SCIENCE CHINA Physics, Mechanics & Astronomy. Very-High-Cycle-Fatigue of in-service air-engine blades, compressor and turbine

Transcription

1 SCIENCE CHINA Physics, Mechanics & Astronomy Article January 2014 Vol.57 No.1: Special Topic: Airworthiness and Fatigue doi: /s Very-High-Cycle-Fatigue of in-service air-engine blades, compressor and turbine SHANYAVSKIY A A * State Centre for Flights Safety, Airport Sheremetievo-1, PO Box 54, Moscow Region, Chimkinskiy State, , Russia Received May 6, 2013; accepted November 26, 2013; published online December 10, 2013 In-service Very-High-Cycle-Fatigue (VHCF) regime of compressor vane and turbine rotor blades of the Al-based alloy VD-17 and superalloy GS6K, respectively, was considered. Surface crack origination occurred at the lifetime more than 1500 hours for vanes and after 550 hours for turbine blades. Performed fractographic investigations have shown that subsurface crack origination in vanes took place inspite of corrosion pittings on the blade surface. This material behavior reflected lifetime limit that was reached by the criterion VHCF. In superalloy GS6K subsurface fatigue cracking took place with the appearance of flat facet. This phenomenon was discussed and compared with specimens cracking of the same superalloy but prepared by the powder technology. In turbine blades VHCF regime appeared because of resonance of blades under the influenced gas stream. Both cases of compressor-vanes and turbine blades in-service cracking were discussed with crack growth period and stress equivalent estimations. Recommendations to continue aircrafts airworthiness were made for in-service blades. subsurface, fatigue crack, Al-alloy, superalloy, resonance, quantitative fractography, crack growth period, stress level PACS number(s): b, a, v, Cd Citation: Shanyavskiy A A. Very-High-Cycle-Fatigue of in-service air-engine blades, compressor and turbine. Sci China-Phys Mech Astron, 2014, 57: 19 29, doi: /s Nomenclature a crack length in the depth area area of crack origination zone c crack length by the surface of the semi-ellipticallyshaped crack h 1 spacing of meso-beach-marks 1 (MBM1) HV Vickers hardness N f number of cycles to failure (durability) (N p ) 1 crack growth period calculated based on spacing of beach-marks MBM1 (Δq i range of stress for bifurcation area of S-N curves tensile deformation ratio of biaxial cyclic stresses ( 2 / 1 ) *Corresponding author ( shananta@mailfrom.ru) a stress amplitude ( e ) 1 tensile stress equivalent value U ultimate tensile stress % offset yield strength 1 tensile stress opening the fatigue crack 2 tensile or compressive stress acting in a perpendicular direction to the 1 direction i tensile stress opening the fatigue crack mean stress for bifurcation i-area wi 1 Introduction For aircrafts, the main structural elements subjected to high frequency of cyclic loads are compressor and turbine blades of gas-turbine-engines [1]. Modern aircrafts can have in-service time since they are new, more than hours. Science China Press and Springer-Verlag Berlin Heidelberg 2013 phys.scichina.com link.springer.com

2 20 Shanyavskiy A A Sci China-Phys Mech Astron January (2014) Vol. 57 No. 1 Blades operating time before rejection during repair can be more than hours. Under the frequency, for instance 1 khz, the operating period under cyclic loads will be approximately cycles. It is evident that blades damages take place in the Very-High-Cycle-Fatigue regime. In aircraft gas-turbine-engines (GTE) design, different materials are used for various stages of low- or high-pressure compressor and turbine [1]. Blades of the first and the second vane-stages of engine compressor are often manufactured from Al-based alloys because of low environment temperature and low stage pressure. Rotor blades of lowand high-pressure compressor stages are both manufactured usually from Ti-based alloys. But in the case of temperature being higher than 300 C in the last compressor stages there are high-temperature steels. Turbine blades of different stages are manufactured from superalloys. During the operating period of an engine, growing fatigue cracks of different origins are frequent in turbine and compressor blades. These cracks may arise from the sites of damage dents, tears, corrosion pits, and bends inflicted on an airfoil by an extraneous matter [2]. They can affect the resonant oscillations of thus damaged blade. For a short period, the blade oscillates with a frequency typical of transient operation conditions of the engine. Once a site of highly concentrated stress exists in a blade, it helps faster initiation and propagation of a fatigue crack at the surface. A situation like that can be the case in the different periods of an engine service. In service, the turbine blades typically experience damage with greater permanence than the compressor blades do. Foreign damages are not seen for fatigued turbine blade. In fact, the turbine blades are permanently heated and statically stretched by the centrifugal forces, controlled by the revolution speed of the rotor. Under such conditions, thermally activated creep-fatigue effects become possible within a one-flight loading cycle. Heating of the blade material is a factor of special importance as long as in this case the fracture mechanisms controlling damage and subsequent crack propagation may both vary depending on temperature. Each of the mechanisms controlling the blade durability operates during its own limit period of time. This means that the strength criteria, controlling the failure, will differ in nature, depending on the service-life stage at which a blade fails. Consequently, the service-life dispersion can reveal several maximums (instead of one), each maximum corresponding to a certain dominating fracture mechanism sequentially