Aeroengine Safety

Illustrations 17.3.1-1: This evaluation of rotor disks should make the problems with the use of higher-strength material with correspondingly increased operating loads understanable. Experience has shown that material costs generally increase with hardness. Cost-efficiency requires their use in more highly-stressed parts. This makes the quality demands on the material stricter. The weak point size (allowable flaw size) must especially be reduced. This can be seen in the curves in the diagram regarding fracture-mechanical behavior (bottom right). If the crack growth under cyclical loads (LCF) is assumed to be the most important life span-determining criterion, one can see that crack growth is influenced in several ways by stress increases.

• Crack growth already occurs in smaller flaws above a threshold value (“F”> a_th in top right diagram).
• The crack growth rate increases, i.e. the cyclical life span “L” becomes shorter.
• - The spontaneous final fracture occurs after shorter lifetimes and with shorter critical crack lengths (“R”= a_c ). This could also mean shorter control times, and in extreme cases the parts must be replaced immediately.
  

The top left diagram shows the size of growth-capable flaws (a_th = threshold) depending on dynamic loading to the elastic limit. The characteristic lifespan is 10⁴ load cycles to fracture. In this fracture-mechanically estimated example (corresponding to the top right diagram), weak points must not exceed an equivalent crack length a_th . Therefore, the applied non-destructive testing must find larger flaws with sufficient accuracy. At a stress level of 120 MPa, a_c is roughly 30mm. At 90 MPa the weak point size can be up to 200 mm, i.e. several times greater.

Figure "Probability of detection of non destructive tests" (Refs. 17.3-4 and 17.3-8): The diagrams show trends of the probability of detection (POD) of surface cracks. They permit an initial ordering of the reliability of the most common non-destructive testing methods in series use. These diagrams refer to tests of aluminum specimens with small artificial dynamic fatigue cracks. In general, crack size/length and crack depth will have different effects depending on the procedure. A POD of 100% can evidently be expected only in exceptional cases. This means that cracks in the millimeter range, which are dangerous for the highly stressed parts of modern engines, cannot be completely ruled out through non-destructive testing (top diagram). Only a combination of several measures, such as multiple non-destructive testing methods, stable optimized finishing processes, and production monitoring, can be expected to result in the high safety levels required in engine construction.

The “least reliable” with regard to the POD of small surface cracks in the present case was X-ray testing (bottom left diagram). The best POD with small surface cracks was ultrasonic testing.

Because the depicted references in the technical literature are already a bit older, it can be expected that ultrasonic testing and eddy current testing have undergone improvements in technology and analysis in the interim.The use of computers for analysis should have significantly increased POD in some cases (ultrasonic testing, eddy current testing; Ill. 17.3.1-3). In spite of this, even in these cases parts that have passed tests cannot be assumed to be absolutely free of flaws. Depending on the procedure used, the POD depends on a combination of various factors (Ill.17.3.1-4).These must be taken into consideration when optimizing the process for a specific application.

Figure "Detecting etch cracks in bores" (Refs. 17.3-7 and 17.3-9): LCF cracks from operation were found in bolt bores in steel compressor disks (diagram). They were oriented in a radial direction from the bore edges (detail). The diagram of the POD of various series-suitable non-destructive testing methods with regard to crack length corresponds to the trends in Fig. "Probability of detection of non destructive tests". However, the roughly 20 years of development work have evidently had an effect. Eddy current testing, which is now automated, shows the best POD and is clearly superior to ultrasonic testing in this case. The POD of the magnetic particle inspection and penetrant testing are the worst. That this situation is not simply due to the visual recognizability of the indicators is attested to by the surprisingly good POD of light-optical inspection.

Figure "Influence of grain size ar the detectability of flaws" (Ref. 17.3.1-1): The characteristics of a flaw and the base material (matrix) also determine its detectability. There is no claim to universality in the following.

