Figure "Welding flaws" (Ref. 16.2.1.3-8): Spot welding, step welding, and roller seam welding are all types of fusion welding and have typical weak points and flaws. Possible causes for these are grouped into three main categories in the table:
* Welding energy * Electrodes * Metal sheets Once the flaw type has been identified, it is possible to narrow the scope to specific corrective measures.

Figure "Sensitivity of special sheet weldings": Spot welding and roller seam welding have performed very well in the typical sheet metal constructions used in older engine types (page 16.2.1.3-2). A prerequisite for this behavior is that the welds are located in part zones with low dynamic stresses. Even when they are in accordance with specifications, welds are a pronounced weak point for dynamic loads. Especially under flexural modes, such as are typical during vibrations of thin metal sheets (Fig. "Welding flaws"), premature failure can be expected. The comparatively low dynamic strength is due to process-specific characteristics:
Spot weld nuggets act as sharp geometric notches (top detail). The overhanging free end must not be misinterpreted as a soft transition. The notch effect is increased by the simultaneous stiffness jump, usually to double thickness.
At the same time, in the area of this stress concentration, the cast structure of the weld seam acts as a structural notch relative to the forged structure of the base material (right detail, also see Volume 3, Ill. 13-18).
The situation is exacerbated by the process-specific shifting of the joined cross-sections. Under tensile stress, the overlapping gap is pulled open and subjected to extreme loads by the resulting bending moment. These loads increase disproportionally with greater stiffness, i.e. thicker metal sheets.

Figure "Dynamic fatigue of resistance welds": This diagram shows several typical damage cases in which unforeseen or unusual high-frequency flexural modes caused fractures at the transition of roller seam welds.

Figure "Electron beam welding (EB) process": During electron beam welding, the energy is not applied by an arc, as in TIG and MIG, but with a beam of electrons. Filler material is usually not necessary. In order to make possible the energy input via the electron beam striking the metal, welding must be done without a welding gap. The special welding process is suceptible to process-specific flaws, including the development of porosity, spikes, and hot crack fields in the HAZ (Fig. "EB welding flaws"). Flaws that can also be observed in other fusion welding processes, such as gas porosity, cavities, and individual hot cracks, are also possible (Ref. 16.2.1.3-11).
The process takes place in a high vacuum in order to allow full propagation of the electron beam. For this reason, collapse of the vacuum due to leakages or vaporization of alloy components or fouling on the parts is a potential source of flaws.
Welding is usually fully automated after optimal process parameters are inputted and adjusted. This also applies to the relative movement of the part under the beam. Exact beam control along the weld joint is of utmost importance, as it prevents failed bonds and kissing bonds, which are very difficult to detect. For this reason, the equipment assembly has an important task (Fig. "EB welding flaws by misalignment"). The geometry (focal point) and position of the beam relative to the part surface must be strictly controlled in order to ensure the desired weld seam shape and root fusion.
The electron beam with a thickness of about 0.5 mm vaporizes the surface on impact and bores a hole into the weld joint (top diagram, Refs. 16.2.1.3-1 and 16.2.1.3-11). This tubular cavity is held open by the pressure of the metal vapors and follows the welding motion. In the melted wall, a flow forms against the welding direction. If full root fusion does not occur, this flow can create inclusions of metal vapors. Pores (spikes) form before the metal vapors condense. This can result in both pores and bonding flaws in unjoined pore walls (bottom right detail, Fig. "EB welding flaws in a gear wheel"). Because the welding cavity is an essential part of the welding process, it is understandable why vaporizing impurities in the welding joint or from impure materials can lead to flaws.
The high welding rate and narrow electron beam are responsible for the typical seam cross-section, which is narrow and almost parallel below the seam head.

Figure "EB welding flaws": Process-specific characteristics are responsible for flaws in electron beam welds (Refs. 16.2.3.3-3, 16.2.3.3-5, 16.2.3.3-9, 16.2.3.3-11) that do not occur in TIG weld seams (Ills. 16.2.1.3-12, 16.2.1.3-13,and 16.2.1.3-14). Because the electron beam has a very small diameter (Fig. "Electron beam welding (EB) process") and can only transfer the required welding energy if it strikes the metal being welded, the beam must be directed very accurately on the welding joint (Fig. "EB welding flaws by misalignment"), and there can be no welding gap. If these conditions are not sufficiently met, there is a danger of bonding flaws (cold lap, also see Fig. "Savety relevant flaws of casted turbine blades"). These are almost impossible to detect using non-destructive methods if there is only a partial bond (kissing bond).

