A THEORETICAL AND FIELD STUDY OF LOAD TRA NSMISSION THROUGH GROUNDED ICE RUBBLE

(1)

(2)

(3)

(4)

(5)

A THEORETICALANDFIELD STUDY OFLOAD TRANSMISSIONTHROUGHGROUNDEDICE RUBBLE

by

@ AlfredR.Marshall,B.Eng

AthesissubmittedtotheSchool of GraduateStudies in partial fulfillment of therequirements for the degreeof

Master ofEngineering

Facultyof EngiIleeringand AppliedScience MemorialUniversity of Newfoundla.nd

St.John's, Newfoundland Canada

September 1990

(6)

NalionalLibrary ofCanada

C:ilnadi an ThesesService Servicedesthesesceoeecrocs

The authorhasgranted an WTevocabIe non- exclusivelicenceallowingl;~NationalUbrary ofCanadatoreproduce,loan,c:flSlribUteor sen copies of hislher thesis by any meansandIn anyformor format,makingthis thesisavailable toInterestedpersons," .

Theauthorretainsownershipofthet:opyrighl in hiS/herthesis. Neither the thesisnor substantial extracts from it maybeprinted or otherwisereproduced without tlislherper- mission.

L'auteura accorde una licenceIrrevocableet non exclusive permettanll!\faBibl'lOtheq\'e nationaleduCanadadereproduire,pc6U , ol$trtbuerou vercre des copiesde sa these de queique manieraet sous Quelqua lonne que cesoitpour mettre des exemptalres de cette these atadispo sitiondespersoroe s lotere sse ee.

L'autelJl'CXlOSefVeIaprolJ(ieledudroitd'alrteur '\l uiprotegesathese.NiIathesenidesextraits substantialsdecene-ctnedotventeue Imptimesou autrement reproduils sansson aulorisation.

ISBN0-31 :)-65 363 -9

(7)

Aerial view of CaissonRet ainedIsland (CRI)atKaub vik,May1987.

(8)

ToLisa

(9)

ABSTRACT

Oil explorationfromartificialIslands in the Beaufort Seahasshown that grounded ice rubblefields oftenaceumuletearoundsuchst ruct ures,preventing moving icefrom acting directly againstthestruct ure. Asimple,rigid body model,has comm onlybeen usedto estimatethe load "seen"bythe structure,withthe resultthatno loadingis expected unt il globalrubbleslidingoccurs.Itwas realizedthat such an approach is notentirelyrealisticandso a collaborati ve field program was carriedout,followedby theoreticalmodelling. The field workwas doneat a drilling location in the southern Beaufort Sea.on ESSO'sCaisson Retained Island(CRI). The threegroupsinvolved wereESSO ResourcesCanada. Ltd .,National Research Council of Canada(NRC) and MemorialUniversityof Newfoundland(M.U.N.).TheM.U.N.data andsubsequent theoreticalwork is presented here,with the ESSQ and NRC datasummarized in the sitedescription.

Mem orial Universit y collecteddata from pressuresensor rceett ee,ast rainarrB.y, thermocouplearrays, rubble profiling, and ice propertymeasurements. Themost importantda.ta came from the preeeure sensor rosettes and the thermocCluplearraya.

Theseindicated that averagesea.ice pressuresagainst the rubble reached350 kPa, and that a substantial and rapidly fanned refrozen layer existed within the rubble, witha.thicknessexceeding 3 m near the rubble field periphery.

Based on the field measurements,a theoretical modelWMdevelopedtoinvesti- gate the role of therubble refrozen layerin load transmission through rubble.The rubble field was modelled as a system of springs and dampen to representthe elastic, frictional,andviecc ue atiffnetlsof the variousload paths. The deformations and load distribution swerecalculated using a commercial finiteelementpackage called ABAQUS.Each material property was assumed tohave a range of veluee, the lim-

(10)

its of which were determinedfrom the literature, theoreticalconsiderationsand field measurements.

Thesensitivity of the outputtoeach of the inputs was examinedand the following wasconcluded;

1.From the analysis it appears that the elasticresponse of therubble leadstothe highestloadson the embeddedstructure becausethetransmittedloadtendsto decreasewith time(creep).

2.The ice rubblepropert ies that are the most importantfor further studyare the delayed elasticand shearpropert ies of un-refrozearubble,as wella!Jtheir variabilitywithin natural rubblefields.

3.Significantloads maybetransmitt ed through icerubbleto astructure before global!!lidillg of therubblefieldoccurs.

4.Thepresenceof the refrozenlayer givesthe potentialfor substautial loadsto be imposedon the structure duringrubblefield movement .

5.Possiblefuture improvementsto the ABAQUScomputer modelinclude the incorporationof;refrozen layerdiscontinuities, bermelope,non-rigid structure behaviour,kinetic and static friction values (rather than an average),and water and icemus.F'urthurresearch is alsorequiredon ice/rubblefailuremodes to eeeblleh designpressuresat therubblefield boundaries.

6. The sample3Dcalculationsshow thatviscous deformation may also be important,althoughit is also clear that the3D methodpresentedneeds improvement.

iii

(11)

KEYWORDS

Sea ice, sea-ice,ice rubble,iceloads,land fas\ice, rubbleSelds,Kaubvik,artificial ielea de,Beaufort Sea,pressure sensors,strainarray,thermocouplearray,refrozen layer,consolidatedlayer,ABAQUS,compu\er model.

(12)

ACKNOWLEDGEMENTS

It has taken threeyears to completethis thesis,duringwhichtimemany people ha.ve helped insome way.I wouldlike to take thisopport'.u!it-yto thankthem,allwell as to acknowledge theircontributions,and I canDillyhope that I have livedup to their expecta tions. To begin with ,Iwouldliketo thankmy supervisor,Dr.IanJ.Jorda.a.n forinviting meto take part in this project,andforhis time,guidance, generous financialsupport,end benevolent sharing ofacademic credit. I wouldalso like to than k Dr.RichardF.McKenna,who acted&IIan unofficialsupervisor, unselfishly providingadvice,assistan ce,and encouragement .Fortheirfinancialsupport,I am gratefultothe Natural Sciencesand Engineering ResearchCouncil(or fundingthe fieldwork, andthe Provir....alGovernmentof Newfoundlandfor awarding me career developmen tfunding.Iwould liketo expressmygratitudetothosemembersof the field projectteamwithwhomIhave had themoot contact,Dr.KenR.Croil.!ldale, CharlesDer,Dr.Mohammed Sayed,Dr.BobW.Freder king ,Dr.Jean-Pau lNadreau, and Eugene Guy;thank you foryour advice and patientreplies to all myquestioM, Finally,Iwould like to thanksome of myfellow graduatestudents,Sanjay Singh, David Finn, John Croes,and Jing Xiao for helpin gme with laterfieldmeaswemenh at Hogan'sPond.

(13)

List of Figures

Ice thicknes s versus tim e after Sept embe r 1. Data.isfro mThesiger

Bay (SachsHr.)(from Parker ,1987) 15

Bathymetriccontour map ofberm at Keubvlk(from Frederkinget&,

1988). .. 17

Detai lsof the ORIshowing a) plan view orthe caissonring,b) cross sect ion through a caisson,and c) crosssect ion througha completed island(from K.R. Croasd aleand Associates,1985). 18 Planviewof CRIshowing ice pressure sensor locetlcee[fromCroMJal e et aI,1988). . . •.. ....• ..• •.. ... •.. . •.. 20 Microcelloutputcovering the period of highestmaximumpressure {from Croasdaleet itJ, 19S5}... . . .... . . ... . .•..• 21 Shearbar datashowing one ofthe high pressure events. Note thll.t although thepressure magnitudeis similar tothatmeas uredbythe mlercee llon the same date,the shearba.rshows t.he event much more clear lybecauseof theincreased sensitivity(lowerfull scale)(from Orc ee- dale etel,1988). . ... .. . •, •• ..•• ,.•..,. , ... . . . 22 NRCpanel and surveylocations in the latewinter rubb lefieldatKaub- vik(from Frederkingetel,1988)....,•.. . .• ,...•,.• .. 24 Internalrubble preeeueee as measuredbytheNorthwest Exxon array duringthe January 8th event(fromFrederkin get&.I,1988). . 25 The pressure distrib utionat one timeduring the January 8th rubble build ing event (fromFrederkingetal,1988). ... • •.• . •.•..• 25 10 Arct~penel rosettedata (fromFred erkinget al,1988).• . • . .. . . 26 11 Data fromhorizont ally segmented pressurepanel (from Frederking et

el,1988)... . • • .... .. • •.•• • ...•.... 27 12 Sur veyedrubbleelevationslKlutbeastof the CRI... • ....• 27 13 ElectronicDiatanceMeaaurement surveyresults 8howing horizontal

movement and (inbrackets)eettlements over the winter(fromFeed- erkinget al,1988...•.• •.•.•. .. •. . ..••.•.. • .... 28

viii

(16)

