UDK 620.179.1.:539.38:621.791.05 Izvirni znanstveni članek ISSN 1318-0010 KZLTET 33(6)459(1999) N. GUBELJAK ET AL.: MEASUREMENT OF STRAIN CAUSED BY RESIDUAL STRESSES… MEA SURE MENT OF STRAIN CAUSED BY RE SID UAL STRESSES IN A WELDED JOINT USING NEU TRON DIF FRAC TION MERITEV DEFORMACIJ POVZROČENIH Z ZAOSTALIMI NAPETOSTMI V ZAVARNEM SPOJU NA OSNOVI NEVTRONOVSKEGA ODKLONA Nenad Gubeljak 1, Jelena Vojvodič Tuma 2, Hans-Georg Priesmeyer 3 1University of Maribor, Faculty of Mechanical Engineering Maribor, Smetanova 17, Slovenia 2Institute of Metals and Technology Ljubljana, Lepi pot 11, Slovenia 3Institute for Experimental and Advanced Physics, Christian-Albrecht-University Kiel and GKSS Resear ch Cen ter Geesthacht, Germany Prejem rokopisa - received: 1999-10-15; sprejem za objavo - accepted for publication: 1999-11-19 The dis tri bu tion and amount of re sid ual stresses can sig nif i cantly con trib ute to the fa t igue frac ture be hav iour of welded joints in struc tures. In te rior re sid ual stresses, which in ter act with the plane strain state, are more dan ger ous than re sid ual stresses at the sur face of the welded joint. If the sur face of the welded joint is me chan i cally treated (e.g. sharp en ing), then sig nif i cant dif fer ences be tween the stress states at the sur face and within the vol ume oc cur . Hence, dif fer ent non-destructive meth ods (e.g. gamma ra di a tion, neu tron dif frac tion) have been de vel oped to de ter mine strains caused by re sid ual stresses in the vol ume of polycrystalline ma te ri als. How ever, the mea sure ment of strains caused by re sid ual stresses be comes dif fi cult for large sam ples. In this pa per the mea sure ment pro ce dure on a sam ple taken from the weld joint is pre sented and the ef fect of re sid ual stresses on fa tigue crack prop a ga tion is as sessed for low and high cy cle load ing fa tigue. Key words: re sid ual stresses, neu tron dif frac tion method, fa tigue crack prop a ga tion, weld joint Porazdelitev in višina zaostalih napetosti lah ko značilno vpliva na utrujenostno obnašanje zvarnih spojev in varjene konstrukcije. Notranje zaostale napetosti so zaradi vzajemnega delovanja z ravninskim deforrmacijsk im stanjem bolj nevarne od zaostalih napetosti na površini. V primeru, če je površina zvarnega spoja mehansko obdelana (npr. b rušena) se pojavijo izrazite razlike med zaostalimi napetostmi na površini in v notranjosti materiala. Zato so bile razvite razl ične neporušitvne metode (npr. gama sevanje, neutronska difrakcija) s katerimi se določajo deformacije, ki so posledica zaostalih napetosti v polikristalnih materialih. Zaradi omejene penitracije neutronov (do 20mm) postaja meritev zaostalih napetosti prob lematična pri večjih debelinah. V članku je predstavljen postopek meritve deformacij na iz zvarnega spoja izrezanih prei zkušancih povzročenih z zaostalimi napetostmi in prikazan vpliv zaostalih napetosti na nizko in visokociklično utrujanje. Ključne besede: zaostale napetosti, neutronski lom žarkov, utrujenostno širjenje razpoke, zvarni sp oj 1 IN TRO DUC TION The presence of tensile residual stresses during service loading of welded structures causes the initiation of fatigue cracks from defects or flaws in welded joints. The three to four times lower stress intensity factor range ?Kth is caused directly by residual stresses 1. Most of the publications 2 ,3 consider the effect of residual stresses in the direction of fatigue crack propagation. Neverthless, tensile residual stresses in the thickness direction are crucial for the initiation of fatigue cracks from the defect or flaw in the welded joint. If the thickness of the welded plate increases, then the effect of residual stresses becomes more significant, because the possibility of a three axial stress state increases. The aim of this paper is to determine strains in specimens of a weld joint caused by tensile residual stresses in the volume of polycrystalline materials, and assess the effect of residual stresses on fatigue crack propagation in low and high cycle loading fatigue. 