Experimental attempts for creep and fatigue damage analysis of materials state of the art and new challenges

Transcription

1 Experimental attempts for creep and fatigue damage analysis of materials state of the art and new challenges Zbigniew L. Kowalewski Institute of Fundamental Technological Research, Polish Academy of Sciences Pawińskiego 5B, 0-06 Warsaw, Poland Abstract Development of creep and fatigue damage was vestigated usg destructive and non-destructive methods materials commonly applied power plants or automotive dustry. In order to assess such kd of damage the tests for a range of different materials were terrupted for selected time periods (creep) and number of cycles (fatigue). As destructive methods the standard tension tests were carried out after prestrag. Subsequently, an evolution of the selected tension parameters was taken to account for damage identification. The ultrasonic and magnetic techniques were used as the non-destructive methods for damage evaluation. In the fal step of the experimental programme microscopic observations were performed. Keywords: creep, fatigue, damage, non-destructive methods. Introduction Many testg techniques commonly used for damage assessments have been developed up to nowadays. Among them one can generally distguish destructive and non-destructive methods. Havg the parameters of destructive and nondestructive methods for damage development evaluation it seems to be reasonable their further analysis that should provide possible mutual correlations []. This is because of the fact that typical destructive vestigations, like standard tests, give the macroscopic parameters characterizg the lifetime, stra rate, yield pot, ultimate tensile stress, ductility, etc. without sufficient knowledge concerng microstructural damage development and material microstructure variation. On the other hand, non-destructive methods provide formation about damage at a particular time of the entire workg period of an element, however, without sufficient knowledge about the microstructure and how it varies with time. Therefore, it seems reasonable to plan damage development vestigations the form of terdisciplary tests connectg results achieved usg destructive and non-destructive methods with microscopic observations order to fd mutual correlations between their parameters. This is the ma issue considered this paper. The microscopic observations of microstructural changes carryg out usg traditional scanng electron microscopes are possible only after specimen failure, and additionally they suffer on high cost. The non-destructive methods are much more convenient and that is why they are more frequently used engeerg practice for periodic evaluation of the material degradation. The resolution range of particular non-destructive tests covers the entire scope of structural effects and crystallographic defects [], Fig.. Considerg the detailed conditions and scopes of applyg various methods significantly limits the possibilities to use them and makes it difficult to identify and analyse the fatigue damage evolution the experimental way. This leads to a necessity of the constant improvement of existg non-destructive testg methods and of developg new measurement techniques that would be able to detect and carry out quantitative assessment of the structural failures resultg from the development of processes leadg to material fatigue and deterioration of its mechanical properties. Figure : Dimensions of microstructural forms/defects and nondestructive methods suitable to monitor them []. The conventional non-destructive techniques (based on the measurement of ultrasonic wave velocities for example) are 7

2 sensitive to a material damage the last stage of material life and do not give the answer to the question when a given element should be withdrawn from the exploitation? Therefore, there is the need to develop a new method based on a sgle or combation of non-destructive techniques allowg to estimate the microstructural and mechanical degradation observed at different stages of the exploited material.. Experimental attempts for creep damage analysis.. Multiaxial creep tests The results from uniaxial creep tests are not able to reflect complex material behaviour. Therefore, many efforts are focused on tests carryg out under multiaxial loadg conditions. The well known method of creep rupture data from such tests analysis is through isochronous stress-stra curves obtaable from the standard creep curves [3]. It gives comprehensive graphical representation of material lifetime. The curves of the same time to rupture determed on the basis of experimental programme are compared Fig. with theoretical predictions of the three well known creep rupture hypotheses: (a) the maximum prcipal stress rupture criterion (Eqn ), (b) the Huber-Mises effective stress rupture criterion (Eqn ), (c) the Sdobyrev creep rupture criterion (Eqn 3). For the biaxial stress state conditions, realised the experimental programme, the rupture criteria mentioned above are defed by the followg relations: R max 4 3 R () e ( ) (3) R max e ormalised shear stress & ormalised axial stress Figure : Comparison of the isochronous creep rupture surfaces (t R = 500 [h]) determed for alumium alloy ( - experimental results;, 3, 4 - theoretical predictions usg the maximum prcipal stress criterion; the effective stress criterion; and the Sdobyrev criterion, respectively) As it is clearly seen, the best fit of the alumium alloy data is obtaed usg the effective stress rupture criterion. It has to be () notg however, that the lifetimes predicted by this criterion are still quite far from experimental data... ew concepts of creep analysis To assess damage usg destructive method the specimens after different amounts of prestrag were stretched to failure [, 4]. Afterwards, the selected tension parameters were determed (yield pot, ultimate tensile stress) and their variations were used for identification of damage development. Ultrasonic and magnetic vestigations were selected as the non-destructive methods for damage development evaluation. For the ultrasonic method, the acoustic birefrgence coefficient was used to identify damage development the tested steel. In the case of magnetic method a several damage sensitive parameters were identified, e.g. amplitude of Ub or Ua envelopes reflectg the Barkhausen effect (HBE) or magnetoacoustic emission (MAE) variation, respectively, their tegrals, and coercivity. Havg selected parameters of destructive and nondestructive techniques, possible relationships between them were evaluated. The representative relationships are illustrated Fig. 3a, b, c. As it is seen, except of the specimen prestraed up to 0.5% due to plastic flow, all results are ordered, and as a consequence, they can be well described by adequate functions dependg on the type of prior deformation. Diagram (a) Fig. 3 shows relationship between tegral of the Ub envelope amplitude - Int(Ub) and ultimate tensile stress. The magnetic parameter is normalized to the value captured for the non-deformed specimen. umbers figure denote the level of prior deformation. Figure 3a allows concludg that tegral Int(Ub) of the Barkhausen noise may be used to estimate a level of the ultimate tensile stress of plastically deformed specimens. It has to be noticed however, that the case of the material prestraed due to creep the situation is more complicated, sce a non-unique relationship between Rm and Int(Ub)norm was found. The relations Fig. 3a dicate that the steel after plastic deformation, leadg to higher values of Rm, can be characterised by lower values of magnetic parameters. This is because the prestraed material contas more dislocation tangles that impede doma walls movement. On the other hand the lower values of magnetic parameters can be attributed to the lower magnitudes of Rm for the steel after creep. The results make evident that the MBE tensity varies significantly due to microstructure modification, however, different ways dependg on prior deformation type. For deformation higher than % this tensity decreases after plastic flow and creases after creep. Strongly non-lear character of plots Fig.3a makes impossible direct estimation of mechanical parameters when only sgle magnetic parameter is used. Addressg the issue for practical application of the MBE measurement assessment of mechanical properties for the damaged steel one can conclude that it is possible only then if at least two magnetic parameters will be taken to account. It can be seen Fig. 3a that relative decrease of the Int(Ub) with prestrag denotes plastic deformation while rapid crease of the Int(Ub) associated with prestra crease is observed for early stage of creep damage development. Better correlation was achieved between Rm and coercivity Hc, Fig. 3b. As it is seen, except specimen prestraed up to 0.5% due to plastic flow, all results are ordered, and as a consequence, they can be well described by adequate functions dependg on the type of prior deformation. The ma disadvantage of the relationships between Rm and Hc, is related to the fact that it cannot distguish a type of prior deformation for small prestra magnitudes. 8

3 Similar remarks can be formulated for the relationships between Rm and acoustic birefrgence coefficient Bak, Fig. 3c. (a) (b) (c) R m [MPa] R m [MPa] R m [MPa] % 4,5% 3% 9,3% 7,5% 0,5% 5,5% 5,9% 7,9% 4,6%,85% 0% 3,5% 0,85% 0, 0,4 0,6 0,8,0,,4 0% 0,85%,85% 3,5% 5,9% 4,6% 7,9% Int(U b ) norm [j.w.] 9,3% 9% 7,5% 0,5% 5,5% 3% 4,5% 0,9,0,,,3,4,5,6,7,8 9% 0,5% H c norm [j.w.] 7,5% 5,5% 4,5% 3% 0% 5,9% 7,9% 4,6% 0,85% 3,5%,85% 9,3% -0,006-0,004-0,00 0,000 0,00 B ak [-] Figure 3: Variation of ultimate tensile stress of the X0CrMoVb9- steel versus: (a) tegral of the Ub envelope amplitude; (b) coercivity; (c) birefrgence coefficient (triangles steel after creep; circles steel after plastic flow) The relationships between selected destructive and nondestructive parameters sensitive for damage development show a new feature that may improve damage identification. In order to provide more thorough analysis reflectg physical terpretation of the relationships obtaed further vestigations are necessary. Programmes of such tests should conta advanced microscopic observations usg not only optical techniques, but also SEM and TEM. 3. Experimental attempts for fatigue damage analysis The fatigue of structural materials is particularly important for