14 Planview ofCRI and rubblefieldwiththe positionof the M.U.N.field

measuremenhindicated . .. ... . .. .. 31

15 Dill.gra.m. of mercuryfilled pressuresensor. .• . . .. . . .•. . ... 32 16 The principalstress anglefor array2. . . ... • . . 33 17 Stress towards the CRI,averaged from thethree rosettes. . • . ... 34 18 The expanded plotofstresses -neeeuredby stresssensor Sla dearly

showsthat periodsof steady pressure were interrupted by rapidly fluc- tuating, apparently brittle, deformations.Duringtheentireperiodthe maximum principa.lstress did Dot reach400 kPa. 35 19 'ihi s photograph shows a fresh verticaldeformat ion100 m northeast

ofthestr ess eenecra.Notethe polarbear tracks14·'·eforeground." 36 20 Showing the physicalmodel of '" landfast ice sheet usedforthe com-

puter prograJD. . .•. .• . •.. ... 38

21 Plotofsea ice pressure averagedfrom mercuryfilled pressuresensors I~'~k. . .. ... . . .. ... .... .... ~ 22 Plot of landfastice movement rateat a point30 km from shore,pre.

dietedfrom wind and temperaturedata . . • .... .. 40 23 Plotof landfast sea ice movement versustime,predicted fromwind

and temperatur edata . ... .. . . .•.• . . ... .... . . 41 24 Plot ofsea. ice temperature versustime.Temperat urewasmeasured

0.6m deep in theice. . .... . ... . . 41 25 Componentoflandfastice movementrate dueto thermal expansion. . 42 26 Plotof predicted landfastice movementrate due towind stress.... 42 27 Showingthewindstressed wedgethatisassumed toload therubble

fieldat Kaubvik. . . .• • • . .. ..•••.•. . . . ..•.•.... 44 28 Strain4.l.Tay dimensions andorienta.tion.... ... . ....•.. .. 46 29 Maximum andminimumprinciplestrains,inner rosette..• .••. .. 47 30 Diagramshowing the install edpositions of the thermocouplearrays

andhow therefrozen layer thicknessiucreesedduringthe field period. 50 31 From thetemperatureprofilethe interfacebetweenrefrozen andun-

frozen rubblecan dearly bedistinguished.Profile taken from theshort array,midnightday73.• ... . .. .•... ....••.. .... 51

(17)

32 Plotofthe averagethermal gradientdirectlyabove the refrozenlayer freezingfront.The topcurve isfor the valley andthe bottom curve is for thesail. . .... .... . .. ... ... 52 33 The graphabove showshow dramaticthe rate of refrozenlayergrowth

differsfrom sea icegrowth 3.ttwo locations. .. . 53 34 Twodimensionalgeometr icmodel of an"aged" rubble field(ie has a

refrozen layer). •.. . ... . •.. ... 58 35 Diagrams showing how relat ivestiffness of therefrozenincompari-

sontoun-refroze n rubble is expected to affect the deformationof a groundedicerubblefield... .•,•. ..•. 59 36 Theeprfng-dampc~system commonlyused to rheologically modelice

compliance.Thisarrangementis known as a Burger'sBody.... .. 60 37 Diagramshowinga spring-dampersystemto modeltleformationsin

groundedicerubble.An-ntire rubble fieldiscomposed of aseries of these "units"connected togeth er. ..• ... . . ... . .. 61 38 Simplified spring-dampersystemusedin this study... 62 39 Single unit with eight elements..•. . ...•...•..•... 63 40 Figure sbowing how average compaction of the upper portionofuu-

refrozenrubbleis relatedtorefrozenlayerdeformationand bottom sliding (linear assumption)....•. .• ... . . . ...•..• 64 41 Graph of average consolidation pressure withi..ihe ua-refrosen rubble

versus settlement strain rate.. .. ...•.•.•... . .. . .•... 80 42 Simpletheoryprediction of reaction force versusborizontalposition

in the rubble fieldfor the baselineCMe.Reactionforceistheapplied forceminus thetoad transmitted to the seabed. ... . . ... 87 43 Average and extreme load distributions obtainedbyvarying the spring

stiffneesesin the spring/dampermodel.• . •.•.•... . ..••.. 90 44 Spring/damperresults for baseline case with variation ofcompaction

stiffness to givemaximum,minimum,and average loadtransmission. 91 45 Spring/dampermodelresults with variationortherefrozenlayer stiff-

nessonly, to produce the maximum rauge of transmitted loads. •.. 92

(18)

46 Spring/dampermodeloutputfor baseline case and variationofun- refrozenrubble shear stiffnessonly,to givemaximumrangeof trans-

mittedloads. .... . . ... . ... 92

47 Simple theory versus spring/dampermodelpredictionsfor thebaseline case,using maximum and minimumice/soil frict ion....•.• •... 93 48 Simple theorypredictionsofreactionforce versussail height ... 94 49 Simpletheory predict ionsofreactionforcevenuswaterdept h.. ... 95 50 Influenceofrefrozenlayerthicknessonload distribu tion..•... 96 51 Reaction Coreeat the str ucture(only)versus rubblefield ext ent Cor

simpletheory and thespr ing/ damp ermodel.... .. • .... . ... 96 52 Appliedloadva reaction force at the st ruct ure(only), forthe baseline

case.Simpletheoryand spring/dampe rmodel predictions... .... 97 53 Reaction force versustime (or averagespringandsliderst iffnesses with

damperschosento giveaverageand fastest rates o( loadredist ribution.98 54 Spring/dampermodeland simpletheorypredictionsforstea dystate

(time=co)react ionforceat the structureversusextent of the rubble fi~.• ...•• • •••• .• . .. . . . ... . .•.. ... . 00 55 Spring/dampermodeland simple theory predictionsfor reactionforce

at the structureversus applied load. ... .. .• ..• • • ... . .•. 100 51) Semi-infinite,unrestrained,three dimensional rubblefield represented

bythespring-dampermodelwithout a structu re. . .• •..• .• •. 101 57 Transmittedforce versuspositionin the rubble field(orinitialelastic

conditio ns.•. .•.... .•...• .•....• ... . ..•.. 102

~8 Initial refrozenlayer elasticdeflect ionversusposition (or averagecon- ditions..••.. . .. . .•. . . •... • . .• . .•. .. .•..• 102 59 Transmittedforce versuspositionintherubblefield for finalsteady

stat e (creep)conditi ons. ... ..•..• • •.. • •. . .. ... 103 60 Refrozen layer velocit yversus positioninthe rubblefield for final

steady state (creep)conditions. . .•.... .. .• •. 103 61 Representingthe ice structure interacti on problemas aneleet lcplate

of thicknesst pusheda distanceCpast a fixed,rigidpegofdiameterD.104

(19)

List of Tables

Summazyofisland /icefield measurementprogru mdescribedin the open literature.• ••••• .• •••..•• . . • • .•....•.. . • Results (rom&Ul;CTand thermistormeuurementsofrefroze nlayer thickness. •.••• •. .. . .. . . .•• •..• •. .• •. •. . • . . Rates

o f

refreezing forthe short thermocouplearray(sealevelat0.0 m depth)..• .•. ..• •.•• ••..••. .•.•. • • • •..••• 51 Rates of re£reezing for the long thermocoupleura.y(aea.levelI.t4.5 m depth).. ...•.. . .. . ... . ...•... .•. 52 List ofmateri al elastlcst iffnesses covering morethan fiveordersof magnit ude. •.•. .•.•. ... ••... ..•• .. . ••....• . 68 6 Measured ice/ice frictioncoefficients. ...•..•. ... ...•. 11 7 Firs t yea.t ridgeaailendkeelslope anglesfrom directmeasurements. 72 8 Internalangle of friction fromshearboxtest! on fragmentedice...• 73 9 Experimenh Jlymeasuredice/soil Iricticnccefficients ••.• ••.•• 74 10 Summaryofthe property values and rangesusedinthe theoretic al

model.•• • •••• ••• •.•. ....• •. .••... • ... 83 11 Inputvalues(orthe baseline rubble field. .•• .••.... ....•• 86

xii

(20)