Flaw type:

• Gas pores (Ills. 15.2-14.1 and 15.2-20), cavities (Ills. 15.1-7, 15.2.1.3-31, and 15.3-7).
• Cracks (Ills. 15.2-1, 15.2-26, and 16.2.1.3-14). * Separations: laps (Fig. "Material separations in forgings"), cold welding (Fig. "Savety relevant flaws of casted turbine blades"), oxide skins (Ills. 15.2-9, forging laps (Fig. "Material separations in forgings"), lack of fusion in welds (Fig. "Fracture of a diffusion welded titanium blade")
• Inclusions: ceramic particles and hard metal particles (tool breakage) from the melting and/or casting processes.
• Segregations from the melting process (Ills. 15.2-11, 15.1-13, and 15.3-12).
• Structural anomalies (Fig. "Flaw causing titanium rotor fracture i") such as “hard a” in titanium alloy

Flaw geometry:

• Flat (separations), smooth, or fissured.
• Sharp-edged (crack, inclusion) or spherically smooth (gas pore).
• Open to the outside or enclosed, e.g. under the influence of compressive stresses.
• Jagged pattern (crack).
• Size/spread of flaw.
• Volume of the flaw, also relative to the cross-section of the part.
• Open or close to the surface.

Flaw location, part geometry and topography (in the part and/or relative to the “testing direction”):

• In corners.
• Depth under the surface.
• At the surface.
• Along the direction of testing (crack under X raying) or across it.
• Hollow part.
• Wall thickness/cross sections
• Accessibility with testing devices.
• Roughness of the surface.

Flaw characteristics:

• Magnetic behavior.
• Electrical conductivity.
• Elastic properties.
• Thermal conductivity.

Properties of the base material (matrix):

• Magnetic behavior.
• Homogeneity.
• Grain size and distribution.
• Grain orientation, texture, “fiber direction”.
• Elastic properties.
• Material-specific weak points.

The top details schematically show the size of ceramic flaws in the structure of a PM material. These flaws were artificially inserted into specimens in order to determine the probability of detection (POD, Fig. "Probability of detection of non destructive tests"). The diagram clearly shows that the POD worsens with increasing grain size. This understanding is of fundamental importance for the safety of highly-stressed parts such as rotor disks. When changing the material and/or raw part production process (e.g. remelting, forging), it must be determined whether the structure has also been altered in a way that worsens the POD. In this case, it must be assumed that there will be larger weak points; i.e. insufficiently detectable flaws, in the part. These may be more capable of growth than the ones assumed earlier. If this is not taken into account in the design, it can dangerously shorten the life span of the part.

Figure "Limiting influences non-destructive testing": The standard series-suitable non-destructive testing methods have individual strengths and weaknesses (Ref. 17.3-4). These are determined by the functioning principle of the specific procedure and very important for selecting a suitable method. They are briefly described in the following:

Visual inspection:
Advantages: This inexpensive and simple procedure can also detect surface flaws on complex contours such as edges. Documentation with photography is possible. Surprisingly good probability of detection (POD) is possible (Ill. 17.3.1-3). Effects such as discoloration or topographical anomalies indicate flaws.

Drawbacks: Highly dependent on the human factor. No automization. Findings can be covered by subsequent changes (working). Of limited use on poorly accessible part zones such as bores and disk surfaces in integral rotors (Fig. "Damaging metal splashes abd sparks").

Liquid Penetrant Inspection, Fluorescent Penetrant Inspection:
Advantages: High POD in good situations (compare Ills. 17.3-2 and 17.2.3-3).Relatively simple and inexpensive process. Tests the entire accessible part surface.

Drawbacks: Only applicable on flaws that are open to the surface. POD is strongly dependent on the surface conditions. If penetrant remains on the surface, it will cause background fluorescence. This masks flaw indications. This situation can occur in case of

• excessive roughness
• porosity
• fouling (e.g. loading with blasting media, Fig. "Shot peening undesirable effects i") or oxide coatings.

