1989 and 1990 from the Arctic Ocean. The same type of .... Pieces of ice can be seen flying in the air quite some distance from the point of contact. Sam.ple ... a cooler, and was stored in a freezer for 3 days before being transported to Memorial.
J OllTllalofGlaci%gy, Vol. +5, Ao. 151, 1999
Microstructural change in ice: Ill. Observations from. an iceberg itnpact zone K.].
M U G GE RID GE,·
1.]. JORDAAN
Ocean E ngineering Research Cell /re, FaCllL£J' d Engineering and Ajlplied Science, l\JemoriaL L'l1il'frsily d. '\fewjolllldLand, StJ ohn's, Nwfo ul1dland AlE 3X5, Canada
ABSTRACT. During a full- scale iceberg-impac t slUdy conducted inJuly 1995 o n the L abrador coas t, C anada, a sampl e of ice was retri eved from th e impac ted surface o f a n iceberg. Th e sample was thin- sec tioned a nd the obsen 'ations of th e co ntact-zone microstructure a rc presented in thi s p a p er. Thin sec tio ns were prepa red fr om two slabs c ut pa rall el to th e impacted surface. In each of th e thin ections ta ken fro m the impac ted-surface slab, fin e-g ra ined materia l was fo und to surro und pa rent- size g r a in s (as obse f\ T d in the seco nd slab ). A bo und a r y between the p a re nt gra ins a nd the g rains of modifi ed microstruc ture was found r unning a pproxim ately pa rallclto the impac ted surface in each of"th e thin sec ti o ns taken fr om th e impac ted-surface slab. Thi s bo und a ry was prono unced towa rds th e edges of the co n tac t zone. Latera l m O\'e ment of grains o ut ward along thi s bounda ry was obse n 'ed in thin sccti ons nea r the ed ges but not nea r th e ce ntre of th e co ntact zone. Th e thin sections we re co mpa red to the res ults of medium-scale indentation tes ts in 1989 and 1990 from the Arctic O cea n. The sa me t y pe of fi ne-g ra ined m ate ri al a nd laye r fo rmati on of m oclifi ed micros truc ture was found in th e contac t zon es.
INTRODUCTION During summ er 1995, full- scale iceberg-impac t tes ts were co nducted on Gra ppling Isla nd (nea r Cart wri g ht ), L abrador, Canad a, b y th e Ce ntre fo r Co ld O cean R eso urces Enginee ring (C -C ORE ), M em ori a l Uni ve rsity o f Newfoundl a nd, a nd K . R . C roasd a le and Associa tes Ltd (Cracker a nd o thers, 1997). Th e aim of thi s fi eld test prog ram was to m eas ure the load s generated during th e im pact of LOwed iceb ergs again st a fi xed in strumented p a nel. The need for full- sca le testing of iceb erg impac ts was recog ni zed in the ea rl y 1980s, b ecause uncerta inties assoc iated with extrapolating ice-crushing streng th meas urem e nts from sm all-scale la bo ra tory tests LO full- scale ofE h ore structures led to co nservative a nd thus costl y des igns (Cracker, 1995). Th e goa l of this paper is to prese nt th e micros tructural changes found in a sample ta ken from the contac t zon e of a n impac ted iceberg. A sampl e of ice containing a contac t zone was co ll ected following an impac t test at th e tes t site on Grappling Island . Th e sa mpl e was stored in a freeze r, a nd later thin-sec ti o ned in a co ntrolled laboratory setting in St J o hn's, Newfo undl a nd. Th e mi crostructure of th e co ntac t zone is presented through a se ri es of thin sec ti o ns ta ken along th e impacted surface of the co ntac t zone. The thin sections show the va ri ati on in microstructure a lon g th e imp ac ted surface o f th e contac t zone. A seco nd se t o f thin sections was ta ken fro m ice directl y below the first sla b of ice to obse rve the microstructure behind th e centre a nd edge of the contac t zon e.
Present address: C -CORE, M em ori al Uni\Trsity o f Ne wfound la nd , StJo hn's, ;\Tewfoundl a nd AIB 3X 5, Can a da.
T he microst ruc tura l changes fo und in thi s cO lllac t zone a rc of interest for co mpa r ison with la b o ratory-tested sp ecim ens. This paper is p rese nted as o ne of three compa ni on papers with Mela nso n a nd oth ers (1999) and r..1cg lis a nd oth ers (1999). Th ese pap ers present so me of the rece nt labo ratory ice-test progr a ms co nducted a t r..lemori a l Uni ve rsit y o f":\Iewfound la ncl .
