0 penilaian0% menganggap dokumen ini bermanfaat (0 suara)
23 tayangan19 halaman
This paper presents a theoretical analysis of crack nucleation in isotropic polycrystalline ice due to the elastic anisotropy of the constituent crystals. The singularity of the associated stress concentrations near a grain-boundary facet junction provides the mechanism f or inducing microcrack precursors. Model predictions show that the stress required to nucleate cracks in compression is about 2. Times that in tension.
Deskripsi Asli:
Judul Asli
Crack Nucleation Due to Elastic Anisotropy in Polycrystalline Ice
This paper presents a theoretical analysis of crack nucleation in isotropic polycrystalline ice due to the elastic anisotropy of the constituent crystals. The singularity of the associated stress concentrations near a grain-boundary facet junction provides the mechanism f or inducing microcrack precursors. Model predictions show that the stress required to nucleate cracks in compression is about 2. Times that in tension.
This paper presents a theoretical analysis of crack nucleation in isotropic polycrystalline ice due to the elastic anisotropy of the constituent crystals. The singularity of the associated stress concentrations near a grain-boundary facet junction provides the mechanism f or inducing microcrack precursors. Model predictions show that the stress required to nucleate cracks in compression is about 2. Times that in tension.
Cold Regions Science and Technology, 18 (1990) 29-47 29
Elsevier Science Publishers B.V., Amsterdam -- Printed in The Netherlands
CRACK NUCLEATION DUE TO ELASTIC ANI SOTROPY IN POLYCRYSTALLINE ICE S. Shyem Sunder and Mao S. Wu Massachusetts Institute of Technology, Department of Civil Engineering, Room 1-273, Cambridge, MA 02139 (U.S.A.) (Received June 12, 1989; revised and accepted August 21, 1989) A B S T R A C T This paper presents a theoretical analysis of crack nucleation in isotropic polycrystalline ice due to the elastic anisotropy of the constituent crystals. The singularity of the associated stress concentra- tions near a grain-boundary facet junction provides the mechanism f or inducing microcrack precursors, i f similar nuclei do not already exist. The first-order microstructural stress f i el d created by the elastic an- isotropy mechanism is linearly superposed on the ap- plied stress field. This total stress f i el d causes the pre- cursors to nucleate into microcracks. The analysis of the nucleation stress is based on a solution to the general problem of an extending precursor in a com- bined stress f i el d including the effects of Coulombic frictional resistance. The local material resistance is characterized in terms of a critical value f or the max- i mum principal tensile stress which can be deter- mi ned f rom the surface free energy of either the grain boundary or the solid-vapor interface. Model predictions show that: (a) the stress re- quired to nucleate cracks in compression is about 2.5 times that in tension, unlike other microstructural models which predict them to be equal; (b) elastic an- isotropy rather than dislocation pile-up as proposed by others may govern crack nucleation in tension over a wide range of strain rates; (c) the stress required to nucleate a crack in compression is strongly depen- dent on crystal orientation and, as a consequence of the random orientation of crystals in isotropic poly- crystalline ice, there can be a distinct beginning and end to the microcrack nucleation phase when stress is increased and i f failure does not occur prema- turely; (d) the grain size effect due to the elastic an- isotropy mechanism is similar to that due to the dis- location pile-up mechanism over the typical range of grain sizes encountered in nature; and (e) a general- ization of the limiting tensile strain criterion* which accounts f or the anisotropy of the constituent crystals is an excellent phenomenological approximation of the nucleation surface under multiaxial states of stress. I N T R O D U C T I O N The ductile-to-brittle transition in polycrystalline ice is significantly influenced by the nucleation of microcracks. For example, the stress required to nucleate such cracks governs the tensile strength for typical grain si zes and over several decades of strain rate. In addition, under compressive states of stress the nucleation of microcracks rather than their growth is primarily responsible for the progressive accumulation of damage in the ductile-to-brittle transition regime. Gold (1972) has observed the microcrack den- sity during compressive creep of columnar-grained polycrystalline ice and postulated the existence of two independent families of cracks on the basis of a statistical analysis. The probability of nucleation for one family of cracks increases directly with creep strain, while that for the other is independent of creep strain. He proposes that stress concentrations due to the pile-up of dislocations are responsible for the first family of cracks, while the growth of nuclei due to diffusion or a gradually increasing local stress is responsible for the second family of cracks. These latter cracks form mainly at grain boundaries and *Proposed by Shyam Sunder and Ting ( 1985 ). 0165-232X/90/$03.50 1990 Elsevier Science Publishers B.V. 3 0 S. SHYAM SUNDER AND M.S. WU are characteristic of purely brittle behavior. The dislocation pile-up mechanism has subse- quently been adopted by Schulson (Schulson, 1979; Schulson et al., 1984) and Cole (1986) to explain crack nucleation under tension and compression, respectively. The study of dislocation etch pits in deformed ice by Sinha (1978) which shows that dislocations do pile-up at grain boundaries appears to be the only microstructural evidence which sup- ports the operation of this nucleation mechanism. Schulson et al. (1984) explicitly associate nuclea- tion-controlled tensile failure with the brittle be- havior of ice, while Schulson (1987) goes on to state that nucleation occurs under the same uniaxial stress in compression as in tension. During compressive creep tests on columnar- grained polycrystalline ice, Sinha (1984) found no evidence of dislocation pile-ups around micro- cracks in hundreds of grains. He concluded that stress concentrations resulting from dislocation pile- ups may not be the mechanism responsible for crack nucleation. In turn, Sinha (1982, 1984) postulated that grain boundary sliding causes stress concentra- tions at triple points between grains and at jogs in the boundaries and that these stress concentrations are responsible for crack nucleation. Since grain boundary sliding is responsible for the anelastic or delayed elastic component of strain in the creep model of Sinha ( 1979 ), he goes on to state that this mechanism of crack nucleation does not involve purely elastic deformation and hence does not cor- respond to truly brittle behavior. There is as yet no consensus in the ice literature as to which of the two mechanisms (dislocation pile-up versus grain boundary sliding) is responsible for the nucleation of strain-dependent microcracks. The internal stress and strain fields in a polycrys- talline aggregate are very inhomogeneous due to crystal anisotropy and dislocation movement. Both these sources of irregularity can lead to stress con- centrations particularly at grain boundaries and cause crack nucleation. Cole (1986) proposes that the elastic anisotropy of ice crystals is specifically responsible for the strain-independent crack distri- bution observed by Gold (1972) at high rates of loading. Cole ( 1988 ) has subsequently compared the dis- location pile-up mechanism with the elastic aniso- tropy mechanism. His analysis shows that while the basal plane contains a sufficient density of glissile dislocations to form the pile-up, the time required to move the dislocation into the appropriate config- uration (approximately 125 s) is much greater than the experimentally observed times for the nuclea- tion of first cracks (typically less than 20 s). He concludes that "the first cracks to nucleate are not a result of the pile-up mechanism, but rather a re- sult of the elastic anisotropy mechanism". His sub- sequent analysis shows that the two mechanisms cannot be readily distinguished on the basis of either the nucleation stress or the grain size dependency. The analysis of the elastic anisotropy mechanism by Cole ( 1988 ), however, is based on a highly sim- plified model which considers just two isolated ice crystals subjected to a uniaxial stress.The elastic strain energy of each crystal is computed by assum- ing that its response is uncoupled from the other crystal. The difference in strain energy of the two grains drives crack nucleation