SYMBOLS AND NOMENCLAT URE Ar Rubble field area.(m²) aT Delaved eluticity exponent (Is) b Const ant(0.34)

C Constant

Co Un-refrozenrubble compactioncreep modulus (m '/! N) Cd Drag coefficient

Om Refrozenlayercreep modulus (m3/ sN3) Os Un-refrozen rubble shearcreep modulus(mS/ sN3) D Structur ediameter(m )

Dc Compactioncree p damper (Ns/m) DR.L Refrozen layer viscous damper(N3s/m ) Ds Shear,creep damper (N3s/m) d Grain diameter(mm) dl Referencegraindiameter (1 mm) E Young's Modulus(N/m²)

Ee Un-refrozenrubble comp action modulus (N/m²)

E; Young 'sModulusfor ice (salineice) (N/m²)

E,.l Rubblefieldextent (m) F Force (N) F. Appliedforce(N) FL Force permeter width(Njm) F. Forceonstructu re (N) G ShearModulus (Pa )

Or Un-refrozenrubble shear modulus(N/m²) 9 Acceleration due to gravity(9.81 m/s²)

H Sallheight(m)

Hi Averageheightof rubble above water(m ) Hit Average depth of rubble below water(m) H. Averagesail height (m)

H", Water depth(m) K Constant

n Power law80wexponent(usual1:r~:Ifor ice) N. Number of sails

P Porosity (%) Q Activation energy(J/mole) q Heat 80w rate(J/s) q/ Latentheat of freezing (Jfkg) R Universal gas constant(J/moleK) SI,2 Shift function for Sinha'sequation Sa Compaction epring conetant (N/m)

xiii

(21)

5lj Stress vector(N/m²)

8m Refrozen layer spring constant(N/m) 85 Shearspring constant(N/m) Tt,u•.I2 Temperatures(K) TJ Freeboard thickness (m) Tm Refrozenlayer thickness(m) T. Submergedthickness(m)

t Time(s)

t; Ice thickness (m) v Velocity (m/ s)

W, Groundingweight of keelonseabed(N) W. Sail weight at underside ofrefrozenlayer (N)

"1 Shearstrain i' Shearstrain rate(/6) DAv, Averagedispl&eement(m)

Dc Deform ation atthe center of theindenter (m) Dm Displacement ofrefrozenlayer(rn) AT, Thermalgradient eC/m) e, Elasticstrain

t<l Delayed elasticstrain

tv Viscousstrain fij Strainrate vector 41 Referencestrain rate (/s) .\. Thermal conduct ivity(W/m k) flil Ice/ice frictioncoefficient fli. Ice/seabedfriction coefficient

flM Refrozen/ Un-refrozenrubblefrictioncoefficient v Poisson'sRatio

Pi Densityofice (saline) (kg/m^S) P", Densityof sea water (kg/m^S)

" Appliedstress(N/m²)

,, 1

Reference stress(1MPa)

"" Normalstress (N/m²) Shear stress(N/m²)

"J Failureshear stress (N/m²)

xiv

(22)

INTRODUCTION

Continuingdemand foroil andgas has pushedthe searchfortheseresourcesinto harsher and moreremote areas of the world,including the Arctic,Petroleumisoften foundoffshore but thepresenceof sea icein theArcticbas madeit difficultto use conventionalfixed or Haating explorat ion platforms.Consequentlyart ificialisland technologywasdevelopedto provideyear round stabledrillingplatform s.In 1973 thefirst artificialisland was builtinthe shallow wete reof theSout hern Beaufort Sea.

This was little morethan a leveled-offpile of fill material (sand /gravel )placedin 2-3 mof water.Since then, artificialislandtechnology has progressedfromsand bagre- tained islandsto the latestgenerationof caisson and caissonretainedislan ds.Caisson retainmentreduces thevolume of fill material and practicallyeliminateseroeicn and theriskof iceride-up. All these developmentswererequiredlUIexplora tionadvanced intodeeperwaterandmoremobile ice.Theincreased icemovementencountered by such islands resulted in a phenomenon not originally encountered, the formationo(

grounded ice rubblepiles around the island.

Ice rubbleis a compactmMSof ice blocks, fragments, snow and slush that is created when a moving icesheet (ails continuously against a fixedobject. Ifthe rubble is unable to clear around thestructure (wide structure)it will accumulate as a pile. The pile extends above and below the waterline and can become well enough grounded to resist subsequent ice movements from other directions.J(such grounding OCCUflla "field"of rubblepiles may accumulatearound the structure.

Initially,itW&8feared that grounded ice rubble fields might increase global loads on structure sby increasingthe effective structurediameter.However,subsequent ex- perienceand field measurement s have indicated that pressureserereduced,and it has become generallyaccepted that such (onnationsare in fact protective.Nevertheless, fullscale propertiesand pressuredist ribution swere not wellenough known to allow a

(23)

thorough engineeringanalysisof loadtransmission through such grounded ice rubble.

The use of externalcaissonsin artificial island construction,not only reduces construction time and costs but also makes island instrumentationmuch easier.By 1986 E550'sCaisson Retained Island(CRI) was equipped with 40 external ice pressuresensors, 10 geotechnical eenecre,and internalstraingauging,making itAllideal platform for the study of islAlld/iceinteraction.Forthisreason,during the winter of 1986/87.BSSO ResourcesCanada Ltd.collaborated withtheNational Research Council of Canada (NRC) and Memorial Universityof Newfoundland(M.U,N.)to conduct an ambitiousfield study of a. grounded ice rubble field.This wu done at ESSO's CRI whileon location at Kaubvik in the SouthernBeaufortSea. The col.

laborativeprogramwas setup to documentthoroughly a fullaccle casehistoryof sea. ice/ rubble/s tr uctureinteract ion.The collected datawes to be publicly available and was intendedto guide subsequent theoretical work on load trenemieeicn through grounded ice rubble.A wide variety of datawas collectedand the field workwes arranged sothat each groupWll.lIresponsibleforcollecting, processing,and doeumenting a certain portion of the data. This data was subsequentlyreleased inthree separate data reports ;Croasdale et al, 1988 (E5S0),Frederkinget al, 1988 (NRC), Jorda.an etaI,1988 (M.U.N.),followed by asummarizing paper by Marshall etai,1989.

(24)

2 OBJECTIVES

Asoutlined in theint roductionthe Kaub vik st udy was thework of manyin dividuals from several groups.The author'sspecific reepcnsibilitieaweretOj

•Assistwit hthe fieldwork,

•Precess,analyze,and document thefield data collected byM.U.N.,

•Reviewallthedata and otherliter ature,and

• Develop a theoreticalmodelfor calculati ngforcetransmission throughgrounded ice rubble.

The objectives forthis thesis werethus to;

1.PresenttheM.U.N.data and analysis,

2. Summarize the associate dliterature,

3. Describea theoreticalmethodfor calculating load transmissionthroughice rubble,

4.Com parethetheoreti calcalculations with thecase study.

Thepresence oftheaeparateM.U.N.data report gave the&uthor theluxu ry of beingable toconcen trateinthisthesison thedatamoredirectly concern ed wit h the subsequent theoret ical work,eincereferen ce forless critical det ailscanbemade to Jorda.a.netel,1988.

(25)

3 LITERATURE REVIEW

3.1 Previous Island-IceMeasurement Programs

Thefirstartificial islandin theBeaufor t Seawuconst ructedbyESSOin 1973(Albery etal,1984),only13win•.'re prior to the Kauhvik program.. Instru ment ationto measure ice forces and movements Mound island,immediately became important and most,ifnotallislandshad somesortof icemonitoring program . Man y ofthese programs ,however,weresetupto collectspecific_deai~ndata suchlUImaximumice pressu re,orfor saIetypurposes suchMwarning of icepressurebuild-up (Tem pleto n III,1979). Act ually analyzing movement eventsand ice pressures,bycorrelating them wit h icefailuremodes, wind data,currents,and otherenvironment al forces wasofsecondar y importance;andin manyC~grounded rubbledidDotform.The resul ts ofmanyofthese programs arealso confide ntialand this somewhat reduces theliterat uretobereviewed . Thepubliclyavailable literature on previousisland / ice fieldmeas urementprograms has been reviewedand is sum mari zed cbronologiullyin table1.

