Vous êtes sur la page 1sur 524

,

J
IAHR International

Symposium on Ice

QUEBEC 1981

Symposium international

sur la glace de I'AIRH

QUEBEC 1981

PROCEEDINGS

COMPTES RENDUS

VOL. II

I) S AR MY COLD R\Z.G\ONS RESEARCrl


t\i\~l; F.i'iG \ t'.i tJ.:I~\NG LABORAtORY
Ar iN ', Library
72 Lvme Road
HanDver,: NI1 03755 ... ~.~
To order the proceedings write to:

Pour commander les comptes rendus, ecrire a:


Prof. Bernard Michel
Departement de genie civil
Universite Laval
Cite universitaire
Quebec, Qc, Canada
G1 K 7P4

Reprints from this publication may be made, provided credit is given to the authors and
reference is made to the Proceedings of the IAHR International Symposium on Ice
Quebec, Canada, 1981.

Celie publication ne peut etre reproduite que si les auteurs en rec;:oivent Ie credit et qu'une
reference soit faite aux " Comptes rendus du Symposium international sur la glace de
I'AIRH, Quebec, Canada , 1981 ".

Di .... ision des publications


Servi ce des relatio ns publiques
RPDP 826J
Universith Laval. Quebec. Canada
International Association for Hydraulic Research

Association internationale de recherches hydrauliques

International Symposium on Ice

Symposium international sur la glace

Quebec, Canada

July 27-31,1981

Du 27 au 31 juillet 1981

Editorial Board - Bureau d'edition


Bernard Michel

Kersi Davar

Robert Frederking

Robert Gerard

Rene Hausser

Richard Kry

Jacques Michel

Universile Laval, Quebec, Canada

Minislere de l'Environnemenl, Gouvernemenl du Quebec

SPONSORS BY - PARRAINE PAR


International Association for Hydraulic Research
Association internationale de recherches hydrauliques

CO-SPONSORS BY - CO-PAR RAINE PAR


UNESCO
L'UNESCO
International Association of Hydrological Sciences
L'Association internationale des sciences hydrologiques
World Meteorological Organisation
L'Organisation mondiale de meteorologie
International Glaciological Society
La Societe internationale de glaciologie

IAHR COMMITTEE ON ICE PROBLEMS


CO MITE DES GLACES DE L'AIRH
T. Carstens (Norway - Norvege) (Chairman - president)
O. Starosolsky (Hungary - Hongrie) (past Chairman - ex-president)
G. Ashton (USA - E.-U.) (Secretary - secretaire)

L. Bengtsson (Sweden - Suede)

V. Degtyarev (USSR - U.RSS.)

M. Drouin (Canada)

G. Frankenstein (USA - E.-U)

R. Gerard (Canada)

VE Lyapin (USSR - UR.S.S.)

M. Maatlanen (Finland - Finlande)

H. Saeki (Japan - Japon)

J. Schwarz (Fed. Rep. Germany - Rep . fed. allemande)

P. Tryde (Denmark - Danemark)

E. Zsilak (Hungary - Hongrie)

NATIONAL ORGANIZING COMMITTEE


CO MITE NATIONAL D'ORGANISATION
B. Michel (Chairman - president)
B. Harvey (Treasurer - tresorier)
D. Carter
D. Chartrand
M. Drouin
R. Frederking
R. Gerard
R. Kry
N. McNeil
C. Verreault
D, Witherspoon

NATIONAL SCIENTIFIC COMMITTEE


COMITE NATIONAL SCIENTIFIQUE
R. Frederking (Chairman - president)
K. Davar
R. Gerard
R. Hausser
R. Kry

LADIES' COMMITTEE
COMITE FEMININ
Mariette Michel (presidente)
Ghislaine Carter
Monic Frenette
Suzanne Godin
Suzanne Harvey
Madeleine Ouellet
Claire Verreault
Marielle Verrette

SECRET ARIES
SECRETAIRES
Diane Dussault
Jeanne Roy
I

TABLE OF CONTENTS - TABLE DES MATIERES


Sponsors by - Parraine par

Committees - Comites

Preface - Preface

List of participants - Liste des participants

Session A - THERMAL REGIME - REGIME THERMIQUE


B. Michel " History of Research on River and Lake Ice in Canada"
L. Bengtsson " Experiences on Winter Thermal Regimes of Rivers and
Lakes with Emphasis on Scandinavian Conditions" 11
Discussion by :
G.D. Ashton 32
Discute par

Author's reply to previous discussion

34
Reponse de I'auteur a la discussion precedente

Discussion
A
by : 0 . CaIk'Ins

35
DIscutt:; par :

Author's reply to previous discussion

38
Reponse de I'auteur a la discussion precedente

Discussion by :

R. Gerard 39
Discute par

Author's reply to previous discussion

40
Reponse de I'auteur a la discussion precedente

Discussion by :

EV. Kanavin 42
Discute par

Discussion by :

R.P. Asvall and S. Roen 43


Discute par

Author's reply to previous discussion

53
Reponse de I'auteur a la discussion precedente
Discussion by :
Discute par
O. Starosolszky 54
L. Bengtsson "Heat Losses from an Open Water Surface at Very Low
Air Temperature A Laboratory Experiment" 55
G.D. Ashton "River Ice Suppression by Side Channel
Discharge of Warm Water" 65
Discussion by :
A. de Haas 74
Discute par
Author's reply to previous discussion
74
Reponse de I'auteur a la discussion precedente
Discussion by: 75
S. Beltaos
Discute par

Author's reply to previous discussion

76
Reponse de I'auteur a la discussion precedente

Discussion by:

R. Gerard 77
Discute par

Author's reply to previous discussion

78
Reponse de I'auteur a la discussion precedente
Discussion by:
K.S. Davar 77
Discute par

Author's reply to previous discussion

Reponse de I'auteur a la discussion precedente 80


V. Matousek "A Mathematical Model of the Discharge of
Frazil in Rivers" 81
Discussion by:
L. Bengtsson 99
Discute par

Author's reply to previous discussion

100
Reponse de I'auteur a la discussion precedente
T. O'D . Hanley and "Acoustic Detector for Frazil "
S.R. Rao 101
Discussion by: AM D 108
Discute par : . . ean
Authors' reply to previous discussion 108
Reponse des auteurs a la discussion precedente

Discussion by :
109
R.S. Arden
Discute par

Authors ' reply to previous discussion


109
Reponse des auteurs a la discussion precedente
Discussion by: 109
S. Daly
Discute par
Authors' reply to previous discussion
109
Reponse des auteurs a la discussion precedente
Discussion by: 110
D.M. Foulds
Discute par

Authors' reply to previous discussion

110
Reponse des auteurs a la discussion precedente
F.D. Haynes, " Periormance of a Point Source Bubbler
G.D. Ashton and Under Thick Ice"
P.R. Johnson 111
Discussion by: 122
C.A. Wortley
Discule par
Discussion by:
G.P. Williams 123
Discute par

Discussion by:
124
D. Foulds
Discute par

Authors' replies to previous discussions

124
Reponses des auteurs aux discussions precedentes
P. Tryde "Ice Formation on the Walls of a Water Tunnel
Excavated Through Rock in Permafrost" 125
Discussion by:
F.D. Haynes 136
Discute par

Discussion by:

G.D. Ahston 137


Discute par

Discussion by:

M. Mellor 138
Discute par

Discussion by:

K.S. Davar 139


Discute par

Author's reply to previous discussions

140
Reponse de I'auteur aux discussions prececentes
L. Votruba "Relations between Climatic Conditions and
Winter Regime of Water Bodies" 141

Session B - ICE HYDRAULICS - HYDRAULIQUE DES GLACES


D.M. Foulds "Peaking Hydro Generating Stations in Winter" 152
Discussion by: 160
D.G. Harkness
Discute par

Discussion by:

T.E. Wigle 161


Discute par

Author's replies to previous discussions

162
Reponses de I'auteur aux discussions precedentes

Discussion by: T.A. McClimans


162
Discute par

Author's reply to previous discussion

162
Reponse de I'auteur a la discussion precedente

Discussion by: D. Calkins


162
Discute par

Author's reply to previous discussion

162
Reponse de I'auteur a la discussion precedente

Discussion by:

L. Billfalk 162
Discute par

Author's reply to previous discussion

162
Reponse de I'auteur a la discussion precedente
O. Gybrke, "Problems of Ice Release and Flow Conditions
E. Decsi and Upstream of Low-Head River Dams" 163
E. Zsilak
Discussion by :
I. Brachtl 172
Discute par
Authors ' reply to previous discussion
174
Reponse des auteurs a la discussion precedente
Discussion by :
G. Frankenstein 176
Discute par
Authors' reply to previous discussion
177
Reponse des auteurs a la discussion precedente
H.T. Sken and "Wintertime Flow and Ice Conditions in the
N.L. Ackermann Upper St. Lawrence River" 178
Discussion by :
TA McClimans 188
Discute par

Authors' reply to previous discussion

188
Reponse des auteurs a la discussion precedente
Discussion by :
D.F. Witherspoon 189
Discute par
Authors' reply to previous discussion
190
Reponse des auteurs a la discussion precedente

Discussion by :

D. Calkins 191
Discute par

Authors' reply to previous discussion

192
Reponse des auteurs a la discussion precedente
T.E. Wigle , " Winter Operations International Rapids
J. Bartholomew and Section of the St. Lawrence River" 193
C.J.R. Lawrie
Discussion by :
D.F. Witherspoon 203
Discute par

Authors ' reply to previous discussion


204
Reponse des auteurs a la discussion precedente
Discussion by:
R.D. Conner 205
Discute par
Authors ' reply to previous discussion
206
Reponse des auteurs a la discussion precedente

Discussion by :

T. Dafoe 207
Discute par
Authors' reply to previous discussion 208
Reponse des auteurs a la discussion precedente
Discussion by : 209
D.M. Foulds
Discute par
Authors ' reply to previous discussion
210
Reponse des auteurs a la discussion precedente
R. Boivin, "Influence de la couverture de glace sur les echanges
O. Caron et d'eau salee et d'eau douce dans un estuaire a maree: Ie
M. Drouin cas de I'estuaire de La Grande Riviere, au debut du
remplissage du reservoir de LG 2" 211
Discussion by:
N. McNeil 221
Discute par
Discussion by :
R. Lariviere 222
Discute par
Authors' reply to previous discussion
Reponse des auteurs aux discussions precedentes
223
Discussion by: TAM CI' 223
Discute par : .. c Imans
Authors ' reply to previous discussion
223
Reponse des auteurs a la discussion precedente
K. Hiramaya "Hydraulic Resistance of Ice Cover" 224
Discussion by: J C Tt'l 235
Discute par : . . a Inc aux
Author 's reply to previous discussion
235
Reponse de I'auteur a la discussion precedente
Discussion by :
H.T. Shen 235
Discute par
Author's reply to previous discussion
235
Reponse de I'auteur a la discussion precedente
Discussion by:
S. Ismail 236
Discute par
Author's reply to previous discussion
236
Reponse de I'auteur a la discussion precedente
Discussion by:
C.D. Smith 236
Discute par
Author's reply to previous discussion
236
Reponse de I'auteur a la discussion precedente
M. Jensen " Ice Problems at Villjarv Power Plant
Measures and Results " 238

Session C ICE HYDRAULICS HYDRAULIQUE DES GLACES


F.E. Parkinson " Field Observations of Ice Conditions on the Liard/
Mackenzie River System" 252
Discussion by:
Discute par
J.w. Kamphuis 260
Author's reply to previous discussion
261
Reponse de I'auteur a la discussion precedente
Discussion by :
D. Foulds 262
Discute par
Author's reply to previous discussion
263
Reponse de I'auteur a la discussion precedente
Discussion by: F. Clement 264
Discute par
Author's reply to previous discussion
265
Reponse de I'auteur a la discussion precedente
N.K. Gidas " Recherche sur les meilieures solutions contre les inon
dations de la Matapedia causees par les debacles" 266
Discussion by:
J.C. Tatinciaux 267
Discute par
Author's reply to previous discussion
276
Reponse de I'auteur a la discussion precedente
F.M. Henderson and "Flood Waves Caused by Ice Jam Formation and Failure" 277
R. Gerard
Discussion by:
S. Beltaos 288
Discute par
Discussion by :
D. Calkins 289
Discute par
Discussion by: 291
J.W. Kamphuis
Discute par
Discussion by: 291
F. Parkinson
Discute par
Discussion by:
O. Starosolszky 291
Discute par
Authors' replies to previous discussions
293
Reponses des auteurs aux discussions precedentes
J.C. Tatinciaux "Stability of Floes Below a Floating Cover"
and M. Gogus 298
Discussion by:
S. Beltaos 309
Discute par
Discussion by :
G.D. Ashton 310
Discute par
Discussion by:
Discute par
o Starosolszky 310

Discussion by:
R. Hausser 310
Discute par
Authors' reply to previous discussion 311
Reponse des auteurs aux discussions precedentes
C. Zanting, "Regulating Effect of Reservoir of the Control of Ice
S. Zhaochu, Run on the Yellow River
x. Jianfeng and 312
C.L. Wang Wencai
Discussion by:
GD. Ashton 323
Discute par
0.1. Gordeev and "Computation of Trajectories of Ice Floes
V.V. Degtyarev Movement on the Rivers" 324
N.L. Ackerman, "Transportation of Ice in Rivers" 333
H.T. Shen and
RW. Ruggles
Discussion by:
S. Beltaos 343
Discute par
Authors' reply to previous discussion
344
Reponse des auteurs a la discussion precedente
Discussion by:
S.E. Daly 345
Discute par
Authors' reply to previous discussion
346
Reponse des auteurs a la discussion precedente
R.M. Vogel and "The Effect of Floating Ice Jams on the Magnitude and
M.J. Root Frequency of Floods along the Missisquoi River in
Northern Vermont" 347
Discussion by:
B. V.C. Kartha 360
Discute par
Authors' reply to previous discussion
360
Reponse des auteurs a la discussion precedente

Session D - MODELLING THE ICE PHENOMENA - MODELES DE GLACE

D.J. Calkins,
D.S. Sodhi and 361
D.S. Deck
Discussion by:
S.S. Lazier 372
Discute par
Authors' reply to previous discussion 373
Reponse des auteurs a la discussion precedente
S.F. Daly and "Force Distribution in a Fragmented Ice Cover" 374
D.M. Stewart
Discussion by:
J.C. Tatinclaux 384
Discute par
Authors' reply to previous discussion
385
Reponse des auteurs a la discussion precedente
Discussion by:
S. Beltaos 386
Discute par

Authors' reply to previous discussion

387
Reponse des auteurs a la discussion precedente
G. Garbrecht, " Formation of Ice Jams in the Elbe River
H. Fahlbusch and A Case Study" 388
w. Mertens
N.D. Elhadi and "Dispersion in a Covered Channel with Varying
K.S. Davar Roughness at the Top Cover" 398
Discussion by:
S. Beltaos 408
Discute par

Authors' reply to previous discussion

409
Reponse des auteurs a la discussion precedente

Discussion by :

J.C. Tatinclaux 410


Discute par
Authors ' reply to previous discussion
411
Reponse des auteurs a la discussion precedente
N. Marcotte " Regime thermique des glaces en riviere
Etude de cas" 412
Discussion by : 423
S. Petryk
Discute par

Author's reply to previous discussion

424
Reponse de I'auteur a la discussion precedente
S. Petryk, " Numerical Modeling and P.redictability of Ice Regime in
U.S. Panu, Rivers ' 426
B. V.C . Kartha and
R. Clement
C.D. Smith " Model Study of Ice Movement at Idylwyld Traffic Bridge" 436
General discussi?n .of the session by : D.J. Calkins 445
Commentalres generaux de la session par:

Souvenirs 449

IN VOLUME II

DANS LE VOLUME II

Session E - ICE MECHANICS - MECANIQUE DES GLACES


M. Melior "Glaciers Mechanics " 455
Discussion by: 474
L. Gold
Discute par
S. Beltaos and " Field Investigations of a Hanging Ice Dam " 475
A.M . Dean
Discussion by:
R. Gerard 486
Discute par

Authors' reply to previous discussion

487
Reponse des auteurs a la discussion precedente

Discussion by:
488
J.C. Tatinclaux
Discute par
Authors' reply to previous discussion
488
Reponse des auteurs a la discussion precedente
G. W. Timco "A Comparison of Several Chemically-Doped
Types of Model Ice" 489
Discussion by:
G. Frankenstein 400
Discute par
Discussion by: M. Maallanen 500
Discute par

Author's reply to G. Frankenstein

501
Reponse de I'auteur a G. Frankenstein
Discussion by: 501
R.T. Weiss
Discute par
Author's reply to previous discussion
501
Reponse de I'auteur a la discussion precedente
Discussion by: JC T r I 501
Discute par : .. a inC aux
Author's reply to previous discussion
502
Reponse de I'auteur a la discussion precedente

Discussion by:

J. Molgaard 502
Discute par
D.A. Sandell " Carbamide Ice Growth in a Large Test Basin " 503
Discussion by:
G.W. Timco 515
Discute par

Author's reply to previous discussion

515
Reponse de I'auteur a la discussion precedente
M. Nakawo and "The Salinity of Artificial Built-Up Ice Made by
R. Frederking Successive Floodings of Sea Water" 516
Discussion by:
A. Assur 525
Discute par
Authors' reply to previous discussion
525
Reponse des auteurs a la discussion precedente
F.U. Hausler "Multiaxial Compressive Strength Test on Saline Ice with
Brush-Type Loading Platens" 526
Discussion by:
Y.S. Wang 537
Discute par
Author's reply to previous discussion 537
Reponse de I'auteur a la discussion precedente

Discussion by:

N.K. Sinha 537


Discute par

Authors' reply to previous discussion

538
Reponse de I'auteur a la discussion precedente
X. Bomeng " Pressure Due to Expansion of Ice Sheet in Reservoirs " 540
N. Urabe and "Strain Rate Dependent Fracture Toughness (K IC )
A. Yoshitake of Pure Ice and Sea Ice" 551
Discussion by:
T.D. Ralston 564
Discute par
Author's reply to previous discussion
564
Reponse de I'auteur a la discussion precedente

Session F - ICE MECHANICS - MECANIQUE DES GLACES


P.R . Kry "Scale Effects in Continuous Crushing of Ice" 565
Discussion by :
C.R. Neill 580
Discute par
Author's reply to previous discussion
580
Reponse de I'auteur a la discussion precedente
N.K. Sinha "Comparative Study of Ice Strength Data" 581
Discussion by:
Y.S. Wang 593
Discute par

Discussion by :

F.U. Hausler 594


Discute par

Author's replies to previous discussions

595
Reponses de I'auteur aux discussions precedentes
P. Duval, "Primary Creep and Experimental Method for Testing
M. Maitre , Ice in Various Conditions of Strain Rates
A. Manouvrier, and Stresses" 596
G. Marec and J.C. Jay
Discussion by :
M. Mellor 603
Discute par

Authors' reply to previous discussion

603
Reponse des auteurs a la discussion precedente
Discussion by :
LW. Gold 604
Discute par
Authors' reply to previous discussion
604
Reponse des auteurs a la discussion precedente
Discussion by:
R. Frederking 605
Discute par
Authors' reply to previous discussion
605
Reponse des auteurs a la discussion precedente
L. Lainey and " Parametric Studies of Sea-Ice Beams under Short
R. Tinawi and Long Term Loadings " 607
Discussion by :
Discute par N.K. Sinha
620
Discussion by :

J.-P. Nadreau 621


Discute par

Discussion by :

R. Frederking 622
Discute par

Discussion by:

P.R. Kry 623


Discute par

Authors' reply to previous discussion

624
Reponses des auteurs aux discussions precedentes
P. Oksanen " Friction and Adhesion of Ice" 628
Discussion by : 638
J. Molgaard
Discute par

Author's reply to previous discussion

639
Reponse de I'auteur a la discussion precedente

Discussion by :

C.D. Smith 639


Discute par

Author's reply to previous discussion

639
Reponse de I'auteur a la discussion precedente

Discussion by:
640
E. Palosuo
Discute par

Author's reply to previous discussion

640
Reponse de I'auteur a la discussion precedente
H. Saeki, T. Ono " Mechanical Properties of Adhesion Strength to Pile
and A Ozaki Structures" 641
L. Billfalk "Formation of Shore Cracks in Ice Covers due to
Changes in the Water Level" 650
Discussion by :
D. Nevel 661
Discute par

Author's reply to previous discussion

661
Reponse de I'auteur a la discussion precedente

Discussion by:

R. Frederking 662
Discute par

Author's reply to previous discussion

662
Reponse de I'auteur a la discussion precedente
N. Yoshimura and " The Estimation of Crack Pattern on Ice by the New
K. Kamesaki Discrete Model " 663
Discussion by :
R. Tinawi 673
Discute par

Author's reply to previous discussion

673
Reponse de I'auteur a la discussion precedente
I.N. Sokolov, "Studies of Ice Action on Pumped Storage
Ya. L. Gotlib, Power Plant Structures" 674
P.G. Dick and
G.M. Ryabkin

Session G - ICE AND STRUCTURES - GLACE ET STRUCTURES


L.W . Gold " Designing Ice Bridges and Ice Platforms" 685
Discussion by :
P. Johnson 698
Discute par

Author's reply to previous discussion

701
Reponse de I'auteur a la discussion precedente
G. Tsang " Fin Boom Ice Gate for Ice Control and Winter
Navigation" 702
Discussion by:
W.E. Webb 715
Discute par

Author's reply to previous discussion

715
Reponse de I'auteur a la discussion precedente

Discussion by :

R. Perham 716
Discute par

Author's reply to previous discussion

717
Reponse de I'auteur a la discussion precedente
P. Yee, " The Lake Erie - Niagara River Ice Boom - An
T.E. Wigle and Operational Experience" 718
A. Hollmer
Discussion by:
G. Tsang 726
Discute par
R. Abdelnour "Model Test of Multi-Year Pressure Ridges
Moving onto Conical Structures " 728
Discussion by:
J. Schwarz 750
Discute par

Author's reply to previous discussion

751
Reponse de I'auteur a la discussion precedente

Discussion by:

D.O. Hodgins 752


Discute par

Author's reply to previous discussion

752
Reponse de I'auteur a la discussion precedente

Discussion by :

G.R . Pilkington 752


Discute par
Author's reply to previous discussion
753
Reponse de I'auteur a la discussion precedente
Discussion by: T R Ch . 753
Discute par : .. an
Author's reply to previous discussion
753
Reponse de I'auteur a !a discussion precedente

Discussion by:
754
D.V. Reddy
Discute par

Author's reply to previous discussion

754
Reponse de I'auteur a la discussion precedente
H.R. Croasdale and "Ice Forces on Large Marine Structures" 755
R.W. Marcellus
Discussion by:
T. Lavender " 766
Discute par

Authors' reply to previous discussion

767
Reponse des auteurs a la discussion precedente

Discussion by:

Discute par P.R. Kry 768


Authors' reply to previous discussion
770
Reponse des auteurs a la discussion precedente
C.J. Montgomery "Estimation of Ice Forces from Dynamic
and A.W. Lipsett Response" 771
Discussion by:
F.D. Haynes 781
Discute par

Author's reply to previous discussion

781
Reponse de I'auteur a la discussion precedente

Discussion by:

G.R. Pilkington 782


Discute par
Authors' reply to previous discussion
Reponse des auteurs a la discussion precedente 782
M. Maattanen "Ice-Structure Dynamic Interaction - Ice Forces
Versus Velocity, Ice-Induced Damping" 783
Discussion by:
R. Frederking 793
Discute par
Author's reply to previous discussion
794
Reponse de I'auteur a la discussion precedente
Discussion by:
AW. Lipsett 795
Discute par
Author's reply to previous discussion
796
Reponse de I'auteur a la discussion precedente
J . Karri and "Measurement of Horizontal and Vertical Ice
P. Jumppanen Loads on Pil e Type Structures" 797

Session H - GENERAL - GENERALE


o. Starosolszky " Thermal Regime and Ice Forecasting for Fresh
Water Bodies " 809
V. Balanin, " Estimation of Ice Conditions and Organization of
V. Tronin, Shipping on Rivers and Reservoirs During the
V. Malinovsky, Extended Period of Navigation" 825
Y. Sandakov and
B. Ginzburg
S.M. Aleinikov, " Protection of Hydraulic Structures from Icing " 836
BE Lyapin,
M.1. Zhidkikh,
A.V. Panyushkin and
N.G. Khrapatyi
R. Gerard "Ice Scars: Are they Reliable Indicators of Past Ice
Breakup Water Levels?" 847
Discussion by :
F. Parkinson 855
Discute par
Discussion by :
B. VC. Kartha 855
Discute par
Discussion by :
O. Starosolszky 856
Discute par
Discussion by :
M. Root 856
Discute par
Author's reply to previous discussions
858
Reponse de I'auteur aux discussions precedentes
R.T. Weiss, " Determination of Ice Rubble Shear Properties" 860
A. Prodanovic and
K.N. Wood
Discussion by :
B. Ladanyi 871
Discute par
Discussion by:
S.T. Lavender 871
Discute par
Authors' reply to previous discussion
872
Reponse des auteurs aux discussions precedentes
I. Mayer " Ice Hydraulic Stability Analysis:
Experimental Determination of Pressure
Distribution under Ice Floes" 873
SHORT NOTES NOTES
S.P. Chee and "Underside Configuration of Ice Covers" 881
M.R. Haggag
R.M. Desmond, " Ice Growth in Rivers " 884
B.V. Karlekan and
S.G. Kandlikar
H. Hamza, "Elastic Creep Bending Analysis of Floating
D.B. Muggeridge Ice Covers" 889
and T.E. Laidley
T. 0 '0 . Hanley "Heat Transfer During Freezing in Calm Water" 894
J.B. Kennedy and " Response of Floating Ice Sheets Under Impact
H.J. Iyengar Loads" 900
A.P.S . Selvadurai "On the Theoretical Modelling of Floating Ice
Sheets which Exhibit a Composite Structure" 905
G.W. Timco "A Technique for Chem ically Simulating a Snow
Cover on Model Ice" 910
P. Vasseur, "Numerical Modelling of Dentritic Ice Formation in
L. Robillard and Supercooling Conditions" 913
B. Chandra Shekar
E.V. Kanavin " Fifty Years of Experience in the Field of Ice
Problems for Aiver Engineering" 917
A.D . Kerr "Remarks to the Buckling Analysis of Floating Ice Sheets" 932

R. Frederking , " IAHA - Recommendations on Testing Methods of Ice


V.P. Gavrilo, - 3rd Report of Working Group on Testing Methods
D.J. Hausler and in Ice" 938
K. Hirayama
GLACIER MECHANICS

Malcolm Mellor U.S. Army Cold Regions Research USA


and Engineering Laboratory,
Hanover, New Hampshire

Introduction
A talk on glacier mechanics at an ice engineering symposium seems
to deserve a word of explanation, e spec ially when it is given by some
body with dubiou s credentia ls and no discernible motivation.
The fact is that the IAHR, which t akes care of international
collaboration in ice engineering, has in recent years developed close
contacts with the Int ernational Commission on Snow and Ice, or ICSI,
which take care of international COllaboration in glaciology general
l y. The two orga nizations have been joint sponsors Eor a number oE
meetings, and as a co-sponsor o f this symposium I CSI was asked by IAHR
to arrange a talk on glacier mechanics. There were no takers amon g
the real glaciologists, and so the Secretary of ICSI wound up with the
job.
Being in no way qualified to give an authoritative review of
glacier mechanics, I shall try to outline what the subject is about,
and how it might relate to the mainstream of ice engineering.

Glaciers
A glac ier is a perennial mass of ice formed from snow. The term
cov ers all ice masse s of this kind, from dirty little permanent snow
beds tucked away in mountain cirques, to the enormous continental ice
sheet o f Antarctica. A glacier is called a valley glacier if it fl ows
along a valley, a mountain glacier if it is stuck up on a mountain,
and a piedmont glacier if it forms a puddle of ice on flat ground at
the foot of a mountain. A continuous mantle o f ice that covers a

455
wide acea and flows in all dicections is an ice cap, unless it covecs
Gceenland oc Antacctica, in which cases it becomes an ice sheet. Ice
caps and ice sheets can have all the othec kinds of gla-ciecs as
components, a nd they can also focm ice shelves by flowing out and
floating on the sea in deep watec coasta l embayments.

Glaciec Me c h a nics
In discussing glaciec mechanics, it is convenient to subdivide
the subject into the kinematics, dynamics and enecgetics of glaciec
motion.
Kinematics.
Kinematics, which deals with geometry, displacement , stcain ,
velocity and stcain cate, was the first bcanch of glaciec mechanics
to catch the attention of eacly scientists. The geomocphology of
deg la ciated aceas allowed cectain deductions to be made about dimen
sions, depths and flow dicections of former glaciecs. Simple ODsec
va ti ons gave pcovided quantitative infocmation about sucface motion,
so that by the nineteenth centucy typical flow velocitie s were known,
and tcansvecse pcofiles of sucface velocity had been estab li shed .
Mack Twain leacned about kinematics in his attempt to tcavel by gla
ciec in Switzecland;
"I macched the expedition down the steep and
tedious mule-path and took up as good a position as
could upon the middle of the glaciec - because Baedekec
said the middle pact tcavels the fastest. As a measuce
of economy, howevec, I put some of the h ea viec baggage
on the shocewacd pacts, to go as slow fceight."
Aftec the tcavelecs had camped ovecnight waiting for the g l aciec
to start , they again consulted theic book;
"Pcesently Baedekec was found again, and I hunted
eagecly foc th e time-table. Thece wa s none. The book
simply said the glaciec was moving all the t ime . This
was satisfactory , so I sh ut up the book and chose a good
pOSition to view the scenecy as we passed along. stood
thece some time enjoying the tcip, but at last it oc
curred to me that we did not seem to be gaining any on
the scenery. I soon found a sentence which thcew a
dazzling light on the mattec. It said, 'The Gorner Gla
cier travels at an average rate of a little less than an
inch a day. ' made a small calculation; inch a day,

456
say 30 feet a year; estimated distance to Zermatt , 3-1 /H
miles. Time required to go by glacier , a little over
five hundred years! said to myself, 'I can walk it
quicker - and before will patronize such a fraud as
this, I will do it!"
In the Alps, serious speCUlation on glacier motion began early
in the eighteenth century , when it was believed that a glacier moved
soley by rigid-body motion, that is, by sliding of ice is a block on
the rock bed. Ideas of ice as a viscous deformable solid developed
late in the eighteenth centu ry, and during the nineteenth century the
understanding of differential motion became quite sophisticated.
However, s tudies of three-dimensional displacements and strains,
including snow densification, particle trajectories, velocity pro
files, and general flux divergence did not deve lop until the mid
twentieth century, when drilling, borehole deformation measurement,
and e lectr omagnetic methods of subsurface exploration became avail
able.
It is now possible to determine the surface contours of glaciers
with relative ease, using air or grou nd photogrammetry, or even
satellite imagery. The rock bed contours can be mapped by electro
magnetic sounding in so-called "cold" ice even at great depth; wet or
"warm" ice is more difficult to penetrate, but pulsed microwave
systems are now being used succe~sfully. Surface displacements and
velocities are quite easy to measure by photogrammetry as long as
fixed references are available, and consequent ly surface fields of
velocity and strain rate can be mapped. Surface velocity is more
difficult to measure on large ice sheets. Once upon a time, the only
expedient was repeated first-order astronomical fixes, but nowadays
surface displacement relative to the rock bed can be sensed, satel
lite fixes are available, and satellite imagery can be used. Inter
nal deformation is still very difficult to record, but from inclino
meter measurements in boreholes we have a collection of vertical
velocity profiles and shear strain rates. Volumetric strains develop
as surface snow is buried and compacted progressively, and the
straining continues in deep ice as air bubbles are compressed and
finally eliminated. Volumetric strain rates are obtained from ver
tical strain profiles by estimating time-depth relationships in the
snow and ice layers.

457
As far as size and shape are concerned, glaciers range in area
fro~ less than 1 km 2 to more than 10 million km 2 , and in volume from
a fraction of a cubic kilometre up to tens of millions of cubic kilo
metres. The ice thickness ranges from virtually zero to more than
4000 metres. Surface slopes are typically small, say in the range
2xl0- J to 2xl0- 1 (0.2% to 20%), and there tends to be an inverse
correlation (a linear one) between ice thickness and surface slope.
This inverse proportionality between slope and thickness implies that
the flowline profile of ice lying on a flat bed would be parabolic,
and in fact the surface profiles of ice caps do tend to approximate
parabolic or elliptic shapes.
Studies of glacier kinematics tell ui that mid-stream surface
velocities on valley glaciers and ice streams typically range from
less than 0.1 to more than 1 m/ day. The speeds of some glaciers vary
seasonally, or in some cases over longer time periods as the glaciers
"surge". During surges, glacier velocity can exceed 10 m/day. When
the rate of accumulation or ablation on some part of a glacier system
is significantly perturbed, the glacier is thought to respond by pro
pagating a kinematic wave of mass flux, which travels much faster
than the ice itself, much like a flood wave in a river. Although
glacier flow has long been treated in steady-state terms, it may well
be that true steady-state flow is the exception rather than the rule.
In typical situations where the glacier is "wide", and flow is
approximately two-dimensional, simple continuity requires that mean
velocity be inversely proportional to ice depth, but in reality there
is usually addition or removal of mass at the glacier surface. Addi
tion of material by snow accumulation tends to produce longitudinal
strain rates in a compressive sense, while removal of material by
ablation tends to produce extensive or tensile, longitudinal strain
rates. An exact analysis of flux divergence in a columnar element of
the flow can only be made approximately because the distribution of
strain components with depth is unknown. In extending flow, where
velocity increases with downstream distance, tensile fractures can
occur and form crevasse fields. It appears that crevasses often form
when tensile surface strain rates approach 10- 9 s-1, which is far
below the strain rates needed to produce anything akin to brittle
tensile fracture in the lab. However, the glacier can produce ten
sile strains much larger than those usually induced in laboratory
experiments . For example, if ice moves into a crevasse zone at 0.1
m/day, and iE new c~evasses Eo~m at a spacing oE 10 m, it takes 100
days to st~ain the ice up to the point oE tensile failu~e. At a ten
sile st~ain t:ate oE 10- 9 s-I, this means that the i c e fails at a
stt:ain oE 9x10- 3 , ot: about 1%. This is considet:ably highet: than the
tensile failut:e stt:ain Eot: bt:ittle Et:actut:e, but it is about equal to
the ductile Eailut:e stt:ain in com~t:ession.

Veetical pt:ofiles of velocit y ace qualitatively similat: t o peo


Eiles in viscous fluid. The ovet:all simple sheae strain eate oE a
glaciee, as given by the sut:face velocity divided by the depth, seems
to be oE the oedee of 10- 9 to TO- 8 S-I, but theee is actually con
sideeable vaeiation oE steain t:at" with depth. Floeehole measut:ernents
show most oE the sheat:ing ta~ing place below mid-depth, and it can be
deduced that simple sheae steain cates in the basal ice layecs at:e
about 10- 8 to 10- 7 s-1 if tem[)ecatuee is clos e to the melting [)oint.
[To avoid confusion, I might stcess that these cough values ace given
in tet:ms oE simple sheat: cathec than the ap[)copciate tensoc compon
ent, which diEEees by a Eactoc of 2. Glaciologists usually expcess
stcain cates in terms of the second invaciant as octahedcal stcain
cate, which is smallec than th e "simple" cate by a factoc of 16.J
BeEoce leaving the subje c t oE sheac stcains, it is intecesting
to note that well-settled snow lying on a slope cceeps downhill with
a velocity pcofile that is close to lineac, with cates oE sim[)le
sheac typically in the cange 2x1U- 8 to 4x1U- 8 S-I o t:, in othec wocds,
cates vecy simi lac to those in the basal layecs oE glaciecs.
Some glaci e t:s, oc pacts oE g laciecs, decive a substantial com
ponent of vel oc ity Ecom sliding oE the ice ovee the cock bed, with
sliding velocities up to 1 m/day oc so. The b"d tempecatuce has to
be at the melting point Eoc sliding to occuc, and flow oE sucface
meltwatec undec tile glaci ec seems to be an impoctant Eactoc. Details
oE the sliding pcocess ace not well established, but c ege lation and
ct:eep enhancement at stcess concentcations ace plausible mechanisms,
and various mystical Eciction Eactocs have been invoked by theoceti
cians.
Dynamics.
Dynamics can be taken as dealing with focces and stcesses in
glaciecs, and with theic celation to displacements and steains. The
study of glaciee dynami cs began in the nineteenth centucy, with
attempts to apply cegelation theocies, and also with systematic
stcess analyses based on viscous Elow theocy. Thece wece some

459
expe~iments by ea~ly investigators, but ~elevant field resea~ch did
not develop until just before the Second world War, while systematic
labo~ato~y investigation of st~ess/strain-rate ~elations really got
sta~ted during, and soon afte~, WWII.
The laboratory expe~iments we~e mainly constant load c~eep tests
in uniaxial comp~ession, and they were designed p~imarily to estab
lish a relation between st~ess and minimum st~ain ~ate, which was
believed to be a steady-state c~eep ~ate. Fo~ st~esses in the ~ange

0.1 to 2 MPa, the ~esulting empi~ical st~ess/st~ain-~ate ~elation was


exp~essed as a simple powe~ law, with an exponent of app~oximately 3
to 4. Fo~ st~esses lowe~ than 0.1 MPa some investigato~s found the
exponent of the powe~ ~elation to be unity, implying Newtonian vis
cosity, but since thei~ expe~iments we~e of ve~y sho~t du~ation and
limited to small and unequal total st~ains, these ~esults could not
be accepted as valid. One expedient fo~ ext~apolating the c~eep

cu~ves obtained f~om sho~t-duration tests was application of an


empi~ical equation known as And~ade's Law, but this p~ocedu~e is now
believed, by some investigato~s, to give e~~oneous results. The
effect of temperature on c~eep rate was desc~ibed by an Ar~henius

relation with activation energy of approximately 70 kJ/mole, though


it was obvious from the sta~t that the semi-log Arrhenius plot was
strongly nonlinear at tempe~atu~es above -10'C.
In o~de~ to deduce st~ess/st~ain-rate ~elations f~om field data,
st~ess has to be calculated, since it is not ~eally feasible to
measu~e st~ess. Fo~tunately, the basic p~oblems of glacie~ mechanics
a~e g~avity body-fo~ce p~oblems, in which stress is dete~mined la~ge

ly by the bounda~y conditions. In pa~ticula~, the c~ucial component


of shea~ st~ess in a vertical plane through a flowline is given
approximately by the p~oduct of ove~burden p~essu~e and su~face slope
a. In othe~ words, the shea~ st~ess T at depth z is given by

Z
T = fopgsina dz ,

irrespective of whether the ice has the ~heological p~ope~ties of


~ock, ~ubbe~, or Wisconsin cheese. Thus the obse~ved shea~ st~ain

~ates obtained from a bo~ehole by ~epeated inclinomete~ measurements


can be compa~ed with calculated shea~ st~esses to obtain st~ess/

st~ain-~ate ~elations, which tu~n out to be in ~easonable acco~d with


those gene~ated f~om lab expe~iments involving uniaxial st~ess

states.

