Aeroengine Safety

A compressor outlet housing and tailpipe tab from an older engine type; they were produced as welded constructions.

Figure "Welded parts": In the case of fusion welds, process-specific (unavoidable) weak points (often called discontinuities; Ref. 16.2.1.3-7) are expected. Even the strength properties, especially the dynamic strength, of the cast structure of the weld bead are usually below those of the base material. In addition, the welding heat also influences the surrounding unmelted base material (heat affected zone = HAZ). The top diagrams show two examples of the (schematic) hardness pattern around the bead, as a characteristic of the material strength. The left hardness pattern with a drop in the weld and HAZ can be observed in hardenable materials such as Ni alloys (Waspaloy, etc.; Ref. 16.2.1.3-12), high-alloy steels (e.g. A-286), and aluminum alloys (e.g. AlMg alloys, Ref. 16.2.1.3-4).

The right diagram shows the hardness pattern in a martensite-forming steel (e.g. St-52; Volume 3, Ref. 12.2.1.3-4). In this case, clear hardness peaks form in the coarse grain zone (recrystallized zone, Fig. "Hot cracks by heat treatment") of the HAZ. If cooling is sufficiently fast, undesirable high increases in hardness can be expected in the weld bead. This type of hardness pattern can be prevented through suitable temperature control during welding and/or subsequent heat treatment.
In addition to material-specific structural effects, the strength of a specification-conforming weld can also depend on a multitude of other factors that are typical for certain processes, and are dealt with briefly in the following.

Separation and hollow spaces: The size and type of preventable flaws (e.g. gas pores, cavities, cracks, kissing bonds) depend largely on the available, serially-implementable non-destructive testing methods and the accessibility of the weld bead. These must be the point of reference for the utilizable minimum strength, which is fundamental to a design.

Notches: There are many possible types of notches (also see Volume 3, Ill. 13-18) that can occur around weld beads (Ills. 16.2.1.3-2 and 16.2.1.3-7). These include form notches such as pores, cracks, undercuts, seam reinforcements, and shifted welds. The second major group includes structural notches, which can be traced back to differences in structure between the weld bead (cast), HAZ (heat-affected base material), and the unaffected base material.

Residual stresses (Fig. "Residual stresses of welds") are dependent on many influences, including part geometry, processes, process parameters, temperature control, and subsequent heat treatment. Even after stress relief annealing (Fig. "Changes by residual stresses from machining"), residual stresses can be expected.

Unfavorable material properties can result in an increased sensitivity to corrosion. In unstabilized CrNi steels, the HAZ becomes sensitized. This occurs due to carbide formation (Cr carbides) and makes the grain boundaries susceptible to corrosion. This promotes intergranular corrosion (IC, IGC; Volume 1, Ills. 5.4.1.1.-3 and 5.4.1.1.-4). Increased sensitivity to stress corrosion cracking, together with a decrease in fracture toughness, can result from structural changes.

The cast structure of the weld bead often has lower toughness than the base material. This can be observed in the impact strength (notch impact strength) and in a lower fracture toughness. This means that shorter cracks can become critical and lead to spontaneous failure. Even under impact loads, such as during a containment incident, reduced fracture toughness may become safety-relevant. The directionally-solidified structure of the weld bead and/or a strength decrease in the HAZ can increase the crack growth rate and determine its direction.
If rewelding or repairing of the bead is required, the weld bead structure, which is not as good as that of the uninfluenced base material, can promote the formation of hot cracks.

Figure "Welds as notches": Unworked welds usually change the part contour and act as form notches. The following problems are common:

Specification-conforming weld bead development: Even unworked weld seams that conform to specifications can reduce the dynamic strength of a part considerably due to their shape. The frame in the illustration shows the notch effects of various typical weld beads/welding configurations. To aid understanding, each of these includes a sketch of the weld, as well as a cross-section of a reworked smooth welded plate with identical dynamic strength (without regard to other effects such as structural influences). Naturally, the differences increase along with the notch effect.
Clearly, even welds conforming to specifications, with no reworking, will result in a dimensionally-dependent decrease in dynamic strength.
Undercut (top left diagrams): This refers to grooves that run parallel to the weld bead and are created during the welding process. They are usually the result of excessively high current and/or poor electrode positioning (Ref. 16.2.1.3-17). Undercut can occur both at the bead head and near the bead root. It primarily occurs in the base material. The dynamic strength losses caused by less pronounced undercut can usually be tolerated (Fig. "Influence of weld mismatcheson dynamic fatigue", Ref. 16.2.1.3-12). However, pronounced undercut must be classified as a flaw (Ref. 16.2.1.3-7).