The mainobjective ofmany ofthesemeasuremen t programswas to collect design relate d data, suchasicethickness,movement , pressure,and globalloadson struct ures . Thelarge number of groups involved resulted in theusec f&widevariety of instrument types and meaaurement procedures,whichsometimesproducedconflict ingresults. An enormousamount ofdatawu generatedwhichcan Dotbecomprehensivelycondensed here. Itis,however,intendedtosumm ariz eenoughinformationsothattheimportant featuresof the Kauhvikdatacan heassessed. Thesources list edin table1,provide the followingtypesof dat a:

(26)

Table1: Summ&r)' orisland/ ice fieldmeasurement programsdescribedinthe open literature.

YEAR ISLAND NAME TYPE SOUR CES

1974 Adgo F-28 sand/gravel Nelsonetal,1974 1975 AdgoP-25 sand/gravel Metge, 1976;Nelsonetal,1976 1975 Netserk8-44 sand/gravel Met ge, 1976;Nelson

et al,1976; Kry, 1977 1976 NetserkF·40 sand/ gravel Strilchu k,1977 1977 Arnak L-30 sand/gravel Semeniu k,1 977 1971 Kannerk0-42 sand/gravel Semeniuk,1977;FlLvrat , 1982 1980 Fairw ayRock natural Kovacs et aI,1981 1979-80 Issungnak sand/gravel Fence Consult&DuLtd.,1981;

Frederkingetai,1982;

McGo nigal,1951:McGonigal et al, 1986;Shinde,1981; Shinde etai, 1982:VlLudrey,1980 198~83 HILDa Island nlLtural Daniel ew:icz et al, 1981,1982,

1983,1988;Metse etal, 1981 1982-84 Tanuit aluon DePaolietall1983;Pilkington et

al, 1983;Sande~n,1984 1982-85 Ada.ma Island nat ural Frederkinget al,1983,1984,

1986a..1986b,188& ; Sayedetal, 1986&, 1988;St&nder,1985

1984 KlLdluk caisson Johnsonetel,1985

1985 Amerk ~09 caisson Sayed etal,1986b; K.RCroasdale okAssoc.Ltd.1985 1985 Mukl uk land/gravel Cox etai, 1988 1985-86 Molikpaq caisson Wrish tetai, 1986

1986 Miouk land/ grlLvel K.R.CroudaleIt'Alsoc.Ltd, 1986

(27)

Ice Pressures: Small pressuresensors,large pressure panels, biaxial pressuresensors,struct ure mounted panels, strain gaugingof stru cturalmembers.

Ice Movements: Wire linereels,radar,traditionalaurvey method s, elect ronic distancemeasurements, time lapsephotography

RubbleMovement s: Electronicdistancemeeaur emente,stll.rldpipe instaUat ions,inclinometel'3.

Ice/Rubble Strains: Electronicdistancemeasurements (large scale), smallstt ain arrays.

Rubble Geometry: Aeria lphotography,levelsurveys,auger drilling,electromagnet ic sensing,thermistor atrings.

Other: Anemometers(wind speed,profile,direction),tide gauges, cu rrent meters,borehole jacks,amall scale properties measurements , water temperature,geotechnical measurements,islan d mounted inclinometers.

The collecteddataisvariablein almOlitevery respect .Prior to1982,useful pressure data withinrubblewesonlyobtainedat twolocations,Netaerk B·44,and Arnak 1-30.This,and subsequent data fromcaissonstru ctures, indicatestr.,,~external ice loading does resultin pressureatthest ructure, but itIegenerallyreduced.Measure- mentsat Tarsuit also showedmeasurablepressure at depthsgreaterthan 4 m below thewaterline.At manyofthesites no significantrubble fieldfonn ed;tber.e location", falI into twoc1asllC3.

1.Islandsclose toebore,insballowwater,which are not subject to large ice movements.

2. Steep-sidedstructuresin deep water where accumulationsofrubble are unable to ground firmly.

Thus, the struc tu res moetprone to generate extensiverubblefieldsare those with gentleelceee,and farenoughoffshore(>10km) tobesubject to large ice movements.

Structuresin this category aresand/gravel islands and shallowcai8lOns caca derwater berms.

(28)

Table2:Result!from augerandthermi stormeasurements of refrozen layer thickness.

AUGER THERMISTORSTRING DIFFERENCE

1.66m 1.92 m

+

¹⁶^%

1.70m 1.30 m -24%

2.69 m 1.26m -53%

Thelargestrubblefieldreport edto haveformedaroundanartificialislandwas 1500 m x 800mand formedat Issungnakduring thewinter of1979·80.Aspartof theIssungnak fieldprogram,however ,a rubbleaccumulationatanaturalshoal was alsostudiedand thisreachedB900 mx6900m,

ReCrozenlayerthicknesswasmeasured atIesungnakby"feel"duringauger profiling, and byusinginsituthermist orstrings. The maximum thicknessin February 1980,measuredby a.uger, was 3.7 m.The differen cebetween the therm istor data and theauger data was alsomeasured at three locat ionswith the resultspresented intable2.

Itis apparent Cramthesereeultethatthe two measurementmethodscanproduce very differentresults. TheuseoCthermistorstringsalso allowedthemeasurementof refrozenlayer growth rate.Onaveragethesewere:

Ncv • March 6.0nun/day(growth ) March -June-16 .0mm/d ay(melting)

Thelargest rubble nilswere measured at FairwayRockand Mukluk Island,both of whichbad rubble piles up to15 mhigh.On average ,the maximumsail height.at otherlocati onswas lessthan12 m,

Rubblemovement sweremonitoredat threelocations,Issungnak, Tarauit, and Ame rk (ESSO's eRI).Movements were detec tedatall locat ionswithshortterm

(29)

[hcu ta-daye] movementsas largeas 0.3m.At Tarsuit, movements weredescribedII!

being tens ofemat therubblefielriperiphery, and less near thecaisson.

Islandmovementwa,.,measuredatAdgo P·25 usinginclinometerswhich.indicated that atotal movementof 5 cm occurred. This wascalculated byM. Metge (1976) to require anaverage globalforce of 984 kN/m£rom the surrounding ice.

Interpretationof icepressure sensor data is notcompletelystraight forward and var ious asaumptlonaare usually made,in orderto calculat e such parame tersII!global load.Forexam ple, average pressure canbeobtained by combiningtheoutputfrom several sensors. This is used togetherwith icethickness measurementsto calculate themaximumaveragethrust ,butbecause bott om iceisrelative ly "warm"andsoft, a reducedthickness is often usedin thecalculation.The maximumpressurerecordedin floating seaice was measuredat Netserk B·44,wherea peak pressurereached 1.5 MPa.

A maximum Averagepreeeureof1.1 MPa wascalculated Irom datacollected atArnak L·30.to general ,however, maximum pressureduring anyseasonrarely exceeded 1 MPa.At several locationsin landfaatice, the maximumthrustwas associated with therm al expansionof the ice.

Icemovementswere monitoredas part of most of the field programs.This wu considered important because standard indenter equationa allow direct calculationof icefcrceefrom the indentation geometry andice velocity. Increasing ice velocityia associated with increasing iceforces. During non-landfast conditions, ice movements can easilybe ofthe order of kros/day ,but upon becoming landfast,theice ia greatly restrained. Wind,currents, tides,andtemperat ure haveallbeen found to influence landfast ice movement,withthe magnitudeof movement....180 being influenced by locat ion. At Adams Is::and,3 km offshore,average ice velocitieswere leasthan 10 cm/day, while atNetserk F·40,some32 km offshore,average velocit ies wereof the orderof1 m/ day.Attempts tocorrelate drivingforceswith icemovement hAvegiven

(30)

variableresults. At Adgo F-28.movementswereclaimedto correlatewithwind, although nosignificant movementoccurredduring two of the largerwind events.At somelocat ions tide effectsdominated,whileat others,thermalexpans ionappeared to be the major contributor toice movement.Aninter esting mechanism calledtidal jacking,was proposed to accountformovements at AdamsIsland . Tidal jacking would appear to be a processwherebyw-ter-filled,wedge-shapedcracks,formed by bendingfai lure of the ice sheet nearthe shore,freeze,thereby causing the iceto move horizontallyduring tide reversal. It isapparentfrom theli:ereture thatthe process of landfastice movementiscomplex and cannot yetbe accurately predicted.

