arc column. Whenever the symmetry of the field is disturbed by some other magnetic force, it will tend to move the self-induced field surrounding the arc and thus deflect the arc itself.

e. Except under the most simple conditions, the self-induced magnetic field is not symmetrical throughout the entire electric circuit and changes direction at the arc. There is always an unbalance of the magnetic field around the arc because the arc is roving and the current flow pattern through the base material is not constant. The magnetic flux will pass through a magnetic material such as steel much easier than it will pass through air, and the magnetic flux path will tend to stay within the steel and be more concentrated and stronger than in air. Welding cur-rent passes through the electrode lead, the electrode holder to the welding electrode, then through the arc into the base metal. At this point the current changes direction to pass to the work lead connection, then through the work lead back to the welding machine. This is shown by figure 6-62. At the point the arc is in contact with the work, the change of direction is relatively abrupt, and the fact that the lines of force are perpendicular to the path of the welding current creates a magnetic unbalance. The lines of force are concentrated together on the inside of the angle of the current path through the electrode and the work, and are spread out on the outside angle of this path. Consequently, the magnetic field is much stronger on the side of the arc toward the work lead connection than on the other side, which produces a force on the stronger side and deflects the arc to the left. This is toward the weaker force and is opposite the direction of the current path. The direction of this force is the same regardless of the direction of the current. If the welding current is reversed, the magnetic field is also reversed, but the direction of the magnetic force acting on the arc is always in the same direction, away from the path of the current through the work.

unbalanced magnetic force due to current direction change

f. The second factor that keeps the magnetic field from being symmetrical is the fact that the arc is moving and depositing weld metal. As a weld is made joining two plates, the arc moves from one end of the joint to the other and the magnetic field in the plates will constantly change. Since the work lead is immediately under the arc and moving with the arc, the magnetic path in the work will not be concentric about the point of the arc, because the lines of force take the easiest path rather than the shortest path. Near the start end of the joint the lines of force are crowded together and will tend to stay within the steel. Toward the finish end of the joint, the lines of force will be separated since there is more area. This is shown by figure 6-63. In addition, where the weld has been made the lines of force go through steel. Where the weld is not made, the lines of force must cross the air gap or root opening. The magnetic field is more intense on the short end and the unbalance produces a force which deflects the arc to the right or toward the long end.

unbalanced magnetic force due to unbalanced magnetic path

g. When welding with direct current, the total force tending to cause the arc to deflect is a combination of these two forces. These forces may add or subtract from each other, and at times may meet at right angles. The polarity or direction of flow of the current does not affect the direction of these forces nor the resultant force. By analyzing the path of the welding current through the electrode and into the base metal to the work lead, and analyzing the magnetic field within the base metal, it is possible to determine the resultant forces and predict the resulting arc deflection or arc blow.

h. Forward blow exists for a short time at the start of a weld, then diminishes. This is because the flux soon finds an easy path through the weld metal. Once the magnetic flux behind the arc is concentrated in the plate and the weld, the arc is influenced mainly by the flux in front of it as this flux crosses the root opening. At this point, back blow may be encountered. Back blow can occur right up to the end of the joint. As the weld approaches the end, the flux ahead of the arc becomes more crowded, increasing the back blow. Back blow can become extremely severe right at the very end of the joint.

i. The use of alternating current for welding greatly reduces the magnitude of deflection or arc blow; however, ac welding does not completely eliminate arc blow. Reduction of arc blow is reduced because the alternating current sets up other currents that tend to either neutralize the magnetic field or greatly reduce its strength. Alternating current varies between maximum value of one polarity and the maximum value of the opposite polarity. The magnetic field surrounding the alternating current conductor does the same thing. The alternating magnetic field is a roving field which induces current in any conductor through which it passes, according to the induction principle. Currents are induced in nearby conductors in a direction opposite that of the inducing current. These induced currents are called eddy currents. They produce a magnetic field of their own which tends to neutralize the magnetic field of the arc current. These currents are alternating currents of the same frequency as the arc current and are in the part of the work nearest the arc. They always flow from the opposite direction as shown by figure 6-64. When alternating current is used for welding, eddy currents are induced in the workpiece, which produce magnetic fields and reduce the intensity of the field acting on the arc. Alternating current cannot be used for all welding applications and for this reason changing from direct current to alternating current may not always be possible to eliminate or reduce arc blow.

reduction of magnetic force due to induced fields

j. Summary of Factors Causing Arc Blow.