Overlap (top diagrams, second from right): The bead head overlaps with the base material at the bead edge, but does not bond.

Weld reinforcement (top right diagram): This is material buildup at the bead head resulting in a bulge that meets with the base material at a steep angle (reinforcement angle). The influence of the weld reinforcement on the dynamic strength depends on the reinforcement angle (diagram, Ref. 16.2.1.3-18 and top sketches in the frame). For some common bead shapes, the notch effect number a_k is provided. It reaches 1.8 in the case of pronounced weld reinforcement. This corresponds to a significant reduction of dynamic strength due to the bead head shape.

Illustration 16.2.1.3-3 (Ref. 16.2.1.3-19): In this case, the weld was evidently in an area that was highly stressed by vibrations of the fuel line. The crack ran parallel to the weld bead. There were no noticeable flaws or unusual bead formation. This indicates that the cause was unusually high dynamic loads and/or assembly-related twisting. The crack pattern is in line with the notch effect that is usually expected at the weld bead transition to the base material in this type of weld (form notch, structural notch, Fig. "Welds as notches").
This example underlines the importance of generally positioning fusion welds outside of part zones under high dynamic stress.

Figure "Influence of weld mismatcheson dynamic fatigue" (Ref. 16.2.1.3-12): This diagram uses an example to show the influence of weld joint mismatches on dynamic strength. Although the material in question is a high-strength titanium alloy, similar conditions should apply to Ni alloys and steels. Only mismatches greater than 15% of the sheet thickness will show a clear drop in dynamic strength beyond the typical range of dispersion for welds. A mismatch of 20% will result in a dynamic strength loss of about 50%. In addition to the actual cracking, weld mismatches have other negative effects. The laterally stressed longitudinal notch created by the mismatch determines the direction of crack growth and accelerates it. This promotes very rapid, uncontrollable cyclical crack growth with a premature sudden forced fracture. For this reason, this type of welding flaw is an intolerable safety risk in highly-stressed parts such as pressurized housings (e.g. combustion chamber liners; Fig. "Pressure stress of engine housings").

Figure "Pressure stress of engine housings" (also see Volume 3, Ill. 11.2.2.2-9): In a two-engine fighter of an older design, a weld mismatch of the longitudinal bead along the horizontal flange (bottom left diagram) of the combustion chamber liner led to the formation of an LCF crack. Because the crack first spread into the notch on the inside of the wall and only spread outward at a late stage, it was not detected in time. The housing explosively ruptured and several can-type combustors escaped. Fortunately, the explosion was directed toward the outside of the fuselage (top diagram), so the aircraft was able to land safely using the parallel engine.

Illustrations 16.2.1.3-6 and 16.2.1.3-7: A special problem are so-called tack welds. These are usually weld points or short weld beads. They are used to position metal sheets that will be welded or soldered together. Experience has shown that these welds are frequently not given the attention they require. Tack welds are subjected to various high operating loads and are therefore a predestined point of origin for dynamic fatigue cracks (middle diagram in Ill.
16.2.1.3-7, Ref. 16.2.1.3-15).
Due to poorly defined process parameters (e.g. temperature control, torch position) and their small size, tack welds can remain undetected, unlike relatively pronounced undercut and/or weld reinforcements in joining beads.
If tack welds are welded over, they may have a negative influence on this area. The resulting poorly weldable structure and tension residual stresses promote cracking in the joining bead running over the tack weld.
If tack welds are located outside of the actual joining bead, such as a roller bead (Fig. "Tack welding risks of dynanic loade parts", middle diagram), they act as special stress concentrations (Fig. "Risk of tack welds" bottom diagram). This effect should be especially pronounced in soldered connections with a modulus of elasticity lower than those of the weld and base material (Fig. "Tack welding risks of dynanic loade parts", bottom diagram). In case of flexure (flexural modes, etc.) in soldered or spot-welded sheet metal structures, the tack weld is especially exposed to vibrational stress.
Another special problem is fixing soldering gaps in heat-treated steels. These can harden and/or suffer a significant strength loss in the heat-influenced zone. The bottom diagram in Fig. "Tack welding risks of dynanic loade parts" shows an extreme case. During a repair development process, it was temporarily attempted to use tack welds to fasten a cylindrical wear sleeve to a steel rotor disk, which would later be soldered. This would have resulted in a dangerous loss of strength and risk of disk failure due to influences such as annealing, hardening, embrittlement, and tension residual stresses.

