Purdue University

Purdue e-Pubs
International Compressor Engineering Conference School of Mechanical Engineering

1978

The Metallography of Impact Fatigue

T. O. Smith

Follow this and additional works at: http://docs.lib.purdue.edu/icec

Smith, T. O., "The Metallography of Impact Fatigue" (1978). International Compressor Engineering Conference. Paper 253. http://docs.lib.purdue.edu/icec/253

This document has been made available through Purdue e-Pubs, a service of the Purdue University Libraries. Please contact epubs@purdue.edu for additional information. Complete proceedings may be acquired in print and on CD-ROM directly from the Ray W. Herrick Laboratories at https://engineering.purdue.edu/ Herrick/Events/orderlit.html

THE METALLOGRAPHY OF IMPACT FATIGUE

Ian o. Smith Senior Lectur er in Physic al Metallu rgy * Univer sity of Queens land, St. Lucia, 4067, Austra lia.

INTRODUCTION Compre ssor valves are subjec t to two causes of failure due to the cyclic loads which they experie nce. The first type of failure is induced by the bending stresse s and represe nts a case of conven tional fatigue failure . As such, measur es for the preven tion of this type of failure are relativ ely well unders tood. The second type of failure involv es the fractur e of small chips from the edge of the valve and is termed impact fatigue failure as it is associa ted with the repeate d impact of the valve agains t its seat. The phenom enon of impact fatigue has not been examine d in detail and very little is known of the micros tructur al proces ses which lead to failure . Svenzo n (1) has develop ed a specia l testing machin e, the s.I.F.T ., which allows impact fatigue failure to be initiat ed under contro lled conditions and fatigue limits to be determ ined for variou s materi als. The impact fatigue streng th of severa l valve steels has been establi shed using this machin e (1). Howeve r, some eXperim ental difficu lty exists in determ ining the intens ity of the impact when materi als of differe nt elastic and streng th proper ties are tested . Fracto graphi c studies of the impact fatigue failure of valve steels (1) have shown that the fractur e initiat es below the surface of the specimen in the region between the valve edge and the contac t area with the seat. The small effect of surface conditi on upon impact fatigue confirm s the probab ility of subsur face crack nuclea tion. The first extern al signs of crackin g were always perpendic ular to the edge of the specim en. The region between these radial cracks then falls away to give the typica l final appeara nce of an impact fatigue failure . Stress analys is (1) of the situati on when a valve impact s a seat square ly shows that the stress levels are very low,of the order of 100 N/mm2. Howeve r, larger shear stresse s are produc ed if the valve impact s the seats obliqu ely. Highspeed cine-ph otograp hy has shown that, due to torsion al and flexura l vibrati ons, almost all impacts are obliqu e.

An attemp t to examin e the micros tructur al change s

in valve steels upon exposu re to impact fatigue (!) failed to reveal any observ able change s. No change in hardne ss of the steels could be detecte d after impact fatigue .

The aim of the presen t investi gation was to examin e in greate r detail the effects of impact fatigue upon the micros tructur e of the materi al and this paper report s the prelim inary results obtaine d to date. EXPERIMENTAL since the conven tional valve steels are extrem ely complex in micros tructur e, a number of simple r materi als were examine d in an attemp t to isolate the effects of impact fatigue . The materi als examin ed were anneale d comme rcial purity alumini um, anneale d SANDVIK 5R60, Nitino l in the anneale d austen itic conditi on and anneale d SANDVIK 12Rl0. sample s of SANDVIK 20C valve steel which had failed by impact fatigue were also examin ed. Transm ission electro n microsc ope sample s were prepare d from the region between the contac t ring and the specime n edge since this is the area of crack initiat ion. Compa rative samples were prepare d from the area of the sample which was clamped during testing . The sample s were examin ed at lOOkV in a JEM-lOOC electro n micros cope. RESULTS By adjustm ent of the intens ity of impact , it was possib le to produc e impact fatigue failure s in all of the materi als examin ed. The difficu lty in establi shing an absolu te measur e of the intens ity of the impact makes it imposs ible to derive any compar ative values of the impact fatigue strengt h of the variou s materi als. Alumin ium Figure 1 shows typica l disloca tion arrange ments in the alumini um sample s, both for the region experiencing impact fatigue and the gripped and hence unstrai ned area. Figure l(a) shows the structu re of the materi al which has not experie nced impact fatigue . The structu re consis ts of well-de velope d subgra ins and two large interm etallic partic les

* This hlork

hlas performed hlhiZe on Zeave at the SteeZ Research Centre Sandvik AB.Sandviken. Shleden.