Worsened wettability, for example through silicon-based fouling (Ills. 16.2.1.7-3 and 17.3.1-8), casts the presumed POD into question.
If cracks are smeared or pressed shut after they form, the functioning principle of this type of inspection is compromised. Typical examples are grinding cracks (Chapter 16.2.1.1), flaws in chipped surfaces, or shot-peened zones (e.g. closing of porosity in cast parts). If parts have powerful macro-compressive residual stresses (e.g. in prespun forged disks; Volume 3, Ill. 12.6.1-13.1), they can press the crack sides together so much that the testing fluid is not able to penetrate.
If cracks are filled with corrosion products or oxides, detection is difficult or often impossible. This situation may occur, for example, in the case of intergranular corrosion in etching baths.
If shallow and wide cracks are open to the surface, the penetrant may be washed out during rinsing, preventing them from being detected.

Ultrasonic Inspection:
Advantages: With good conditions, this process has a high POD (compare Ills. 17.3.1-2 and 17.3.1-3). It can be used in all homogeneous materials. It can detect internal flaws and also determine their size and orientation. Flaws from foreign material such as material contamination (segregations) are also detectable. The process can be automated (Fig. "Probability of detection of non destructive tests").

Drawbacks: Complex part contours limit its application. In some cases, an “ultrasonic contour” may be necessary before final working. Attachment of the transducer depends on the contour.
Flat flaws at the part surface (in the near zone) are not verifiable. Certain flaw positions within parts require suitable testing configurations. Flaws in unexpected positions may not be detected (“blind spots”, Ills. 15.2-16, 15.2-21, 15.2-22, and 17.3.1-5).

Structural inhomogeneities can decisively compromise the POD due to the scattering of the signal (“grass” in the display). The coarse grain (Fig. "Material operating properties") typical of fine cast parts (e.g. integral turbine disks in small gas turbines) does not permit a POD that would correspond to the high part stresses.
When dealing with extremely high material strengths, even the slightly more coarse-grained forged variant of a part may no longer permit the required POD that was possible in a variant made from very fine-grained powder metallurgical material (Fig. "Testing of high strength material"). This would preclude an seemingly risk-free material change.
Porous and inhomogeneous materials such as thermal sprayed coatings cannot be inspected with ultrasonic testing.

X-Ray Inspection:
Advantages: This process is independent of the condition (roughness, fouling) of typical finishing surfaces. It permits the detection of internal flaws. In addition to gaping flaws (cracks, pores), it can also detect fouling that weakens the X-rays differently than the base material. These include inclusions such as core fragments (Fig. "After-treatment flaws of casted turbine blades") or blasting residue in cooling air bores (Fig. "Abrasive blasting problems"). The procedure can also be used in coarse-grained materials such as cast Ni. It reveals pore- and cavity fields in cast parts (Fig. "Types of hollow material flaws") and welds (Fig. "Fouling of Titanium welds") in an analyzable way. With the required accessibility (X-ray photo), the use of X-ray micro focus makes a 20x magnification possible. This makes it possible to detect very small flaws such as “layered cavities” in blade walls (Ref. 17.3-1, Fig. "After-treatment flaws of casted turbine blades"). This procedure makes it possible to photographically document the findings over a longer period.

Drawbacks: The prerequisite for flaw detectability is sufficiently large absorption differences of the X-rays. This means that the detectable flaw size depends on the weakening of the penetrated cross-section. The narrower the cross-section or wall, the smaller the detectable flaws. The flaw location and shape should allow the X-ray to pass through without hindrance. Cracks oriented across the X-ray direction and/or jagged closed cracks are difficult to detect. Because the process principle requires a photographic film at the X-ray exit, accessibility is a key requirement for a good POD.
Pronounced material inhomogeneities, such as coarse grain in cast parts, can scatter the X-rays so much that the POD is too low for small flaws.

Magnetic Particle Inspection:
Advantages: Simple, fast, and inexpensive procedure for detecting surface cracks in parts made from magnetic materials. Also reveals non-magnetic material fouling near the surface. Cracks beneath a sufficiently thin non-magnetic coating (e.g. galvanic Cr coating) can be detected in magnetic materials (Fig. "Aftertreatment cracks on Cr-coating").

Drawbacks: Flaws on inside edges, such as in spline shafts, are masked by the disruption of the field (Fig. "Cracking in steels processed in burnishing baths"). Cannot be used on insufficiently visible surfaces. Cannot be used with non-magnetic materials or parts with excessively thick non-magnetic coatings. Not suitable for poorly magnetizable part zones such as small bores. Limited documentability of the findings.