3.2 Load Tran sm is sionModels

The force requiredtomove (causesliding failureof) a groundedicerubble field over a levelseabedis universallycalculated&5;

(1) where:

pi,

=

ice/seabedfrictioncoefficient

W,=groundingweight and canhecalculatedas~1)1l0W8i

W.=

(1- fa)

^[H;·"^{- H. (, .}^-^{, ol)}^·^g·^A,

where

P

=

^rubble^porosity(%)

H;=aver&.ge heightof sail ice abovewater(m) H.=nerage depth of keelice below water(m) Pi

=

densityof ice (saline )(kg/m3) P",=density of sea wate r(kg/m3 )

(2)

(31)

9

=

accelerationdueto gravity(9.81m/ s") Ar=Horizontalarea of rubblefield(m")

Ifone assumesthatP,Pi,PUll9, andJJ;, arebasically constants, the ability ofa rubblefieldtoresistsliding depends on threegeometricvariables,fl.,HI..andA•.

Sliding resistan ce increases withgreatersail height , shallower wat er,and larger area.

This is straight forwardbutit shouldbe notedtbatare a.is veryimportant.Sea ice loads against a stru cture(orrubble field) are inessence calculatedea follows;

F=FL·D (3)

WhereFLis force per metre diameterand Dis thestructure (or rubblefield) diamete r. The applied force there forevariesindirect proportionto thediameterwhile the sliding resistanceisproporti onalto thediametersquared. Thusanygrounding pressurewill producea stable rubblefield,provided that the field islarge enough.

Slidingstabilitycalculationsofearly sand/gravelisland streat ed the rubble-island systemas a singleunit inorderto calculateglobalslidingresistance.Thisapproach has since beenexpandedin order tocalculat eload(F.)transmitted to a structure throughgroun dedrubble[Kry, 1977, 1980; Allynetal, 1979,1982;Bercha et el,1980, K.R.Croasdale and Associates, 1985).Thismethodusesa rigidbody assumpt ion for therubblefieldandshall be referred to hereas thesimpletheory.The rigid body assumpt ionimpliesthat noloadwill reach the structure untilglobalrubble sliding occu rs, at which point ;

F.=Total external force - Rubblefield slidinA: resistance Thismethod of calculati on hasbeenused invariousways. K.R.Croasdale&:AsllO- dates(19M)useda.two-dimensionalapproachand considered("DIy therubb ledirectly

10

(32)

between thestructureandthe applied load. Kry, Bercha and otherstook a three- dimensionalapproach,consideringthe entire rubble field. The three-dimensional approach led to concernsthat a poorlygrounded rubble field couldincrease global loads on a structureby increasingtheeffectivestructurediameter. Allyn andothers extended thethree-dimensionalapproachby including the berm slope,edge failure andfrict ion (shear)along the sides of therubble field in a computerprogram which calculatedtheloads on a circularsand/g ravelisland. Sayedetat(1984)includedan assumption aboutice rubble properties.Theyeasumed thatnewlyformed ice rubble is a vertically and horizontal lyhomogeneousMohr-Coulombmaterial at criticalequi- librium.The stress distrib utionwithinthe rubble was calculated for a narrow rubble fieldagainsta.structure(two-dimensionalcase) butalthough this approach was more complex thanprevious models, the calculated loads were stillzero untilthe applied load exceededthe totalrubbleslidingresistance .

None of these analyses take into accountthepossibility thatrubble field defor- mationsunder loading may generate (orcP.8 againstan embedded structurebeforethe sliding resistanceisreached. Inordertoovercome thelimita.tionsof the rigid body assumption, severalgroupshave resorted to numericalmethods for calculati ngload distributio ns.

Evginand Morgenstern (1984) ledtheway,usingfinite elemen tanalysis to examine the behaviour ofan eight sided,caisson ret ained, island(Tarsuit ) when subjected toiceloads.They included the case ofanintactice sheet resting ongrounded icerubble, andfrom theirresultstheyconcluded tbat the presenceof arubble fieldreduces theamountofdisplacementandincreases the overallmaximum force an islandceu withstand. However,uponexami ning theirmethods itis apparentthat this conclu- sion is subject to severalqualificat ions.Itappears tha.t onlyone20elMticCMewa.s examined, for whichanarbitrar y value forrubble elasti citywas used.It alsoappears

11

(33)

t.hat. tbepossibilit.yofslidinga.t. therefrozen/ un·refroun lniertece and rubble/ seabed interlacewu not ccaeidered; a.nd theint.a.ct icesheet(refrozen layer) wa.sesaumed to bethe samethickness as the surro undingseaice.Thi. study cleulybrought fort.h the conceptthatanice rubble fieldca.nbetreateda.sa non-rigidbody,butgeneralization.

from th e results mustbemade very uutiously.

In1986,Williams etelused numericalcomputersimula.tiontoeelcel aeridge and rubblebuildingforces.Theyuseda discreteelement methodtosolveincre mental ly the dynamicequilibrium equa.tions foreachoftherubbleblocksaswellas forthe advancin gicesheet,Tbismethod issimilar to finite elementanalysisexcept thatitis usedforsystemsof discret e bodies.Thereweretwoexamplespresented , of icerubble accumu lat ionin front ofastruct ure,bothstarti ngwithaninta cticesheet,Each example wurun foronly 100 seconds andalthou ghnoground ed rubblefield formed (a t mosttwoblocksweregrounded)itisobviousthatlongerrunswould haveproduced moregroundedrubble.Thisapproachrequirestheinput ofmaterialpropert ies,a.s wella.sinit ial a.ndbound&r)'conditi ons. Deformati on.wit hin therubbleare thus calculated and, withcorrectinpuu,thUmethod should give reason ableresultsfor newlyformedrubble.

Finally,in 1988,Ca.na.dianMarineEngineeringLtd.carriedout afinite element a.nalysisto examineloadtransferthrou pa grounded rubblefieldwitharefrozen 1a.yer.

Onlythe 2D,rigid.tructurecase was examinedwit h an 80m widerubblefield.Itis int erestingtoDotethatthere£rozen layerwasauumednot toreach tbe outer edgeof the rubb le field.Twocueswere presented,the differencebeingthehorizontalextent ofthe refrozenla.yer (7.5m and 78 m).The Young'sModulusfor the consolidate d (re£rozen)layer wa.s auumedto be0.2OPa.,someten times largerthanthe value usedfortheun-refr ozenrubble.Some effort wasmadeto detenn ine theseproperties (romtheoret ical coDeiderations,a.swell as £rom smallsuJetestdat a. Unfort unately,

12

(34)

problems were encounte redwiththerubble/seabed slide r elementsand so theywere removed.The rubble wast.hus fixed to the seabed ,and so it was estimatedthatthe finiteelementanalysisreeultewere inerror by morethan an orderofmagnitude. The geometryand justi ficationof materialproper ti es weremoreadvanced tha nthose used by Evginand Morgens tern(1984),but theslider element problemsandlimited cases examinedmeantthat fewgeneralizationscouldbe made.

13

(35)

4 SITE DESCRIPTION

4.1 General

Section 4.1givesan overviewofthe environmentalconditionsinthe Southern Beaufort Seawit h site specificdet ailsprovidedwhereeverpossible,whilesection4.2 reviews the data collectedbyESSOand NR Cduringthe winte rof 1986/87.

4.1.1 Wea th er

AtTukt oyaktu k the average airtem peratu reisbelow0 degreesbetween lateSepte m- berandearlyMay. The average February airtemperatureis-29 degrees andwinte r temperaturesare persistently ratherthanext remeJy cold. Thereis littleprecip itation and the averageannualsnowfall at Sachs Harbouris75 em/ yr. Prevailingwindsate fromthe nort hwest andeast -southeee t withthe one houraveraged,50yearreturn period wind ,being105 kmh.Thewontmonthforstonn sisOctober (sto rms present 8%ofthe time)and duringthe remai nderof thetimelowpressure systems arepresent 2·3% ofthe time.meaning thattheBeaufortSea isnotasto rmy area (P ilkingtonet a1,1983).

4.1 .2 Watermovement

Wate r movementinthe Southe rnportionof the Beaufortseais influenced by the westwardmovingBeaufortGyre,the northeasterlyflowof the MacKenzi eRiver, and theeasterly flowofthe 'Trans-PolarDrift .Wate r current velocitiesareusually very low andonly extrem e currentsexceed50 cm/sec(1knot). Wind induced watermovement atthesurface maymask thegeneral water flow andmayinduce strongshortterm currents . Tidalfluctu ations aregeneral lysmall andaverage IS em.although in 1944 anext remestormsurgecaused waterlevelstorise3matTuktoyaktuk (Pilkington etal,1983).