111

can be seen, The subgrain boundar ies are usually attache d to these particl es. The structu re in the vast majorit y of the impacte d samples was the same as in the unstrain ed regions . However , in

SANDVIK 5R60 Figure 2(a) shows the microst ructure of the unstrained regions of the samples . The structu re is typical of a well anneale d austeni tic steel with a low disloca tion density being present . Most of the areas of samples which had been subjecte d to impact fatigue had a similar structu re to this with no signs of deforma tion. However , in some areas of some samples the very localize d damage depicte d in Figure 2(b) and (c) was found. No positiv e connection between these bands and any element of the microst ructure could be establis hed. The bands are related to the crystall ography of the grains and follow (111) planes, and so change directio n when crossing a grain boundar y. Figure 2(c) shows a possibl e associa tion of bands with a grain boundary particl e. Within the intense deforma tion bands the disloca tion structu res are complex and some recover y appears to have occurre d resultin g in the formatio n of network s as shown in Figure 2(a).

Figv.re 1 (a) An eleatro n miarograph showing the subgrai n struature of the unstrain ed region of the aluminium S0117p le.

Figure 2 (a) An electro n micrograph revealin g the microst ructure of the unstrain ed region of a SANDVIK 5R60 sample.

Figure 1 (b) An eleatro n micrograph showing a region of increas ed disloaa tion density in the viainit y of a seaond phase partial e in the impacte d region of the aluminium s0117ple.
some localize d regions some effects of impact fatigue could be seen as shown in Figure l(b). This shows a greatly increase d density of tangled dislocation s in the vicinity of a large second phase particl e. A compari son of this figure with Figure 1 (a),;reve als a signific antly increase d dislocation density in the impacte d materia l in localised regions . This was especia lly evident in the regions near interme tallic particl es in the samples examine d. Thus, the metallo graphy of the aluminiu m impact fatigue samples indicat es localize d deformation in the region of some large interme tallic particl es. The deforma tion is very inhomQg eneously distribu ted and the great bulk of the materia l appears to have undergo ne little or no deforma tion.

Figure 2(b) An electro n micrograph showing an intense deformation band in the impacte d region of a SANDVIK 5R60 sample.

112

Figure 2 (a) An eleatro n miarograph showing the possibl e assoaia tion of a deforma tion band with a large grain boundary partial e in SANDVIK 5R6 0.
The impact fatigue of SR60 appears to result in very intense localize d deforma tion. No positiv e evidenc e of the associa tion of these intense bands of deforma tion with any element of the microstructu re was obtaine d. Nitinol Nitinol was chosen for the investig ation because of its extreme ly high damping capacit y. Damping capacity had been identifi ed in the investig ation of a range of valve steels as an importa nt propert y determi ning impact fatigue resistan ce (1). It was propose d that materia ls of high damping capacit y suffered less oblique impacts due to the damping of the torsion al and flexura l vibratio ns and so suffered lower stress levels during impact. Unfortuna tely, the difficu lty in establis hing absolute values of the impact intensi ty preclud ed any assessm ent of the relativ e resistan ce of Nitinol to impact fatigue .

Figure 3(b) The corresp onding dark-fi eld electro n miarograph showing the presena e of martens ite in the deformation band in impact- fatigued nitinol .

Figure J(a) An electro n migrograph showing disloca tion arrays associa ted with large seoond-phase particle s_ in impaat- fatigued nitinoZ .
Figure 3(a) and (b) are a bright field-da rk field pair of microgr aphs with the dark field image formed from a martens ite reflecti on showing a localized deforma tion band which appears to have interacted with or origina ted at a second phase particl e. Figure 3(c) shows another example of localize d deforma tion in the form of disloca tion arrays which are typical of austeni te which has reverted from martens ite (2). Again as in the other materia ls examine d, the deforma tion was extreme ly localize d and the majorit y of the sample appeare d to be unaffected by the impact fatigue . SANDVIK l2Rl0 In this materia l most areas of the foils examine d showed no evidenc e of deforma tion and the impact fatigue damage was confined to narrow and isolated bands. Figure 4(a) is a low magnifi cation micrograph of a deforma tion band in impacte d 12Rl0. The crystall ographi c nature of the band can be seen in

Figure J(a} A bright- field eleatro n miarograph showing the association of a deforma tion band with a seaond-phase partial e in impaat- fatigued nitinol .