replacing one another during the service period [1,2]. The blades of compressors and turbines of GTEs normally experience dynamic tensile stresses, caused by rotor revolutions, together with dynamic bending and torsion effects of the gas flow. The blades oscillate with a frequency that varies along the blade to create biaxial stressed states. The blades of different stages differ in their natural oscillation frequencies. The latter changes from several hundred oscillations per second for the first-stage fan blades to several thousand, for the last-stage rotor blades of a compressor or the first turbine stage. Design service goal (DSG) for blades of various stages estimated based on the knowledge about material fatigue limit, w2, usually determined on the smooth specimens have no evidence in fatigue cracking at the durability more than 10 8 cycles. Because of different types of damages by the blade surface in time that can occur for in-service blades, the safety factor in the range of five to ten decreases the stress level, w2, which recommended designer for DSG estimation. Fatigue fracture process in metals is usually taken into consideration, applicably to different scale levels: micro- or nano- (Very-High-Cycle-Fatigue or VHCF-regime), meso- (High-Cycle-Fatigue or HCF-regime), and macro-scale level (Low-Cycle-Fatigue or LCF-regime) [2 4]. Transitions from one to another scale level are strongly expressed in accordance with introduced bifurcation diagram for metals, Figure 1. Figure 1 The bifurcation (a) diagram of metals fatigue has constructed according to the tension diagram [2 4] and (b) schema of probability of metals cracking variation for the bifurcation area ( q 2. Bifurcation areas are specified at transitions from nano-( w1 w2 ) to meso-( w2 w3 ), and macro-( w3 w4 ) scale level of metals fracture.

3 Shanyavskiy A A Sci China-Phys Mech Astron January (2014) Vol. 57 No The bifurcation diagram allows describing fatigued metals behavior based on a uniform methodological principle applicable to systems in which evolution occurs far from the equilibrium position. The synergetics concept allows connections among themselves all of which experimentally demonstrated data on the research of metals fatigue at different scale levels and also explaining the increase and decrease of dispersion of fatigue durability in the process of increase of cyclic stress level at achieving critical stress levels. Such a manner of metal fatigue behavior consideration is lawful, even if in the process of evolution metal undergoes only one unstable condition and consequently has only one bifurcation area between two boundary conditions when it is not loaded (one border) and when it completely fails (the second border). Numerous investigations of the dislocation structure of a metal indicate that the accumulating process of damage in the metal during its cyclic loading appears ordered and self-organized [5,6]. The critical condition of a material, which corresponds to the initiation of the crack and its growth during a loading cycle, may be associated with the level of damage density that will be the same however different are the ways or conditions of the cyclic loading. Based on this idea, one can use a single, synergetical approach [2,5,6] to the analysis of the crack-growth stages. A metal with a growing crack represents an open dynamic system, which is far from equilibrium; the system is exercising a series of sequential transitions from one to another stability state and the continued energy exchange with the environment. Once the system has come to a certain critical condition, the homogeneous equilibrium is not stable any more; therefore, inhomogeneities appear in the system, to be defined as dissipative structures [6]. An open system evolves by passing through the critical states, referred to as the bifurcation points, to, alternatively, a stability condition [2,3,5,6]. As long as the system experiences fluctuations, it cannot avoid instability immediately before a bifurcation point. The newly activated processes of damage accumulation develop or, alternatively, die out, depending on whether the system is capable of the selforganized absorption of energy in the ways that shift the construction element toward a greater stability, i.e., longer life times. The general principle of self-organization is that altering the parameters of the extrinsic (applied) influence (the kind or way of cyclic loading) does not result in the occurrence of the structures hierarchy but rather in switching on the fracture mechanisms operative in the system. Bifurcation diagram (see Figure 1) shows a cascade of transitions from one to another bifurcation area indicated as stress level rang ( q i. There is no fatigue limit for metals but bifurcation area were transition from one to another mechanism of metals fatigue cracking takes place (see Figure 1(b)). In the bifurcation areas fatigue durability distribution is bimodal. It means that at the same stress level of cyclic loads can be seen two possibilities in material manner of crack originates. The probability of each manner of crack origination changes in the direction of stress level increasing. At the left border of the bifurcation one mechanism starts and its probability of appearance is not far from zero. Then, for the stress level increasing the new mechanism probably increases its appearance but previous mechanism dies and its probability of appearance decreases. That is why after the transition from bifurcation area, there is only one new mechanism of material cracking. From the discussed consideration followed the conclusion that the bifurcation area reflects material instability in its behavior. In the range of stress level ( q ) 2 probability of subsurface cracking, that dominated in