BACKGROUND De \'el opment of the o il a nd gas indu str y operating off th e cas t coast or Ca na d a h as been spu rred by the rece nt Hibe rni a proj ec t, sta rting fro m the ea rl y ex pl orat ion drilling conducted in th e 1970s a nd ea rl y 19808. T he first oil has b een recm'ered at the Hibe rni a site ahead o f schedule, in Nove mber 1997, and th e Te rra Noya proj ect is well undcrway. Other fi el d s such as \\' hiter ose, Bonne Bay, H ebron and Ben Nevis a rc exp ected to prov ide significant future di scove ri es. Oil a nd gas operations in th e G ra nd Ba nks a rea require p a rticul a r attenti on to e nvironmental conditi ons in th e a rea. Th ese incl ude iceb ergs a nd seyere sto rm-generated w ind a nd waye co nditi ons. Th e consequences o f iceberg impac ts ca n be se\'ere, a nd co nsid erati on must b e g ive n to th e probability of these impacts a nd th e assoc ia ted loads based o n the typ e of structure, location of p roducti on site a nd tim e of o p era ti ons. On c of the ma in ac hi eveme nts of th e ice group at iVl emori a l U ni\'er: it y is th e compl etion of a co m prehensi\T m odel for es tim ating ice berg d es ign loads fo r ya ri ous types of produc ti o n systems (?-.1e m oria l Unive rsity of New foundl a nd , 1996). This includes es ti m ation of globa l a nd loca l ice load s ror different produc ti o n systems at a number ofp ra bability of exceedance \·alues. It was show n tha t design globa l a nd
449
J ournalojGlaciology local loads could be reduced substanti ally compared to those previ ously used by industr y, leading to more economical designs. Th e reduction is associated mainly with improved analysis of the ice-failure process. Pas t assessments have genera lly been based on a conser vati ve design premise, for exa mple using conservative estim ates of ice press ure. As knowl edge and understanding of the ice-failure processes improve, th ere is p otenti al for m a king more realistic assessments, a nd for reducing unce rtainty in the desig n-load estimates. Fu ll-scale test data and information from impacted icebergs are key to obtaining a be tter understa nding of th e behavior of ice during impact. Th e fai lure process of ice is complex, with a number of processes occurring during th e interaction. Fractures in the form of spalls reduce the nomina l a rea of contact, a nd highpressure zones a re produced during the impact. High-pressure zones a re areas of high pressure surrounded by areas of relati vely low pressure (jordaan and others, 1999). Highpressure zones accommodate large shear stresses as well as high confin ing pressures towards the centre of th ese zones. As the compressive failure occurs within the high-pressure zone, evidence of microcracking and press ure m elting at the grain bo undaries has been obser ved in the field during medium-scale indentati on tests a nd small-sca le la boratory tests. As the high-pressure zone (s) fai l, recrystalli zation of grains a nd refreezing can occur. Small ice pieces a re extruded at the free edges of the co ntact zone with the release of the confining pressure. Extrusion of fin e ice partieles has been observed in the fi eld at the M olikpaq struc ture in the Beaufort Sea a nd during medium-scale indenta tio n te ts. A schematic of a high-pressure zone and the ex trusion of crushed ice is shown in Figure 1. See Melanson and others (1999) for furth er disc uss ion on high-pressure zones.
High Pressure Zones Under High Confinement
Ice Feature Structure
Fig. 1. Sclzematic qf high-pressure zone.
PROCEDURE Im.pac t e xperim.e nt
Th e test site was located on Grappling Isla nd approximately 14 km by boat from Cartwright. A steel panel, measuring 6 m by 6 m a nd weighing approximately 30 t, was instrumented to measure impact loads. This p anel was 450
m o u nted on a near-ve rtical elifT face with most of th e p a nel below the water surface. The panel co nsisted of 36 triangular subpanels, each of which had a single load cell located at eve ry corner (a total of 108 load cells for the panel ) capabl e of meas uring a uniaxial load. A schem a tic of the test configuration is shown in Figure 2. A rope was attached to th e iceberg, passed thro ugh a snatch block located just above th e centre of the pa nel, a nd then connected to the towing vessel. A p endulum line was used to a lign the iceberg with the centre of the panel, as shown in Fi g ure 2. Following impact, the p endulum line was released a nd the iceberg was towed into position for another impact or moved ofT site so that testing could proceed with a new iceberg. Because towing a sing le iceberg to the site often took several hours, each iceberg was typically impacted a number of times, providing it was sti ll of significant size.
Grappling Island
Fig. 2. Schematic qf iceberg-impact test configuration (plan view). A total of 30 iceb erg impacts was recorded on th e panel. The icebergs were in the range 500- 1500 t, with impact sp eed s reaching 1.5 m S- 1 (Crocker, 1995). Photographs were taken of the icebergs as they impacted th e panel. Plate 11 sh ows the second impact of "Steve B", an iceberg w ith a m ass of approximately 700 t at the time of testing. Pieces of ice can be seen fl ying in the air quite some distance from the point of contact. Sam.ple c ollec tion and preparation
Iceb ergs genera lly impacted the panel well below the w a terline, a nd their contact zones norm a lly remained underwater. Samples co u ld be retri eved on ly if the iceberg rolled and exposed the contact zone, which did not occ ur often. After the fifth impac t of iceberg "M a rlene", a contact zone was spotted above the surface of the water. Iceberg " Marlene" had a mass of approx imately 500 t at the time of tes ting. It is believed that the impact occurred with the eliff face r a th er than the p a nel, since the impacted surface was brown in colour with traces of marin e p lant life. For this reason, there a re no load-time data availa ble for this impact. The sp eed of the iceb erg was approxim a tely 1m S- I The a mbient temperature was approximately lO oC and th e water temp e rature approxim ately 5°e. Th e temperature of the ice in the area of the contact zone is estimated to be between - 5 ° a nd - lO °e. A bl ock of ice containing the contact zone was extracted from iceberg "M a rlene" using a cha in-saw. The contact zone on" Marlene" is seen as a brown spot in Plate 12. The block of
• Fo r colour plates see section before previous pap er.