In EBW seams, especially, metallographic cross-sections can reveal that melt evidently enters into previously created grain boundary cracks in the HAZ (bottom left detail, Ref. 16.2.1.3-23). This phenomenon can be explained by cracking in the heating phase, and can thus be referred to as heating cracks. In EBW, the high energy concentration creates extreme temperature gradients, which cause correspondingly high heat stress in the base material. This could result in the formation of fields of micro-cracks oriented in the direction of the welding process.

The following mechanism of crack initiation is likely: as long as the welded zone has not yet melted, high compressive stresses build up during the heating process due to the restricted heat strain. These stresses are in equilibrium with tensile stresses at the seam edges (HAZ). If the latter overstress the grain boundary strength, they will form hot tears. An indicator of this cracking mechanism is cracks that run diagonal to the welding direction. This crack orientation is logical if the cracks first form at the surface, where the beam striking point travels ahead of the welding cavity (Fig. "Electron beam welding (EB) process"). The frequently observed diagonal crack field (Fig. "Influences at micro-cracking") should be seen in connection with “pulling” of the cracks (notch effect) due to the course of the temperature field during the welding process.
This damage model also seems to fit a type of behavior observed in TIG welds. The more intensive the heat input, i.e. the faster the welding speed, the more pronounced the micro-cracking becomes (Fig. "Influences at micro-cracking").
The softened grain boundaries in the HAZ that are torn open then partially fill with the available melt and are healed.

Other typical process-specific flaws are long pores in the root (spikes). They are created by metal vapors at the tip of the cavity that become trapped by the melt (Fig. "Electron beam welding (EB) process"). They can also cause bonding flaws. If the pore collapses after the metal vapors condense, but the pore walls do not bond (Fig. "EB welding flaws in a gear wheel"), it creates a type of cold lap (Fig. "Savety relevant flaws of casted turbine blades").

Figure "EB welding seam flaws" and Figure "Repair of EB welding flaws": The diagrams provide a summary of the causes for typical problems and flaws in electron beam welds, preventive measures, and possible reworking options (repairs).

Welding jig: Even if the jig is sufficiently exactly aligned, unfavorable heat strain behavior during welding (Fig. "EB welding flaws by misalignment") can cause it to warp. The joint surfaces can lift up or move out of the plane of the beam. This creates a risk of bonding flaws and kissing bonds (Fig. "EB welding flaws", Ref. 16.2.1.3-27). These flaws also result from undesired beam deflection. This, as well as changes in focus, can result from magnetization of the jig. If the jig is made from a different material than the work piece, it must not be melted by the beam. If work piece melt mixes with jig melt, it will result in unallowable strength behavior.
Insufficient coverage of the root can result in intensive melt spray. If the sprayed melt strikes unprotected part zones, they will have a damaging effect (Ills. 16.2.2.6-5 and 16.2.2.6-6). They cause dynamic strength losses due to local tension residual stresses, cracking, and structural changes (Fig. "Damages by splashes of liquid metal").

Seam preparation: Excessive roughness, circumferential grooves, and insufficient planarity (concave ring surface) can result in a damaging gap between the contact surfaces, at least in local areas. This then creates a danger of bonding flaws.
If the welding joints are not sufficiently clean and/or air is trapped between them, this fouling can cause porosity and cracking in spite of the vacuum (Fig. "Repair of EB welding flaws").

Base material: Material conditions suitable for welding (a usual criterion is the tendency to hot cracking), which especially include suitable grain size (in Ni alloys, usually fine grain, Fig. "Influences at micro-cracking"), are prerequisites for high weld seam quality.

Process parameters: Process parameters, such as beam intensity, focusing, and welding speed, that are not optimized to the welding cross-section and material, can compromise the seam geometry (sagging, weld penetration, poor root fusion) and also promote flaws such as cracking and porosity. These flaws tend to occur especially at the end of the seam where the beam runs out.