113

the direct ional change s when bounda ry. Figure s 4(b) and format ion bands in which the formed some of the austen ite

it crosse s a grain (c) show intens e dedeform ation has transto marte nsite.

alumin ium.

Figu:re 4(c) An electr on micrograph showing a deform ation band in impac t-fatig ued SANDVIK 12R10. Figure 4 (a) An electr on micrograph showing a deform ation band crossi ng severa l grains in impact ed SANDVIK 12R10.

Figure 5 (a) An electr on micrograph shahling the typica l structure of SANDVIK 20C valve steel. Figure 4(b) An electr on micrograph showing a deform ation band in impac t-fatig ued SANDVIK 12R10.
of No convin cing eviden ce of a clear associ ation carbid e partic les with the initia tion of deform igin of ation bands could be establ ished as the.or the the bands was never in the viewab le area of foil. SANDVIK 20C of the Figure s 5(a), (b) and (c) are microg raphs The steel. valve 20C ued t-fatig impac struct ure of micro structu re of this mater ial consis ts of e temper ed marten site and relati vely large carbid exity compl This x. comple very is partic les and s of makes it diffic ult to be sure of the effect (c) and S(b) Figure er, Howev e. fatigu impact y in show a higher than averag e disloc ation densit the the vicini ty of carbid e partic les. As such, for ed obtain those unlike microg raphs are not

Figure S(b) An electr on micrograph showing the micro struct ure of impac t-fatig ued SANDVIK 20C.

114

CONCLUSIONS 1. Impact fatigue failure occurs in anneale d commer cial purity aluminiu m, anneale d SANDVIK 5R60,N itinol, and anneale d SANDVIK 12Rl0. Impact fatigue causes the formatio n of highly localise d areas of deforma tion in these materia ls. A possibl e explana tion of the highly localise d deforma tion involve s the interac tion of the stress waves produce d by the impact with discontinu ities in the microst ructure of the materia l.

2.

3.

ACKNOWLEDGEMENT The author wishes to thank Profess or R. Kiesslin g Directo r of Researc h, Develop ment and Quality Control at Sandvik AB for the provisio n of facilities to carry out this work and permiss ion to publish this paper. REFERENCES

Figure 5(c) An electro n micrograph showing increas ed disloca tion density in the vicinit y of large carbide particl es in impacte d SANDVIK 20C.
DISCUSSION The deforma tion which occurs during impact fatigue is of a highly localize d nature. This probabl y explain s why measure ments of hardnes s before and after impact fatigue fail to show any signific ant change. The deforma tion is much more localize d than that found in normal tension- compre ssion fatigue where there is intense disloca tion activity in persist ent slip bands but also deforma tion in the surroun ding matrix materia l (3). The very intense deforma tion could be likened to that observed near growing fatigue cracks in iron and stainless steels ( 4,5). The presenc e of these localize d regions of deforma tion probabl y indicat es that the shear stress operating in these planes greatly exceeds that in the neighbo uring regions . One can specula te that it is an interac tion of the stress waves produce d upon impact with some discont inuity in the materia l which locally causes high shear stresse s to produce the band. There is some evidenc e for the associa tion of the local areas of deforma tion with large particles so these particl es may represe nt such a discontinu ity. The lack of general deforma tion indicates that the average level of stress is low as indicate d by the stress analysi s perform ed by Svenzon (1). It seems likely that a stress analysis of a sample contain ing discont inuities may be necessa ry to establis h the potenti al local stress levels involve d in impact fatigue . Thus, a possibl e explana tion of the origins of impact fatigue failure involve s the initiati on of localise d deforma tion at discont inuities in the matrix by the interac tion of the stress waves with these discont inuities . Such an explana tion appears to fit the experim ental facts known about the phenome non of impact fatigue failure .

1.
2. 3. 4. 5.

M. Svenzon , Doctora l Thesis, Uppsala Univers ity (1976).

J. Perkins , Met. Trans. (1973)27 09. L.M. Brown, Meta Sci. 11 (1977) 315. J. Awatani , K. Katagir i and T. Shirais hi, Met. Trans. 7A (1976) 807.

J. Awatani and T. Shirais hi, Met. Trans. 7A (1976) 1599.

115