Aeroengine Safety

Figure "Representative samples for part quality": The selection of test samples for determining operation-relevant material characteristic properties for engien part design can be more difficult than it might seem at first. The problems with this can be illustrated with the aid of an example of lost-wax cast parts:

With many engine parts, the removal of a sample suitable for installing mounting devices (bolts, etc.) is not possible simply due to the lack of sufficiently thick cross-sections (blades, for example). For this reason, samples for tensile tests are often cast with the same metal flow as the blades. This has the advantage that the samples are made from the same casting batch as the blades. These samples make it possible to properly affix the loads, ensure even temperature distribution, and have a defined test profile. The decisive disadvantage is that the structure in the tested cross-section of the sample may be considerably different from the life span-determining cross-section of the actual engine part. Possible differences include: grain size, grain/grain boundary orientation, cavities, pores, cracks, and deformations, as well as reactions with the casting form and/or the core. This is due to the cross-section differences, the temperature gradients during cooling, and in some cases stresses that are created in cast parts by restricted thermal strain inside the part, and between the part, core, and casting form.

Tests that make use of a representative original part are optimal with regard to structural peculiarities and production influences. This primarily refers to turbine rotor blades with shrouds. In some cases, the shroud and root can be used as fasteners for simulating centrifugal forces (middle diagram). Unfortunately, the life span-determining blade cross-section in the test is often not the same as that during operation. This may be due to the fact that during a test with constant temperatures along the length of the blade, it may not be possible to achieve the same fracture location as during operation. Therefore, the temperature distribution during the test must ensure that the fracture occurs at the desired location. This means that the fastened zones must be intensively cooled, especially the relatively thin cross-section near the shroud. This has often proven to be difficult. An additional problem is applying the test force in a way that prevents any overlapping flexural stresses from occurring. Problems such as twisting of the blade and the location of the center of gravity of the blade cross-section must be taken into account.

Engine parts with sufficiently thick cross-sections (e.g. integral turbine disks, right diagram) allow the removal of integral samples. Of course, samples should also be removed from the life span-determining cross-sections in order to ensure that the test results are meaningful. In the case of LCF loads on turbine disks, this usually applies when the sample can be taken from the center of the hub without destroying the part. Other suitable areas are cast-on hub cross-sections on the side where they were cast on. This ensures that any flaws and weakpoints such as cavities, coarse grains, and hot cracks are included, as these tend to concentrate in the thick cross-sections that are highly LCF-stressed by a combination of centrifugal force and thermal strain. In this way, the results from these samples should be “on the safe side”. One disadvantage of integral samples is the elaborate sample removal and preparation process. Poor sample results do not necessarily mean that the whole part is bad. This makes difficult decisions regarding the evaluation of these expensive parts impossible to avoid.

Creep data should be taken from the creep life-determining blade region with the typical structural formation (also see Fig. "Grain size influencing static and dynamic fatigue"). However, the cross-sections in this area are usually too thin to provide samples.

Figure "Regenerating creep-damaged parts": In creep-stressed engine parts, the extent of damage, i.e. the used and residual life spans, can be estimated if certain conditions are met. This is a prerequisite for possible rejuvenation.

Estimating residual life: In order to conduct a “theoretical” estimation of the residual life span with the aid of the accumulated damage, the past load spectrum must be known and one must have sufficient experience with the part strength (see Fig. "Lifespan verification by cyclic spin test "). Unfortunately, these important data are usually missing.
A reference value for the creep stress that actually occurred during operation can be a dimensional change due to creep. Possibilities include length changes in rotor blades or increases in the diameters of disks and rings. However, this seems simpler than it actually is. Often, only a small, maximally creep-stressed part zone contributes to the bulk of the creep strain (Figs. "Life span decrease by overtemperature" and "Pore formation as creep damage"). In many cases, residual stresses build up, e.g. between the blade edges and supporting cooler cross-section (Ill. 12.1-14). This can lead to deformation, but with no analyzable lengthening. One possibility may be a 3D-deformation measurement with a laser-supported scanning process. However, determining the creep deformation may require a documented measurement of the new part.
In any case, it can be concluded that a great deal of experience with the part being analyzed is necessary in order to use creep strain to draw sufficiently accurate conclusions regarding the residual life for possible reinstallation.
Another possibility is determining the creep porosity (also see Fig. "Scatter of creep life span data") in the life-determining engine part zone. In special cases, especially with uncooled turbine rotor blades, this can be accomplished non-destructively through the inspection of relevant surface areas of every single part. In order to accomplish this, the most heavily creep-stressed area must be at the surface. The success of a residual life estimation on the basis of a creep-porosity evaluation depends primarily on proper sample extraction (top left diagram). Naturally, this must be done in the relevant cross-section.