Figure "Strength reducing influences on welds": Unlike weak points (discontinuities, Ills. 16.2.1.3-1 and 16.2.1.3-2) welding flaws, i.e. deviations outside the specifications/design data, must be prevented with sufficient reliability. This diagram is a composition of typical welding flaws (Refs. 16.2.1.3-4, 16.2.1.3-24), some of which are dealt with in more depth in other sections (e.g. hot cracks, Fig. "Hot cracks by heat treatment"). Some of the flaws are material-specific, including the embrittlement of titanium through oxygen absorption or oxide formation. This occurs when air comes into contact with the work surface due to insufficient shielding gas cover at material temperatures above 600 °C. Another material-specific flaw is the sensitization of CrNi steels, with a risk of later intercrystalline corrosion (Fig. "Welded parts"). Martensite steels are at higher risk of embrittlement through hydrogen absorption from moisture in the weld pool. The top detail shows an SEM image (Fig. "Scanning electron microscopy (SEM)") with typical signs of hydrogen embrittlement in steel (Ref. 16.2.1.3-21). Even titanium alloys show brittle cracking around pores (Ref. 16.2.1.3-22, Fig. "Effects of welding flaws on LCF") in solute gases such as hydrogen.
The damaging influence of welding spatter or arc burns (Chapter 16.2.2.6) located outside of the weld must not be underestimated. Unfortunately, these are often found in more highly stressed part zones.
The welding root is a unique problem zone. This is partially due to an accessibility-dependent greater susceptibility to flaws. Also, there is a limited shielding gas supply in this area. Additionally, visual inspections of this area are often restricted, at best.

Figure "Effects of welding flaws on LCF" (Ref. 16.2.1.3-12): This chart plots the cyclical life span over the stress concentration of flaws in specimens with a fusion weld. The load is a sinusoidal cyclical stress (Fig. "Influences of weld flaws on fatigue"). The scatter band applies to high-strength titanium alloys and limits the incubation time to noticeable crack growth (detail). The operating loads are in the LCF range. One can see that the welding porosity typical for this material (Ills. 16.2.1.3-15, 16.2.1.3-17, and 16.2.1.3-24) only shortens the LCF life by up to 20%. In many cases, a stress concentration in the given scatter range is tolerable.
It must be remembered that dwell time fatigue (DTF) can seriously accelerate cracking (Volume 3, Ill. 12.6.1-20). The life span during crack growth is influenced by an overlaying of residual stresses and cyclical operating loads.
This also does not take into consideration spontaneous brittle micro-cracking due to pore cracking (Ref. 16.2.1.3-22) caused by gas absorption and unfavorable structures. This can appear as sustained load cracking (SLC) resulting from residual stresses (Fig. "Pores causing cracks in titanium alloys").
In cases of DTF and SLC, there is no pronounced relationship between pore size and cracking (Ref. 16.2.1.3-22), thus limiting the applicability of the diagram considerably.

Figure "Hot cracks by heat treatment": Typical welding flaws, especially in hardenable high-alloy steels and Ni alloys, include cracks. These are primarily hot cracks or hot tears (Ref. 16.2.1.3-10, Ills. 16.2.1.3-11 and 15.1-8), cold cracks and stress relief cracks (Fig. "Location of welds and finishing cracks" and Table 16.2.1.3-1).

Hot cracks are often the criterion for the weldability of a material (Fig. "Weldability influecing repairability"). They can occur both during the heating and cooling phases (top diagram, also Fig. "Dynamic fatigue of resistance welds" ). They have similar damage mechanisms. Softened or melted phases on grain boundaries (HAZ) and/or between dendrites (melt/weld metal) locally reduce the strength so that cracks occur in the tension stress phase. The left detail shows the SEM image (Fig. "Scanning electron microscopy (SEM)") of a cracked grain boundary. The typical doughy characteristics (Ref. 16.2.1.3-21) that explain the damage mechanism can be seen clearly (SEM image, center of diagram). In order to develop specific corrective measures, it is important to know the various aspects of hot cracking. The literature (Ref. 16.2.1.3-4) differentiates between solidification cracks between dendrites in the weld metal and remelting cracks in the coarse grain zone of the HAZ. Grain growth following recrystallization can be an important factor if the grain boundaries are enriched by lower-melting phases during the growth process. The melting point of these grain boundaries drops during cooling, even though the melt is already solidified. In both of the described cases, cracking occurs during the cooling phase and can be categorized as cooling cracks.