Eddy Current Inspection:
Advantages: Highly sensitive, automatable, and continually documentable process with outstanding POD (Ills. 17.3.1-2 and 17.3.1-3). Can be used in electrically conductive materials. Well suited for flaws near the surface. These can be separations, or also small metallic or non-metallic inclusions (Fig. "Dangers of tool-fractures") and inhomogeneities. Well suited for inspecting bores (Ill. 17.3.1-3).

Drawbacks: Reacts to minor surface changes, making signal analysis difficult even for specialists. This means that even allowable deviations in roughness or structural inhomogeneities will be noticed. The process works in a relatively thin surface zone. Flaws located deeper in the part are not detected. The eddy current sensors must be fitted to the part contour and follow it. This may require an extensive and elaborate adaptation of the procedure. Coatings prevent the use of this method for the base material. Coatings themselves must be sufficiently homogeneous and electrically conductive in order to be inspected.

Figure "Correct application of non-destructive testing i": The depicted situation of ultrasonic testing of a forged blank for a turbine disk with an ultrasonic contour caused the disk to fracture in the engine (Fig. "Segregation induced turbine disk failure"). The testing procedure (ultrasonic = US) was not specifically adjusted for the geometry of the flaw as a specialist would expect it to occur. The flaw`s curved shape and location in the annulus (top diagram) were typical for the casting process of the pre-material (creation of the segregation) and the forge forming (Fig. "Flaws in forged rotor disks"). In this case, the US transmitter and receiver were on the same testing head (middle detail). The echo, which was reflected by the flat flaw at an angle, could not reach the receiver. For this reason, the large flaw (segregation; several square centimeters) was not detected. Optimization of the testing procedure (US transmitter and receivers across fom one another) guaranteed the required probability of detection.
Part-specific selection, adaptation, and optimization of a process is an interdisciplinary task that requires a great deal of technical knowledge. The following should especially be considered:

• Material: Structural characteristics such as grain size and inhomogeneity (ultrasonic testing); magnetic behavior; absorption of X-rays; thermal conductivity and capacity (thermography); electric conductivity (eddy current testing); typical expected flaws from raw part/blank production and the finishing processes that were used (e.g. tendency to have grinding cracks).

• Flaw type to be detected: Surface flaws; volume flaws; separations (cracks, laps/forging laps); shrinkage and gas pores; non-metallic inclusions such as segregations and carbides; etc.
  

• Expected flaw location: Surface; inside; in material agglomerations (e.g. disk hub); in thin cross-sections (e.g. the wall of a hollow blade); on outer or inner edges; bores; in part zones that were worked or treated with a specific procedure (e.g. grinding cracks on ground fir tree roots, etc.).
  

• Flaw geometry and size: Even; spherically curved; round, size near the limit of detectability, etc.
  

• Required minimum probability of detection (POD): This depends on the loads on the part zone and the required failure safety.
  

• Part geometry: Accessibility (e.g. visually/photographically, with probes required for testing process); compromising the flaw detectability (e.g. in the grooves of splined shafts); “ultrasonic contours”.
  

• Highly-stressed, life-determining part zones with especially high quality requirements must be known. This includes the maximum allowable number of flaws that is part of the design basis (e.g. the hub area and around the bores of disks and blade roots).
  

• Influences from finishing and working processes: Smearing or hammering shut of pores and cracks (shot peening or oxide blasting); covering flaws with coatings or oxidation; possible necessity of pre-treatment of the part being tested in order to open smeared cracks, etc. (thermal cycles, etching, Fig. "Opening of cracks before penetrant testing").
  

Figure "Design for quality assurance,": The design engineer is especially responsible for the achievable quality of a part. Quality depends on the selection of production processes (e.g. welded construction or integral cast part), which testing methods will be used, and what limits are placed upon their use by the characteristics of the part. Examples include parts that are optically inaccessible or poorly accessible, contours that are not suitable for ultrasonic testing, and/or highly-stressed part zones that are difficult to reach with X-rays. A typical example is the root of a weld seam (housing detail in left diagram, friction welding of a rotor in Fig. "Damaging metal splashes abd sparks"). In case of poor inspectibility, these processes can already represent a given “quality risk”.