460
It was the ability to calculate bed shear stress as the product
of density, depth and surface slope that led to the intriguing find
ing that the bed stress of stable glaciers throughout the world is
always of order 0.1 MPa (or 1 bar, or 15 Ibf / in 2 ), with a range from
0.05 to 0.2 MPa. This phenomenon has already been mentioned in the
context of an inverse correlation between depth and surface slope,
but we might risk labouring the point by noting again that, if H is
ice thickness and x is distance upstream from the edge of an ice
sheet, the constant stress equation pgH dH/dx = constant integrates
to give a parabolic relation between x and H. Knowing that the value
of the constant has to be below about 0.2 MPa for a stable ice pro
file, we could, for example, design stable side slopes for a large
open-pit excavation in ice.
The relati v e constancy of bed shear stress prompted J.F. Nye,
about 30 years ago, to apply plasticity theory to glacier flow, a
development that was remarkably stimulating to theoreticians and
glaciologists generally. Since then glacier flow has been treated in
terms of nonlinear viscosity, ostensibly in conformity with constitu
tive equations developed from laboratory tests.
A particularly interesting type of glacier flow is exhibited by
ice shelves, which are floating slabs of ice, typically some hundreds
of metres thick, attached over part of the perimeter to land gla
ciers. Since the shear stress has to be zero at both top and bottom
surfaces, an idealized ice shelf simply thins out by spreading under
vertical and horizontal components of normal stress which can be
estimated in simple terms for the idealized case. The flow of real
ice shelves is complicated by anchorage of parts of the margins, and
by gradients of thickness, but stresses can still be calculated.
Comparison of calculated stresses with measured strain rates for
various Antarctic ice shelves has provided a stress/strain-rate rela
tion for very low stresses (0.04 to 0.1 MPa), and this turns out to
be an extension of th e empirical relation for higher stresses, with
the power law exponent still equal to 3, thus firmly contradicting
the small strain lab experiments which give smaller exponents for low
stress.
Another topic for glacier dynamics is compressibility and volu
metric stress/strain-rate relations. After making the transition
from snow, glacier ice is bubbly, and if it has been formed by com
paction of cold, dry snow its bulk density is initially only about

461
o .8 Mg/m 3 . At depth, the stress field becomes close to hydrostatic,
and the air bubbles of the ice are compressed by overburden pressure,
their size being regulated more or less b y the gas laws. The com
pressibility oE the solid ice itself is small, and it is character
ized by an elastic bulk modulus of approximately 9 GPa. There is a
phase boundary between Ice I-h and water, and the freezing point
depression under pressure near O' C, given by the Clausius-Clapeyron
equation or by experime!1t, is O.074'C / MPa . The compression of snow
i!1 th e upper layers of a glacier accu mu l a tion area is much more com
plicated, involving large strains , non-hydrostatic stress states, and
nonli ea r viscosity . Many s tudies ha v e t r e a ted the self-weight com
paction of accumulating snow as a continuous v iscous proc e ss but, in
fact, snow adjusts its density to the overburden pr es sure ~uite

rapidly, and thereafter the volumetric straining is almost imgercep


tibly slow. Thus the process is more of a plastic collapse under
pr es s ur e , so that verti~al profiles of bulk density, or specific
vol u e , give a characteristic relation between d e nsity and pressure
,"hi c h can be described by a qua s i-plastic compression modulus that
jtse lf is a fun ct i()o of den si t y.
Ice fracture has not received a great deal of attention in
glaci e r s t udies. T he for~3tion of crevasses, which nas already b e en
m_:ltioned in terms oE .s train ra t " , can be treated in terms oE stress
t h ro ugh thlC con stit u tive equation, and tile depth of crevasses can be
an31 y zed by taking into account the compr e ssive stress component, and
th e c < n " " -!'lent Co!no r es sive creep, which increase with depth, ulti
m t e ly overrt din g th e eEf e cts of extending longitudinal strains . The
r e petitive tidal ELexure oE floating gla c ier tongues and ice shelves
h~ s also been tr ea t ed as part of the analysis of iceberg calving,
using f3ilure criteria Eormul a t e d in terms of stress.
Pluid mechanics has to be taken into consideration Eor some
Jla c ie ~ s , since water pcessure at the bed can reach as much as 50~ of
the ice overburden pcessure . fluctuations in meltwater Elow rates
and pressures appear to be the main c e so ns for seasonal, oc e v en
diurnal, fluctuations in glacier velocity.
En e ro e tics
En e r ge ti c~ deals mainly with considerations of mechanical work
or, for continuously working systems, mechanical power. However, all
Eo~ms of energy in the system, and all energy transformations, have
to be accounted for. In engineecing mechanics genecally, the energe

462
tics approach is often very useful for tackling complicated proDlems,
especially in non-dissipative systems, but in glacier mechanics there
has not been much systematic application of energetics.
The general principles are fairly obvious. The glacier consumes
potential energy as it descends a slope or spreads to lower surface
elevations. It acguires kinetic energy as very large mass moves at
very low speed. However, the flow process is strongly dissipative.
The internal deformation of the ice represents work done and power
consumed, the resulting heat being dissipated either by flow along
temperature gradients, or by internal melting of the ice when tem
perature gradients are small or non-existent. Sliding of ice on the
bed, which actually involves an ice layer of finite thickness, pro
duces a similar transformation of mechanical energy to heat. Local
addition or removal of mass at the glacier surface represents
addition Or removal of potential energy, and in principle subglacier
melting or mass redistribution is a factor. Heat flows into the ice
by geothermal flux from the underlying rock, and either in or out
through the upper surface, depending on the surface energy balance.
The vertical temperature, gradient either inside the glacier is also
affected by long term variations of surface temperature, caused
either by climatic change Or by advection of heat as an accumulating
surface descends to lower and warmer elevations.
Energy considerations suggest an intriguing possible mechanism
for flow instability and glacier surging. Internal shearing of the
ice produces heat, and if this cannot be dissipated by convection the
ice is either warmed or melted. Both warming and melting lower the
creep resistance of the ice, so that if the flow stress remains con
stant the rate of dissipation of strain energy increases and the flow
becomes progressively faster. For two-dimensional flow at great
depth, where normal stress approximates hydrostatic pressure, the
dissipation of strain energy gives a power per unit volume of
(1/2) Txy Yxy' where Txy is the simple shear stress and Yxy
the corresponding "engineering" strain rate. Because the flow stress
is very low, this specific power is also low. If we take Txy = 0.1
MPa and Yxy = 5 x 10- 6 s_ 1 , the specific power is 2.5 x 10- 3 W/ m3 ,
and in the absence of any heat flow the ice would be warmed at a rate
of 1.3 x 10- 9 C/s, Or about 0.04C/yr. Glaciologists have apparently
concluded that this effect is too weak to produce instability, but
one might guest ion whether they have used a realistic relation for

463
tempecatuce se nsitivity of cceep neac the melting point. If the ice
is alceady at the melting point, a specific powec of 2.5 x 10- 3 w/m 3
wou ld give cise to internal melting at the rate of about 8.2 x
10- 12 /S or 2.6 x 10-"/yc. In other wocds, in any block of the ice,
about 0.026% is melted per ye ar. Experiments sho w tilat creep rate
increases with liquid water content in ice at the melting point, so
presumably this is another feedback pro ce ss which could cause insta
bility.
Disgre s sing a little fcom the sub ject of energetics, th e re is
another hypothesis foc glacier sucging which postulates a double
valued celation between stcess and strain-rate, leading to the possi
bility of alternation between slow-flow and fast-flow modes in a
glaciec. An explanation of the physics of the celation has not been
put forwacd but, as we shall se e in a moment, thece is expecimental
justification for the idea if conditions ace such that s train soften
ing can occuc.

Glaciec Mechanics and Ice Engineecing


Although glaciec mechanics may s ee m fac cemoved from the major
pra ct ical concerns of ice engineering, it i s actually quite cl ose ly
related.
First o f all, thece ace engineecing pcoblems involving glaciecs.
Twenty yeacs ago we wece buccowing into the Gceenland Ice Cap and
bucying s tations o n the Antaccti c Ice Sheet, and deep dcilling in
glaciecs has been going on foc abou t thcee decades. In some pacts of
the wocld there ace pcoblems acising fr o m the damming of watec by
advancing glaciecs, with con seque nt cisk o f catastcophic dambucsts.
In mountain aceas, fluctuations in glaciec cegime can cause ice
avalanches, some ceaching tculy catastcoph i c pcopoctions. Hydcoelec
tcic pcojects and mining opecations sometimes involve tunneling and
dcilling in glaciecs. Thece ace some oce bodies that might be mined
by op en pit excavation in glaciec ice. In taCkling problems like
the s e, a sound kn o wledge of glacier mechanics is essential for good
engineering.
Anothec ceason foc linking glaciec mechanics to ice engineecing
is that the two areas of stud y tend to deal with t he same material in
two vecy diffecent canges of stcess and stcain rate, thus providing
complementary bodies of data. In laborator y studies, the main con
cern of glaciologists tends to be long-term flow under low deviatoric
stress, as measured by the constant-load creep test. By contrast,

464

--
the main conceen in ice engin eer ing tends to be deformation and
euptuee undee eapidly applied loads, with constant rate strength
tests dominating the scene. However, ceeep tests and steength tests
ought to give essentially the same infoemation if they are propeely
conducted, and consequently glaciologists and ice engineecs can learn
a lot feom each othec by lOOking at the various COllections of data
in appeopeiate way s . Along thi s line, I should like to finish up
this talk by discussing so me cecent work which compares the data for
ceeep tests and stcength tests.

Defoemation and railuee Undee Constant Steess and Constant Strain

eate.

In a eecent CRREL study, David Cole and 1 set out to show


expeeimentally that constant steess and constant strain-rate tests
give much the same information, and foe the range of conditions that
was coveeed, the expectations weee mOre or less borne out. The full
stoey is told somewheee else, and heee I shall just piCk out a few of
the points that ace eelevant to compacison of glaciological data and
ice engineering data.
riguee 1 gives steess /s teain curves obtained at _5C trom fairly
conventional steength tests made at constant deformation rate on a
closed-loop electro hydeaulic testing machine that has high inherent
stiffness and eapid feedback eesponse. ror steain rates up to about
10- 4 s-l, we usually see two p e aks, oe stress drops, on the curve.
The fiest, which we call the initial yield point, is the point at
which internal microceacks form in large numbees, as indicated by
eapid eise in the eate of acoustic emissions. The second peak, which
we call the secondaey yield point, is the point at which deformation
eesistance begins to deop iereveesibly towards an asymptotic limit
foe laege steains, a limit we call the eesidual strength. The secon
daey yield point occues at an axial steain close to 1%, while the
steain at the initial yield point increases with the applied strain
eate, eanging in these tests teom about 0.03% to almost 0.6%. As
steain eate inceeases, the initial yield point eventually dominates
and becomes the only identifiable yield point. However, at the
lowest steain cates applied in these tests, 10- 7 to 10- 6 s- l , the
initial yield steess became equal to, oe geeatee than, the secondary
yield stress.
We can look next at the results of constant stress tests made on
identical ice, also at _5 C . The applied stresses in these tests
weee chosen to conform with the peak stresses of the constant eate

465
tests, while the strain rates of the constant rate tests were chosen
to conform with the minimum strain rates of the constant stress
tests.
Figure 2 gives the results of high stress tests in the form of a
log-log plot of strain rate against time. The classic creep curve,
which plots integrated strain against time, is very flattering to bad
data, but it tends to suppress information about what is really going
on. In Figure 2, we see strain rate decelerating from an initial
rate down to a minimum, in what is usually called primary creep.
After the minimum there is acceleration, usually called tertiary
creep, which seems to be tending to a limit. The~e is no secondary
creep, if by that term we mean a protracted period when strain rate
is constant. The locus of the minimum point for this family of
curves is a straight line with a slope of -1 on the log-log plot,
which implies inverse proportionality between minimum strain rate and
the time needed to reach it. Making certain assumptions, this
further implies approximate constancy o f plastic strain at the mini
mum point, something which turns out to be true when the data are
plotted as strain rate against strain, as in Figure 3.
There is not much doubt that the minimum points of Figure 3
correspond to the secondary yield points found in constant strain
rate tests. Both represent a maximum value of the ratio stress to
strain-rate, or the inverse, both o ccur at about 1% axial strain, and
both give data which define a single relation between stress and
strain rate.
Another interesting thing about Figure 3 is that the strain rate
at 0.2% strain is proportional to the strain rate at the minimum
point. with a ratio of about 2.8. If this proportionality could be
confirmed for a broader range of conditions, it would open up the
possibility of predicting the minimum strain rate from short-duration
tests run to a s train of only 0.2%.
Getting back to the question of correspondence between creep
results and strength test results, we have no difficulty in identify
ing the secondary yield point in the creep curves, but where is there
any sign of an initial yield point ? The difficulty in answering this
question comes from the fact that we did not anticipate it during the
experimental program, and the data salnpling for creep at small
strains was too infrequent, which is why Figures 2 and 3 are chopped
off for strains below 0.1%. However, when we dig back through the

466
admittedly weak data, and ~e-plot on semi-log scales to avoid exag
ge~ation of the scatte~, the~e is fai~ly convincing evidence of an
initial yield point in c~eep, as can be seen in Figu~e 4. Fo~ those
who p~efe~ t~aditional c~eep cu~ves, the~e is an indication oE an
initial yield point in some of the t~aces taken E~om the X-y plotte~

(F ig. 5).
With two yield points, a complete c~eep cu~ve has a c~eep cu~ve

buried within a creep curve. Tests running to total st~ains of 5% o~

more give classic c~eep cu~ves showing the so-called p~ima~y, secon
dary and tertiary stages, and the effects of initial yield get lost
in the prima~y part of the curve, as can be seen E~om Figu~e 5. How
ever, tests that terminate at total st~ains of about 0.2% can also
give classic creep curves, with primary, secondary and tertiary
creep. Because most high stress tests are run to large strains,
while low stress test s almost invariably t e rminate at small st~ains,

it seems quite possible that the ice flow laws of glaciology are
based on inconsistent sets of data.
Another bothersome point about glacier mechanics is that glacier
flow is considered to conform with a stresG/strain-~ate relation
obtained from labo~atory experiments which terminat e at st~ains less
than 0.2% o~ so. Most oE the low st~ess tests terminate when the
creep curve appears to straighten out after the initial deceleration,
and the final strain rate is commonly referred to as "steady-state
creep." In the light oE general exp erime ntal evidence it seems
incautious to infer that this is a steady-state condition and, since
the glacier has all oE geological time at it s disposal, it is hard to
see why small-st~ain data should be d irectly applicable. If there is
such a thing as steady-state flow, it is much more likely to be
represented by the apparently constant strain-rate which de ve lops in
laboratory tests at axial st rains of 10-15%, when recrystallization
has had a chance to create preEer~ed crystal o~ientations.

Anyway, although ou~ understanding of the mechanical behaviour


of ice is still fa r from complete, we do have prima facie evidence of
a clea~ co~respondence between the results oE constant st~ess and
constant strain-rate expe~iments. This means that, within cectdin
limits, complete creep curves and complete st~ess/strain curves can
be used inte~changeably. In principle, one can even generate creep
curves f~om families of stress/strain curves, and, vice versa, for
example by reading off values of strain and strain rate for constant

467
stress level. However, in order to re-interpret glaciological data
and ice engineering data, we need the basic results rather than
results which have already been processed on the basis of question
able assumptions. Few research groups have published raw data or
maintained accessible data records, and it would be very useful if we
could all do a better job in this respect.

Conclusions
To conclude this sermon, I would like to suggest that for a full
understanding of ice mechanics we have to cover the complete field,
in which glacier mechanics is the oldest established branch. Most of
our knowledge of constitutive relations and failure criteria at low
stresses and low strain rates is derived from studies of glacier
mechanics. By re-analyzing some of the glaciological creep data, we
can probably gain more insight into the rate dependence of strength.
Finally, if we are smart enough to interpret the behavior of glaciers
in quantitative terms, we can perhaps use the glaciers themselves as
giant laboratories for measurement of the mechanical properties of
ice.

468
14
(2.20 , lOS 5 11

40
(1 .34,10 5 5- 1)
01
(6.09, 106 s l l 43 0. 125 -' 122
17811.16',11 ( 5 .311.10"',1)

52 (1.15,10 5 5 1)
(1.63,10- 6 sll
'"
II0l16',

0.-

o 7
A,ial Strain (%) 1 0'",--.-----r---,r---r--~-_r--,__,

23CD

(t , 14 l l d3~I)

e'

22
4
_(7 95 ... 10 ..- 1)
24CO
(t :o: 1 0311.10 4 5")

;;, 46
(132 lI0-5 5 1)
(113)[.16
13
5
$"')

10
(2.65 II; 10- 6 5- 1)

Aliol Strain (%) 546


(16 4 II. 10 ,1)

Figure 1

469
, 1'1'\ ", 1'1 I "' I , 1'1'1 "

~
o 100
a: 97 ' (185)
c (235)

L+ 94
(to)

96
(08)

Time (s)

Fi,:rure 2

470
, I I 1 r ',----r 1""- ' .,.--,- T1 I I 1 T
0

,,\
\~
~, '~~,
"'~ ,~
" '
~" .~
.
..

<

;;,
~8 ~ '2
~~
m;",
;;
"
I
<1

471
('III J I r- r
1'" If I !

.'" 1:
r r'I~' , ~

v -'
0 <..>
~
,., N
aT 2 2
N
d

GO

'" (, _9) ~IOH U!OJIS


'2
'" <1J
H

"
til

r
r
~tll

;::. t
!

~
aT
d

:!l
9

~Q
(,.i) aroH UIOJIS
2
J '2

-.~

~2
.,
'2
I,.S) ;)rOH V!OJ LS
,~1
'2 '2

472
' f'

473
"GLACIER MECHANICS" BY MALCOLM MELLOR

DISCUSSION
BY:

L. Gold, National Research Council, Division of Building Research,

Ottawa, K1A OR7 .

found the author's information on a minimum in the strain-rate - or


a yield - during the first 0.1% strain to be of great interest. We
found this unusual behaviour during creep in columnar-grained ice
several years ago. It corresponds to the initial yield at about 0.1%
strain for columnar-grained ice when it is tested under conditions of
constant nominal strain rate. I have attributed this unusual
behaviour to the fact that normally we test ice that has not been
subjected to previous deformation. During the initial straining of
the ice, modes of deformation must be established that will allow the
ice to conform with the imposed strain. Because of the limited
degrees of deformation ava ilable to individual grains, internal
stresses develop which ultimately result in initiating modes of
deformation such as cavity formation, formation of low angle boundar
ies, polygonization in the grain boundary region, grain boundary
migration and, if the stress is sufficiently large, crack formation.
These features are quite evident in columnar-grained ice because of
the significant constraints that are imposed on such grains. It is
of great interest that it can also be observed for granular ice for
which the constraining effects are less severe.

474

I
FIELD INVESTIGATIONS OF A HANGING ICE DAM

S. Beltaos, Research Scientist National Water l~esearch Insti tute Canada


(Formerly - Research Officer Alberta Research Council Canada)
A. M. Dean, Jr .. Electrical Engineer U. S. Army Cold Regions
Research and Engineer ing Labora tor y U.S.A.

ABSTRACT

A hanging ice dam that forms annually in the lower Smoky River, Alberta, has been the
object of continued inves tigation during the period 1975-1979 . The study aims at documenting
physical dimen sions and material properties of the dam; elucidating the mechanisms of its
formation and removal; and assessing its effects on the progress of breakup in the river. This
paper presents a summary of the results obtained to date.

INTRODUCTION

A hanging ice dam is a downward project ion of river ice, produced by deposition of frazil
slush under an existing i ce cover [15]. Typically, a hanging dam form s at a low speed section of
a stream, located immediately downstream of a high speed section. During freeze up, the latter
remains open while an i ce cover forms at the former section. Frazil ice produced in the rapid
flow sec tion agglomerates into slush and pans that are transported under the cover of the
tranquil section and deposit where the flow speed is sufficientl y low. Deposition continues until
either the upstream supply is discontinued or the flow ve locity under the accumulation increases
to a value ca pable of transporting the entire amount of incoming ice. The .l imiting velocit y
varies bet w een I m/s and 1.5 m/s depending on the composi tion and dimensions of the
transported material [ 14, 161.
Hanging dams are often mentioned in ice engineering literature [ 3, 5, II, 16. 171 but there
e x ists little documentation of their behaviour and effects. Gold and Williams [ll] described a
90 m deep and 1200 m long hanging dam in the Ottawa River. Such massive accumulations of ice
can obstruct the spring ice run and initiate major ice jams as well as being ca pable of interfering
with river structures.
The possibility of a hanging dam occurring in the Smoky River, about 40 km above its
confluence with Peace River (see Figure I) was first detected in 1974 by British Columbia Hydro

475
staff; the existence of the dam was confirmed by soundings in early 1975. Because of possible
effects of this hanging dam on breakup near the town of Peace River (Figure I). a long-term
investigation was initiated by Alberta Research Council. The main objectives of this study are to
document the formation of the dam, assess its effect on the breakup process in the Smoky and
Peace Rivers and examine whether impact for ces by moving fragments of such dams need be
considered in the design of river structures. This paper presents a summary of the results
obtained during the period 1975-79.

USA

Fig. 2 Oblique air photo of hanging cam site (looking


up stream, Dec. 2, 1975; note hummocked ice
Fig. I Location map. surface and open wa ter lead in rapids
upstream ).

SMOKY RIVER HANGING DAM

The st ream configuration in the vicinity of the hanging dam site consis ts of a deep and wide
section preceded by a section of rapids upstream; thi s sequence exhibits features conducive to
hanging dam formation and , to a degree, is illustrated in Figure 2. Also shown in Figure 2 is the
hummock ed and perceptibly elevated surface of the hanging dam. A longitudinal profile of the
dam, obtained in March 1975, is shown in Figure 3. The f razil accumulation is roughly triangular
with a base of 300 m and a maximum depth of IJ m below the water suface. These dimensions
vary from year to year. In January 1976, the length and maximum depth of the dam were 300 m
and 16.3 m while corresponding values for February 1977 were 700 m and 11.0 m. River cross
sections, located as indicated in Figure 3, are shown in Figure 4 where the deepening and
widening of the river near section 0-0 (deepest section) are well illustrated.
Open-water flow cond ition s at the dam site were documented in July 1975. It was found
that, due to channel expansion, two large eddies were present near the river banks, as sketched in
Figure 5; there was no evidence of the channel bed depression having been filled in by sediment

476

z
g
00

1 <3
i1J
I
~
I
~
Ii
~I
'tIIfII1~r

:ll
Ii
I
sur &k:e .....U (7'lS

J. i! &l
iii ill Ie
I I !
:ll~iil
1i~1e
II I
" l!I
reI 11I ~
J.
ill
I
;
I l ~? ~ c::::
2

4Ei
...., ,_.

X- SEC-335
' --9 frat,!

X-SEC-l51

~ Sll 50 00

~ Nelle ()epIh MNSIUf8d from tree wa lei'" surlace


go" abO.oeic.e
iii
II:

"
-~ -~ 0 roo .Jx,
DISTANCE FROM CROSS-SECTION 0-0 1m)

Fig. 3 Longitudina l profile of ha nging dam (M arch 25

an d 26 , 1975)

HORIZONTAL DISTANCE (1'I"61res)


_ _v-- - ";>
/ / ,.... /'J...F
~~; /'r ..
?-= '9?~ m

~l ~"'~~
. . . ~~
''''" 3"~ ~.- . 50
l - I. __ 50
.~~ ._
tOO "

<{;:. Fig. 4 Rive r cross sections at hanging dam site (March


25 and 26, 1975)

Fig. 5 Sketch of flow pattern a t hanging

dam site (Ju ly 30,1975)

""'
""'
The same was reported for the Ottawa River dam [l 11. A vertical veloci ty profile, taken in the
live stream at section 0-0, indicated an average velocity of 1.2 rnls which is comparable to
values measured in the rapids section upstream of the dam site. This explains why the river bed
depression is not filled with sediment in the summer: as flow veloci ties and, hence, transport
rates are comparable upstream of and at the dam site , deposition is not likely to occur.

FORMATION

Formation of the hanging dam was documented in November 1978 by means of an 8 mm


movie camera, programmed to expose one frame per minute during daylight hours and installed
near the top of the west valley wall. The resulting film shows freeze up events for a period of
six days and provides a fair description of the dam format i on mechanism. At the dam site, the
ice cover is initiated at the edd y areas where frazil floes and pans recirculate and eventuall y
become shore fast. Gradually, the firm ice Cover extends outward from the banks toward the
midstream and somewhat upstream. As this occurs, the eddies also move upstream which enables
continued build up of the co ver. Eventually, onl y a narrow strip cf open water is left at
midstream, corresponding roughly to the live stream under open-water condi tions (see Figure 5).
This strip is finally bridged by an arching mechanism similar to that studied by Calkins and
Ashton [ 6 I. For the 1978 freeze up, surface flux of i ce pans began during the night of November
8 to 9 and a complet e ice bridge across the river formed by the morning of November 14
Temperature records at Watino r II indicate that No vember 8 was the first day of sustained frost
while the average air temperature dur i ng the formation period was about -13.5C.
It was mentioned earlier that the limiting frazil slu sh deposition veloc ity is in the range 1.0
to 1.5 m/ s [ 14, 16 1. Neglecting seepage through the dam and using the flow areas from Figure
4, average velocities under the dam are about 0.12 m/s for March 1975 (discharge=)) m)/s).
3
Allowing for a freeze up discharge of about 100 m Is (November 1974), the corresponding freeze
up velocity is estimated as 0.20 m/s. This is much less t k ln the limiting deposition value which
suggests that verti cal growth of the dam is limited by a discontinuation of ice supply due to
free z ing over of the rapids usptream. Vertical velocity profiles under the dam were measured in
1977 and 1978 using a magnetic flow meter. With the exception of one profile, the measured
values [4 I are well. below 1.0 m/s which reinforces the above suggestion .
The measured velocit y profiles showed further that the absolute roughness of the dam
undersurface is highly variable, being somet i mes less and sometimes more than that of the river
bed. From semi-logarithmic plots of the velocity data, the average friction fa c tor and
equivalent sand roughne ss were estimated as 0.08 and 0.8 m respectively.

MA TERIAL PROPERTIES

At the time this study was initiated, no information could be found on material properties
of frazil accumulations. Such information was thought important in engineering applications,

4 78

I
such as assessing effects on ice breakup, forces on river structures and flow through
accumulations. The following is a summary of pertinent findings to date.

Compost tion

The non-submerged portion of the hanging dam ("overburden") consists of a hummocky


accumulation of snow and weak granular ice with a maximum thickness of 2 m; the latter can be
classified as S5 Ic e "drained congealed frazil slush", using the terminology of Michel [151. A
6 em thick layer of solid ice topped the overburden near the river banks, extending to
approximately the live stream boundaries under open-water conditions. These findings are
quali tatively similar to those concerning the Ottawa River dam [Ill.
Near the free wa ter surface, there is a 0.3 to 0.9 m thick layer of solid ice, underlain by
the main (submerged) accumulation of frazil. The latter is fairly dense frazil slush, similar in
composi tion to the overburden material; its pores are saturated and its cohesion is much less than
that of the overburden. The overburden originates from saturated frazil that rises above the
water surface and drains as the accumulation grows in depth. The solid ice layer near the water
surface forms from the slush as its crystals have random orientation and sizes comparable to
those of the submerged frazil. The conductivity of this layer was found to vary in the vertical
direction (1976), being 7.3 and 13.7 ~mho/cm at respective depths of 0.1 and 0.3 m which
suggests impurity migration (total sample depth=0.37 m; see also [4]). The thin top layer of ice
near the banks is ordinary ice that forms at the eddy areas prior to significant frazil deposi tion
underneath and consequent emergence above the surface. That no such layer has been found in
midstream suggests that frazil accumulation in the live stream area is much faster than in the
eddy areas.
The ice particl es in the saturated slush are between spheroid and discoid in shape with a
major diameter of 1-6 mm. The size distribution by weight is approximately: 60 percent in the
range 1.1 to 2.4 mm; 35 percent in the range 2.4 to 4.8 mm; and 5 percent in the range 4.8 to
6 mm.

Shear Strength and Bearing Capacity

The shear strength of the slush was measured by means of shear va~es attached to a series
of 1.5 m long extensions. Torque was applied and measured with a commerCially available torque
wrench. Figure 6 shows shear strength values (-r ) measured in 1976, plotted versus depth at
f
three locations; two vane si zes were used for comparison (two holes spaced I m apart were
drilled at each location). The scatter in Figure 6 is typical and illustrates both the crudeness of
the measurement technique and the natural variabili ty of the strength. No consistent variation
of 'f is evident in Figure 6 but later data have shown that 'f increases generally with height
above the bottom of the accumulation (see, for example, 1979 data in Figure 7); this trend is
sometimes obscured by scatter. Figure 8 gives a summary of shear strength measurements,

479
plotted in the form of depth-averaged T f versus accumulation thickness. The shear strength is
seen to vary from year to year.

11 lfl, eoT TQf.!. CF- FRAIl!.

Fig. 6 Vertical profiles of shear strength Fig. 7 Vertical profiles of shear strength
(March 10 and II, 1976) (March 14, 1979 ; 10 . 2 cm vane)

-;;;
Cl.
~ 60
~
0 ~'''-''m'
40
'"
.2 Fig. 8 Depth-averaged shear strength
>'" ~r'14'1979 versus accumulation thickness

r
TI
Q)
Ol 20 Mar. 14-15,
["
<lJ 1978
Feb 22-23,

.i{ 1977

s;;;
a. 0 10 20

0 '" Thickness 01 Accumulation (m)

The results of plate bearing tests exhibited large scatter, but average values increased with
height of observation from the bottom of the accumulation (h ), being 300, 150 and 90 kPa at
f
values of h equal to 11.2, 5.3 and 2.3 m respectively.
f

Densi ty and Parosi ty

The dry density of the slush (p ) obtained from the drained weights of known volumes,
f
increased with h as shown in Figure 9. The porosity of the accumulation (Efl is given by:
f

(I)

in which 0i=density of ice. From Figure 9, E f is calculated as 0.51 and 0.33 at h f =2m and 12m
respec ti ve I y.

480

l600~ - , l~
' : ' '1
Iu : "".": i
a 0 10 . . 20
Height Above BotTom of AccurT1\JlatIOf), hI (ml
Fig. 9 Dry densi ty versus height above
bottom of accumulation .

Using average values of 1976 data on of (Figure 9) and 'f (Figure 6) at corresponding
heigh ts hI' an empirical correlation was obtained, as follows:

3
in which, f is in kPa and P is in Kg/m , Equation 2 applies in the ranges T f=30-75 kPa and
f
3
P =450-620 Kg/m ,
f
Figure 9 may be used further to determine the stress-densi ty relationship for the
accumulation. The vertical stress gradient due to buoyancy is:

in which p=vertical stress; g=acceleration of gravity; and pw =density of water, USIng Equation I
and integrating gives:

(4)

An approximate calculation using graphical integration (see Figure 9) resulted in the stress
density relationship depicted in Figure 10 along with relevant findings for snow [13]. For the
same stress level, frazil densification is about 1.5 times that of the lower bound of Mellor's [13]
data. This is primarily caused by differences in temperature and water content and, to a lesser
degree, by particle geometry effects. The mechanics of densification change significantl y near
OOC in a saturated media where pressure melting and regelation [71 s trongly affect the
deformation of frazil ice and allow densification at a lower stress level than would be found in
dry, colder media. Colbeck et al [8] discussed this difference and reported test results that
0
compare saturated with dry snow at _2 C. If the same densification is considered linear and
applied to the present resuJ ts, the transformed data fall much closer to Mellor's (Figure 10).

Intrinsic Permeabili ty

Permeability was calculated based on flow rate of a lOW motor oil through a cylindrical
sample under a fixe d head [10]. Th e equation used is (2):

481
10'.-----. -- ----r-----,------, 1r.r----.-----'
'

o
"
~ ----:-/=-- Mellor (1974)
Fig. 10 Comparison of stress
~ 10 4 [0-48'CJ density relationship for
if, o the frazil accumulation
o with Mellor's data for
"0.

:~
o
snow.
a:
(5 IO~
.0' o
::E
0: o Present 00 10
Saturated \I S Dry Tronlo/crmOhon

10' L-____- L______~----~------~~----~--- ---~

200 400 600 800

PI' De nsily (kg/m l )

(5)

in which Q=flow rate; kf=intrinsic permeability; A=cross-sectional area of test cylinder; ~ =fluid
viscosity at test temperature; p=fluid density at test temperature; h'=head=distance from top of
input reservoir to tip of drain tube; and L'=length of test cylinder. Measured k values were
-6 -6 -62 f
16. 3x 10 ,15.6x 10 and 15.0x 10 em
at h values of 2 m, 7.6 m and 12.2 m. The hanging dam
f
permeability is between those of coarse sand and fine gravel [12] which appears reasonable since
the frazil particle size is consistently between I and 6 mm. Snow with 1 to 2 mm particles has a
-6 2
k valueof2xl0 cm [9].
f

BREAKUP
Breakup observations have been carried out annually during the period 1975-79. Detailed
information may be found in [ ~]; only a brief summary will be given here.
The hanging dam obstructs the progress of the spring breakup and initiates ice jams, most
of which are major. Removal of the dam is usually forced, that is, it shears off at the sides
(roughly at the live stream boundaries) and is subsequently broken into small pieces upon the final
release of the jam upstream. There was one instance, however, when the upstream ice passed
under the dam; the lat ter remained in place for several days and was removed gradually by water
erosion (1977). Twice ([976, 1979), removal of the dam was followed by surging ice runs that
were only arrested 2 km upstream of the Smoky River mouth (about 38 km downstream of the
dam site); on both occasions, major jams formed there and gradually broke through into Peace
River. The effect of the dam on breakup near the town of Peace River (Figure J) can be either

482
beneficial or detrimental depending on prevailing ice condi tions in the Peace River itself.
Continued annual observations are deemed desirable so as to obtain a more complete record of,
and assign frequencies to, various events of interest.
To develop a c ri terion for the removal of the dam, an approximate force analysis was
carried out [4 J, as outlined briefly below . Upon release of the jam upstream, the main
horizontal force on the dam is a net hydrostatic pressure caused by the advancing water wave
(Figure 11); other forces, e.g. hydrodynamic force and pressure of advancing ice jams are
relatively very small in this case. The dam shears 0[[ when the applied for ce exceeds its
resistance on two vertical surfaces which se parate the grounded portions of the accumulation
near the banks from the floating portion in midstream. Analysis has shown that the dam will be
removed when

(6)

in which W=distance between the two shear surfaces; T=average shear stre ss over the sheared
area; and ST=toe slope of the upstream jam just prior to release. Detailed breakup data taken in
1975 indicated that ST was in the range 0.0043 to 0.007. Using W~70 m (see Figure 4), Equation
6 gives T=3 to 4.8 kPa which is generally lower than measured midwinter values shown in Figure
8. It is noted that a decrease in strength is lik e ly during the spring breakup if the water
temperature ri ses above OOC. If T and W do not change appreciably from year to year, Equation
6 would suggest that there is a limiting value of ST' betw ee n 0.004 and 0.007, that mus t be
attained before the dam can be removed. This is consistent with the 1977 finding, i.e. that the
dam did not "break" and the upstream ice passed under it: the avaolable data for 1977 indicated
that ST could not have exceeded 0.0039.

former toe section

Prior to jam relea se:


Shortly after jam release:
negligible horizontal component
significant horizontal component
of net hydrostatic pressure
of ne t hydrostatic pressure.

Fig. II SketCh of assumed mechanism of dam removal

483
SUMMARY AND CONCLUSIONS

A hanging ice dam that forms in the lower Smoky River has been the object of annual field
observations and the results have been reported in the previous sections.
The hanging dam site is a depression of the river bed, preceded by a section of rapids. The
mode of dam formation is essentially as has already been described by others; site-specific
peculiarities have been identified, based on a visual record obtained by means of an automatic,
time-lapse photography appara tus.
The streamwise profile of the dam is roughly triangular, with a base of 300 to 700 m and a
depth of II to 16 m; the dam consists of porous frazil slush with ice particles 1 to 6 mm in size.
The in si tu shear strength of this material varies f rom year to year; in anyone year, it increases
with height above the bottom of the accumulation and generally does not exceed 80 kPa. A
similar variation was found for the dry density of the material. The intrinsic permeability of the
2
dam is about 15.5xl0-6 cm and decreases slightly with height above the bottom of the
accumulation. Velocity measurements under the dam indicated average values of 0.08 and 0.8 m
for the friction factor and equivalent sand roughness height of the dam underside respectively.
During spring breakup, the dam initiates an ice jam upstream. Usually, final release of this
jam is followed by removal of the dam and occasional ice surges that are only arrested near the
river mouth, 38 km downstream. On one occasion, jammed ice upstream released and was
transported under the dam rather than dislodging it. To explain the mechanism of dam removal,
a preliminary force analysis has been carried out and partly documented using available data.
The effect of the dam on spring water levels near the town of Peace River can be either
beneficial or detrimental depending on simultaneous ice conditions in Peace River itself.
Continued observations are deemed desirable in order to develOp an adequate statistical record.

ACKNOWLEDGEMENTS

A portion of the work reported herein was carried out as part of a continuing field research
program on river ice hydraulics; this program is conducted by the Transportation and Surface
Water Engineering Division of Alberta Research Council in cooperation with Alberta Environ
ment, Alberta Transportation and University of Alberta.
Occasional assistance provided by B. C. Hydro observers (F. Sampson, B. Tutt, M.
Vanderkraan) is gratefully aCknowledged. G. Childs, G. Putz and T. Ridgway of Alberta Research
Council participated in the field work. Review comments by T. M. Dick, Y. L. Lau and G. Tsang
of Environment Canada are appreciated.

REFERENCES
[1) Atmospheric Environment, 1978. Monthly Record - Meteorological Observations in Western
Canada". November, Vol. 63, No. 11, Part 1.
[2] Amyx, J. W., Bass, D. M. and Whiting, R. L., 1960. "Petroleum Reservoir Engineering
Properties". McGraw-Hill, New York.

484
[3] Barnes, H. T., 1928. "Ice Engineering". Renouf Publishing Company, Montreal.
[4] Beltaos, S. and Dean, A. M. Jr., 1981. "Field Investigations of a Hanging Dam". NWRI
Hydraulics Division Unpublished Report (in prep.).
[5] Bolsenga, S. J., 1968. "River Ice Jams - A Literature Review". Research Report 5-5, U.S.
Corps of Engineers, Lake Survey District.
[6 I Calkins, D. J. and Ashton, G. D. , 1975. "Arching of Fragmented Ice Covers". Canadian
Journal of Civil Engineering, Vol. 1, No.4, pp. 392-399.
[ 7 I Colbeck, S., 1976. "Thermodynamic Deformation of Wet Snow". U.S. Army CRREL Report
76-44.
[8] Colbeck, S. C., Shaw, K. A. a nd Lemieux, G. 1978. "The Compression of Wet Snow." U.S.
Army CRREL Report 78-10.
[ 9] Colbeck, S. C. and Davidson, G., 1973. "Water Percolation through Homogeneous Snow".
Internat. Symposium on the Role of Snow and Ice in Hydrology, Vol. I, pp. 242-257.
[ 10 I Dean,A. M., 1976. "A Method for Determining the Permeability of Frazil Ice". U.S . Army
CCREL, Technical Note (unpublished).
[J I ] Gold, L. W. and Williams, G. P., 1963. "An Unusual Ice Formation on the Ottaw a Rive r".
Journal of Glaciology, Vol. 4, No. 31, pp. 569-573.
[121 Hough, B. K., 1969. "Soil Moisture". In Basic Soils Engineering, Ronald Press, New York,
Chapter 3, 76 p.
[13] Mellor, M, 1974. "Mecanique de la neige". Proc. Grindelwald Symposium, April, IAHS
lASH Publ. No. 114.
[I4J Michel, B, 1971. "Winter Regime of Rivers and Lakes" . U.S. Army CRREL Monograph IU
Bla.
[151 Michel, B., 1975. "The Formation of Ice Covers". Universit e Lav a l Report GCS-75-09-05.
[16] Michel, B., 1978. "Ice Accumulat ions at Free ze-Up or Break-Up". Proc . IAHR Symposium
on Ic e Problems, Lulea, Sweden, Part 2, pp. 301-317.
[17] Sampson, F., 1973. "The Ice Regime of the Pea ce River in the Vic inity of Portage
Mountain Development, Prior to and During Diversion". Proc . Seminar on Ice Jam s
in Canada, University of Alberta, Edmonton, Published a s NRC Technical Memoran
dum No. 107, pp. 158-188.

485
DISCUSSION

R. Gerard, Canada

This paper is a sum mary of the results of a large amount of field work carried out at a very
difficult site. The authors are to be congratulated for producing the first thorough
documentation of the significant engineering characteristics of a hanging dam. It is mentioned in
the paper that hanging dams are of inter est because they can obs truc t the spring ice run, as we ll
as interfere with enginee ring struc tures. An extremely good example of the latter is described in
a recent paper by Cheung and Gui llaud (1981). In this example it was backwater created by the
hanging dam that was important. Prediction of similar backwater effects was also fundamental
to the study described by Hopper &. Raban (1980). A knowledge of hanging dam characteristi cs
is obv iou sl y very important and has quite broad application.
An imp ortant proble'n in predicting hanging dam development is calculating the amount of
frazil produced in the open water upstream and estimating what portion of the volume produced
will be incorporated in the hanging dam. In this regard, because the size of the dam was
apparently limited by a discontinuat ion of the ice supply, it would have been of i nterest to
document the variation in the open water area upstream and to atte,npt to predict the size of t he
hanging darn from estimates of the frazil produced. If nothing else, such a calcu lation would
have provided sorne more information on which to base the important decision on whether the
growth of the dam was lim ited by ice supply. The more-or-Iess constant waterway area under
t he dam suggests that pe rhap s it wasn't ic e supply limited . If ind eed it wasn't, the limiting
deposition veloc it y was 0.1-0.2 mis, a value qUite different [rom that acc epted in the literature,
and therefore a very important observation. With regard to the criti ca l deposition velocity,
wouldn't a more sigllificant parameter be the shear ve locity a t the ice surface, a parameter
probably better deduced [rom the mea su red ve locity profile s than from the average friction
factor given in the paper.
~o mention is made in th e paper of the nature of the stream bed . Although it ha s no direct
influence on the nature of the accumulation, it is of interest w hen pondering why there should be
slJch 2. large hole in the stream bed at this location.
Doubtless tlie breakup force analysis is offered with some "tongue-in-c heek". However, it
would be useful if the authors would indicate whe ther the stated failure mode (i.e shear at the
live boundaries) was obse rved as it happened or was deduced from evidence found after the
failure, and just when and how the val ue of ST was measured. Is there any possibility t he "keyed
in" nature of the accumulation cou ld have governed the measured slope, rather than the shear
failure that wou ld have had to occur before the ke ying in becam e effecti ve?
In the introduction mention was made of assessing the forces imposed on river structures by
such accumulations, yet the su bject is not mentioned again in the paper. A comment on this
po int would be useful. It would seem the force could be significant on a large ri ver where
accumulations 10 m or more thi ck cou ld mo ve along the river. For example, this was a concern

486
reported by Tom Lamb for pier design for a bridge over the Mackenzie River near Fort
Providence, where thick frazil accumulations can apparently be expected.
[18] Cheung, J. L. and Guillaud, C. H., 1981. "Effects of Ice Progression During Construction of
Muskrat Falls Hydropower Development". Proc. 5th Canadian Hydrotechnical
Conference, Canadian Society for Civil Engineering, Fredericton, New Brunswick,
May.
[19] Hopper, H. R. and Raban, R. R., 1980. -"Hanging Dams in the Manitoba Hydro System".
Proc. of Workshop on Hydraulic Resistance of River Ice, Canada Centre for Inland
Waters, Burlington, Ontario, September.