i\I!ugge,-idge andJordaan: Microstructural change in ice: III ice was immediately transferred to a cooler, a nd was stored in a freezer for 3 days before being tran sported to Memori al University for microstructural ana lysis. Thin-section analysis of contact-zone material
Th e sample was stored at - 30°C at M emorial University to prevent significant modification of the microstructure prior to prepa ration of thin sections approx im atel y 3 wee ks later. Thin sections were prepa red from two slabs ta ken from the sampl e containing the contact zone. A schematic o f th e ice sa mple showing th e contact zone a nd the numbered thin section s is give n in Figure 3. Th e contact zone was a pproximately ova l in shape, with th e m aj or ax is measuring 160 mm a nd the minor axi s 110 mm. Th e ice sa mple was first c ut para llclto the pla ne o f the impacted surface to form a slab of ice approximately 80 mm thick. This slab was called the "impacted-surface" slab and deno ted IS. The approximate ce ntre line of the co ntact zon e was m a rked to indicate the relative positio ns of th e thin sec ti o ns along the impacted surface. Thin secti ons numbered 10 through I, and la through Id, were taken from th e impac ted-surface slab as shown in Fig ure 3. A second slab, denoted slab 2 (S2), 80 mm thick was c ut para ll el to th e first slab. Th e ce ntre line o f the impacted surface was m a ked slab 2, and th e general a lignment of thin sec tions ta ken from this slab in rela tion to th ose ta ken from the impacted-surface slab is shown in Figure 3. The labelling of thin sec ti o ns sta rts with eZI (contact zone I), followed by th e slab, IS o r S2, and the thin-secti on number. For ex ampl e, thin sec ti o n 8 from the impacted-surface slab is denoted e ZI/IS/TS8.
o
11~
16an Plan view of Impacted surface (contact zone of Iceberg)
Impacted Surface (IS) Slab
Impacted Surface
Slab 2 (S2)
8an
Centreline of impacted surface
Elevation view of iceberg sample
Fig. 3. Schematic qf icebelg sample containing contact zone and position qf thin sections. All th e thin sec ti ons were mic rotomed to approx im atel y I mm thickn ess or less a nd th e n ph otog raph ed be tween cross-pola ri zed filters th e same d ay they were prepa red. Eac h section wa s ori gin all y cut with a ba nd-saw to a pproxim ately 10 mm thickness. The micro to ming process bega n by a ttaching th e secti on to a glass plate with three or fo ur drops of water appl ied a long the perim e ter o f the thin section. Th e thin sec tions were mi crotomed until a smooth, level surface was obta ined. They were then turn ed ove r on th e g lass pl ate so th at th e Oat side of th e thin secti o n was aga inst th e g lass p late. Th e thin section was attac hed to the glass pl a te with a
continu ous bead of wa ter a nd th en mi e rotomed to a thickness of < I mm. Thin sections 5 a nd 7 o f the impacted-surface slab a nd thin section 2 of slab 2 wer e lost during th e mie rotoming process. Th e thin sections from this sampl e were found to be extremely fragil e, p a rticul arl y the ones closest to the centre of the contact zon e. Each of the thin sections menti oned was photog raphed be tween cross-polari zed filters at a number of sta nda rd m ag nifications. The o ri entation of the thin secti ons w as such tha t the top of th e photograph r e presented the to p o f the impacted surface for thin sec tions ta ken from th e impacted-surface slab. Th e ori entati on of thin section 3 ta ken from slab 2 was inve n ed in the photog r aph. Globa l photogr aphs o f all thin secti o ns were taken. A number of additiona l photog raphs wer e ta ken of each thin section a t two m ag nification settings. A scale in mm is shown in all of th e pho tographs. A few additional photographs of se lected thin sections were taken with pl a in li ght shone from the side to look a t th e pred omin ant crac k p a ttern. These photographs we re taken approxim ately I yea r la ter. The thin sec ti o ns we re LOO fragil e to mic rotome on both sides, so onl y th e to p side was mierotomed. E\'en though th e storage tempera ture was - 30 °C, some sublimation occ urred betwee n the thin section a nd th e glass plate. Sampl e ph otog raphs have bee n included since th e pattern of cracking ca n be easil y see n. Th e other deta il s in th ese photographs will be ignored, since th ey arc readil y visible in the photographs taken of the same thin secti ons di rectl y a ft er the origina l thin-sectioning process.
OBSERVATIONS AND INTERPRETATION For brevity, onl y selec ted ph otographs a rc presented. Photo gr a phs o f the globa l \'iew of partic ul a r thin sections a rc shown in Plate 13. M agnifi ed views of se lected thin sections a rc show n in Pl ate 14. Selec ted ph otog r aphs under pla ina nd side-lighting conditions arc shown in Plate 15. In a ll o f th e g loba l views, th e microstructure a t th e immedi ate ed ge of th e thin secti ons co rresponds to the froze n bead of water used to weld the thin sec tion to th e glass pl ate. In th e global \'iews of the thin sectio ns taken along th e impac ted surface of th e contact zone (Pl ate 13a- d ), fin egrained material wa s fo und to be prese nt in all of the thin sec ti o ns. Thin sections taken from . Iab 2 (Pl ate 13e and f) indica te th e size of the pa rent grains of th e iceberg, whi ch is simi la r to that of som e of the unaffected g rains in th e thin sec ti o ns ta ken from th e impac ted surface. The ave rage size of th e pa rent grains is 10- 15 mm . The average size of g rains within th e fine-grain m ateri al is consider a bl y less than I mm in di a m eter. During th e microtoming process, it was noti ced th at th e thin sec ti ons cl oser to the centre of the contac t zon e were more frag ile than those nea r the ed ges. A b o unda ry betwee n th e fin e-gra in ed m ateri al a nd th e pa re nt g ra ins was fo und in each of th e thin secti ons take n from th e impacted-surface slab. Thi s bo und a ry ran approx imatel y parallel to th e impac ted surface a t about 80 mm depth a t the ce ntre o f the contac t zone a nd 50 mm depth nea r the edge of the co ntact zone. Amo ngs t the fin e matri x of mall g ra ins, large pa rent-sized g rain s were found. Pla te 13a shows an example o f a la rge a mo unt o f fin e-grained m ateri a l found to exist be tween large un a ffec ted grains. A sig nifica nt diffe rence is see n in thi s main bound a r y pa ra llel to the impacted surface when o nc compares thin sec451