Illustrations 17.3.1-7: The principle of fluorescent penetrant testing requires that flaws are sufficiently open to the surface. Some finishing processes such as chipping and blasting (shot peening, abrasive blasting) can press cracks shut or smear over them in a way that their probability of detection is greatly reduced. Foreign materials such as cutting oil, processing baths, and water vapor can prevent sufficient penetration of the testing fluid into the flaws. If the part was heated in oxygen (e.g. in an air atmosphere), oxides can form in cracks and compromise the test findings. Experience has shown that these effects can be minimized or eliminated through pre-treatment of the part before testing. There are three primary procedures for this:

A temperature cycle that heats the Ni alloy in the range of hardening (about 800°C; top diagram). Two effects appear to be primary in opening cracks in this case:

Relaxing residual stresses in burrs, causing them to deform and stand up (top diagram, upper details).

Compression of the burrs due to the thermal stresses of rapid heating. This is due to the thin cross-section and the poor heat removal properties of the burrs (top frame, bottom details).
Another possibility is etching. This can also act in two ways: Burrs pressed together by plastic deformation have compressive residual stresses on the “surface side” after rebounding (Fig. "Straightening difficulties"). The good access for etching media means that the material-removal here is especially strong. This breaks down the compressive stresses, causing the burrs to stand up (deform; Fig. "Typical flaws on galvanic coatinga", Ref. 16.2.1.8-16) and open the crack (bottom frame, upper details).
In addition, crack opening should aid testability through removal of the thin burr cross-sections and the crack edges near the surface (bottom frame, lower details).

It is less common to use mechanical methods to open cracks. It could be used in rotating disks and rings in the form of spinning (bottom frame). This situation has the highest probability for generating sufficiently high stresses to ensure crack opening in all life-relevant part zones. During spinning, penetrant fluid is sprayed onto the part. The rest of the inspection and analysis then occur just as they would normally on the stationary part.

Figure "Difficult crack detection by fouling on baths": The problem is the sudden, epidemic-like appearance of surfaces that have been poorly inspected with penetrant testing.
This “epidemic” can announce itself in various ways. The surface of some parts may be completely or mostly fluorescent green under UV light. This evidently concerns so-called background fluorescence. This prevents cracks from being detected because they also appear green and there is no longer any contrast.

The opposite situation is also possible, in which the penetrant fluid suddenly pearls and runs off the surface as though it had been greased, even though the wetting properties of the fluid are essential to its function. The surface now appears metallically blank in UV light; i.e. dark violet with no flaws. This type of image is suspicious. One cannot be certain whether there might be a crack that the testing fluid was not able to penetrate. Usually, this effect is countered by intensive manual cleaning. Experience has shown that this must be repeated several times. Therefore, it is very time-consuming and costly, as well as being generally unsettling.

Usually, the described effects can be traced back to surface fouling from the finishing process (Fig. "Problematic changing of long used baths"). Therefore, the first task is to identify the media that is causing the problem. All cleaning baths, especially those in the crack-testing process, must be inspected for contaminants. These may have collected and formed a thin film on the bath surface. Every inserted part is coated with this film upon removal (bottom detail), making it unsuitable for testing. In many cases, the contaminants are silicon compounds such as silicon oil. The effect is known from the windshield wipers of passenger cars- in spite of clean wipers, annoying streaks may form.

Unless they are strictly prohibited, silicon oils are often used in production. They are used as defoaming agents in cooling lubricants or as release agents in forms (e.g. for fiber-reinforced plastic parts). The parts carry this silicon oil into the cleaning baths. Although it is washed off as desired, it collects on top of the bath (Ills. 16.2.1.7-2 and 16.2.1.7-3).
The following corrective measures have proven effective:

• No use of silicon oils or substances made from unknown or untested ingredients (e.g. hand cremes, lubricants, defoamers, release agents) in the finishing process!
  