which, in effect, cor- responds to a nucleation event driven by shear on the grain boundary. The analysis also implicitly predicts that crack nucleation occurs at the same uniaxial stress in compression and tension. This paper presents a theoretical analysis of crack nucleation due to elastic anisotropy in polycrystal- line ice. The microstructural stresses due to the elastic anisotropy of individual crystals in an oth- erwise isotropic polycrystal are analyzed using a first-order approximation of the approach devel- oped by Eshelby (1957). A similar first-order ap- proximation has been used by Evans (1984) to analyze the grain-boundary residual stresses in iso- tropic ceramic polycrystals stemming from the thermal expansion anisotropy of individual crys- tals. The singularity of the stress concentration near a grain-boundary facet junction due to elastic an- isotropy provides the mechanism for inducing mi- crocrack precursors, if similar nuclei do not already exist as proposed by Gold (1972). The precursors can nucleate into microcracks through the local in- tensification of the applied and microstructural/in- ternal stress fields. The analysis of the stress required to nucleate mi- crocracks and the incipient growth direction is based on a solution to the general problem of an extending precursor in a combined stress field. This solution CRACK NUCLEATION 31 uses the maxi mum principal tensile stress ( MPTS) criterion for mixed-mode crack extension which has been proposed by Erdogan and Sih (1963). The lo- cal material resistance may be characterized in terms of the surface energy of either the grain boundary or the sol i d-vapor interface, bot h of which are similar in magnitude for ice (Ketcham and Hobbs, 1969). The effect of Coulombic frictional resistance is in- cluded in defining the effective stress driving the precursor in the shearing mode of deformation. Further, it is postulated that the first crack to nu- cleate is associated with the most favorable orien- tations of both the precursor and the adjoining grains. The analysis procedure is applied to study the problem of crack nucleation in polycrystalline ice under uniaxial and biaxial states of loading involv- ing bot h tensile and compressive stresses. The ef- fects of grain size on the nucleation stress are also explored. The predictions of this microstructural model are then compared with the phenomenolog- ical model based on a limiting tensile strain crite- rion that has previously been proposed by Shyam Sunder and Ting ( 1985 ). ANALYSI S OF MI CROSTRUCTURAL STRESS FIELD The analysis of microstructural stresses due to the elastic anisotropy of individual crystals in an oth- erwise isotropic polycrystal is based on a first-order approximation of the Eshelby (1957) proce- dure.The first step, in the complete three-step Esh- elby procedure (see Fig. 1 ), is to separate from the solid those microstructural features which create the major part of the microstructural stresses of inter- est. To calculate the stresses in the vicinity of grain boundaries due to elastic anisotropy in a two-di- mensional system, this step corresponds to separat- ing the grains surrounding the precursor of interest since they will dominate the stress field. The behav- ior of each of these separated grains is defined in terms of the elastic stress-strain relations for single ice crystals, i.e.: ~,=s,o (I) where a is the applied stress vector, c, is the vector ISOTROPIC MATRIX X n 0 _ Fig. 1. Schematic of three-step Eshelby procedure for analyz- ing microstructural stress field. of strains induced in each of the individual grains expressed in engineering notation as c = [ ~ Em ~=, ~ ' y z , 7 z x , 7 x y ] r with T denoting the transpose opera- tion, and Sg is the compliance matrix of a single crystal in the chosen frame of reference. If the cho- sen frame of reference does not coincide with the standard frame of reference for the hexagonal ice crystal, then Sg is given by: S~= Rs TS,,Rs (2) where Rs is the three-dimensional rotation matrix and S, , is the compliance matrix of the single ice crystal in the standard frame of reference both of which are defined in Appendix A. This compliance tensor is anisotropic since ice possesses a hexagonal crystal structure with five independent elastic con- stants, not j ust two as for isotropic materials. The dynamic elastic constants of single ice crystals have been determined by several investigators (see, e.g. Gammon et al., 1983 ). The remaining matrix is assumed to act as an iso- tropic medium, having the same elastic properties as the polycrystalline body, i.e.: ffm ~--SmO" ( 3 ) where Cm is the vector of strains induced in the ho- mogeneous matrix and Sm is the elastic compliance 32 S. SHYAM SUNDER AND M.S. WU tensor of the isotropic polycrystal given in Appen- dix A and has the same components in all orthogo- nal coordinate frames. The polycrystal elastic con- stants can be determined through independent testing or from the monocrystal data as outlined by Gammon et al. (1983). In the second step, the separated grains are de- formed to fit into the cavity in the matrix by apply- ing surface tractions on each grain. Microstructural elastic stresses are induced in the grain due to the misfit strains and are given by: O'0 ~,~- Cg ( Em - - Eg ) (4) where Cg is the elastic stiffness matrix for single ice crystals in the chosen frame of reference which can be determined from the crystal stiffness matrix Cgs in the standard frame of reference and the associ- ated rotation matrix Rc given in Appendix A. Using Eqs. 1 and 3, Eq. 4 can be rewritten in the following more convenient form: ~ro= ( CgSm- - I )o (5) The surface tractions on each grain can then be ob- tained from the equilibrium conditions on the surface. In the last step, the separated grains are welded back to the matrix and the surface tractions on their boundaries are relaxed. This step creates an addi- tional stress field tr c in the whole body. The result- ant microstructural stress field due only to the ef- fects of elastic anisotropy can then be expressed as the sum of,70 and tr. The total stress field crt is the sum of the applied stresses and the microstructural stresses. A complete solution obtained with the Eshelby procedure by Evans (1984) for the problem of ther- mal expansion anisotropy in ceramic potycrystals shows that the microstructural stress distribution in the vicinity of a grain-boundary facet junction may be expressed as: a( x ) = [I + Fl n( l / x ) ]ao,avg (6) where I is the identity matrix, F is a diagonal matrix with the elements representing shape functions that take into account the crystallographic orientations of the adjacent grains, angle between facets, etc., and ~ro,avg is the average of the microstructural stresses given by Eq. 4 for the two grains adjoining the pre- cursor under consideration. The variable l is the length of the grain boundary facet which Cole (1988) has estimated to be about 0.56 times the grain size d, and x is the distance from the grain boundary facet junction. Due to the logarithmic term in Eq. 6, the stresses become singular near the grain junction as x tends to zero. Unlike thermal stress concentrations, stress singularities caused by a geometrical discontinuity along a bimaterial interface in elasticity are often much more powerful than l n( l / x ) . These singular- ities are generally expected to be of the form ( l / r ) ~, where 0 < 2 < 1 and r is the radial distance from the grain boundary facet junction (see, e.g. Bogy and Wang, 1971 ). The intact material cannot sustain a singular stress field since it exceeds the molecular forces of cohesion. Consequently, the stress singu- larity at the grain boundary junction provides the mechanism for inducing microcrack precursors. In the case of highly irregular grain boundaries as is the case for an actual material specimen, stress sin- gularities may occur at geometric discontinuities such as steps and ledges on the grain boundary in addition to those occurring at junctions. Although the average microstructural stress does not depend on the grain facet size, the scale effect contained in the singularity allows the stress to be sustained over a larger area of grain facet as the facet length increases. Consequently, the microcrack pre- cursor will tend to be larger for larger grains. Al- though the details of this process are not well-known even for ceramic materials, it is reasonable to con- sider ratios of precursor size 2a to facet length or grain size that vary in the approximate range given by 0.1 <2a/ l <0. 