14

(36)

--~ -9~ ·'"CONF IDENCE LIMITS

DAYS SINCE SEPT. l

Figure1:Ice thicknessversus time after September1.Datais from'I'heaiger Bay (SachsHr.](from Parker, 1987)

4.1.3 Ice Conditions

Freeze-up usuallystarts by the first week inOctober and the maximum ice thickness of 2 m is reached by themiddleofMay(see figure 1).Themeasured ice thicknessat Kaubvik was Ui m by April10,1987 which is less than usual for the timeof year.

The Kaubvik site is usuallylandfastbymidwinterand then remains landfast until the springbreakup at the endof June (McKenna et el, 1988).Between the start of freeze-upandlandf""tconditions thesiteissubject to moving first-yearpack ice withoccasional multi-yearfloes andridges. It is thus duringthe initialnon-landfast conditions,whenice movementis of the orderof kilometers per day (Spedding, 1979), that rubble fieldsform.

Duringthewinter of 1986/ 87Ka.ubvik became lanclfast bythesecond week in January. Landfastice,although generallystable,does move and movement rates in 20 m waterdepths can reach 3m/hr (McKennaet al,1988). These majormovements have been associatedwit hcoastalstorms (Agerton et al,1979).

15

(37)

4.1.4 Ba t hym et ry

Theaveragewidthofthe cont inental shelf is 70kmand ext endsto the 50 m depth contour.Kaab vlkis30km from the nearestshore, in 18.5mofWAter,andisthus wellinsidethe out eredge ofthecontinentalshelf.

A circular, 8at topped berm10mhigh was builtunderwater,ontowhich the caissonring was set down.Theouts ide designdiameter of the berm was380m, sloping upto a148m diameter level surface forthecaieeona.After construc tion of theberma bathymetr icsurvey wascarri ed outgivingthe berm contour mapin figure2.

4.1.5 Geotechnica l

Core samplesofthe sea bed atKaubvik weretest edbefore constructionofthe berm inorder to calculate island settlementratesandsliding resistance.Thebotto msed- imentsconsistmostlyof soft clay with some sand layers. Thebermmaterialwas obt ained from Issigak and consisted of sand and fine gravelwith a lower bound shear frict ion angleexpectedto be34~(Shinde, 19M).

4.1.6 Ca isson Struct ure

ESSO'sCRI isoneoftheearliest arctic caisson islandsand was first used at Kadluk during1983/ 84. In 1986it was stationed atKa ubvik,some 120 km northw est of Thktoyaktuk(69" 52.5'north,135~25' west).The CRI consists of8,water ballasted, steel caissons arrangedin a ring and heldtogether by tensioned wireropesL"sbown in figure3.Eachcaisson is 49.2mlong,13.2 m wide and12.2mhigh,with a mass of 5000 tonnes (K.R.Croasdale andAssociates Ltd, 1985).The centralcoreofthe islandis filled withsand which providesthe workingsurface for drilling operations, as well as sliding resistanceagainsticeforces.The designglobalice pressureis 1700

16

(38)

- WA'n'RlltPnl, m - --- GROONO£OIlU891.(

aOUNDAIIY

o 100m

' - - - J SCALE

Figure 2:Bat hymetriccontour mapof berm at Kaubvik(fromFrederkingetal, 1988).

kPa(global loadofapproximately45000 tcanee)and thedesignlocal ice pressure is 4800 kPa(Croudale et ai,1988).

Whenthe caissonrin,is set down it is preciselypositioned30that caisson1 faces directlynorth(Croudale etel,1988). The CRIwas originally designedto include33 iceload senson,variousgeotechnical eenscre, as well as structuralresponse sensors(Hawkinset ai,1983).By thewinterof1986/87 32iceload senson and 10 geotechnicalsensors~ereACtive.

4.2 Ot herKaubvi kdat a

In orderto present acomplete case studydescription,the other twoprojectdat a report s{Crcesdele et al,1988andFrederking et al,1988)aresum..narizedhere. As withtheM.U.N.data, themain emphasisis to describethe resultsthat are relevant to thesubsequent theoreticalwork.

17

(39)

£ ..

==bl.- - - - -

(b )

(0)

-' m

SEA FLOOR

Figure3;Details of theCRJ showinga)pla.n viewofthe caissonrin~.b)cross sectionthrougha.caisson,and c)cross sectionthrough a completedisllLll.d (from K.R.Croasdale and Associa.tes,19M).

18

(40)

4.2.1 ESSOData

EssaRescuzces Canada Ltd wasresponsible for operating the CRI mounted sensors.

The sensors of concern for this projectwere the external icepressure sensors and the sand core geotechnicalsensors. Analysis of the collected geotechnical data showed that ice forces did notsignificant ly affect the measured pore pressures, so the following summary only deals with the ice pressure data.

By the winterof 1986-87therewere two typesof ice pressure sensor, microcells and shearbar panels.Microcellsaresmall (0,02m¹)and weredesignedto measurepoint pressures while the shearbar panels (1.05 m1 )were designed to measure pressures averaged over alarger area. Duringthe fieldperiod, 27 microcellsand 5 shearhar panels were operating,with the greatest concentrationof sensors being in thenorthern quadrant.Althoughpressuresfromthe north(open ocean) were expectedto be the greatest ,allcaissonshad some sensorsand the southerncaissonshad a total of 8 operational microcells(see figure4).The microce1lswere mounted at the waterline and weldedHush withthe outercaisson pLating. The microcells have thefollowing specifications:

Type: Straingauged cantileve r diaphragm Diameter: 16.5em

Full Scale rating: 6895 kPa Accuracy: 5% full scale (340kPa)

Until drillingoperati onsceased in earlyJanuary,the sout hernmicrocell dat a was sampledevery10 seconds.Immediate processing was done toreduce data volume and only the 5minutepeak, average,andvariance values were act ually stored. After shutdownof drmingoperat ionsa battery powered remote dataacquisition systemwas installed whichrecordedpressuresat a fixed loggingrat e, without any preprocessing.

This systemoperated untilMay.

Althoughthere were no ehearbar panelsinthe southernquadrantit is interesting

19

(41)

Figure 4: Planview of eRI showing ice pressure sensor locations (from Crcesdale et .1,1988).

to comparethe outputfromthetwo sizes of sensor. Figure 5 containsthe maximum peakeRIpressurerecorded during the entire winter (caisson 8). The dropin pressure fluctuationsafter October 26 appearsto indicate the presence ofan icerubble accumulation.In comparison,the maximum peak pressuremeasured by a shearbar panel OD CaiSSOD S was less then0.5 MPa (see figure 6).This is due to the difference in sensor area. During ice failure,direct ice-structurecontactsare transient and lo- calized.The small microcellscan frequentlyhavetransient ice contactoveralarge percentageofthe sensor area, whereasthe shearbars will averagethesesmall contacts overamuch larger area.The microcellout putis therefore moreerratic butthe time averaged pressures on bothshouldbe the same.During non-failureevents(icecreep) the output from bothshould be similar.

Once ice rubblehad formedagainst the eRIno compressive failureof therubble or refrozenlayer was observed at the CRr/rubblecontact.Pressuresm~asuredby the microcells duringthe latter part of the wintershouldthusberelat ively independent

20

(42)

~

l JII~' ¹

^I,.". ^~

h ~-

I

Figure 5: Microcelloutpu t coveringtheper iodofhighestmaximum pressure[from Croasdale etal,1988).

21

(43)

J A,l. ^~.

) '\ ^('\

~

I

Figure 6:Shearbar datashowing one of thehigh pressure events.Note that although th e pressuremagnitudeis eimiler to that measured bythe micrccell00the same date, theshearbar showsthe event much more clearly because of the increasedsensitivity (lower fullscale){fromCroasdaleet al, 1988).

22

(44)

of sensor size.Thetime ofgreatestinterestis theperiodc<ncidingwith theM.U.N.

data,March 12•May:i. Duri ngthistime, however, verylittleactivity was det ect edat the southern quadran t .Itappearsfrom theplots (seeappendix1) that theresoluti on ofthe data was±60kPa which makes it difficulttodetect small pressurevariations.