• If parts can no longer be tested for cracks using penetrant fluid, then all cleaning baths must be replaced. The baths ahead of crack testing are the main priority! Bath containers must be cleaned before refilling (fouled edges)!
  
• Not every unusual behavior of a part surface is caused by fouling interfering with crack inspection. If in doubt, the responsible technical department should be consulted. They will have the ability to use specimens to attain some certainty.
  
• Sprays, lotions, soaps, hand cremes, etc. for private use cannot be brought to the workplace. The only care items that can be used here are those that are officially permitted by the responsible technical departments.
  

Figure "Failed detection of large cracks": It often happens that apperently safely detectable cracks and flaws are not found using testing methods that would otherwise seem to guarantee safe detection. Experience has shown that this is usually related to typical situations related to the human factor (Ref. 17.1-8):

Findings outside of the “horizon of expectation” (top left diagram): This influence occurs most often with especially large (albeit very rare) overlooked crack indications. For example, a crack several centimeters in length above the root platform of a compressor rotor blade was overlooked. The Cr-steel blade belonged to an older engine type. In a reproduction of this case using a blade with a dynamic fatigue crack created in the lab, the fluorescent crack indication during the routine magnetic particle inspection was clearly visible to “neutral” observers even from a distance. However, even on this blade, the inspection personnel did not notice the crack.
In another case, (Fig. "Disk cracking during heat treatment"), a 20 cm circumferential crack in the flange of a forged Ni-alloy turbine disk was overlooked during penetrant testing.

The following factors affected both of these cases:

• The crack size was many times greater than expected. Usually, crack indications of about one millimeter are sought. For this reason, the crack was not recognized as such. Confusion with “water spots” are likely in the case of penetrant testing.

• The crack location was in an unexpected area of the part. This situation is likely when the tester does not know or understand the part-specific highly-stressed zones in finishing (e.g. heat treatment, grinding) and/or operation. In other words, this case also involves a certain lack of training.

• The location of the crack is in a poorly inspectible area that itself tends to produce crack indications. Inner edges are a typical example. This includes circumferential cracks in flange hubs (Fig. "Damage by Ni coating lowering fatigue strength") and/or on coating transitions (Ills. 16.2.1.8.3-9 and 16.2.1.8.3-10). Axial cracks (torsion) in longitudinal grooves on splined shafts (Fig. "Damage risks by etching baths" can 17.3.1-4) can also be classed into this category. In these cases, a lack of experience and/or training seems to be a factor.

• Frustration and/or excessively monotonous routines: One must consider the situation of a tester who spends years looking for crack inidications up to the detectability limit (Fig. "Probability of detection of non destructive tests"), but never finds one. Then, after many years, a part with a crack indication several centimeters in length appears before him. In this case, it is understandable that the indication does not fit into the routine search pattern.In this case, might it not be possible to purposely include parts with artificially induced (dynamic fatigue) cracks typically found during operation? This could improve awareness. However, this possibility is usually discarded out of concern that one of these flawed parts could slip through the system.
  

• Poor surface conditions: A typical effect is background fluorescence caused by penetrant fluid that remains on the surface. Usually, the surfaces involved are heavily abrasively blasted surfaces (top right diagram; roughening, loading effect). Pronounced machining grooves (e.g. turning grooves) or porous coatings such as oxide coatings and adhering contaminants present poor conditions for penetrant testing. The causes for these problems should first be sought in the work preparation and a lack of technical expertise on the part of the personnel responsible for the inspections.
  

• Testing methods that are not suitable and/or not properly adapted: The case shown in Fig. "Correct application of non-destructive testing i" is typical (bottom left diagram). It concerns an ultrasonic testing configuration that did not take into account a type of expected flaw. Here, one cause is a lack of a proper overview (Fig. "Correct application of non-destructive testing i") during the conception of the inspection process. For example, the type, size, shape, and location of typical raw part flaws must be known and incoporated into the process.