2 or 0.05 <2a/ d<O. l O. This is the typical length over which the influence of the sin- gularity is a maximum. The first-order approximation of the Eshelby procedure is based on the average microstructural stress field given by tro,avg, i.e. the contribution of ~rc to the microstructural residual stress field which is responsible for the singularity term of Eq. 6 is ne- glected. Once a precursor has formed, the asymp- totic stress field ahead of the precursor is taken to follow the well-known square-root singularity of linear elastic fracture mechanics and not the singu- larity due to geometric stress concentrations which C R A C K N U C L E A T I O N 33 is derived from the analysis of an intact or defect- free material. The precursors can nucleate into microcracks through the local intensification of the total stress field which is the sum of the applied stress field and the first-order approximation of the microstruc- tural/internal stress field given by Eq. 5. If the mag- nitude of stress intensity is small, then either the formation of the precursor may be prevented or if the precursor does form it may not be able to nu- cleate into a microcrack (a crack typically of the size of a grain). Equation 5 predicts that the precursor is in general subject to both normal and shear stresses. For example, when polycrystalline ice is loaded in uniaxial compression the equation pre- dicts a normal tensile stress on a precursor that is parallel with the loading axis and when the basal plane orientation lies in the range given by 0<( <25 or 65 <( <90 (see Fig. A.1 in the Appendix). CRACK NUCLEATI ON CRI TERI ON The microcrack precursor or nucleus is in general subject to a combined stress field, not only tension or pure shear. In addition, its extension may take place at an angle with respect to the orientation of the precursor. Thus, the analysis of the stress re- quired to nucleate microcracks and the incipient growth direction must be based on a solution to the general problem of an extending precursor in a combined stress field. Asympt ot i c stress f i el d As previously stated, the asymptotic stress field ahead of the precursor is taken to follow the well- known square-root singularity of linear elastic frac- ture mechanics, not the more general geometric stress singularity which is derived from the analysis of an intact or defect-free material. Further, the pri- mary effect of crystal anisotropy is to determine the total remote stresses acting on the precursor; its in- fluence on the local asymptotic stress field is con- sidered to be negligible. In the case of thermal anisotropy, Evans ( 1978 ) recognizes that the superposition of the logarithmic A 5C 0 O It. >- I--- Z lit I " _z CO iii t r o~ a iii . . d < t r O Z 0.9 02- 0,7- 0 . 0 0 . 5 0 . 4 . 0 , 8 0 . 2 0 . 1 # 0 . 0 0 .~}0 I " 1 I I I o ! \ < ' = i ~ l G l i , A I N ARIIUI Y ~ ~ . , / t l # / i i l ~ t ~ l / f T H R O I ~ I H C P I ~ K I N t N F I N l l l l P I j I I T E { l ~ F l l l ~ # i i / I N G L E ' G R A I N / f / l I I I t 0. 05 0 . 1 0 0 . 1 5 0 . 2 0 0 . 2 . 5 0 . 9 0 RELATIVE CRACK LENGTH, 2a/I Fig. 2. Nor mal i zed st ress i nt ensi t y f act or pl ot t ed as a func- t i on of t he r at i o of pr ecur sor size t o t he grai n facet l engt h ( das hed l i nes for single grai n, grai n pai r and grai n array, re- pr oduced f r om Evans, 1978). stress field on a material with defects is an approx- imation and that the stress intensity factor tends to be overestimated as a result. He goes on to compute the mode I stress intensity factor for precursors at grain boundary junctions using the asymptotic stress field given by the logarithmic singularity for an iso- lated single grain, a pair of grains, and an array of grains. Figure 2 (reproduced from Evans, 1978) shows the normalized stress intensity factors for each of these cases. The larger intensities for the grain pair and grain array are achieved by orienting the crys- tals in the most favorable manner. Superposed on the figure is the normalized stress intensity factor from linear elastic fracture mechanics for a through crack of length 2a in an infinite plate, i.e.: K , [ ' z c a ' ~ '1~ {7) 34 S. SHYAM SUNDER AND M.S. WU Thi s convent i onal stress i nt ensi t y measure, whi ch charact eri zes an asympt ot i c stress field with a square-root singularity, is seen t o follow t he general t r end pr edi ct ed by t he model of Evans ( 1978 ) f or the range of precursor sizes under consi derat i on. Equat i on 7 is al most exact for t he case where t wo grains are favorabl y ori ent ed. In general, t he prob- ability of at t ai ni ng hi gher stress i nt ensi t i es di mi n- ishes for an i sot ropi c pol ycryst al since it is neces- sary for several grains ( not j ust t wo) to be favorabl y ori ent ed. The curve f or t he grain array in Fig. 2 is in essence an upper bound. Furt her, since the model predi ct i ons based on superposi t i on can be some- what over est i mat ed (Evans, 1978), t he conven- tional i nt ensi t y measure is in fact a good represen- t at i on for bot h t he grain pai r and grain array. For t he pr obl em under consi derat i on, t he geo- met ri c stress singularity applies onl y f or t he mi cros- t ruct ural stress field i nduced by crystal ani sot ropy. Once the precursor has f or med and a significant part of t he stress concent r at i on has been rel i eved, t he asympt ot i c field governi ng mi cr ocr ack nucl eat i on will be square-root singular. Furt her, t he appl i ed stresses generally domi nat e t he t ot al stress field. The stress i nt ensi t y associ at ed wi t h a pr ecur sor subject onl y t o t he appl i ed stress field in an i sot ropi c solid shoul d be given by t he convent i onal i nt ensi t y mea- sure in Eq. 7. Clearly, this woul d be t he measure of choi ce for a precursor in t he absence of mi crost ruc- t ural stresses due t o crystal ani sot ropy. Finally, ex- t ensi on of t he pr ecur sor in a combi ned stress field is significantly mor e compl ex and, in fact, t he for- mul at i on by Evans ( 1984) uses t he convent i onal measures of stress i nt ensi t y in t he tensile and shear- ing modes of def or mat i on t o defi ne an effect i ve mi- crost ruct ural stress. Ma x i mu m pri nci pal t ensi l e st ress cri t eri on Two al t ernat i ve solutions, t he maxi mum pri nci - pal stress cri t eri on of Erdogan and Sih (1963 ) and t he mi ni mum st rai n energy densi t y f act or cri t eri on of Sih ( 1974) , are of t en adopt ed f or t he pr obl em of an ext endi ng pr ecur sor in a combi ned stress field. In t he f or mer approach, a critical stress i nt ensi t y defi nes t he i ni t i at i on stress while t he di r ect i on of ext ensi on is post ul at ed t o occur in a di r ect i on per- pendi cul ar t o t he local maxi mum tensile stress. In
i PRECURSOR
Z Y,q ) Y z Fig. 3. Definition of coordinate system and incipient growth direction for an inclined precursor in a combined stress field. t he l at t er approach, t he critical st rai n energy den- sity fact or specifies bot h t he i ni t i at i on stress and t he di r ect i on of ext ensi on. The f or mer cri t eri on f or ms t he basis of t he crack nucl eat i on analysis adopt ed here. A compar i son of t he t wo cri t eri a f or t he pr ob- l em of crack nucl eat i on in ice is pr esent ed in a f or t hcomi ng paper by Wu and Shyam Sunder ( 1990) . The f or mul at i on t akes i nt o account t he com- bi ned stress field generat ed by bot h t he appl i ed and mi cr ost r uct ur al stresses, as well as t he effect of Cou- l ombi c fri ct i onal resistance in defi ni ng t he effect i ve stress dri vi ng t he pr ecur sor in t he sheari ng mode of def or mat i on. Pr ecur sor ext ensi on is consi dered t o be a mi xt ur e of modes I and II and t o t ake place under t he act i on of a bi axi al stress field including t ensi on, compr essi on and shear. Thi s is consi st ent wi t h t he mi cr ost r uct ur al stress field i nduced by an externally appl i ed tensile or compressi ve l oad which generally i ncl udes bot h nor mal a nd shear stresses. Consi der t he pr ecur sor shown in Fig. 3 whi ch is in t he x - z plane, or i ent ed at an angle 1/with t he x- axis, and subj ect ed t o t he combi ned stress fi el d Or. Assumi ng a t wo- di mensi onal stress field, t he "t o- t al " r emot e nor mal stress O'pp and shear stress trpr acting on t he pr ecur sor are gi ven by: o ,io 2fl (8) CRACK NUCLEATION 35 trpr = - ( a:~ 2 o ~ ) s i n 2 f l - a x z c o s 2 f l +_ lurpp = ~'~ + f l a p p (9) where/~ is the coefficient of friction and #trpp is the Coulombic frictional stress which opposes sliding whenever the normal stress is compressive (nega- t i ve), i.e. the sign in Eq. 9 is positive when 12>0 and negative when 12 < 0. The effective shear stress given in Eq. 9 is set to zero whenever the magnitude of the sum of