Fig. "EB welding flaws" seems to show that cracks do not only occur during the cooling phase, but also during heating. This diagram shows the EB welding of a forged Ni alloy. The melt has obviously penetrated into the open grain boundary of the HAZ. It is possible that the softened grain boundaries in the HAZ, i.e. next to the weld, are subjected to tension loads during the rapid heating of the bead area under the arc or energy beam (electron beam, laser beam). These tension loads must balance out the compressive stress in the heated zone. This process should be promoted especially by the intensive rapid heat input of EBW and laser beams. In these cases, one can observe the melt that fills the crack heals it. However, if cracks still remain (Ref. 16.2.1.3-1), they are referred to as heating cracks.

Solutions for hot tears:

• Optimization of the welding process, and especially the temperature control over the heat input (welding speed, Fig. "Influences at micro-cracking") and heat removal (cross-section, equipment).
• Optimization of the material: selecting weldable material (Fig. "Weldability influecing repairability").
  
• Fine-grained semi-finished parts (Fig. "Influences at micro-cracking") require excellent processing conditions during raw part production (part size, temperature control, etc.). The influence of the position of the finished part in the raw part is also important. This factor is largely determined by the reforming and temperature patterns, and therefore the structure (dependence on forging process, see Fig. "Forging process caused characteristics").
  
• Optimized heat treatment condition (e.g. solution annealed, Fig. "Influences at micro-cracking"): Tightening specifications regarding tolerable foulings and trace alloy elements.
  

Figure "Weldability influecing repairability": The trend is that the higher the high-temperature strength of a material is, in connection with its Al and Ti content (Ref. 12.2..1.3-16), the more difficult it is to weld. The primary criterion is hot cracking (Fig. "Hot cracks by heat treatment"). In this regard, high-strength powder-metallurgical materials must be viewed especially skeptically. The impossibility of welding wear-resistant layers onto integral labyrinth tips may prevent the realization of an intermediate labyrinth ring made from an especially high-temperature-resistant PM material. This is especially unpleasant if it is not until repairs are done that it becomes clear that repair welding on worn labyrinth tips is not possible. Usually, high-strength parts are subjected to correspondingly high loads during operation. In these parts, even very small welding-related flaws can lead to crack growth. The worse their weldability due to micro- and macro-cracking, strength losses, residual stresses, and deformations, the more elaborate and scrap-prone welding work will be, if it can even be accomplished in accordance with the quality demands. Not only the welding procedure itself, but also the subsequent quality assurance and testing for internal micro-cracks (hot tears) are prerequisites for welding to conform to specifications. The operating behavior of welded repairs may be compromised relative to the original conditions due to a loss of strength in the weld and transition zones.

Figure "Location of welds and finishing cracks" and Table 16.2.1.3-1 (Refs. 16.2.1.3-4, 16.2.1.3-5, 16.2.1.3-10, and 16.2.1.3-24): In addition to hot cracks, there are several other cracking mechanisms (Fig. "Hot cracks by heat treatment"). This diagram, combined with the table at right, are intended to provide an overview of the types, locations, and causes of cracks. In case of damage, this should aid with the identification of the damage mechanisms, which is a prerequisite for targeted corrective measures.
In addition to hot cracks, which occur most frequently in the materials used in modern engine construction (especially Ni alloys), there are also cracks that only form after the welding process and are referred to as cold cracks (delayed cracking). Important initiators for cold cracks are hydrogen embrittlement, structural anomalies, tension residual stresses, and corrosion.
Stress relief cracks are another type of delayed cracking. These do not originate from fluid or doughy grain boundaries, but form in a solid state at increased temperatures (heat treatment) through creep processes (Fig. "Changes by residual stresses from machining").