Consequently, most of thelate winterdata is straightlineanddose to zero(see appendixA). Themostdramatic exception occurred onApril15th, when both sensors active on caisson4measuredpressu res exceeding1.0MPa.ThisIs-ted about 2hours andalthoughno simi larchange in pressurewas observedat any of tileothercaissons, itappea rstobeanactualeventratherthen an equipm entmalfunction .Similarly, earlywinterpressureeventswereusually dete ctedbymorethan onesensorbut often not by all the expectedsensors. Interpretat ionof this data with regardto average pressuresis thereforenot so straightfo rward.

Inconclusion,no pressures over300kPa were recorded atany ofthesouth ern caissonsduringtheM.U.N.fieldperiod,except for theApril 15th (julienday105) event,andingeneral, pressur eswerelessthan100kPa. Thelow pressures,combined with60 kPa resolution ,make it difficultto comparethemicrocell data withother data on anevent by eventbasis.

4.2. 2 NRCData

The NationalResearchCouncilof Canada wasresponsiblefor collecting rubbl efield pressu repane landsurveydata.Fieldwork started inlateDecember1986 andended in ear lyMay. Fourteenpressurepanels wereinst alled inthe rubble in 4 groups[ar- rILYs).As thesewereinst alled while the sea icewasstillactive(not landfast ) three ofthearrayswereplacedto the northwestoftheeRI(see figure7),whererubble form ati onWIISmostlikely.The fourt h arraywas placedto th esoutheast to coincide with the M.U.N.instr uments.As expecte d,ice movementcausedrubble building to

23

(45)

I~

.. "'~'f': lDEAlP...elS

:~~'i~(,;G~~ .

"."lUI

\<.1 ....:::

E~'Oi'IP'''''lS lel;\~·~'~

^{..IU I}

.. SUlI\IEVflEFlecrOll ,. ST~e$SPAllln.(INOlCAtlNG OAlEOI1AllOl<) - -lEYU SURveH INE, Eg.I""

Figure 7: NRC panel and surveylocat ions inthelate winterrubble fieldatKaubvik (from Frcderkinget a.1.1988).

OCCUItwice in theweek following panel installation. One event extended the rubble

fieldeastwar d but the secordeventoccurredalongthe northwestedge of therubble

field.New rubble even overtoppedold rubble,buryingthetwo outermostpressure panels. Tileresulti ng internalrubble pressureswere successfullyrecorded and demon- strate that sustained pressuresof severalhundred kPa. can betransmitted intothe rubble, see figure 8.As figure 9 shows,however.the distributionof pressures within the rubble was notstraightforward.The greatest pressures wereactuallymeasured somedistanceback from the rubble field edge.Many factors couldcontribute to this, includ ing local refrozen layer bending, incomplete sensor freezein, a thinner refrozen layer in the newer(extern al) rubb le,orit could indicatethattheCRtwasa. etreee concentr ator .teeprofile drilling doneat that timeindicatedthat the"consolidated layer" was about 2.5m thick,and althoughthis was measuredby"feel" ,ratherthan temp erature measureme nt ,it is still a good indicationthata substantialrefrozen

24

(46)

.

_TIME.JULI_ANOAYS

00

e~£NSORI

00

-

00

j

^/"SENSOR&

,

,I.-

, ...

"

_f/S£"OR'

^SEHSOR2

a

~':~--- '

00

"

-I t.a

Stressvers ustlll1t, north-westEuQnlll nels.

Figure8:Inter na l rubblepressuresasmeasuredby the Nort hwestExxonarray durin g theJanuary8th event(fromFrederking etaI,1988).

Figure9:The pressure distribu tion at one time durin gtheJanuary 8th rubble build.

ingevent(from Frederking etal,1988).

25

(47)

Figure 10: Arctecpanel rosette data[from Prederkinget al, 1988).

la.yerexisted at that time. The Arctec panelrosettenearthe caisson recorded almost equal biaxial stress (see figure 10)indicating a maximum principlestressonly20·30 kPa greater thanthe minimum principlestress.

Some ofthepanels were dividedintohorizontal segmentsanddata {rom these indicatedthat vertical pressuregradientsexisted (see figure 11). Unfortunately,over time,zero drift and apparent malfunctions reduced thereliability of the datu..The southeastarray dat aofgreatest interest isthatwhichcoincides with the M.U.N.

data(March .May).By thattime,some pressurevariations areapparentbutthe magnit udes are consideredto be unreliable.

NRCalso collected 4typesofrubble survey dat a; rubble elevat ions,rubblefield movement , rubble profiling, andrubbleblock size, Standardsurveying techniques were used to measure rubble elevat ion along 7 survey lines (see figure 7).Atypical resultis shown in figure12 and fromthese itwas determined that theaverage sail heightsoutheastofthe CRtwas 4m. Thiswas calculated using equation4 to ensure that averageheightH,couldbeusedtocalculateaveragenilweight(assumesimilar

26

(48)

600

'VrL

'00

s

'00

.,;

v~

~

-' 00

~f

-600

,

12 TIME,JUti ANOAYS

I'

²⁰ ²⁸

"

Figure 11:Data.fromhorizontallysegmentedpressure panel(from Frederkingetal, 1988).

sail shapes) .

where:

H=Sail height N.

=

Number ofsails

(EH ')'·'

H.= N; ⁽⁴⁾

;

J

The highestsail inthe rubblefield(noton asurveyline)was about9m high.

!lori ;tonta lDistanc e(m) Figure12: Surveyed rubble elevationssoutheast of the CRY.

27

(49)

Figure 13: Electronic Dista nceMeasurement survey results showing horizont al movementand(in brackets)settlem ent!over the win - (fromFrederkiagetal, 1988.

Inordertomeasurerubblemovement,15survey posts weremounted in the rubble inearly January (usually atopsails).The horizontal andvert ical positions ofthese postswere measured atvarious timesthroughoutthe winter using anelectronic dis- tance measuringdevice(EDM). The results are shown in figure13,from which it is apparent that the rubblesettled at alllocationsand, with one exception, the rubble movedhorizontal lyin anortherly direction.

The horizontal movementis also, with oneexception,away from the CRI and down the berm slope. Horiecat al movement of up to 0.33 mWi13measured but unevensettlement (tilting) ofrubble sails mayhavecontributedto this.

Rubbleprofiling was done by drilling holesinto the rubble. This showedthat "val- leys"between sails generally reached within 0.2 m of sea level,and theconsolidated layer in Januarywas 2.5m thick in places. A floating block at a depth of 5.5m, 25 m from theeRIalso indicated that not all the rubble wascompletely grounded.

Therubble blockmeasurementsshowed that 0.2 m thick ice fer-ned groundedrubble 28

(50)

in 8.5 m deep water,the rubbleclosest to the CRI was the oldest (thinnest blocks), and rubbleform at ionappearedtoheve ceased when thesea Ice reacheda thickness of0.75 m (thickestblocks).

29

(51)

5 FIEL D RESULTS AND DISCUSSION

5.1 General

The rubble field finished growing when the surrounding ice becameJandfast in ea.rly January,withthe majoraccumulatio nsof rubblebeingto theeMtand northwestof the CRI.The first fieldvisit was scheduled (orMarch and so,with. the CRt so heavily prote ctedonthe eastand northwest sides, it was decidedto concentrate efforts to the south(see figure14).The rubblemounds to the south were not so extensiveor massive,andthe rmal expansion of thelandfast ice sheetduring thesecond half of the winter was expected to result inloading fromth&.t direction.A total of three field tripsWallrequired ,covering atwo monthperiod startingin early March 1987. The M.U.N.team collectedthe following data.

1. Sea icest resses from 9 mercuryfilled stress sensors.

2.Creepandstrain in the ice rubblefrom one strain array.

3. Rubble thickness from 8 augerholes.

4.Rubble temperatur eprofilesfromtwo thermocouplearra.ys. S.Torque readings from one10minstrume nted auger hole.

6.Iceproperti es from 9 cores.

Inadditionto this,'fuktoyakt uk air temperature, wind speed,andwinddirection data were obtain ed from Envitl""':mentCanad a(see Jordaaneeal, 1988).