the development of railway and air transport as well as power engeerg. Fast troducg of new technologies those sectors frequently leads to serious crashes. It is enough to mention two crashes of the first jet-propelled passenger aircraft called Comet, crash of the fast German railway Eschede 998 or two jet crashes oo the first transoceanic airway le Poland 980 (Okęcie Airport Warsaw) and 987 (Kabaty near Warsaw). All those tragic crashes were caused by the material fatigue, the nature of which has not been examed sufficiently. Experimental determation of damage development requires on-le recordg the material responses to the given cyclic loadg durg the whole experiment. Difficulty conductg of this task is strengthened additionally by a lack of the effective fatigue damage measure describg properly effects of material degradation. Failure mechanics is relatively a new field of studies and the reference fatigue damage dicators described scientific works should be treated only as suggested, but not verified and accepted problem solution. Monitorg of the changes mechanical properties takg place under cyclic loads by means of recordg stress and deformations of the measured sample section consecutive cycles was an efficient technique enablg not only to defe the number of cycles required to damage the specimen, but also to defe the fatigue damage dicator and its evolution the fatigue process. Behaviour of materials with the range of high cycle fatigue (HCF), which means the stress amplitude is below the yield limit of a material, can be divided to two basic types terms of mechanisms of damage development [5]. Behaviour of the first group of materials is described by the ratchetg generated by local deformation around voids, non-metallic precipitations and other defects of structure. Behaviour of the second group of materials undergog cyclic loadg is described by cyclic plasticity generated by slips on the level of gras and local slip bands. In both cases the changes of measured deformations are a sum of local deformations the whole volume beg measured. Cyclic loadg the range of HCF causes itiation of various mechanisms of damage development. Initial defects the form of clusions itiate a localisation of damage, and as a consequence, its development consecutive loadg cycles. Their imagg (Figs. 4 and 5) helps not only to defe the damage development mechanism, e.g. the form of debondg or decohesion, but also provides the basis for modellg of their mechanisms development. The metallographic methods are laborious and problematic due to the specific conditions of sample preparation. Usually they are maly used after completion of the loadg process or after its programmed suspension the desired phase. Their significant limitations are due to the laborious procedures of sample preparation and the necessity to observe them after load removal. A few years ago, a unique research station has been constructed, with a servo-hydraulic testg mache equipped with digitally controlled signals of load, displacement, total or plastic stra, which can be used for monotonic stretchg and compressg of specimens the range of ±00 k and for low- 9

4 cycle fatigue tests (LCF) temperature range up to 00 o C. The strength testg mache is connected to the set of scanng electron microscopes for examg the sample surface under monotonic load and microscopic observations with the resolution of m. The second electron microscope enables an examation of the specimen surface under constant load, and microscopic observations with the nano-metric (nm) resolution. The specimen, after fixg and imposg the planned range of fatigue loads, is descended, together with the stress-testg mache frame, to a large vacuum chamber with the capacity of m 3 where the metallographic tests of the entire sample surface are carried out. The device is equipped also with EDX system (Energy Dispersive X-Ray) enablg to carry out local chemical analysis, and with EBSD system (Electron Back Scatter Diffraction) to image the metallographic orientation the tested field. The first research station of such a type was stalled 009 at the Institute for Material Science of the Erlangen-uernberg University Germany. damage dicators, Fig. 6, damage measure can be defed by the followg relationship: a m a m (, ) (4) D Hence, a defition of damage parameter takes the form: ( ) ( ) ( ) m (5) max m where accumulated stra up to the current loadg cycle, (m accumulated stra at the first cycle, (max accumulated total stra calculated for all cycles, which damage measure is cluded the form of equation (4). Figure 6: Illustration of stra damage dicators durg fatigue conditions Figure 4: Microscopic image of failure due to fatigue carried out on X0CrMoVb9- steel Figure 5: Microscopic image of failure due to fatigue carried out on A356-TiB Among many fatigue testg programmes one can distguish two basic directions: (a) vestigations conducted by physicists and metallurgists focusg on tryg to learn the mechanisms governg the process of fatigue; (b) theoretical and experimental vestigations order to create a phenomenological theory to allow quantitative description of the phenomenon. Both of these trends are currently developg parallel. 