Table 16.2.1.3-1

Figure "Micro hot cracks in welds": This diagram depicts two typical examples of micro-hot-cracks in Ni-based alloys:
Fretting zones are protected by welded wear-resistant layers (usually stellites). These are typically found on rotor blade shrouds in low-pressure turbines. In this case, the contact surfaces of the joined shrouds must be welded on (top diagram). In addtion, it may be necessary (e.g. during repairs) to weld on labyrinth tips that were worn off by rubbing. A complicating factor related to hot cracking is the fact that the base material is a cast alloy that has the high Ti and Al content necessary to guarantee creep strength. The tendency for hot cracking in these materials is therefore very high, which results in a poorly definable weldability (Fig. "Weldability influecing repairability"). For this reason, a certain number and size of hot cracks are tolerated in the specifications. This weakness must be compensated constructively through the location and geometric limiting of the weld, as well as sufficiently low operating loads.
The bottom diagram shows the hollow shaft of a high-pressure part. This forged part is made from an Ni alloy (Ref. 16.2.1.3-4) and is assembled from three parts by electron beam welding. At one of the circumferential seams, serious micro-hot-cracking occurred, primarily in the HAZ (detail). This concentrated on the seam side of the coarse-grained forged ring for the flange (Fig. "Influences at micro-cracking"). The relatively low operating loads in the seam area made it possible to tolerate tightly limited crack concentrations and crack lengths. The cracks only appear as shadows in X-ray inspections, necessitating especially experienced interpreters for these findings.

Figure "Influences at micro-cracking" (Ref. 16.2.1.3-4): Hot tears in fusion welds can be observed especially clearly in electron beam welds on forged Ni alloys (top diagrams). Cracking occurs primarily in the HAZ, but can also spread from this area into the weld. In general, the finer the grain of the material, the less pronounced the hot tears (middle left diagram). This behavior is related to reduced grain boundary coverage and alloy components that lower the melting point. The finer the grain of a material, the larger the grain boundary surface on which the coverage can disperse. This lowers the concentration of these elements on the grain boundaries. Below a certain grain size, no micro-cracking will occur.
Electron beam weld seams appear to be especially affected by micro-hot-tears. This seems to be due to several factors. At least until friction welding was introduced, highly-stressed rotor parts that were used in areas with high operating temperatures were welded using electron beams. These parts require high-strength forged Ni alloys. Due to their special composition (high Al content), they are prone to hot tears (Fig. "Hot cracks by heat treatment"). The bottom diagram seems to point to an additional influence. Surprisingly, hot tear formation increases in narrower welds (in this case, TIG welds). Narrower seams require higher welding speeds with correspondingly intense heat input. This is especially high during EBW, as reflected in the typical narrow seams created by this method. It can be speculated that these cracks occur in the heating phase (also see Ills. 16.2.1.3-10 and 16.2.1.3-25). Relative to the surrounding material, the material in the seam area is very rapidly heated and expands, creating compressive stresses as long as the material doesn`t melt. In the surrounding cooler material, powerful tension stresses are created as a balance. If the doughy grain boundaries break open, they result in micro-cracks, which evidently at least partially fill with melt and heal up (Fig. "EB welding flaws").

The diagram at middle right shows the importance of an optimal material state, i.e. material-specific heat treatment.
The increase in the hot tears along with the annealing temperature probably does not have universal applicability. Experience has shown that other materials, such as high-alloy steels, appear to deviate from this trend.
In the depicted case of a hardenable and forgeable Ni alloy, the micro-cracking increased along with the solution annealing temperature, contrary to expectations. Was this effect due to the solution annealing lowering the strength levels? This would mean that the tension stresses during heating could more easily overstress the material. It is also plausible that grain growth may promote more pronounced grain boundary coverage with a corresponding lowering of the melting point. This effect is assumed to occur in the HAZ during grain growth following recrystallization (Fig. "Hot cracks by heat treatment"). This is in agreement with the negative effect of a higher annealing temperature and increased recrystallization.
Another interesting factor is the influence of suppliers. This shows the importance of raw part production (forging, Fig. "Forging process caused characteristics"). One explanation may be that the raw parts have different degrees of reforming and hardenings, which are transferred into the engine part. This would have a decisive influence on the recrystallization rate, i.e. the grain growth (Fig. "Flaws in forged rotor disks").

Figure "Influences of weld flaws on fatigue" (Ref. 16.2.1.3-12): Unlike in Ni-based alloys and steels, the most common flaw type in welds on titanium alloys is porosity. Welding porosity can have various causes:

• Quality problems with the base or fill material
• Moisture in the melting pool
• Surface fouling
• Formation of spikes in electron beam welds (Ills. 16.2.1.3-24 and 16.2.1.3-30).