(1) Arc blow is caused by magnetic forces. The induced magnetic forces are not symmetrical about the magnetic field surrounding the path of the welding current. The location of magnetic material with respect to the arc creates a magnetic force on the arc which acts toward the easiest magnetic path and is independent of electrode polarity. The location of the easiest magnetic path changes constantly as welding progresses; therefore, the intensity and the direction of the force changes.

(2) Welding current will take the easiest path but not always the most direct path through the work to the work lead connection. The resultant magnetic force is opposite in direction to the current from the arc to is independent of welding current polarity.

(3) Arc blow is not as severe with alternating current because the induction principle creates current flow within the base metal which creates magnetic fields that tend to neutralize the magnetic field affecting the arc.

(4) The greatest magnetic force on the arc is caused by the difference resistance of the magnetic path in then the base metal around the arc. The location of the work connection is of secondary importance, but may have an effect on reducing the total magnetic force on the arc. It is best to have the work lead connection at the starting point of the weld. This is particularly true in electroslag welding where the work lead should be connected to the starting sump. On occasion, the work lead can be changed to the opposite end of the joint. In sane cases, leads can be connected to both ends.

k. Minimizing Arc Blow.

(1) The magnetic forces acting on the arc can be modified by changing the magnetic path across the joint. This can be accomplished by runoff tabs, starting plates, large tack welds, and backing strips, as well as the welding sequence.

(2) An external magnetic field produced by an electromagnet may be effective. This can be accomplished by wrapping several turns of welding lead around the workpiece.

(3) Arc blow is usually more pronounced at the start of the weld seam. In this case, a magnetic shunt or runoff tab will reduce the blow.

(4) Use as short an arc as possible so that there is less of an arc for the magnetic forces to control.

(5). The welding fixture can be a source of arc blow; therefore, an analysis with respect to fixturing is important. The hold-down clamps and backing bars must fit closely and tightly to the work. In general, copper or nonferrous metals should be used. Magnetic structure of the fixture can affect the magnetic forces controlling the arc.

(6) Place ground connections as far as possible from the joints to welded.

(7) If to back blow is the problem, place the ground connection at the start of welding, and weld toward a heavy tack weld.

(8) If forward blow causes trouble, place the ground connection at the end of the joint to be welded.

(9) Position the electrode so that the arc force counteracts the arc blow.

(10) Reduce the welding current.

(11) Use the backstep sequence of welding.

(12) Change to ac, which may require a change in electrode classification

(13) Wrap the ground cable around the workpiece in a direction such that the magnetic field it sets up will counteract the magnetic field causing the arc blow.

(14) Another major problem can result from magnetic fields already in the base metal, particularly when the base metal has been handled by magnet lifting cranes. Residual magnetism in heavy thick plates handled by magnets can be of such magnitude that it is almost impossible to make a weld. Attempt to demagnetize the parts, wrap the part with welding leads to help overcome their effect, or stress relieve or anneal the parts.


a. General. Only rarely are there failures of welded structures, but failures of large engineered structures do occur occasionally. Catastrophic failures of major structures are usually reported whenever they occur. The results of investigations of these failures are usually reported and these reports often provide information that is helpful in avoiding future similar problems. In the same manner, there are occasional failures of noncritical welds and weldments that should also be investigated. Once the reason is determined it can then be avoided. An objective study must be made of any failure of parts or structures to determine the cause of the failure. This is done by investigating the service life, the conditions that led up to the failure, and the actual mode of the failure. An objective study of failure should utilize every bit of information available, investigate all factors that could remotely be considered, and evaluate all this information to find the reason for the failure. Failure investigation often uncovers facts that lead to changes in design, manufacturing, or operating practice, that will eliminate similar failures in the future. Failures of insignificant parts can also lead to advances in knowledge and should be done objectively, as with a large structure. Each failure and subsequent investigation will lead to changes that will assure a more reliable product in the future.

b. The following four areas of interest should be investigated to determine the cause of weld failure and the interplay of factors involved:

(1) Initial observation. The detailed study by visual inspection of the actual component that failed should be made at the failure site as quickly as possible. Photographs should be taken, preferably in color, of all parts, structures, failure surfaces, fracture texture appearance, final location of component debris, and all other factors. Witnesses to the failure should all be interviewed and all information determined from them should be recorded.

