Residual Compressive Stress Strengthens Brittle Materials

Effect of retained residual compressive stress on the ductility of the still hot and ductile subsurface the surface layer of brittle materials in reducing applied surface glass. As cooling continues, the sub­ tensile stress. Spring test data demonstrate the beneficial effect surface glass contracts against the rela­ tively cool and rigid surfaces, resulting of residual compressive stress in· the surface of hard steel. in residual tensile stress in the core and corresponding resiqual compres­ sive stress in the surface.' JOHN 0. ALMEN with glass specimens. Among the · ad­ The relative static strength of a Research Consultant vantages of glass for evaluating resid­ specimen of normal glass, which was ual stresses is that at ordinary tem­ carefully annealed to avoid residual UNDER CONDITIONS of br:ttlc fracture, peratures glass is completely brittle. stresses, and that of a similar glass specimen surfaces are weaker than For this reason, the residual stress pat­ specimen_that had been quenched on sound subsurface material. When such tern remains unaltered by plastic flow both surfaces as described in the pre­ relatively weak surfaces are residually to the instant of fracture. Also, any c:!ding paragraph is shown in Fig. 1. stressed in compression, the specimens apparent unorthodox behavior of glass As shown by the insert in Fig. 1, these will support greater tensile loads and will not mistakenly be attributed to specimens were ordinary -! in. plate will exhibit greater ductility because cold work or altered structure. glass, 3 in. wide and 18 in. long. Each the applied surface tensile stress is re­ Residual compressive stress in glass was placed on supports spaced 12 in. duced by an amount equal to the re­ surfaces is developed by rapid cooling, apart, then downward acting loads tain~d residual compressive stress. usually by air jets, from some elevated were applied at the mid-points. It is The strengthening effect of residual temperature at which glass is plastic. seen that the annealed glass failed at compressive stress in brittle m~ terials The surface of the glass in cooling 7,000 psi calculated stress, and that is perhaps most easily demonstrated contracts, thereby plastically deforming the residually stressed glass did not

Product Engineering- July, 1953 189 fail until a calculated stress of 32,000 process has been commercially ap­ because surface vulnerability increases psi was reached. plied for upward of a century. Early as ductility decreases. The beneficial Residual stress in quenched glass applications were the strengthening of effect of residual compressive stress in has been measured photoelastically by wine glasses and similar glassware. the surface of hard steel is shown in Littleton (Ref. 3) . In Fig. 2 arc During the kerosene age, extra strong Fig, 3. In these charts, the bars show shown the results of these measure­ prestressed lamp chimneys commanded the relative fatigue durability of three ments as applied to the specimens of premium prices. Today almost every groups of coil springs that differ only Fig. !, , The surface residual compres­ automobile uses quenched glass. Also, in hardness and prestressing treat­ sive stress of the quenched specimen extensive use of such glass is seen in ments. is seen to approximate 25,000 psi, In the numerous unframed glass doors The lower chart shows a summary beam loading, the surfaces of the used in stores and other commercial of the fatigue durability of thirty-two quenched specimen were nominally buildings, production springs in terms of their stressed to 32,000 psi. Since the ac­ Prestressed steel has also seen long average life, which was 15 7 ,000 cycles, tual or resultant stress is the algebraic and varied use in many machine parts when subjected to repeated compres­ sum of the residual stress and the that require extremely hard steel to sive loads. The chart also shows their nominal stress, the resultant stress resist wear, but in which brittleness life variability, which ranged from a ranges from must be avoided because the parts are r.iinimum of 45,000 cycles to a maxi­ also highly stressed, (Ref. 4) . Such 32,000 + 25,000 = 5'7 ,000 psi mum of 355,000 cycles. demands have been and continue to be compression, as shown on the upper These springs had been heat-treated met by rrocessing to obtain hard sur­ surface of the diagram Fig. 2, to to Rockwell C52-53 hardness followed faces and ''tough" cores, that is, by by preset, that is, the springs were 32,000 - 25,000 7,000 psi = case hardening. The hard case, whether compressed to solid height at 575 F. tension, as shown on the lower surface obtained by carburizing, nitriding, or This operation caused plastic yielding of the diagram Fig. 2, with a maxi­ cyaniding, would be just as brittle as in the most highly stressed regions, mum subsurface tensile stress of 21,- through hardened steel of the same thereby reducing the maximum stress 000 psi. hardness except for the fact that, when under subsequent loads. They were Thus it is seen that the quenched properly applied, the hard cases result­ then shot peened at an intensity of glass specimen failed at the same re­ ing from these processes are, like 0.017 A2 and again preset, this time sultant tension surface stress as the quenched glass, residually stressed in at room temperature, annealed glass specimen, but the compression. Incidentally, the core in To measure the effect of high hard­ former supported a load four times such specimens is "tough" only be­ as great as the latter. This increase in cause, being submerged, it is not af, ness on fatigue durability, four springs load carrying capacity was possible be­ flicted with surface weakness. commercially identical with the thirty· cause brittle materials fail only by Shot peening, which produces resid­ two production springs were prepared tension and subsurface tensik- strength ual compressive stress in the peened as indicated in the upper bar charts is greater than surface tensile strength. surfaces, is more effective in reducing of Fig, 3. These experimental springs were quenched from the same tempera­ The advantages of prestressed glass brittle fractures when applied to hard have been known and the quenching brittle steel than to soft ductile steel ture and in the same medium as the production springs but the tempering treatment was omitted. Their hardness was Rockwell C61-63. Two of the four springs were then shot peened at ~zt~mal food an intensity of 0.026A2. This intensity -,- - ·ienaion Compression - was higher than that used in peening the production springs because the hardness of the metal to be peened was , I 'Resultant strass higher, and the peening treatment pre­ (I/Jllnt:hllr/_ 9/DSS ceded presetting because of the possi­ bility of damaging (Ref. 5) the hard ' 'Calr:u/11~. . oxt.I 1tr~ I _ unpeened steel by presetting. To avoid -quMich•d .glass loss of hardness, the four springs were finally preset at 300 F instead of 575 F used on the production springs. When fatigue tested under the same loads that were applied to the pro­ duction springs, the non-peened hard springs failed in less than 10,000 cyles, whereas the minimum lift of ;::;.. :':-:; ?' ""\f: ,, Y· .,: , ,Stren_, th~~•'o~d • "IO <· ,, the shot peened hard springs exceeded