Research has revealed that the influencing of the dynamic strength (LCF strength, see top diagram) is decisively dependent on the distance of the pores to the weld seam surface. Depending on their diameter, the closer the pores are to the surface, the lower the dynamic strength, i.e. number of load cycles to fracture.

Phenomena that are considerably rarer than pores include cracks or metal inclusions. Hard metal inclusions (WC) can originate from remelted chips that contain broken cutting edges from tools. For this reason, the use of chip removal machining is prohibited for the production of raw parts for rotor parts. The frequency of material-independent flaws such as joint mismatches, undercut, and insufficient root fusion is in the typical range for fusion welds.

The bottom diagram shows the influence of subsequent surface machining and crack-like flaws such as lack of fusion (Fig. "Dynamic fatigue of resistance welds") on dynamic strength. The dynamic strength of normal machined surfaces is similar to that of welds containing the usual weak points. These include small pores that are sufficiently far from the surface. On the other hand, it can be assumed that these weak points will be covered by the influence of the usual surface machining. This is not the case with surface porosity, which can evidently lower the dynamic strength to 30%. Lack of fusion has a similar effect on dynamic strength. For this reason, these flaws are especially dangerous for part safety.

Figure "Residual stresses of welds": The residual stresses in a fusion weld overlay with the later operating loads. Since these include high tensile stresses, they affect the mean stress and considerably reduce the dynamic strength of parts that are under significant dynamic stresses (Fig. "Residual stresses reducing fatigue strength"). For this reason, suitable heat treatment (stress relief annealing, Ills. 16.2.2.4-14 and 16.2.2.4-5.1) and/or surface treatments (shot peening, etc. Fig. "Increasing fatigue strength by compressive residual stresses") are recommended to break down tensile stresses or induce compressive stresses in the surface.
The diagram schematically depicts the typical averaged even residual stress state of a V-weld seam. It shows the residual stress distribution along (“X”) and across (“Y”) the weld seam, within it, and at a greater distance from it. The development of these conditions during the welding and cooling processes is complex because they change during welding as the seam length changes (Fig. "Welding causing deformations and residual stresses"). The level of the residual stresses also depends on the geometry of the part (resilience/stiffness) and the process parameters (Fig. "Residual stresses reducing fatigue strength"), and can be minimized through their optimization.

Figure "Residual stresses reducing fatigue strength": The top diagram (Ref. 16.2.1.3-12) indirectly shows the influence of welding residual stresses (Fig. "Residual stresses of welds") on the utilizable dynamic strength of a high-strength titanium alloy. A comparable effect can also be observed in steels (Ref. 16.2.1.3-4). The influence of the welding residual stresses on the dynamic strength can be compared to rough machining.
Stress relief annealing raises the dynamic strength into the usual range for welds with typical weak points.
Macro-residual stresses require structures in which deformation is restricted. For this reason, it must be critically questioned whether, despite their longitudinal seam, the narrow specimens used are actually capable of building up welding stresses that will realistically demonstrate their influence on the dynamic strength. The greatest residual stresses can be expected perpendicular to the weld seam (Fig. "Residual stresses of welds"). However, these are not accounted for if the testing direction is parallel to the seam. Without stress relief annealing, safety dictates that it must fundamentally be assumed that welding residual stresses will occur in the liquid limit range.
Based on these considerations, it can be assumed that welding residual stresses have a considerably more negative effect on parts than shown in the diagram. A corresponding degree of caution should be taken with welded constructions.
The bottom chart shows the influence of the welding process on residual stresses, in this case in a high-strength titanium alloy (Ref. 16.2.1.3-12). TIG- and MIG-welds are roughly in the same scatter band. However, it is interesting that the electron beam weld seam has considerably greater residual stresses, even if this only represents a single test. This would not be expected from the typically narrow EBW weld seam with comparatively low heat input, which indicates that the narrow specimens may be problematic, as mentioned above.

Figure "Weld quality by cover gas": Metal melts readily absorb gases. During solidification, these partially go out of solution and form pores. Dissolved gases remaining in the metal lattice can have an embrittling effect. This can cause two different types of damage:

• (Gas-) pore formation (Fig. "Influences of weld flaws on fatigue")
• Embrittlement and delayed cold cracking (Fig. "Location of welds and finishing cracks"). In martensite steels, this primarily concerns hydrogen embrittlement. Cold cracking in titanium alloys in connection with pores is dangerous (pore cracking, Ills. 16.2.1.3-8, 15.2-14.1, and 15.2-20).
  