If flaws are discovered after welding, especially in the seam head area, they can often be repaired through additional welding. This proven method is possible because the vacuum has prevented oxidation of the weld seam. In the right column of Fig. "Repair of EB welding flaws", repair welding mechanisms are linked to special welding flaws. Of course, these safety-relevant repairs must all be tested, documented, and certified (Fig. "Minimizing scrap rates throuch reworking").

Figure "EB welding flaws by misalignment": A prerequisite for the input of welding energy during EBW is that the electron beam strikes the metallic surface. Therefore, the joint being welded must be positioned precisely below the very thin (about 0.5 mm, Fig. "Electron beam welding (EB) process") electron beam. This ensures a bond across the entire surface. These conditions may be lost due to thermal strain during the welding process. The diagram (Ref. 16.2.1.3-25) shows an example of heat strain/transverse shrinkage of an EBW seam on sheets made from CrNi steel. Naturally, the transverse shrinkage following cooling only represents a part of the thermal strain that occurred during welding, and therefore only concerns an order of magnitude. Understandably, strain decreases with the welding speed, i.e. narrower seams (less heat input). The same applies to the cross-section thickness, i.e. stiffness and heat capacity. At lower welding speeds, the strain is in the range of tenths of a millimeter. In the diagrams, the example of a circumferential weld seam is used to show the effects of seam length on the welded joint. If the jig is not optimal, the rising heat levels and increasing thermal strain can cause the weld joint to shift out of the beam plane, which compromises seam quality. A shift in the range of tenths of a millimeter is sufficient to cause flaws.

Figure "EB welding flaws influencing fatigue": Pores in EBW seams cannot always be prevented. One cause may be unfavorable conditions for non-destructive testing. Material characteristics and welding conditions may make pores unavoidable. It is also possible that pores will be discovered at a later time, e.g. during an overhaul, with no damage occurring. In these cases, it is important to estimate the influence of the flaws. This may allow calculation of a tolerable limit for the part loads during operation. The bottom diagram shows the results of cyclic stress tests in flat specimens of electron beam-welded hardened steels (gear shafts) with welding pores (left diagram) and bond flaws. Using the various fracture load cycle numbers, fatigue strength values were estimated as shown in the top diagram. These were normalized with the fatigue strength of pore-free specimens (bottom diagram). One can see scatter bands in which the weakening caused by bond flaws is considerably greater than that caused by pores, as expected. This result shows that, in the case of known dynamic loads, it is useful to specify the size of allowable welding pores (Ills. 16.2.1.3-30 and 16.2.1.3-31).

Figure "Strength of EB welding pores in Titanium" (Ref. 16.2.1.3-14): The effect of welding pores on dynamic and static strength not only depends on the pore size. Another important factor is the fracture toughness, i.e. the resistance that a material has to crack growth. This toughness is usually lower in the cast structure of the weld seam than it is in the uninfluenced base material (bottom diagram). This means that the fracture strength and LCF strength react more strongly to flaws such as pores (grey scatter areas in the top chart) in the welded structure. When estimating the allowable pore sizes, therefore, one must take into account the worse fracture-mechanical behavior of the weld seam relative to the base material. The best approach, although also the most costly and time-intensive, involves part-specific specimen tests (Fig. "Pore size influencing LCF Life").

Figure "Pore size influencing LCF Life" (Ref. 16.2.1.3-14): This diagram shows the influence of pores on the LCF behavior of EBW seams in a high-strength titanium alloy. In addition to the grey scatter bands from specimen tests, calculated lines (dashed) are plotted to show the constant stress concentration and aid estimation of the effects. The constant stress concentration corresponds to the fracture toughness in the specimens for the respective scatter band. Here, as well, the pore size-dependent scatter bands present a possibility for defining an application-specific maximum allowable pore size.

Figure "EB welding flaws in a gear wheel": This diagram shows a dynamic crack in a gear shaft with a flawed EB weld. The crack runs tangentially into the circumferential bead in the rotor disk (top left diagram). It originates in pores (spikes) and cold laps in the root (details). Due to the close proximity of the second gear, the root of the weld seam could not be visually tested, and X-ray testing was also severely restricted. Even if this type of flaw can be found sufficiently safely with serially-implementable non-destructive testing methods (X-rays), weld seams should generally be positioned in easily testable areas outside of part zones under high dynamic stress.