J ournal q/Glaciology
ti ons ta ken fro m nea r the ce ntre with those ta ken from the outside edges of the contac t zon e. In thin sections ta ken near th e centre, the boundary was n o t as distinctive, a nd lateral moti on of g rains a long the bounda ry was not v isibl e. In thin sec tions ta ken near th e edge of the contac t zone, the boundary was prono unced. The gra ins abO\-c the bound a r y appea r to have shifted outwa rds in a la teral directi on. Pl a te 14a- d illustrates the moyement of g r a ins along the boundary in thin sec tions I a nd la taken from the impacted-s urface slab. Thi s is consistent with the ex trusio n process a t th e free edge of the high-press ure zone (see Fig. I). In the m agnifi ed views of thin secti on la , shown in Pla te 14c and d, th e la rge, greencoloured g ra in in the bottom ri ght qu arter of the ph otograph can be seen to be offset a long the boundar y b y approximately to mm. Thi s boundar y m ay be considered to be simil ar to a n extrusion plane with a "crack-like" boundary and the g rain s within bein g forced outwards tow a rd s a free surface. Th e microstructural cha nges observed in th e thin secti ons ta ken from the co ntac t zone of the impac ted iceberg suggest th a t thc process of gene r a ting fin e-gr a in ed materi al plays a n importa nt role in the d eform ation m echanisms during iceberg impacts. Th e incr eased depth o f the zone of fin e-gra ined m a teri al in the centre of the contac t zone, as compa red to the edges of the contact zone, is co nsistent with the configura tion of a single hig h-press ure zone. The centre of the co ntac t zone lVo uld correspond to the centre of th e high-pressure zone wh ere th e confining press ure is higher. At the edges o f the zone th e co nfining press ure is significantl y redu ced , corresponding to a lower prop ortion of fin e-grained m ateri al. Latera l m oti on of lin e-g r a ined materi a l occ urs as th e confining press ure is suddenl y r eleased. The p rocess which genera ted the fin e-gra ined materi al in th e thin secti ons taken from th e impacted surface is compl ex. Thi s m a terial is opaque in nature and appears to exist in bands which nolV past un a ffected grains. For example, thin sec ti on 8 from the impacted surface (see Pl ate 13a) shows mostl y opaque fin e-g r a ined materi al whi ch looks grey in colour whcn viewed between cross-pola ri zed filt ers. Th e sma ll gra in structure may be caused by fin e ba nd s of microfractures which occ urred within the grains. U nder high conlining pressure in the centre of the co nt act zone, it is expected th a t some pressure melting wo uld occ ur during the impact. There is likely a lso to be dynami c rec rystalli zation along g ra in boundaries a nd refrerz ing of the g rains. See the disc ussion in Mcglis a nd others (1999) on the occ urrence of simil a r shear softening in other materia ls (e.g. myloni tes in rocks). In a ll thin secti ons ta ken from th e impac ted surface there appears to be a layer in which most of th e microstructu ra l cha nges in the ice occ urred. These thin sec tio ns represent a "s napshot" of the ice a fter impact, a nd if th e iceberg were to co ntinue its intcracti o n with the structure, the laye r would prog ress furth er into the iceberg. The gener al layo ut of these layers m ay remain th e same, changing o nl y in space a nd time. "Vh at is observed in th e thin sectio ns ta ken from the impac ted surface is a laye r of highl y rearra nged and broken-down g rains followed by parent g r a in materi al. Nea r the outer edges of th e contact zone, the laye r is narrower, but there is still a sig nificant amount o f fine-grained materi a l conta ined in the laye r (see, e.g., Pla te 13d ). It is also noted tha t in so me cases thin sec tions ta ken fro m the edges of the contac t zone have gr a ins simil ar in size a nd orienta452
tion to the pa rent g r a ins at the top o f the thin sec tion close to the interaction surface. \Vithin the laye r o f most micr os tructural cha nge fo und in th e thin sectio n s from the impac ted-surface slab, th ere were parent-size gra ins among st fin e-grained material. Based on the ori entati on of groups o f bubbles in the p a rentsize g rains, some of th ese grains were found to have ro tated during the impac t process. Th e bubble ori entation in th e p a rent grains shown in thin sec tions taken from slab 2 is approx im ately 45 ° from norm a l. The bubbles a re seen as elongated elliptical shapes in Pla te 14c and d, which gives the m agnifi ed view of thin section la from the impacted surfa ce, a nd in Plate 13f, which gives the globa l \·iew of thin section 3 from slab 2. Th e unaffec ted g rains within the laye r of most microstructural change with the sa me ori entation as the pa rent grains (based on the bubble ori entation ) may h ave their c axi s in an ori entation which makes them less likely to rotate or break down into smaller grains. There is fin e-g rained material a lo ng the main boundary which di vides th e pa re11l grains from the grains which have undergone the m ost microstructura l cha nge. This can be seen in Pl ate 14a and b which gives th e magnili ed views of thin section I (CZI/IS/TS1). Fine-grained