• Oxidation has an especially embrittling effect on titanium alloys and leads to structural changes (a-accumulation). The main cause for dangerous oxidation conditions in welds is insufficient protective gas cover (middle diagram).In the following, only the welding processes used in engine construction are discussed. There is no discussion of other commonly used processes that are done in air with coated welding electrodes.
  

The most commonly absorbed and damaging gas, at least in steels, is hydrogen. This can come from various sources:

• Moisture from the air resulting from insufficient protective gas cover (middle diagram, Fig. "Fouling of Titanium welds"), condensation water on the parts, or moisture in the protective gas. Even unsuitable gas lines can allow moisture to enter the protective gas (Ref. 16.2.1.3-5). In one reported case (top diagram), a thick synthetic hose was used that acted as a semi-permeable membrane for air moisture. This allowed concentrated amounts of air moisture to enter into the argon used to protect TIG welding. The result was extensive cold cracking in housings made from low-alloy steels.
• Fouling of the welding joint or a filler material, for example, with lubricants, etc. (Ref. 16.2.1.3-18).

Nickel alloys under the influence of CO and N primarily form pores. Increased content of oxidizers such as Al, Ti, and Nb reduces the tendency to form pores, but may promote hot tear formation (Fig. "Strength reducing influences on welds").

Protective gas also influences the seam geometry (bottom diagram, Ref. 16.2.1.3-5) and process parameters, such as the possible welding speed. The solidification conditions (Fig. "Influences at micro-cracking") influence porosity and probably also micro-cracking. Titanium alloys, especially, react extremely sensitively even to minor fouling of the protective gas.

Figure "Fouling of Titanium welds" (Refs. 16.2.1.3-27 and 16.2.1.3-31): Damages resulting from external influences on the welding process are embrittlement, gas pores, and cracking. They can be caused by reactions with gases (oxidation, nitride formation), as well as through solution of gases. Gas absorption can occur in the melted material, and therefore in the weld (especially hydrogen), as well as through reactions with solid and fluid media. Fundamentally, especially with titanium
alloys, every fouling of a part surface in a welding zone presents a potential threat to part safety. In the following, typical influences are discussed without any claim to completeness:

Absorption of gases: These primarily originate in atmospheric air that has entered the protective gas. Insufficient protective gas cover can be prevented through adherence to optimized process parameters (protective gas throughflow, torch spacing, bottom right diagram; Ref. 16.2.1.3-27). The potentially damaging gases in air are oxygen, nitrogen, and hydrogen (from air moisture). Oxygen causes oxidation (embrittlement) and influences the structure (e.g. stabilization of a-structure). Embrittlement can occur both in the weld seam and in the surface of the HAZ in the form of a brittle coating. Embrittlement causes a significant reduction in plastic deformability, characterized by the angle of flexure reached by a specimen before fracture. Increased hardness relative to the uninfluenced base material is an indicator of embrittlement. It is interesting that the absorption of nitrogen causes a considerably greater increase in hardness than the absorption of oxygen does (bottom left diagram for pure titanium).
Even protective gas itself can, in unfavorable conditions (e.g. excessive flow rate), be trapped by the melt and create pores, primarily in the root area.

Surface oxidation present before welding: This can result from a previous heat treatment or an overly hot chipping process. The oxidation can be mechanically removed or etched off (pickling, usually HF-HNO₃ solution). It is surprising that there have been no reports of this fluorine causing stress corrosion cracking due to a reaction layer (Fig. "Chlorine in process baths causing stress corrosion") comparable to that caused by chlorine.

Greases and organic residue include deep drawing greases and cooling lubricants (Fig. "Risks from coolong lubricants"). The carbon in these media leads to carbide formation. The hydrogen from these hydrocarbons dissolves in the melt and subsequently in the weld seam, causing porosity. In this case, there is an increased danger of cracking due to hydrogen (pore cracking, sustained load cracking, Ills. 16.2.1.7-8 and 15.2-20). If these media contain oxygen, there is a risk of oxidation.

Cleaning media containing Cl, such as “Tri” and “Per”: These degreasers must not necessarily be present as liquid residue in order to cause stress corrosion cracking in titanium alloys (Fig. "Chlorine in process baths causing stress corrosion"). They form very thin reaction zones containing Cl at the surface of the part (Volume 1 Chapter 5.4.2.2), which can result in cracking under tension residual stresses during welding and at temperatures above 450 °C.