Reply by S. Beltaos (Canada) and A. M. Dean, Jr. (U.S.A.)

The writers would like to thank Dr. Gerard for his comments and interest in the paper and
for bringing to their attention two additional references.
The discusser has correctly suggested that, from the available information, one cannot
completelY rule out the possibility that deposition velocities could be as low as 0.1 to 0.2 m/s.
Though such values are much less than what has been reported in the past [14, 16], a certain
variability is conceivable through changes in the characteristics of the transported frazil. More
detailed studies in the future should help clarify this question. Shear velocities were calculated
from measured velocity prof ties and used to obtain local friction factor values. These varied
considerably, indicating a highly irregular bottom surface of the dam. The average friction
factor given in the paper could be used to estimate the average shear velocity in cases where the
flow area and discharge are known without resorting to detailed velocity profile measurements.
The shear velocity applicable to an average velocity of 0.1 to 0.2 m/s is estimated as 0.0 I to 0.02
m/s.
The bed material is generally gravel. However, a scoop sample taken at about the deepest
section of the dam site under open-water conditions, was medium sand. To consider the question
of the origin of the "hole", a soils investigation would seem prerequisite but access would be a
major problem at this particular site.
The breakup force analysis is tentative and was offered mainly as a means of stimulating
more detailed studies in the future. The assumed mode of failure was deduced from in si tu
observations during and after breakup and by taking into account the site geometry. Side shear is
very difficult to perceive visually when the dam begins to move because the initial movement is
very slow. The writers' view is that side shear failure occurs first and the dam moves
downstream; the "keying-in" mentioned by the discusser probably occurs somewhat later when
the dam has moved some distance downstream. The value of ST used in the paper is the
maximum water surface slope measured upstream of the dam before its movement.
It may be too early to comment on the forces applied on river structures by moving
"chunks" of hanging dams. If such forces are caused by a bearing-type failure and if our bearing
strengths obtained from small-scale tests are assumed representative, the effective pressure

487
should be no more than about 250 kPa. This may produce significant forces when a structure is
at tacked by very thick chunks such as those mentioned by the discusser. However, the above
figure should be considered hypothetical until further evidence answers such questions as what is
the actual mode of failure of hanging dam fragments impacting on structures; and how applicable
are small-scale test results to prototype conditions.

J. C. Tatinclaux, U.S.A.

1. The empirical equation 2 was arrived at from data of March 10-11, 1978. Was it confirmed
by data obtained at other dates as shown in Figure 8.
2. Figure 8 shows that the depth averaged values of T for February 1977 are lower than for
f
any other year; however, 1977 is the only year when the hanging dam did not fail. Could
the authors comment on this apparent contradiction?
3. Regarding the statement given at the top of page 9, it seems to me that the limiting value
for ST was obtained from the field data, and that Equation 6 thus yields the co rresponding
range of limiting values of the shea r stress T; not the other way around.

Reply by S. Bel taos (Canada) and A. M. Dean, Jr. (U .S. A.)

The writers would like to thank Dr. Tatinclaux for his comments and interest in the paper.
Below are itemized replies to Dr. Tatinclaux's questions.

1. Equation 2 was based entirely on data taken on March 10 and 11, 1976. Dry density was not
measured in subsequent years.
2. !he values of ST and T were deduced from the 1975 breakup observations; corresponding
mid-winter values of T are not available (see Figure 8). Moreover,:; cannot be considered
f
equal or uniquely related to T f because of possible strength deterioration prior to breakup.
Because of the many unknowns involved, the present analysis is tentative and has been
offered mainly as a means of stimulating more detailed study in the futur e . At the same
time, the analysis has identified ST as a relevant index of the driving forces while the
available data show tha t the 1977 value of ST was much less than the 1975 value.
3. The discu sser's statement is correc t. If futur e st udies advance the state of knowledge to a
point that -:;: can be estimated, Equation 6 (or a more advanced criterion) could be applied
to predict the value of ST needed to cause breakup of a hanging dam.

488
A COMPARISON OF SEVERAL CHEMICALLY-DOPED

TYPES OF MODEL ICE

G.W. Timco Hydrau Ii cs Laboratory Ottawa, Ontario


Research Officer Nat ional Research Counci I Canada

ABSTRACT

At the recent POAC '79 conference, the author presented a progress report on a

project initiated at the National Research Counci I of Canada to look for an alterna

tive dopant to sodium chloride for refrigerated model I ing basins. This report

included a discussion of the growth mechanism of model ice, a chemical survey to

define potential dopants for model I ing basins, a description of small scale tests on
several different chemically doped ice sheets, and a description of the physical pro

perties of large ice sheets doped with some of the better candidate dopants (methyl

alcohol, sodium acetate, carbamide). At that time it was reported that there were
other candidate dopants which remained to be tested in the large test tank. This
testing is now complete. The results indicate that there are several chemicals

(including carbamide, I ithium chloride, formamide, sodium propionate) which produce


ice sheets with a better strain modulus to flexural strength ratio than sodium

chloride doped ice. The advantages, disadvantages and proper range of scal ing for

each of these chemically doped types of ice are presented.

489
1.0 INTRODUCTION
In refrigerated towing basins, it is important that the model ice has a strain
modulus (E) to flexural strength (of) ratio in the range 2000-5000. In the past, this
has been accompl ished by using a saline-doped ice sheet and a "warming" technique for
model ice in the flexural strength range of 50-60 kPa [1,2]. However, from an econo
mics point of view, it would be much more desirable to use lower strength ice in order
to have a larger scal ing factor for the tests. For lower strength sal ine ice, this
results in an unacceptable E/O ratio of less than 1000.
As such, an experimental
f
program was initiated at the National Research Council of Canada to look for an alter
native chemical dopant which, when added to the solution before the freeze, would pro
duce an ice sheet which had the proper E/U ratio for lower flexural strengths.
f
There were basically five parts to this program: (I) A review of the morphologi
cal properties of chemically impure ice to determine the factors which control the
strength of the ice. This study found that the ice strength was controlled by the
thickness of the mechanically hard incubation length ("upper layer") of the ice sheet,
the size of the individual cr ystals, especially in the upper la yer, and the size and
al ignment of the impurity pockets in the ice sheet [3,4] . (2) A chemical survey to
define those chemicals ",hich could be used in a refrigerated. towing basin. The
results of this survey indicated that the homologous series of chlorides, sulfates,
alcohols, salts of fatty acids, surfactants, emulsions, sugars, amides and simple
amino-acids were suitable for testing [4,5]. (3) Small scale tests to determine the
effectiveness of these chemicals in reducing the strength of the ice. The results of
the s e test s indicated that the most effective groups of chemicals were the R.OH
alcohols (R" alkyl radical = C H + ), the salts of the fatty acids (R.COONa) and the
n 2n I
amides (R . CO.NH ) [4,6]. (4) A test prograln to measure the strength and strain
2
modulus properties of ice grown from large ice sheets containing these chemicals to
check the suitabi I ity of the E/Of ratio properties and the structure of these types of
ice for use in a refrigerated model ling basin. This study forms the basis of this
paper; and (5) A complete te s t of all of the mechanical properties of the "best" model
ice found from thi s study [7,8].

2 .0 EXPERIMENTAL
The ice tested in these experiments was grown in a swimming pool (6 m x 3 m X I m)
which was built inside a large walk-in cold chamber. In adding the chemical dopants
to the water, the amount added ",as determined by .,eight, and an air bubbler system
"as used to ensure thorough mixing of the chemical in solution. Before growing an ice
sheet, the solution was pre-cooled thoroughly at an ambient temperature of +2"C . In
the test s reported here, the ice s heets were wet-seed nucleated. For this, the air
temperature was lowered to -20"C after the pre-cooling period. When an ice skin
started to form over a part of the pool, the refrigeration equipment was shut off, and
the ice on the water's surface was quickly cleared using a ful I-width screen which
was pulled along the length of the pool. The solution was then wet-seeded using two
water bottles attached to air I ines equipped with #ISO (air)- 1/40100 (water) spray
nozzles. This produced a fine misty-like fog in the room which settled on the water's
surface and nucleated the ice sheet to produce a fine cr ys tal I ine pattern. When an
ice s kin had formed covering the whole surface, wind deflector s were placed above the
surface to ensure that the initial ice formation grew under quiescent conditions, and
the refrigeration equipment was restarted to freeze at -20C. After ~3 hours of
freezing time, the end wind deflectors were I if ted to provide a wind chi II factor to
speed the ice formation. Typical freezing times were 16 - IS hours at -20C. This
produced ice of 4-5 cm thickness. Fol lowing the freeze, the refrigeration equipment
was shut off and the chamber was allowed to slowly warm up. Since this ice is
impurity doped, this procedure increase s the liquid impurity volume in this ice. This
"warming technique ll reduces the strength of the ice [1,2].
Periodically during this warm-up period, the strain modulus (E) and flexural
strength (Df) of the ice were measured. The strain modulus was measured using the
plate method whereby the ice sheet was loaded in the centre of the pool wi th a known
increas ing load (P) and the resul t ing deflection (0) was determined us ing a standard
dial indicator. This produces a load-deflection curve from which the characteristic
length (9,) of the ice can be determined from [9,IOJ

( O.SP ] 1h (I)
9. = l21TpgO

where P is the density of water and g is the gravitational acceleration. The s t ra in


modulus (E) is then determined by

(2)

where h is the thickness of the ice cover and ~ is Poisson's ratio (see [IOJ for more
detail). It should be noted that these equations are applicable only when the sides
of the pool are at least 39. from the point of loading. Since this was not always
achieved for the larger E values (E ? 120 MPa), these measured values were corrected
to el iminate the edge effect by using the procedure de scribed by Edwards and Kotras

[IIJ. As such, all strain modulus values should be representative of an infinite ice
sheet.
Immediately following the determination of the strain modulus, the flexural
strength of the ice was measured by cutting a set of 4-6 cantilever beams (length
L ~ 20 cm, width w ~ 10 cm, thickness h ~ 5 cm) along one end of the pool and using
instrument push-pul I gauges to load (p) the beams by pushing down on the free end .
The flexural st rength of the ice was determined from

491
6PL
(3)
~

Typical loading times to failure were on the order of 1-2 seconds.


Since the above described procedure was repeated several times during the day as
the ice sheet warmed up, a series of E - of points were obtained for each concentra
t ion of dopant tested . As such, the resul ts of the present experiments are presented
in the form of strain modulus (E) - flexural strength (of) graphs which show individ
ual curves for each concentration of the particular dopant being tested. I t should be
noted that each of these curves tending towards lower E - o f values actually shows the
effect of the warm-up period on the physical properties of the ice sheet.
Finally, in some cases (depending on the availability of a cold chamber for this
type of study), samples were taken from these ice sheets and thin sections were made
to examine the crystallographic features of the ice (12).

3.0 RESULTS AND DISCUSSION


In this section, the results of these experiment s are presented along with a
brief discu s sion of the general appl icability of each individual type of ice for use
in refrigerated modelling basins .

3.1 Sodium Chloride


The first series of tests were performed on ice grown from a wet-seeded 0.6% NaCI
solution, primarily to verify the present test techniques in comparison with those
used in other model test basins. This type of solution is used by the HSVA ice test
basin in Hamburg, Germany, and it was chosen because its behaviour has been well docu
mented in the I iterature. According to Sch,larz (1,2] during the warm-up period after
the freeze, the ice exhibits a broad maximum in its E/o ratio on the order of 2000
f
for strengths about 50 kPa. This general behaviour is in agreement with that observed
by the present experiments (Fig. I). This is the optimum E/o f ratio for NaCI doped
ice, and represents an accurate scal ing factor (\ ) on the order of 15-20.

3.2 R.OH Alcohols (Methyl Alcohol)


From the R. OH group of chemicals, onl y the 10.lest member of the series, methy l
alcohol (R = CH ) was used as an ice dopant in these tests . In this case tests were
3
done at only one concentration (0.3 %) using the wet-seed technique for nucleation. The
E - of results of these tests are shown in Fig. 2. For this type of ice, the Elo
f
ratio is well above 2000 for strengths down to ~40 kPa. From the mechanical proper"
ties point of view, this type of ice would be a better representation than sal ine
doped ice as a model ice for refrigerated towing basins. However, structurally this
ice is not suitable since it consisted of a hard upper layer 001 cm thick from which
unconnected dendritic platelets "Iere suspended vertically (20 0 ) into the melt. As

492
SODIUM CHLORIDE METHYL ALCOHOL

300 300
o o
a.. a..
:::;; :::;;
I I
(/) 200 (/) 200
::J ::J
...J ...J
::J ::J
o o
o o
:::;; ::<
zlOO zlOO
<t <t
a: a:
t- l
(/) ~:'- 0.6"l0 FIG .I (/) FIG.2
O~ __L -_ _L -_ _L -_ _L -_ _L--...J O~ __L -_ _L -_ _L -_ _ L-~L-~

o 40 80 120 o 40 80 120
FLEXURAL STRENGTH - kpo FLEXURAL STRENGTH - kpo

SODIUM ACETATE SODIUM PROPIONATE

300 0.4"10 -1!


0.60;'0-'
? 300
0.4 % ---+

I. if.' j
0 0
a. a.
:::;; :::;;
I

~ .. /
(/) 200 (/)
I 200
::J 000 ::J
...J " ,; ...J
::J , \6 ::J
0
0 d" X.
0
0

~/;
::<
ZIOO ::>'00
z
<t <t
a: a:
t _-X X- I
(/)
X-X"X- '\ 0.8 "lo
FIG.3 (/) FIGA
0
0 40 80 120 40 80 120
FLEXURAL STRENGTH -kPa FLEXURAL STRENGTH -kPo

Fig s . I - 4 : Graphs of the strain modulus (El - fle x ural s trength (Of) c haracteristic s
of wet-seeded (1) sodium chloride; (2) methyl alcohol; (3) sodillm acetate;
and (4) sodium propionate doped ice

493
such, this type of ice was very porous and did not scale properly in density. Since

this type of crystal structure was also produced by other members of this R.OH alcohol
group (ethyl alcohol = C2 HS ; propyl alcohol R = C Hy)' it would seem that these
3

chemicals would not be suitable for use as dopants in refrigerated towing basins

3.3 Salts of Fatty Acids (Sodium Acetate, Sodium Propionate)


From the salts of the fatty acids (R.COONa), two members of the series were

tested - sodium acetate (CH .COONd) and sodium propionate (C H .C OONa). For these

3 2 S
dopants, tests were done for seeded ice for concentrat ions up to 0.8% for the acetate
and 1% for the propionate.
Fig. 3 shows the E - Of results for ice gro"Jn from wet-seeded solutions doped
with sodium acetate. In this figure there are several features to note. For each of
the four concentrations tested, the effect of the warm-up on the ice is eviden t by
follo"Jing each of the curves from the top right-hand corner to the bottom left-hand
corner of the figure. For ice grown from the low concentration solution (0.2%), the
strain modulus was very large but decreased markedly with increasing warm-up time. In
this case, howe ver , the I imi ting fle xu ral strength of the ice was "100 kPa which is
far too strong for use in model testing. With increasing impurity content in the
melt, this E - Of curve shifted tm,ards lower l im iting flexural strengths, with,
however, an apparent c lockwise r otat ion, such that for solution conce ntrati ons of
0.8%, the ic e was quite weak (down to 10 kPa) but had a low E/o ratio of 400 at this
f
strength. This behaviour of a lateral shi ft to lowe r strengths accompanied by a
clock w ise rotation of the E - of curve with i n creasing soluti on concentration appears
to be quite gener a l and was observed for sever al of the doped icestested. A general
behaviour of this type is reasonable if one considers the effect of increasi ng
impurit y vo lume in the ice w ith increasing initial solut ion concentration. Clearly,

the amount of dopant in solution is extremely important." For this ice, the E/o
f
ratio is great er than 2000 for strengths to le ss than 2S kPa for ice grown from solu
tions containing 0.6% sodium ~cetate. This lower strength appears to be the limiting
strength for ice gro"n from this concentration of sodium acetate in solution. As
such, from the Elo f ratio point of view, thi s ice is a better model ice than sal ine
doped ice for refrigerated model I ing basins. Structurally, however, the ice was not
as homogeneous as sal ine ice and it consisted of a two-layer system . From this point
of v i ew, it y}Quld appear that this type of ice would not be a suitable model ice for
a model I ing tOCJing tank.

*Because of this pronoun ce d influe nc e of the solution concentration on the physical


propert ies of the ice sheet, the author recommends that each ice tank do a series of
t~sts such as those described her e to find the optimum solution concentration f or the
r ()rtic ulur dOpA n t used irl the tank. This is necessar y because of differences in the
pre-cool ing procedures, seeding techniques, growth rates and warm-up times among the
various tanks.
494
Fig . 4 shows the E - of results For ice grown From wet-seeded solutions doped
with sodium propionate . In this case for ice grown from solution concentrations of
0.4 - 0 . 5%, the E/oF ratio is above ZOOO For strengths to less than 30 kPa . For ice
grolJ'Jn from sodium propionate so lutions doped with more than this amount, however, the
E/oF ratio drops to less than ZOOO . Structurally this seeded ice was quite acceptable
For a model ice and as such ice grown From solutions containing 0.4% - 0.5 % of this
dopant would appear to be a better model ice than sal ine ice For reFrigerated towing
basin s .

3.4 Amide s (Formamide, Carbamide)


From the admides (R.CO.HN ) group, two members of this series were tested
Z
(Formnmide H.CO . NH ; carbamide (urea) NH .CO . NH ). For these dopants, tests were done
2 z 2
on wet-seeded ice For concentrations up to 0.5% For the Formamide, and 1.3 % For the
carbamide.
Fig. 5 s how s the E - OF results For ice grown From wet-seeded solutions doped
with Formamide. In this case, For ice grown From 0.5% solution concentration of
Formamide, the E/oF ratio is above ZOOO For Flexural strengths of 50 kPa, but drops
markedly For lower Flexural strengths. Structurally, this ice was columnar grained
with an average grain diameter of ~0.3 cm. It would seem that this type of ice is a
comparatively better model ice than sal ine ice For reFrigerated towing basins.
Fig. 6 shows the E - OF results For ice grown From wet-seeded solutions doped
wi th carbamide . In this case, For ice grown From solution concentrations of 0.6% and
0.9%, the I imiting Flexural strength of the ice was ~50 kPa and 30 kPa respectively.
However, For ice grown From 1.3 % carbamide solutions, the E/oF ratio remained above
ZOOO For strengths down to less than ZO kPa. It would seem that, both mechanicallyand
structurally, this ice would be an excellent model ice For use in reFrigerated towing
basins.

3.5 Miscellaneous Chemicals (Sodium SulFate, Lithium Chloride)


In addition to the chemicals From the three groups mentioned above, two other
chemicals were tested, based on their perFormance in the small tray tests [6]. Sodium
sulFate (Na S0 ) was chosen For testing since, considering its relatively high average
2 4
molecular weight and its high eutectic temperature, it was very eFFective in reducing
the strength of ice . Lithium chloride (Liel) was also chosen For Further study since
it was an extremely eFFective chemical in the strength reduction of ice.
Fig . 7 shows the strain modulus - Flexural strength graph For wet-seeded ice
grown From aqueous solutions doped with 0.8% and 1% sodium sulfate. In both cases,
the strain modulus decreased markedly as the ice sheet warms up with, however, only a
s light change in the corresponding Flexural strength . As such, the E - OF curves are
very steep ,"ith I imiting Flexural strengths on the order of 30-40 kPa. These lower

495
I I 1 1

FORMAMIDE CAR BAMIDE (UREA)


301) 300
0 .34 "10-i ./
0 ... X 0
Q. Q.
::;;
/ "" ::;; 1.3"10
~
~/
1 00 I 0 0
'" 2 00
::J
--' 1 ~
\6
'!-o
'" 200
::J
--' ? 1/ X
I
i /"
.
~\
::J ::J
0 I 0
0 I 0
~ ~
ZIOO zlOO
4 <l
C! ;'/......- O. 5 0/0 C!
>- >
FIG.S FIG.6
'" /'
/
'"
0 0
0 40 80 120 0 40 80 120
FLEXURAL STRENGTH - kPo FLEXURAL STRENGTH - kPo

SODIUM SULFATE LITH IU M CHLOR IDE

300 300

o o
Q. Q.
::;;
I
::;
1
~
. _ _ -X.,!-O0 0
",200 '" 200 t"t .
::J
--'
::J
--'
liO.25 "1 ~\6 0

"
::J ::J
o o
o o
::;; ::;; 0.35 "10 ...fJ
Z I 00 zlOO '\ 'r ' /'
~.

Cl: Cl:
> >
FIG .S
'" O~--L- __L -__L -__L----'~---'
'" O~ __ ~---'L----'L----' __---'__ ~

o 40 80 120 o 40 80 120
FLEXURAL STRENGTH - kpO FLEXURAL STRENGTH - kPo
Fig' 5 - 8: Graphs of the strain modulus (E) - flexural strength (Of) characteristics

of wet-seeded (5) formamide; (6) carbamide; 0) sodium sulfate and


(8) I ithium chloride doped ice

496
strength values were achieved, however, only after an extremely long warm-up period.
Therefore, although the E - Of curves look promising, it would seem that ice grown
from this dopant would not be suitable for use in a refrigerated model ling basin.
Fig. 8 shows the strain modulus - flexural strength graph for wet-seeded ice
grown from aqueous solutions doped with I ithium chloride for concentrations up to
0.35 %. For this ice, the strain modulus is very large for flexural strengths down
to ~40 kPa, whereupon the strain modulus decreases rapidly. In order to achieve the
lower strength values 40 kPa), a long warm-up time was required. As might be
expected, the structure of this ice was very similar to that doped with sodium
chloride. From an applications point of view, this ice could also be considered as
an excellent model ice for use in a refrigerated model basin. This would be so
expecially if the tests were performed in the strength range above 50 kPa where the
E/o ratio is on the order of 4000.
f

4.0 SUMMARY AND CONCLUSIONS


In this study the mechanical and structural properties of ice grmm from solu
tions containing relatively large amounts of various chemical impurities have been
investigated in an attempt to find a doped ice which would have the desired proper
ties for model ice in refrigerated to\,ing basins. Although there are several
candidate chemicals which appear to be better than sodium chloride, there are two
which are far superior to NaCI for producing model ice.
Ice doped with carbamide (urea) appears to offer a very good E/o ratio of ~2300
f
for strengths from 60 kPa down to 20 kPa with a reasonable warm-up time and crystal
structure. In additi on, it is relatively economic (50 kgm = $30 Can.), presents no
toxic hazard [13J, and since it is commonly used as a fertilizer, it can be readily
disposed of if required . A recent study by Minsk [14J on the corrosiveness of ch emi
cals on concrete has shown that carbamide (urea) attacks concrete only slightly
(rated I on a scale of 0 to 5 with 0 representing no corrosive action and 5 represen
ting severe corrosive action), and therefore should be usable in a concrete environ
ment with I ittle difficulty . Determining the concentration o f carbamide in solution
is straightforward since either a high accurac y hydrometer or refractometer can be
used (see Appendix). Finally, carbamide reportedly hydrolyses in water forming a
weak basic solution. This suggests that there might be a "fatigue" effect with this
dopant, although tests by the author on a series of ice sheets doped with carbamide
did not indicate any such effect [?l.
Ice doped with lithium chloride also offers an excellent Elof ratio, especially
for strengths in the range 40-60 kPa. This dopant, however, is more than ten times
more expensive than carbamide (50 kgm LiCI = $440 Can.), although of course, less
I ithium chloride is required in solution than carbamide. Lithium chloride is

497
generally non-toxic (it is sometimes used to replace NaCI in salt-free diets [13])
although it does have some hazard associated with it since it is a I ithium compound.
Determining its concentration in solution is straightforward since a conductivity cell
can be used. In terms of its corrosiveness, it should fall somewhere between sodium
chloride and carbamide.
In summation, the choice of which of these two chemicals produces the best model
ice is not clear-cut. The carbamide ice appears to offer the best overall properties
since its E/o f ratio is good, it's non-toxic, non-corrosive, economical and readily
disposed of if required. On the other hand, if the additional expense and longer
warm-up time of the lithium chloride ice can be tolerated, it offers the best strain
modulus - flexural strength properties of any of the types of ice tested. The choice
of which chemical to use will have to be made by the individual ice tank operators
depending on their requirements . In any event, it is clear that model ice doped >lith
either the carbamide or lithium chloride dopant is far superior to sodium chloride
doped ice for use in refrigerated model I ing basins.

5.0 ACKNOWLEDGEMENTS
The author would 1 ike to thank Mr. R.A. Martin for technical assistance in
this program .

6.0 REFERENCES
I. Schwarz, J . , "On the Flexural Strength and Elasticity of Sal ine Ice", Proc. 3rd
Int. Symp. on Ice Problems, IAHR, pp. 373-386, Hanover, U.S.A., 1975.
2. Schwarz, J., "New Developments in Modell ing Ice Problems", Proc. 4th Int. POAC
Conf., POAC '77, pp . 45-61, St . John's Canada, 1977.
3. Timco, G.W., "Morphological Characteristics of Ice Grown from an Impure Melt",
NRC Div . of Mech. Eng. Report LTR-LT-91, Ottawa, Canada, June 1978 .
4. Timco, G.W., "The Mechanical and Morphological Properties of Doped Ice: A Search
for a Better Structurally Simulated Ice for Model Test Basins", Proc. 5th Int .
POAC Conf., POAC '79, pp. 719-739, Trondheim, Norway, 1979.
5. Timco, G.W., "A Chemical Survey to Determine Potential Dopants for a Model Ice
Test Basin", NRC Div. of Mech. Eng . Report LTR-LT-95, Ottawa, Canada, January 1979.
6. Timco, G.W., "Flexural Strength of Ice Grown from Chemically Impure Mel ts", Cold
Regions Science and Technology i, pp . 81-92, 1981 .
7. Timco, G.W., "The Mechanical Properties of Saline-Doped and Carbamide (Urea)-Doped
Model Ice", Cold Regions Science and Technology 1, pp. 45-56, 1980.
8. Timco, G. W., "The Mechanical Properties of LiCI-Doped Model Ice", NRC Div. of
Mech. Eng. Report LTR-HY-79, Ottawa, Canada, May 1980.
9. Wyman, M., "Deflections of an Infinite Plate", Canadian Journal of Research ~~,
pp. 293-302, 1950.

498
10. Frederking, R., "Mechanical Properties of Ice and their Applications to Arctic
Ice Platforms", Proc. Ice Tech. 75 Symp., Paper K-i, Montreal, Canada, April 1975.
II. Edwards, R.Y. and Kotras, T.V., "Results of Model Ice Properties Research",
Arctec Inc. Final Report 0057, Columbia, U.S.A., July 1972 .
12. Timco, G. W. and Martin, R.A., "Physical Properties of Chemically Impure Ice
Sheets", NRC Div. of Mech. Eng . Report LTR-LT-113, Ottawa, Canada, February 1980.
13. Sax, N.I . , "Dangerous Properties of Industrial Material s ", Reinhold Publ. Co.,
New York, 1965.
14. Minsk, L.D . , "Freezing and Thawing Te s ts of Liquid De-icing Chemicals on
Selected Pavement Materials", Cold Regions Science and Technology ~, pp. 51-58,
1979.

Appendix: Techniques for Determining the Urea Concentration in Solution


To determine the concentration of carbamide (urea) in solution, either a high
accuracy hydrometer or refractometer can be used. In experiments conducted by the
author, a set of solutions were made containing various amounts of urea in concentra
tions from 0-1.5%. The s e solutions were tested to measure both specific gravity at
20C (using a high accuracy hydrometer - range 0.996-1.011), and refractive i nd'x at
22C (using a precision refractometer - range 1.3320-1.3412). The results are shown
below. From the se graphs, it is clear that urea concentrations can be determined
easi Iy to an accuracy of O.I % using either method.

1.003 1.3350
x

~ 1.002
o
"'a.r 1.3 34 5
X
w
0
>
f-
>
<l
a::
1.00 I
,L

X --
R EFRACTIVE
INDEX 1.3340 ~
z
w

tD f-

U
U 1.000 <l
~
a::
~

", /x
U w
W a::
a. 1.3335
(/)

.998 I I 1.3330
o 0.4 0.8 1.2 1.4
CARBAMIDE (UREA) CONCENTRATION (%1

499
A COMPARISON OF SEVERAL CHEMICALLY-DDPED

TYPES OF MODEL ICE

G.W. Timco Hydraul ic s Laboratory Otta \v'a, Ontario


Research Officer Nat ional Research Counci I Canada

Written Discussion from: G. Franken s tein, U. S.A. CRREL.


The author should receive an award for his efforts in developing a model ice
that retains an acceptable E/Of ratio for all desired strengths.
CRREL has experimented with urea doped ice for approximately one year. We
find that if an ice sheet formed ylith 1.3% urea was properly seeded (at ~-O.2C)

and warmed it indeed woul d produce the re sul ts as the author s hows in his Figure 6.

However, if not seeded correctly the ice produced would be extremely strong. A
tremendous advantage in using urea ice is that almost an y desired strength can
be ob t a i ned.
The author should emphasize his recommendation which appears on the bot tom of
page 6. He also should have added that each laborator y report their results so that
everyone can gain from their experience. We found that one must have patience when
first attempting to use urea ice. The results are worth the effort that it yJi11
take to develop your laborat ory freezing technique.
It is hoped that the author has inspired other researchers to continue his
efforts to hopefully develop a model ice that would sati sfy al I of the required
scal ing laws.

Written Discussion from: M. M~Mttanen, University of Oulu, Finland


The author has continued his thorough and sy s tematic research on the character
istics of different dopants for model ice. Preliminary re su lts (pOAC-79) have been
verified in the present paper and a more c omprehensive study of the mechanical
properties of carbamide-doped ice is given in his reference 7. The success of this
research can be seen in the enthusiasm and extent in which carba mide-doped ice has
been adopted by ice model basin operators.

500
Most of the questions that were presented two years ago have been cleared out,
even the difficulty in determining the carbamide concentration is solved. There is
a minor pitfall in the flexural strength test procedure. The cantilever test beam
has the length to width ratio about 2 which is significantly less than what is
recommended by IAHR for index strength tests. A wide beam gives lower strength
values than the standard index strength beam. As the same beam shape was used for
all different dopants the results give correct answer for dopant comparison pur
poses but overestimate the E/o ratio.

Quest ion from: G. Frankenstein, U. S.A. CRREL


How do you measure "EII.

Have you compared measurements of "E" using the beam or plate methods.
Answer:

The strain modulus values were measured using the plate method.
In this series of experiments a comparison was not made between "Ell values

measured using the beam or plate methods.

Question from: R.T. Weiss, Exxon Production Research, U.S.A.


Please describe the experimental work leading to the conclusion that NaCI is
not a suitable dopant. Which NaCI concentrations were studied. The properties
of LiCI ice were found to be very sensitive to concentration. Were sufficient
tests conducted \-lith NaCI to rule out the possibility of a similar sensitivity
to concentration.
Answer:
Sodium chloride was studied extensively in the small tray tests which indicated
its relative effectiveness as a strength reducer for ice. In addition, a series of
tests were performed measuring the strength and modulus characteristics of 0.6 %
sodium chloride in the large tank. This concentration was studied since it has
been shown by Schwarz (On the flexural strength and elasticity of saline ice,

Proc. 3rd, Int. Symp. on Ice Problems, IAHR, p 373-386, Hanover, U.S.A. 1975) that

this is the optimum concentration for sodium chloride as a dopant. The present

tests comfirm the results of Schwarz. With regard to the concentration sensitivity

of NaCI as a dopant, this was studied and discussed in detail in Sch\-Iarz's paper,

and I refer you to it for more detail.

Question from: J.C. Tatinclaux, U.S.A . CRREL.

For the determination of the bending strength of the doped ice, the author
performed cantilever beam tests with beams of aspect ratio L/h=4. Isn't that aspect
ratio somewhat low with the risk o f not satisfying the conditions of application
of equa t ion 3.

501
Answer:

In performing these test s , care was taken to en sure that the L/h ratio was in
the range of 4-6 in all ca ses. This was done since it was estabI i shed in a separate
series of experiments that, for model ice, the apparent flexural strength is

independent of the L/h ratio in this range (for more detail see "On the test methods
for Model Ice" by G.W. Timco in Cold Regions Science and Technology (in press)) .

Comment from: J. Molgaard, Memorial University of Newfoundland, Canada


I would like to add to the remarks made by a preceeding discussor .
The coefficient of friction of ice does usually depend on sl iding veloci ty,
it could also in some circumstances vary with load. The statement that ~ is not
a problem in scal ing is probably premature. The dependence of ~ on other factors
may wei I be of some concern.

502
CARBru~IOE ICE GROWTH IN A LARGE TEST BASIN

O.A. Sandeil U.S. Coast Guard Academy USA

Abstract

Carbamide NHzCONH z (Urea) has been proposed as a dopant for model ice because
of its desirable physical properties and non-corrosive and non-toxic che mistry. A
se ries of ice sheets using two different urea conce ntrati ons was run at the CRREL
Te st Basin to develop growth Curves over time at constant temp erature . The heat
transfer coefficient for the test basin was estimated using nonlinear regression
techniques and was used, alo ng with temperature, as the key parameter in the heat
transfer analys is differential equation whose solution was the ice growth prediction
model. Variation in crystal structure was observed to profoundly affect ice
st rength and modulus because of the amount of urea entrapped. The conditions whi ch
bring about these variations were explored and partially identi fied . A fractional
factorial experimental design was employed.
Key Words:
Carbamide
Urea
Modulus
Heat transfer coefficient
Heat transfer analysis
Crystal structure
Regression

503
I

Introduction
Carbamide (urea NB2CONB2) has been proposed by Timco [8J as a dopant for use in
test basin !.later to provide desirable scaling properties for ice model testing. It
avoids the extremely corrosive effects of salt (NaCl) while retaining desirable
scaling properties and the capacity to vary its strength over a considerable range
by controliing temperature. Experience in growing urea ice in the large test basin
at the U.S. Army Corps of Engineers Cold Regions Research and Engineering Laboratory
1n Hanover, New Hampshire is reported herein. There was a need to develop ice
thickness growth curves for the new test basin at CRREL for use in scheduling ex
periments. The effects of urea concentration, seeding, and temperature were also of
interest. A haif replicate of a 2 3 factorial experIment was employed to assess
these effects on four ice sheets grown to a thickness of 9 cm under carefully con
trolled conditions. A nonlinear regression procedure was used to determine the heat
transfer coefficient characteristIc of the test basin. This parameter was then used
in the mathematical model of the ice growth curves. An interesting anomaiy occurred
with one of the ice sheets which resulted in investigation of the crystailography of
urea ice.
Experimental DeSign and Procedures
The half replicate of the 2 3 factorial design Ls s hown in Figure 1. An ad

vantage of the experimental environment of the test basin was the abillty to main
tain constant temperature (-20C and -10C). This greatly simplified the solution
to the differential equation discussed In the next section. There was some varia
tion in temperature due to other demands on the refrigeration plant (especially
Sheet NO.4) but generally this experiment can be considered to have been con
trolled. The procedure used was to cool the entire water mass to under LOC by cir
culation. The basin measures 9 m x 36 m x 2.4 m deep and is cooled by refrIgeration
coils in the ceiling 5.2 m above the water surface. There are also coils near the
tank bottom at one end between which the water can be circulated. These are cooled
by anhydrous ammonIa. Vertical circulation is achieved by bubbling aIr frOID the
tank bottom. When conditions were ready for ice growth to begin, the bubbling was
stopped and (if desired) the surface was seeded using steam. Measurements were
taken at regular intervals using a micrometer caliper until a thickness of at least
9 Cm was reached. The temperature was then raised to +5C untii strength and modu
lus properties reached desired model values. This latter process lasted from 24 to
36 hours. The old ice sheet was used to cool the water for the next sheet by again
bubbling air fro," the tank bottom. When the water temperature throughout the tank

504
EXPERIMENTAL DESIGN
Y. Replicate of 2' Factorial

Temperature IT)

Urea
%(U) .6 1.2 .6 1.2

iL
c
o
.:;
f
!
..
I I)

a..

Defining Contrast TUP

Figure 1. Experimental De si gn, 1/2 Replicate of 23 Factorial


reached the desired l evel, the old ice was removed to a melt tank and t he new sh eet
was i ni ti a ted either by seeding (if desi red) or allowing ice to f orm na t ura lly .
It became appa r en t that the thermal regime was very im portant to the sub s eque nt ic e
c ry sta l st ruct ure . Unless water temperature was unifo rmly be l ow +.4 C, the
dendriti c growth necessary to entrap brine pockets did not occur. This phenomena
was nOt as dependent upon seed ing as it was on thermal reg ime .

Regression Model of lee Grow th


Figure 2 depicts th e nature of th e data a nd a linear least squ ares fit for each
of the four ice shee t s together. There is also shown an addi ti onal sheet grown for
another exper iment but used to assess re pea t abil it y of the data. Several locations
ove r the sheet were moni tor ed during the thickness tests and this var iabllit y
appears as scatter about the re gression lines shown in Figure 2 . The mo re
appropriate model for this r egress i on , however, was the heat transf e r a nalysis
Equation (1) for lee thickness growth (without snow cover) which is well known
(Neuman [5], As sur and Weeks [2], and Ashton [IJ to be:

505
1. 6 X 10

o
Seeded - 20" C 0 .6 % Ureo
Unseeded - 10 C 0.6%
1. 4 6 Seeded - 10 C 1. 2'Y.
.. Unseeded-IOC to-15-C 1.2 %
S ded -20C 1.2%
1.2 0

B
E 1.0

0 .6

0.4

0 .2

Linear Reqre..ion r-otbH

o 0 .4 0 .8 1.2 1.6 x 10
H, Timo (hrsl

Figure 2. A least Squares Fit of Thickness vs. Time for the


Five I ce Shee ts

T - T
m a (I)
~ + I
k.
L
Ii

where

d~
de = the r a t e of cha nge of ic e thickness ~ with r es pect to time

Pi the density of ice


the heat of fus ion of i ce
T!1] the melting point of ice
Ta the ai r t empe ra ture
k'L the therma 1 conductivity of ice and J

h the thermal (heat transfer) coefficient o f the ice/ai.r interface.


For the purpo se of this study it was felt that coefficients of fresh wa ter ice
~er e close enough to tho se of urea at 0 .6% and 1. 2% to be used for the prediction

506
equations desired. Thus, Pi = 916 kg m- 3 , A = 3.34 x 105 Jkg- 1 , Tm = O"C,
ki 2.24 \{M-l C- 1 were the physical constants used. In the natural environ
ment, Ta, varies over the daily cycle and must be treated as time average tempera
ture and numerically integrated using finite difference equations to determine ice
thickness over discrete time increments. Fortunately, in th e test basin experiments
this temperature could be treated as a constant. It is interesti.ng to note that

according to this model, the only parameters which c~n be controlled to increase the
rate of grow th are the air temperature 1a and the heat transfer coefficient h.
The latter quantit y in nature depends upon wind velocity and sunllght. In the test
basin it depends on the interface between the refrigeration coils located on the
ceiling and the ice surface 5.2 m below. This is primarily influenced by air clrcu
lat10n in the room.
Since air temperature Ta can be considered a constant, Equatlon (l) can be
integrated with respect to time to obtain:

2k.
l
hT)2 + 2k .T) (T T ) t = 0 (2)
l PiA m a

with Tm 0 this becomes

2k h T t
2 a
hn + 2k n - 0 (3)
i PiA

Since T) ~ 0, the positive root can be obtained by means of the quadratic for
mula as

L/2
A-I) J Ih (4)

This result was parameterized for various values of h with Ta = -20"C to ob


tain the family of curves shown in Figure 3. The empirical data from Tes t Sheet
No. 1 was also plotted and falls between h ; 5 and h = LO.
A more precise estimate of the heat transfer coeffic ient was obtained uslng
Equation (4) and regressing the ordered pairs of data (T)i,t!) from test sheet
No.1. A Marquardt [14) nonlinear leas t squares procedure y ielded an estimate of h
= 6.06. In this manner an estimate of heat tra nsfer coefficient for sheets No.2
and 3 was also obtained using Ta = -LOce. These results are shown in Table 1.
Using a two-independent variable (time t and temperature Ta) non linear regressIon,
a pooled estimate of heat transfer coefficient was obtained as show n in the table.

507
30

E
<.>
II)
II)
Q)
c:
-'"
<.>
.c. 20
I
Q)
<.>
H

4
Time (days)
Figure 3. Thickness vs. Time for the Different Thermal Heat
Transfer Coefficients

Table I. Estimated Heat Transfer Coefficie nt s

Data from Sheet No. Heat Transfer Coe f. h


Watts/(ml)("C)
I 6.06
2 6.36
3 5.47
pooled 5.93
It should be noted that the procedure described above will cause th e aggregate
effect of all so ur ces of va r labilit y to re side in th e estimated value of heat
tran s fer coefficient h. Nevertheless, it would appear that a value of h = b des

5 08
crlbes the approximate effective heat transfer characteristics of the test basin at
CRREL Ice Engineering Facility under its May 1980 configuration. It should be noted
that some retrofit modifications have been made to the refrigeration plant since
tha t time and additional tests of the kind described above are needed to det e rmine
the validity of this estimate of heat transfer coefficient.
The primary result of this experiment is the equation:

n + (5 )

where
n ice thi ckness in meters

ki thermal conductivit y of ice = 2.24 W/mC

h hea t transfer coefficient = 5.93 W/m 2 C

Ta air temperature in C (assuming that temperature of test basin is

maintained constant over growth of i ce s he e t)

p density of ice = 9.16 kg/m 3

A heat of fusion of ice = 3.4 x 105 J/kg

time in seconds.
Note that the values of ki' P and A are those of fresh water ice. If
corresponding coefficients for urea ice of appropriate concentration are available,
these should be used.