Figure "Design of EB welds" (Ref. 16.2.1.3-25): This composition is intended to give design engineers tips for configuring EB welds. The criteria are susceptibility to flaws and production costs. In each set, the left diagram shows the condition of the part before welding.

Figure "Process of fly wheel inertia or friction welding" (Ref. 16.2.1.3-1): This section deals with inertia welding, which is the most commonly used friction welding method for engine rotors. The weld quality is ensured by the continuous documentation of the machine parameters. This is necessary because kissing bonds (Ills. 16.2.1.3-37 and 16.2.1.3-39), although rare, may occur and cannot be sufficiently safely detected using serially implementable non-destructive testing methods. The process parameters, which influence one another, have a very tight scatter pattern. Therefore, even small deviations from the tested and documented machine data can be recognized before welding flaws occur. Inertia welding can only be used to join pipe cross-sections and, less effectively, solid cross-sections (Fig. "Cracks in friction weldings"). First, one side of the work part is clamped into a flywheel and accelerated to the required RPM, i.e. kinetic energy of the flywheel. Next, the static, firmly clamped opposite work piece is pressed against the rotating piece with a rapidly increasing and then constant pressure. The pressure, torque, and friction result from the self-regulating friction conditions. For this reason, the machine data (process data) for inertia welding must first be determined and optimized in sufficiently realistic part-specific tests. In addition, there are other important influences that may need to be monitored. One example is titanium alloys, for which a sufficient protective gas cover must be guaranteed.

Figure "Flaws of friction welding": Experience has shown that, if the tested and proven process parameters for friction welding are observed, dangerous flaws are extremely rare. In spite of this, various welding flaws are possible in friction welds. These can require especially costly and extensive actions if they are not detected by the process monitoring that acts to ensure quality.

The top right detail shows a slice through a friction weld on a ring-shaped cross-section with typical characteristics. Compared with the weld seams of other processes, the most obvious characteristic of this untreated seam is the typical pronounced, more or less curled weld flash. If the material and geometry are symmetrical, the flash will be identical on both sides of the joint. The flash can have pronounced axial cracking (hot tears) due to the extreme plastic deformations. As long as these cracks are removed along with the flash, they will not affect the quality of the weld. The repeatable formation of the flash is a sign of the stability of the process. If different materials or butts with different stiffness are joined, the flash will not be symmetrical to the joint.

In the seam plane, there is a heavily deformed structure with fine grains that are directed outward. The heat-influenced zone (HAZ) on both sides of the joint may already have more coarsely grained, recrystallized structure relative to the uninfluenced base material. Due to the process, there should not be any signs of solidified melt.
The bottom diagram is a composite of potential problems, weak points, and flaws.

Hot tears: Although there is no melted zone in the joint, hot tears can occur if the material properties are unfavorable. Doughy grain boundaries are torn open by the high welding forces and/or thermal stress, even without melting (Fig. "Influences at micro-cracking"). Experience has shown that coarse-grained Ni alloys are prone to hot tears. This problem occurs in small turbine disks and turbochargers that have been welded to a steel shaft (diagram). The cracking is especially intensive if segregations and micropores have accumulated in the material zones that were last to solidify, such as, typically, hub attachments on the rear side. In the case of these unavoidable cracks, either the joining mechanism must be changed (e.g. soldering), or the weld must be moved to an unproblematic zone. If this is not possible, then tests and sufficient understanding of the expected operating loads are necessary to use a fracture-mechanical assessment in order to ensure safe operation despite the cracks.

Insufficient fusion/kissing bonds: These dangerous flaws, which are somtimes large and can cause joint failures (Fig. "Cracks in friction weldings") are very difficult to detect using non-destructive methods. It is doubtful that process monitoring can sufficiently safely detect the formation of these flaws. Potential causes are influences that locally alter the welding process by changing the friction conditions. Potentially dangerous media include high-temperature dry lubricants such as hexagonal boron nitride, molybdenum disulfide, and graphite on butting surfaces.