.~ ~~;. •. J .:.· •• ~ . • • -· ,, • • •• • •• • :. : ~~-·- ~· the maximum life of the production ,~·· ··.,, ,,·:.i,. _ Fig.' .2-Reconstructed -diagrams .to show-stress d"trlbution - springs. M:ore significant, because of r.:,, - _ ·_ in .glass ,beam specimens when '1hey •' were. Ioaded to failure. .,~ ... -~.J' • ,.,,. ... # . . '!-·'. ~ ;' •, • ·• ·.; .~ ( the small number of hard specimens tested, is their greatt'r maximum life:.

190 Prudud Engim:cri11g - July, 19:;, than 800 F, induction hardening, and 2 EACH HARD EXP. SPRINGS flame hardenil'lg. Among additional ~ ~ n n~ mechanical processes are superficial -..---.----.,....-) rolling, honing, tumbling, several ma­ I,., I~ LL , ., I) ;,-i _____....__....___, chining operations, and certain cold 0 "' "' 0 c: 5 i-' ______.L...1, __ __. __ ___. __ ___.., forming and straightening production 2 ~ . N olzo z "' 0 "' 1----.,...... ---,----,----,----,----r---.,.. operations. Processes that increase brittleness by 32 PRODUCTION SPRINGS Minimum inducing residual tensile stress in steel LL,...... _ ...... _, .. ~~I::;~ Average surfaces are welding, flame cutting, ., N O I"" -10 >- "' z "' C! .... ------I grinding, and probably the opposite Maximum of stretching, that is, axial compres­ ' i ! 0 ! !- ______,__. !--)-..-! -r-r-1- 1 200,000 400POO 600,000 sive loading sufficient to cause per­ I .!' I i :,,. ! I manent deformation. Most of the cur­ .,"' I.;c I- I=\ ~ I r rently used cold straightening and cold ~ w ~ : ! ~ I 1' forming processes reduce brittle strength and ductility of steel because ~ !ti1 mm______-~-- -- ____ _ they induce residual tensile stress in ,,_)~:"._/" ._.,,,~~· _;(,?"~ ~:' . :. . ·,). . . ~ ' ·~: th·" the surface metal. :.·i. 'F::g . •~Relative fatigue dural;,ility of grou.ps of coil sp_rings · Because of the enhanced ductility that differ only in hardness and .:,.; prestressing ' treatments. "•' and strength of potentially brittle -: .;."' . . ' ' ,..,, . . -~-· .. ·: metals processed to develop surface residual compressive stress, such treat­ ments should be applied to all hard This increase in fatigue durability sion, Eaton Manufacturing Company, machine parts, especially when used cannot be attributed to work harden­ as part of a cooperative research pro­ at low temperatures. Since the damag­ ing because the hardness of very hard gram that included the l<.esearch Lab­ ing effects of surface residual tensile steel is not increased by shot peening. oratories Division, General Motors stress are increased at low tempera­ The greater intrinsic strength of the Corporation. tures, processes that induce such hard.er steel could not be realized in Similar effects may be expected stresses should be avoided. the non-peened springs because of sur­ from any other process, including BIBLIOGRAPHY face weakness. But when the surface stretching, that protects the surface 1. J. T. Norton, "X-Ray Methods in tensile stress wa, reduced by the resid­ with a layer of compressively stressed the Field of Residual Stress", Experimen­ u.al compressive stress that was in­ metal. Conversely, processes that de­ tal Stress Analysis, Vol. 11, No. 1, pp duced by peening, some of the greater velop residual tensile stress in steel 157-160. 2. J. 0. Almen, "Fatigue Weakness of subsurface strength became available. surfaces increase brittleness. Surfaces", Product Engineering, Novem­ It is possible that had these hard Several thermal and mechanical ber, 1950, pp 118-139. processes, other than those mentioned 3. J. L. Littleton, Jr., "A New Method springs been strain peened to increase for Measuring the Tensile Strength of the magnitude of the residual com­ in the preceding paragraphs are known Glass", Physical Review, Vol. 22, 1923, pressive stress (Ref. 2), even greater to induce residual compressive stresses pp 510-516. 4. J. O. Almen, "Fatigue Failures Are fatigue durability would have resulted. in steel surfaces. Additional thermal Tensile Failures", Product Engineering, The tests on these production and processes are drastic quenching of low March, 1951, pp 102-103. experimental springs were made dur­ hardenability steel, steel quenched 5. J. 0. Almen, "Torsional Fatigue Failures", Part II, Product Engineering, ing World War II by the Spring Divi- from tempering temperatures higher March, 1952.