Crystallography
Over the course of this experiment it was observed that Sheet No.4 (-20C,
unseeded, 1.2% urea) was unusually strong (700 - 800 kPa) as compared with 150-250
kPa for the others during the growing period. Upon subsequent ~arming to reduce
strength, this sheet remained at approximately 400 kPa (se e Fig. 4) despite apparent'
deterioration to the point of wetness over the entire sheet. This was in sharp
contrast to the more normal trend shown for sheet J (-l OC, seeded, 1. 2% urea) in
Figure 5. Elastic modulus mea s urements were similar ly disparite. At first glance
it would appear that the seeding vs. non-seeding condition would explain this
anomal y . However, the unseeded Sheet No.2 did not display this unus ual
characteristic. Upon further experimentatlon it became apparent that the cause of
this dramatic effect ~as the initial temperature re g ime of 0.7C as compared with
less than 0.4C for all other sheets grown.
A thin section analysis of samples t ake n from Sheets 3 and 4 using the methods
of Langway [ J] (cros sed polaroid photography revealed a striking difference.
Sheet No.3 (see Fig. 6) had the desirab le columnar growth associated with
horizontal c-axis crystal orientation which enables entrapment of urea pockets and
5~o
800~---.---.r---,----.----~---r----'----r---'----'-~

Sheet NO. 4, -toC to -tSC


Seeded 1.2 % Urea
60
6 T emperature (ge)

40
;>
~ ..
'"~ 400 "
.. 0

.
in
u
H
20
..
~

E
f

200
_~-----b--------'1 0

24 32 40
o 16
T i me (hou rs )

Figure L~. Variati o n of Ice Stre.n g th v s . T emp e r at llre Throughout


the Hi s tory of i ce she e t No . 4

200 Ice Sheet No_ .3, -10C Seeded 40


9.06 em Thickness

32
160

6 Temperature ( QC) 24

:: 140
~ 16 ~

~
E
H 80
:- -L
o _ _ o. -- O ------------ ~ -- -.----~--=-.:_ --- 0
~

ll~ ---

40 /
-8

a
Time ( hours)

Figure 5. variation of Ice Strength vs. Tem perature Throughout


the History of Ice Shee t No.3

510
Figure 6 . Thin Section of Ice Sheet No.3 Showing Desirable Lower Columnar
Strue ture

511
the corresponding capacity to be weakened upon warming. Sheet No.4, on the other
hand, (see Fig. 7) had very large crys tals or iented with c-axis vertical. It would
appear that the incubation thickn ess required before dendritic growth can occur was
never reached in this case as sugges ted by Timco [61. Urea concentrations measured
from melting upper and lower strata of the ice sheets revealed that markedly
different amOunts of urea were entrapped in thes e sheets and explain the wide
variation in strength.

Figu~e 7. Thin Section of Ice Sheet No.4 With Noncolumnar St~uctu~e

~12
Conclusions and Recorc.menda t ions
The statistical procedures used in this experiment provi,k',j Lit:> :nl:';l."lS of
developing an ice thickness gro~th prediction model (Equation 5) by estimating the
heat transfer coefficient characteristic to the large test basin at CRREL. Because
of the nature of the heat transfer analysis equation, nonlinear regression
techniques were required. Experience gained using carbamide as dopant in this large
scale modelling environment indicated the need for care in preparation of the
thermal regime prior to the initiation of the ice cover. It is recommended that
mass water temperature be brought down to -+D.IC before ice is seeded or crystals
are permitted to begin formation from, of course, much colder air temperature above
the water surface. Failure to achieve this homogeneous thermal reg ime may result in
ice crystallography (i.e. c-axis vertical) which will be unacceptable for model
testing.
An analysis of variance was performed on the three factors of this experiment.
The effect of urea concentration and seeding on ice growth rate were not significant
(cr = 5%) in comparison to the unexplained variability of the experiment.
Temperature, of course, was highly significant. Considering the large scale of this
experiment (i.e. the test basin) and the small sa mple size (i.e. four ice sheets)
these analysis of variance conclusions are not based on powerful tests. However,
they may be regarded as ample for the purposes intended for ice g rowth prediction
model with its pooled estimated of heat transfer coefficient.

Acknowledgements
the author wishes to acknowledge the assistance of the ice Engineering Branch
at CRREL for their support and cooperation in the conduct of this experiment. The
insight and interpretation of ice crystallography from Dr. Tony Go~ of CRREL and the
suggestions of Dr. Gary Timco of NRC Canada regarding carbamide are especially
appreciated.

513
References

l. Ashton, C.D., 1979. riller ice, American Scientist, Vol. 67, No. l, p. 38-45,
January to February.

2. Assur, A. and W.F. Weeks, 1963. Crowth, Structure and Strength of Sea ice,

Int. Assoc. of Sci. Hydrology, Publ. No. 6l, p. 95-108, 1963.

3. Langway, C.C., Jr., 1958. Ice Fabrics and the Unillersal Stage, Tech Report 62,
August, U.S. Army Snow Ice and Permafrost Research Establishment, Wilmette,
Illinois.

4. Marquardt, D.W., 1963. An algorithm for least squares estimation of nonlinear


parameters, J. Soc. Ind. Appl. Math, 2, 431-441.

5. Neumann, K., 1949. Die Wachstumsgeschwindigkeit Von Eisdecken in Abhangigkirt


Von Lufttemperatur and Windgeschwindigkeit, Ann. Meteorol 2, p. 144-155.

6. Timco, C.W., 1979. The Mechanical and Morphological Properties of Doped Ice,

hoc. POAC 79, Vol. I, p. 719-739.

7. Timco, C.W., 1980. The Mechanical Properties of Saline-Doped and Carbamide

(UREA) Doped Model Ice, Cold Regions Science and Technology 3, No. I, p.

45-56.

8. Timco, C.W., 1978. Morphological Characteristics of Ice Crown from an Impure

Melt N.R.C. 0111. of Mech. Eng. Report LTR-LT-91, Ottawa, Can., June.

514
CARBAMIDE ICi:: CRO>lTH IN A LARCE TEST BASIN

by D.A. SANDELL

Discussion by: C.W. Timco, Hydraulics Laboratory


National Research Council of Canada

The concept of determining the bulk heat transfer coefficient (h) for model ice
tanks as suggested by Sandeil in this paper wi ll provide mode l tank operators with a
useful parameter for c haracterizing the rate of ice growth and the efficiency of the
freezing system. The author is to be commended for this contribution. Sandell's
paper suggests that instead of simp l y specifying the air temperature to describe the
growth conditio ns Eor model ice in a refrigerated baSin, one should, in addition,
specify the heat transfer coefficient for the ba s in in question. This would then
provlde a standard method in comparing the model Lee growth characteristics among

the various ice tanks. (Ciearly, a basin with a low "h" value will require more
dopant in solution than a basin with high "h" value in order to achieve the same
type of model ice.) Sandeli's re commendation of pre-cooling the solution to +O.ioC
before seeding is too conservativ e. Since th e dopant in the solution depresses the
freezing point and ice growth in the ~up~r cooled regime is necessary for dendritic
structure, it would seem that the solu tion should be pre-cooled to -0.2 to -O.SoC
depending upon the dopant concentration.

Response to discussion by C.W. Timco, NRC Canada


Experiments carried out at CRRE L subsequent to those reported herein confirm
TimeD's point. A thermal regime super coo ied to -0.2 to -O.SoC, as he suggests, wili
not only enhance the dendritic struc ture ; it will also increase the freezing rate
and the apparent heat transfer coefficien t. Those con templating the use of this
s tatisti cal procedure for estimating heat transfer coefficien t wou ld do well to
follow his advice. The values of h reported herein may underestimate the actual
values for the CRREL test basin. appreCiate the interest he has shown in making
this discussion.

D. A. SANDELL

515
THE SALINITY OF ARTIFICIAL BUILT-UP ICE ~~DE BY
SUCCESSIVE HOODINGS OF SEA I'IATER

~L Nakawo*
Research Associate Divisi o n of Building Research Canada
National Research Council o f Ca nad a
R. Frederking Ottawa, On tario, Canada
Research Officer

ABSTRACT

One method of thickening an ice sheet is "free flooding": ice is built up


by successive flooding and freezing of sea water layers. The salinity of the
built-up ice is of great interest because it plays an important role in
establishing me c hanical properties.
Detailed observations on salinity of flooded water and built-up ice were
carried out during construction of an ice platf orm. Ice salinity was generally
about 207. 0 which is significantly lower than the salinity of the original sea
water (- 30%0)' Almost half of the "lost salts" d i sappeared during the free zing
period of a layer; the remainder were lost during subsequent floodings.
Detailed salinity, thin section and dye migration measurement s were used to
postulate processes of horizontal as well as vertical desalination during
construction of the ice platform.

* Present address: Applied Ph ys ics Section, Fac ulty of Engine eri ng,
Hokk a ido Uni versi ty, Sapporo, Japa n 060

516
Floating ice platforms have provided a very suc cessfu l means for carr ying out
exploratory drillin g in the Canadian arctic islands [1] . Since the fir st ice platform
was constructed and used for drilling in 19 74 the trend has been toward increasing rig
loads and l engthe ned drilling per iods . Consequentl y, efforts have been devoted to
devel oping an improved basis for ice platform design [2] as well as platform
cons tructi on [ 3] .
The technique used in constructi ng an ice platform is "free floodin g", i .e ., ice
built up in la yers by successive flooding and freezin g of sea wat er pumped fro m
beneath the ice cover . Grain structure and salinity of built-up ice plays an
important role in establishing its mechanical properties. Observat i ons have shown
that the salinity of the built-up ice is significantly lower than that of the ori ginal
sea water . An explanation of this desalination process is of considerable interest.
This paper presents field data on spatial and temporal distribution of salinity in
built-up ice and discusses proces ses that control the se distributions.

Description of Observation Site


The field observations were made 14 to 20 Janu ary 1979, at Panarctic's Desbarates
8-73 well s ite which was located 76 42' and lOS 57' W, northeast of Melville Isl and.
Two platforms were constructed at the site: one for the drilling opera ti on and the
other for the drilling of a relief well if necessary. The actual measur ement program
was carried out on the relief platform.
Two pumps located near the centre of the platform lift sea '<ater from beneath the
ice and discharge it on the su rface . Average sea water salinity was 30.8 0.6% .
Distribution o f the wat er was co ntrolled by periodic adjustment of th e discharge
direction of the pump nozzle. The resulting platform was elliptical in shape with
maximum and minimum diameters of approximately 200 m and 100 m respectiv e l y. The r e
was no confinem ent to the "ater flow and "free floodin g" resulted in a platform with
maximum thickne ss at the centre, tapering to the natural ice thickness at the edge,
i.e., the platform surfac e sloped slightly dOh~wards from the centre (i nclination
about 0.4).
At the time of th e fi e ld program the total ice thicknes s of the platform was
about 3 m at the centre and 1.8 m at the edge (natural ice thickness). The duratio n
of a single flood ,<as 0.5 to 1.0 h, "ith a resulting l ayer thickness of 10 t o 20 ml'l.
Flooding frequency was once every 3 to 5 h. Altogether 31 floods "ere made during the
observation period for a total ice build-up of almos t 0.5 m.
Detailed observations we re made alon g a r ad i al line runnin g ou tward about 60 m
from one of the pumpS. This line co rrespon ded to one of the short axes of the plat
form so the obs erva ti on area covered a representative sec ti on extending from th e
centre to the edge. The field obse rva ti ons included measurements of salini.t y of the
sea water and ~ui lt-up ice and also temperature in th e built-up ice. A core was

517
recovered and returned to the laboratory for further salinity measurements and
analysis of grain structure.

Results
Salinity of the surface layer was m e~s ured at a number of positions in the obser
vation area several times during the observation period. The surface layer samples
were recovered shortly before th e next flood. Horizontal distribution of the surface
layer salinity is shown in part (b) of Figure J. Note that up to 30 m, surface layer

20 I I I I I I
18 r- (a) -
'" E
..... E r 0 0 0 -
> 16 ----o-----...
..:
14 r -
~ V)
0 '0
V)
0' '0
..........
'-"2
..:""
12 r ,,
",
"'u -
..... - 10 I Figure I
>:c
":0
I - Variat ions of la yer
I I I I
I "" thickne ss (a) and
50 salinity (b) with
I I I I I I
A, DISCHARGE D WATER AT PUMP (b) increas ing distance
o
B, RU N NIN G WATER ON SUR FA CE from pump
40 - C, RUNNING WAT ER ON SURFACE -
>
..... 1
u _
<I:z
o
~ o -
cr:~
=>..:
V) V)

I I
20
o 10 20 30 40 50 60 70

DISTANCE FROM PUMP, m

sa linit y is about 25%0 but be yo nd 30 m it in c r eased to about 35%0. This figure a l so


show s the sa linities of pump di sc ha r ge water and ~ater running on the surface during
flooding. These data show a n increase in s alinity wit h inc r easing distan ce from the
pump. Part (a) of h gu re I pre se nt s a ve r age Ja yer thickness as determined by mea s ur
ing t ota l ice build-up over the ob serv a tion peri od a nd di viding by the number of
floods in that period (31). Each flood did not comp letel y cover the a rea be yo nd 30 m,
which accounts for the reduction in average l ayer thickness bey ond 30 m.
From a core taken 18.5 m from th e pump, vertical thin sections were made and a
vertical salinity profile established (Figure 2). Each salinity value represents the
average of that particular layer. The surfac e sa linity of the vertical core (2 5% 0)
corresponds we I I wi th the average surface sa I ini t i r,; shown in Figure I. The vertical

518
SALINITY, 0 / 00

15 20 25
o

LA Y E R
0.1
}- 22

~ . rigure 2

}- 18 Ve rti ca l salinity
u
profile obtained
0.2 for a core sampl e
::::l
taken 1979-01-20
" a t a distanc e of
0
t

~
}- 13 l8.S m from pump

0 0.3
""'w-
I
t
"
cw
0

0.4

profile s hows that to a depth of 0.4 m salinity wa s in the range of 20 to 22 %0' Below
0. 4 m, ther e wa s a further Jeerc::,(" i. n salinity. The boundaries of the individual
flood layers can be quite c learly d istinguished as dark lin es in the thin sections.
An enlar ."!.'J:lc nt of the profile at a depth of 0.28 m (reference level) is shown in
I j !'.ure 3. I'h ~ bound ar ic"s as shown in the thin sec tion (lre i.ndicated by the dashed
Jin es . The salinity profil es Iii thin a la yer show a characteristic "d oub le S"
d istribut ion , i.e., salinity is hi gh on the layer houndary, dec reas e s to a minimum at
the qu "rt cr point in the layer, increases to a peak at th e mid - point in th e l ay er ,
decreases to anoth e r minimum at the three -quarter point, and finally to another
maximum on the boundary. Grain stru c ture also shows J characteristic pattern: fine
grainerl granulHr ice on the boundaries, bands of elongated grains at the quarter and
three-quarter points in the layer, and a very irregular band is in th e middle of the
la)'er .
519
SALI N ITY, 0/00

20 25 30 35

E
E

20
cu
U
Z
cu
'"cu
LL
40
CU

""
~
o
""
LL 60
I
>-
(L
cu
o
80

Figure 3 Detailed sa l inity profile fOT four flooded layers in the core sample

Measurements were also made of temp erature and salillity changes with t i me ill the
upper layee; of the built-up ice . P8rt (a) of Fi".1ue ~ indicates timc variations of
temperature at different depths in the built-up ice durinK sevcr~l flooding nnd freez
ing cycles. The tempe rature curves lab e lled L.ayers A, Band C "ere measured "i th in
the mos t recently flooded layer. At the I" " ' ~ illning of a flood the se temperature s were
high as the relatively "arm ( _2C) sea wnter was discharged onto the ice surface.
A rapi d de crcase in temperature fo ll owed thc co mpl et ion of the flooding. At great~r

depths in the i ce, t e mperatu r e fluctuation s "ere a tt e nuated and a pha s e s hif t was
evi de nt. Bel ow a dept h of about 200 mm ice temperature did not respond to individua l
fl oodin gs .
Part (b) of Fi gur e 4 pre s e nt s re sults of periodic salinity mc~s u re me nt s of
part i cu lar la ye r s during severa l f l ooding and freezing cyc les. In the case of eac h
flood the first point "as the s a l inity of the sea "ater. Subsequent samplings sho"ed
a gradual decrease in salinity to about 25 %. at the end of the fr eezinR period.
During successive floodin gs ( L.ayers B and C) there was a further decrease in th e
sa linity of the underlyin g ice ( Layer A) to 20 t o 22 '0 0 (the same sali.nity as found at
intermediate depth in th e vertica l pro f i l e (Figure 2)).
In addition to th e sa li nity, t cmperature and thin section observ atio ns, a dye
ex periment (using Rhodamin e B) was carried ou t to visually trace the directions o f
520
T I ME. h

0 4 6 12
0
\ (a)

,,
\
\
-5
,,
u
,,
.... ,,,
.
a:
- 10
=>
>
,,
r
I
DEPTH
,
.q:
a: ,,\ II
/'
",'
., Fi gure 4

w
CL - 15 , I
. ,,
". '
125 mm

" ",\
::;;: I \ 111 mm
....> I
I
\
\
I
I
, Time va riati on of

'" 65 mm
t e mp e r a tur e (a)

\ \,
-20
\
, " 5 1 mm
" I
I "
and sa linity (b)

\ , - I
-..J
~ ~m
for three fl ood ing
' .... .... 9 mm
(LAYER B) " , (LAYER C)
c yc l cs measur ed
'
- 25 1 979 - 0 1 -1 8 a t a
di ' tance of IR. S m
35
from pump
o
o

>
>
z
-'
.q:
Vl

FLOODING A FLOODING B FLOODING C

15
brine mov ements . Th e resu l ts of the experiment a re illustrat ed in Figur e S. After
the current su rface l ayer , m, had fr ole n, a trench, a, was cut normal t o th e wat er
f l ow directi on. The trench was then fill ed ,.i th dyed ,.at er and a IIQI,ed t o fr ee le . The
next flooding produced la yer y. After l ayer y had frole n , a ver tic a l sect i on was cut
across the trench. Fi gure 5 shows that the dyed area ex tend ed horilontally, in th e
water fl ow direction, a nd downward, indi ca tin g that th e re was brine moveme nts in th ese
directions .
Another dye experiment was carried out in the area where the ver t ica l core was
recovered (Figure 2). In this case dye was mixed "i th the wat e r in la ye r 22 . Two
521
WATER FLO W DIRECTION
<

ICE SURFACE

Vertical
Figure 5

se ction
f
sch emati c and
photogr a ph of
d ye migrat io \l

e xperiment

da ys 1ater, when the core "as recovered, the dyed zone extended dowTII,ard from layer 22
to layer 18. !l'hen another core "as rec overed t"o months later, the dyed zone ext ended
further dowm;ard to layer 13. By this time, however, the colour intensi t y of the dye
had greatly reduced. Also no teworthy "as the absence of an y apparent upward migration
of dye.

DLscus s i on
The dye experiments demonstrated, if on l y in a quali ta tive fashio n, that brine
moves vertically downward and horizontally in the built-up ice. It is possible,
however, to make sOme first order quan tit ative estimate s of these brine movements. The
resul ts presented in Figure show ed an increase in salinity with dis tan ce from the
pump. Combining the salinity data with the ave ra ge layer thickness, a mass balance
was carried out on the salt in a sector of radius 60 m from the pump and arc I radian
(see Tabl e I for resul ts). Area I comprise s the !One from the pump out to 30 m;
area 2 the zone from 30 to the pos iti on wh ere the average surface layer sal init y i s
equa l to th e sea water (43.5 m); and area 3 the zone fr om 43 .5 m to 60 m. Row I of
th e table give s the mas s of salt in each area for an equivalent layer of sea water;
row the actual mass of salt in the surface layer of i ce (calculated from measured

522
Table 1. Salt content distribution in an average la ye r

Ar ea 1 Area 2 Area 3 Tot a l

Row Salt in \~ater, kg 398 374 412 1184


RON 2 Sa lt ill Ice , kg 323 340 457 1120
Ro\.. . Difference, kg -75 -34 +45 -64

sali nit y and layer thickness); and row 3 the differen ces . Area 1 experienced about a
20% loss of salt (desalination) , area 2 about a 10% decrease and area 3 about a
10% gain in salt. Over the three areas th e re was a net loss of 5%. From these
numbers there is evidence of a horizont a l redistribution of salt, decrease in areas
and 2 and increase in area 3, as well as an over-all loss of 5% from the surface la ye r
due to vertical dr a inage. In area 1 during the initial free zins period, de sa lination
" as equall y the result of vertical and horizontal brine movement. To quantify this
bre ak do~'1l more accurately, experimental meas ure ments of permea bility of built -up ice
would be needed. Visual evidence of horizontal brine movement was indicated by the
observation of high salinity brine see p in g from the outer edge of the top flooded
la ye r at the end of the freezing period. Simil a r obs e r ved indi ca tions of horizontal
brine movements were noted during flooding e xperiments at Point Barrow [4]. The driving
force for thi s movement could be intern a l pres s ure generated within a la yer during
freezing. The results in Table were for the initial freezing of the s urfa ce layer .
If an average salinity for are a corres ponding to succe ss ive floods (20 \.) were used,
the proportion of de sal ination due to vert i ca l brine drainage would increase to 10%.

As shown in Figure 2 the salinit y of the built-up ice below the surface layers is
about 20%. which is supported by the observa tions presented in Figure 4. It appears
that after th e initi a l desalination (down to 20%.) furthe r vertical brine movements
are by a displa cement process, i . e., salinity over the dep t h of 0.05 to 0.4 m does not
change. For depth s below 0.4 m the decrease in salinity can be attributed to the
effects of temperature and temperature gradient in promoting br i ne movements [5].

The de tailed sal init y profile and thin section illustrating grain structure
(Figure 3) can be explained as follows. The high salinity and fine grained s tructure
of the i c e at the top and bottom of each layer would result from rapid freezing when
the sea water COmes in contact with the co ld ic e s urface and air. Subsequent ic e
growth would occur from both th e top and bottom of the layer, but at a slower rat e
which would allow more of the brine to be eje c ted fr om thi s part of the layer. Here
one would expect a lower salinity and a larg er grained columnar structure . The centre
of the layer, the l as t part to freeze, would have a high er salinity and a more
irregular gra in structure.

523
Conclusions
An analysis of the resu lt s suggests that there a re three s t ages of desalin a tion .
In the first stage, freezing period of the top l ayer, salinity is reduced from 31%. to
25%. as a result of vertical and horizontal brine movements. The second stage sees a
further reduction in salinity to 20%. duri ng the next two floods. The salinity remains
approximately constant at t hi s value unt il such depth is reached wh ere the ice
temperature is higher than -15 C. During construction the temperatu re reaches this
level at a depth of about 0.5 m [3]. Thi s leads to the third sta ge of desalination
which is caused by warming of the ice. An extension of the third s ta ge of desalination
occurs in the spring wi th seasonal \{arming of the ice cover.

In addition to providing information on the desalination processes, the results


give n in thi s paper shOl< the difficul ties that ,;o uld be experienced in tr yi ng to
s imul a te in the laboratory the ope n syste m process f or building up the ic e that occ urs
in th e field.

AcknOl,ledgements
Th e authors "ou l d lik e to ac kn ow l edge the suppo rt provid ed by Pa na r c ti c Oi l s Ltd.
and FENCO Consultants Ltd., and the assistance of D. Wrig ht , Tec hni ca l Offic er,
Division of Buildin g Research, National Researc h Council of Canada, in carrying out
the field measuremen ts.
This paper is a contribution from the Division of Building Research, ~"tio nal

Research Council of Ca nada, a nd is pub li shed with the approval of the Iii r ec tor of the
Division.

References

[1] Baudais, D.J., Mast e rson, D.M., and Watts, J.S. A syst e m for offshore drilli ng
in the arctic islands. 25t h Annual Technic a l Meeting of the Pe t roleum
Soc i e t y of CIM (Ca nad i an In stit ut e of Mining and Me t a llur gy), Calgary,
Alt a. , ~lay 1974, Paper No. 374029.
[ 2] Ma s t e r son , D.M., And erson, K.G., a nd Strandberg, A.G. Strai n measu r eme nt s in
fl oa tin g i ce platforms and their a pp licati on t o platform des ign. Can. J .
Civ. Eng., Vo l. 6, No . 3, p. 394-405, 19 79 .
[3] Nakawo, M. He at exchang e at surface of bui lt-up ice platform during
constru ction . Cold Regions Science and Technol ogy, Vol. 3, p. 323-333, 1980.
[4] Dykins, J.E., and Fun ai, A.!. Poi nt Barrow Trials - F.Y. 1959; Inv es tigations of
thickened sea ice. U.S. Nava l Civil Engineering Lab ., Por t Huenem e, Cali f.,
Tech. Report RI S5, 1962.
[5] Adams, Jr., C.M., French, D.N., and Kingery, W.O. Solidif icatio n of sea ice.
J. Glaciol., Vol. 3, No. 28, p. 74 5-761, 1960.

524
THE SALINITY OF ARTIFICIAL BUILT-UP ICE MADE BY SUCCESSIVE

FLOODINGS OF SEA WATER

by M. Nakawo and R. Frederking

Discussion by Andrew Assur, USA CRREL


The detailed study of salinity distribution after flooding is
certainly a welcome contribution. The desalination is quite modest,
i.e. the resulting salinity is still quite high. At low temperatures
(approx. -25C) this would not matter, but under warmer conditions a
gradual deterioration could take place. What are the observations?
What are the salinity changes after several months?

Reply to discussion
Some observations carried out in early May, when ice temperatures
were approaching _10C, still did not show signs of deterioration or
further significant desalination. By late June, however, there were
obvious signs of deterioration and ice salinities were of the order
of 5%0.

525
MULTIA XIAL COMPRES SIVE STRENGTH TESTS ON SA LINE ICE
WITH BRUSH-TYPE LOADING PLATENS

Franz Ulrich Hausler Hamburqische Sc hif fba u \oIes t Germany


Versuchsanstalt GmbH.

Abstract

A series of un i- and multiaxial compressi ve s trength tests on saline ice, whi ch


~Ias frozen in the laboratory under simulated natural co nditi ons , was conducted

on an electronically controlled triaxial c l osed -l oo p testing machine with brush


type loading pla tens . The brush-type loading platens were especially designed
for ice and pro vide a very low transverse strain restr ict ion combined ~1ith a
high st iff ness in the loa ding direction. Tests were carried out with a constant
strain rate within the probe parallel to the main loading axis 11hile the o ther
tvlO axes were kept stress controlled to maintain a con stant stress ra ti o between
the three axes. Twenty-two stress ratios were in vesti gated considering the
an isot ropy of the columnar grained sa line ice while the stra in ra te, temperature
and salinity were kept consta nt.

526
Introduction
During the last two decades large amounts of natural resources have been explored in
polar regions and these activities are still increasing. For exploitation of these
already known resources and for further explorations ships and marine structures are
necessary, which can withstand ice loads. For marine structures the ice load condi
tion usually defines the design load, a fact, which is also valid in more moderate
zones, where ice occurs only occasionally. It is possible to build structures and
vessels, which are strong enough for polar ice conditions, but up to now the estimate
of ice loads acting on these has been based mostly upon empirical data from model
tests and from some few full-scale measurements. For theoretical calculations of ice
forces, a large set of data on the mechanical properties of ice is needed.
Failure of ice determines load maxima and ice covers fail usually under mu1tiaxial
stress states. Considering this, the lack of multiaxial strength data on ice and in
particular on saline and sea ice makes it nearly impossible to reliably predict ice
forces by purely theoretical methods. In order to reduce this lack of data, a series
of uni- and multiaxial compressive strength tests was conducted in the ice labora
tory of the Hamburg Ship Model Basin (HSVA). Results from these tests are presented
in this paper.
Like a lot of other materials, ice can withstand some stresses without any failure
or yield, while under some other stress conditions it fails. These failure stress
states form in the tree-dimensional stress space a closed surface, so-called failure
surface. The failure surface looks similar to a flat rugby ball, the longitudinal
axis of which lies in the hydros tatic axis0, 2 : 03' The most prevailing part
of this "rugby ball" is positioned in the tria xi al compression octant of the stress
space. All stress states within the surface are possible, while stress states outside
the surface are impossible, because the ice fails before they are reached. The exact
shape of the "rugby ball" is influenced by different parameters like temperature,
salinity, strain rate and crystal structure. The overall objective of the investiga
tions presented here was to determine the triaxial compr'ession part of the failure
surface of saline ice frozen under simulated natural conditions, the strain rate, the
temperature and the salinity being kept co nstant. To reach this goal a testing
apparatus was needed on which a large variety of stress states could be realized.
At present the commonly used testing apparatus for mu1tiaxia1 testing of solid
materials is the standard triaxial cell (e.g. Jones (1978) [7] or Fokeyev (1976) [1]).
Since the fi rst tests conducted by v. Karman in 1910 [8] the triaxial cell has been
developed to a high standard, but only stress ratios of the form 0, ~ <J
2
: <J are
3
possible. [n addition strain measurements directly at the tested specimens are rather
di fficult to per fo rm. The testing apparatus of Frederking [2] on which some of the

527
first multiaxial tests on i ce have been co ndu cted , is restri c ted even more, because it
i s designed for the plane s train co nditi on only . In connection with the construction
of prestressed conc rete shelters for nuclear plants a varie ty of multiaxial strength
tests on concrete have been performed during the last two decades. Some new designs
of mul tiaxial testing machines have been developed, which provide for a la rger variety
of s tre ss states than the s tandard tria xia l ce ll (see Gerstle, et a1. (1976) [3]). For
the tests presented here one of th es e new deSigns, the "Brush-Type Loading Platens"
developed by Hil sdor f (196 5) [6) was adopted.
Test ing Facilities
Tests were pe rformed in the ice laboratory of the Hamburg Ship Model Basin. The i ce
laboratory has two refrigerated rooms. Their tempe ratu re can be se t independen tly
between +20 C and -30 0 C. In one of the two refrigerated rooms a 2.20 m wide , 2 .70 m
l ong , and 1.05 m deep freezing basin i s instal led. The ba s in is insulated carefully
and has an adjustable overflow pipe. He reby the ice covers frozen in the basin
correspond to a piece out of a wide, homo ge neous natural ice cover.
The secon d cold chamber i s used for spec imen preparation on a band saw and on a lathe,
and for performance of uni- and mul tia xial ice strength tests on a tria xial t esti ng
ma chin e. The ve rtica l axis of the testing ma ch ine consists of a screw driven TREGEL
RME lOO' l oading frame, while the horizontal axies consi st of a HSVA-de s igned quadratic
s~e ~frame with two LUKAS-Servo-Se t hydraul ic cy linde rs (Fig.I). All th e three axies
have a maximum load of 100 kN and can be
electronically closed loop controlled for
the force-, velocity- and strain- rate mode
independently from each other. For load
application the tes ting machine is equipped
"lith bru sh-ty pe loading platens (" bru shes ") .
Brushes have already been used successfully
for mul tiaxial material strength testing on
concrete (e .g. Kupfe r, 19 73 [9) or Grasser,
et a 1., 19 75 [4 J). The brushes used here
have been designed especially for ice.
Every brush cons i s t s of 169 bra ss bars
(bri st l es) which are 200 mm long. Cve r a
1ength of 60 mm the bri s t l es are fi xed in

Figure 1.: Tr ia xial testing machine

528
Figure 2.: Brush-type loading
pI a tens wi th defl ec ti on
transducers, tri axi a 1 arrangement

Figure 3. : Uniaxial compressive


strength test with
brush-type loadir.g platens

0.4 mm thickness provide for uni


distances between the brist
les . The main advantage of brush
type loading platens is, that
because of their flexibility the
bristles follow evel'y transverse
deflection of the loaded specimen
with a minimum of resistance. The
undesirable effect of transverse
strain constraint,known from stiff
loading platens is avoided (6] . On
the oth~r hand the axial stiffne~s of the bristles is similar to the one of a stiff
loading platen, which is desirable in order to obtain a high stiffness of the loading
system (see Sinha and Frederking, 1979 [101). Another advantage of the brush-type
loading platens is the poss ibility to attach deflection transducers near the top of
the bristles (Fig .2) . Hereby deflections of the loaded specimen can be measured even
in triaxial tests, when all six planes of the cubic specimen are covered by loading
platens. Except for evaluating the specimens' mass and size all measurements were
done by means of electric transducers and measuring instruments. The measuring
instruments like carrier frequency amplifiers and digital thermometers are installed
in a heatable measuring room, which is positioned beside the two cold chambers.
The electronics of the closed-loop control system of the testing machine are installed
here too. The ice laboratory is con nected to a Hewlett Packard 21 MX computer on
which all time dependent measuring data are digitized and recorded on-line.

529
I

Testing Procedure
In order to keep testing conditions constant the specimens to be tested were taken out
of only one 21 cm thick ice cover which was frozen in the ice laboratory's freezing
basin under reproducable conditions. The freezing process was initiated by seeding.
The raw specimens were stored at a temperature of -30 0 C. Specimen preparation was
conducted at -22 0 C. The cubic specimens had a side length of 69.8 mm with a tolerance
of :t 0.15 mm. The non-uniformity of side lengths in one direction of every. single cube
was less than :t 0.05 mm. (For comparison: the non-uniformity of distance between the
two parallel loading planes of the brushes of one axis was :t 0.02 mm). After prepara
tion the specimens were stored again at _30 0 C, now packed in airtight plastic boxes.
For one day before the tests the specimens were stored at the testing temperature in
order to achieve a good uniformity of ice temperature within the specimens.
Just before the test the specimens' mass and dimensions were measured. While position
ing the ice cube in the testing machine, the loading platens were driven to a pre-load
of 0.5 kN corresponding to a compression of 0.1 rPa to assure a good contact between
specimen and loading platens. In addition the top of the bristles to which deflection
transducers were attached was fixed to the specimen by a drop of freezing water. In
uniaxial tests a parallelogram guided deflection transducer was attached directly onto
the specimen (Fig.3). During the following strength test the hydraulic x - axis of
the testing machine was driven strain rate controlled as "master axis". The actual
force value of the x - axis, optionally reduced by a factor of a up to 1, was taken
as nominal value for the y- and z-axes, which were driven closed loop force
controlled. The time history of the three forces and of 6 deflections was recorded.
Immediately after the test air temperature near the brushes and ice temperature
inside the tested specimen and inside a control probe were measured. Salinity was
evaluated from the melt of the tested ice cube.
Evaluation of test results
Since all measuring data had been recorded on-line in digital form it was easily
possible to perform the standard computations by the aid of a computer program,
which finally produced a printer output and a plot of the results of every individual
test. An example of the computer plot is shown in Fig. 4.
The average stresses existing in the specimen were computed following the equations:
Ox Fx / (ly x lz) (1)
0y Fy / (lx x lz) (2)
0z Fz / (lx x ly) (3)
with 0x,y,z being the stresses, Fx,y,z being the forces and lX,y,Z being the cubes'
side lengths related to the testing machine's coordinate system.

530
With respect to the controlled
strain rate it is important to
3.200
note that the correct strain
formul a
2 . 00 = u/(c- u) (4 )
with being the strain, u the
1.800 deflection and c the basis
length, was simplified by neg
.aoo
lecting the change in the basis
length under deflection
a .000 ~=::.:====:::::==:::===;;==::::~ =u/c. (5)
EP5X2[(!']
20 . 000
EP5YI[+] Considering the small strains
EP5Y2( .]
EP5Zl(.] occuring in the ice cubes this
12.000
EP5Z2(.] simplification is admis sible .
[""1M]
The elastic modulus was computed
000
by fitting the best straight
-04 .000 line through the first 20 pairs
of stress -st rain values of the
-12.000 six uniaxial tests and calcula
ting the slope of the straight
-20.0_ 000 .:000 12.000 20.000 1i ne.
EP5Xl( "M/M]
BMf! MTK 76-B H5VR 83.0132/524
HR. 163.1 YO" 1'.Il.BO 16.6' UHR

Figure 4.: Computer plot of stress Ox and 5 strains versus controlled strain EX for
-4 -1 .
saline ice (strain rate EX 2 x 10 s ,stress ratlo 01 : 02 : 03
0y : 0 X : 0Z = 0.67 : 1. 00 : 0.33; <x1: 1st strain parallel to x-a xis etc.)

Resul ts

The coordinates of 37 points of the failure surface in the three-dimensional stress


space have been determined for saline ice at TI = _10 0 C ice temperature and 11ith
3 0
PI = 910 kg/m specimen's mass density and wi th 5(10.6 / 00 sal inity in the original
ice cover . The ice was columnar grained. The c -a xis was horizontal and random
ted. The crysta l' s size inc reased from 0.5 to 1 em at 7 cm to 1 t02 cm at 15 cm
below the ice covers' surface, corresponding to the top and to the bottom of the speci
mens in their or iginal position. The strain rate parallel to the master axis was
4 1
EX = 2 x 10- s- . With respect to the planar is otropy only 22 stress ratios had to
be investigated to get the 37 points mentioned above. In 6 of the 22 stress ratios
the axis parallel to the grovlth direction (index "1") was taken as master ax i s ("x").

531
0,", ", / "0, /
/ /

15

:,t :(,:670 1

10

5
. I. ~"O.3JG]
//' _O _0 33~

o~ o 5 10
0, [MPaJ
20

Figure 5.: Projection of the failure surface of saline ice onto the 01 - 02 - plane
(51 = 10.6 0/ 00 , iX = 2 x 10- 4 s-~, TI = -10 0 C)
For every stress ratio 3 tests were performed, i.e. in the total 66 tests. In 4
cases no maximum stresses could be determined, because stresses were still increasing
at the end of the tests. The maximum stresses, averaged over all tests performed for
each individual stress ratio are taken as the coordinates of the failure surface in
the three-dimensional stress space (Tab.1). Fig. 5 shows a projection of the failure
surface onto the 01 - 02 plane. In the part above the 45 0 axis the master axis was
perpendicular to the growth direction, while in the part below the 45 0 axis the
master axis was parallel to it. In Figures 6,7, and 8 only that part of the failure
surface is shown, for which the master axis was perpendicular to the growth direction.
The isometric drawings in Figures 7 and 8 give a three-dimensional impression of the
failure surface's shape. Figure 6 shows its projection onto the 02 - 03 plane. Here
it can easily be seen that the biaxial strength of the saline ice tested here can
reach up to 4.5 times the uniaxial ice strength for stresses only perpendicular to the
growth direction (0 1 = 0). If in addition compression acts in the growth direction
(triaxial, 0 1 f 0) ice strengths of up to 8.5 times the uniaxial strength perpendi
cular to the growth direction have been determined (refer to 01=0.33 O2=1 03=0.33).
If the ice is loaded biaxially with compressive stresses parallel to the growth

532
direction increasing from 0 % to 100 % of the stresses perpendicular to it, only ice
strengths of less than 1.33 times the uniaxial strength have been measured.