VHCF-regime for material behavior as In Part Closed System (PCS) [3,7 10], decreased but crack origination at the surface, that dominated in the HCF-regime for Open System (OS), became more evident with probability 100% at the border of bifurcation area. In VHCF regime when duration to failure exceeds 10 9 cycles, behavior of metal appears close to thermodynamic PCS because metals have subsurface cracking that depended on material properties only. Fatigue limit, w2, is only shows itself in a certain range of cyclic loading. With the number of loading cycles grown beyond that range, the fracture events might even occur again in VHCF regime after 10 8 cycles. The tests under gradually lowered stresses brought the material down to a first fatigue limit, w1. Between the two fatigue limits, the fracture events consisted in initiating a subsurface fatigue crack in the structural component. This problem was discussed applicably to the aircraft components [1 3]. It was shown that in-service fatigue crack initiation of titanium and aluminum blades has taken usually place from the surface in area of their lifetime to failure up to 10 9 cycles. It depends on the surface damage intensiveness, which has influence on the crack initiation as a stress raiser. In the case of high frequency for blades of titanium alloy VT3-1 and VT8 fatigue cracking took place in VHCF area that attested itself fatigue crack origination subsurface [11]. Blades of titanium alloy VT3-1 have had fatigue cracking because the contact by the blade-shrouds was lost as a result of intensive wear. Blades of titanium alloy VT8 had crack subsurface initiation of the blade-shrouds because of heterogeneities of the transition area between fused material CT4 and main material VT8. Consequently, blades subsurface crack origination had not systematic behavior due to high level of stresses or had appearance because of blades DSG in the bifurcation area. Below two cases of blades in-service subsurface cracking have considering applicably to compressor vane blades of Al-based alloy VD-17 and turbine blades of GS6K superalloy.

4 22 Shanyavskiy A A Sci China-Phys Mech Astron January (2014) Vol. 57 No. 1 2 Blades fatigue cracking of VD-17 Al-based alloy The statistical analysis of the in-service fatigued vane blades of Al-based alloy VD-17 of the first and second stages low-pressure compressor of aircraft engine D-30 has shown that up to their lifetime to failure hours there was crack initiation at the blade surface from the corrosion pits [1,2]. Blades failure for in-service lifetime more than hours has also taken place because of the corrosion pitting influence on the crack initiation at the surface, Table 1. This conclusion was followed from the results of blades surface investigation. All surfaces had widespread damages because of corrosion pits. In the crack initiation location, the designed stress level is near to 20 MPa and the attested average frequency is near to 2800 Hz that influenced the blades stressing during 60% of the in-service lifetime. At the lifetime, more than hours blades experienced not less than = cycles. That is why the blade in-service fracture takes place in VHCF regime with crack origination at the surface. The phenomenon of corrosion pittings influence on the crack origination in VHCF regime was considered with specimens test in salt water [12], and it was shown that environmental deterioration by the metal surface can be more effective than damage accumulation subsurface in VHCF regime. That is why crack origination at the surface was dominant up to cycles. But it is not clear what kind of cracking will dominant at- or subsurface with number of cycles and more. It was interesting to understand: is it true that the blades have only fatigue crack initiation at the surface with number of cycles approximately because of corrosion pits or, may be, after any in-service lifetime blades with damaged surface have fatigue failure in bifurcation area ( q ) 2 with increasing probability of blades subsurface cracking inspire of intensive surface damages in VHCF regime? This question has gown up for practice because of different ways of in-service cracking prevention for blades. In the case of environment deterioration dominant for blades in-service long time it was need anti-corrosion surface layer improving. In the case of subsurface cracking it needs to be estimated DSG for this type of blades in VHCF regime. To answer the question there were used for investigation two of in-service fatigued blades with lifetime to fatigue cracking and hours. One blade with the in-service lifetime hours have had only crack but the blade with hours has been in-service failed (see Table 1). The in-service failed blade is shown in Figure Fracture surface patterns of blades origin The fractographic analyses, performed on the scanning electron microscope EVO-40 of the Karl Zeiss with resolution better than 3 nm, have shown that after the in-service lifetime more than hours (more than cycles) there was crack initiation in blades subsurface inspite of intensive pitting by the blade surface. Four types of the fracture surface patterns were discovered on the crack initiation subsurface sites: (1) grain boundaries, Figure 3; (2) inter-metallic inclusions, Figures 4(a) and (b); (3) intersected persistent slip bands; (4) quasi-cleavage with the river relief, Figure 4. The origin area of the blade 4 (see Table 1) was performed by the triple grain boundaries, as shown Figure 3. There were discovered persistent slip bands on the grain boundary surface. Part of the grain-surface covered specific layer with amorphous structure. Out of the intergranular area, there was performed fracture surface because of transgranular crack propagation of the same manner that appeared for through crack propagation. Fracture surface patterns area typical for the crack growth with low speed. The discovered features of the origin area for the blade 7 (see Table 1) were similar to the fracture surface patterns that were discovered earlier in the tested specimens from the D16T Al-based alloy [13]. At the fracture surface Table 1 Data base about damages and fatigue cracking of the investigated vane blades of Al-based alloy VD-17 after different in-service lifetime since new (SN) and after the last repair (AR) Date Aircraft number Lifetime SN/AR, hours (flights) Type of damages and type of blade cracking ЛСО /1 7692/1279 Corrosion pits; failure ЛСО911015/1 7858/ ЛСО / МС / (10600)/1629(891) ЛСО /2 9445/ МСО / / СО / (8966)/2455(1135) Corrosion pits, crack ЛСО330316/ /2819 Corrosion pits; failure ЛСО / / ЛСО /3 9662/ ЛСО /3 8089/ ЛСО / (8582)/2696(1306)