materi al along the main boundary line m ay be expl ained by th e "chipping off" of g rains as two or m ore grains a re fo rced to slide along each o ther. The res ulting fragments a r e expected to have been roll ndr d through th e process of sIipping of the grains a long the shear plane bo undary. The fin e-grained structure a long the boundary m ay al so help to facilita te further slip. This mate ri a lmay a lso be a res ult of nucleation of new grains. C racking is a ppa rent throug hout th e thin sections a long the impacted surface a nd in thin see tions taken from slab 2 (see Plate l3e and f). Th ere a re randoml y orienta ted cracks a pparent throug hout all the thin sections. These a rc believed to be pre-existing crac ks b ecause of their ra ndom ori entation a nd thei r presence in the parent ice. Cracks w hich are out- of-pl a ne may no t b e \·isible in th e thin section s or may appear as thick g rey lin es as in thin section CZI/S2/TS3 sh own in Plate 13f. R andomly o ri ent ated c racks ha\·e been observed in natura lly occurring iceberg ice. Th e crac ks p a ra llel to th e impac ted surface a re ass umed to be a direct result of the iceberg im pact. It is interes ting to examin e the cr acking paUerns readil y vi sibl e in the selected photographs in Plate 15a a nd b ta ken under plain- a nd side-lighting conditions, respec tively. As m enti oned earli er, since these photographs were ta ken some time after the sections we re mad e, o nl y the pattern of cracking should be noted . Other deta ils in these photographs will be neglected , si nce they arc readi Iy visible in the photog r aphs of th e sa m e thin sec ti o ns ta ken aft er the o riginal thin-sectioning process. In Pl ate 15, very fin e dots in th e o rder of 0.1 mm in di ameter should be ignored. These are th e result of sublimati on and cr ys ta llizati on of the ice over time between the thin secti on a nd the glass pl a te. Cracks throughout thin secti on 8 and thin secti on Id from the impacted-surface sla b a re seen in Pla te 15a and b. In Pl ate 15a, two large "recta ng ul ar-shaped" gra ins measuring approx im a tely 35 mm ac ross can be seen in the centre of thin secti on 8. Th ese actually consist of a number of grains, as can be seen under cross-pol a ri zed filters shown in Plate 13a. The gra ins which make up each of th e larger g ra ins app ear to ha\·e moved as a group. This movement has bee n determined throug h close examinati on of the crac k lin es through the two la rge grains. It appears that the two large
Nluggeridge alldJordaan: IIlficmstructuml change in ice: III gra ins wer e I)]"e\'iou sly side by side and th at the grain on th e left has m O\'ed down rcla ti\'e to the oth er la rge grain. T his move ment of groups of g ra ins suggests tha t there is some rea rrangement and brea kdow n of grains w ithin the ee J1lra l region of the contact zone which may e\-cntu all y ex trude out th e edges of the contac t zone as the ice ack a nces. In Pl a te 15b, many microcrac ks can be see n in thin secti on Id ta ken ri'om th e impacted-surface slab. Th e laye r o f highly m odifi ed microstructure abO\'e the m a in bounda r y is bright in colour due to the refl ecti on o f li ght from th e many crack surfaces. The la tera l move ment of the gra ins a long th e bo undary ca n easil y be seen as crac k li nes are interrupted a nd offse t.
DISCUSSION As stated in M elanso n a nd o thers (1999), the fa ilure or ice in compression consists of fr actures, with m ost of th e load being ca rried through hi g h-press ure zones tha t retain th eir integrit y, a t least for a sho rt p eriod of tim e. Th e press ure di stributi o n over these zones has been obser ved to be para bolic in shape, with low co nfining pressures nea r the edges, a nd high pressures near th e centre, Exce pt a t ver y low speed s o f illleractio n, a distinct, na rrow layer of hig hl y da maged ice has inva riabl y bee n obse r ved to form adj acent to the structure. Examples of this have been give n in the p resen t paper, including the first exa mples to our knowledge of the fo rmati on of this laye r in a real iceberg impae l. Studi es of th e mechanics of the interac tion process haw bec n carri ed out Uo rdaa n and oth ers, 1999), using the me thods of da mage th r ory. In thi s, state va ria bles represe nt the prior stress hi story of ice a t various locati o ns, ta king into acco unt also thc states o f stress. In the simul ati ons conducted , it was fo und that da m age associated with miero frac turing occ urred first nea r th e edgrs. th en prog ressed inwa rd . So ftening associated with pressure effec ts then occ urred nea r th e ce ntre a nd adva nced toward s the edge. This res ulted in a reaso nabl e simul a ti on of laye r fo rm ati on, with sudden fa ilure of the laye r ta king pl ace when th e two zones m ee t. A co mpa rison of the obse n 'ati ons from the current thin sec tions a nd those from th e sm a ll-sca le labo rator y tests desc ribed in M elanso n a nd o th ers (1999) a nd i\legli s a nd oth ers (1999) is give n here. Fig ures have been chose n to illustrate severa l po ints. In compa ring the current thin sec ti ons a nd the labora tory tests described in l\Ielanson a nd others, th e tests with higher stra in ra tes we re the ma in foc us since th ey represented rates ex pected in full scale. Thin sections from the centra l region of the impac ted surface of the CO lllact zone desc ribed above were compa red to the tests at 20 MPa co nfining press ure, which m ay correspond to the centre of the COlllae t zone (high-press ure zone). In partic ul a r, Pl ate 13a (thin sec tion 8 taken from th e ce ntral-regio n contac t zone) was compa red to Pl ate 2 b. Th e microstructure in Pl ate 2b is reasonably simil ar, indicating th e fi eld tests I ikcly had stra ins in the 5% region. In Pl a te 2, a substanti a l breakdO\m of gra in-size can be seen with strain s of > 2% . In co mpa ring thin secti ons from the present paper a nd th e labora tory-tested ice d esc ribed in M egl is a nd others, example thin sections (i'o m th e edge a nd from near th e centre o f th e contact zone a re co mpared to ice sampl es tested under co nditi ons simil a r to those expected a t th e edge a nd a t the centre of a high-press ure zone. Thin sec tion I from the impac ted-surface slab shown in Plate 13d co mpa res favo ur-