the first two terms, i.e., 12 is smaller than or equal to the frictional stress term. The asymptotic stress field in the vicinity of the precursor can be expressed as the sum of the stress fields due to the deformations in modes I and II, respectively: aij = K x / ( 2zcr ) l / 2 f i j ( O ) +KII / (27rr)l/2filij(O ) (10) where K~ and KII are the stress-intensity factors in modes I and II, respectively, while f~o and f I I i j a r e trigonometric functions of the angle 0 defined in Fig. 3. The stress intensity factors for standard crack ge- ometries are available in the literature (see, e.g. Sih, 1973 ). For a through thickness precursor of length 2 a in an infinite plate, they are listed below: K I = 7 p p ( 7 ~ a ) 1 / 2 ( 1 1 ) Kn = O'pr (Tt'a)I/2 (12) If avp is compressive, it is assumed that the crack is closed and consequently KI = 0. The asymptotic distribution of the tangential and shear stresses given by Eq. l 0 can be expressed as follows: C r o o = ( 2 7 ~ l r ~ 7 ~ c o s ( O ) [ g , c o s 2 ( O ) - 3 K i i s i n O ] (13) ( o ) trr0 2(2rtr)t/2 cos [KI sin 0+Kxx(3 cos 0 - 1 ) ] (14) where r is the radial distance from the tip of the pre- cursor and 0 is the angle measured anti-clockwise from a line extending along the precursor in the solid as shown in Fig. 3. The stress aoo will be the principal stress i f trio= 0. This is the case for 0=0m where 0m is found by equating Eq. 14 to zero (see, e.g. Brock, 1986): 2 1/2 t a n Om Kt~+ ~_[(K~ .,_~q 2 4K n L ~KII , ] A (15) The principal stresses corresponding to the two val- ues of 0m are then given as: 1 2 O'1 ' O ' 2 - (27~r)T~ COS ( ~ ) I g I c s ( ~ ) - 3KI, sin ( ~ ) 1 (16) According to the maximum principal tensile stress criterion, the precursor begins to grow when the maxi mum value of the principal tensile stress given by Eq. 16 has the same value as that for growth in an equivalent mode I problem, i.e. equal to K l c / ( / t r ) 1/2. The growth condition is obtained by equat- ing Eq. 16 with this quantity, i.e.: K i c = K , c o s S ( ~ ) - 3 K i t c o s 2 ( ~ - ~ ) s i n ( ~ ) (17) For growth under pure mode I loading, the critical value of the stress intensity is given by: KI = plane strain (18) and: Kic = (2Ey) w: plane stress (19) where y is the grain boundary surface energy Y~b, i f extension occurs along a boundary, facet or the solid- vapor surface energy 7sv, i f the precursor extends into the crystal. Values for these two surface energies have been determined by Ketcham and Hobbs (1969), i.e. 7gb=0.065 J m -2 and 7sv=0.109 J m -2. For exten- sion along the grain boundary Kit is equal to 36.83 K P a m I/2 under plane strain and 34.83 KPa m t/2 36 S. SHYAM SUNDER AND M.S. WU under plane stress with E= 9.33 GPa and v = 0.325. The corresponding numbers for extension into the crystals are 47.69 and 45. l0 KPa m ~/2. Since both surface energies are similar in magnitude (and stresses derived from the two estimates differ only by about 29.5%), precursors on the grain boundary can easily extend into the crystals if there are obsta- cles to their growth on the grain boundary. Typi- cally precursors may grow either parallel or perpen- dicular to the basal plane under these conditions. Although the surface energies are in general ex- pected to be temperature sensitive, experimental data on this variation is unavailable at the present time. Once the crack nucleation criterion is satisfied, the precursors start growing until either (a) the rate of energy supply is inadequate, or (b) they are ar- rested at obstacles such as neighboring junctions or crystals with unfavorable orientations and form a stable microcrack. The microcracks nucleated in this manner generally have a length which is propor- tional to the grain size and which coincides with the wavelength of the microstructural stress field. Cole (1986) has found that the average crack length is slightly smaller than (viz. approximately 0.6 times) the mean grain size. The instability condition for the propagation of microcracks (as opposed to their nucleation) must be based on a driving force supplied by the applied stress field that can overcome the material resis- tance offered by obstacles. This resistance may be characterized in terms of conventional polycrystal fracture toughness measures and is known to be about 2-3 times the surface energy based resistance for ice determined in this paper. In general, there is a transition from the smaller to the greater resis- tance as the length of the nucleated crack increases. Even if the applied stress field is biaxial, it is in general not necessary for the total stress field to be biaxial due to the contribution of the microstruc- rural stress field. However, in the specific case where the c-axes of the grains adjoining the precursor lie in the reference two-dimensional plane the total stress field is biaxial. Analysis of the microstruc- rural stress field under an applied biaxial load shows that out-of-plane s h e a r s a r c n o n e x i s t e n t ( s e e Figs. 18 and 19 in the Appendix), while the out-of-plane normal stress has no effect on a two-dimensional in- plane precursor. The present analysis relates di- rectly to this situation, but is a good approximation for more complex conditions (where the c-axis may be oriented arbitrarily) since the applied rather than the microstructural stresses dominate the total stress field. D I $ C ~ O N OF ~ L P R E D I C T I ON S Cr a c k nucl eat i on under uni axi al t ensi on and compr essi on Cole (1988) has compared the stress levels at which cracks nucleate due to the dislocation pile-up mechanism and the elastic anisotropy mechanism. Figure 4, reproduced from this paper, shows the variation of crack nucleation stress with grain size for the two mechanisms. From this figure he con- 0 . . O3 O3 LU I ' - O3 z 0 t-- < W ..../ o z 2 0 p . = 0 . 6 L 2 a / d = 0 . 1 0 K l C = 0 . 0 3 4 8 M P a e m 1/2 4 = 9 0 ! E L A S T . A N I . M E C H . ( C O M P , ) i ~ ELAST. ANI. MECH. (COLE, 168) ~ \ ' P, LE- UP MECH. (COLE, I ~ ) \ \ \ \ i E L A S T . A N I . M E C H . ( T E N S . ) I I I I 2 4 ~ S 1~ GRAIN SIZE (mm) Fig. 4. Model predi ct i ons o f stress requi red t o nucl eat e t he first crack i n pol ycryst a~i ne i ce as a f unc t i on o f grai n si ze. C R A C K N U C L E A T I O N 37 eludes that the mechanisms cannot be distin- guished on the b a s i s o f e i t h e r the nucleation s t r e s s or the grain size effect. The microstructural models for both mechanisms assume that crack nucleation in tension and compression occur at the same stress level. Predictions of the stress required to nucleate the first crack in uniaxial compression and tension, re- spectively, for the model presented in this paper are also shown in Fig. 4. The calculations are based on E=9. 33 GPa, u=0.325, Kxc=34.8 KPa m 1/2, the ratio of precursor size to grain size equal to 2a/ d=0.10, and the coefficient of friction #=0. 6. The most favorable orientations of the two grains ad- joining the precursor has been adopted for predict- ing the stress required to nucleate the first crack. In particular, both grains have the same c-axis orien- tation and the loading is parallel to the c-axes. Un- der these conditions, there is no property mismatch between the two grains containing the precursor. This is consistent with the findings of Evans (1978) who has shown that the largest microstructural stresses that can develop for a grain pair occur for connecting grains with closely similar orientations. The predicted nucleation stresses and grain size de- pendencies are similar to those of Cole (1988), but there are important differences. The stress required to nucleate the crack in ten- sion is seen to vary in the range of 0.7-2.2 MPa for grain sizes between 1 and 10 mm. Over the same range of grain sizes, Schulson (Schulson et al., 1984; Schulson, 1987) observes nucleation stresses vary- ing between 0.7 and 1.4 MPa. The experimental data were obtained at temperatures varying between - 5 and - 20 C and strain rates of 10 - 6 and l0 -3 s -l. The data show that there is a tendency for the nu- cleation stress to increase with decreasing tempera- ture and increasing rate. Both the magnitude of the nucleation stresses in tension and the grain size de- pendencies predicted by the present formulation are in good agreement with measurements. The predicted stress required to nucleate a crack in compression varies between 1.75 and 5.5 MPa over the same range of grain sizes. Kalifa et al. (1989) have experimentally observed the nuclea- tion stress in uniaxial compression for isotropic po- lycrystalline ice of 5 mm grain size at - 10 o C and a range of strain rates between 10 -5 and l0 -3 s- i. The data show that the nucleation stress