5 Shanyavskiy A A Sci China-Phys Mech Astron January (2014) Vol. 57 No Figure 2 Overview of the in-service failed vane blade at hours since it was new of the first compressor stage of the aircraft engine D-30. White arrow pointed out place of the fracture surface origin. Figure 3 View (a) of the main origin area of the failed vane blade, placed subsurface in the depth of 200 m, and (b) the fracture surface features in the area of origin location inside of dashed line. Arrows pointed out crack growth direction. Figure 4 Fracture surface patterns (a), (b) in Back-Scatter-Regime of electron microscope under different magnification of the cracked vane blade of the second compressor stage of the engine D-30 with lifetime Arrow pointed out the origin location has shown inclusion in magnified view (b) inside of the circle. besides discussed origin location there is place of the corrosion pit of about 30 m in the depth (see Figure 4). This depth can be indicated by the scale level shown on the photo. Inspite of the pit influence on the fatigue crack origination as geometrical and chemical stress raiser at the blade surface, there was dominant stress concentration around an inclusion caused crack origination subsurface (see Figure 4(a)). It can be seen by the river relief which have patterns orientation from the inclusion in all directions. There was discovered another place of subsurface crack

6 24 Shanyavskiy A A Sci China-Phys Mech Astron January (2014) Vol. 57 No. 1 origination in not failed blade 7. Subsurface origin area had patterns reflected the multi-sliding process under the long-time in-service material before the crack appearance. There was discovered cascade of the intersected slip bands in origin location. One else area of subsurface crack origination was revealed, also, in the blade 4. There was transgranular material cracking because of quasi-cleavage by the sliding plains in the discovered area, Figure 5. This manner of Al-based alloys fatigue cracking was earlier discussed for different specimen surface state [14]. Focus of the discussed origin area had the same specific layer with amorphous structure that was discovered for the main area of origin. To explain the nature of this specific amorphous layer structure, X-ray analysis was performed. The material composition of inhomogeneity area with amorphous structure (see number 1 in Figure 5) was investigated in the scanning electron microscope with the Inca facility of the Cambridge instruments. In the area 1 with amorphous structure there was discovered oxygen in several times more than in other areas 2 and 3 are not placed far from the origin area 1, Table 2. The aluminum percentage was in two times less in the area 1 than was discovered for areas around origins because of oxygen dominating. It resulted in the oxide film in area 1 which is material defect. That is why there is less value of aluminum in area 1 because of oxygen is dominant element that has been introduced in material during manufacturing procedure. Performed analyses have shown that in different material Figure 5 (Color online) View of the fracture surface patterns of the second origin of the vane blade with in-service time since it was new. Places of the X-ray analyses numbered 1 3. zones, there were small in size and thin casting defects. The defect areas in size were less than threshold value that have been estimated for using non-destructive tests during blades manufacturing procedure. This type of defects in size cannot be revealed now with acting methods of non-destructive tests. The discovered features of the origin areas for investigated blades reflected the material in-service critical state when the limit for lifetime to failure for material reaches. Several of the discovered origin areas could be influenced crack subsurface initiation but only one of them will be the first for the crack start. Therefore, it is fatigue limit for in-service blades in VHCF area. The crack origination from each of the discovered areas of quasi-cleavage has balance in probability. In dependence on the intensiveness of the material stress-state in areas of the blades probable crack origination, the critical in-service lifetime for blades will exceeds in different times but after hours since the blades new. That is why it was need to introduce non destructive periodical inspection of in-service blades to reveal fatigue cracks in time. 2.2 Crack growth period and stress equivalent Nearly normal dispersion pattern is typical of the fatigue lives as of the crack-initiation periods for aviation-structure components in case that they are damaged naturally by routinely applied cyclic loading. Here, we identify a single loading cycle with a complex of loads applied in a period between the starting and shutting down of the engine. Hence, the single loading cycle involves a complex of loads that differ in the deviation amplitudes and average levels. However, being aware of the chance