• Misinterpretation of crack indications (bottom right diagram): These problems can also occur in connection with insufficient technical knowledge and/or the selection of an unsuitable testing process (Fig. "Damage by Ni coating lowering fatigue strength"). Insufficient informing of the testing personnel with regard to the characteristics and backgrounds of the parts being tested could also play an important role. The diagram shows a case (Fig. "Aftertreatment cracks on Cr-coating") in which crack indications in a magnetic particle inspection were mistakenly attributed to an allowable and typical type of crack formation in a non-magnetic galvanic Cr coating. In fact, the cracks were deep cracks in the base material of the magnetic steel toothed gear.
  

Figure "Linear penetrant indications which are no cracks,": Evaluating the indications of an inspection process requires special technical knowledge and experience. The analysis and evaluation of ultrasonic and eddy current indications require an especially high degree of technical knowledge. Automated processes are especially demanding in this case.
The problems are dealt with in the following using penetrant inspection as an example. The first factor is suitable surface conditions. Indications such as water spots, background fluorescence, or wetting problems (Fig. "Difficult crack detection by fouling on baths") indicate that the flaw indication is not optimal and that the probability of detection is insufficient (Fig. "Probability of detection of non destructive tests"). This may necessitate repeating of the testing process. In some cases, one must verify whether the testing process still conforms to regulations or whether it has been unallowably altered.

There are a large number of different flaws that could be behind an flaw indication. The main task of the tester is classifying the flaw indications according to the specifications. These are primarily black-and-white decisions, but evaluation of flaw sizes can be very difficult if indications are disrupted, especially if they are near the threshold value for tolerability. The flaw type can also determine allowability if sizes are comparable. For example, cracks are often evaluated differently than oriented porosity that also has a linear indication.

Experienced testers can draw conclusions regarding the type and depth of flaws on the basis of additional treatment of individual indications. It can be very important for the allowability of a part, for example, whether a flaw is a crack-like separation or just porosity.

Skilled cleaning of the crack indicator with a bit of remover can result in a renewed flowing-out of the penetrant from flaws, changing the intensity of the indication and providing information regarding the depth of cracks or the size of a cavity field under the surface. This method can also be used to differentiate artifacts from real flaw indications.

The intensity, distribution, frequency, pattern, and shape of the indication provide information regarding the type and size of a flaw. In this way, an oriented crack field in a ground surface is characteristic of grinding cracks. The prerequisite for this type of evaluation is that the finishing processes used in different areas of the part are known along with their typical damage types. Even tiny indications in a ground surface are alarming and must be inspected closely, for they may be cracks that have been smeared shut. Therefore, the tester should know typical flaws of the specific finishing processes, as well as their usual appearance. This includes their position relative to the working conditions on the part and the expected indications.

Naturally, this also applies to material-specific indications in raw parts and blanks. They should be known to the tester in their usual appeaance and part-specific location.

If there are any doubts regarding the evaluation of an indication, one should seek the aid of the responsible technical department (e.g. metallography, Fig. "Damage analysis using metallography"). Local polishing and/or microscopically analyzable impressions (e.g. SEM, Fig. "Non destructive microscopic inspection") may yield a satisfactory explanation.

Figure "Importance of specifications quality of testing results": Clear, unmistakable and practical specifications are of decisive importance for sufficiently safe evaluation of indications during testing processes. In this case, penetrant testing is used as an example.
If, for example, specifications mention a specific maximum allowable length of an indication, it leaves room for interpretation of the findings. First, there are questions regarding what constitutes a linear indication. If this is near the limits of detectability, then even minor deviations in the process can change a linear indication into a point-like indication (through run-out, etc.).
On the other hand, linearly positioned casting pores that have been cut open can appear as a linear indication, i.e. like a crack (Fig. "Linear penetrant indications which are no cracks,"). Cracks that have been partially smeared shut (e.g. grinding cracks or cracks that have been blasted) may look like a row of pores, making them appear less dangerous to part strength.

Because the tester must follow the letter of the guidelines, it is important that the composer of the specifications has sufficient technical knowledge and experience. Uncertainty and misinterpretations must be prevented as much as possible through sufficiently precise flaw definitions.

Confusing guidelines are especially likely to occur in translated documents, e.g. in the case of license acquisition.