Brittle phases: Friction welding can be used to join very different materials. It is important that the required plastic deformation (flow behavior) of both materials during the welding process is in roughly the same temperature range. Unfavorable combinations of different materials can lead to the formation of brittle phases during welding. Even subsequent heat treatment and/or sufficiently high operating temperatures have an embrittling effect in this way. One example is the combination of Ti alloys with Ni alloys or steels. In this case, the weld seam itself can become a weak point. Under impact stress (e.g. as consequential damages) the result may be spontaneous fracture.

Embrittlement with brittle phases can also occur in titanium alloys in contact with air (Fig. "Weld quality by cover gas"). This must be considered especially with linear friction welding, because the oscillating welding motion and tipping of the friction surfaces increases the risk of oxygen intake on the welding surface (Ref. 16.2.1.3-1).

Spontaneous cracking/stress corrosion cracking (SCC): High-strength titanium alloys are very sensitive to this type of corrosion cracking around notches and under the influence of halogens such as chlorine (Fig. "Stress corrosion cracking by process baths and hand sweat"). These axial cracks (Fig. "Cracks in friction weldings") tend to originate in the typical acceptable cracks in the flash. Apparently, even minimal residue from hand sweat (finger prints) is sufficient to cause cracking in the welding joint area. It is also possible that very thin Cl reaction zones at the surface could have a similar influence. These extremely thin layers could form during cleaning of the titanium parts before welding, using baths containing Cl (“Tri”, “Per”; Fig. "Chlorine in process baths causing stress corrosion").

Welding spatter: Similar to fusion welding, friction welding can also create welding spatter. However, in this case, the spatter does not consist of molten drops, but rather glowing particles from the friction process. Sufficiently hot titanium alloy particles burn as they fly through the air. This additional heat could even make melting possible. If these particles strike part zones that are under high stress during later operation (e.g. inner surfaces of rotors, such as hubs and disk membranes, Fig. "Damaging metal splashes abd sparks"), the dangerous drop in dynamic strength (Fig. "Preventing contact with welding splashes") can cause LCF fractures. For this reason, even during friction welding, these zones should be suitably covered as they would be during electron beam welding.

Figure "Cracks in friction weldings": Friction welding can create considerable form notches due to the process mechanism. These can affect both the operating safety as well as consequential damages in the finishing process.
The inner side of a hollow shaft is especially interesting in this regard, if the flash cannot be machined. This notch will be even more dangerous if there is a lack of fusion between the sides of the flash (top left diagram). An additonal stress-increasing phenomenon is a non-reworkable shaft deflection on the inside.
If the part in question is a solid shaft (top right diagram), the friction conditions become worse towards the center (low relative movements, flash outflow restricted). This makes it more likely for flaws to occur in the center of the welding plane. As long as this part of the shaft is in an area that is not subjected to high stresses, it may not affect the operating safety of the part. However, if the flaw reaches or spreads into a more highly stressed zone, sudden shaft fractures can be expected.
The stress on the inside of the pipe is fairly low under flexural stress. However, if significant centrifugal forces and/or heat stress occurs, LCF crack growth may well originate from an inner, sufficiently large flaw. This may occur if materials with different thermal strain coefficients are joined together. One example is a turbine disk made from an Ni alloy that is joined with a steel disk (bottom left diagram, Fig. "Flaws of friction welding"). The stresses can be especially high if the weld is too close to the back side of the disk. This type of crack will only be externally detectable, e.g. through penetrant testing, when it has already dangerously weakened the cross-section. For this reason, these cracks cannot be caught before they cause shaft or disk fractures.