that the crack can grow to the critical size, one should never describe a crack-initiation period in terms of a traditional statistic approach in case that cracking was initiated by an occasional artificial damage in service. Instead, growth period of a fatigue crack may only be used as a flight-safety criterion to continue airworthiness so that the critical state of a cracked blade is never achieved in flight. They measure the growth period of a fatigue crack in blades by the number of flights or of the engine. Each cycle leaves behind the macroscopic or mesoscopic fatigue beach marks (BMs), depending on the nature of damage that the loads applied during the flight produce [1]. A crack only propagates in a blade during a short part of a start-and-stop cycle of the engine. Then it does not advance during the Table 2 Chemical composition of vane material discovered in different areas 1 3 of fracture surface shown in Figure 5 Area C O Mg Al S Ca Mn Fe Cu

7 Shanyavskiy A A Sci China-Phys Mech Astron January (2014) Vol. 57 No flight unless the blade enters the resonance again. This long-term rest of a crack is marked by a step on the fracture surface, which shows itself as macroscopic or mesoscopic BMs, depending on the degree of damage. A BM itself with a plateau between this and the next adjacent BM is characteristic of the damage caused by a one-flight cycle. It should be pointed out that BM were seen on the fracture surface when crack start to grow as surface crack. The blade fracture can exhibit a block of BMs instead of a single BM. The number BMs depend on how many times in a flight the blades experienced resonance. The number and formation patterns of the BMs repeat along the crack path. The BM blocks are self-similar geometrically, which makes it easy to express the crack-growth period in the number of one-flight loading cycles as equal to the number of the fatigue-bms blocks. Doing so there was measured BMs spacing, h, on the fracture surface of in-service fatigued compressor blades that reflected crack increment during one flight and, then, it was reproduced in number of flight-cycles, Figure 6. Crack growth for investigated blades was during several hundred flights. In the blades 4 and 7 cracks propagation took place during 556 and 320 flights respectively. In the blade 8 crack growth was not up to the fast fracture (see Figure 6(b)). The crack was discovered in the blade during repair. That is why crack growth was in the blade during approximately 250 flights, being less than in the failed blades. The stress equivalent was calculated for the first and third types of origins based on the Murakami s equation [7] modified by Bathias for the VHCF area [9]: (1) 1/6 w2 [( LnN f )( HV 120)] / ( area). It was known test result for the investigated material (from the Russian standards) that was discovered on the specimens tested without fatigue fracture under the stress level w2 =155 MPa up to cycles. It was also known from the designer the stress level for blades in the zone of in-service fatigue fracture near 20 MPa, and the number of cycles to in-service blades failure near cycles. In the eq. (1) for blades mean value was HV =130 kgf/mm 2 that discovered from the hardness measuring on the vane surfaces. Therefore, were compared two curves constructed on the basis of the test results (used two points) and calculated by the eq. (1). Calculations were performed for the next origins: 1 main origin for the blade 4; 2, 3 second and third origins of the blade 4; 4 main origin of the blade 8 that was cracked but not failed in service. It turned out that in all cases studied by the eq. (1) the stress level was in near to zero at the lifetime near to cycles. In the case of more longer time than cycles the eq. (1) cannot be useful. Its possible region for application is up to cycles. The estimated stress level by the Bathias s equation was in several times lower for in-service failed blades than discovered for them based on the in-service stress measurement. Constructed S-N curves based on the material VD17 tests and in-service measurements of the stress level in stator blades were summarized, Figure 7. Analyses of the discovered dependences have shown that the better result for the compressor blades lifetime to failure estimation and the stress level calculation at the crack initiation in VHCF regime with the corrected parameters of the Bathias s equation is the next: Figure 6 Beach marks spacing, h, and crack growth period, N p, versus crack growth length, а, in the vane blade of the compressor (a), (b) first and (c) second stage of the engine D-30 for (a), (b) different origins of the failed and (c) the cracked blade. Figure 7 Dependences of the stress value on the lifetime to failure for different origins basis of the Bathias equation and constructed based on the material VD17 tests data. area value of the investigated blades (a) (d) discovered on the

8 26 Shanyavskiy A A Sci China-Phys Mech Astron January (2014) Vol. 57 No. 1 (2) 1/6 w2 [( LnN f )( HV 120)] / ( area). The corrected parameters of eq. (2) were discovered as less values for lifetime which described several equations that were received from description material behavior with using different values of area -parameter for several origins. The performed investigation shows, that for vane blades of VD-17 Al-based alloy after the in-service lifetime hours the material critical state reaches by the criterion of subsurface crack origination. Blades with more longer in-service time than hours can only operate with regular non-destructive inspections to attest their state with interval not longer than 250 flights or 500 hours. 