abl y with Plate 4 b. The labo ra tory cree p tes t was a t a low confining press ure (Pc = 5 MPa ) with a deviatoric stress of 15 MPa, a strain ra te of 10 I to 10 2 S I and a stra in o f about 10 %. Thi s va lue of devi atoric stress is representa ti ve of th e va lue expected towa rds the edge of the co ntact zone. Plate 4 b shows ex tensive microcracking simil ar to that seen in Pl ate 15 b (th i n sec ti on Id taken from the edge of th e contact zo ne with side lig hting). At hi gher co nfining press ure corresponding to the centre of the contact zone, thin sec ti o ns from l\I[eglis and others showed th e sam e breakdown of g rains as seen in Pl ate I3a (thin secti o n 8 ta ken [i'om the centralregion contact zo ne). This is shown in Plate 7a which illustrates tests at 50 MPa eonfinemel1l, 15 l\IPa applied devi atoric stress a nd a stra in of a bout 4% . The microstructure was simil a r to that seen in M elanson a nd others as d isc ussed above for 20 MPa eo nfinemen t. In summary, the results show n in th r prese nt thin section ta ken from th e co ntac t zones have microstructures consistel1l with confining pressure ra ng ing fro m approx imately 5 MPa at th e edges to about 20 MPa near the ce ntre. Highl y d amaged ice is likely to have th e expected shear of approx im ately 15- 25 MPa a nd strains of >4%. To all ow co mment to be m a de on the genera l observations represented in the thin sec ti o ns ta ken fro m th e contact zone as compa red to res ults from some medium-scale indentation tes ts, measurements o f th e layer ofm odili edmicros tructure due to th e impact we re m ade and a sketch was drawn (sec Fi g. 4a ). Th e depth rro m the top of each thin section, corresp onding to the impac ted surface of the co ntac t zone, to th e bo unda ry betwee n th e pa rent ice a nd the ice \\'ith changed microstructure was noted and indicated as points in Fig ure 4·a. Eac h point co rresponds to the average depth of the bo und ar y a nd th e a pprox im ate pos iti on of each thin section. A c urve was drawn using the points as a g uide to estim ate th e shape orthe bound ary. As menti oned earlier, th e bound ar y was observed to have th e greates t depth (approx im atel y 80 mm into the ice ) nea r the centre of the contact zone. Nea r th e edges of the contac t zone, th e depth of the bounda ry from the impac ted surrace is approx im ately 50 mm. Th e . hap e of the bo und a r y is reasonably fl a t in the 16em
i
i i
~
, I I
I
I
i 8em
Layer
I I
~ .~ I
a
••
IMPACTED SURF'"AC E
Mostly Mhttvrll!' of F"lne
Gr-Q.,ned o.nd Po-ren t Size GrOined Ma. t erl(ll
b
r".
PARENT ICE
Fig. 4. Schematic drawings rfjmijile qfla)'er with 1I10stll1icro slntclllml change. ( a) Schematic qfp rqfi.le based 011 measuremel1ts rf /a)'eT depth ill thin sections. ( b) Sketch qf prrfile shape rfthe contact zone. 453
J ournal qfGlaciology middle, steep during the transitio n from the centre of th e contact zone a nd level at the edge of th e contact zone. The bounda ry was more pronounced away from the centre of the contact zone, as illustrated in Figure 4 b. Th e definition of the bounda ry appea rs to be linked to the a m ount of latera l m ovement of the g r ains. Grains near the centre of the co ntact zone had little la teral movement. At the edges of the contact zo ne, the lateral motion of g r a ins was significant, a nd offset of ma ny individua l grains was seen above the bound ary. Fine-g rained m aterial sur ro unding unaffected pa rent-size grains was observed in the laye r b oth near the centre a nd near th e ed ge of the contact zone. Th e centre of the contact zone is considered to be the approxim ate centre of a high-press ure zone, with lower presSUITS at the edges where extrusion o f fin e ice pa rticles occ urs. The thin sectio ns were compa red to the results of m edium-scale indentation tests in th e Arctic O cean in 1989 (Frederking a nd others, 1990), wh ere crushing and extrusion processes and hi gh-pressure zones we re evident. A photograph of the ice face a fter an indenta ti on test from Fred erking a nd others (1990) is included in Pl a te 16. Pl ate 16a shows the contaet face with a view of th e crushed layer. Plate 16b shows a thin section ta ken from the a rea indicated by a box in Plate 16a. In Pl ate 16a, th e shape of the layer is cl earl y shown. The centra l region of the layer, which is included in the a rea indicated by the box, corresponds to the blue zone of th e contact face. On both sides of this cen tral area, a layer of op aque white ice was observed continuing to the edges of th e contact face. The ice in this zone has been crushed a nd pulveri zed into small grains. Near the centre, most of the large g ra ins remain, but with incipi ent d a mage on the g r a in bo undari es and in some grains (Frederking and others, 1990). It was noted by Frederking a nd others (1990) tha t th e ed ge of the contact zone was soft, while the centre was quite solid even in areas where the ice had been highly pul verized. Thi s finding is in agreement with the microstructure of the ice sampl e shown in Plate 16b. The layer of highly pulve rized ice can be see n on either side of the central region. The bo unda ry between the fin e gra ins a nd th e pa rent g ra ins is ve r y disti nct away from the centra l region, particul a rl y at the bottom right of Pl ate 16b. In