varies between 1.9 and 3.0 MPa with a mean value of about 2.5 MPa and that there is no significant effect of strain rate. The mean value of 2.5 MPa at a grain size of 5 mm is almost identical to the model prediction in Fig. 4. More recently, Cole (1989) has observed a nucleation stress varying between 3.3 and 3.9 MPa under almost identical conditions. Although these values are somewhat larger than the model predic- tions, they fall within the limits of inherent statisti- cal and modeling uncertainties. It is expected that experimental data on the nucleation stress as a function of grain size and temperature will become available in the near future. The most important consequence of the model predictions in Fig. 4 is that the nucleation stress in compression is different from that in tension when considering the elastic anisotropy mechanism. In fact, the ratio of the nucleation stresses in compres- sion and tension is independent of grain size (as will become clear later) but varies with the coefficient of friction. As shown in Fig. 5, the ratio lies between 1.7 and 2.8 for 0<#< 1. For the typical value of # = 0.6 considered in this paper, the ratio is equal to 2.5. The orientation of the grain pair with respect to the loading axis has a significant influence on crack nucleation. Defining ~ as the angle between the c- axis of the ice crystal and the z-axis of the coordi- nate system in Fig. 1, the results show that the nu- cleation stress in both compression and tension are Z I.U m [ L : [ 0 4 _o I I 1 r I [ K l C = 0 . 0 3 4 8 MP a - m 1/2 = 9 0 Z 1 F - ~ 0 I I I I I I I I 0 ~ . 0 ~ , 1 0 . 2 0 . 3 0 . 4 - 0 . 5 0 . 6 0 . 9 ( ~ . 8 0 . 9 . 0 Z C O E F F I C I E N T O F F R I C T I O N , p Fig. 5. Rati o o f nucleation stress in uniaxial compressi on to that in uni~ial tensi on plotted as a function of the coefficient of friction. 38 S SHYAM SUNDER AND M S WU u~ u J or p - u~ z k- < W ._J o z 0 0 ',~ \ = 2 7 g = 0 . 6 2 a / d = 0 . 1 0 = 0 K l C = 0 . 0 3 4 8 MPa , , m 1/2 = 9 0 o l I I I 2 4 6 8 1~ GRA I N SI ZE ( mm) Fi g. 6. T h e e f f e c t o f c r y s t a l o r i e n t a t i o n o n c r a c k n u c l e a t i o n i n p o l y c r y s t a l l i n e i ce. funct i ons of this angle (see Fig. 6 ). In bot h t ensi on and compr essi on t he first crack t o nucl eat e is asso- ci at ed wi t h a grain pai r t hat has its c-axis al i gned in t he di r ect i on of loading, i.e. ( = 90 . The results f or al i gnment at 0 are ver y similar, but when t he c- axes are aligned at an angle of 45 t o t he l oadi ng axis t he nucl eat i on stress is a maxi mum. The influ- ence is significant under compr essi on where t he nu- cl eat i on stress ranges bet ween 3.8 and 12.2 MPa compar ed wi t h t he range of 1. 75-5. 5 MPa f or t he most favorabl e ori ent at i on. On t he ot her hand, it appears t hat in general t he effect of c-axis al i gnment is less significant in t ensi on. An i mpor t ant consequence of r andom crystal ori- ent at i on in pol ycryst al l i ne ice is t hat in compres- sion t here can be a di st i nct begi nni ng and end t o t he mi cr ocr ack nucl eat i on phase when t he stress is in- creased and i f fai l ure does not occur premat urel y. In part i cul ar, f or a grain size of 5 mm t he results in Fig. 6 show t hat cracks can nucl eat e over a stress range of 2. 5-5. 5 MPa. Beyond 5.5 MPa fewer cracks will nucleate. Cole ( 1989) has obser ved this phe- nomenon experimentally, where he finds t hat cracks nucl eat e upt o a stress of 5.6 MPa ( compar e wi t h t he pr edi ct i on of 5.5 MPa ) beyond whi ch no mor e cracks nucl eat e unt i l fai l ure occurs at 8.5 MPa, Bot h Kal i fa et al. ( 1989) and Cole ( 1989) fi nd t hat t he fai l ure stress is bet ween t wo t o t hree t i mes t he initial nucl eat i on stress at a st rai n rat e of 10-3 s - I and a grain size of 5 ram. Schulson ( 1987 ) has obser ved compressi ve strengths at a strain rat e of 10 -1 s -1 varyi ng bet ween 4 and 6.2 MPa for grain sizes in t he range of 1. 5-10 mm and a t emper at ur e of - 10 C. These dat a suggest t hat t he fai l ure stress is about 1. 3-2. 3 t i mes t he initial nucl eat i on stress. As such, it can be concl uded t hat initial crack nu- cl eat i on is in general not strength l i mi t i ng in compr essi on and t hat addi t i onal crack nucl eat i on will occur pr i or t o fai l ure as t he stress is increased. Ef f ect of f ri ct i on, precursor si ze and cryst al ani sot ropy on cr ack nucl eat i on The i nfl uence of t he fri ct i on coeffi ci ent on t he nucl eat i on stress is shown in Fig. 7. As expect ed, t he tensile nucl eat i on stress does not depend o n / t while t he compressi ve stress can change by about 0. 8- 2. 5 MPa as t he grain size decreases. Schulson ( 1987) has used a fri ct i on coeffi ci ent of 0.6 t o pre- di ct t he pr opagat i on- cont r ol l ed compressi ve strength of polycrystalline ice at a strain rate of 10-3 s - 1. For a strain rat e of 10- l s - t, he argues t hat t he coeffi ci ent of sliding fri ct i on must be about 15% l ower or 0.51 on t he basis of research on ice f r i ct i on by Barnes et al. ( 1971) . Thi s val ue is close t o t he reference val ue of / ~=0. 6 adopt ed in this paper. In any case, t he var i at i on in model predi ct i ons for a coeffi ci ent bet ween 0.51 and 0,6 is negligible. The rat i o of t he pr ecur sor size t o t he grain size also has an effect on t he nucl eat i on stress. Fi gure 8 compar es t he pr edi ct i ons for t he reference rat i o of 0.10 and a val ue of 2a/d-- 0.05. The smal l er precur- sor size leads t o an i ncrease in nucl eat i on stress, by appr oxi mat el y 0. 3- 0. 8 MPa in t ensi on and 0. 7- 2 MPa in compressi on. The larger increases occur at grai n sizes close t o 1 mm as woul d be expect ed f r om an i nverse square-root dependency on precursor size. The pr ecur sor or i ent at i on fl and t he i nci pi ent CRACK NUCLEATION 7 (L u) o9 Lu 4 n- t - O9 z _o 3 F- < ..I ~ 2 z 0 2 4- 6 8 ]0 GRAIN SIZE (mm) Fig, 7. The effect of f r i ct i on on model pr edi ct i ons of t he stress r equi r ed t o nucl eate the f i rst crack i n pol ycr yst al l i ne ice. growt h di r ect i on 9 f or g = 0 . 6 are shown as func- t i ons of the c-axis al i gnment i n Fi g. 9. I n tensi on, f l i s appr oxi mat el y equal t o 70 whi l e 0 i s around - 4 0 f or the opt i mum or i ent at i on correspondi ng t o ~=90 . These results are consi stent wi t h the analysis o f Ma i t i and Smi t h ( 1983) as shown i n t hei r fig. 4. However, i n compression f l i s close t o 30 whi l e 0 i s about 70.5 i n the most favorabl e or i - ent at i on. Once again, these results are i dent i cal t o the values predi ct ed by Ashby and Hal l am ( 1986) f or the chosen f r i ct i on coeffi ci ent. The same fi gure also shows t hat bot h the opt i mal precursor or i ent at i on and the i nci pi ent growt h di - rect i on are r el at i vel y i nsensi t i ve t o the or i ent at i on o f the basal plane wi t h the l oadi ng axis. Conse- quent l y, all the cracks whi ch nucleate as the stress i s increased wi l l possess ver y si mi l ar ori ent at i ons, i.e. damage accumul ati on i s hi ghl y ani sotropi c ( t hi s statement assumes t hat the precursor or i ent at i on i s not an i ndependent l y cont rol l ed var i abl e) . Cole ( 1986) has experi ment al l y observed t hat i n uni ax- i al compression the cracks t end t o cl uster about the 39 i 03 03 LU rr t - z _o k- W z ] I 2a/d - 0.05, TENS. o i 2ai d - 0.10, T E N S . , 0 2 4 I I I I ~ = 0.6 Kl C = 0.0348 MPa=m 1/2 = 9 0 2a/d = 0.05, COMP. 2a/d ,,, 0.10, COMP. t I I 6 8 ]0 GRAIN SIZE (mm) Fig. 8. The effect of t he rat i o of precursor size to grain size on model predi ct i ons of t he stress requi red to nucleate t he first crack in polycrystalline ice. l oadi ng axis wi t h an average angle of 23 and a st andar d devi at i on of 17.5 . The dat a also show a mode in t he range of 10-15 (see fig. 25 of t he ref- er enced r epor t ) . Bot h t he average and t he mode are consi st ent wi t h t he opt i mal pr ecur sor or i ent at i on whi ch vari es bet ween 23- 33 f or di f f er ent crystal or i ent at i ons and t he effect of t he i nci pi ent growt h di r ect i on whi ch t ends t o r educe t he average angle of t he compl et el y f or med mi crocrack. A nor mal i zed stress measur e acr(r~a)~/2/K~c can he def i ned and pl ot t ed as a f unct i on of t he precur- sor or i ent at i on fl f or vari ous val ues of t he fri ct i on coeffi ci ent as shown in Fig. 10. Not e t hat f or a gi ven fl it is in general not necessary f or t he val ue of whi ch yields t he smallest nucl eat i on stress t o be equal t o 90. The opt i mum pr ecur sor or i ent at i on f or nucl eat i on of t he first crack cor r esponds t o a mi ni mum of t he nor mal i zed stress measure. In t he case of compressi on, t he opt i mum fl is clearly i den- tifiable and vari es in t he range of 30- 52 depend- ing on t he fri ct i on coeffi ci ent . However , in t he case 4 0 S. SHYAM SUNDER AND M,S, WU J (.9 z < n - O 6`3 ~0 0E 5`3 O 4`3 UJ O : 3`3 (..) 