3 Turbine blades of GS6K superalloy of the engine M-601 In service of the engine M-601 there were seen in-flight rear fatigue failures of turbine blades (TB) of superalloy GS6K. The fatigued blades of various engines had flown in the range of [ ] hours at the moment of their failure. The introduced frequency for the blade in-service loading was approx. 4 khz. That is why the lifetime to failure for the fatigued blades took place in the VHCF regime: ( [ ]) = [ ] cycles. Below more detailed analysis of the material cracking was performed for the in-service blades that had flown 1493 hours or 1084 flights since it was new. by the introduced manufacturing procedure for TB in Figure 8. Material average chemical composition has been compared with their local distribution based on the X-ray analyses in scanning electron microscope. Results of the analyses have shown that by the main chemical elements there has not principal difference in their distribution as recommended for investigated superalloy GS6K. Fractographic analyses of the in-service failed TBs have shown that the crack origination in all cases took place subsurface but not far from the blade surface, Figure 9. First flat facet (FFF) is the main fracture surface pattern for all blades independently on the in-service number of flights or operating hours. Cracks have origination in one of the points by the border of the grain or through grain without formation point of origin. Two grains had quasi-brittle cracking with strongly expressed border which divided subsurface and surfacesly TB fatigue cracking. That is why below will be considered more precisely only features for in-service failed TB which had flown 1493 hours. Fracture surface features in the FFF have been compared with subsurface pattern of fracture origin areas of EP741 NP alloy specimens fatigued in the stress transition range from VHCF regime to High-Cycle-Fatigue (HCF) regime 3.1 Facture crack origination TB material GS6K is strengthened primarily by precipitation of a gamma prime phase ( ). Microstructure of all discussed, fatigued in service, has been attested with scanning electron microscope EVO-40 of the Karl Zeiss, and it was shown that in each case studied the variation in shape of -phase, the distribution of grain size, and the mean grain size were in the same range that recommended Figure 8 View of typical shape and distribution of a gamma prime phase ( ) in investigated blade of GS6K superalloy. Figure 9 Areas of fatigue crack origination in TRB after in service time (a) 1493 hours or 1084 flights and (b) 370 hours or 264 flights.

9 Shanyavskiy A A Sci China-Phys Mech Astron January (2014) Vol. 57 No [15]. It is clear that compared features of the FFF are the same in both cases, Figure 10. There shown indication 1, 2, and 3 on the Figure 10 for different cracked grains. The subsurface FFF has strongly expressed border because of transition from one to another mechanism of material cracking. The mode-iii mechanism of twisting under compression is the way of the crack subsurface origination for metals in VHCF regime that was briefly discussed in the paper [16]. The FFF subsurface formation starts because of material cyclical weakness in a local volume under compression with twisting due to residual stresses and, simultaneously, diffusion in the weakened volume rest gases or other chemical elements. Nevertheless, concluding remarks have to be done that TB fatigue cracking takes place in VHCF regime because of specific fracture mechanism which directed to FFF subsurface formation due to the mode-iii mechanism of twisting under compression. 3.2 Crack propagation In VHCF regime of specimens there can be seen short period crack propagation inspite of dominant period of crack origination [9]. The dominating of crack origination period has evidence for full subsurface fatigued specimens. TB subsurface cracking took place in the area of two or three grains only, and, then, through crack propagation has occurred with semi-elliptically shaped crack front. That is why it was interesting to estimate number of flights for cracked TB in VHCF regime but for the stage of surfacesly crack propagation. The well-known pattern as meso-bms (MBM) were performed on the TB fracture surface because of material cracking under mixed Mode I and II due to bending and torsion, Figure 11. There were discovered two types of MBM: (1) MBM1 with small spacing; (2) MBM2 with large spacing which included itself some number of MBM1 with small spacing. The MBM formation process had so regular manner that the MBM1 in a number of 8 12 could be systematically seen as one block for one MBM2 (see Figure 11). The block of MBM2 reflects the blade material reaction on the cyclic loads variation during fatigue crack propagation in one flight [1]. As a result of the measurement MBM1 spacing, h 1, in one direction of the semi-crack-length, c 1, along the TB surface the unified dependence of the spacing, h 1, on number of cycles (N p ) 1 was discovered, Figure 12. To estimate number of flights the value of (N p ) 1 was divided on ten (number of cycles between two neighbored BM). Based on flights number discovered in one and another direction from the origin for semi-elliptical crack the in-service life for propagated crack was estimated in the range of ( ) flights. This period for the investigated TB has not principal difference with the earlier established interval flights for in-service fatigued blades of superalloys which failures took place at 10 9 cycles and more with crack origination at the surface [1,2]. Figure 10 View areas of subsurface FFF in the in-service failed (a), (c) turbine blade and (b), (d) in-test fatigued specimen of EP741 NP superalloy [15] formed respectively after in service time 1493 hours or 1084 flights and cycles with stress level 960 MPa and R=0.05. Numbers 1, 2, 3 indicate cracked grains. Photos (c) and (d) are magnified view of (a) and (b) respectively.