15

10

o.no,
5
----

10 15 20
02 [MPa}

Figure 6.: Projection of the failure surface of saline ice onto the 02-03 - plane.
o -4 -1 0
(51 = 10.6 / 00, 'x = 2 x 10 s ,T[ = -10 C).
Yaung' s moduli determined from the uniaxial tests are El = 15.1 : 2.9 GPa parallel and
+ . . . . .
[2,3 = 4.56 - 1.94 GPa perpendlcular to the growth dlrectlon. In the unlaxlal tests
the elastic part of the stress strain curves ended at stress values of 20% to 35% of
the maximum stresses. The strains measured parallel to the ma s ter axis \vere rather
small . At maximum stress only strains of 0.1 % to 0.4% in the uniaxial tests and of up
to ca. 2% in multiaxial tests have been measured. Similar results are known from
previously performed uniaxial tests [5].
Discussion, Comments and Conclusions
Due to the anisotropy of the columnargrained ice, biaxial compressi ve strengths under
stresses perpendicular to the growth di rection are much higher than for isotropic
materials. They are also much higher than under biaxial stresses, the one being
parallel and the other perpendicular to the growth direction. This is in accordance
with the findings of Frederking (1977) [21, who reported among other data, strength
values for plane strain compression of columnar grained fresh water ice at -10 0 C and
at a strain rate of i: = 10- 4 s-1 of 2.5 times the uniaxial strength. For the more

533
isotropic granular s now ice an increase of only 251 is reported. This value
corresponds to the data on concrete under bia xial load presented by Kupfer (1973)[9].
As mentioned above, the deflection transducers measured the deflections directly at
the specimen. This provides for more exact strain values than the commonly used
"nominal strain" - method, because the disturbing influence of the contact zones

Number Failure Stresses


Ratio of Stresses of 01 O
2
Nr . Code 03 Tests [MPa] . [MPa]

1
300
o o 3
10.05 0 o
2
030 o 3
o 2.06 o

3
310
a x 0.33xo x
o 3
12.44 3.72 o
4
320
ax 0.67 xo x
o 3
6.30 3.46 o
5
330
1.00xo x
o 3
2.48 2.47 o
6
130
0 .33 xo x
o 3
. 73 2.10 o
230
O.6 7xo x
o 3
l.84 2.73 o
8
031 o Ox O. 33xo x
3
o 3.46 1. 19

9
032 o Ox 0.67 xox
(3) o 6. 98 4.72
10
033 o l. OOxo x
(3 ) o 9.36 9.40

11
311
Ox 0.33x o x 0.33 xo x
3
17.33 5. 77 5.85
12
321
0.67xo x 0.33x o x
3
15.73 10.36 5.39
13
331
0.33xox
3
4 .49 4.49 l. 54

14
322
ax 0.67x ox
3
18. 54 13 .20 12.50

15
332
1.00x ox Ox 0.67xo x
3
8 .65 8 .6 5 5.83
16
333
1.00xo x
1.00xo x
3
14.18 14.20 14 .25

17
131
O. 33xo x ax 0 . 33xo x
3
l. 26 3.74 l.28
18
132
0.33xo x a x 0.67xox
3
2.78 8 .2 3 5.50

19
133
0.33x ox 1.00xo x
( 3) 5. 79 17.12 17.07
20
231
0.67x ox ax 0.33x o x
3
3.65 5.45 1.86
21
232
0.67x ox Ox 0.67xo x
3
5.44 8.15 5.51
22
233
0.67x ox 1.00xo x
3
7.73 1l.62 1l.63

Table 1.: Hultiaxial compressive strength tests on sa line ice: stress ratios and
o -4 -1 0
failure st resses. (SI = 10.6 /00 , EX = 2.0 x 10 ~ ,T] = -10 C; the

master ax i s is indicated in the stress ratio columns by "ax" without any factor)

534

.J
between specimen and loading platens is excluded.
The "brush-type loading platens" used in the tests presented here have proven to be
a good means for the determination of multiaxial compressive strength data of ice,
especially if they are combined with a closed-loop contro l system. Tensile stre ngth
tests with brush-type loading platens, which have already been performed on concrete
are as yet impossible with ice because of the lack of an appropriate glue to adhere
ice specimens to the loading platens.

0,

["'Pal

15

-1--..
5
10

10

Figure 7.: 0, [MPal


I sometric drawing .............

of the failure surface


of sa line ice, view along
(01 = - 02 3)-axis

Figure 8.: Isometric drawing


of the failure surface
10
of saline ice, view along the

hydrostatic axis (01 = 02 = 03) 0, [MPa]

535
Acknowledgements

The in ves tigation s presented above were made possible by th e finan c ial assistance of
the German Ministry of Scie nce and Tec hnol ogy (BMFT). The basic equ ipme nt o f the screw
driven axis of the testing machine is a l oa n from the Deutsche Fo rsc hung sgeme ins c haft
(DFG). The contribution of Dr. S. Stockl and Mr. H. Aschl t o the design of the testing
machine is most gratefully ack nowledged. The author thanks Dr. J. Schwar z for his
many suggestions and encouragements and ~Ir. l'. Neper for his mo st accurate prepara
tion of the specimen s and his ass istance dUI"ing the tests. He also th a nks all the
colleagues, who contributed to th e work presented in this paper.

References

[1] Fokeyev , N.V., 1976: Determinat ion of the compressive stre ngth o f artifical
i ce s pec ime ns of different sa liniti es under co ndition s of combined stress.
Proc ee ding s o f the Arctic and Antarctic Research In s titute, Vol . 33 1, 1976 ,
pp. 189 - 20 2 (o rig. in Russ., Engl . t ra ns l. Ottawa, 1980).
[ 2 ] Frede rk ing, R., 1977: Pl a ne s train compressive s tre ngth of co lumnar grained a nd
granular-snow ice. Journal of Gl acio l ogy , Vol. 18, No. 80, 1977, pp. 505-516.
[3] Gerstle, K.H. et aI., 1976: Strength of Concrete under Multiaxial Stress States .
r1c Henry Symposium, October 1976, Mexico City, pp. 103-131.
[4] Gl'asser, E., D.H. Lin se a nd H. Aschl, 1975: Festigkeit und Verformung von
Beton bei mehra c hs iger Beanspruchung. SBB-Tagung des Deut sc hen Au ssc hu sses
fUr Stahlbeton (DAfStb) am 13. / 14. Oktober 1975 in Berlin, Band 3/6 , 17 p.
[5] Hausler, F.U., 1980: Dru ckfestigke itsversuche mit Salzwasser-Eis. Procee dings
INTERr1ARITEC '80, September 24/2 5, 1980, Hamburg, pp. 403-413.
[ 6 ] Hilsdorf, H., 1965: Bestimmung der zweiachsigen Festigkeit des Betons.
Deutscher Aussc huB fUr Stah lbeto n, Heft 173 , Berlin 1965 , 68 p .
[7] Jones, S.J., 1978: Tria xial testing of pol ycrys talline i c e. Third Intern at ional
Co nferen ce on Perwa fro s t, Edmo nt on, Alberta, Jul y 10- 13 , 1978.
[ 8 ] Karaman, Th. v., 19 11 : Fest igke itsversuc he unter a ll se itigem Dru ck, VOl-Heft
Nr. 42, 19 11, pp. 37 - 68 .
[9] Kupfer, H., 1973: Das Verhalten des Betons unter mehrachsiger Kurz ze itb e l ast ung
unter besond erer BerUcksich ti gung der zweiachsigen Beanspruchung. Deutscher
AusschuB fUr Sta hlbeton, Heft 229, Berlin 1973, pp. 1-10 5.
[10] Sinha, N.K. and R.M . W. Freder king, 1979: Effect of system stiffness on stre ngth
of ice. Proce ed ing s Vol. 1, The 5th International Conferen ce on Po r t and Ocean
Engineering und er Arc tic Cond itions, August 13-18, 1979 (POAC 79), Trond heim,
pp. 708-717.

536
MULTIAXIAL COMPRESSIVE STRENGTH TE STS ON SALINE ICE
WITH BRUSH-TYPE LOADING PLATENS

Franz Ulri ch Hausler Hamburg isc he Schif fbau


Versuchsanstalt GmbH.
West Germany

Question from: Y. S. Wang, USA


1. What were the shape s of the stress-strain curves?
Were the sam ples failing in a ductile or brittle fashion ?
Did the stress-strain curves look linear in the beginning of a tes t? What I am
getting at is whether you have seen any initial lagging of the force due to
possible mismatch of loading surfaces.
2. How much strain did the samples generate before failure? I assume that under
different stress combinations the y will be different.
Answer :
1. A typical stress-strain curve i s shown in fig. 4 of the paper.
The samples did fail both in a ductile and in a brittle fashion depending on
the actual stress combination.
An initial lagging of the force has not been observed.
2. Failure strain, here defined as the s train at the moment, when maximum stress is
reached, was as mentioned in the paper 0. 1 to 0.4 % under uniaxial and up to 2 %
under multiaxial stresses.

Question and comment from: N.K. Sinha , Canada


Although brush-type loading platens and the use of cubic geometry for the specimen
were claimed to work well for concrete, it is a mistake to have an apriori
assumption that the system will work well for ice also without examin ing systemati
cally the mode of failure in ice and the interactions of the ice specimen and the
test system. Polycrystalline ice exhibits cracking activity. This is known as high
temperature embrittlement because of the high homologous temperatures involved.
The cracking activity has a profound effect on the outcome of a test. The
illustration below shows results obtained on a 10 x 10 x 10 cm specimen using the
brush-type platens and on a 5 x 10 x 25 cm specimen using polished steel platens.
Both tests were conducted using closed-loop test systems. Unia xial compressive loads
were applied perpendicular to the axis of the columns of laborator y made fresh water
columnar-grained 5-2 ice. In both cases ice was made at the Division of Building
Research, National Research Council of Canada. The differences in the two results

537
are considered significant. Could the author make any comments to clarify the
di fferences.

5.8

COLUMNARGRAINED 51 ICE
4.5
AVERAGE GRAIN SIZE' 4 TO, mm
IEMPERATURE 'IOC

4. e CONSTANT STRAIN RAIE I x 10" s'l

3.5

3.8

SINHA. 1981
15 x 10, . 1, eml
2.5
':'
E
::i::
~ 2.B
<J)
<J)
LLl
~
<J)
1.5 eml

I.B

l5

U
~ ~ I!l
Time (sec. )

Repl y:
The HSVA test result shown in the illustration is one of only four tests which have
ever been performed on fresh water ice with brush-type loading platens. Even though
a comment on the shown difference is possible:
Doubtless there are stress concentrations at the edges of every single bristle. In
the case of virgin pure fresh water ice this may lead to a higher cracking activity
compared with the load application by means of polished steel platens. The friction
between the polished steel platens and the sample gives a lateral strain constraint,
which reduces the occurance of cracks at the specimens end faces. The lateral strain
constraint of brush-type loading platens is close to zero.

538
In the case of saline ice, whi ch has been investigated in the test program presented
here, the difference due to micro cracking caused by stress concentrations at the
bristles should be, if existant, much smaller, since saline ice contains already
a lot of natural disturbations like the brine pocket s , which give a large number of
initial pOints for cracks. A direct comparison between brush-type and rigid polished
platens would give here a final answer.
What concerns the cubic geometry: due to the neglectable lateral strain constraint
of the brush-type loading platen s the unia xial compress ive strength is independent
of the specimens height. This allows the use of a cubic geometry, which is needed
for the multiaxial te s ts.
After all we sh ould not forget, that the HSVA brush-type loading platens are
deSigned for multiaxial tests on ice. They allow to perform true tria xi al tests
at an arbitrary stress state, with a minimum lateral strain constraint. If the
adhesive-problem is solved even multiaxial tension tests are poss ible (already
performed on concrete). Strain measurements directly at the specimen can ea Sily be
performed even in triaxial tests. Thi s allows a closed-loop strain (-rate) control
at all stress combinations. Regarding all these advantages a certain error caused
by st ress concentrations at the bristles is acceptable.

539
PRESSURE DUE TO EXPANSION OF ICE SHEET

IN RESERVOIRS

Iu Bomeng,Engineer Water Power Research In s ti tute,Northeas t Peopl e 's


Designing Insti tute,Ministry of power Rep ublic
Industry of China

ABSTRACT

The paper deals With temperature fluctuations of ice sheets associated


With changes in air temperature, main factors affecting ice p ressure
and variation of pressure due to expans~on of ice. Data obtained at
five reservoirs were analyzed. The maXlmUffi thru s t of i ce sh3et occurs
in the season when ice is th~ c kest, after succes s ive days of tem perature
rise, and with mean initial ice temperature at about - 5 - _6 0 C. 'IIi th
initial temperature still lower, the pressure is small e r. It takes 30 r.r
to develop the maximum pre Ssure with sus tained rise in tem perature.
0
After the mean ice temperature has reached -1.5--2.0 C or so, f urther
rises in temperature do not bring ab0ut pre ssure increase. A formula ha s
been derived for computing ice pres s ure dlrectly from data on a~r t ~ mr

erature. The computed and observed values a g ree i n t h~ mai r. .

INTRODUCTION

Winter in the northern sections of China is severely co l d and a n umbe r


of hydraulic structures suffered damages to dlfferent extents because
no adequate provision had been mad e to allow for iCe thrust. Thus,study
on ice pressure is of much signifi cance to t he d esien of hydraullc
structures in this part of the country.

In recent years, systematic in-situ obs e rvations have been carried


out in the Northeast and informati on conCernlng ic e thrust as related
with temperature fluctuations and movement of ice has be en collected.
A comprehensive analysis is made on the obs e rved data, to fin d out the
540
relationship between air and ice temperature,and between ice temperature
and thrust. A method of computing static pressure exerted by ice sheet
is also presented.

TEMPERATURE OF ICE SHEET AND ITS RELATION TO AIR TEMPERATURE


]!'actors contributory to the the mal condition of ice sheets are manifold,
air temperature,sunshine and snow cover being most sign~ficant.

In accordance with field data,fluctuations in air and ice temperature


are of three types:
(a) Fluctuation under ordinary condition in a s~ngle day,exhibiting a
marked periodicity on a one-day cycle expressed approximately by
sine curves in most cases.

(b) Temperature drop over successive days,with small rises during day

time.

(c) Susta~ned temperature rise for two or more consecutive days. During
daytime,air temperature rise s abruptly on the f~rst day,followed by
a small drop at night. The same happens again on the seco nd and
so metimes the third day.

There is a l ag between change in air temperature and rise or fall of


ice temperature,the time ~nterval being larger a t g reater depths. This
is particularly so with the presence of snow cover.
Air temperature affect s ice temperature mainly with~n 30-40 cm of the
surface. Further below, the fluctuat~on is rath er mild. be~ng pract~cally
linear (figs.l & 2). It is seen that the ma gn~tud e of thrust due to ex
pansion is chiefly governed by fluctuations in temperature of the upper
layer of ice sheet.

The variat~on o f ice temperature is a complicated process of heat


exchange between the ice sh ee t and the surrounding atmosphere. Correlat
ion analyses have been ma de on tem pe ratur es a t di ffe rent depths of the
ice sheets measured at several r ~fe rvoirs and the data on air temper
ature obta~ned from the loca l observatory. The follorJng relat~onshi ps

have been establi s hed,though Cl p T'rox ~ :!Iat ", l y.1'or t he cas e of no sno w COVeY':

(a) Relation betwt:en ai r tum perature ( t " .oC) a nd t ,,'I1 pLY'at!:re of tl ,e

ice surface at [) a .m. (t ~ o .0C) on H. E SClme da y:

541
(1)
Ice tempera {ure ('C)
-8 ('0) Vanation 01' .LCe tEcmperature
/ at 8 s.m. ~ith depth i s esstint
3 iRlly llnear \f l ~~ .l i 2),and
5
~ay be written thus:
'"
<.>

- V (2)

j
~

~ Jo 'd;ere ti h --- tenperature at .3.

c a r tal~ depth ,f ~hd ice sheet


~ 6J . at Sa.", (oC); hl --- depth In
~ oA); lce sheet (cm); and h

~/oo j tnickness of ice stlee t (cm).


i!: , ;r} \ c) Relatlon between incr ement of
~I'l' o
aJl t em pe rature (Ata' C) b'om
hjl. fllLClualion of lemperature of ice S R.m. to 2 p.m. a~d avrage
sheet durinJ successive d.ays of
rising atmospheric temperature I n<; ~ <.m <,;nt of iC e temper3tllre
(Yinlze reServoir, )farclr Ij -/6, 19Ttf) (At1l,oC) measured at the top
1- B: 00 l1arcA 15 2 -14 : 00 /'farc/' I j
4- 8 f1arch t6 la yer (0 -15 cm):
3-17,30 !la-rclt IS :00

5 -It,: 00 I'farc/r. /6
For t : ~l, e r stu re chal e; in an

.;)til = O,5.'.t (3)


s

For tem;:>eratllre risl; In suceessi Ve days

LltLl = O. G2Llt" (4)

(d; '<ela tio :] bet.v"en S" e l 'R f,8 r8te of t,,:~perature rise In the top layer
(Sl,oC/hr) and that in tLe second (S2' C/hr):

(5)

Jlffect of im tlfll t em'leratur of ice

It is Generally reeo Gnlz ed that,w.Lth the same rate o f temperature rise,


t~, e lower the ini tlal temperature of lce (:neasureu at 3 a.m.), the larger
wlll the ice tt:rust Le. In-sltu observations,however,8lve results other
wise.
Cbs,,:-ved values of mean ini tlal ic e tempE;rature and average increment
542
Ice temperature ('C)
-16

/oj

iO 20
'-'
~

'"
""...," .0
'"
.~
'"
<>
~
'""
"'< do
.~
/.. 2 .. J.
~ J'~rage Utcre-u,,1 0/ ice
le"'JN!lUlure .IIi ('C)
[e71t?erature fluctuation in a single !ii.' p-i.-Ali rddi,,"sltip at
dar<ErlOJIjnu reserVoir. January 6 , /977)
1~8:00 2 -1/:00 1--/~:OO 1.I~/7:00
s.Il,,!.ewtUf feser,,,ir, Iff~-IJI'f

01 i c'" t"mpera ture were plotted agains t ice pressures, as in fig.3. I t can
be seen that, wi t h tile same increment, the lower th e ini tial temperature,
the smaller is the ice pressure,mainly because there exist irregular
cracks ~n the ice sheet due to di1ference in tempera1;ure between the up
per and lower part of ice. with temperature fall and non-unifo rm con
1;ract~on,cracks d~vel op. ~hose runn~ng thrcu8h th e sheet w~ll be filled

with water t o form new masses o f ice,exertinG pr8ssure during temper

ature rise. "'he cracks at the top of the natural ic e sheet which do not

run through contract or expand periodically in step with temperature

fluctuations. When ice ex~ands , a part of the expans~on take s place


freely. But when the spaces are all occup i ed by the expanding ice,further
increase in temperature will produce ~ncrease in pressure. The lower the
ini t~al temrera ture, tt'E larger ore the crac~s to allow unrestrained ex
pansion. Thus, wi th the same ra te of telllperature rise, the lower the ini t
ial temperature,th e smRller will the p r essure be. This certainly does
not mean th~t ice pre~sure will ur.der all circumstances be larger with
hieher ini t~al temperature. When ice temperat u re rises, stress-relaxa tion
becomes morc pronounced. If U: e mean ini t~al ~ce temperature exceeds a
certain level, t,, ~ pressure will no l onger increase with an increase in
ini tial t emperature. The maximum pressure Ilsually occurs when the mean
initial ~ce temperature is -5~_6C or so. It was in such a way that the

543
hi c he st ic c pr~s8ures oc curred sEveral t~ ~ es durlng the f irst 20 days
of February l n Jilln and l n the f irst 20 day s of Ma rch ln HCllongjlang.

Di sar re emen t between res ults obtal ned from tests o n small samples and
fleld obs e!'vatl0ns,and t he effect of cracks liave been rn~ntl0ned by a
number o f res"R r ch worke rs. Th e pro blem here i s hl) '; ! to estlma t e the
e ffe ct cor rectly.

In c o~n uting preSsure due to ~x n ansion of i ce Sheet, th e character of


.indi vidual ic e masaes aJld the co nrti ticn of the ice-tielC1 as a whole
s hould both be t <:ken i n to c ons i deration. Th e re latl on be tween i c", p re s
su r e OJ) and initla l t ~ mp l;ru.tu re (ti) can be ex;> n .sseu thus:

(6)

wF.e: r e a , C a t~ d n are constants .

Effec t of ra t ~ aLa '~ ;J ru t.i.on of t...rn ;Jerature r l se

It has be e n proved by I n-situ and l " bora 'Lory tests th i-lt ,0 cf,'- " coml ~tion s

beET j.dentical,th e hlEher the rd-l;c of t ~ '-ne rRtvr c r ise,the larC" l ' is
the i c", pressure . With
the same rate o f temper ~'----/\
ature r ise ,ice !lr essure
increases with ti"te. But
~

~
~

<\)
/I '
wh en the mean ic e temp e r "
' ~
I
-:;0 / I
'C; -
ature ha s reached a ce r 'lI i" f
'
I.
tain va lue,the !lress ure .....
;:j

V r
r ~

Qv\ '"I 11\2


10
no l onge r i nc reas es ; on ~
the con trary,it tends t o
1}
.....'"

v
decrease. The thresh old
"
<:j
I
was found t o be in the ~
vicinity o f -1. ':l- - 2.0 0 C /
I
(see fig.4). Thu s , ice t) 00 I~ 00 }O flO ./ ou 8 00 I" . 0 0 20 IX) l ()O 8 oc) I~ .00 :to 00

pressure i s related to febru ary 16' f ebruary 19 february 20


both the rate a nd the II!! " fLuctuation o( tempaalUl'e and pressure
duration o f temperature durinJ successive days o( temperat ure ,-ise
(ErLonJhu reservoir, february 1!i77 )
nse. I - mean temp era/ure 0/ ice
e pre ssu..re
2 - ( f
Duri ng t e mp ero ture fall in succ <;ssi. ve day s , Llle ic e pre Ss ure is

rathe r low, or even null. ';lith air tem pe rature va r y ing unC1 e r orC1inary

544
co ndi t ion i n a s ingl e day , ~ cO!l1para ti vt; ly h .!.: ~ h rr(; s su r" maj he e ;;pe c t ~d
i f the i niti al i ce t c",peraturc a nd the rut e of t ePlp t' r a t u r e n .se ::.l r e bo t h
r e l a t lve l y hi gh , ave r a ge i cu p r ess ur e l nc r ea s l ne wlth a n increa se ln a l r
tem perature. Be tw een 14 :00 a nd 16 : 00 nr , ho wev e r, the p r ess ur e beGi ns t o
dro p (see f i g . 5) . From t he 0 bserved da ta o f SOel " re se r VO l r s , l t has al s o
be e n f ound t ha t the hi ghe s t p res s ure oc c urs unde r t he t oll o wlng condi t
o
i ons : comparati ve ly hl gh in i tia l tempera t ure , being _5 ", _6 C o r 50 , ,,- nd
sus t a i ne d te!l1r e r s ture ri se i n
co ns ec utlve days . Ge neral ly ,max
i mum ice pres s ure oc curs o n t he
Se co nci da y of te r~pera Lure ri s e
(s ee fig .4) , o r- o n the t hi r d , if
COVered bY snow (fi g . 6) . The

- -, ..
~
~I" O~ 14 :OQ 11:'
JD~
Ii",~ (,il') effec t of the du n, "l o n o f s us tain
ed rise i n t emperA ture s tlO uld be

....~

-6t1 4(JC.
-4
. .. _etl."
take n in to ac co unt

.....
L------
~
-6
~ l ()C-r

Ef f e c t o f s uns hi ne and s nON Co ver

---
-.....
-~:~
10'"
It
A compari so n of t he re l e van t o b

~" - IZ Ge rvecl data rev eals t r,at ice


fig. , . FluciuaUoll of ,c e teMperatu.re p r e ss ur e on an ove r ca s t da y i s
OJId pressu.re du.n,,! a Jilf!l lb..] mu c h les s t han that in f i ne
,." orriillar)' weo.tier (IrloJlJllu
r"s~r"o"r. J.lfu .. rf 6. '177 )
weat hers . Hen ce , the effect o f
s unS hi ne shou l d be
ta ken i nto co nsider
ati on in co mputin g ice
p ress ure. Ice thrus t
at Yi nhe and T ai y ang~
Sheng reservoir wa s
l owe r than e l s ewhere.
This is a t tri bute d
mainl y to t he presen c e
of sno w co ve r (lO-l 5cm
t hi Ck) thTo ug ho ut the
wi n ter. As r ega r ds ice
tem peratuTe flu c tuat
i ons wit hin a singl e

545
day, under ordinary conrlltions,the increment from 5:00 to 14:00 hr on
an average day at the fo~cr two reservoirs generally does not exceed
0.5 0 C. At the other reservoirs,the increment is 2.0 0 C or so. At Yinhe
and Taiyangsheng,the averagc lncrement has never exceeded 1.9 and 2.7 0 C
respectively for two and three consecutive days of temperature rise,
whereas at the others,the increment exceeded 3 or 4 0 C.

Effe ct of contact and other factors

Beside being influenced by temperature,condition of ice-field and mech


anical properties of ice,the pressure exerted by a natural ice sheet is
also affecten by the conjition of restraint such as bankform and shape
of the water surface,as well as geographical location and general topo
graphy. A coefficient is to be introduced in the formula appropriately
to account for the overall effect of variolls factors.

CO~! PUTA~ION S

Basic formula

The baslc formula for computlng ice pressure is as follows:


d
Pi = K.Ks(t i + C)nSib\T - e)/t i
a
(7)
in which Pi average ice pressure of a certain layer; ti --- mean
ini tial ic e temperature in that layer; Si --- average rate of temper-
at\~ re rise I n that layer; T duration of su~tained rise in temper
ature; K --- coefficient to account for overall effects; Ks --- factor
of snow cover; a, b, c, d, e and n --- constants.
DiagrRm for computation

stre~ses pr0rtuced by ice sh e et at dlfferent depths differ,owing to dif


ference in increment of lce temperature and overall condition of the
ice sheet. The stress is higher at the top than below. At the free sur
fRce,th~ stress is slightly lower, and at the depth of 10-30 cm,the
stress is highest. It then diminishes gradually with further increase
in depth, in the form of a reverse curve, with zero pressure at the
bottom (see fig.7). In order to simplify the computation,the relation
between temperature and pressure is established for the two top layers
of 15 cm each,and the stress further below is expressed in the simplifi
ed form of a triangle (fiC.S).

546
Ice pressure ("j cm- 1 ) p,
0 2

'"
0

, \
\ i
'"
<.> 10
~ ~z
\

:s
::- 10 \
j- t-
~
10 /
.. .;
V
j
V
ftJ-,r Diagram for
fo / /
computation of
ice pressure
60
l) V Accoriinf. to tr.e ri lR['TafJ' ,1;l:e D.vt:rE.ge

70
/7 ressure of ice sheet (P,ke cm- 2 ) can be
d" termlned thus:
80 I (8)

90 ~ .ller-e Pl and P 2 are averaBe lce pressure


for the layers 0-15 and 1 5-30 cm r e spect
ively (kg cm-?).
fig - 7_ Jiart"ahon t-ce pressure 0/ By inserting eq. (7) into (8) ,wt: o::>tain:

wi/It depih

/ . Ta.ipi'njcJri re.seryot",. P ~ 0. 5 K.K s [( '0 , (. \-t il+ ,,)n 5 i l bit


7 /.)
v 1 i'.
a +
(february t5 , 197* )
2_ Er/o'jilU resery.ir (Ja.nuary6, 1917) (ti2+C) nSi2 b/ti2 a ) (T d -e)
J. Taljan!.Ilreng reserYOt'r

(f~brua,.J' 2f, 1979, ice slteet covered


(9 )
bYS110W, 12 -1$ eM lI,iel<) '~heini t ial temp emture , rate o f t,,~?

erature rl8e o:s ><ell as aVerCl_s e ice


pressure for ~epth9 of 0-15 em and 15-30 cm durine; the days when the
average pre s s ure o f ic t: exceeds 1.0 kg cln- 2 arc' selectea and the con
stants are determined uy method of mathema ti ca l s t atis ti cs ,refe rrlng to
test data. Thence,th e relatlonship hetween ice pressure Rnd temperature
is established. T!y ln sc rtin{," eqs . (1) to (5) i r.trJ the forctr"in8 :Connula,
transforming , rearrancin g and simnlifyin6,we obtain :

(10)
in whi ch \, --- lnl tinl fur terl!'erc:.tu re at ) 3.m. (IJ C) , ger c rc.l.ly liot
exceeding -leoC; 4ta --- increment of r ise in air temperature (oC ) from

547
8:00 to 14:00 hr,that from 8:00 of the first day to 14:00 of the second
or third day being taken for sustained rise in temperature; the highest
air temperature should not,however,exceed OOC; T --- duration of sus
tained rise in temperature (hr) corresponding to Jt. For ordinary wea
ther T = 6 hr,for successive days of rising temperature T = 30 hr,and
with presence of thick snow cover,temperature rise over 3 consecutive
days may be considered,hence T = 54 hr; K --- coefficient to account
for overall effects (K a 4-5); Ks --- factor of snow cover,in case of
no SDOW Ks = 1; Ch --- conversion factor related to thickness of ice,
as shown in Table 1.

The K-value in the above formula reflects the effect of various fac
~ors,including different conditions of contact. K-value is based on
data obserYed at the five aforementioned reservoirs and should not be
a constant (see Table 2).

~ ,,! rt'Serioir
/( yatu~

Tinhe and Taiyangsheng reservoirs are in mountainous regions,with


capacity of over 100 million m3 ,and have broad and extended water sur
faces. Under similar conditions of restraint,K-value is the same for
both reservoirs. Shanghewan reservoir is of small size and lies on the
plain. Its total capacity is 5.8 million m3 The valley slopes are
gentle and the upstream slope of the dam is provided with smooth as
phalt facing,which exerts less restraint,so the K-value is low. Taiping
chi reservoir,also in level country,is of large size,Wlth vast area of
water surface,so that the K-value is high. However,K remains more or
les8 con8tan~ under definite condition of restraint. Earth dams of large
reservoir8 on the plains are almost universally subjected to ice
thrus~.,thatat Taipingchi being most severe. In many other places in
the Northeast,there occurred quite a number of slope damages by ice in
winter 1973-1974. The K-value obtained through measurement of ice pres
sure at Taipingchi can therefore be taken as a design value for similar
reServoirs. Due account should however be taken of the slight increase
in iCe pressure as a result of the excavation of a trench in the ice
548
at a distance from the field of observation, which produced atres8
concen~ration. The value of K may be ~aken a8 4-5 for design purpose.
All in all,for conditions in ~he Northeas~ in general,higher yalues of
K may be assumed for large reservoirs on the plains in Jilin and lower
values may be adop~ed for large reservoirs in Heilongjiang and other
reservoirs in the mountains. Still lower values may be appropria~ely
used for small reservoirs. Owing to the complexity of factor8 affecting
K-value,which has a most important bearing on the proper eyaluation of
ice pressure,it is necessary to collect more field data.
The effect of snow accumulation should not be taken into account in
computations unless covering by snow throughout the winter seasons i8
assured. In the present computation,with snow cover to a thickness of
10-1~ cm,the value of Is is taken as 0.5-0.6.
Comparison of computed and observed values

With average ice pressure aboye 1 kg cm- 2 for the reservoirs in question
selected and referring to observed data on air temperature in the local
ities,computations for pressure due to expansion of ice sheet have been
carried out by means of eq.lO. A comparison of the computed and observed
values has been made,as shown in Table 3 and fig.9. It can be Seen that
they agree closely, the correlation
. factor being 0.96. In view of the fact

J
//
V that the slope protection of Taipingchi
dam shifted 2-3 cm as a result of ice
thrust whiCh reached its highest value
2 V on February 15,1974 and considering the

I~~
strength of the frozen mass,it was found

IV
" 0
that the reSisting force was 35.7 tim,
wh~ch approximates the observed maximum
of 32.5 tim.
2 :1 4
Press u. re , COMputed. (Ir$ C11l-1 ) CONCLUDING REMARKS
Taip'"JcAi " [rl.ngltU. 0 ("f/he

laiyan9sli.enJ shllnyilewa1l
0 The pressure exerted by ice sheet over a
fig, 9, Relation between computed reservoir is governed mainly by the
and obseryed values of lce particular features of temperature
pressure
fluctuation and deformation of the
natural ice sheet as well as the condit

549
7able J
pde aicillress of Ice pre IS ure (A'j c",-<)
/?eserVOt"r
Year 11000tit Paj' ice s.feel(cwr oJseryeJ Co "'pu ted
I 20 66' 2.J 2.J
2 7J /.J / . 6'
Taipilljclti 19711 2 9 N 2.0 2.0
/" / 5 N 4 2 ~. 2

J 61 /. 2 / .J
62 / . j) /5
I
"7 62 /.} /. :J
9 62 /. I; 1.1
Z] 2< 76 2. / 25
2"J 76
N
I 2
09
1/
/. /
Etlol1j/tu 1977
4 71 12 / .2
9 71 /.0 /. /

2 /2
/7
"0
1
/. 2
/1
/ .2
/.3
/,j I j) #2 2 1 2 .1'
22 ,fo /.$ /.<
n ~2" <1 0 ..,.5 2. 5
/ J)77 12 / 42 o ,/ 07
2-J 4J /. z /. J
2 ..l l - ,32 /06 /./ 14<
Yinhe 19T! Z5 - 26 /0,/ / ./ /0
2 - 3 //0 /. :J /.0
3
/5 /7 //0 2.2 2. /
$-9 J'5 /.$ /2
2 / .J) -20 /02 /2 /:1
TQ.iyal/pltellJ 1979 .13 ~JJ /04 /. ] /. :I
J 7 -I /02 /.] /. J
/ 71 c.j) /.2
2 71 1.2 /. :J
J- 4 72 2 .4 2 . .,
7$ / 3 /$
sltaJIJlJeWall
~/.1 73 /.2 /-2
1975 2
/9 73 /.5 /.2
2~ 7J o. j} /.0
26 ?J /./ /.2
2J' 73 /.4 /3
ion of restraints during its expansion. stress has been laid on observ
ations and researches in these respects. Ment10n should also be made of
the applicability of the formula presented in this paper and the validity
of the relationship between ice temperature and ice pressure for the
conditions in the northeastern sections of our country. These may serve
as reference in engineering design,while more experimentation and
research on the properties and stress-strain behaviour of ice under
comolicated circumstances are to be carried out.
Comrades Liu X1hao,Hong wanshao,wang Liang,Li Y1ng,sun Yuliang,Dong
-----------------------
Yongshen,Tang Jiashou and others took part in the work.
550
STRAIN RATE DEPENDENT fRACTURE TOUGHNESS(KIC)

OF PURE ICE AND SEA ICE

N. Urabe Technical Research Center Kawasaki-~ity,


Kanagawa
A. Yoshitake Nippon Kokan K.K.
Japan

ABSTRACT

Critical stress intensity factor K of sea ice and laboratory-grown fresh


IC
water ice was measured by means of three-point bend tests on edge-notched
rectangular parallelepiped specimens (Scm X 10cm x 4Scm) as a function of strain
rate. In the sea ice case, a special loading apparatus was also set up on the
sea, and in-situ three-point bend type fracture toughness tests were carried out
on natural-thick ice specimens (20cm x 40cm x 170cm). Height of the supporting
rollers in the apparatus was adjusted so that the self weight of specimen was
cancelled out with respect to the buoyancy force.

The strain rate was varied up to the corresponding relative cruising speed
of floes and ice breaking ships. The fracture toughness . value decreased with
increasing in strain rate.

In order for K to be a material constant, the small scale yielding


IC
condition was examined on pure ice data, including fracture toughness values
which were available in the literature. The K values measured at a relative
rc
ly high strain rate (K greater than 10 kPa/m/sec) are the material constant,
I
since the small scale yielding condition was met.

For the sea ice case, the K values were shown to have possesed a close
IC
relationship with the microstructure of ice. It was shown that there was no
size effect of specimen, if the influence of microstructure on the K were taken
IC
into consideration. It was also confirmed that the fracture stress could be
found, when the fracture toughness value and the size of crack-like flaw were
once determined.

551
It was sugges t ed in the present stud y th a t t he fractur e t o ughness test
mi gh t be a stand ard testing meth od to ob tain the strength and to understand the
fracture pheno me na of ic e .

INTRODU CTION

Recent of f s hore expl o it a tion of hydroca rbon resources in th e Arctic Sea


a rea has con sidera bl y incr eDs~J and f i xed struc ture s or i ce br eak in g ships
,.. . erk ing in tne area are threat ened with menace of sea i ce in the wint er season.
In order to es timate the interacti o n forces betwee n i ce and struct ures, it is
primar il y imp or tant to unders tand prop er l y the fract ure phenomena o f sea i ce .

Mechanical properties of ic e have be e n s tudied exten s i vel y by means o f


c ompression tests, be nd ing t es t s and so on. Howeve r, it i s hard to under sta nd
the fracture phenomena of ice fro m thos e test resu lt s, s ince man y crack-like
flaws ex i s t in l ake ice and sea ic e . Therefore, line ar el astic fracture
mechan ic s, wh ich can deal with substances containing crac ks , sho uld be applied
to und ers tand the fra c ture phenomena of ice. Several results of measurements
f o r critical stress inten si t y factor, K ' are now avail a ble i n literatu re fo r
IC
pure ice [1, 2, 3, 41 , and Baa l ce [5, 6).

~n o rder for K to be a va li d va lue, th e most important re s t r i c ti on is that


1C
th e size of the pla s ticall y deformed vo l ume in the vic init y of a c r ack tip should
be s ma ll er than an y dimen s ion of the s pecimen. If thi s sma ll sca le yieldin g
condition is sat i sf ied, then K is a mater ia l constant, and it does revea l the
1C
proper fra cture initiation characteristics o f ic e.

Til e grain s i ze in i ce s pecimen is larg e if compared I.olith mc tollic materials,


thu s, th e ef fect of size o f s pecim e n on the K shou ld also be examined.
rc
In thi s paper, r esu lts of strain rate dep endent K va lu es measur ed on pure
IC
i c e and sea i ce a re reported, and the possi bilit y i s examined for the fracture
toughness t est to be standa r d t c wting method t o su r vey the fra c t ure phenomena
of i ce.

552
FRACTURE TOUGHNESS OF PURE ICE

Experimental Procedures

Pure ice sheets were grown in a \.Jater tank placed inside a cold room .
The temperature of room air was controlled at around -20C, and an ice sh~et of
13cm to 15cm thickness was formed in three or four days. The surface of the
water was seeded by spraying with pre-cool ed "ater so that an ic e sheet
containing fine columnar grains (average diameter being about Smm) t.;a s al\.Jays
grown.

Specimens sized 5cm x IOcm x 45cm we re prepared from the columnar grained
portion of the ice sheet by means of a band saw. A saw-cut edge-notch was
introduced at the middle of the 5cm x 45cm s urface of the s pecimen. Then the
notch root formed by the saw was reshaped by a razor blade in order to increase
the notch acuity. 11<0 kinds of specimens were cut: one so that the direction
of the long axis of the columnar grains was normal to the lOcm x 45cm surface,
and the other so that the long axis of the columnar grains was normal to
5cm x 45cm surface.

Three-point bending type fracture toughness tests were then carried out on
a universal tester placed in the cold room. Fig. 1 shows the experimental set-up.
Both load versus time, and load versus load-paint-displacement curves were
recorded graphically until the fra c tu r e was initiated. All the sp ec imen prepara
tion and fracture toughness tests were performed at -20 0 e.

Fig. 1 Specimen geometry and experimental set-up


for fracture toughness t es t of ice.

553
The critical stress intensity factor was calcula t ed from Equation (1),
derived by Brown and Sra wley [71.

(1)

where P f is the f r actu re load, S is the loading span l engt h, a is the notch
l ength , B is the spe c i men thickness, and W is the spec i men width .

Test Results and Discussion

The K values of pur e ice tested at - 20C ar e plotted against stress


rc
intensity factor.rate Kr
r
(K
= dKr/dt ; t is time) in Fig. 2. The Kr c decreased
wit h increasing K , and no dependenc e of K on notch plane orien tat io ns with
r rc
respec t to the direction o f the long axis of co lumnar grains was observed .

The K values of pure i ce have been mea su red by Liu and Miller [21,
rc
Goodman [31 , and Hamza and Mug gerid ge [41. They, however, used different
specimen geometries, and loading conditions . Those conditions are summariz ed in
Fig . 3. The Krc ' s , which were obtained in the -20C test on the spec i mens
containing average co lumnar grain s iz es of 5mm 10mm , are plotted in Fig. 2 as
a function of Kr with the present test results.

400
0 L1u & Mlller(2)

't~"
300 A Goodman (3)
1.0
~

V Hamza & Muggerldge(')

0.8 !!'I
~ 0,. Present Study

It, }t~
E
200 (. Klc converted from lid
<II
0.6 E
Q.
.:c ....DI
.:c
~
:.: 0.4 ~
:.:
100
90
80
70
60 y~ 0.2
50 3
10" 10 10' 10 10' 10 5 10
8

Fig . 2 Rel ationship between cri t ica l stress int ensity factor Kr and Kr
for pure ice of average g ra in si ze being Smm - 10mm , tes~ed at -20C .

554
Spec.Geometry (unit: mm) GrainSize KI(kPa"'"l/sec) Authors

E11~
P
/'
.... Growth
't' Direction
-5mm 10' - 10 3 L1u & Miller(')

g
PI2 P/2

Uo n 5-10mm 10' - 10
3 Goodman(3)
135 ;
P/2 PI2

~l
P/2 P/2
n -8mm 10 - 10'
Hamza &
Muggurldge(')

QJ:~
P/2 P/2
-5mm 10' - 10' Present Study

Fig . 3 Summary of fracture toughness test s for pure ice.

The data plots show the mean value of the K with the spreads of data. As can
rc
be seen in Fig. 2, K va lu es seem t o have good correlatio n wi t h Kr on a lo g- l og
rc
c hart , notwit hs tanding the different tes tin g conditions .

The correla ti ons between 10g(K ) and 10g(K ) were also exam ined at o the r
rc r
testing temperatures (-4"C to -4 0"C) . rf temperature T and Kr are given in the
uni t of degree Celsius and kPa/m/sec, respectively, th e K ca n be es timated by
rc
Equa tion (2).

. -( 0.08 10gT + O.OJ)


(155 10gT + 61)K (2)
r

For the K t o be the valid va lu e , in ot her words. not t o vio l a te the


rc
concept of linear e l astic frac t ur e mechanics, the plastic zone size formed near
the notch tip shou ld be smaller t han 5% of the notch length. specimen thickness
or specimen width, which ever is smaller [8]. As ice has a high homologous

555
temperature, creep deformation is suspected t o take place in the vicinity of the
crack tip prior to the onset of fracture. Riedel and Rice [9) have recen tly
proposed an equation to obtain the creep zone size for elastic-nonlinear-viscous
material. For the material which obeys a Nor ton type creep la w of the form
n
: Aa , where i s creep rate, a is stress level, and A and n are material
constants, the size of the creep zone rc is given by Equation (3).
2
K 2
. IC
(3)
21TE2

where E is Young's modulus, t i s time, an is amplitude fa c tor, and Fc(e) is an


angular function. According to Weertman's theory [10), n is equal to 3 for pure
ice. Substituting the other numeri ca l values at - 20C into Equation (3), as
done by Goodman [11), Equation (4 ) is finally derived.