It is possible to increase the degree of safety by including clear diagrams or photos of typical findings with corresponding assessments, rather than merely providing written descriptions.

Figure "Changing of new parts before operation": It often occurs that cracks are found in new parts before they are installed, for example after long periods of storage and repeated crack testing. The delamination of coatings or unallowable dimensional deviations are other similar findings. Experience has shown that it will generally be assumed that the flaws were overlooked during the tests in the finishing process. This suspicion is not entirely unfounded, as every testing process can be expected to have a certain amount of outgoing quality issues (Fig. "Probability of detection of non destructive tests"). However, this assumption is not always justified, by any means, and can lead to additional damages as well as slowing down the search for causes and solutions.
In these cases, one should always consider the possibility of delayed cracking. This effect, which has several causes, also leads to cracking during storage. Typical causes of delayed cracking include (Fig. "SCC of titanium-alloy in contact with chlorine"):

Hydrogen embrittlement (“1”) can cause cracks after a short time (within hours) or after longer periods following hydrogen absorption (Volume 1, Chapter 5.4.4). Hydrogen absorption can occur in baths (e.g. cadmium plating), through moisture, or gas. Known causes are moisture during fusion welding (Fig. "Dangers of residual stresses") or case hardening in the gaseous phase (toothed gears, Fig. "Hydrogen embrittlement by coatings").

Relaxation (“2”): Unallowable dimensional changes (middle left diagram) are possible as a result of creep (Fig. "Annealing time effect on residual stress level"), which can also occur at room temperature. Examples include deformations in welded constructions and cast parts with complex shapes.

Stress Relief Cracking (“3”): This type of cracking is especially common in welds (see Fig. "Location of welds and finishing cracks" and table 16.2.1.3-1). In steels, it is often related to hydrogen embrittlement. In Ni alloys, relaxation can lead to the growth of hot cracks from the fusion welding process.

Separation of coatings (“4”): This can appear in brittle coatings such as thermal barriers as a lifting up of the corners. This effect should be seen in connection with relaxation. In tough, dense, and thin coatings such as galvanic silver coatings, blistering may occur (Fig. "Shot peening as testing method"). This is probably due to oxygen being released from around the bonding surface. The gas was absorbed during the galvanic process and can be released by the coating and/or the base material.

Dimensional changes through volume changes (“5”) can be traced back to changes of unstable structural states during storage. They are very rarely observed. There are two reasons for this:

• The use of suitable countermeasures such as freezing or annealing hardened steel parts that are deemed to be at risk.
• The very minor volume changes are only noticeable in parts with very tight tolerances, such as regulator pistons (Fig. "Volume changes during heat treatment").

Corrosion (“7”): This includes stress corrosion cracking, pitting corrosion, and intergranular corrosion. These attacks are possible in the case of poor storage conditions. They are promoted or caused by, for example, residue from aggressive media used during the finishing process. This includes flux from brazing that combines with condensation water to even attack corrosion-resistant materials (Fig. "Corrosion of brazed joints").

Crack opening under operating loads (“8”): Typical examples are smeared grinding cracks in the fir tree roots of turbine blades (Fig. "Chipping surface metallographic section") or flaws that were masked by intensive blasting. The cracks open under thermal and mechanical strain (also see Fig. "Opening of cracks before penetrant testing").

Figure "Problems with photographic documentation": The documentation of visual findings, such as shots of topographic anomalies and flaws (e.g. cracks) or UV test indications (e.g. penetrant testing or magnetic particle inspection), can lead to later misinterpretations.

• Shadows on grooves and burrs can look a lot like cracks (left diagram).
• The size of the part and flaw should also be apparent to “outsiders” (right diagram). An object with a widely understood size or a measuring device (ruler, graph paper) are usually sufficient.
• The location of the flaw on the part should be recognizable.
• Problematic color differences should be avoided. Examples include discoloration due to tarnishing and oxidation.

The authenticity of photographic documentation is being increasingly called into question due to the possibilities of computer-aided picture editing in this day and age. Uncertainty is already created if findings are made to stand out from the surrounding area through the use of contrast and/or color changes. This makes their use as evidence problematic.