1~ 9`3 I 8'3 7 0 w F - o U N I A X I A L L O A D I N G , p, = 0 . 6 I [ I I I I T E N S I ON J C OMP R E S S I ON I t I I I I I I 1@ 2 0 39 4.0 .%0 6 0 "70 8,~ 9Q ORI ENT AT I ON OF BASAL PLANE, u. / ,-,I z < hE 0 < r r " O 813 6 0 4-0 2 0 (3 - 2 0 -4-0 - 6 0 - 8 0 ] . ~ 0 I I I I ] I I I C OMP R E S S I ON Z - T E N S I ON W ~" I l I I J l I I Z Fig. 9. The effect of crystal alignment on optimal precursor orientation and incipient growth direction under uniaxial loading. o f t e n s i o n t he mi n i mu m oc c ur s a r o u n d / ~ = 70 o i n- de pe nde nt o f f ri ct i on but t he no r ma l i z e d stress i s al mos t c o mpl e t e l y i ns e ns i t i v e t o precurs or ori ent a- t i o n ove r a wi de range o f angl es ( b e t we e n 50 a nd 90 ). It i s i mpo r t a nt t o n o t e t hat t he no r ma l i z e d stress i s i n d e p e n d e n t o f precurs or ( o r gr ai n) s i ze and t he cri t i cal stress i nt e ns i t y f act or. Cons e - que nt l y, t he rat i o o f unnormalized stresses i n c o mpr e s s i o n a n d t e n s i o n wi l l be a c o ns t a nt , e qual t o t he rat i o o f no r ma l i z e d s t res s es i f t he s e vari abl es are t he s ame. Fi nal l y, t he ef f ect s o f el as t i c a ni s o t r o py o n t he nuc l e at i on stresses are s ho wn i n Fi gs. 11 a nd 12. The fi rst f i gure s ho ws mo d e l pr e di c t i o ns o f nuc l e a t i o n stress as a f u n c t i o n o f grai n s i z e bo t h i nc l udi ng a nd e xc l udi ng t he mi cros t ruct ural s ~ i n d u c e d by el as t i c a ni s o t r o py ( g i v e n i n Eq. 5 ) . Th e ef f ect o f O E F - Z W F- Z O9 O9 W OC I - O9 0 W _N _I < CC o Z 20 ]5 10 K1 C = 0 . 0 3 4 8 MP a e m 1/ 2 [ [ ] I I COMPRESSION 0. 6 I / t I ~t=O I I I I I I I 1 0 10 20 30 40 50 ~0 70 80 PRECURSOR ORIENTATION, D Fig. 10. Normalized crack nucleation stress under uniaxial loading plotted as a function of precursor orientation for varying values of friction coefficient. el ast i c a ni s o t r o py i s qui t e s mal l i n t he case o f t en- s i on, c aus i ng a r e duc t i o n o f a bo ut 15% i n t he nu- c l e at i on stress. Howe ve r , i ts ef f ect i s s i gni f i c ant i n t he case o f c o mpr e s s i o n, where t he n u c l e a t i o n stress decreas es by a bo ut 33%. Res ul t s no t pr e s e nt e d i n t he f i gure i ndi c at e t hat e v e n wh e n t he c oe f f i c i e nt o f f ri ct i on i s s et t o zero t here i s a di f f erence o f a bo ut 25%. I n al l cas es t he mi cros t ruct ural stresses bel p t o accel erat e crack nuc l e a t i o n. Fi gure 12 s ho ws t he rat i o o f nuc l e a t i o n stress i n c o mpr e s s i o n a nd t e n s i o n pl ot t e d agai ns t f ri ct i on. In general , t he rat i o i s greater wh e n a ni s o t r o py i s ex- c l ude d t ha n wh e n i t i s i nc l ude d as i n Fig. 5. For t he ref erence f r i c t i on v a l ue o f 0. 6, t he rat i o o f stresses i s a ppr o x i ma t e l y 3. 15 ( c o mp a r e wi t h 2. 5 wh e n an- i s ot r opy i s i nc l ude d ). Cc ==k ~ u ml mr I WI dI d t m ~ B gl=Wd The crack nuc l e a t i o n surf ace i n t he cas e o f an ap- pl i e d bi axi al stress f i el d i s s h o wn i n Fig. 13 as a CRACK NUCLEATION 41 5 O3 6 O ,,, 4 rr I . - O3 Z o 3 I.- W . J ~ 2 z 0 0 ELAS. ANI.: PARAMETERS A8 IN FIG. 4 I I ~ I 1 i wiTHOUT EaS. ANI. IC~MP.I \ \ \ ~ ~ - ~ W I T H E L / ~ WITH ELAS. ANI. (TENS.) I I I r 2 4 6 8 10 GRAIN SIZE (mm) Fig. 11. The effect o f excl udi ng mi cr ost r uct ur al stresses i n- duced by elastic ani sot ropy on model pr edi ct i on o f the stress requi red t o nucleate the f i r st crack i n pol ycr yst al l i n ice. Z LU n o o O3 O3 LLI n-" t-- U~ Z o ,< =, ' 0 Z ANALYSIS WITHOUT ELASTIC ANISOTROPY I I I i I I I i 4- K1C = 0.0348 MPatrn 1/2 2 3 - ' . . . . . . . . . . . . . . . i (~ I I I I I I I I ~ . ~ 0 . 1. ~ . 2 0 . 3 ~ . 0 . ~ 0 . 6 ( ~ . ' 7 ~ . 8 Q . 9 t . COEFFICIENT OF FRICTION, Fig. 12. The effect of excl udi ng mi cr ost r uct ur al stresses i n- duced by el ast i c ani sot r opy on t he r at i o of nucl eat i on st ress i n uni axi al compr essi on t o t hat i n uni axi al t ensi on. f unct i on of grai n size. The analysis assumes t hat crack nucl eat i on occurs onl y in t he pl ane of loading. The results show t hat t he dependence on stress st at e is negligible under bi axi al t ensi on. However , under (9 z 0 0 i i i i i i > Z I'-- 0 z 14 = 0.6 3 I I I I 'f ,-' t - ~xx (MPa) 0 j ~ ' [ I I j ~. - - " ~l l C _ ~ - [ t " 3 - MPTS C R I T E R I O N - ~ -4 -6 J I i / / " , ~Y -6 -5 - -3 2a/d = 0.10 Kl c = 0.0348 MPaem 1/2 :J -2 - t 0 ] 2 ELASTIC ANISOTROPY ANALYSIS Fig. 13. Crack nucl eat i on surface unde r a bi axi al st ress fi el d as a f unct i on of grai n size. b i a x i a l c o m p r e s s i o n , i n - p l a n e c r a c k n u c l e a t i o n m a y b c s u p p r e s s e d c v c n u n d e r a m o d e r a t e b i a x i a l i t y i n t h e s t a t e o f stress. T h e n u c l e a t i o n s u r f a c e c o n t r a c t s a s t h e g r a i n s i z e i n c r e a s e s , a l t h o u g h t h e g e n e r a l s h a p e o f t h e s u r f a c e is p r e s e r v e d . T h e c r a c k n u c l e a t i o n s u r f a c e a s a f u n c t i o n o f fric- t i o n is s h o w n i n F i g . 1 4 . A s e x p e c t e d , f r i c t i o n p l a y s a s i g n i f i c a n t r o l e u n d e r s t a t e s o f s t r e s s i n v o l v i n g c o m p r e s s i o n b u t h a s n e g l i g i b l e e f f e c t u n d e r p r e - d o m i n a n t l y t e n s i l e s t a t e s o f stress. T h e v a r i a t i o n o f t h e o p t i m u m p r e c u r s o r o r i e n t a - t i o n a n d t h e i n c i p i e n t g r o w t h d i r e c t i o n w i t h c r y s t a l a l i g n m e n t is s h o w n i n F i g . 1 5 f o r r e p r e s e n t a t i v e s t a t e s o f b i a x i a l stress. T h e f i g u r e s h o w s t h a t a s t a t e o f s t r e s s i n v o l v i n g t e n s i o n t e n d s t o y i e l d r e s u l t s t h a t a r c s i m i l a r t o t h o s e i n u n i a x i a l t e n s i o n a n d a s t a t e o f s t r e s s i n v o l v i n g p u r e c o m p r e s s i o n y i e l d s r e s u l t s c l o s e r t o t h a t i n u n i a x i a l c o m p r e s s i o n . I n all c a s e s , t h e d o m i n a n t p r i n c i p a l a p p l i e d s t r e s s t e n d s t o g o v - e r n t h e p r e d i c t e d a n g l e s . T h e m o d e l p r e d i c t i o n s a r e a l s o c o m p a r e d w i t h t h e l i m i t i n g t e n s i l e s t r a i n ( L T S ) c r i t e r i o n f o r c r a c k n u - c l e a t i o n p r o p o s e d b y S h y a m S u n d e r a n d T i n g ( 1 9 8 5 ). A c c o r d i n g t o t h i s p h e n o m c n o l o g i c a l ' c r i t c - r i o n , c r a c k n u c l e a t i o n o c c u r s w h e n t h e m a x i m u m v a l u e o f t h e p r i n c i p a l ( m a c r o s c o p i c ) s t r a i n s is t e n - 42 S. SHYAM SUNDER AND M.S. WU 0 r r 0 Z 2a/ d = 0. 10 d = 5 mm K l C = 0. 0348 MPa e m 1/2 3 I i I I 1 3 I [ I 6 x x l ( MPa ) I t o -I - 2 / / -3 / ~"e I{ Y -5 - 6 " F I I I 61 I I -6 -5 - - 3 -2 - [ g l 2 EL AST I C A NI S OT ROP Y A NA L Y S I S Fi g. 14. Cr a c k n u c l e a t i o n s ur f a c e u n d e r a bi axi al s t r e s s f i el d as a f u n c t i o n o f t h e f r i c t i on coef f i ci ent . sile and its magni t ude at t ai ns a critical value. The critical st rai n is t he nucl eat i on stress in uni axi al t ensi on di vi ded by t he Young' s modul us. Appl i cat i on of this cr i t er i on f or uni axi al compressi on suggests t hat t he nucl eat i on stress in compr essi on is equal t o t hat in t ensi on di vi ded by Poi sson' s ratio, since under this compressi ve stress t he lateral tensile st rai n at t ai ns t he critical value. Thus t he l i mi t i ng tensile st rai n cri t eri on predi ct s t hat t he nucl eat i on stress in compr essi on is about 1/ 0. 325 or 3.08 t i mes t he nucl eat i on st ress in ten- sion. Thi s is somewhat larger t han t he pr edi ct ed val ue of 2 5 when ani sot r opy is i ncl uded (Fig. 5) but al most i dent i cal t o t he val ue of about 3.15 when crystal ani sot r opy effects are excl uded (Fig. 12 ). The use of t he l i mi t i ng tensile st rai n cri t eri on based on t he Poi sson' s rat i o f or an i sot ropi c solid is somewhat ambi guous since it cannot di st i ngui sh whet her t he const i t uent crystals are i sot ropi c or an- isotropic. The results suggest t hat t he l i mi t i ng t en- sile st rai n in uni axi al compr essi on is smal l er t han t hat under t ensi on in t he presence of t he mi cros- t ruct ural stress field. Consequent l y, it is possible t o defi ne a new par amet er called t he "l at er al st r ai n ra- t i o for crack nucl eat i on ( LSRCN) " and assigned t he symbol p, whi ch is si mi l ar in concept t o t he Pois- w" . J Z < n- O 03 n" O uJ n- o_ . J < O 0 BIAXl AL LOADING, I a = 0.6 90 80 70 60 50 4~ 39 29 19 0 ' ' ~ I I I / I - T - C, X = - 0 . 