Fingerprints: The chlorine contained in common salt will cause stress corrosion cracking at temperatures above 450°C and sufficient tensile stress (Fig. "Stress corrosion cracking by process baths and hand sweat").

Tungsten particles: During welding, particles of the tungsten electrode can fall into the melt due to carelessness, or break off when they come into contact with the solidifying melt.

Particles containing iron or nickel: These may be surface rust, grinding dust, smeared metal from grinding, or wear products from wire brushes, steel wool, and equipment. These particles can be dissolved by the melt. Sufficiently high Fe concentrations in the melt can have an embrittling effect (Ref. 16.2.1.3-31). Accumulations of titanium-iron eutectic form in the HAZ, which can initiate micro-cracking. These accumulations are additionally sensitive to corrosion.

Similar effects can be expected in Ni alloys that form low-melting eutectics with titanium.

Residue from machining processes: This is primarily abrasive grain or shot (loading effect, Volume 1, Ill. 5.3.1-7). Especially SiC reacts with Ti and Ni alloys at high temperatures. Al₂O₃ particles can act as hard impurities.

Figure "Recognizing embrittlement of Ti-welds": External indicators (Ref. 16.2.1.3-27) of embrittlement in welded titanium alloys as a result of air contact are an important aid for non-destructive quality assurance.

Discoloration and, in extreme cases, topographic changes are the most important characteristics for decisions. Tarnishing occurs due to very thin oxide layers. The thickness of the oxide layer determines the tarnishing color and is therefore an indicator of the intensity of the oxygen contact. Evaluation of the embrittlement risk based on the tarnishing requires experience and technical knowledge. Even if the part surface is still metallically silver in the weld seam zone, the weld may still have absorbed oxygen. This is the case if the protective gas trail provides sufficient cover behind the melt pool, but the melt pool itself is not sufficiently covered by the protective gas around the torch. In these cases, observation of the electrode during welding is helpful. The electrode is altered through oxidation by air sucked in by the protective gas (injector effect). This occurs in a way that is recognizable for specialists (welders).

The tarnishing makes stripes that are yellow (low oxidation) to blue and matte grey in appearance. This is a clear sign of dangerous oxidation. It is especially dangerous if small sections of the oxide layer have broken off and exposed a rough surface. There may be some confusion if there are matte silvery fields between the lines of color (silvery hiatus), as these may be mistaken for a sign that the part is oxidation free.

Tarnishing outside of the weld seam zone is not necessarily an indicator of poor protective gas cover. On the contrary, yellow and intense blue lines parallel to the weld seam (tram lines) are common in TIG and MIG welding.

The hardness of titanium alloys increases along with oxygen and nitrogen absorption (Fig. "Fouling of Titanium welds"). For this reason, hardness testing can determine whether or not damage has occurred. With the aid of a portable or stationary Vickers hardness tester, it is possible to determine the required hardness patterns at the seam surface, insofar as this is accessible. A hardness increase of more than HV = 50 relative to the hardness of the recrystallized base material (i.e. no hardenings) indicates unallowably poor strength values and is not acceptable. In this case, the seam must be completely removed (Ref. 16.2.1.3-30). A more reliable indicator is the micro-hardness pattern in a metallographic cross-section (Fig. "Damage analysis using metallography"). However, this requires taking a specimen destructively.

Conductivity or eddy current testing: These tests are possible because oxidation reduces the electrical conductivity in the surface area. This allows conclusions to be drawn regarding possible damage.

Structural analysis: Metallographic cross-sections Fig. "Damage analysis using metallography") reveal sufficiently pronounced oxide layers and oxygen-stabilized structural components (a-structure).

Destructive hardness testing: A transverse bending test that determines the attainable bending angle is very revealing. Embrittlement leads to smaller bending angles. However, these tests require suitable flat specimens that can only be obtained from parts in very rare cases. Transverse tension tests are not usable because the fracture usually occurs outside of the weld seam, and no noticeable plastic deformation occurs in the weld seam.

Microscopic inspection of a laboratory fracture (SEM, Fig. "Scanning electron microscopy (SEM)") is capable of detecting embrittlement on typical fracture structures and enables conclusions regarding types and causes based on characteristic properties (such as hydrogen embrittlement).