(4 )

The units of rc are rom, if K and KI are in kPa/; and kPa!;/sec, respectively.
IC
Combining Equation (2) and Equation (4), the creep zone sizes for pure ice at
-20C were obtained, and are given in Table 1. From Table 1, the KIC ' S which
we re obtained at KI greater than 10 kpa!;/sec are valid, and are useful
parameters to desc r ibe the fracture phenomena of ice.

Table 1. Creep zone s ize for pur e ice tested at -20C at various K .
I

KI (k Pa,r;:nIsec)
rc (mm)

The test results for KI being les s than 10 kpa!;/sec were reexamined by the
J-int egral concept (12), which can be applied even at a large scale yielding
condition. J values were calculated according to Equation (5 ) .
IC

2U
J IC = 8(W-a) ( 5)

where U is the area under the load versus l oad-point-displacement curve. And
K values were converted from J value (K = IJ E! (1-V2 ); v is Poisson's
rc IC IC IC
ratio ). This value was plotted in Fig. 2 using a solid mark, and the K at the
rc
lowest KI shows almost the same value as the K obtained previously.
IC

556
The sp ec imen geometries shown in Fig. 3 wer e prepared in conformity with
the method described in ASTM E399 [81, which is specified for metallic materials.
Howev er , the grain size of ice is quite lar ge with res pect to the standard speci
men size given in the code. The grain size for a specimen of metallic mat eria l
is so small that the specimen can be considered a homog eneo us substance.
Vittoratos and Kry [131 have shown that there is a large effect of the size of the
specimen on the compre ssive strength of fresh ice. Therefore, the effect of
specimen size on the fracture toughness test should be examined. These tests
were performed only on sea ice. since it is eas y t o get a heavy thick ice
specimen from a naturally grown ice sheet.

FRACTURE TOUGHNESS OF SEA ICE

Experimental Procedures

Large Scale Tests

Fracture toughne ss tes ting of sea ice has been performed since 1979 at
"Saroma u lagoon, which is located at the most northern island in Japan. The
water is sea water (since the lagoon has an opening facing the Sea of Okhotsk),
and it begins to freeze about mid-December and thaws around the end of April.
The tests were carried ou t from the beginning of February until mid-March. \,hi le
th e temperature of the sea water was fairly constant a t about -2C, the
atmospheric temperatur e ,,,as variable. However, the test s were performed at th ose
I
times when the atmospheri c temperature \.,ras about _2e.

Fig. 4 shows th e load ing set-up of the three-p oint bending type fracture
toughne ss test, which was constructed on the sea ice sheet. The height of
s upport ing rollers, separated by 160 cm in the loading fixture , was adjusted so
that the ice specimen .!as kept as i f it were floating in the sea. Therefore,
the self weight of specimen was equal to the buo yancy f orce . The snow ic e was
removed , and large scale sp ecimens sized 20cm x 40cm x l70cm were prepared so
that the long axis of columnar grains was perpendicula r t o the 20cm x l70cm
face of the specimen.

A salol-cut edge-notch was introduced at the middle of the specimen surfac es


in three ways; normal t o the t o p, to the bott om , and to the side surfaces.
The notch root was r es haped by a razor blade. The notch depth was kept less
than lOcm so that the notc h tip was s ituated a t the proper position with respect
to th e desired ice structure. Both load versus time and load versus load-point
displacement curves were recorded on recording devices placed in a warm shed.

557
Actuater
Loadcell~ Steel Frame
-;::j AMPLIFIER J
/ I
iDATA RECORDER i

~
~ '<"~--'l---'
y XY RECORDER J
--- Lf Xt RECORDER I
L S:160cm I SeaWater

Fig. 4 Specimen geome try and experimental set-up


for in-situ fr ac ture t oughnes s test of sea ice.

Small Scale Tests

Small sized specimens having dimension s of Scm x IDe m x 4 5cm were cut from
the to p portion and from the bott om portion of the sea ice sheet. The long axis
direction of the co lumnar grains was alway s taken to be normal to the 5cm x 45cm
surfaces. The frac ture toughness te s t s we re carried out in the s ame way as the
pure ice tests.

Test Results and Discussion

K values were calculated by Equation (1). The results for the large scale
IC
and the small scale te s ts were plotted as a function o f KI in Fig. 5 and Fig. 6,
respectivel y .

It is impo ssi ble to check the small scale yielding condition for K '
IC
similar to the case of pur e ice, since creep data for rrSaraman sea ice is not now
available. But careful inspection ahvays showed a flat and crys tallographically
fractured surface. An abrupt load drop within the ela s tic portion in the load
ver su s load-paint-displa cement cur ve was also observed at the instance of
fracture. It lVas also co nfirmed that ~ressure-melting and related phenomena did
not take place at the loading position nor at the supporting roller positions
of the speci men. These re s ults suggest that the concepts of linear ela s tic
fracture mechanics were properly applied to the estimation of the fracture
phenomena of s ea ice.

558
100

~ 50 -o,------~O~---o~o~

Fig. 5 Relationship between K an~ Kr for large scale sea ice specimen
rc
with various loading d,rect,ons. (T = _2C)

Sea Ice

r<'~~"l:::: :
,..., 100

~III
a.
~
u
o 0 00

~ 50
_0 ;:P-O--~~
"b0 0 0 0 0
---08~0 o 0
00 0 _____ DO
0
0
0
0 __

O~~~-W~L---~-L~~~~__-L~~-Lww~
10 10 2

1<1 CkPam/sec.)

Fig. 6 Relationship between K C and Kr


for small scale sea ice specimen,
cut out from the top p6rtion and from the bottom portion of
the ice sheet. (T = _2 C)

559
Fig. 5 indica te s that the K va lu es are largely influenced by both the
1C
init ial notch-plane ori entation and the notch tip location. Th e K values whose
1C
notch tips were located in the bot tom portion of the sea ice were greater than
those values for the upper side. When the initial not ch plane is orientated
normal to the side surface , the K va lue s show intermediate values .
1C

From linear elastic fracture mechanics, the relation between the fracture
toughness value , the crit i cal fracture stress ocr' and the flaw size a is give n
by Equation (6).

(6 )

The flaw sizes at the top portion and the bottom portio n of the sea ice sheet
were calculated, according to Equation (6), using the K values and oc r val ues
rc
which had been obtained from the three-point bending tests on unnotched specimens.
The ca lcula ted flaw sizes were abou t 2.5mm and 1.9cm for the top and the bottom
por ti on of the ice she et, respectively . Moreover, th e calculated fla w sizes were
shown to be constant without regard to th e K .
1

SEA WATER

Fig . 7 Typical micrographic structures


of IISaroma" sea i ce.

560
Fig. 7 shows typical structures of "Saroma lT sea ice. The average diameter
of the subgrains, which have the same brine c e ll orientation ~",ith horizontal
section, changes from 4mm at the top portion of the columnar ice to 2.5cm at the
bottom portion, respectively. These locations correspond to the notch tip
positions for the top and the bottom portions respectively. The flaw sizes of
the sea ice, determined from the fracture toughnes s values. correspond well to
the average diameter of the subgrain size. This result suggests that if the
flaw size is determined by subgrain size, the fracture stress could be prediCted
by fracture mechanics.

As can be seen in Fig. 6, the K values of the small sized specimen Cut
IC
out form the top portion of the sea ice sheet show almost the same values as
those of the ~arge However, the K
scale specimen notched at the top surface.
IC
values of the specimen cut out from the bottom portion of the sheet show a
discrepancy with those of the large specimen notched at the bottom surface.

For the sea ice case, brine cells act as flaws [6]. Therefore, the notch
length becomes longer when the subgrain size is large enough if compared with
the original saw-cut notch length. In this study, the notch length was about 8 cm
for the large scale tests, and about 2cm for the small scale tests. Therefore,
the K values were strongly influenced by this effect for the small scale
Ie
specimen cut out from the bottom portion of the ice sheet, since the subgrain
size was as large as 2.5cm. Thus, the fracture toughnes s value should be
corrected according to equation (7), taking into consideration the subgrain sizes.

ofii3e f (a /\,) K
12+2.5 f(0.45) K
(7)
(JV"1ia f (a/W) Ie 12 f(0.2) IC

where a e is corrected notch length, and f( ) is the function given by Equation (1)
in square bracket.

The correction factor in Equation (7) is 1.7, and when applied to the K
Ie
value obtained on the small scale tests at the bottom portion of the ice sheet,
it yields almost the same value of K obtained for the large scale tests.
Ie

The correction factor is 1.06 for the small scale specimens cut from the
top portion of the ice sheet.

561
CONCLUSION

Fracture toughness measurements were carried out on pure ice and sea ice.
The main results obtained are as follows,

(1) The linear elastic fracture mechanics concept is shown to be applicable if


tests are carried out on the standard specimen recommended by the ASTM
E399 code, at KI being grea ter than 10 kPa/ID1 sec.

(2) For the sea ice case, the K value is related to the subgrain size, and
IC
no si ze effect cf fracture toughness is observed if this fact is taken into
consideration.

(3) The fracture toughness test might be a standard testing method to obtain
the ice strength because the critical fracture stress can be obtained from
the fracture toughness value.

ACKNOWLEDGEMENTS

The authors are very grateful to Messrs. H. Takeuchi, Y. Kiuchi and


T. Miura for their eager technical assistance during the experiments.

REFERENCES

[1] D. J . Goodman and D. Tabor; Fracture Toughness of Ice: A Preliminary


Account of Some New Experiments, Journal of Glaciology, 21 (1978) p65l

[2] H. W. Liu and K. J. Miller; Fracture Toughness of Fresh-Water Ice, Journal


of Glaciology, 22 (1979) p135

[3] D. J. Goodman; Critical Stress Intensity Factor (K ) Measurements at High


IC
Loading Rates for Polycrystalline Ice, Physics and Chemistry of Ice, ed. by
P. Tryde (Springer-Verlag, 19 79) p129

[4J H. Harnza and D. B. Muggeridge; Plane Strain Fracture Toughness (K) of


IC
Fresh Water Ice, Proceedings of Port and Ocean Engineering under Arctic
Conditions (1979) p697

[5J K. D. Vaudrey; Ice Engineering - Study of Related Properties of Floating


Sea-Ice Sheets and Summary of Ela s tic and Viscoelastic Analyses, Technical
Report R860 (Civil Engineering Laboratory, U. S. Navy Construction Battalion
Center, 1977)

562
[6J N. Urabe, T. Iwasaki and A. Yoshitake; Fracture Toughness of Sea Ice, Cold

Regions Science and Technology, 3 (1980) p29

[7] W. F. Bro~~, Jr. and J. E. Srawley; Plane Strain Crack Toughness Testing

of High Strength Metallic Materials, ASTM STP 410 (ASTM, 1966)

[8] ASTM; Annual Book of ASTM Standards, Part 10, E399 (ASTM 1978)

[9J H. Riedel and J. R. Rice; Tensile Cracks in Creeping Solids, Fracture

Mechanics-Twelfth Conference, ASTM STP 700 (ASTM, 1980) pl12

[lOJ J. Weertman; The Eshelby-Shoeck Viscous Dislocation Damping Mechanism


Applied to the Steady-Creep of Ice, Ice and Snow (1963) p28

[llJ D. J . Goodman; Fracture To~ghness (K ) Measureme:~~2f~~ Polycrystalline


IC
Ice at High Strain Rates (K greater than 600 kNm s ) , Report on
I
Studies for Physics and Chemistry of Ice (Ministry of Education, Japan,
March 1980) p85

[12J J. R. Rice; A Path Independent Integral and The Approximate Analysis of


Strain Concentration by Notches and Cracks. Journal of Applied Mechanics,
35 (1968) p379

[13] E. S. Vittoratos and P . R. Kry; A Comparison of In-Situ and Laboratory


Uniaxial Fresh Water Ice Strength, Proceedings of Port and Ocean Engineering
under Arctic Conditions (1979) p683

563
"STRAIN RATE DEPENDENT FRACTURE TOUGHNESS (K ) OF
IC
PURE ICE AND SEA ICE" BY N. URABE AND A. YOSHITAKE

DISCUSS ION
By:
T. D. Ralston, Exxon Production Research Company, Houston, Texas, U. S. A.

The field tests were co nducted on ice beams that had an inhomogeneous crystal
structure through the ice thickness. This structure ,.ould lead to a non-uniform
stiffness distribution and hence a non-linear stress distribution. How should
one compute the fra ct ure tou ghness for notched beam tests in this ice ?

DISCUSSION BY T. D. RALSTON ON "STRAIN RATE DEPENDENT


FRACTURE TOUGHNESS (K ) OF PURE ICE AND SEA ICE"
IC
AUTHOR'S REPLY
By:
N. Urabe, Technical Research Center, Nippon Kokan K.K., Ka,,'asaki-city,
Kanagaw8, Japan

The stress field in the vicinity of crack tip under the ten s ile stress in
an isotropic material is given by the following equation,

where 0 is the stress compo nents in the polar coordinate, Kr is the stress
ij
intensity factor of mod e I and fI,(O) is an angular function.
1J
Paris and Sih (P. C. Paris and G. C. Sih, ASTM STP 381 (1965) pJO) have shown
that the above equation is valid even for an anisotropic material, since the
I
anisotropic parameters f o r elastic cons tants can be involved in f (8).
ij

We did not measure the stiffness variation through the i c e thickness, and
assumed it is constant through the thicknes s . But we can compute the fracture
toughness value of the anisotropic materials by means of Finite Element Method
using above equation.

Ice crystal is al so inhomogeneous, since the grain size is quite large.


Therefore, we must use large specimens in order to obtain les s scattered data.

564
SCALE EFFECTS IN CONTINUOUS
CRUSHING OF ICE

P. R. Kry RESEARCH DEPARTMENT CANADA


Esso Resources Canada Ltd.
Calgary, Alberta T2G 283

ABSTRACT
Continuous crushing of ice sheets, the steady and continued penetration
of an ice sheet by an indentor, has been studied at two scales with semi
cylindrical indentors in S2 ice sheets. Geometric similarity and strain rate
similarity were achieved by matching the ratios of indentor width to ice sheet
thickness and penetration rate to ice sheet thickness. The small scale tests
in a cold room used indentors of 0.12 m and 0.25 m diameter penetrating machined,
confined ice sheets 0.01-0.05 m thick. The bubble free ice had typical grain
diameters of about 1 mm, and was tested at various penetration rates and at
temperatures of _lOoC and _2 0 C. The large scale tests were performed on a
natural lake using a 1.2 m diameter indentor to penetrate 0.25 m thick ice sheets.
The typical grain size increased with depth through the thickness from 1 mm
to 5 mm. The ice sheets were all between _3C and OoC during the tests and
penetration rates spanned more than an order of magnitude. Standard strength
tests which were performed on ice grown in the cold room and on ice sampled
from the lake test sheets showed the two ice types were equally strong. Each
test sheet was penetrated for a distance equal to 5 - 20 ice sheet thicknesses.
In the large scale tests, the highest forces were generated when the
ice failed in a mode termed ductile flaking. In this mode, the ice failed
by the formation of one or more in-plane cleavage cracks which continued to
open as the indentor advanced. The ice pieces underwent a ductile deformation
until they eventually broke off forming flakes. Large flakes were 1/4 to 2/3 as
thick as the ice sheet and extended 2 - 3 ice sheet thicknesses away from
the indentor and 3 - thicknesses along the indentor. During ductile flaking,
the effective stress on the indentor was generally a continuous but

565
variable function of the penetration. Its median value was usually
continuously exceeded for penetration lengths equal to 1/2 - 1 ice sheet
thickness.
Four factors influence continuous crushing in the ductile flaking
failure mode: the indentor width (penetrating a constant ice thickness),
the scale (preserving geometric similarity), temperature of the ice, and
effective strain rate of the penetration. In general, the effective
stresses on a wide indentor are less variable than those on a narrow
indentor penetrating an ice sheet under the same conditions. There were
two manifestations of a scale effect. At the larger scale, ductile
flaking occurred over a wider range of penetration rates, and for similar
conditions, the effective stresses on the larger scale indentors were
significantly less than those on the smaller scale indentors. The
effects of temperature and strain rate are interrelated. As the penetration
rate decreases or the temperature increases, the ductile aspects of the
failure process become more important than the brittle aspects. In
ductile flaking, this leads to less variability in the effective stresses,
and generally an increase in the mean stress.

I NTRODUCTI ON
To explore for new hydrocarbon reserves in the Southern Beaufort
Sea, Esso Resources Canada Limited has used artificial iSlands as drilling
platforms in water depths up to 20 m (1), (2), (3). To safely operate
from such a platform , or any bottom founded facility in ice covered
waters, requires consideration of the effective stress which an ice
sheet can exert on the structure. The stress depends upon the failure
mode of the ice. The failure mode depends both on the structure's size
and geometry and on characteristics of the ice. For example, when ice
is very thin, it is susceptible to buckling . When an ice sheet moves
against a sloping sided structure, it will fail in flexure. However,
crushing will occur in the case of thick ice sheets which move against a
vertically sided structure or against a vertical edge of ice.
Around artificial islands, an observed late winter failure mode for
the ice sheet is continuous crushing. Sufficient small scale movement of
the ice sheet due to tidal effects or small scale horizontal motions
occurs to prevent the formation of a frozen bond between the iSland and
the ice sheet. However, the ice sheet is sufficiently static that it
can achieve thicknesses of 2 m. A steady motion past an island of such

566
an ice sheet leads to continuous c rushing, at least until suffi cent
uncleared ice rubble hds formed to change the failure mode to that against
an unconsolidated rubble pile (4).
Several previous investigators have addressed the crushing failure of
ice (5-11). Primarily these have related to relatively narrow struc tures
which have diameters less than five times the ice thickness. Hirayama
et al (8) performed the most thorough small scale tests on cont inuous
crushing and were th e first inve st igators to assign importance to in
plane cleavage cracks in the failure of an ice sheet moving relative to
an indentor. They also used probability distributions to represent
the data acquired in the continuous crushing experiments.
The principal problem in determining effective crushing stresses
during crush ing of a thick ice sheet past an artificial isl and arises
from the scale of the interaction. It is extremely costly, laborious,
and time consuming to perform experiments in ice thicknes ses more than a
quarter meter with indentors more than a meter wide. However, it is
only by relating tests of these larger scales to the masses of data at
smaller laboratory scales that an extrapolation to full scale Arctic
ice conditions can be made. Ther e is evidence that increased
scale decreases effective failure stresses in ice (12) and in coal (13).
Two extrapolations are requi red in general. One from thin ice to thick
ice, which is considered in this paper, and one from narrow structures
to very wide struct ures (14) .
Thi s stud y addre sses the failure mechanisms involved when an indentor
continuousl y penetrates and crushes an ice sheet, the associated loads,
and the effect of geometric scale on the fracture mechani s ms and lo ads.
The approach involved both small scale tests in a cold room and large scale
tests on an ice covered lak e. The differ ence in sca le between the tests
i s approximately a fact or of 10. Ther e is approximately another factor
of 10 in terms of ice thickness between the lake tests and the Southern
Beaufort conditions. Efforts were made to ensure the ice tested in the cold
room was mechanically the same as ice tested at the larg er scal e. Penetration
rate s were varied over a wide range in both cases to ensure appropriate
duplication of failure modes at the two scales.
The penetration rate determines the strain rate characterizing the
ice failure (IS). However, there i s no agreed upon relationship between
penetrati on rate and strain rate. Hirayama et al (8) pragmatically used curve

567
f i rt- ing t ~du d (lu e' t o i de.ntify and def i. ne s Lrain r lle in t erms o f til i c kll Q:S ~ , JLl
m t ~ r . :I nd v i' l tlL:. i lv to il b i t r a t \,.' pUWt:: r s . How veT , [ll ~ r e .u e U :H "elt"l in i e s in

eXL r apol.a ting t ili ; r e l.l L j UI1 t o Arc. t i c () ndir jni1 ~ . Th t \.) l e t i ,,..t I p la up st r" i ll o r
p La ll L' :; t r e ~s ;1Il:d.yses of tilL' icL' shc t: inuL'llt.ltioll pru u L0 J!J , whjeh are most

amenable to the froze n-in co ndi d on, sugges t the diame t er i s the a ppr o
priate linCHr dimension to noncii lllcnsiona1ize pen e tr a tion rate ( 16) .
Howev er , thi s i s by no means clear in the case "here the failure mechanism
is gov er ned by cleavage cracks as noted by HiraY;lIl1a et a l (8) in the
continuous crus hin g mod e . For test result s to be useful at larger
scales the question of the definition of s t ra in ra t e must be satisfactorily
reSOlv ed .

TEST PROCEVUR ES
Fr es h SZ ice (17) "'as used for all ind e nt at i o n tests.

The bubbl e - free fine i ~ ( " in i ce used in the col d room t es ts "3S grown

~ r aj il s i ze "'as
0
in a t<lnk at Hn ni r t e mpera t ur e of _10 e. Horizontal
of the ord e r of 1.5 mm. Ice sheets test ed were ma chin ed t o LI ese tol era nces
(': 0.1 mm thickness, ~ Zmm width, edges square or parallel "'i thin 0.00 5
radions) .
Th e ice for th e large scale tests conducted on t he sur face of logle
Lake wa s grown in tes t ponds 8.5 m x 7.6 111 cut in th e na tural ice cover in
an area maintained free of snow during the ent i re' \;inter. The ponds "cre
prepar ed approxima t ely thrce weeks prior to t es tin g by removing the natural
ice (usually 0.6 m thick) a nd seeding the open area with. s now blower.
Sub seq ue nt thin sec ti ons ver ified the growth of SZ i ce "ith ~ rdin

sizes increas in g from lnun a t a depth of 5 LIn t o 4. 5 mm at ,j oL'( it h "f 2S


cm. The seeded l a yer was ge nera ll y l ess th an 3 <:m in till <:kn" as .
A pro gra m of s t and ard s trength tests '<;IS peero rmeJ on i ce from th e
co ld room and s amples from Eagle Lake test pond s . '(';11)1 8 J swnmari le s se le ct ed
results from thi s program. Overall th e program de mons t rated that within
the limit ations of test technique there is no difference in mechanical prop erties
bet"e e n Eagle Lake ice and cold room ice. Gener~lly l ess sca tter was observ ed
in cold room ice results consistent with be tL er co nt ro l in its growth.
The un co nf i ned comprcss lve strength of the i ce at _IOoe and at the transition
strain rate "US 7 MPa ~ 15 %. Strength de crca s cd to ha l f that value as th e
temperature incre aseJ to _zoe.
The co ld room ind e nt ati on t es t s "ere performed by inst,llling machined i ce
s hee t s 1 m "iJe and O.S m high within a co nstraint sys t em "hich prev e nt ed

568
Temperature Cold Room Eagle Lak e

a , a
-I -I
c ,IP a J,.s "IPa ks

5.4 0 .84 6 .4 0.6V

4. 8 0.86 5.4 0 . 75
TABLE 1 Uniaxial s trength
6. 7 0.12 8.7 0.20 a at the s train ra te of
co ld room and Eagle Like
-10 7.2 O. IS 6 .4 0.21
ice at th'O temp e rature s
loa ded in the plane of the
6.6 0.13 5.9 0. 13
original ice sheet
4.1 0.022 3.3 0.23

4.3 0.022 5.6 0.021

-2 3.6 0.19 3.8 0.22

3.5 0.21 3.2 0.51

2.6 0.16 6.2 0.23

bu ckling and simul a ted the elastic properties of a semi-infinite ice .heet .
The c onstraint system \y'as mounted on the lower pluten of J. testin g meJc hi.n e
and a semi-cylindrical indentor f as tened to the upper platen wa s forced into
the ice sheet at a constant rate. Penetration lengths of 25 em were possible
for indentors of 13 cm a nd 26 cm diameter. In each t es t continuous records
of indentor penetration and appli ed load \,'ere mad e . Temperature was co ntrolled
within O.soC.
The Eagle Lake tests were performed by using hydr auli c rams to push a
1. 2 m diameter semi-circular indentor for penetration lengths of 3.7 m.
A variable displacement pump provided controlled penetration rates for the
indentor. The continuous crushing loads from the test pond ice wer e reacted
over a large area plate to the thicker nat ural ice cover. The rams ana
indentor were supported by a gantry sitting on the natural ice cover.
Hydrauli c pre ss ures a t the rams were monitored to provide load data and
reel poteniometers connecting the indentor to the na tural ice sheet measured
penetration . Ice sheet temperatures were recorded by thermistors frozen
into the ice sheet and thicknesses "ere obtained through slots cut
with a chain saw.
Table gives the parameters used in the Eagle Lake and cold rOom
tests.

569
Indentor lee Sheet Temperature Penetr'a tion
Diameter thickness
c rat~I
m m (mm s )
0.013
Co Id Room 0.13 -10 0.002
0.025
0.26 -2 2
0.051

Eagle Lake 1.22 0.24-0.30 -2 0.2-10

TABLE 2 Parameters used in penetration tests.

FA ILURE MOVES
Three failure modes were observed in the cold room indentation tests.
These are schematically indicat'ed in Figure 1. The effective stress is
defined as a=F/Dh, where F is the total force on the indentor in the direction
of ice sheet motion, 0 the projected indentor width normal to the ice sheet
motion and h is the ice sheet thickness.
At very slow penetration rates, the failure mode observed has been
called Ductile Failure. The ice fails as a plastic material once an initial
penetration has occurred. The load is essentially independent of futher
penetration. The effective stress increases at smaller aspect ratios
(D/h),lower temperature, and higher penetration rates .
This failure mode can be well described by plastic failure theories of a
continuum. Large deformations of the ice sheet occur significant distances
from the indentor and the ice sheet behaves as a continuum.
At moderate indentation rates, the failure mode changes into one which
is called Ductile Flaking. Implicit in the name is the observation that
the failure of the ice sheet is controlled by the formation of very large
flakes which undergo significant deformations suggesting the occurrence
of ductile processes. Formation of cleavage cracks in the plane of the
ice sheet were observed to reduce the loads achieved during the indentation
process. Both at Eagle Lake and in the cold room large flakes were typically
observed to be 1/4 to 2/3 as thick as the ice sheet and to extend 2-3 sheet
thicknesses away from the indentor and 3-5 sheet thicknesses along the indentor ,
As penetration of the indentor continues, significant deformation of the indi
vidual flakes are observed as they are pushed apart and away from the indentor .
During ductile flaking, the effective stress on the indentor is generally
a continuous but variable function of the penetration.

570
c:J : tf'~/! ~c
DUCTILE
~

(slow) '"
~

'"
(TIME)
(PENETRATION)

DUCTILE
FLAKING
(middle)
D ",~!" ~ :...
FIGURE I Schematic representation of
three failure modes observed
in cold room te st s.

(TIME)
IPENETRATION)

BRITTLE
FLAKING
(fast)
CJ ~~kct ~

(TIME)
(PENETRATION)

At the high es t penetrat ion rates, a failur e mode termed Brittle Flaking
is observed. TIlis is characterized by flakes being formed as the indentor
penetrates the ice sheet. However, fl a ke formation is generally a one
step proc ess, that is, a crack initi ated at the indentor propagates to th e
ic e surface forming a small flake. The i ce near the indentor ass umes th e
for m of a ,,'edge. Flak es generally tend to be smaller during Brittle Flaking
than during Ductile Flaking and effective stresses on the indentor are
genera lly much l ess than those observed during Ductile Fl aki ng. The effective
stresses on the ind en tor are a discontinuous function of the penetration,
generally dropping near t~ zero after eac h flake forms and rising to an
arbitrary peak before the next flake forms.
In the cold room tests, Duct ile Flaking stresses were significantly
higher than Brittle Flakin g stresses. Despite the fact that the highest
stresses were generated during the Ductile Failure mode, interest was
concentrated on the Ductile Flaking failure mode, since this was the only
failure mode observed in the large scale lake test s, and appears to be
the failur e mode around artificial islands.
Ductile Flaking is illustrated by the photos in Figure 2. Both at
Eagle Lake and in the cold room in-plane cleavage cracks are evident along
with large flak es which have undergone extensive ductile deformation. In
Figure 2 the cold room ice sheet thickness was 2.5 em while the Eagl e Lake
iee sheet thicknes s was 27 em.

571
FIr.VIlE 2 Photograph s of section s
of ice which ha ve failed by Duc til e
Fl uki ng. The upper photo i s from
a cold room test w.ith S1i('et thi (r..
ness 2.5 cm. The lower photo is
from an Eagle Lake t es t pond wit h
sheet thickn es 5 27 cm.

RESULTS
Flaking occurred over a wider range of penetration rates in the [u g le
Lake tests than in the cold room test s. Fi gure 3 demonstrates this result .
Strain ra te has been arbitrarily defined as the penetration r0tc divided by
the ice shee t thickness. However, indentor di ame ter divided by ice she e t
thickness is the same for cold room test s and Eag le Lake tests, so that
most common definitions of strain r a te wou l d ge nera te the same t)'pe of
relationship.
Figure 3 illustrates the ductile f a ilure was not observed in l arge
scale te s t s a t rates an order of magnitude less than those for which
it was observed in the cold room. Ductile Flak ing , which leads to fa ilure
stresses generally less than Ductile Failure i s the favoured mode as the
ice sheet th ic kness increases. One mi ght speculate that this i.s reasonable
if the gene rat ion of large in-plane cleavuMu cracks is a flaw controlled
process. As ice thickness increases, the absolute number of flu\,-s would
be expected to increase and result in a higher probability of initiating
cleavage cracks and the Ductile Flaking failure mode.
The temperature of the ice in the cold room tests was colder than
that in the Eag le Lake tests . Therefore, a lt hough geome trical similari t.y

572
DUCTILE FLAKING OCCURS
XXXXX)(XXXX
BRITTLE FLAKING OCCURS
DUCTILE FAILURE OCCURS
........... UNKNOWN

COLD ROOM TESTS (-10" C) +++++++++++++ - - - - - - - - - --OOOOOOO(()()()(XXXXXXXXXXXXXXXX

EAGLE LAKE TESTS ++++++oOOOOOOOOOCoooa OOO::::>tXIOOOOOQ:x> ::;::c C:::l ::l XIOOOOOOOOO+++++++++

10- 4 10- 3 10- 2 10- 1


PENETRATION RATE (thickness 15ee .)

FIGURE 3 Occurence of failure modes in co ld ro om a nd Eagle Lake tests

exists in the tests , an exact equivalence does not. However, since increa si ng
temperatures generally enhance ductile defor mation processes , it would be
expected that cold room tests at telllp eratures equivalent to th e Eagle Lak e
test s would result in Du ctile Failures oc curring at high er penetration
rates than sho"'11 in Figure 3. The difference du e to scale would be en hanced.
The stress-penetration data for the Eagle Lake tests are the largest
scale data obtained in a controlled experiment for the secondary failure procesS.
To concisely present these r esults, the test results \,'ere group ed in four
classes by penetration rate. For each class a single stress penetration
curve was constructed by sequentially ordering the stress -penet ration dat a
fo r each portion of a test.
St res s-penetration traces for the four nominal penetration rate s are
pr es ented in Figure 4. The highest effective stress mea sured was 3. 5 MPa
at a rate of 0.001 s-l . A comparison of the top two traces in Figure 4
indicates some evidence for the occurrence of Brittle Fl a king at the
higher rates.
Figure 5 is an example of a stress-penetration r eco rd gen erated
during Ductil e Flaking fai lur e of an ice shee t in a co ld rOom test. The
penetration rate of the 12.5 em diameter i ndentor through the 1.25 em
thick ice sheet lVas 0 .01 thickness per se cond: nearly equi valent to the strain
rate 7 x 10- 3 sol graph in Figure 4 . The analogous nature of the stres s

573
EAGLE LAKE : STRAIN RATE 2.10- 1 EAGLE LAKE: STRAIN RAn ;;; 7. 10'1
4.0 4.0

S 3.0 S 3.0
T T
"SE "SE2 .0
S 2.0
S
M
M

'0 .. 36 40 44 48 52
PENETRATION (ICE THICKNESS)
56 60 64

EAGLE LAKE: STRAIN RATE 2 l! 10 ' !ACRE LAKE : ST"AIN "AT! 1 10'
4.0

S 3.0 '.0
T ,s
"E ,
R

~ 2.0 ~ 10

.'0
M
P

4 , ,.
r PENETRATION (ICE THIC KNESS)" "
PENETRATION (IC E TH ICK NESS,

FIGUR f: Continnous crushi ng stresses measured at l arge sca le . Re s ult s grou~ed


accordin g to pe ne tration rate divided by ice sh ee t thickness (7x lO- 4 s- 1 , 2xlO - 5 - 1 ,
7xIO- s- and 2XIO - 2s- l ). Total penetration distance normalized by ice sheet
3 I
thickness penetrated.

penetration curves and the observed ice behaviour illustrated in Figure


2 veri fies the occurrence of the same failure mode a t th e two geometri c
scales .
Table 3 is a st ati stical swrunary of the results of a ll tests in wh ich Ductile
Flaking was the observ ed fai lure mo de . The test con diti ons were ice she et

FIGURE 5 Example of continuous 4.0F==:;::==;:==;:C::O=L=D~"=O=O=M=T~E=S=T=25=r==;==:;=.:~

crushi ng stresses during Ductile

Flaking in Cold Room.

oL-~~ __ ~ __ ~~ __U-__ ~ ____ ~ __ ~ __ ~~

a 4 12 16
PENETRAT ION (ICE THICKNESS)

574
h
TC l CUndll i o ns
0 T i: "
,
h
,
Pe fl cl..-al ion
(J pk QO. 01 0 . 16
Stresses
0 . 50
,,
( ur ) ( ~t) (QC) (t II " ,: 10. . (t hi ck (th?~L Vlra) (.'tPa) (~\Pa) (IoIPil) (MPaJ (IoIPa)

---.-..--
Il C:t.SCIo IlC "IO 'ic s ) !l e s se e )

2. 57 U.I 11 -2 .. , 1. 10 4.0 4.0 ,. , 2.6 2.55 0.72

1. Sl U. 2S4 -2 !.l .H O. ijO 2.' 2.8 '-, 2.1 2. 04 O. lS

Z. 55 U 12 7 -IU 10.0 O. lO I7 3. 0 '.0 '-' 1.8 1. 79 0 . 55

2.57

2.49
U. J 27

O. 2S<4
-10

- 10
32 . D

l to. 2
0 . H7

0.78
"10 6.0

,..
3.9 , ..
5 .8 . I

'.1
2.6

2. '
2.76

2.32
1.20

0.70

1.27 0.254 - 10 10 H. S 0.37 4S 2.8 1.9 I.' I. 35 0.53

L. 27 0. 127 -10 10 19 .0 0. 117 20 5. 2 ' .5 2.8 2. 1 2.22 0.63

1.2() 0 . 127 -10 9.6 U.~3 . 6 . 6 , .8 3 .2 3.06 0. 83

.\0 . 2 \.22 -2 0.7 14.5 0. 6 4 II 3.0 2.6 I., U 1. 37 O. SO

29 .7 1.22 -2 lU.3 0.87 2.4 2.5 1.8 1.2 1. 18 0 . 59

26.9 1.22 -2 54.7 0.65 42 3 .5 '.3 I. ' 1.0 1. \3 0.74

23.6 1.22 -2 20 62.0 0.19 160 2. 5 2.0 1.1 0. 6 0 . 69 0.42

Table 3: Statistical swrunary of a ll Du cti Ie Fl aki ng failure stress data


grouped by te s t condi tions .

thi ckness h, indentor diameter 0, tes t temperature T, and penetration rate


divided by i ce sheet thickness i: The penetration dat a has been normali zed
h
by the ice sheet thickne ss: Eh is the total penetration cO.S is the average
penetration di stance for which the stress continuously exceeded the median
st re ss and N is the number of times the s tre ss exceeded the median stress in
the total penetrat ion distance. The s tre ss es are char act erized by the peak
st res s; a ' the s tres s exceeded for 1%, 16% and 50% of penetration,
pk
00.01, 00 . 16' 0 0 . 05 , the mean stress and its standard deviation as. a
Figure 6 is a plot of stress data from Table 3 as a function of
penetration r ate illustrating the effect of scale and temperatur e .
Three series of te sts are plotted: cold room te sts at _lOoe and _2 0 e
and Eagle Lake t es ts . For each series, three stress level s are plotted:
those levels exceeded for 50%, 16% and 1% of the penetration respectively.
4 1
For example, at a penetration rate of 7 x 10- s- , during the Eagle Lake
tests the stresses exceeded 1 .3 , 1.9 and 2.6 MPa, for 50%, 16 % and 1% of
the penetration respectively . The results show that at a larger sca le,
geometrically similar indentation tests lead to lower stress level s .
Difficulty in quantifying the scal e effect reduction arises since the
failur e mode changes at low penetration rates in the co ld room but not
at large sca le . Figure 6 also shows the reduction in stresses associated
~ith a tendency to Brittl e Fl ak ing at the highes t penetration r ate s both

575
8.0r-------------------------_
COLD ROOM TESTS (- 10 0 C)
@ @@ COLD ROOM TESTS (_2 0 C)
@ 0 0.01
0.16
E o l:J. 0 EAGLE LAKE TESTS (_2 0 C, 0 C)
@ l:J.
0 .50
@ 0
F
F
E 6.0
C
T
I
V
E
4.0
S
T
R
E
S
S2.0

M
P
a

OL-_____ ~ __________ ~ ___________L__________L___ ~

10 '3 3x10 ' 3 10 " 3x10-'


PENETRATION RATE (thickness / sec.)
rigure 6: Probablistic stress levels "s a function of penet .."tlun rate.

at Eagle Luk e and in the cold room. The limited dat a on the effect of
temr c r ~ ltu re i llu stra t e a n incr t.=i.I::i e in stress levels at warmer temperatures.
all other parameter s constant. However, a t th e warmer tcnlpcrature t here
..as no t end e ncy to Brittle Flak i ng as "as observed at th e col der tempera t ur e
du e to incr eased duc t ility of the ice . Th at is, the relative s tr e ss
levels are influ enced by a modifi cation i n the failur e mod e.
The effec t of ind entor ..idth on Du ctile Fl aki ng s tr esses is to
decrease both the variability and the mean Str ess . Thi s is demon s tr ated
0
in Tab Ie 3 t lnough com p" riso n of tes ts at _2 e and _100 on 2.5 cm t hi ck
s heets and t es t s at -lOoe on 1.2 cm t hick s he e t s .
In any statis tica l analyses an underlying assumption is mode that the
amoun t of dat a is sufficie nt to represent the popu l a tion sampled . The
pa ramet er N, the s t ress cycle number (Table 3), is the key indicator in
determinin g the sufficiency of the data. If th e stress "ere a si nusoidal
function of th e penetration, N "ould r epre sent th e nwnher of cycles " hic h
occurred in a given penet ration distance. For ductile flaking, l arger N

576
values shou ld correspond to more representativ e sa mple s. Results of statistical
analyses of the componen t test s which are combined to yie l d a single record
for each set of test conditions indicate that if the stress cycle number
is approximately 5, stress l evels are reproduced "ithin 30% by a second test ,
"hereas if N ~ 20, stre ss l eve l s are reproduced within 15% by a second
test. Since the mean duration of exceedru, ce of the median s tress is
0.5 - 1 ice thicknesses, at least 20 thicknesses of penetration are
required to defin e stress l evels within 15%. There is some doubt of the
usefulness of attempting to refine the accuracy further since natural
variability in ic e strength may be of simi liar order.

CONCLUSIONS
Within limitations of measurement and test t echnique, the ice at Eag le
Lake and the ice grown in the co ld room "ere mechani cally equi valent S2 ice.
Maximum uniaxi a l strength tested in a hori zonta l pla ne '..as 7 MPa at _lOoe
dropping to 3.5 MPa at _2 0 e.
Three failure mod es were identified for i ndentatio n of an i ce sheet :
Ductile Failure, Ductile Flaking and Brittle Flaking. Ice failing in th e
Ductile Failure mode appea rs to behave as a continuum wit h signi ficant
deformations occurring far from the indentor. Dl,ctile Flaking is characterized
by the formation of large flakes (2-5 ice sheet thicknes s es) generated by
cleavage cracks "hich prop agate in several steps. Brittle Flaking occurs
with no evidence of ductile behaviour; flake s form ed are small an d occur as
cracks form in a one step process. Ductile Failure OCCurS at the lowest
penetration rates; Brittle Flaking at the hi ghest.
Ductile Flaking OCCurs over a "ider range of strain rate at larger
scale. The highest fai lure s tress of 3.5 ~IPa observed at Eagle Lak e "as
associat ed "ith this failure mode. The range of pene tration rate s for
which Du ctile Flaking occurs in the cold rOom tests i s very restricted,
however thi s predominant failure mode at large scale "as dupli cated in
small scale l aboratory tests.
St r esses during Ductile Flaking at small scale were s i gnifi cantly
higher than Ductil e Flaking stresses at l arge sca le. Quantification is
made difficult by the difference in range of penetration rate for which
Ductile Flaking Occurred at the [1<0 scales .
No single strain rate definition adequ ately scales penetr a tion rate
for all failure modes observed in contin uou s crushi ng. It is sugge sted that
the ratio of penetration rate and ice sheet thickness is most appropri a te
for Ductile Flaking.