2 f / J C- C, ; L= 0.2 I l I I I I I I @ 10 2 0 3 0 4 0 5 0 6 0 7 0 8 0 9 0 ORIENTATION OF BASAL PLANE, 1 o o ( D 6 e Z 4 ~ < 0 rr -2e O -4e F- Z - 6 0 uJ - 8 0 - 1 O O O Z I I I I I I I C- C, X = 0 . 2 T - T , X= 0 . 2 T - C, X = -0.2 10 2 0 3 0 ~ 0 5 0 6 0 7 0 8 ~ 9 0 Fig. 15. The effect of crystal alignment on optimal precursor orientation and incipient growth direction under biaxial loading (Note: 2=tr=/a~x). son' s rat i o v for def or mat i on. Thi s par amet er is equal t o 1/ 3. 08 or 0. 325 when crystal ani sot r opy ef- fects are excl uded and equal t o 1/ 2. 5 or 0. 40 when t hey are i ncl uded. Figure 16 compar es t he pr edi ct ed nucl eat i on sur- face wi t h t hat based on t he l i mi t i ng tensile st rai n cri t eri on correspondi ng t o a grain size of 5 mm and # = 0.6. The cur ve based on p = 0.325 t ends t o over- pr edi ct t he nucl eat i on surface, while t he use of p = 0. 40 pr ovi des an excel l ent mat ch under states of stress i nvol vi ng compressi on, i.e. biaxial compres- sion and t ensi on- compr essi on. Unde r bi axi al t en- sion, bot h model s over pr edi ct t he nucl eat i on sur- face. Under this stress state it is preferabl e t o i gnore biaxiality and use a si mpl e cut - of f equal t o t he nu- cl eat i on stress in uni axi al t ensi on. Fi gnre 17 con- fi rms t hat when t he effect s of crystal ani sot r opy are C R A C K N U C L E A T I O N 4 3 (.9 Z 0 n - O UJ t r W Z < r r I - - o z 2 a / d = 0 . 1 0 d = 5 m m 3 i i 2 [ -1. - 2 - 3 -4- - 5 - 6 - 6 p . = 0 . 6 K l C = 0 . 0 3 4 8 M P a , m 1 / 2 i i i .J i i a =-.2- - , I I,~J~1~"-,~ " / ' - , ~ x I ( M P ~ - 2 - [ . t I _ / 2 _~ LTSC (p- o.4) -2 - M P T S C R I T E R I O N . " / - - I l I I / V ~ " . r ~ I I - 5 -4- - 3 - 2 - [ 0 1 2 C~ _z Y 0 < m o W t r u J > Z < r r I - - 0 Z : 3 - 1 [ - - -4- - 5 - 6 J I I J - 6 -5 - 4 - 3 - 2 E L A S T I C A N I S O T R O P Y A N A L Y S I S Fig. 16. Comparison of the predicted nucleation surface un- der biaxial loading with the phenomenological limiting ten- sile-strain criterion of Shyam Sunder and Ting (1985). i 2 - t TSC (p..o~) & t-'xx(MPa) ~ ' ' ' ' ' ' ' ~' ~F - I/ 0 i i - - " ~ T I I ~ J Y I - 2 - t 2 - [ - 2 - 3 - I I -t 0 1 MPTS CRITERION (WITHOUT ELASTIC ANI~K)TROPY) Fig. 17. Comparison of the predicted nucleation surface ex- cluding microstructural stresses induced by crystal anisot- row with the phenomenological limiting tensile strain crite- rion of Shyam Sunder and Ting ( 1985 ). excluded the use ofp=0. 325 provides an excellent match under biaxial compression and tension- compression. Once again, a simple tension cut-off is adequate in biaxial tension. The use of the limiting tensile-strain criterion for crack nucleation with p=0. 40 instead of the Pois- son's ratio also provides an excellent match to the nucleation surface obtained experimentally by Kal- ifa et al. (1989) under cylindrical triaxial compres- sion states of stress (see Eq. 22 and fig. 6 of the ref- erenced paper). C O N C L U S I O N S This paper has presented'a theoretical analysis of crack nucleation due to elastic anisotropy in poly- crystalline ice. The singularity of the associated stress concentration near a grain-boundary facet junction provides the mechanism for inducing mi- crocrack precursors, if similar nuclei do not already exist. The first-order microstructural stress field generated by the elastic anisotropy of individual crystals in an otherwise isotropic polycrystal is lin- early superposed on the applied stress field. This to- tal stress field is used in the analysis of the stress required to nucleate microcracks and the incipient growth direction based on a solution to the general problem of an extending precursor in a combined stress field. The local material resistance is characterized in terms of a critical value for the maximum principal tensile stress which can be determined from the critical mode I stress intensity factor; this in turn is related to the surface energies of either the grain boundary or the solid-vapor interface. The effect of Coulombic frictional resistance is included in the analysis and it is postulated that the first crack to nucleate is associated with the most favorable ori- entations of both the precursor and the adjoining grains. On the basis of the model predictions, the follow- ing specific conclusions can be drawn: (a) The stress required to nucleate the first crack under uniaxial compression is different from that under uniaxial tension. The typical model parame- ters considered here predicts this ratio to be about 2.5 for polycrystalline ice. A review of the recent ex- perimental data of Kalifa et al. (1989) and Cole (1989) support this stress ratio. The simple model 44 S. SHYAM SUNDER AND M.S. WU of Cole (1989) for the elastic anisotropy mecha- nism cannot predict this result. (b) The nucleation stress in tension is smaller than that for the dislocation pile-up mechanism as has also been predicted by Cole (1988), but it is significantly larger in compression. Thus in general the elastic anisotropy mechanism is expected to control crack nucleation at the high end of the quasi- static loading regime and low temperatures where the contribution of dislocation creep is a minimum. At lower rates of loading and higher temperatures it must compete with cracks that nucleate due to elas- tic stress concentrations stemming from the creep anisotropy of the constituent ice crystals. (c) The range of values over which the nuclea- tion stress varies in tension as a function of grain size is in good agreement with the experimental data of Schulson (Schulson et al., 1984; Schulson, 1987 ) at strain rates of 10 -6 and 10 -3 s -t. This finding supports the hypothesis by Cole ( 1988 ) in the case of tensile loading who states that the first cracks to nucleate are not a result of the dislocation pile-up mechanism but rather a result of the elastic aniso- tropy mechanism. (d) The stress required to nucleate the first crack in compression is almost identical to the mean value of the nucleation stress obtained experimentally by Kalifa et al. (1989) and somewhat smaller (though within the limits of statistical and modeling uncer- tainties) than that obtained experimentally by Cole (1989). Both data are for isotropic polycrystalline ice with a grain size of 5 mm. (e) The stress required to nucleate a crack in compression is strongly dependent on crystal ori- entation. As a consequence of the random orienta- tion of crystals in isotropic polycrystalline ice, there can be a distinct beginning and end to the micro- crack nucleation phase when stress is increased and if failure does not occur prematurely. This phenom- enon has been experimentally observed by Cole (1989) in tests conducted on similar ice with a grain size of 5 mm and a strain rate of 10- 3 s- 1. The pre- dicted and experimentally observed maximum nu- cleation stress are almost identical. (f) Crystal orientation has little influence in ten- sion. This is particularly true in the typical case where nucleation of the first crack causes failure in tension. (g) The friction coefficient has little effect in ten- sion but significantly influences the nucleation stress in compression. (h) The microstructural stress field induced by elastic anisotropy helps to reduce the stress that must be applied to nucleate cracks. Once again, the effect in compression is significantly more than that in tension. (i) The nucleation surface as predicted by the phenomenological limiting tensile strain criterion of Shyam Sunder and Ting (1985) agrees well with model predictions for states of stress involving compression. Under purely tensile states of stress, a simple tension cut-off based on the nucleation stress in uniaxial tension is adequate since the effects of multiaxiality are negligible. A generalization of the original limiting tensile strain criterion based on the concept of a "lateral strain ratio for crack nuclea- tion, p" (similar to Poisson's ratio for deforma- tion) has been proposed to distinguish whether the constituent crystals are isotropic or anisotropic. The predictions of this generalized phenomenological model are in excellent agreement with the nuclea- tion surface obtained experimentally by Kalifa