(2) Background data. Investigators should gather all information concerning specifications, drawings, component design, fabrication methods, welding procedures, weld schedules, repairs in and during manufacturing and in service, maintenance, and service use. Efforts should be made to obtain facts pertinent to all possible failure modes. Particular attention should be given to environmental details, including operating temperatures, normal service loads, overloads, cyclic loading, and abuse.

(3) Laboratory studies. Investigators should make tests to verify that the material in the failed parts actually possesses the specified composition, mechanical properties, and dimensions. Studies should also be made microscopically in those situations in which it would lead to additional information. Each failed part should be thoroughly investigated to determine what bits of information can be added to the total picture. Fracture surfaces can be extremely important. Original drawings should be obtained and marked showing failure locations, along with design stress data originally used in designing the product. Any other defects in the structure that are apparent, even though they might not have contributed to the failure, should also be noted and investigated.

(4) Failure assumptions. The investigator should list not only all positive facts and evidence that may have contributed to the failure, but also all negative responses that may be learned about the failure. It is sometimes important to know what specific things did not happen or what evidence did not appear to help determine what happened. The data should be tabulated and the actual failure should be synthesized to include all available evidence.

c. Failure cause can usually be classified in one of the following three classifications:

(1) Failure due to faulty design or misapplication of material.

(2) Failure due to improper processing or improper workmanship.

(3) Failure due to deterioration during service.

d. The following is a summary of the above three situations:

(1) Failure due to faulty design or misapplication of the material involves failure due to inadequate stress analysis, or a mistake in design such as incorrect calculations on the basis of static loading instead of dynamic or fatigue loading. Ductile failure can be caused by a load too great for the section area or the strength of the material. Brittle fracture may occur from stress risers inherent in the design, or the wrong material may have been specified for producing the part.

(2) Failures can be caused by faulty processing or poor workmanship that may be related to the design of the weld joint, or the weld joint design can be proper but the quality of the weld is substandard. The poor quality weld might include such defects as undercut, lack of fusion, or cracks. Failures can be attributed to poor fabrication practice such as the elimination of a root opening, which will contribute to incomplete penetration. There is also the possibility that the incorrect filler metal was used for welding the part that failed.

(3) Failure due to deterioration during service can cause overload, which may be difficult to determine. Normal wear and abuse to the equipment may have result-ed in reducing sections to the degree that they no longer can support the load. Corrosion due to environmental conditions and accentuated by stress concentrations will contribute to failure. In addition, there may be other types of situations such as poor maintenance, poor repair techniques involved with maintenance, and accidental conditions beyond the user’s control. The product might be exposed to an environment for which it was not designed.

e. Conclusion. Examination of catastrophic and major failures has led the welding industry to appreciate the following facts:

(1) Weldments are monolithic in character.

(2) Anything welded onto a structure will carry part of the load whether intended or not.

(3) Abrupt changes in section, either because of adding a deckhouse or removing a portion of the deck for a hatch opening, create stress concentration. Under normal loading, if the steel at the point of stress concentration is notch sensitive at the service temperature, failure can result.


a. There are two other welding problems that require some explanation and solutions. These are welding over painted surfaces and painting of welds.


Cutting painted surfaces with arc or flame processes should be done with caution. Demolition of old structural steel work that had been painted many times with flame-cutting or arc-cutting techniques can create health problems. Cutting through many layers of lead paint will cause an abnormally high lead concentration in the immediate area and will require special precautions such as extra ventilation or personnel protection.

b. Welding over paint is discouraged. In every code or specification, it is specifically stated that welding should be done on clean metal. In some industries, however welds are made over paint, and in other flame cutting is done on painted base metal.

(1) In the shipbuilding industry and several other industries, steel, when it is received from the steel mill, is shot blasted, given a coating of prime paint, and then stored outdoors. Painting is done to preserve the steel during storage, and to identify it. In sane shipyards a different color paint is used for different classes of steel. When this practice is used, every effort should be made to obtain a prime paint that is compatible with welding.

(2) There are at least three factors involved with the success of the weld when welding over painted surfaces: the compatibility of the paint with welding; the dryness of the paint; and the paint film thickness.