2 RE SID UAL STRESSES DE TER MI NA TION USING HIGH-RESOLUTION NEU TRON DIF FRAC TION High-resolution neutron diffraction has become a powerful method to determine strains caused by residual stresses in polycrystalline materials. The high penetration of neutrons allows the measurement of lattice plane distances by coherent Bragg scattering within the grains of polycrystalline engineering materials. Lattice plane distortions occur in the elastic regime, where the lattice plane distance has to be measured with an accuracy of 10 -4 . Neutron time-of-flight (TOF) spectroscopy has been used to determine strains caused by residual stresses. The relation between the neutron wave length ?, the lattice spacing d and the scattering angle 2T is given by the well-known Bragg formula: dhkl = X 2 -sin(Gh kl) (1) which can be written for a time-of-flight diffractometer as: KOVINE, ZLITINE, TEHNOLOGIJE 33 (1999) 6 45 N. GUBELJAK ET AL.: MEASUREMENT OF STRAIN CAUSED BY RESIDUAL STRESSES… dK> h1 mn L 2 -sin(0AA/) (2) Equation (2) shows that the neutron wave length X and the neutron time-of-flight t are proportional to each other. The lattice planes are defined by their Miller indices Ml, h is Planck's constant, /n„the neutron mass, L the neutron flight path and t the neutron flight time. With the mo der n crystal or time-of-flight diffractometers the scattering angles 2® hm or the flight times tm can be determined to yield the required accurancy in dm. Strains are determined as relative deviations of lattice spacings dm (sample with residual stresses) from the equilibrium state d0hkl (sample without residual stresses): hkl 0 hid 0 hid (3) Strains are then converted into stresses using the generalized Hooke's law: (4) G . = y C ..nE n where Cijkl are the stiffness coefficients of the material. This calculation has not only to take into account the dependence of the crystal's elastic properties on orientation, but also models of intergranular force coupling as well as the influences of preferred orientation (texture), microstructure and plastic anisotropy 4. As can be seen from eq. (3), the incorrect determination of lattice spacings for the stress-free case d0hkl will cause systematic errors for the stressed state. 3 MEA SURE MENT PRO CE DURE The measurements were carried out at the 5MW research reactor FRG-I of the GKSS Research Cen t e r using a (three point bend) fracture toughness specimen machined out of a welded plate. The specimen without residual stresses was stress relieved at 530°C for 2 hours. The GKSS Fourier correlation spectrometer FSS (Fourier-Strain-Spectrometer), operating in the reverse % X gauafl volumna plane of Čîrnen" Fig ure 2: Po si tion of gauge vol ume (2x2x30 mm) in the welded joint Slika 2: Položaj obsevanega volumna (2x2x30 mm) v zavarjenem spoju time-of-flight mode, has been optimised to measure strains in engineering components 5. The neutron flux at the sample position was 3x10 6 n/cm 2sec. Scattered neutrons were detected by a bank of 16 Lithium-6 loaded glass scintillators in time-focussing geometry. The nominal resolution was 3x10 -3 . Figure 1 shows a schematic diagram of the time-of-flight diffractometer FSS. The measurement was čjjč| east pj detector Fig ure 1 : Sche matic view of mea sur ing re flec tions in ten sity o f neu trons on spec i men Slika 1: Shematski prikaz meritve gostote odbojev neutronov na preizkušancu 460 "b s Ě4 1-Q (211) (31fl) STO 1000 1500 2000 elisile!;: 3S00 se« 3500 b) Fig ure 3: Com plete time-of-flight dif frac tion pat tern mea sured at a) east and b) west de tec tor Slika 3: Difrakcijski vzorec izmerjen v celotnem času prileta nautronov na a) vzhodnem in b) zahodnem detektorju KOVINE, ZLITINE, TEHNOLOGIJE 33 (1999) 6 E hki N. GUBELJAK ET AL.: MEASUREMENT OF STRAIN CAUSED BY RESIDUAL STRESSES… 0(4»]) OC211) .........< FT™)"" ..................Ň.LV&-... < ? (351) trsile Č ÖOD> minprs55iDn «Mkl Fig ure 4: Dis tri bu tion of strains in lat tice planes of the fer rit ic phase (bcc) Slika 4: Porazdelitev deformacij v kristalografskih ravninah za feritno fazo (bcc) performed using a gauge volume of 120 mm 3 (2x2x30 mm). The basic assumption was that the strain in the weld pass direction (30 mm) does not significantly change for a certain point. It is assumed that the peak of the tensile residual stresses is 6 mm away from the surface of the specimen, as shown in Fig. 2 . The highest tensional stresses are expected in this region. The time-of-flight (TOF) neutron diffraction method gives the average lattice distance for the whole measured volume. Using the TOF method it is possible to determine strains for many different lattice planes in the spectrum. As shown in the spectrum, see Fig. 3 , several reflections were measured simultaneously, so it is useful to employ more than one Bragg peak for the strain determination. Each Bragg peak belongs to one of the lattice planes. The full spectrum covers the (110), (200), (211), (220), (310) and (321) reflections of the ferritic cubic-bcc structure. 