3.. Fatigue damage evaluation In many cases the process of fatigue damage is controlled by more than one mechanism. The fatigue tests carried out on metallic materials and metal matrix composites have shown that the damage process occurred due to combation of cyclic plasticity and ratchetg mechanisms. Therefore, usg adequate The results published so far [e.g. 6, 7] confirm the correctness of the adopted methodology for damage analysis of the materials after service loads, that takg to account parameters responsible for cyclic plasticity and ratchetg. 3.. Previous and new concepts of fatigue testg In order to assess damage degree due to fatigue of the material the as-received state and after exploitation the Wöhler diagrams may be elaborated that represent the number of cycles required for failure under selected stress amplitude. The results of such approach are illustrated Fig. 7 for the 3HMF steel. As you can see, the Wöhler diagrams dependg on the state of material differ themselves, thus identifyg the fatigue strength reduction due to the applied loadg history. Unfortunately, such method of degradation assessment of the material undergog fatigue suffers on very high cost and additionally it is time consumable. Figure 7: Wöhlers diagram for the 3HMF steel before (0h) and after exploitation (80 000h and h) 0h 80000h 44000h 0

5 Therefore, searches are conducted for new solutions that would provide a better assessment of the fatigue damage development. In order to obta this effect, an adequate damage parameter must be defed on the basis of the measurable dicators of its development. Selected proposals discussed section (3.) have been validated experimentally and the representative results will be presented here. Force controlled high cycle fatigue tests (0 Hz frequency) were carried out on the servo-hydraulic testg mache MTS 858. Durg the tests, se shape symmetric tensioncompression cycles were applied to keep constant stress amplitude equal to 70 MPa and 350 MPa for metal matrix composite (Al/SiC) and X0CrMoVb9- steel, respectively. Tests were performed at ambient temperature. Each cyldrical specimen manufactured from both materials was subjected to cyclic loadg until fracture. A movement of the subsequent hysteresis loops along the stra axis was observed with an creasg number of cycle (Figs. 8 and 9). Simultaneously, a width of the subsequent hysteresis loops became almost unchanged for composite and became greater for the steel. Such behaviour identifies the ratchetg effect. both mechanisms a variation of the damage parameter can be calculated usg equations (4) and (5). In graphical form it is shown Fig.. Lookg at the diagrams Figs. 0 and one can say that the lear damage accumulation rule cannot be applied for both tested materials. Moreover, it can be seen that the rate of damage parameter variation for composite depends on the SiC particle content. It creases with an crease of the SiC particle content at the itial stage of fatigue, Fig. 0. Figure 0: An fluence of SiC content on damage parameter variation (stress amplitude 70MPa) Figure 8: Hysteresis loops for selected cycles the results for Al/SiC Stress [MPa] ,004-0,00 0 0,00 0,004 0,006 0,008 Stra Figure 9: Hysteresis loops for selected cycles the results for X0CrMoVb9- steel Sce ratchetg is the domant mechanism of the composite deformation, the mean stra was taken to account durg a damage parameter calculation the stable growth period. Hence, the damage parameter can be defed usg equation (4) assumg that its first term is neglected. It is worth to noticed that the rate of damage is relatively high at the begng of the period. Afterwards, it becomes slower (Fig. 0). In the case of X0CrMoVb9- steel besides of ratchettg also cyclic plasticity mechanism is responsible (crease of hysteresis loop width) for damage development, Fig. 9. Takg to account Damage parameter D 0,9 0,8 0,7 0,6 0,5 0,4 0,3 Material P9; Specimen 4-; y = 0,099Ln(x) + 0,7097 σ a = 350 MPa 0, D_(ampl+mean)_sum 0, Log. (D_(ampl+mean)_sum) 0 0,0E+00,0E-0 4,0E-0 6,0E-0 8,0E-0,0E+00,E+00 / f Figure : Damage parameter variation for X0CrMoVb9- steel (stress amplitude 350MPa) The examples of fatigue testg may be treated as an alternative method for damage evaluation of materials subjected to cyclic loads. Studies which a variation of the hysteresis loop width and its movement were recorded for cycles under fixed constant stress amplitude have demonstrated that this procedure gives a possibility to assess safe operation period for materials question and there is no need to perform so many experiments, as it is required for the Wöhler diagram determation. The proposed method of assessg fatigue damage evolution makes it possible to: determe damage dicators; defe damage parameter; assess fatigue and stress levels to fd ranges which an accumulation of damage can be described by the lear law. Another very promisg attempts the fatigue damage analysis are related to relatively new techniques such as Electronic Spackle Pattern Interferometry (ESPI) or Digital Image Correlation (DIC). Some prelimary results of the ESPI usage will be presented here. The cast alumum alloy AlSi7MgCu0.5 for automotive enge heads [6] was tested after exploitation cycles.