For creep pore analysis, properly polished and/or etched sections/surfaces are microscopically (metallographic and/or SEM) analyzed, which demands a great deal of experience. Experience has shown that creep fractures can be prevented with a sufficient degree of safety. The most heavily creep-stressed zones of cooled turbine blades are often on the inside. Near cooling air bores, especially, centrifugal forces combine with tensile thermal stress. Therefore, creep pore analysis must be destructive. Understandably, special preparation methods such as repeated polishing and etching are necessary to prevent pores from being smeared shut or confused with holes left by carbides that have broken out. Of course, destructive analysis is only meaningful with an acceptably small amount of engine parts. However, this amount must be large enough to ensure the required accurate statistical conclusions, i.e. it must be representative of the rest of the batch. Because creep damage and, therefore, creep pore formation scatter widely (Fig. "Scatter of creep life span data") within a set of blading (the same stage of the same engine), even if there is an available standard for the evaluation of porosity, life span estimates can also be expected to scatter widely. This means that safety demands that only a part of the potential residual life span can be used if the part is reinstalled. Estimation of the reusability of run blades has been shown to be sufficiently safe in the case of forged materials. These materials form well-analyzable creep pores (Fig. "Creep mechanisms in hot parts"). With forged blades, the procedure is limited to older engine types and, in some cases, has proven itself over the course of years. Rejuvenation also seems possible in cast materials (e.g. 713C) as long as they form creep pores, according to Ref. 12.5-9.

Another method of determining residual life is a creep test on a damaged engine part. As simple and safe as this procedure seems at first glance, it is rarely meaningful, and rarely are the results transferable to other parts. There are several reasons for this:

• The strength test must stress the area damaged in operation enough for failure to occur here. However, the stress distribution may make it impossible to, for example, use a tensile test to test the part zone that was stressed by centrifugal force during operation.

• The damaging loads must correspond as closely as possible to the operating loads in all parameters. If the selected loads are too high in order to conduct acceptably short tests, the activated creep mechanism will often differ from the one that occurred during operation (Figs. "Creep mechanisms in hot parts" and "Verification test of residual life spans"). This means that the operating damage may not show any life span reduction, even though there might be no usable residual life remaining.

In some cases, structural changes in the base material (orientation and g'-phase) and coatings (Fig. "Metallographic findings of thermally stressed parts") can reveal creep damage. However, these are usually rather effects of the temperature influence and have limited relevance for conclusions regarding the creep stress.

Rejuvenation of creep-damaged engine parts:

A special problem is the selection of rejuvenatable blades and verification of the rejuvenation effect. If the creep damage is merely limited to reversible structural changes, e.g. in the g'-phase, a rejuvenating effect can be achieved through heat treatment. However, if grain boundaries are already damaged by creep pores and micro-cracking (Fig. "Pore formation as creep damage") an HIP treatment is required (Hot Isostatic Pressing; top right diagram). In an argon-filled autoclave, damages that are not oxidized and are not open to the part surface, such as pores and microcracks, will close. This “fusing” process requires gas pressure greater than 1000 bar and temperatures over 1000°C. The HIP process is frequently followed by a heat treatment, if the required rapid cooling cannot be achieved in the HIP press.
Because rejuvenation through HIP is only possible as long as the argon cannot enter into the damage spaces, it is necessary to inspect the parts to be rejuvenated for any flaws that are connected to the surface. This is possible on the ouside of the part with the aid of penetrant testing. However, if there are cracks in the cooling air bores (not uncommon, Fig. "Damaging colder part zones by thermal fatigue"), then the rejuvenation effect will be severely limited, at best. One possible solution may be de-oxidizing annealing followed by sufficiently thick coating before the HIP process.
Rejuvenated parts can initially tend to unexpectedly large creep strain due to renewed and pronounced primary creep (Fig. "Life limiting 'creep fracture'").