577
ACKNOWLEDGEMENT
Permission of Esso Resources Canada Limited to present this paper is
gratefully acknowledged. The project was undertaken on behalf of the Arctic
Petroleum Operators Association CAPOA project 106) and their support is
gratefully acknowledged. The assistance of many people was required to
conduct this program in particular the author is indebted to Mr. R. F. Lucente
and Mr. R. E. Hedley.

REFERENCES
1. GARRATT, D.H . and KRY, P.R. 1978, Construction of artificial islands
as Beaufort Sea drilling platforms, Journal of Canadian Petroleum
Technology, Apr.-June, pp. 73-79.
2. CROASDALE, K.R. and MARCELLUS, R.W. 1978. Ice and wave action on

artificial islands in the Beaufort Sea, Canadian Journal of Civil

Engineering, 5, pp. 98-113

3. KRY, P.R. 1979 Implications of structure width for design ice forces.
Physics and Mechanics of Ice. Editor P. Tryde, International Union
of Theoretical and Applied Mechanics Symposium, Copenhagen, Aug. 6-10
pg 179-193
4. KRY, P.R. 1980, Ice forces on wide structures. Canadian Geotechnical
Journal, Volume 17, No.1, P 97-113
5. KORZHAVIN, K.N., 1968. Development of Methods for Determining Ice

Pressure on Bridge Piers in the U.S.S.R. Trans. Novosibirsk, Inset

Railw. Transp., Transl. by U.S. Army Cold Regions Res. and Eng.

Lab., Hanover, N.H. Apr. 1972

6. AFANAS'EV, V.P. DOLGOPOLOV, Y.V., and SHVAISHTEIN, Z.I. 1971. Ice

Pressure on Individual Marine Structures. Ice Physics and Ice

engineering, ed. G.N. Yakolev, Leningrad U.S.S.R., Transl. by

Israel Program for Sci. Trans. Jerusalem, Isr.

7. NEVEL, D.E., PERHAM, R.E. and HOGUE, G.B. 1972. Ice Forces On

Vertical Piles. U.S. Army Cold Regions Research and Engineering

Laboratory, Report No. 77-10, IIp.

8. HlRAYA}~, K., SCHWARZ, J. and WU, H.C., 1974, An Investigation of

Ice Forces on Vertical Structures. Iowa Institute of Hydraulic

Research, University of Iowa, Report No. 158, pp. 153.

9. BLENKA.RN, K.A. 1970. Measurement and Analysis of Ice Forces on

Cook Inlet Structures. Offshore Technology Conference, Houston,

Texas, April 22-24. Volume II. paper number OTC1261, p. 365-380.

578
10. NEILL , C.R. 19 76. Dynamic Ice Forces on Pi ers and Piles. An
Assessment of Design Guide lines in the Light of Recent Research.
Canadian Journal of Civil Engineer ing . Volume 3, pp. 305-341
11 . MICHEL, B., and TOUSSAINT, N. 19 77. Mechanisms and theory of
indentation of ice plates. Journal of Glaciology, 19(81) pp. 285- 300
12. WEEKS, W.G. an d ASSUR A., 1969. Fracture of Lake and Sea Ice. U.S.
Arm y Cold Regions Rese arc h and Engineering Laboratory, Research
Report 269.
13. BIENIAlvSKI, Z. T., 1968. The Effect of Specimen Size on Compressive
Strength of Coal. International Journal of Rock Mechani cs and
Mineral Sciences. Vol 5, pp. 325-335
14. KRY, P.R. 1978. A statistical predi c tion of effe cti ve ice crushing
stresses on wide structures. Proceedings, International Association
for Hydr aulic Research, Symposium on Ice Problems, Lul ea, Swede n,
Part I , pp. 33-47
15. ~IICHEL B. 19 78. The Strength of Polycrystalline ice. Canadian
Journal of Civil Engineering. Vol. 5 No.3, pp . 285 -300
16. REINICKE, K.M., and REMER, R. 1978. A procedure for the determination
of ice forces-illustrated for polycrystalline ice . Proceedings,
International Association for Hydraulic Res earc h, Symposium on Ice
Problems, Lulea, Sweden, Part I, pp. 217-238
17. ~I1CHEL, B. and RAl'1SEIER, R.O. 1971. Classification of River and Lake
Ice. Canadian Geotechnical Journal, Vol. 8, No. 36, pp. 36-45

579
Discussion of "Scale Effects in Continuous Crushing of Ice"

C. R. Neill

I have 2 points regarding scaling:

1. The author has assumed that time scale is 1/ 1,


that is, in interpreting similar penetration
rates, he has expressed these in ice thicknesses/
second. But 1 sec. model may not represent 1 sec.
prototype.

2. If in fact the d ifference in results is a scale


effect, it ought to be possible to recognize a
dimensionless variab le which was not similar at
model and at prototype scales. Can the author
suggest what this dimensionless variable might
have been?

Reply:

Although the time scales may indeed differ between


the two se ts of tests reported, the experiments were
conducted over a sufficiently wide range of penetration
rates that any appropriate shifts due to differences in
time scale are included in the data .

A variable which ha s not been taken into account


in the scaling is the absolute number of flaws in a
given length of ice crystal. As the ice sheet thick
ness increases, this number should increase and perhaps
accounts for the reduced indentation loads observed.

580
CO~'IPARATIVE STUDY OF ICE STRENGTH DATA

N.K. Sinha Divi s ion of Buil d ing Research Canada


Re searc h Offic er National Res earch rouncil of Canada
Ottawa, Ont ario) Canada

A r eview of s train r at e se nsi ti vi ty of uni axial compressive yield


st rength of polycrystalline ic e inJi cat es th 3t th e r esu lt s obt ai ned by
different inv es ti gator s reflect the s tiffn es s of th e corr esp ond in g t es t
s ystems. It is s hown that th e softer th e system the sof t er i s t he
e ff ec ti ve response of th e ma teri al. Us i ng th e con cept of failur e modu
lu s , it is sho wn that brittle-like splitting failures do not indicat e
pure e la s tic l oadi ng conditions, and that a ppare nt duc tile-to-brittle
tra nsiti on depends on th e resp onse of t he t e s t s ys tem. Thu s , muc h of
the sca tt er in available results for the S3 me t ype of i ce i s at tr i butable
t o differences in th e response of t he differ en t t es t sys t ems .

581
Reported va lues of the strength of ice are rath er sca ttered and often difficult
to interpret . Lack of detailed information on the conditions under wh ic h the measure
ment s were made i s one of the prime difficultie s in int erpreting strength dat a. It is
the purpose of this paper to pre se nt results of t es ts on one type of ice at one
temperature obtained by different investigator s using differ ent test sys tems . The ic e
chosen for this comparison is one of the most common, columnar-grained 5-2 type. It
was loaded perpendicular to the column s and tested at -10C. This choice was dict a ted
primaril y by the number of readily available results from Laval Univer si ty [1-5] and
from the Division of Building Research, National Re sea rch Council of Canada [6-9].
Re s ult s have also been obtained on th e dep endence of upper yield stress on trul y
constant s train rate [10] usin g a closed-loop test sy st e m.
Conventional t es t machines capable o f deli ver in g a constant or near-constant
ac tuator or cross-head di sp la ceme nt rate have been used in a ll test s until very
recently. Cylindrical or rectangular spec ime ns were u ~e d ; and s trengths were plotted
as functions of nominal st ra in rat e , En ~ x/, \-ihere x i s th e a c tuator or cro ss -head
di'rl~cement rate and t is the specimen length .
For yield type of failure the upper yield stress was observed to depend on
nominal strain rate according to t n =P of P , where P and p are constants. It is,
ho~crcr, preferable t o express the above power law in a normalized, dimensionally
balanced form by

(I)

where tl is the unit or reference strain rate} 01 is the unit or reference s tre ss, and
M and m are const ants.
This form of the relation allows a ready comparison of the rate sensitivity of
strength to the dependence of viscous flow rate on st re ss in pure uniaxial creep
us ua Ily gi ven, and aga i n prese n ted in a norma I i zed form, as

wh ere A and c< are constants.


A f 001 r
Th e term "viscous" will be use d in this article to
(2)

describe the non-recoverable (pe rmanent ) part of the tot a l deformation, discu ss ed
elsewhere in detail [II,12l.
Probl ems
Creep t es ts are relati ve ly simple and can be performed \;ithout muc h diffi cu lt y
using a dead load sy stem. Th e s tre ss level can therefore be maintained reasonably
constant so long as the load is kept constant and the specimen is not allowed t o
de form excessively, say not more than a fe\; per cent. Similar comments cannot be made

582

I
for the so -called con s tant strain rnte te s ts. It has he en clearly demon s trated [9]
thnt conventional tcst m<lchinc s arc not capable of m~intainillg a const:1l1t strajn rate
during :1 tcst c;lrr icd out il t a con s tant c ro ss -head rat c. Observ a tions Ilave indicated
that a conventional test machine i s able to impose the s pecificd nomina l deformation
rJ.tc, En' o n the s pecimen only h'hen the specimen offers no more inc rea se in resistance
with furthcr incre ase in deformation. It has been s hOlm that s train rate at yield is
e quivalent to nominal s train rate. Similar observHtions were made by Drouin [I].
This generalization cannot, however, be appl ied to brittle-lik e failure s in which the
peak s train rate could be a fraction of E [9). The use of E a s a n independent
n n
variable can ther e fore be j usti fied on a limited basi s for yield t ype of fa il ure only.
Ihe t es t condition, however, cannot be constru ed as constant strain rate.
'['h e difficulty in making a comparative stud y of lce strength from constant cross
head rate tests and constant load creep r es ults hecomes evident. The difficulties
have been re cognized by many investigators, but the s ubject and possible co nseq uen c e s
hav e never been openly discussed in glaciological lit era ture. The main reason for
thi s 1i es ) perh aps , in the o bserved fac t that the stress exponents p (or mJ and a
were found by different ob se rvers to be c lose to each other and equal to about 3 .
Consequently, ther e never seemed to be any requirement to exa mine the differenc es , if
any, in the values of I' (or M) ob tained Dy different investigators, nor has ther e been
a comparative study of P (or M) a nd A.

Stren gth Result s


Mu guruma [6) examin ed the uniaxial yield s trength of columnar-grained S-2 ice of
a ve ra ge grain di amete r of 2 to 5 mm loaded perpen dicula r to the column axis . The "ork
was carr ied out a t the DBR/NR C laboratory using a low ca pacity (0.02 ~IN) Hounsfi e ld
tensometer. Specimens (0.5 x 2 x 2.5 em) proved to be rathe r inadequate [13] sin ce
the number of grains across the width was not sufficient. Muguruma dis cussed hi s
results in t erms of a power law similar to that of equation (I). He obtained a stress
exponent of 3.0 in his t ests, but did not give a numeri ca l value fo r the coefficient.
2
Figure 8 of Muguruma 's pape r [6) giv e s 2 .4 MN'm- for th e mean strength of six
5
s pecimens of average grain diameter of 4 mm, test ed at -10C and En 2 .1 x 10- s -1.
From the results of his experime nt, equation (I) ca n be expr e ssed as:

3.0
tv1uguTuma
[6) 1.5 2 x 10-
6
[ :~ 1 (3)

Gold and Krausz [7) inve s tigated th e compressive s trength of vari ous t ypes of
natural ri ver ice, including S-2 type at -10C at the S3 me DBR /NRC labor a t ory . The y
used a larger ca pacit y (0.05 MN) Ivykeham Farrance, ~10del 57, soil-testing machine.
583
L

Their n.'sults \" hel\ cO ll vC'r tl'd t o the form o r C'qu ~ lti O!l (I ) :I re gi vt'll hy

1 . ()
11 [ Cold :1 1["71 1Kr,," , c J = I.(l(, x lO - 7 [ -o r 1 (~ )
" I J
J-. X ~lIl1i l1:1tioJ\ of till' t'xpC'ri 1T1l'l1t~ 1I pr(H.: <.:'dul"c:s d i ~( l oscd th :lt Cold :lIlU Kr:ll L,';: IIsed

III 2S ern ilnd il x H x I~) em. The y llll'nt i oneu

t"h:lt the sm: lll cr spcl'i mv ns h:ld to b<.' IIs ed ~l t the.' hi ~l.h(' r r:!l l's uf .". t r:lin to l.d'I.'I) the

10:h.l s Illthill the l' ;ll';lCi t)' o f th e tC'Sti ll ~ 1Il;lchi lll', hu t th e': r~lt<..' :It \... hic h th e cli:lngc 1</;1:-;

rll cX;l liii llin g the str:llil r; lt c s Cllsiti v ity .J:' :) I; lll C'-s tr~lin c omprcs"ivl' ~. lrl' llgth uf

l:lb or~ltoJ' )' -mad C' S -_~ ice of ;rvcr:1 gc gr:lill di ;Hllcter of ;!I)()\ ll :; mnr, FrC'<ierking I '~ I .1 1; 0
performed t est s :It -I noc tin der IIni:lxi;1 1 IOilC:in .: l ' \ ~lhliti on:-; (s]lC'l'imclls S x lO X ). , e m)

It P BI~ / He 011 a 0.1 ]\IN c:1p:rcity fnstroll TTDI'--I-L test sys te m. C01 1\'t'f.,-; jO Jl or : rl. ~k,~i IlV's

results to the f o rm of cqu:ltiull l l) ~jvcs

r: 11 [ Frc d rh illg ) = .' "0


181 _..L (5 )

ncpeJlJcll"'~ of comrress lv C' ~t rclI;~th uf column ;If'~ '.r;l in c J s -.? icc on s traln r ate at

-IO oe for ;1 loauin g (Ol lUiti oll of Coltst ;ln t rate of \.r () <~' h ead disp l :lc cmcnt ~oJas in ves ti

g"tco extcII s i vc l y by Sinha 1.11 at IlBI(/ NRC I"i ll.'! the ~ 'l!l1 C rnstron TTfWI - L te s t machi ne ,
:-:: p n c i.l11e ll g eometry, .1nJ t ype of lCC [IS W~ I S u sed hy frcucrk i ng jR J. rhc d e pe ndenc e of
til e uppe r yic LJ ...; tr e s~ 0 11 11 0 lllilLil -;\ J':lin r:ltc W:l S

(6)

Sinha lllli h;JS abo perfo rmed ~trc n il t h t ests on S .~ ice at - 10C using a 1.0 ~I N
C;t p :lc i t y, closed- l o o p , servo-hyurnu l ic :-, )' s tClTJ at EX XON Pl'ouucl i on Hcs ca rch L;Jhor atory
in 1I0l1 ~ on (spec imens S x 10 x .2S COl ;ITld a\'eral~c ~ r .1j ll cJi~l m~ters ~1 t o 5 mm). For

~ olHii i o n s of trul y COJ1SClnt strain r ;lI P , th e- d \..'lw lld c JlcC of the yie l J stress on str3 in
l".Jtc \yas gi ven by
, -, .91l
~I .~ :;~ \1i1 j
l) f !

r. ~ J
[ = I . Xl (7)

('v prnl ill r<.'s tig~lti o ns of till.... ~t n"' ll ~ h of pul~Tr)'sLl llj n<.' icc h av e he en c3rrje J
out i n tire :,v Ll h,,,"to r )":lt L;rv:.l Un i v e rsity II - I , l 'II , :l IlO t he r c "rl -; c lII"r i 1eJ by
~I ichd [51 . The se iflll' s t i. It i ons '.,'e r e c arrieJ (J il t nn 3 lI'y kcha nl I' "rra n ce ,Iodel T57B
h'ith rn:l x imuJ11 C;lP ~ I l' ity (). d ~ ~IN l _l l , the ~ :trT1c ma k e :!IIU mn le i as th ;lt :1cquircd by DBR /~ RC

584
in 1~ 1 S.s ;lnd lI~ l' d h: " Go ld :11\\..1 I\r :llls : [- I, 1:01" \' ( ) l ll!llI1.I !' -~', l ': lillcd S-.:.: i ...'(' o f ;1\'Cl': l g l'
gl':li.n di :lIlll't cr o f S I1l1n ;It _\(1oc:, in ;IJl'd p (,l'Pl~ l h.l il'lI L I I' t() rill' \'O II II1l 11 ~ , rh i.' dcpl'lh.l en l'l'

11 i'li ch el
(X)
i:
I IS I

Re lati ve l y large spe cim c ns hcre u ::-ed ;lr L;l \'~ll ) :l~ ",;llOhl1 i n T:lb ll' I" 11r ()u i n [II
us ed cyljlluric81 s pcc imcn s S l'1ll di: I1:Il..'tl' l' ;"111(1 10 e lll l o n g; C: lrrcr :lnd ~li ...'hl' l 1' \ u::- e d
cy li. ndr ic al SP CC1 1I1(, IlS 5 em d i ~\lill't,:r and 15 l'lI1 lO! I;: . H:ll!1 :H.: i e r ['~l lI se d r cct: lIl gu l :1 l'

sp ecim e ns 5 x 10 x c5 c m, th e ",':,' ,ioe os " e re ::ns t fr e qu enlly IIS(,U :It \l~R /N I (C .

Tabl e r. Uni<l xio i compres s i v(' st r e ng t h of co lu nlll ,tl"- g r:l ln e d S-.2 i c l' of nV12rage
~ rain si : c of 4 to S mill at - IOoe , i O:1ded perpen d icu la r to the l'011 111111 5

[n ve sti g"tor ( s ) lach i ne Spe cime n 5 t n; ~s


Refe r e nc e C1p ac i ty, Gcome try , Ca c ff i c i l'l1 t E:q >olll' nt
~IN I!IOI r\ m

~luguTuma (6 ] 0. 02 c; x cO x c5 I . Sc x lO- lJ 3.00


rcc t ;ll1gu l: lr
Gol d and Krau sz (7 ] 0.05 SO x 100 x CSO l. OCl x 10- 7 4 . 01l
ond
40 x 80 x 190
rcc t n ngl.l L ~r

~I i
che I (5] 0.0 5 50 !llln uiam Jnd 1 .:;5 x 10- 0 .%
(r ep r e:-it: nti- ng 100 o r 150 mm
La va l r esul t ,) 11.)JIg cy1indric <-l1
or
50 x I UO x 25 0
re c t :1ngular

Fr ed er king (8] 0. 10 50 x 100 x 250 ~, 32 x 10- 7 3,34


rec t ang ula r
Sin ha (9] 0. 10 50 x 100 x 250 3.3 7 x 10- 7 1.03
r ec t a ngu 1:11'
Sinha [10] 1. 0 so x 100 x 250 1.81 x 10- 7 2 ,90
( tr u ly c ons t ant ( cl ose d r e ct a ngu1al'
stra in ra te ) l oo p
sys t e m)

im i l ari ti es - - Ih e i.mpor t a nee of th e co eff i.c i en t an d s tres s expone nt i.n


equ ation (1 ) has been enllJha si:ed . Examination of th e expe r i 111'ntal res ult' s giv en in
e quatio ns (5) t o (8) indicat e , " it h th e excep tio n o f equation (4 ) , " hlch co rres ponds
t o the r esu lt s of Go l d and Kra u sz (! ] that s tress exponent s were determ i ned by
d ifferen t inves ti ga tor s to be clo se to 3. Thu s , ai I tn e st r e 11 gt l1-stra i n ra t c , U1'V",
were found to ha ve th e s am e sh ap e , a s may be seen in Fi i!lI f e I .

585
SINHA, [10]
SINHA, (9]
':'
E fREDER KING. IS]

,.
z GOLD & KRAUS Z, [7]
MICHEL, [5]
b
MUGU RUMA . [6]
SINHA, [ll]

'"
'"

'"co

~
'"
a

o ~----------~----------~----------~----------~
10 -7 10- 5
-1
STRAIN RATE , E, s

Fi gu re 1 Dependence of stren gth of co lumna r -graine d i ce on


stra in rate at _1 0 C

Dif feren ce s -- Although the Curves are s im ilar in Figu re I, their posit ion s are
sign i fi ca ntl y different, The position of the stre ngth -strain rate curves i s determined
hy significa nt differences in the co effi ci ent s of equ atio ns (3) to (8). Thi s may also
be seen i n Table / , In ge ner al, large r capac it y and probably s t iffe r machines t en de d
to give a l ower va lu e for the cocfficien t, and thi s s hift ed the cur ves to lower str ai n
ra tes. Figure I, in conj un ct i on with Tab le I, sho ws c lear ly that a lar ge capacity
mac hine give s a hi gher streng th at the same nominal s tr ain rate than does a low
capa city ma c hine, and that a closed-loop system with equi valent infin it e stiff ness
y ield s th e hi ghest strengt h,
noma ly -- According to th e evi den ce of th e previo us sec ti on, test machines
similar in capa cit y and stif f nes s s hould give s imil ar results, This is e vident in
Figure I if th e results of Fr ederk in g [8] are compar ed wi th tho se of Sinha [9]. Both
used the :;amc 0 ,1 ~IN capa c ity ln stro n TTD~I-L ma ch in e and a lmo st identical testing
cond i tions, On t he ot her hand, th e results of Go ld and Kra usz [7] s hould be s imil ar
to th ose re por t ed by Michel [5] because the same 0,05 NN capa ci ty Wyke ham Farran ce
~lodc l 57 test machine was used; but Figure I sh ows that t he t,,'o sets of re s ults differ
signi ficantly,

586
Th e coe ffi cie nt in e qu a ti on ( 4 ) o htained by Go l d a nd Krausz [ 7'1 i s s i g nif ic antl y
sma ll e r th a n th ose ob t 3 in cd .1 t L;l va l . h' hcrcas th e co rrespo ndin g s t ress expo ne nt i s
s ign i fic ~ lntl y g r c.:1 t c r. Thu s th e res ult s ap pear to be a noma l ou s in COm pJr L50 n with

o th e rs, and t he ~In omu l y C.1n be sec n in ri~urc 1, as \~c ll, \~ 11CrC th e Cl!rVC intercep ts
~ 111 the other cu r ves.
Anomalous ch,]l'actcri s r ics in th e results of Gold a nd Kr au sz [7] cou ld be due to
the fact that equati on (el ) ref le cts the ,lvcLlge dependence for a ll th e lee types
te s ted rather than ju s t S- :! LC (, . Usc of specimens \\;ith differ ent geome t ry by Gold and

Krausz IDay also contribute t o th e ' Inom:II)" Sinh'l and Frcdcrking [ 151 showed that
c ha nge in geometr y C1ffcct s tile rcl:ltiv(' stiffness a nd r:lt e o f l ond in g o f (l sys tem Clnd
th e refore c h:l ngcs the st rC'llgth :It the s: lIn(' nomi n:ll stra i.n rate . Note th ~ t t he comb ina

t ion of I mv ca paci t y m;l(.: hine an d sm;1i I ~ pecjlIlen genmC' t ry usc J oy ~1u g urum a [6} ga ve
results compara ble t o t hosc repo rt ed hv 1,li che l I S ] , Th i s indica t es that i f o nl y the
r es lIl t s of S-~ i re arc L1 ken fr o m t he r e po rt o f GolJ a nd Krau sz [7J, a nd a ll Ol,ance i s
made for e ffect of ~ p cc im e n s i.! e, he tt cr a g reemellt wou ld be founJ \vith t he results of

~ I iche l (equ at ion (R)),

Cr ::.:.Jl R'lt l' " - III I'l' OI'O,ill!! ' I viscoc l:lstic model for i c c, Si nh ' l 1111 described the
irreversible VlSCOllS str ~lill r ;lt c component of the constant stress c re e p rate at -lODe

as

(9)

Equati on (9) is su rp risi ng l y c l o:-;e, bo th in numericll va lu es f or th e coe f fic ient

a nd in exponent t o eC(uation ( 7) f o r s tr eng th oh t :-lined un Jer tru J)' cons t a nt s tr 3 in rat e .


This indicate s strongl y, <=In d Figure i Ii tts tr a t es it g l'ap hi c<l 1 Ly. th :lt the resu lt s for
the s tr es s dependence o f v i scotls f l o\" r:lt C' obtained in co ns t;lnt s t ress cree p t es t s mC'l Y
be com l1arab le with the ex periment ill observa tions on th e dellcn dcnce of )'i eld s tress on

s train r.1te [10] on l y I,hen th e lmposed strain rate is trul y constallt.

ln sllfficiences ill l\c porr eJ I{esu l t ,


Rational analysi s o f d ' ita f ro m the icc literature is d iffi c ult ill " lmo s t all
cases owing to lack of vi.tnl infor m:1 tion heilring 011 the r esu lt s . [mportnnt details

are commonly omitted concerillng ~ tr ~11n and dUl':ltion of te st, in sp it c o f the f~ct th<.1t

(a) all tests take some s pec i fi e time to complete, Clnd (il) spccimen s d eform Juring
testing. This i nform:nion cOl dJ he obt:li ned I"i th littl e :Hlditi o n:d effo rt duri ng the
experi ment s ,

S tress and Strain ' It Fai lu re -- Alth ou g h it '''IS imposs i b le t o o ilt :IiI\ information
o n t he time aspect s o f th e av<li la bJe t es t resu lt s, SOlllC s tr ~ lin Ole; l:-> lIreme nt s \v cn.' lTl,Hlc

for comp arat i v e stu <..ly . Co ld anu Kr:tllsz [ 7 1 provi ucd three stre:-:,~ - s trilin uia,!!,r(lms (sec

587
Fi gure 6 of their pap e r) correspo nding to En = 3.3 x 10- 7 ,1.7 x 10- 5 and
1.7 x 10 - '1 5 - 1 for S-2 ice at -9.s o C. (It is not cer Ll in whether nIl three "l'c , i l11l'ns
had the same ~', eometry.) The strains \:ere ::lo:.:~!SUrl'tl IH'th'cl' 1l the top and th e bottom
platens (given here in I'i g ure 2).
Drouin [1] also provided onc s tres s-s train d iagralil (S<.~ Fi.gllrc 4.75 i n h i, tile,:is)
for cOlumnar-grnin ccl S-2 ice lo aded normal to the column Rxi < at _9.4C and suhjected
to a nominal str ain rate of 6.9 x 10- 8 5- 1 . Th e' Ir'''H' r yield st n'ss and th e str ain at
y ield were obtained fr om this illu s tration and will be u sed lat ~ r. Thi s pair of
results is not sho\oJTl in Figure 2 because of the hlrge amount of strain in vo l ved .
Drouin (I, 14] showed, as l..Jel l, the variation in str:lin rate d1lring, t ests .
Ramseier [4] illustrated three s tres s-strain diagrams (F igure .1 2 of his hes i s )
for S-2 ice at -10 C for nominal strain rat es of 1.8 x 10- 7 , 6. 1 x 10- 6 and
6.7 x 10- 5 s-l. Th e results from the three curves are shoh'n in Figure 2. Stra in s
were Inc :lsured between the top and the bot tom p l at ens, but no information \\ ~\S ~:~ve n

regarding the time aspects of the te sts.


,Ii c hel [5] presented three s tress- s train di agr ams ( Figure 9 of hi s pon er l fur S -C
ice a t -10C for the nom i nal s train rates of 2. 4 x 10- 6 , 2.5 x IO -
S and .1 .2 x 10 - 5 s - l

The up pc r yie ld stresses and the strains obtained from the three curves are sholm in
Figure 2. Deformations were measured but th e time :l ~ pcc t s of th e te s t s ar e .l ot
available.

o SINHA. [l0]
/ SINHA. [9)
I
I
E
(SINHA. r9)1
/
c: n
'" GOLD & KRAUSZ. [lJ
;IiCHEL. [5]
/ RAMSEIER. r4]
/
b / 0
/
/
/
'".....
V> /"
?
/
/
/
I
'"/ / o
/ o
'"o

-

o
o 10 20 30 40 50
4
YIELO OR FAILURE STRAIN. <f. 10

Figure 2 Dependence of yie lcl or failure str " in on the corresponding


st res s observed by vario us in ves ti gato rs for COlumnar-grained
5 - 2 ice a t _IOo [

588
Oependenc e o f s train o n s tr ess a t yie l d o btai ned e a r li er by Sinha [9, 10 ] are
also shown in Figu r e 2 . Time a sp ec t s o f th e tw o s e ts of re s ul ts have al re a dy be e n
d is cussed i n gre a t detail.
Al t hou g h the sc a tt e r is I "rge, Figure s hows that the result s ob ta i ned at Laval
Uni ver s it y [1, 4, 5] a re cons is te n t. This could be related t o th e use of the ""lie
t es t s ys tem, a nd th e results sho ul d th e r e f o re be comparAb l e to tho se of Gold and
Kr a usz [7] who u sed " simil ar machine. I'i~ure 2 5hOl.S that the y a r e Co",p'lTab le,
a lthou gh th e latte r seem t o have o bt a in ed sODuwh a t le ss s tra in for the sa me yi eld
stre s s es . Gold and Kr s ll sz do no t s how any ~In oma l y Ile re , as was po in t ed out fOT
Fi gure 1. The in vest iga ti o ns of ~Ill g ur um" [ 6 ] and FrcJerking [8] cou ld no t be u sed
bec3llse s cr ai ll da t n aTC lackin g .
Re s ult s given i n Fi gur e 2 i ndic a te s t ro ng ly th a t a l arge r ca paci t y, st i ff e r
ma chine ten ds to de form the ma te r i a l l ess in induci ng fail ur e CI t th e ;; al~ll' s t rc=,s l e vel.
It seems th erefo r e th a t th e ha r der t he sys tem the l ess the a ppArent ductility of the
mCl t erl <ll. 11l e closed-loop sys t e m \oJith equ.i valent in fi. llite stiffn ess defo rm s ma terial
le as t .

F3 ilul'e lodlilu s
Since de f orma ti on of ma t e ri a l seems to be co ntro lled by the s tiffn ess o f th e
t es t i ng s ys t e m, i t was decid e d to exa mine the c f f0 ctiv e s t i ff ne s s o f tIl e nl~ t c ri a l at
f :l llure for all the aV di lab l e re s ults. Thi s l ed to the introduction of. Cllllccp t of
"Faj lure rlodulu 5 ," E de fined a s the Tatio o f the upper y i e ld or fai lur e s tr ess a nd
f
c or r C';.:;po nd ing str::l in ,

( I I' .

F:1ilUfC modulu s is cs s cntirdly th e se cant modul us co rr espo ndin g t o th e ma xi ll lm s tress


a nd r i .! '.:lre 3 illl.lstrD.t es t he avail a ble e x perI li\e ntal resu lt s di sc!l~scd in th e pre ,"iou s
s e c ti on i n te rms of th e corresponding imposed s tr ai n rate. Drou i n ' s mC ,1 ::' HrC'lllcnt.

me n tio ned before bu t not in co rpor n ted i n Figure 2, can n Oh' be used. [rr C$ )c ctlve of
the t es t sy st e m, Figur e s hoh' S that the f.1 i lli r e mod u l us in c re ase s hi t h incre as e in
s t ra in ra t e. The val ue of th e modulu s at ;1 give n s tra in rate, ho\\ ev cl", uc pcn tl s on the
s tiffn ess o f th e t est s ys t e m. Th e i lilistra tion r C~lf ril'1t! ~ th a t (1 s ti f f er S~'5 crr

in c r c:l c c C the ~""' "r e nt s t iffnc ss of I.h e mate ri al. As " inhn ' s l 1(1) rC,l tlt s ,,'ere
oh ta ineJ for tru l y co nsta nt s t l'a i n ricH e under <3 clos e d- loo p 'T\o de o f s tr a in con trol l

the l' ffr c ti"l' s tiffne s s of the s ys t em h'as ve r~' hig h and the co rre s pond ing E versus
f
Cllrv e mny bc COll s id c r ed as tllC l i rl itin ~~ on e. In sj,it c of tl, C f Jct tl, at :: iml l J r
rl .1 l" !Li r:l'~: \\!el'C u s cd i n both l abor a torie s, Figure:) "ind ica te s th Rt th e t or a l s y:::tcms :J

La v:ll Un i \'c r .< it y u s e d b y flrouin [I]. Ramsc;<:!' r~ l , illld 1l1 c hl'l (:;1 "'erc ,II sof t " r th all
those U50d at IlBH/NI ~C ill' Co l d a nd Kr<1 us? (7) . Thi< ";I S p r ol>a h i,' hcco lls c the s ti fr nc;:s

589
E
o SINHA. [IOJ. CLOSED l OOP TESTS
SINHA. [9J. NOMINAL STRAIN RATE
29.. 16
z (> GOLD & KRAUSZ. [7). FIG 6
<.:>
RAMSEIER. (4). FIG. 6.12
o MICHEL, [5] . FIG 9
x DROUIN [lJ. FIG 475


'"
o
o
--- ---
(>-

-I
STRAIN RAT[. f. s

rl gll TC :) OC,c ndcncc of failur e Hlotl Hl us on stra in rat e

If the tOl;ll ... ., s Cm oc pc nds also on the "' ti ffn . ~., chaT J.ctcri.stlcs of the v3rio u s
" (Im"tl ~nt ... of the loaJi ng c o l umn, for ex ample, th e load cell, platens , etc rlS]. Tt
shOl>1d :,!:" in he- l' (') int e J out thot the Hnomal), in th e resu l ts of Gold and Krausz sho"'n
in r i ':f r e I is () rrc' $(' nt in r igu re 3 .
\11 mo 11 11 r eport ed in Figur e .~ :-Ire co ns id e rahly !u',\'or than the YOllng's moclulus
() f ; 1hOIl r ~) . ~ r. \ . det e r mined c ;!r Iier [11, 12 1 f o r th e same type of ice, loading
Lli f eet-jnn , :llld l Cl11pC r;lturc . Thu s the hr itt Ic-1 ik e , premature fai.lurcs occurrin ~l. at
4
:-i ra in r:ltc<..; i n the T;l nl (C of a l' out JU- s-l m.1}' not he cons id e red as pure e lastic t y pe
of I O:ldn}.~ or :1S t rt lly "hrittlc-t yp c" f n i lur c ~ . Thi s :v :p lies ver y s tr () Jl~:.ly to thc
i ll '; cq\ Il(a ti on s o f Co l d and Kr nH "Z [71, Ra IDs c i c r [41 and ~lichel [51.
!lrcffiHt un, f:d l u re s. h'crc not ctl h y the author [() l to occur a t (-t nnminol s troin r at c
1
of ?i / 10- 'i-l o r hi gher For ~I con vell tionCiI s ys tem. For similarly prepared s pecime ns
the a uth nr [10 ] found ahrupt s plitting type of failure at a con s tant s tr a in rate
o f ~, x )( ( . , s- I Thlls the occurrence of th e so -called hrittle - Ilke failure s also
,kl,en,1s nn th e stiffne s s of the test sy s r m for the same end conditions.

(on c ll l sion~

Iin i:lxi:JI str e ng th rc'->ults o htained hy diFferent investigators dcpend on the


, t i ff nl'ss u the test syst e m, Il se d. Th e shape of the s tr e ngth- s tra in rate c urve docs
no s eem to be affected by the relative s tiff ness o f the sy s tem, but it s position on
t.ll e :-i ra in ratc axis tS, The I l miting position of thi s curve i s determined hy a test

590
S)'StC'11I .such ;IS th e closetJ-loop sys tem, h'ith effective infinite st if f nes s . This limit
ing rcLLtion h;I S ;1 on c-t o -oIlC cn rrc spol1uence "'i th the JepC'nJcllcc of non-rccovcrahle
vi scolls flow Ull str es . :. in lmi ax i;t\ cre ep .

The duct iii t )' of icc depends O il the s ti ffness of the te st s )'s tem. 1\ te s t sys tom
with infinite s tiffn ess deforms the material l eas t. The sof ter the sy s tem the soft er
i s the re spons e of thc mater i ;11.

rili lure mouII\u s increases wi th incre as e in strain rate for a given sy s tem.
It ;l l so jn c rc;l~cs with incre;ls c in stiffness at a given str ain rate. All th e faiIllr e
moduli were found to he co ns ideru bl y le ss th a n You ng's modulus of icc. Apparently,
brittle - like, pre~lture failur e s do not repre se nt pure clastic l oading . The strain
rute ut which premature failure s tarts as we ll as that at which th e apparent ductile
hrittle transition begins, a l so depends on the stiffn ess of the test system. Premature
failure occurs at a lower strain rate in a s tiffer system.

The author i s indchted to Or. Y.S . lVang an d EXXON Production Re se arch Labor a tor y,
Iioust on , for their cooperation in permitting usc of the closed-loop ma chine . This
pap er is " contribution fr om the Divi s ion of Building Research, National Research
Council of CDnoda, and is published with th e approval of the Direct or of the Division.

Re f e re nces

[I] ilrollin. ~I., Le s pous se es d'origine thermique exerc~es par le s couverts de glace

sur l es structure s hydrau lique s . Ph.D. Thesi s , La va l Uni ver sity, (1uebec,

C IIl"d a, 197 1 .

[ 2] Curter, Il., and ~Iichel, B., Loi s et me ca ni s me s de 1 'apparente rupture fragile de


la glace de riviere e t de lac. Ra pport 5-22, O~partement de Genie Civil,
Universite Laval, Que bec (Que), 1971.
[3J Michel , B., and Paradis, M., An al yse s tatistique du fluage seco ndaire de la glace
de riviere et de lac. Rapport GCS 76-02 , O~partement de Gen ie Civi l,
Universite Laval, Quebec (Qu~), 19 76 .
141 Ramseier, R.O., GrOl,th a nd mechanical properties of river a nd l a ke ice. Ph.D.
Thesi s , Laval Uni ve r s ity, Quebec, Canada , 19 76 .
[5] ~Iichel, B., The s trength of polycrystalline ice. Can. J. Civ. Eng., V. 5, No.3,
1978, p. 285- 300.
[6] Muguruma, J., Effects of surface conditions on the mechanical properties of i ce
crystals. J. Ph ys . 0., Ser. 2, V. 2 , 1969, p. 1517-25.
[7] Gold, L.W., a nd Krausz, A.S ., Investiga tion of the mech anica l properties of
St. La'Hence Riv er lee. Can. Geotech. J., ~ (2), 1971, p. 163 -169.

591
I

ISJ l l' l'\"krkill .l~, H ., PI:t l l l' - .. .;,tLli l l !".'U IIIPI l~;-;.i\'l' ~tr('n g. tll o f (U IUll1l1 ; lr-\J.r;llll t' d , lil t!

gr;IIlIlLII'- ":' llo\\, i l'l' . . 1. LI :-H: i o i., lr; [X II) . 1~) /7 , p. S()!i- 5 1(I,

\ ~l J S inh :-l, N .r-.., I{; t te sl'lIs it iv lty of ( o illpre...;s i v l' ",tn.-'Ilgrh o f co l tl mn :tr-gr(lill l 'd I CC.

L x p c rilll l' llt a l ,\lcch:IJli (.'s, .~ I l h ), 1~)Hl, p. 2 W1- 2 1H.

1101 S inh :l , N. " . , ('o n ..:, t:lllt s tL l ili r ;ltl' :1 11.1 COl b t;IIlt s tn' s..:, r:lre CUlIIIH'<.'ssiv c s tr (' ll,I'.!h

OJ' (.{) I\lllIJl:lr- ~! r ; lill ed 1(.' . To h e flllhl i ... h l' d.

111 \ ."; inll:1, ~.I\. Hh l'o l p.~~y 0 1' C()llI l tl ld r- ~l": 1 ;tll' \1 icc'. 1:.x pc r' imc ll t:1 1 r,1L'l'll :111i cs, J:'<. ( 12 ),

I Q-'S, p. 1()I- 1"\ 1.

! ~ )7 ~, p. IS .... 17 1 .

11 :;1 Cnld , 1..1\' . , The L lillll'l' PI'( h.T<";S i ll C ()\lllllll:II"- j.! r ; l i n l' d i Cl' . Ph . I) , 'llH'o..: j::. , \ k(;ill

l i ll iv~ ! ' :-.:.ity. 1 ~ )7( ) .

fl ll IlI'(Hlill, 1\1 . , L:lh n r :!t ll l"Y jI1Vc' ..;,ti.~ : lti ()1l DO i cc' th e J"ln: 11 P Il',,;s lI rc :-:.. 11f(K f Alm I cl'

SY IlIP0:-: i 11111, l.enill g r ;l d , ...!h- ~~ ) S l'pt c !nher 1~) 7?. p. 7"'!-HO .

IISI S i ll h :l , N .K., :llld l-"rl'd l' I'kin g . Ie, LfFc c t o f t l':-:. t ~)":-:. tClll ~ti r rll {,:~ :-:' 0 11 :-:. tn..' llgt ll o f

ic c' . Proc. S th Int c rl1<1ti ulI: 11 COnfl'rCll c'c 011 110 rt ;IJld (k C: 11l I:Ilgil wl' ri ng lIn ue r

A r c, t ic C p nditi()I1~, T rO lldh c illl , \~) r',\:I)', l,~ 17 Al l gus t 1~)7~). p . 'OX - / 17 .

592
Dis c uss ion 011 'T COlllpiJrlltivc Study of lec Strcng"tll Du LI '! by N. 1\ . Silllltl

Oiscu>scd by Y. S. Wang
E X:\ Qn Pr oduetion I\.Q ~ c ~1rch C'ompnllY
Hous t on , T X, U.S.A.