10 28 Shanyavskiy A A Sci China-Phys Mech Astron January (2014) Vol. 57 No. 1 Figure 11 Two types of meso-beach-marks have been indicated as (a) MBM1, with small spacing, and (b) MBM2, with large spacing, incorporated some number of MBM1, difference in spacings due difference in blade loading. The ratio N p /N f between crack growth period, N p, and lifetime to failure, N f, for investigated TB was [160/1084] 100%=14.7%. Consequently, in the VHCF regime there is the ratio N p /N f being approximately the same that in the HCF regime when subsurface cracking has transition to through crack propagation. 3.3 Stress level The TB stress state is biaxial bending-torsion in the section of the blade fatigue failure with principal stress ratio near to = 2 / 1 =0.3 [17]. This value was used to estimate the stress equivalent value on the basis of the data taken from the paper [18]. In the paper, there was introduced diagram for different flaw sizes and the stress equivalent value, ( e ) 1, at different biaxial stress ratios. The FFF-area-size a = 235 μm of investigated TB was used for ( e ) 1 estimation. The discovered value was near to 230 MPa inspite of designed value not more than 48 MPa. For TB evaluating the fatigue limit quantitatively, the area -parameter model proposed by Murakami et al. [3] in the case of subsurface crack origination was used: 1/6 w2 [ km 1.56][( HV 120)] / ( area). (3) The model can be used when area <1000 μm. In the invested case of the blade (see Figure 10) fatigue failure parameters of the eq. (3) are H v =430 kgf/mm 2 and area = 235 μm. That is why calculated value of the equivalent stress is near to w2 [1.56 ( )] /( 235) 1/ МPа. The same calculation of the stress equivalent by the eq. (3) was performed for the other two blades ( 2, 3) fatigued in service at 1669 и 370 hours had having other sizes of FFF- area. It was shown that the stress value for blades 2 and 3 is 317 MPa and 345 MPa respectively. Consequently, the mean value of the stress equivalent for the three fatigued blades was 320 MPa. The superalloy GC6K has the fatigue limit in High- Cycle-Fatigue regime near to 250 MPa at the number of cycles This value is not far from the calculated value if the factor k m =1 will be used. In this case mean value of calculated stress by the eq. (3) will be near to 212 MPa. The designed stress equivalent for the blade was near to 48 MPa under the cyclic loads with the introduced frequency 4.42 khz. This value is near to 7 times less than the value calculated by the eq. (3). To understand the contradiction between estimations of the stress equivalent value and designed stress value the fatigue diagram of the GS6K superalloy was analyzed for simulating the fatigue resistance of this alloy in the area of lifetime to failure more than 10 8 cycles, Figure 13. The new fatigue diagram was constructed based on known tests data of the smooth and notched specimens (points in Figure 13) at the environment temperature 950 C for GS6K superalloy. Specimens fracture was at the surface. That is why to construct fatigue curve for the VHCF range where the subsurface cracking takes place, the data for five failed specimens in the area of High-Cycle-Fatigue were approximated and the discovered line was extended (dark line in Figure 13) in the range of number of cycles near to The extended fatigue curve was shifted in the right hand (dashed line in Figure 13) in area of lifetime to failure in 10 times more based on well-known tests-data for VHCF taken from [8]. In the range of VHCF for subsurface cracking lifetime has one order more than for cracked metals when conventional fatigue limit determines. At the moment blades cracking they had approximately cycles since new. This number of cycles was used for Figure 12 MBM1 spacing h 1 and number of cycles (N p ) 1 versus crack length c 1 in the direction of crack propagation from origin on the maximum distance to the fast fracture.