et al. (1989) under cylindrical triaxial compression states of stress. (j) The optimum precursor orientation and in- cipient growth direction are relatively insensitive to the effects of crystal orientation. The dominant principal applied stress tends to govern the pre- dicted values for both angles, This suggests that, if precursor orientation is not an independently con- trolled variable, average microcrack directions will tend to cluster about a value which depends purely on the state of stress, i.e. for a given state of stress damage will tend to be highly anisotropic. ACKNOWLEDGE~NTS The authors would like to acknowledge financial support from AMOCO, ARCO, BP America, CHEVRON, CONOCO, EXXON and MOBIL through the MIT Center for Scientific Excellence in Offshore Engineering, and the U.S. Department of the Interior, Minerals Management Service. Addi- tional funds were made available by the Henry L. Doherty Professorship in Ocean Utilization CRACK NUCLEATION 4 5 awarded to the first author. The thoughtful review comments of David Cole and John Dempsey are most appreciated. A P P E N D I X Ro t a t i o n M a t r i x R For the chosen Cartesian coordinate system {x, y, z} and the standard frame of reference for the hex- agonal ice crystal {p, q, r }, the direction cosines be- tween the axes may be defined as: x y z P I i ~v~ 1 n] q l z m2 n 2 r 13 m2 n3 The rotation matrix R, appropriate for transform- ing the compliance matrix from the {p, q, r} frame to the {x, y, z} frame is then given by: /22 m22 n2 2 2m2n2 2n2/2 212m2 /3 2 m 3 2 /13 2 2 m 3 n 3 2 n 3 1 3 2 1 3 m 3 1213 m2m3 n2n3 m2n3+m3n2 n213+n312 12m3+13m2 13l] mam[ n3n~ m3n]+m~n3 nal~+n~13 13m~+l~m3 1212 rnlm2 n]n2 mln2+m2n] nll2+n2ll llm2+12ml Denoting Rs by: where A, B, C and D are 3 X 3 submatrices, the ro- tation matrix Rc appropriate for transforming the stiffness matrix from the {p, q, r} frame to the {x, y, z} frame is given by: El a s t i c c o n s t a n t s o f s i ngl e c r y s t a l a n d p o l y c r y s t a l The stiffness and compliance matrices are de- rived from the data of Gammon et al. (1983) for the temperature of - 16 C. For other temperatures Z ,,( O9 5 W 0 l.- W 0 uJ cO W O C : l - d 0.4- 0.3 0.2 0.[ 0.0 -0.1 -0.2 -0.3 -0 . UNIAXIAL COMPRESSION: Oxx = -1 M P a n ~ 0.4 Z 0.3 < 0.2 5 t,.U O.t 0 I.- 0. @ w :D -0.1 o9 U.l -0.2 (/) ILl n- -0.3 1- 03 (~ -0.4- [ I I I I I I I ( ~O , Z Z I ~ O , X X J I I I I J I I 0 20 40 60 80 100 120 140 180 180 ORIENTATION OF BASAL PLANE, 1 I l l I I I l ~ r o , x z ao, yz ~o,xy I I I I I I I I 0 ~ 0 60 80 100 120 140 180 IB0 Fig. 18. Variation of average microstructural stresses with crystal orientation under uniaxial compression. in the range of - 5 to - 4 0 C, the equation sug- gested by Gammon et al. (1983) may be used: ( 1 - a T ) X( T) =X( To)( 1 ~ a T o ~ where X represents an elastic constant at the tem- perature T (in C), To is the temperature (in C) at which a measure of T is known, and the parame- ter'a' has the value 1.418X 10 -3 ( C) -~ C o mp l i a n c e ma t r i x of si ngl e i ce c r y s t a l At - 16 C the s i n g l e c r y s t a l compliance matrix Sp is, in ( GPa) - l , given by: 46 S. SHYAM SUNDER AND M,S. WU -0.10318 -0.04287 -0. 02316 0 0 0 0.10318 -0. 02316 0 O 0 0.08441 0 0 0 0.33179 0 0 SYM. 0.33179 0 0.29210 Compl i ance mat r i x f or isotropic pol ycrystal At - 16 C the i sotropi c polycrystal compl i ance matrix S. , is, in ( GPa ) -~, gi ven by: BI AXI AL COMPRESSI ON- TENSI ON: axx = "(~zz = -1 MPa E t ~ 0 0. 3 ~: ~. 2 5 0 tad - 0. t ~O iii ..0 -~ ,2 ILl - ~ . 3 0 9 n- _o n ~.4- I Z 0 , 3 . < ~ ~. 2 ..J UA ~. t O UA 3 - ; ~ . t { , 9 Idd CO -G . 2 O3 u J E - 0 . 3 ( : j - 0 . 4 { L I I I I I I " ' , ~ 0 ZZ / . / \ , , ' ' , , ' t , ( ~ o , y y , / ' 0 " 0 , X X ~ J ] t O 2 0 4~3 6 C 8 0 100 ] 2 0 t t 1 4 0 1 6 0 1 8 0 ORI ENTATI ON OF BASAL PLANE, L / f - - . \ ' o , x z 1 I I L I / ' , , l 0- 0, \ , /I / - . / . . . . % y~ { I I L . [ I [ ~ 1 i I 6 20 4 0 6 0 8 0 1 0 ~ 1 8 0 1 4 0 1 6 0 1 8 0 Fig. 19. Variation of average mi crost ruct ural stresses with crystal ori ent at i on under a equi-biaxial t ensi on- compr essi on state of stress. I 0.10716 -0. 03486 -0.03486 0 0 0 0.10716 -0. 03486 0 0 0 0.10716 0 0 0 0.28401 0 0 SYM, 0.28401 0 0.28401 S t i f f n e s s ma t r i x o f s i n g l e i c e c r y s t a l At - 16 C the single crystal stiffness matrix Cg~ is, in GPa, gi ven by: 13.929 7.082 5.765 0 0 0 13.929 5.765 0 0 0 15,010 0 0 0 3.014 0 0 SYM. 3,014 0 3.4235 .J Mi cr ost r uct ur al st r esses due t o el ast i c ani sot r opy The microstructural stresses due to uniaxial compressi on and biaxial compres s i on- t ens i on are comput ed from Eq. 5 and shown in Figs. 18 and 19, respectively. REFERENCES Ashby, M.F. and Hatlam, S.D., 1986. The failure of brittle solids cont ai ni ng small cracks under compressi ve stress states. Acta Metall., 34(3 ): 497-510. Barnes, P., Tabor, D. and Walker, J.F.C., 1971. The friction and creep of polycrystalline ice. Proc. R. Soc. London, A324: 127-155. Bogy, D.B. and Wang, K.C., 1971. Stress singularities at in- terface corners in bonded dissimilar isotropic elastic ma- terials. Int. J. Solids Struct., 7: 993-1005. Brock, D., 1986. El ement ary Engineering Fracture Mechan- ics. Martinus Nijhoff, Boston, Mass. Cole, D. M. , 1986. Effect of grain size on the internal fractur- ing of polycrystalline ice. CRREL Rep., 86-5, U 3. Army Cold Reg. Res. Eng. Lab., Hanover, N. H. , 79 pp. Cole, D.M., 1988. Crack nucleation in polycrystalline ice. Cold Reg. Sci. Technol., 15( 1 ): 79-87. Cole, D. M. , 1989. Micxofracture and t he compressi ve failure of polycrystalline ice. Proc. 1UTAM/ I AHR Syrup, Ice- Structure Interaction. St. John' s, Nfld., Preprint. Erdogan, F. and Sih, G. C. , 1963. On t he crack extension in plates under plane l oadi ng and transverse shear. J; Basic. Eng., 85: 519-527. Eshelby, J. D. , 1957. The det er mi nat i on of the elastic field of CRACK NUCLEATION 47 an ellipsoidal inclusion, and related problems. Proc. R. Soc., London, A241: 376-396. Evans, A.G., 1978. Microfracture from thermal expansion anisotropy - I. Single phase systems. Acta Metall., 26: 1845-1853. Evans, A.G., 1984. Fracture in Ceramic Materials: Toughen- ing Mechanisms, Machining Damage, Shock, Noyes Publ., Park Ridge, N.J. Gammon, P.H., Kiefte, H., Clouter, M.J. and Denner, W.W., 1983.Elastic constants of artificial and natural ice sam- ples by Brillouin spectroscopy. J. Glaciol., 29 (103 ): 433- 459. Gold, L.W., 1972. The process of failure of columnar-grained ice. Philos. Mag. A, 26(2): 311-328. Kalifa, P., Duval, P. and Ricard, M., 1989. Crack nucleation in polycrystaUine ice under compressive stress states. Proc. Eighth Int. Conf. Offshore Mechanics and Arctic Engi- neering, The Hague, Netherlands, IV: 13-21. Ketcham, W.M. and Hobbs, P.V., 1969. An experimental de- termination of the surface energies of ice. Philos. Mag. A, 19:1161-1173. Maiti, S.K. and Smith, R.A., 1983. Comparison of the crite- ria for mixed mode brittle fracture based on the preinsta- bility stress-strain field. Part 1: Slit and elliptical cracks under uniaxial tensile loading. Int. J. Fract., 23:281-295. Schulson, E.M., 1979. An analysis of the brittle to ductile transition in polycrystalline ice under tension. Cold Reg. Sci. Technol., 1 ( 1 ): 87-91. Schulson, E.M., 1987. The fracture of ice Ih. J. Physique, Col- loque C1, Suppl. au no. 3, Tome 48: 207-218. Schulson, E.M., Lim, P.N. and Lee, R.W., 1984. A brittle to ductile transition in ice under tension, Philos. Mag. A, 49(3): 353-363. Shyam Sunder, S. and Ting, S.-K., 1985. Ductile to brittle transition in sea ice under uniaxial loading. Proc. 8th Int. Conf. Port and Ocean Engineering under Arctic Condi- tions, Narssarssuaq, Greenland, 2: 656-666. Sih, G.C., 1973. Handbook of Stress Intensity Factors. Le- high Univ. Press. Sih, G.C., 1974. Strain-energy-density factor applied to mixed mode crack problems. Int. J. Fract., 19(3): 305-321. Sinha, N.K., 1978. Observations of basal dislocations in ice by etching and replicating. J. G1aciol., 21 (85): 385-395. Sinha, N.K., 1979. Grain-boundary sliding in polycrystalline materials. Philos. Mag. A, 40(6): 825-842. Sinha, N.K., 1982. Delayed elastic strain criterion for first cracks in ice. Proc. IUTAM Conf. Deformation and Fail- ure of Granular Materials, Balkema, Rotterdam, pp. 323- 330. Sinha, N.K., 1984. Intercrystalline cracking, grain-boundary sliding, and delayed elasticity at high temperatures. J. Mat. Sci., 19: 359-376. Wu, M.S. and Shyam Sunder, S., 1990. Crack nucleation in polycrystalline ice - a comparison of two nucleation cri- teria. Proc. Ninth Int. Conf. Offshore Mechanics and Arc- tic Engineering. Houston, Tex. Febman, 8-23, Preprint, 28 pp.