(3) Paint compatibility varies according to the composition of the paint. Certain paints contain large amounts of aluminum or titanium dioxide, which are usually compatible with welding. Other paints may contain zinc, lead, vinyls, and other hydrocarbons, and are not compatible with welding. The paint manufacturer or supplier should be consulted. Anything that contributes to deoxidizing the weld such as aluminum, silicon, or titanium will generally be compatible. Anything that is a harmful ingredient such as lead, zinc, and hydrocarbons will be detrimental. The fillet break test can be used to determine compatibility. The surfaces should be painted with the paint under consideration. The normal paint film thickness should be used, and the paint must be dry.

(4) The fillet break test should be run using the proposed welding procedure over the painted surface. It should be broken and the weld examined. If the weld breaks at the interface of the plate with the paint it is obvious that the paint is not compatible with the weld.

(5) The dryness of the paint should be considered. Many paints employ an oil base which is a hydrocarbon. These paints dry slowly, since it takes a considerable length of time for the hydrocarbons to evaporate. If welding is done before the paint is dry, hydrogen will be in the arc atmosphere and can contribute to underbead cracking. The paint will also cause porosity if there is sufficient oil present. Water based paints should also be dry prior to welding.

(6) The thickness of the paint film is another important factor. Some paints may be compatible if the thickness of the film is a maximum of 3 to 4 mm. If the paint film thicknesses are double that amount, such as occurs at an overlap area, there is the possibility of weld porosity. Paint films that are to be welded over should be of the minimum thickness possible.

(7) Tests should be run with the dry maximum film thickness to be used with the various types of paints to determine which paint has the least harmful effects on the weld deposit.

c. Painting over welds is also a problem. The success of any paint film depends on its adherence to the base metal and the weld, which is influenced by surface deposits left on the weld and adjacent to it. The metallurgical factors of the weld bead and the smoothness of the weld are of minor importance with regard to the success of the paint. Paint failure occurs when the weld and the immediate area are not properly cleaned prior to painting. Deterioration of the paint over the weld also seems to be dependent upon the amount of spatter present. Spatter on or adjacent to the weld leads to rusting of the base material under the paint. It seems that the paint does not completely adhere to spatter and some spatter does fall off in time, leaving bare metal spots in the paint coating.


Aluminum and aluminum alloys should not be cleaned with caustic soda or cleaners having a pH above 10, as they may react chemically. Other nonferrous metals should be investigated for reactivity prior to cleaning.

(1) The success of the paint job can be insured by observing both preweld and postweld treatment. Preweld treatment found most effective is to use antispatter compounds, as well as cleaning the weld area, before welding. The antispatter compound extends the paint life because of the reduction of spatter. The antispatter compound must be compatible with the paint to be used.

(2) Postweld treatment for insuring paint film success consists of mechanical and chemical cleaning. Mechanical cleaning methods can consist of hand chipping and wire brushing, power wire brushing, or sand or grit blasting. Sand or grit blasting is the most effective mechanical cleaning method. If the weldment is furnace stress relieved and then grit blasted, it is prepared for painting. When sand or grit blasting cannot be used, power wire brushing is the next most effective method. In addition to mechanical cleaning, chemical bath washing is also recommended. Slag coverings on weld deposits must be thoroughly removed from the surface of the weld and from the adjacent base metal. Different types of coatings create more or less problems in their removal and also with respect to paint adherence. Weld slag of many electrodes is alkaline in nature and for this reason must be neutralized to avoid chemical reactions with the paint, which will cause the paint to loosen and deteriorate. For this reason, the weld should be scrubbed with water, which will usually remove the residual coating slag and smoke film from the weld. If a small amount of phosphoric acid up to a 5% solution is used, it will be more effective in neutralizing and removing the slag. It must be followed by a water rinse. If water only is used, it is advisable to add small amounts of phosphate or chromate inhibitors to the water to avoid rusting, which might otherwise occur.

(3) It has been found that the method of applying paint is not an important factor in determining the life of the paint over welds. The type of paint employed must be suitable for coating metals and for the service intended.

(4) Successful paint jobs over welds can be obtained by observing the following: minimize weld spatter using a compatible anti-spatter compound; mechanically clean the weld and adjacent area; and wash the weld area with a neutralizing bath and rinse.

No tags for this post.