11 Shanyavskiy A A Sci China-Phys Mech Astron January (2014) Vol. 57 No condition with inspection interval not larger than 250 flights or 350 hours. (4) Applicably to turbine-blades it was recommended to change their design to exclude in-service possible resonance with high level of frequency or to introduce their in-service periodic inspection since new with interval not more than 120 flights to exclude blades in-flight failure. Figure 13 The fatigue lifetime to failure, N f, against the cyclic stress,. Points show the test results for smooth and notched specimens of the superalloy GS6K fatigued at 950 C. The dark line shows the test results approximation with their extension to the VHCF area. The dashed line shows the corrected test results approximation with shifting S-N curve in the rite hand on the value of ΔN f = 10N f based on of the test results have taken from the review [6]. stress level estimation. The discovered stress equivalent value at cycles, estimated by the dashed line in Figure 13, was near to 180 MPa. Therefore, the performed calculation by the three discussed methods has shown that the stress equivalent for the in-service fatigued blades was in the range of ( ) MPa being in several times more than it was measured for external cyclic loads frequency khz. The resonance frequency for TB was estimated near to 4.42 Hz. Both frequency were estimated in test and measured. Measurements method has accuracy not less than 3%. Therefore, for blades maximum of influenced frequency has [4200+( )] = 4326 Hz and the minimum resonance frequency [4420 ( )]= 4288 Hz. Evidently that small part of the in-service TB can have experience the resonance regime of cyclic loading which can be in (5 10) times more intensive in stress level than with designed value of 48 MPa. Appearance of the resonance cases in service is very rare situation because it appearance has low probability for TB. That is why in-service fatigue failures of TB were very rare events. 4 Conclusion (1) Subsurface fatigue cracking in discussed blades was as a result of reaching them critical state in VHCF regime. The performed investigation shows that the material critical state exceeds for vane blades of investigated Al-based alloy after the in-service lifetime hours in the designed operating condition. The in-service fatigue failures of turbine blades manufactured of GS6K superalloy took place because of blades resonance. (2) Inspite of subsurface crack origination in compressor vane and turbine blades there exists in-service long time of material cracking after transition to the stage of through crack growth. (3) Vane-blades of Al-based alloy with more longer in-service time than hours can only safety operated with regular using of non-destructive tests to attest their 1 Shanyavskiy A A. Tolerance fatigue cracking of in-service aircraft structures. In: Synergetics in Engineering Application. Ufa, Russia: Publishing House of Scientific and Technical Literature Monography, Shanyavskiy A A. Fatigue limit Material property as an opened or closed system? Practical view on the aircraft components failures in GCF area. Int J Fatigue, 2006, 28: Shanyavskiy A A. Modeling of metals fatigue cracking. In: Synergetics in Aviation. Ufa, Russia: Publishing House of Scientific and Technical Literature Monograph, Shanyavskiy A A. Bifurcation diagram for in-service fatigued metals. Proc Eng, 2010, 2(1): Shanyavsky A A. Synergetical models of fatigue-surface appearance in metals: The scale levels of self-organization, the rotation effects, and density of fracture energy. In: Frantsiskony, ed. PROBAMAT- 21st Century: Probabilities and Materials. Netherlands: Kluwer Academic Publisher, Ivanova V S. Synergetics, Strength and Fracture of Metallic Materials. Cambridge: Cambridge International Science Publishing, Murakami Y. Metal Fatigue: Effects of Small Defects and Nonmetallic Inclusions. Oxford: Elsevier, Sakai T, Ochi Y. Very high cycle fatigue. In: Proc. Third Intern Conf VHCF-3, September 16 19, 2004, Ritsumeikan University, Kusatsu, Japan, Bathias C, Paris P C. Gigacycle fatigue in mechanical practice. Marcel Dekker, NY, USA, Berger C, Christ H J. Very high cycle fatigue (VHCF 5). In: Proc. Int. Conf. VHCF5, DVM, June 28 30, 2011, Berlin, Germany, Shanyavskiy A A, Potapenko Yu. In-service very-high-cycle-fatigue of titanium compressor blades of aircraft engines. In: Allison J E, Jones J W, Larsen J M, Ritchie R O, eds. Proceeding of Fourth Int. Conf. on Very High Cycle Fatigue (VHCF-4), TMS, USA, Palin-Luk T, Perez-Mora R, Bathias C, et al. Fatigue crack initiation and growth on steel in very high cycle regime with sea corrosion. Eng Fracture Mech, 2010, 77: Shanyavskiy A A, Koltsun Y I. Near-bifurcation fatigue cracking of the surface-hardened and notched metallic rods. In: Yates J R, ed. Proceedings of Int. Conf. Fatigue 2007, Cambridge, UK, (CD) 14 Shanyavskiy A. Mechanisms of the 2024-T351 Al-Alloy fatigue cracking in bifurcation area after laser shocks hardening procedure. Key Eng Mater, 2011, 465: Shanyavskiy A A. Fatigue cracking of smooth and notched specimens of compacted superalloy EP741 NP in high- and very-high-cyclefatigue regime. In: Berger C, Christ H- J, eds. Proceeding of Int. Conf. VHCF5, DVM, June 28 30, 2011, Berlin, Germany Shanyavskiy A, Banov M. The twisting mechanism of subsurface fatigue cracking in Ti 6Al 2Sn 4Zr 2Mo 0.1Si alloy. Eng Fracture Mech, 2010, 77: Birger I A, Shorr B F, Iosilevich U B. Strength calculation for structures. Moscow, Mashinostroenie, McEvily A J, Endo M. A method for the prediction of the influence of flaws on the fatigue strength under biaxial loading. In: Sonsino C M, Zenneer H, Portella P D, eds. Proceedings of Seventh Intern. Conf. Biaxial/Multiaxial Fatigue Fract., June 28 July 1, Berlin, Germany,