4 DIS CUS SION The set of equations was solved with respect to the strain components. The principal strains caused by residual stresses are shown in Fig. 4 . The deviations in measured strains and statistical errors are large. The large statistical error is a consequence of measuring in a large volume. In the large volume the strains in a particular lattice plane are summed as tensile and compressive contributions in the same direction in the gauge volume. The neutron penetration in ferritic steels is limited but the measurement requires a minimum number of neutrons in the beam. Therefore, if the thickness of a specimen increases then the gauge volume should be larger, but then the prob lem of average stresses arises in the observed volume. A stronger neutron beam is not possible because the measuring is KOVINE, ZLITINE, TEHNOLOGIJE 33 (1999) 6 Fig ure 5 : Ob tained fa tigue crack shapes through-the-thickness of spec i men, a/W=0.5; W=72 mm us ing a) low and b) high cy cling fa tigue un der the ef fect of re sid ual stresses Slika 5: Dobljene oblike fronte utrujenostne razpoke skozi debelino preizkušanca a/W=0.5; W=72 mm z uporabo a) nizko in b) visoko cikličnega utrujanja pod vplivom zaostalih napetosti limited. The strain distribution of the (200) plane in the sample is significantly lower than the other investigated lattice planes. The lattice plane (200) is in the softest direction 6. The greatest Young's modulus is the consequence of the crystals grain orientation due to the direction of the neutron beam. The other lattice planes show significant tensile strain. The maximum strain of the examined lattice plane (211) is 3-4x10 -4 . The effect of tensile residual stresses on fatigue propagation was estimated using low and high cycle loading. Low and high cycling fatigue were performed on a load level of 60%F y-yield load (amplitude load ?F=F max - F min =18%F y ) and on a load level of 20%F y (amplitude load ?F=F max -F m in =18%F y), respectively. The obtained fatigue crack shapes for both cases are shown in Fig. 5 a) and b) respectively. Residual stresses are present in the specimen cut out of the welded joints. For high cycling fatigue the crack shape shows a significant effect of the residual stresses on fatigue pre-cracking only in the tensile residual stress region, where tensile strain was measured. In the case of low cycling fatigue the crack shape shows a decrease of 461 N. GUBELJAK ET AL.: MEASUREMENT OF STRAIN CAUSED BY RESIDUAL STRESSES… the effect of residual stresses. However, the effect of tensile residual stresses near the surface and compressive residual stresses in the mid-thickness is still evident. Therefore, the effect of residual stresses is significant during the initiation of fatigue crack propagation. 5 CON CLU SION The strains caused by residual stresses are determined on the bas is of high neutron diffraction measurements. The interior strains in specified lattice planes are calculated. Except polycrystalline structure and chemical composition of microstructures, the accuracy of measurements of lattice distance depend on the thickness of the specimen. Therefore, in the case of large specimens it is necessary to use large sample volumes for neutron diffraction, if the intensity of the neutron beam is limited. For a large gauge volume the measurement integrates over all lattice planes. Hence, using a large gauge volume causes broadening of a Bragg reflection, if strain gradients are present. The performed measurements show the largest residual stresses for crystallographic lattice planes (220), (221) and (321). Residual stresses in these planes support the initiation of fatigue crack propagation at high cycling levels. The tensile residual stresses contribute to a higher susceptibility of welded joints on fatigue cracking. The result of this event is a lower threshold of the stress intensity factor range ?Kth . Therefore, the often unconsidered effect of residual stresses through the thickness of a welded joint presents a damage parameter of the welded structure. 6 REF ER ENCES 1 Japan Welding Engineering Society Stan dar d: "Method of Assessment for Flaws Welded Joints with Respect to Brittle Fracture and Fatigue Crack Growth, WES 2805- 1997 2 Glinka G.: "Effect of Residual Stresses on Fatigue Crack Growth in Steel Weldments Under Constant and Varible Amplitude Loads", American Society for testing and Materials STP 677, C.W. Smith, Ed. 1979 , 198-214 3 Chang J.B., Chiu S.T., Goodemberger D.L.: "Residual Stress Effects on Safe-Life of Welded Structures", Proceeding of 9 th Int. Conf. on Fracture , Sydney, Australia 1, 1997 , 547-554 4 Mohr A., Priesmeyer H.-G.: "Experimental Correlation Between the Elemental Composition and the Lattice Dimensions in Alloys, Using AlCu as a Model Substance", Int. Conf. Neutrons and Their Applications , Crete, Greece, 1994 , SPIE Vol. 2339, 350-355 5 Priesmeyer H.- G.:"Reverse time-of-flight Fourier Technique for Strain Measurements", Proceedings NATO ARW Oxford , Kluwer Dodrecht, 1992 , 329-334 6 Schröder J., Priesmeyer H.- G.: "Residual Stress measurements in an Inertia Welded Sample by Means of Different Neutron Diffraction Techniques", Proceedings of 3 th Int. Conf. about Residual stresses , 253-258 46 KOVINE, ZLITINE, TEHNOLOGIJE 33 (1999) 6