5.2 SeaIceStresses

Nine stress sensorswere arranged in three rosett es,with cne8e090'ineachrosett e orientedtomeasur e stresses towards&surveypoint00Cai48004. Each mercury

30

(52)

LEGEND

aSTRESS SENSOR ROSETTE .. SURVEY POST INRC) - PRESSURE PANEL (NRC)

• RUBBLE PROFILE HOLE OAINSTRUMENTED AUGER HOLE eTTHERMOCOUPLEARRAY

4 STRAIN ARRAY

t

C 3

~oo SCAL E 1m}

Figure14:Plan view orCRland rubblefieldwiLh the positionor theM.U.N.field meuurementlindiuted.

(53)

Figure 15:Diagram ofmercury filledpressure sensor.

filled,diskshaped, sensorwas attachedtoa pressure transducerwhichmeasuredthe pressureof the mercury(see figure 15). The rosett es werefrozenintothe sea ice about240 msout heast ofthe caissoncent re and about50 m apart.The sensorswere thus about 90 m south of the rubble field edge.At the time ofinst allation the sea icc was coredandfoundto be1.5 mthick.The neut ralaxisofsuchasheetislocated at a depth of approximately0.£ m(Duckwort h etal,1988) and so thiswas the chosen deployment depth.An electric chainsawin analignmentjig was usedtocutthe slot for each pressure sensor.The slotwas filledwith freshwate r from a vacuumBask,the sensorwaslowered tothe bottomof the slot,and the waterfroze withinoneminute.

Two thermocouples were also installed,onein the dataloggerand one atadepth of 0.6 m in theice.The datawereread and recordedevery2 minutesby aCD248data logger poweredby a thermal generator.

Over the31day measurementperiod thedata. from thenine8tr~ssensorsand 2 thermocoup leswereread 22,353 times.Unfortunately,an intermittentfaultwhich

32

(54)

c s

, ,

" -J

c

V I'

s

ro ~ ec es so

" " '" '"

⁰⁰ ^'"

'"

^es

JULlA~-'AY

Figure16: Theprincipalstress anglefor array 2.

has been attributedto the data logger, causedlar ge negative values to be recorded onoccasion,mainly withrosette number three. This only affected the stress sensor channels and resultedin alose of 16% of the stress readings, lea.vinga total of 168,000 stress readings and 44,700 temperature readings. During subsequent data processing the erroneous negative pressureswere discarded. Plots of stress versus time were made for each sensor and the principal maximum and minimum stresses,stress angle, and shear stress for each rosette were calculated (see Jordean etal,1988). The lar gest measured stresswa.s MO kPa which corresponded to a maximum principal stress of570kPa.The principal stress directionforall three rosettes was generally northwest/southeast (see figure16).

The rosettes were arranged so that one sensor in each (labeledsensora) meeeured the stresstowards a survey point on caisson 4. The average of these stresses has been plotted {or comparison with theeRrsensor data.. As each of these sensors faces approximatelyNW thesecan be usedto estimate the averagemaximum stress (see

33

(55)

,I,

~I ~ ^'\~

^'~

, II{ \

. ^l~ ^\W

"

. ..

"

JULlANOAY

Figure17: Stresstowards the CRI, averagedfromthethree rosettes.

figure17).Thismethod is expectedto beaccurateto within 10%andthis approach is required becausedata loss from sensorsinrosett e 3 (other than sensor3&)resulted inno principalmaximumstressesbeing calculatedforthelast weekofthe fieldperiod.

From the pressure data. itappears thatthere weretwo types of deformationof the sea ice asit moved past thegroundedrubble(see figure18).Shorttermfiu<:tuations of up to200kPa (see day106) are typicalofa britt letype of failuremode,while slowly varying pressure(days 110 to113) indicates a ductile(creep)type deformation. On the lastfield visitthe fresh ice deformationshowninfigure 19was observed aoout100 m NE ofthestres s sensors.Thisappearsto beabuckling typeof failureand matches description sofsimilar features at Adams Island (Frederki nget at,1983).

The above waterportionof theice sheet is visibly curvedand as thefingerlike projectionsare not indicativeof a fast fractu retypefailure itis hypothesisedthat the ride-upof thisice "tongue"occurred very slowly.Althoughthe featu re infigure 19 was locali zeditwas one or at leastfivefresh verticalice sheet movement!on the southern edge of the rubbleenditcouldaccount forthemeasured fluct uations in ice

34

(56)

JULIAN DAY

Figure18:The expanded plotof stressesmeasured by stress sensorSi a clearly shows tha.t periodsofstead y pressurewere interru ptedbyrapidlyfiuctull.ting,a.pparently brittl e, deformat ions.During theent ire periodthemaximumprincipalstress didnot reach 400kPa.

pressureoverthe daysprecedingthe photograph. The deformation required atleast 3m oficemovementandthe sudden pressurefluctuationsataveragestr essesbelow even100 kPa. indicates that such"failure"mechanisms have very lowthresholdsand may be impor ta nt when consideringlimitingforces.

When ice is stressedit creeps,therefore the presenceof euetainedIlea ice preeeure of the orderofhundreds of kPa means that the ice sheet was moving. La.ndfastice is frozen tothe shoreand its ability to move andapply loads tofixed objectsis therefore restricted.Itisthus relevant to examine the source(s)of thismovementrela.tiveto the measured pressures.

The magnitudeof the loads may be limited in two W;:Ly8:

1. Therestriction on icesheet movementmay limit the force or, 2. The force appliedma.yhelimited bythe drivingforce.

35

(57)

Figure 19: This photograph shows a fresh vertical deformation 100 m north east of the stress sensors.Note the polar bear tracks in the foreground.

36

(58)

Bothof these willbe examined,beginningwitha modelfoe landfast icemovement.

In a manner sirnilar to that used by Croasdale(1975) the landfast icewill be treated as a semi-infiniteice sheet,fixedat theshore,and unrestr ainedatthe seawardedge (see figure 20).

Iceisavisco-elastic material and,during creep deformati on,the pressureexerted by the icesheet is proportional to the cube rootofthe indentationrate. During creep,the widerthestructu re the faster the ice mustmove.in order to generatea givenpressure(strainrate isa functionof the icespeedand stru cturewidth ).As therubblefieldat Kaubvikwas 0.8km wide, significant movements are requiredto generate the measuredpressures.

Itishypothesizedthatmovement of thelandfastice was the result of thermal expansion,and wind and currentstresses.Basedonthis hypothesis,work wasdone to quantify thecontr ibution of eachof thesesources of movementand develop a model to predictthe movement frommeasured data.

Calculat ions (see appendix B) showthat stres sesduetotidalcurrents areex- pectedtohe negligible, and as wind-induced cur rents are notexpect edin ice covered water,itwas decided thatcurrent s would notbeincludedin the analysis.Ba.sedon this,a computer program wa.swritten(also appendixB) which usedthe wind data from Thktoyaktukandsea ice temperaturedata to calculatethe landfast ice sheet movement. The windst resses produce elastic,delayedelast ic, andcreep deformations which were calculatedbased on the work donebySinha(1983&).Thewind stresses were calculatedusinga drag coefficient of.003(Feldman et al,1981),and the 30 km strip of ice between the CRI and shore was divided into5kmwidestrips forthe creep calculations. The thermal expansion was calculate dusing a coefficient of expansion of 0.000051 mlm°C (Michel, 1978)and all the calculationsweredone withoutinclud- ing any restraining influence of the CRI. As the me.curyfilled pressure sensors were

37

(59)

Figure20: Showingthe physical modelof alandfutice sheet usedforthe computer program.

located south of the rubble field,only the north-south component ofthe movement wea studied.The winddatawuaveraged over 12hrintervalsandthe temperature wastake n every 12 hr.The movement rate atapoint30kmoffshore was calculated (loutionofthe CRt)and the results are shown in figure 22.When compared to the measuredpressure data(fi821)some similarity c&nbeeeen.The movement rmi, sm all between daY'114and118whichcoincides with the low pressuresmeas ured a.t tha.t time, the movement ratedropssharply on da.y123coinciding withthe dropin pressureon that day,and the daily cyclingofthe movement curve is reflected in the pressure data.

The resultan tpressures etl.:!.be estima.tedusingthe referencestressmet hod for creepindenta tion as presented bySanderson (1984). Betweendays105and 125 the aver agemaximummovementrate(fivehighestpeaks)was 1.03 m/day.This translates into anindentationstrain ret e of1.49·1O-s fsec foran800 m wideindenter (appe ndix C). Fromthe results of Sanderson (1984) theaveragecontactpressure is-calculatedto bebetween535 kPa (p-anularsea ice]and290kPa (columnarseaice).Themaximum

38