th e centre, there are m os tl y p a rent-size grains, but the orienta tion of the gra ins is not ra ndom (as in und a maged ice ) a nd so me change in microstructure is appa rent. Results of the indentati on tests indicate tha t at higher indentati on rates, which would more closely represent iceberg impacts, a relatively thin layer of crushed and da maged ice with ex trusion of fin e pa rticl es was observed. The thi ckn ess of the laye r was found to vary th ro ug hout the layer, but a narrow ba nd of relatively undam aged ice usua lly in the central pa rt of the co ntact zone was obser ved after each test. In the centra l part of th e contact zo ne, measurements of local ice press ures were about three times higher th an th e ave rage pressure over the entire contact face (Frederking a nd others, 1990). Plate 17a, ta ken fro m Meaney (1 992), shows a horizontal thick secti on ta ken from a n ice face after impact with a fl at rig id indentor (test denoted TFR-Ol) a t H obson's C hoice ice isla nd in 1990. M ea ney (1992) stated tha t a lthough the centre of the horizontal thick secti on was da maged, a discrete crushed layer did not exist. The crush ed laye r show n in Pl ate 17 a had an average thickness o f 57 mm and a m aximum thickness of 155 mm, a nd was found to increase in thi ckness toward the perimeter of the ice face. A sketch of th e zones obser ved on the test face, a nd the corresponding pl a n v iew 454
depicting the layer, is given in Plate 17b from M eaney (1992) and M eaney and others (1996). This sketch shows the distinct bounda r y away from the central region a nd th e lack of a distinct bound a ry in the centre. The laye r which extends to the edges of the test face is a zon e of pul ve ri zed fin e-grained ice. Segments near the edges of tb e contact face which broke off during th e test are show n a s spall ed a rea s in Pl a te 17b. Density meas urements along the crushed laye r showed th at th e density increased towa rd s the centre of th e crushed layer. The substa ntially greater confining press ure in the centre of the inde ntor was thoug ht to contribute to thi s findin g. Th e crushed layer was found to contain la rge ( > 25 mm ), relatively unda maged pa rticl es surrounded by a m atri x of pulveri zed ice, suggesting a g rinding action as pa rticles slide and rota te during extrusion (Meaney, 1992). M eaney a lso noted th e differences produced by th e use o f different shapes of in d en tors. The wedge indentors produced a higher degree of spa lling than th e fl at indentors, a nd the laye r of crushed ice produced with the wedge indentor was indicative of the lower confinement presen t. A sectio n through the impacted surface a fter a sphericalhead indenta tion test a t H obson's Choice ice isla nd in 1989 is presented in Figure 5 fo r co mpariso n to th e present set o f thin sections. The rigid spherica l indento r used had a n area of 0.8 m 2 with a 1.28 m ra dius of curvatu re a nd 1 m overa ll diam e ter. Fig ure 5 shows the same typ e of crushed laye r conta ining fine-grained m a teri al. A prono unced boundary is see n som e di sta nce away from th e centre of the contact zonc. Th e central region contains a mi xture of fine grains a nd p a rent-sized grains a nd ex tend s some way into the ice. Thi s is m ost likely due to the shape of the spherical indentor a nd th e confi nement at the hea d of the inclentor. In disc ussions o f th e indentation tests inJordaan a nd o thers (1990), the ice within the layer is stated to have a lower density tha n the pa r ent ice. The sintered nature of the ice a fter the tests was di scussed a nd th e importance of pressure melting in the deform a tion process was n oted. Observati ons we re found to agree with earli er obser vations from drop-ball tests (Kheysin a nd Cherepanov, 1970). Simila rities found in the va rious indentation tests discussed a nd in the present thin sections are the formation of a layer of fine-grained materi al and the pronounced bo unda ry toward s the edges of the contac t zo ne. In a ll tests, th e centra l region showed cha nges in microstructure within th e ice but no pronounced bo unda ry. The difference in the depth of cha nges in mierostructure in the centra l region was noted for th e vari o us examples m entioned. In the indentation tests, the central region was a thin band. In the present thin sections ta ken from the contact zone, the central region of microstuctural change in the ice extends sli ghtly further into the sampl e. In all tests, there was evidence of extensive cracking, extrusion of sma ll crushed ice pa rticles a nd pressure melting. Based on the observa tions made of the microstructura l changes in the impac ted-iceberg contact zone, a number of suggestions for furth er wo rk are presented . It is important to use iceberg ice in future sm all-scale testing a nd to conduct tests under loading and confinement conditions simil ar to those exp ected in iceberg- structure interaction (e.g. test durati ons of simil a r leng th to iceberg impacts). Comparisons of the microstructura l cha nges under different loading and confinement conditions have gi\"en insig hts into the deformation m echanisms occ urring in the ice. See M elanson a nd other s (1999) and Megli s a nd others (1999) for recent labora-
1I1uggeridge alldJordaan: Microstructural change in ice: III TEST 05
Parent Ice
Indented Ice Interface
Crushed Layer Boundary of Crushed Layer
Edge of Indentation
Fig. 5. Section acTOSS cTllshed layer after spherical-head indentation lesl, Hobson's Choice ice island, 1989 (photo taken Stander and used with permission). tor y test programs conducted at M em orial Uni\"Crsity. Continued laborator y testing in these a reas a long with furth er full- sca le testing wou ld help to improve understanding o r the processes occ urring in the ice durin g iccberg impacts.