Figure "‘Kissing bonds’ in diffusion weldings": A prerequisite for a diffusion connection is an uninterrupted intensive contact of the metal lattices in the welding joints. This is also the basis for process-specific problems and flaws. Every gap between the joining surfaces, no matter how small, results in a flaw (top diagram). The causes of these gaps include dimensional irregularities, thermal deformation, and oxide skins. The typical result is kissing bonds, which are a weakened connection (Ref. 16.2.1.3-33) caused by collections of tiny pores of about 10 mm (Ref. 16.2.1.3-1). This partial fusion does not allow sufficiently accurate and reliable non-destructive testing (Ref. 16.2.1.3-32, Fig. "Fracture of a diffusion welded titanium blade"). Diffusion welding does not permit quality assurance through process monitoring, as can be done with friction welding (Fig. "Process of fly wheel inertia or friction welding"). This is because the formation of the flaws is not reciprocally related to the process parameters. For this reason, it is not recommended to use this type of weld in part zones that are expected to experience significant shear stresses.
For titanium alloys, unlike Ni alloys, diffusion welding has proven successful due to the safe contact of the joined surfaces. An example of this is hollow fan blades (Volume 3, Ills. 13-25, 16.2.1.3-39). According to Ref. 16.2.1.3-1, the danger of fusion problems is further “…minimized, because the solution rate (diffusion rate) of oxygen and nitrogen is so high in titanium (Getter effect), that even the unavoidable oxide skins do not interfere in the early stages and are dissolved.”
The necessary tight contact between the joining surfaces is aided by the superplastic behavior of the titanium alloys at the welding temperatures (about 950°C). Even a minor differential pressure is sufficient to securely press the evacuated and sealed butt surfaces. However, these measures are evidently not enough to completely prevent flaws (Ill. 16.2.1.3.-39). Due to their thermal resistance, Ni alloys require much greater pressure. For this reason, configurations such as axially-stressed conical surfaces (Fig. "Diffusion weld of a ‘dual property’ part") and the high pressures of HIP presses come into use. In spite of this, the safety of these welds in Ni alloys could not be satisfactorily ensured.

Figure "Diffusion weld of a ‘dual property’ part" (Ref. 16.2.1.3-1): In accordance with the dual property or hybrid turbine wheel concepts (top diagram, Ref. 16.2.1.3-35), in turbine disks, the task is to combine an integral cast annulus (high thermal resistance but low toughness) with a forged hub (high LCF strength and toughness at low temperatures). Both parts are made of Ni alloys that are optimally matched to the specific temperature ranges of these part zones. Evidently, there have even been attempts to weld rings with cooled turbine blades (Ref. 16.2.1.3-35).
In at least one case, attempts to use a conical diffusion weld to join a blade annulus with a hub (Ref. 16.2.1.3-1) failed due to bonding flaws that could not be sufficiently safely detected (Ref. 15.2.1.3-32). Ultrasonic testing was hampered by the coarse cast structure of the annulus. Therefore, the number of cycles to failure in overspeed tests scattered unacceptably. Fractures even occurred during startup (bottom diagram). On the other hand, some parts reached life spans of several thousand cycles without damage. Parts with short life spans could not be differentiated from those with long life spans on the basis of any quality-assuring test results.

Figure "Fracture of a diffusion welded titanium blade" (Ref. 16.2.1.3-34): This example shows that even in titanium alloys, problems can occur with insufficient fusion in diffusion welds. The engine damage was the result of a fan blade fracture that occurred at the dovetail root. A dynamic fatigue crack originating in a production flaw (created 10 years before the damage!) in the middle of the blade root led to the damage (top diagram, reconstructed from the information in the literature). The lack of bond (Fig. "Diffusion weld of a ‘dual property’ part") was at the transition between the two sides of the blade. This plane, which is under centrifugal forces, is not affected by any significant detectable shear stress. Possible flexural modes should not have much effect in this area, which is in the neutral axis. Evidently, the dynamic fatigue fracture spread across the welding plane, and thus across the flaw, corresponding to the high loads.
The hollow fan blade was made of two high-strength titanium sheets that were diffusion-welded together. Due to the process-specific difficulty of obtaining flawless connections in the blade root, criteria for the tolerability of certain flaws (weak points) had been established. Weak points of up to 6 mm were considered tolerable. Although a flaw of this type was detected by X-rays in the damaged part during the production process, the evaluation criteria resulted in it being deemed tolerable. Unfortunately, the weak point was in one of the two most highly stressed zones in the part, and was therefore threatened by cyclical crack growth despite its longitudinal orientation. Once this was understood, the evaluation regulations were tightened. The weak point was reclassified as a flaw, and was therefore not tolerable.
The original bond-line defect was elliptical and spread about 12 mm along the blade chord. It grew to 22 mm due to additional loss of bond, and shows the low strength in this area.