6 Opportunity for stra/stress components measurement of a specimen usg ESPI is presented Fig.. The specimen was cut from automotive enge head (cast alumum alloy AlSi7MgCu0.5). Such heads were cast accordg to the standard procedure that ensures degassg. The ratio of the average porosity was 6%. Observg the stra and stress distributions Figs. and 3 a serious problem can be noticed, i.e. an averagg of the accuracy of the stra components the entire volume of the geometrically homogeneous specimen. In this group of materials the development of fatigue damage takes place around various drawbacks, maly the form of voids formed manufacturg processes such as castg. Besides of density and distribution of defects the volume of tested specimen, the specimen size, and location of dividual defects are important factors of the fatigue damage itiation and further development. Development of local stra components around defects leads to ratchetg, i.e. the cremental rise of the stra components beg agreement with actg stress direction. This group cludes not only all cast alloys, but also the whole range of modern metal matrix composites. simulate the behaviour of the structure under loadg is based maly on experimental data from tests under uniaxial stress states. However, for such purposes also data comg from complex stress state experiments are required. The testg technique under complex stress states gives full formation about the mechanical properties of structural materials, necessary for parameters determation durg numerical modelg. In this context either ESPI or DIC are very helpful sce monitorg of the phenomena related to the fatigue, as the process itiatg locally, requires the full-field observations of the displacement brought by cyclic loadg. Both these methods enable determation of the displacement distribution on the specimen surface, and thus, also stra and stress concentration spots resultg from the defect. 4. Conclusions The results clearly dicate that selected ultrasonic and magnetic parameters can be good dicators of material degradation and can help to locate the regions where material properties are changed due to prestrag. Creep or fatigue analysis should be based on the terdisciplary tests givg a chance to fd mutual correlations between parameters assessed by classical macroscopic destructive vestigations and parameters comg from the non-destructive experiments. Such relationships should be supported by thorough microscopic tests, givg as a consequence, more complete understandg of the phenomena observed durg damage development. Application of the full-field observation methods (ESPI, DIC) enables localization of the itial spot of the failure resultg from the cyclic stress and monitorg of its development as well. References Figure : Stra component distribution map for z direction on the plane specimen surface (cross section 8 4 mm) under load of. k Figure 3: Transversal distribution of the stress component along z axis Solvg problems related to development of fatigue damage, degradation of the mechanical properties of structural materials, as well as modelg of the mechanical properties necessary to [] Makowska, K., Kowalewski, Z.L., Augustyniak, B., Piotrowski, L., Determation of mechanical properties of P9 steel by means of magnetic Barkhausen emission, J. Theor. Appl. Mech., 5, pp. 8-88, 04. [] Holler, P., Dobmann, G., DT-Techniques for monitorg material degradation, Mat. Res. Soc. Symp. Proc., 4, pp. 05-8, 989. [3] Kowalewski, Z.L., Isochronous creep rupture loci for metals under biaxial stress, J. Str. Anal. Eng. Des., 39, pp , 004. [4] Kowalewski, Z.L., Szelążek, J., Mackiewicz, S., Pietrzak, K., Augustyniak, B., Evaluation of damage development steels subjected to exploitation loadg - destructive and nondestructive techniques, Journal of Multiscale Modelg, Vol., o, pp , 009. [5] Przybyla, C., Salagehed,., Prasanna, R., McDowell, D.L., Micromechanical Modelg of High Cycle Fatigue Processes, ASM/TMS Symp. on Comput. Mat. Design GE Global Research, August 0-, 007. [6] Rutecka, A., Kowalewski, Z.L., Pietrzak, K., Dietrich, L., Rehm, W., Creep and low cycle fatigue vestigations of light alumium alloys for enge cylder heads, Stra An Int. J. Exp. Mech., 47, pp , 0. [7] Socha, G., Experimental vestigation of fatigue cracks nucleation, growth and coalescence structural steel, Int. J. Fatigue, 5, pp , 003.