CONCLUDING REMARKS A few of the highlights o rthe obsen'ations ofth r mi crostructure found in th e contact zone are noteworth y. A laye r of ice with hig hl y modifi ed microstructure was see n in the thin sec ti ons taken rrom th c impac ted surrace. The boundar y be tween the ice with modifi ed microstructure and the parent ice was found to b e more pronounced away from th e ce ntral region orthe contact zone. Fin e-g r ained material in the laye r on eith er side of the central region was found to move in a lateral di rection toward the ed ges of th e contact zone. Th ese findin gs are co nsistent with the concept of a high-pressure zone with higher press ure in the centre and lower press ure at the edges as ice is extruded. Cracking was e\·ident throug hout the thin sec tio ns in \'a ri ous direction s. Cracking parallel to the impacted surface is tho ught to b e a direct result of th e iceberg-impact process a nd ma y be due to unloading. The findin gs co ncerning the prese nt thin sec tions were simil ar to those of previous medium-sca le inde nta tion tests.
ACKNOWLEDGEMENTS The authors would like to ac knowl edge the support obtained through the Nationa l Energy Board, C a nada. G. B. Crocke r a nd G. M. Engli sh of C -CORE provided valu able information [or the proj ec t, and their enthusiastic cooperation is g ratefull y acknowledged. Th e authors acknowledge th e use or th e facilities at the Institute for Marin e D yna mics, Na tiona l R esea rch Council of Canada. The assistance of B. M. Stone, ]. F. Sweeney, 1. L. Meglis and P. M . Melanson was
~)I
E.
apprec ia ted. Disc ussion s with R . E. Gagno n and S. ]. J ones of the In stitute for ?\[arine D ynamics were a lso helpru!. Th e authors thank E. Stander {or p ermission to use Figure 5. This work was supported in part by the Natural Sciences and Engin eer ing R esearch Council and the Canada ~Newfound land OfT~ hore Development Fund. Th e authors thank th e rev icwers for th eir co nstructivc and useful comments.
REFERENCES Cracke r. G. 1995. Iceberg impact expe riment. C-CORE . \ ;'1("1, 20 3, 10- 12. Crocker, G. B.. K. R. Croasdalc. R . F :'I1 e Kenn a. G. :'II. Eng lish, J Gllzzwcll a nd S. Bruneau. 1997. C-COR F; iceberg impacl rrperimenl /"ha" 2,jlnal reporl. SI J o hn\ :\'Od. l\l emorial niwTsil y of i\C\\ lonndland. Ccmre 101' Cold Ocean Resou rces Engineering. (C-CORE Comran Repon 96-CI6.) Frede rkin g, R ., I.J.J ordaan a ndJ. S. :'IlrCa llum. 1990. Field ICSIS of iee indem,nion al medium sca le, H obson's Choice Ire Isla nd , 1989. 111 /Af/R. '~)'III/}OJillm 011 Ire, Es/)oo. Filllalld. '~ lIgIlJI 20-24. 1990. Proceedings. 101. 2. Espoo, Hel sinki Cni\'crsil ), o fTechnolog)" 931 9++. J ordaa n, I..J. alld 6 others. 1990. Dn'elo/lInenl '!! nf!(' ice load models. SIJohn·s. Nlld, l\lcm o ri a l Universily of Ncwfoundlancl. Ccnlre for Cold Ocean Resou rces Engineering C -CORE ). J ordaa n, I..J. . D. G. l\[als kc\- ilc h and J. L. r..lcglis. In press. Disimegralion of ice uncle I' 1~lsI comprcss i\'e load ing. 1nl. J Fracl. , 97 (1 +),279-300. Kheysin, n. 'lc. a nd X V Che repa n O\·. 1970. IZ\11cne ni yt" slrllkl ur)' I'da v zone uclara I\'e rdogo Icla 0 povcrkhnosl' ledyanogo pokrm'a [C hange of ice slruClure in Ihe zone o /"impaCl ofa solid body aga insl Ihe ice cove r surface]. Probl. Arl.!.. Inlark!. 34-, 79- 8+' r..Iea ncy, R . B. 1992. Loca li zed c rushing in ice-slruclure interaction. (r.. I.En g Ih esis, r..lemoria l Uni\"(Tsil y ofNc\\'founcl la nd .) l\ lca nc)" R ., I..J. J ordaan a ndJ Xi ao. 1996. Analysis o f m edium sca le iccincle nl a lion leSlS. Cold Reg. Sci. Tee/1II0!.. 24 '3),279- 287. :--kgli s, I. L.. P l\J. r../rl anson a nd 1.J J ordaan. 1999. l\li croslruCl ura l cha nge in ice: 11. Creep beha\'ior under Iri"x ia l slress condilions. J Clnciol., 45 (151), 438 148. l\l elanson, P. 1\1. , 1. L. l\ lcgl is. 1..J. Jordaa n anclB.l\1. Sto ne. 1999. I\licroslructural change in ic e: l. Co ns lan t-ci eforma li o n-ra tc leSlS under lriaxial slress co ndili ons. J Glariol., 45 (151), +17.J.22. ~[e m o ri a l U ni vcrsil )"ori'\cwfollndland.1996. Calladiall OJpho!"e Desiglljor Ire EIIl ,irolllllenl (COf)/E ) "colld alll/lInl reJiorl .. IJohn"s, I\'nd , i\J rmo ri a l Lni\'ers il ), o fNcw fo uncll and. Ocean Engineering Resea rch CCnlre.
MSS received JI Jun e 1998 and aeee/J/ed illlevisedJorm 9 A/Jril 1999 455