Tilis popel' is a continuing- cont ributi on by Dr. Sinh" in his eHol'ls to resolve tile ilppClrent
di screpancies in til e results of Inboratory streng-til tests of similur icc by differ'ent
investigators--" fa ct that has pu zz led ice mechanics r ese arcllers. lleeill,"e tile stiffness
of an ice sa mple varies with both st rc ss and Stl'CSS rnte (or struin Hnd str'a in rilte), the
strain history of a sample, tested on a conventiomll ma c'hine, strongly depends upon the
stiffness of the loading system which includes th e stiffness of the ma c llin e f!'Ullle , tile
load cell , the lood platens, and the compliant platens if th ey arc used, et c. Thus, two
samples te st ed on two machin es with diffel'ent stifi'ne," chnl'l1C't c l'istics rn ay be subjected
to two differ ent loading histot'ies under' the same nomina l loading- pl'ogrnn1s. The
strengthS obtained are, therefore, a lso different,

The author compar ed test result s from many investigators and concluded thut under the
same nominal s train ra t e , stiffer mac hin es pr'oduce hig-h er' strengtils and less sa mple
deform a tion. Thi s is becaus e, for s tiffer machines , a la"ger portion of tile nominal
deformation g-oes to the sample as compa red with a sof t er mac iline, TIlliS. sa mples
tested with a stiffe r machine actually ar'e subjected to hig'll er strain r'a t es eo mpared to
samples tested with sorter maChines, Also, fr'om my pe rsonal experience from streng-til
tests with a closed-loop machine, the sample strain at which f a ilure OCCUI'S dec,'eases
with the increase of strain rate, which is consist ent with wh a t tile author tw s found.

The author also pr'esented on int eresting obsel'vution that th e relationship between stress
and il'reversible viscous strain rate in cr'eep tests is ver y close to th e one between
failul'e stress and strain rate in constant strain rate t ests. I am currently working- on
a one-dimensional stress-stra in-strain r a te relationship t o descr ibe the bellUvior of sea
ice under a variety of loading conditions and this fOl'mulation predicts that the stl'ess
slt'a in rate r elationships for the two types of tests 8 1'e the same. I am very mu ch
encouraged by Dr, Sinha's r es ults,

593
COMPARATIVE STUDY OF ICE STRENGTH DATA
by N.K. Sinha

DISCUSSION BY:
Franz Ulric h Hausler, Hamburgische Schiffbau-Versuchsanstalt GmbH, W-Germany

It was shown that a high system stiffness of the testing apparatus provides
for realistic test results. Since the only closed loop system of all the testing
apparatus presented here has by far the highest loading capacity, the conclusion that
the good results gained on this system are based mostly upon its closed loop control
could be questioned . Nevertheless, the results gained at HSVA with a 100 kN spindle
driven and two 125 kN servohydraulic closed loop test systems during a research
program on the multiaxial strength of saline ice lead to the same conclusion.

Unquestionably, the closed loop technique represents today's state of the


art . It gives uS a good tool to achieve reliable results even with rather small
testing machines because it provides for equivalent system stiffnesses which are close
to infinite. But as a result of the experience we have gained at HSVA during the
last three years, it seems important to note that a true constant strain rate will
only be obtained if the whole test system, including the specimen to be tested, is
optimally tuned. If not, oscillations or the other extreme, long response times
may lead to considerably large differences between desired and measured strain
rates, especially at higher strain rates, i.e. at ~ > 1,0 x 10- 3 s-l.

Questions:

1 - How long was the response time of the closed loop control system in the strain
rate controlled mode?

2 - How large were the deviations of the really measured strain rates from the
desired value at the different strain rate levels?

3 - lfuat sort of loading platens did you use?

594

AUTHOR'S REPLY to '.U. Hau s l er:

1 - The tests were carried bel o w 1 0 - 4 5- 1 an d so 110 problems with response time
a ro se.

2 - Within 1%.

- Polished s teel.

AUTHOR'S REPLY to Y.S. Wang:

The autho r alJpreciacc.:s lhe se c mllle-nes.

595
PRIMARY CREEP AND EXPERIMENT A L METHOD
FOR TESTING ICE Il\ VARIOUS CONDITIONS OF
STRAIN RATES AND STRESSES

P. Duval , M. Maitre, A. Manouvrier, Laboratoire de Glaciologi e Fran ce


G.Marec and J.C.Jay * 2 , rue Tres-Cloitres
38031 Grenoble Cedex
* Institut de Mecanique-Gr e noble

Abst ract
A torsion creep test was conducted on randomly oriented snow ice to analyse pri

mary creep . Primary creep includes a reco v erable delayed elastic strain and an
unrecoverable strain which also decreases with time. A new apparatus [or testing

ice in various conditions of strain rate and stresses is described . It has been de

signed for deforming ice samples in compression in two perpendicular directions .


Experimental tests can be performed for constant strain rates. constant stresses

or following a deformation path.

I. Introduction

The plastic behaviour of polycrystalline ice has principally been studied by creep

t e sts under a given load [IJ [2] . On the macroscopic scale, pOlycry s talline ice

affords at first a decelerating creep (primar y creep), then a secondary creep

during which deceleration goes to a very low value. If the load is applied for a
long time in order that strain exceeds 2-3 0/0, creep rate increases and a stead y

state can be attained at low stresses. Acceleration during tertiary creep i s in

duced by recrystallization processes and development of preferred crystal fabric

[3J. But an unlimited acceleration can be ob s erved dur i ng tertiary creep caused

by increased crack formation leading to fracture of the sample14J

For secondary creep. strain rates ij vary with deviatori c stresses T'ij with

the relationship t: =:tij ~ T n-' T 'ij (-1)

ij 2 TJ 2

596

j
-=
where n is the viscos ity , B a parameter which depends principally on temperature
and T the effective shear stress defined by T'-= -..L L (T.'.'"
2 lJ
Equation (l) has b ee n verified by Du v al [5] for isotr opic i ce s with creep tests per
formed in torsi on, compression and torsion -compression for 1 <T <0.5 M Pa .

Equation (1) implies that the principal axes of stresses and plastic strain rates

tensors coincide.

2. Primary creep
The time dependent deformation during primary creep is generally expressed

by : E _ E (2.)
E + E t
o P s

whel'e is the instantan eous strain on l oading. p is the tran s ient strain and
o
5 is the secondary creep rate. From compression creep tests on columnar

grained ices, Sinha [6] deduced that f. p was totally recoverable. Lloyd and Mc

Elroy [7] and Duval CS] intr oduced a non- recoverable component in transient strain

A to rsion creep test was performed with the apparatus used by Duval~] at
- 2. O'C on randoml y oriented snow ice to analyze the primary creep curve.Crystal
size was about 2 mm. Constant torsi on torque was applied during 25 hours; then,

the specimen was unloaded. Both creep and creep recov ery were recorded

(fig. I )

--------t-----------Jh~---------7.~--T--
,~.--('O"-
-,,-,-k - --"
Fi g. I - Experimental creep and recovery curve of polycrystalline snow ice
a t -Z 'C - Shea r stress = 0 . 2 M Pa

597
By assumin g that steady state creep was obtain ed after 25 hours, 's was calcu la
ted. The non-recoverable comp onent of primary creep was obtained by substrac

ting from the total st rai n f. the recoverab l e strai n and the strain corres pond i ng

to steady state (fig. 1). From this test, it appears that the reco verable delayed

elastic str ain is large with regard to the expected elastic strain. But th e non
rec ove rable compo n ent of primary creep E is equally large at time t = 25 hours.
p
Since steady state creep was probably not attained at the end of the test, thus the
non-"" F'~vverable primary creep must be lar g ("t" t han the One calculated abov e.

For conditions w h ere the disl ocation cre e p nl. f!' chanism is dominant. the delayed

e l c,S[ icity is correbted with strain due to grainboundary sliding [6J. It was assu

med by Sinha [61 that transient creep was inversely proportional to crystal size.

F rom Du va l and Le Gac ~-91, prima ry creep does n o t appea r to inc rease whe n
crysta l size decreases and no influence of crystal size was found for steady state

creep. It was shown by Raj and AShby CI 0: that strain due t o grainboundary sliding

with elas tic accomodati on does not depend on crystal size.

Phenumenol ogical models were de ve l oped by Michel[4j and by Duval ell] to des
cribe the:: time and temperat ure d epe n dencf' of plastic flow of isotropic polycrys

tall i n e ices, Expe rimenta l work is needed to improve these models for non

st e arl y and complex loadings. Th ~ app aratu s descr i bed in the paper was bui lt to

ha ve a b, tt ('l' understan ding of the mechanical prop e rt i es of polycr ysta llin e ices
u nd (" r comlJlex l(')<ldln g ~ and for non- steady state.

3. D e sc rio tion of h e bi a :d a l pr e ss

3 . I. Mec hanic a l d e tails

Th e b ia xial press has b e e n designed for deforming ice samp l es in comp ressi on in

two perpe ndi cular directi ons. This is achieved by linking togeth er fi v e rigid pla

'. ,_ '.c v.s ind ic ate d in figure 2. Platens 2 an d 3 ha v e a slide built in them . Ice

samp! , ." can be deformed wi th dim ensio ns varying between 4 and 8 cm in the X

and Y directions and between 5 and 10 cm in the dir ecti on Z. They expand and

contl' oct in the two X and Y direc t ions. Only expansion is possible in the direc

tion Z since Q z = O. Power is pro v id ed by two stepper motors. Reduction gear


boxes convert the dri ve to the low speeds needed for experiments and a mechani

cal screw transforms rota.ry motion into linear motion.

598
rr-9-2_
I
-

~
t--

I
t
,
-(1)
f-0

L
Fig .2 - Sc h ema ti c representation of the biaxial press

l ee samples are enclosed in a compliant pipe in nat ural rubber (thickness :0.2 mm)
to m in imize tangential stresse s and a lubricant is put between platens and the pipe.

Slrain markers are inscribed on three faces of sam ples (Xy, YZ and XZ planes).
Th e,v markers a re circular groov es 7 mm diameter, 0.5 mm deep and 0.3 mm
wide. They permit the strain hom ogeneity to be verifi ed. All th e apparatus shown
in figure 3 is housed in a regulat ed temperature cold room ( L\T < O. 2'C). Tests

can be p erformed i n the temp erature range -1 to - 45'C.


3.2. Contro l and measurement systems

The control electronics of the biaxial pres s includes three part s, a control sys

tem, a measurement system and a recording system.

The control system drives the stepper motors. Working can be manual or auloma

tic by means of a minicomputer built around a 8085 microprocessor. In automa

tic mode , th e biaxial press can be r un either at co nstant stress or f o ll owi ng defor

mation paths. In manual mode, tests are performed at constant strain -r ates .

Measurement systems give strains and displacements in anal ogical and digital

forms. Dimensions of ice samples are taken off the s tepper motors motions by

up-down counting s tep s. L V. D. T. transducers fo ll ow the pl atens motion . Com

pr ess ion stresses (Tx and 0y are obtained from s tr a in gauges transducers

599
directly buill ill pLllclls. Digital dat'Cl. acqui!"::iitiun <lnu rccoI'ding on C"\ pllTlched ta.pe

':-1 i"l' ("()111rollt'd by till' millicoITlputL'l' The uperalor curnmunicd\c.'-i with lhl' mini

COIl1pUtl'I' by mC;'lns Of;-l keyboard and a C. R.T. dala displrty terminal

Fig.3 - Biaxial prC'ss

3.3 Stre s s ';;t nd s L r a i n r a t !,' con Dgl.1 r ' ltion

\Vilh the biaxial press described above, the state of slr ess is plane Stress and

strain rale tens()rs are respectively

ax o o f 0 0
x

a
IJ
o a
y
o i ij
0 Ey 0

o o o
0 0 -(E +E
x

v./ith the incompressibility assumption

The deviatoric stress tensor i.s given by

600
_ - (J IT
2
1 _ --"_
x _ _J...
v
o o

L
a 3
20 -(J
ij o V x o

o o -((J
x
+ IT
y
)

The three invariants of this tenso'- are

1' 1 =0 I'
~
=_ l.. ( a l + a L - a a )
x 3 x Y Y

1
l' (50 a - 2 a - 2 (/) ( a +a )
3
27 x Y x Y x Y
and the effective shear strain rate r is given by

3.4. Relati o nship between stl'ess and strain rilte


It will be possible with this press to check the equation (l) [01' isotropic icc. A

gene rat form for the relationship b e tw e en stress and strain rate tensors w a s
given by Glen P 2] and Morland [I ~'I [or seconda ry creep

2
C
c
iJ' = ,I,
't'l
(I'
t'
I ' ) a'
3 ij
+ ,I. (I'
't' I.'
I ' ) (0'
3 ik a kj - 3" l~ s.) (3)
IJ

with i5 ij =I i[ i = j, but zero if i oj j.

With the stress and strain rate configuration given above, we have

X = +cjJ,(l'~,
).
f I' ) (2 a - a )+ ~ cjJ 1. (I'L I' 3 ) (2 02. + a - 2 a a )
" x y 3 x y x Y

f ~ 4> (I' I' )(2 a - a )+ ~ l/J 2. (I ',- ' I' ) (2 a :Ly + a x~ -2 a (Ty ) (4)
y 3 1 L .3 Y X 3 x
3
and
z
( x
+ E
y
)

Since E x and f yare independent, it will be possible to verify if equations (4) re


duce to equation (1) in a stress range larger than given by DuvalCS] . On the other

hand. this apparatus is especially well adapted to analyze the degr e e of anisotro

py o f polycrystalline ice. But tests will be c ondu c ted on polycrystalline isotropic

i c es to improve the relationships between stress and strain rat e for non-steady

state and complex l o ading[11J .


References
[lJ Glen,J.W.; The c reep of polycrystalline ice. Proc.Roy.Soc . Ser.A , 228,519

538,1955.

[~ Steinemann, S.; Experimentelle Untersuchungen zur Plastizita.t von Eis. Bei

tra.ge zUr Geologie der Schweiz, Hydrologie, la, 1958.

601
[3] Duval. P.; Creep and recrystallization of polycrystalline ice. Bull. Mineral.
102,80-85,1979.

L-~ Michel, B.; A mechan ica l model of creep of polycrystalline ice. Canadian Geo
technical Journal. 15, 155-170 .
[5] Duval , P. Lo is du fluage transitoire ou permanent de la glace polycrystalline
pour divers etats de contrainte. Annales de Geophysique, 32,335-350,1976.

[6J Sinha, N. K. ; Grainboundary sliding in polycrystalline materials. Philosophical


Magazine, 40, n'6. 825-842,1 979.
[7] Lloyd ,G.J. and Mc Elroy R.J.; On the anelastic contribution to c reep. Acta
Met.22,n'3,339-347,1974.
l~ Du va l , P.; Anelastic behaviour of polycrystalline ice. Journal of Glaciology, 21,
n'85 , 621 -628 , 1978.

1.9j Duval. P. and Le Gac H. Does the permanent creep rate increase with cr ystal
size. Journal of Glaciology, 25, n'91, 151 -157,19 80.
~~ Raj,R. and Asby, M.F. ; On grainboundary sliding and diffusional creep. Met.
Trans.,2, n'4,1113-1127,1971.
[II] Duval, P.; Constitutive relations for the non-elastic deformation of polycrystal
line ice. IUTAM Symposium, Copenhague, Per TRYDE, Editor, Springer-
Ve rlag, 51 -59 , 1979.
[12J Glen , J. W. ; The flow law of ice . A discussion of the assumptions made in gla
cier theory, their experimental found ations and consequenc es. lASH 47,
171 -183,1958.

[13J Morland , L. W.; Cons tituti ve la ws for ice. Co ld Regions Science and Techno
logy, 1,101-108 , 1979.

602
PRIMARY CREEP AND EXPERIMENTAL METHOD FOR TESTING
ICE IN V A RIOI ' S CONDITIONS OF STRAIN R A TES AND STRESSES

Dis c ussion by M. Mellor


CRREL lJSA

Th e paper deals willl two d i. in ct t op ics (J) primary c reep e xperiments ,


and (2 ) an a ppar atus fo r biax ial te s t s .

Pr i m a ry cr ee p u f ice c e rt a inl y d ps~' r v es nl 0 1"e in v e s tig at ion, a nd t o r Slon


tests ma y b e ve ry suit c. h l e . T il e o n ly ,' (" sull !-i given here ar ~' th f' sm o othed cur v es
of Fig \lre 1 . so n u l much corn m e n t ('a n })(' m d de. U se o f the t rm " s t ea d y state is l
'

pe rh a p s rnis l ca din g w h (' n t OLal s t r ;,i1l 5 ar e o li ly fr ;!: ct i on s of a perc e nt . F or the


ra ng v of stres se s a n d lenl p t ' r;JlU r es u n d e r C ' n ~ i dC'r a l ioI1 , it i s u nlikely that th e re
c an be tr u C' st e ;jcl y s L l t (.' ro r st r L' SSC S hi:' l u \,.\' 10 ~o . R e c e nt u n i a. x i a l com p re s sion
l(~st s at CRREL g i v e hi! h s Lr ps' ( f l' t' P C'1rv(, :-; [ 0 1' j :::iot ropic w hic h h a v ~ fl n inflection
po int at about I ~~ ax ia l::; O at il , rtnd a t1 n t h t~r rll s t rain s t -' 55 than O. 2 ~~ _ It would
be int f' r('sti ng to k n o w wh e ll'le r tilt " cLda o b t (!Il l'd by Dr Du va l and his c o lleagues
give a n y ind i c a t io n of an in fl e c tl o n;1t .sm i11 1 ti l r:'iin s _

The design of thf' bi a x }("!\ I O<'1d -lIg fr ;ln") (' i s ~u ch Ih a t any po s itiv e friction
between the f'pf' cimen a l d a plat tr. will t e n d t o lie \11 the s ;:trne di recti o n a c ross the
full width o f ec,ch plalen , w i l h tile fri c t io n d i r e c t i o ns opp o site fo r the two pl a tens
in ea ch fa c ing p a ir. Pl a ten i n t er frl c e c u s h io! IS o f s oft f~ lasti c Or p l ;t s tic sh C'e t have
bL' e u fo u nd und e sirab l e in ro c k n-\E.' ch a n i c s b e c au5 c e x trusi o n produce s itl nsile
surfac~ tra c ti o ns . It is 11 0 1 c l e.J r ho w th e rn d r k i ngs o n thl' s p ec imen L lC ( ' S are
uo c d to v(, r i fy str a in ho mo ge n e it y dur in!, a tes t. T o under s t a n d h ow t h ,' "pparatus
rL';d ly h c h .J. vcs, it would pt rl"'laps be usefu l to load a Slil b o f p h o toela s tic plast i c
a n d dn:t1 y ze the fr : ng e patt e rns.

Answer

I. I agr ee w ith your c omm e nt. TIle" c r eE" p rat e measur e d [o r sn la il strain s
d o es not c orrt' sp ond t o t he stead y state a n d is probably hi g h e r. Ilul. [or the d ,, " c rip
t ion of the tran s ient c reep , the co nsequ e nc e i s th;ll th e non-rccove rablt.~ part i s y et
mor e high th a n tha t giv e n in t h e papE'r .

2 . At th i s time, we are m a k i ng e xperiment s to test the apparatu s , The


circula r stra in. mal"k e rs in ~ cribed on th( sp eci men f<J cf' s a l' e deformed into a set
of ell i pses. The uniform i ty of th e di s t o rs i o n of th e marker s s u ggest th a t str a in is
hom o gene o u s.

603
I

PRIM A RY CREEP AND EXPERIMENTAL METHOD FOR


TESTING ICE IN VARIOUS CONDITIONS OF STRAIN RATES
AND STRESSE S

Discussion by L.W.Gold
Division of Build ing Research. Nat i on al Resea rc h Council
Ottawa, Ontario KIA ORI

This paper is primarly a description of a n app a ratus to be used , and it

wi ll be of great interest to see the res ult s ob tain e d w i h it. There is one po in t

upon w hich I would like the authors to c omment. In the exa mpl e g iven in the di s

c uss ion of re covera ble and no n-reco vera ble primary creep, th e y st ate th a t t h e

non- r ecoverable cr e ep i s independent of g r a in size. Figure I indi c ates t hat th e

t ota l strain impos ed on the specimen refe rr e d to in the discuss i on, is l e ss than
3.5 x I 0 -3 , an d one w ou ld not expect second a r y c r eep to h av e yet b e n e st a bl ished,

although th e c ree p Curve is quite linea r aft e r a st rai n of abo u t 2 x 10 - J .

Crystallographic o bse rva t ions I have made on columnar grained ic e ind icate that

as the ice deforms into the sec onda r y Creep stag e, polygonization and recryst all i

zation occur in the grain boundary region. The mOre the speci.men is defo rm ed the

m Ore gr a ins a re modified b y this process. Th is su gg es ts that although secondary

creep rate may not be dep enden t on init ia l g ra in size, it may depend on the

structural changes induced in the grain boundary region by the d efo rm a t iOl'. W h en

the au thors state that the secondary creep rate is i ndepa nd e nt o f grain s iz e . ar e

they referring to i nit ial grain si ze ? If so, did th ey det e rrn ir.(' grain si O:'.:e af t e r

defo rm ati on, or look a t st r uctu ra l cha n ges induc ed in the grain boundary reg io n

by the deformation?

Answer

Indeed, secondary creep was probably not establiShed at the end of the

cr e ep t est described i n figure I. With rega r d t o the ind epe ndance of grai n size on

secondary creep, we ar e referring to the i n it ia l g rain size. But, from Le Gac and

Du va l (1980), no change of grain size was detected after 1 '70 of defo rmation. We

agree th at stru ct ural changes (fo rm ation of sub-boundaries) are induced by the

de form a ti on and specially in the gr a inbo undary regi on.

Le Gac and Du v al P., 1980. Does the permanent creep rate of polycrystalline

i c e in crease w ith crystal s ize ? Journal of Gla c iology, Vol. 25 , n 0 91,

p.151-157.

604
PRIMARY CREEP AND EXPERIMENTAL M ETHOD FO R
TESTING ICE IN VARIOUS CONDITIONS OF STR A IN
RATES AND STRESS E S

Discussi on by R.Frederking
Division of Building Research, National Research Council
Ottawa, Ontario Kl A ORl

In this paper the authors review their approach to primary creep and des
cribe an apparatus developed for strain or s tress controlled experiments under bi
axial loading. The ap p aratus is in genious and it will be very interesting to see the
results produced by it.

The primary cree p ph a se as described in equation (2) compri ses instan


taneous strain E (presumably pure elastic) ; transient strain, E ; and secon
dary strain, EO t. The authors also speak of "recoverable dela/ed elastic
strain" and a "no~-reco vera ble component of primary creep11. Do these two compo
nents contribute to E ? H ow do the authors experimentally dist i nguish between
the non-recoverable p component of primary creep and secondary (permanen t )
creep? Also what deformation pr ocesses in the material are associated wi th these
strain components?

A very intrigu i ng apparatus has been devised. Could th e authors give som e
indication of the l oad and rate capabili ti es of it. The introdu ction of th e 0. 2 mm
thick rubber l ayer between the ice and plat en w ill increase the comp li ance of the
system. In assessing the perfor manc e of the system it wou ld b e h e lpful if direct
cont i nuous s tra i n rnea s urernents cou ld be made on the spec im en.

Answer

1. The time -dep endent deformation E during primary creep can be ex


pressed by

E E + E
p
+ Est.:......
f 0 steady state creep rate
e lasti c stra~
transient stra in

E is give n by
p

E E
r
+
u
P

The recoverable del ayed elastic strain E is measured after unloading.


The minimum creep ra t e ('" the permanent strain"rate, ) is measured during a
s
creep test.

By substracting from the total s train E , the recoverable d e l ayed e l astic


strain E and the s t rain correspond in g to sec ondary creep ( E t ), we obtain the
r s
605
non-recoverable component of the primary creep Eu.

2. With regard to the deformation processes which are associated with


these strain components, it seems that recovery processes control the primary
and the secondary creep. The delayed elastic strain would be associated to the
dislocation movement between subboundaries (piled up of dislocations against
grainboundaries and subboundaries).

3. The load capability of this apparatus is O. I MN and the rate capability


is betwe e n 10- 7 mm/s and 10-c mm / s.

A direct continuous strain measurement On the specimen will be


made to have. with the aid of the microcomputer. a good control of strain
rates.

606

I
PARAMLTRIC S TUDI ~ OF SEA- ICE
BI'A~I S UNDER SIIORT AND LONG rm;,1 LOADINGS

Luc LAINEY Ecole Poly technique Canada

Rc s parch Assistant, P.O. Box 6079, St ation "A",

Rene TINAIH Montreal, Quebec,

Associa te Pro fessor. H3C 3A7

Abstract
'Ihi s rarer describes flexural tests performed on 280 sea- i ce beams with an average
salinity of 5/00 , From the short term tests th e instantaneous flexural strength and
the apparent average elastic modulus were obtained for different temperatures and
loading rates. Long term creep tests allo"ed the study of the creep behaviour for
different temperatures "nd stress level s.

1. Introduction
'"he prediction of the bearing capacity of i ce covers is not a simple one from the
analytical point of view. Even if numerical OT approxi mate methods are used, accurate
knowledge of the mechanical properties of the ice is necessary.

In this paper t"o sets of flexural tests are presented in a modest attempt to shed
some light on the short and long term behaviour of sea-ice. Although the~e are many
va lid argume nts for and against flexul'al tests, in ge neral there are t"o basic rea
sons for the c hoi ce of suc h tests here: first, ice covers when loaded do exhibi t the
flexural behaviour of a plate on an e la stic foundation, Second l y, they provide an
easier met hod for predicting the tensile strength of ice [1] in comparison with uni
ax ial tests. Furthermore they eliminate the problem of interpreting tensile strength
by indirect methods [2] .

The first set of tests consisted of 180 beams loaded instantaneously in flexure at
different temperatures and loading rates. These are known to be the controlling
parameters for flexural strength and elastic modulus [3, 4]. The second set of tests
relate to. the long term behaviour of about 100 beam s loaded at different temperature s
I
and stress levels.
607
All thl' tC' sts c arrieu ou t u :-; ed (o lunlilar S2 typc IS! SI..'<t -i Lc h'itl1 . J s;ilinit y of 5% 0

'1'111:-; va l ue ,.. . ;t:; LhOSC Il be"': ~ lUSL) it is hc l c i vcu to represent th e ~ I verdg e sali ni t y in
th e .'\l',,:t i c sC~I - i(l' !(), 71,

2, 1 :~Jle r illlollt:J I se t - up

L:'IOO f:ltory icc \.1<1:; groh'rl ill a cjrcu lJr tonk ~It ~I t L'IlIPl'I':ltur e of -l (Je. I)etad s of
t his operation (:.111 he round in Refcrcllce s 2, 10 ~ I:), Th e S~ t y pe se a-icc, hlhich '.. . as
~1~)mlll ill thi...:h.llcss, incluu e d a 1:J)'cr (2 rnm tlli ck) of small cTys t a l s due to th e see uin g
pro cess. Th e ~ lve[,~lg c crys tal siz e js Gmm in uiaUleter at ~I Jista nce of 90mm helow
th e t op su r face anJ tIle!'!' op ti c ax i s i~ runJomly oriented i n a hori zo nt a l r1ane.
S' lllnit)' wa s ,ne ,/s u red inlilicuiatl y "ft cr cu tting t he be;]ms Ill.

Eac h forma tlon of icc In the t:IJ,k yie ld e u a bou t I S to 20 rcCL;]ngular beams. Th ese
bea m>. h 1c r c s tored In a fre ez er a t -30C. [lefor: testing Q be am, it \\';)5 brought back
in the co lu roolll for at l eas t f our hour s . ~Iachining of the beams in s ur e d th e unifor
mit)' of th e sec ti ons I<ili c il I<C're "bollt 92 x 92 min. As for their len ~t h th e bea ms
average d 750 111111 .

T he test rig useJ w~s rcportcJ by Mur~lt r2, 10 1 . Il o wcver. the f ollowi. ng imr roveme nt s

were l lltl'oduccJ (see re f e re nce rI31):

- str cngtl1: ning of th e rig itse lf;

heat i ng a nd in su l at.ion of the air jacks in order to mi n imi::.c their se ns itivity to


the tefllper;l tlu'C whe n louJing til e be3 m s~

- (Icljus t.Jble supports in order to insure u un i form contact ,a cross th e heams itt th e
s uppo r t s;
- I in car and acc urate control o f the lOClJ ing rate on the beams,

The experimen tal se t - up is s hahin in Figure 1 \<!hl' re th e s Pan \</;)5 fixed a t !l IO mm and
t wo I'o int loa ds at o n e- thi rd of the s pa n "'e re appl i ed ,

3. Shor l tenll load in g


3.I t xp e r i ment;l l pro( odur c
Aft e r ca l ibra tion of the appara tu s. ea ch bea m "a s s ubj ec t ed to nine successive l o ad
ing s ,dtil loadin g rate s va rying betl<ec n 10 an u 600 kr,, /s. The first va lu e rep r ese nts
a l Oh'er limit below \d l l ch c reep mu st be co nsidered and the hi gher va l ue refers t o an
upper bo und beyond "h ie h d y nam i l' a mplification cou l d be considered . The loadin g
unloadin g cyc l e time is about 1 ,55 J nd the n" rura l period of t he beam i s about .0 07s
[8J. I lcnt.:c t he dy namic ampl j fica t i on is Il L'g l i~eab l e jn thi s c ase.

608
r:ig, 1 - LxperilJl c llt ~1 1 sC' t-up.

P (N)
! t=-5C !20- 03-S0
800 Pf

sal i nity =5 %0

600

400

200

o
o O.lmm Central deflection (w)

Fig. 2 - Central d e flection for different s tre ss r ate s . Short term loadi ng.

609
Referring to the successive loadings mentioned earlier, it must be pointed out
that the stress in the beam never exceeded 2/3 of the elastic flexural strength at
lOa kPa/s. After obtaining different load-deflection curves for different loading
rates, as shown in fi gure 2, the beam was loaded all the way up to failure at a gi
ven load rate. Enough time was allowed between successive loads so that most of the
delayed elastic deformation was recovered. Finally it is important to realize that
the flexural strength at different loading rates and evaluHtion of the elastic modu
lus from deflections is based on linear elastic beam theor y . In this context, the
lOJd i ng rate (lin he expressed in terms of the extreme fiber ~ tTC ~S in th e beam~

11 31.

~FlcXllra l _ str<::'.F_1C.h.

The fle xura l strength ~ is a function of the loading rate and temperature . Figure
3 s hol"s, for different temperatur es, the variation of Of \,'ith a .hich is defined as
.
the rate of stress applic3t10n on t h e beam at the extreme f iber' Pi)
(0=61'" Eac h pOlnt
.

on the curve represents the average value obtained from a number of tests which is
also indicated on the Figure. These values correspond to the tensile stress at the
lo"cr fiber since the tensile ,trcllgth is known to be 2 to 3 times lower than the com
prce.ive strength 191. Figure 4 s hows the flexural s trength as a function of tempera
ture for different loading (stress) rates.

Examination of the results of figure 3 show that for stress rates a var ying between
10 and 100 kPa /s there ex ists a s ubstantial difference in the corresponding val ue of
~ special l y at lower temperatures . This phenomenon is also reproduced in figure 4
as an increase in the ba nd width for the lower temperature s . For loading rates above
100 kPa/s there is a slight uniform decrease in all va lues of ~ at all temperatur es
as shown i n figure 3. Furthermore, figure 4 sho"s a definite discontinuity in the
curves around -22.9"C which is the temperature at which the sodium chloride precip i
tates. This discontinuity relates not only to the loading rate but also to the temp
erature ~nd the purpose of fi gu re 4 is to emphasize the change in behaviour at
-22.9 C.

3.~ Ela s tic mo dulus


Figures 5 and (, s how the va riation of the longitudinal horizontal clastic modulus
with loading rate and temperature respectivel y. Each point on the graph represents
th e average of a number of tests for each tempera ture. The modulus \." a5 meas ured by
computing from figure 2 the slope of the curves bet\.Jccn t\1I0 s. tr . . . ss 1 imits.
These
stresses had a IO\'I'er limit of 4% and an upper I imi t of 67% of the flexural strength
Of ohtained earl ie r at 6- ~
100 kPals . This modulus is really the apparent modulus

610
CT
t
(kPa)~----------------------------------------------'

2500
5
17
ok
* *1 Salinity = 5%0

2000 ~fr\

1500 1.55 ~~~~


.64
4
______ ~3
cf. 5~ ~-2------------~~~:~:~:~~--~
1000 ~ ----------~7~----------~~~-~

~
. 5. .
~---------~n~
6.
!:
__________ ~~2~0~0~C__~~
6
500 - - ~----------L -5 0 C .:
Cr
(kPo/s)
o
o 100 200 300 400 500 600

F i!-!, :) - r: l C.\ ll r-; II :-:. tr l' lIg t ll \I'i :-,tr c:-;.:-:' r :1 t e f or differe nt t (, l1lp cra turc ~,
Shor t t ernl JO;IJill g, el a ~ ri l.: \J11 ;l ly~i s,

CT
t
(k Pal
2500 0 10 kPa Is
6
8
25
50

..o
181

2000
0 75

181 100

!II 300

1500 19 600

5%0
1000

500

o
o - 5 -10 -15 -20 -25 -30 -35 -40

ri g, 4 - 1:l ex ur<J! s t re n g th V~ tC'mpcr 'l t ll r C' f o r d i ffer ent ~ t ress r J tC's,


Sh Ol' t t e rm l o~d ing, e l os ti c :In ;II )' :-'is .

6E
E
(GPo) _40C

~
7

- 30 C
6 -20 C

5
- 5 C 0
~
4

3 Salinity = 5 %0

2 t t
k i

0 0
0 100 200 300 400 500 600 (kPa/s)

Fig. 5 - App are nt e lastic modulus vS stres s rat e for d i ff e r en t tempe 1'a
tu res. Elastic ana lysis.

E
(GPa)~------------------------------------------------'

8 o 10 kPa/s
8 25

7
: ~~
l1li100
ED 150 !.. __ ---
1-- - ~r--r ~E=0. 078t +4.29
6
o 200 __ ji? -50C<t~400C
$ 300 - - III 8 - r:r

3
5

4 'f o
6 __ ---
__ -
_~_--,~;' \E'~:'o", +2."
(40) -5C<t <_40C

-0
(40)
2 ( measurements)
Salinity = 5%0

o ~ __ - L_ _ _ _~ _ _~ _ _ _ _~ _ _~ _ _ _ _~ _ _~~ _ _- L_ _ _ _ _ _ _ _ _ _ _ _~~

o -5 -10 -15 -20 -25 -30 -35 -40 t(O C)

Fig. 6 - Apparent e lasti c modulus vs temperature for differen t stress


rates. Elastic analysis.

612
since the stress variation across the beam depth is linear. Also the modulus is a
function of the loading (stress) rate hence the average apparent value represents
the slope of the curve between the limits specified above.

Problems related to deflection measureme nts in the laboratory at _25C preclude the
evaluation of the elastic modulus at this temperature. These problems, however, had
no effect on the flexural strength.

Examination of figure 5 shows the dependance of the average apparent modulus on the
s tress rate c) for values bet"een 10 and 100 kPa/s. Thi s dependance is decreased for
stress rates between 100 and 200 kPa/s. As for values above 200 kPa /s there is a
slight linear increase in the value of E. Figure 6 s uggests a linear variation of
elastic modulus with re spect to temperature and not a square root vari ation as im
plied by the usual formulae based on brine vOlume calcul a tion [10]. The slope of
thi s function seems to be independant of the loading rate c) and for a fixed sa linity
of 5%0 the fo 11 owing equa tion s are s ugges t ed :

E .078 1t I + 4.29 for c) 600 kPa/s

E .07 41 tl + 2.46 for c) 10 kPa/ s

where E is in GPa and t is in C. No apparent discontinuity was observed around


_23C as in the case of Of .

4. Flexural creep tests


4.1 Ex per imental procedure
The experi mental set-up is identical to the one described above for s hort term load
ing except that the X-V plotter was replaced by an automatic recorder using punched
paper tape . Sublimation "as miniTllized by coverin g the beam with a thin polythene
sheet. Small circular pl a telet s "ere placed under the DCDT in order to prevent local
creep effects at the contact points.

Loading was applied at 100 kPa /s up to the desired level then was left con s tant for
the remainder of the t es t. For each tempe rature the applied long term load was
roughly 10, 25, 45, 65 and 85% of the flexural strength Of measured earlier corres
ponding to c) of 200 kPa / s. For each tem pe rature and stress four test s were pe rfor
med. Onl y the results of t es ts at -5. -20, -25 and -40C are availabl e.

613
r

4.2 ~e riment a l result s


Figur es 7 to 10 show the variation of th e non-dimensional deflection parameter wh/l"
"ith respect to time for temperature s of -40, -2 5, - 20 and _5C re spec tively. So me
te sts were by far longer than indi ca t ed on the figures wi th some mea sur ement s t ake n
af t er 10,000 min (7 day,).

For creep tests a t _40C, figur e 7 s how s three le vels of applied stress on the beRm s
name l y 90 0, 725 and 550 kPa. For thi s v(' r y temperature a large number of tests were
performed at stresses above 900 kPa where th ey all exhihit ed a brittle type fail ure
without an y Si gnificant in creas e in deformati ons. The table bel ow s hows the tlme
elapsed befor e failure took pla ce in the ,e t es ts.

0 OF
N U ~I B ER OF
tJU~IB E R TIME FOR

kPa TESTS FAI LURES FA ILU RE


(mi ll) i

I G.O 8 8 6 t ; -40C
I "i (l l ' 8 8 12 I instantaneous
1 100 <; 9 58 I 0, ; 2200 kPa
900 5 263
I
*AutO::1a t lc recording

The re for e any ductile beha vi our can only take place if the specified s tress is 100'er
than about 40-45% of ~. The same performance was observed for tests a t -25C
(figure 8) where the upper bound for a ductile behaviour is incrcac,ed to 75% of 0, .

The table be low summarizes Lhe resulLs of th C' ,'c teSLS.

0 NU~I!l E R OF N~IBER OF TI ~ I E FOR

kPa TESTS FAI LURE S FAILUHE (lIIi.n)

980 6 6 9 t -25 C

865 4 3 274
Instantaneous
750' 5 700 0, 1150 kPa
;

As for tests at - SoC (figur e 10) the long term behaviour indi cates c l early a failure
due to tertiar y c reep for s tresse s above 45% of ~ .

4. 3 Creep l aws
A creep law o f the power form E: ; Bon """, he IIsed t o represent th e Ion!: t (' rlll heha
viour. Figure II shows the rat e of deflection ~h /l" as a fun ctio n of th e applied
str es s 0 for differe nt t emper at ures. The s lope of these lines [II) repres ent th e
valu e of th e expo nent n for each temperature. A regression analysis indicates f or

611,
wh wh
""j2 ""j2

250tx 1(j6 loool x 10- 6 _ _ __ _ _ _ _ --,_ - - - _

Instantaneous Instantaneous

(Tf = 2200 k Po (T f =1150 kPa

t = - 40 C t= _25C

200 800
t
Sal inity = 5%0

150 600

Premature fai lure


at about 865 k Pa

400

Premature failure
at about 1100 kPa 200

++

115 kPa

o I
time (min)
,I ~
o~ time, (m in), I
o 500 1000 1500 2000 2500 3000 3500
o 500 1000 1500 2000 2500 3000 3500

Fig. - Non- di me n sio na l c r eep de fl e ction pa l'ameter Fig . 8 - Non - d imensi ona l creep defl ec t ion pa r ameter
wh/9,2 as a func ti on of ti me . Te mperature: wh /.~2 as a fun ction o f ti me. Tempe r a tu re:
'"'-"
I-'
_40o ( . - 2S o( .
7>
f-"
o

wh
wh j2
12 t--~------------------------~T
6

5000}X 10
1500
x 10
6
_ 20C /Y I

#1 t =- 5C
Instantaneous
Instantaneous (J f = 475 kPa

(J f = 750 k Po F
1250 4000L
= 5%
t l
Salinity 5 %0
Salinity 0 h :d

1000
,
3000

750

2000

500

1000

120 kPa

o t:::=-,
o 500 1000 1500 2000
; ir
7 j/~~i
kPg

2500 3000 3500


o 500 1000 1500 2000 2500 3000 3500

Fig. 9 Non-dimensional creep deflection parameter Fig. 10 - Non-dimensional creep defle c ti on parameter
wh/~2 as a function of time. Temperature: wh /~ 2 as a function of time. Temperature:
- 20C . _5C.
- -H), -25, -~o ,llld -S oC :1 va l ue of ~ .O ~~ 1. 7:1. I. B() :IIlJ ~.Oll 1"C:-;lh:~: tjvcl .\' . Il nh'l'vcr ,

it mu s t be po int eJ out tlwt th e :-;e v: ilu cs h~ \l c th e followil\~ 1 i11li t:li i o Il S :

;t) Nor ton' s 1;]\,' i s an .:Ip pto;\ill\~ t l' for illulati on of t he so-c:Jllcd st: l ti oll:lry cree p
\-.' l1il.:ll c a u s es th e i.ntrtc.ate prob l em of mcasuring tltc s tLl in (01' {kfol'm ;lt i on) ra t c for
c:Jell stre s s l ev e l [ I ~ I. In f;J.c t this 1':1t (' decreas e s ul'ltil t l..'r tiar y cr eep i s t ili tLJ
t e d. For vcry 10h' temperatures and hig h s tr e s sc s thi s con s tant st ra til r;i tc i s no t
easily obtai ncu du e t o tile brittle beh:lviour of til e ic e. This is 5hO"' 11 in figu re II
by the sl13ued re g i on. On th e othe r hand if du c t i l e be haviour is obtain eu , " st a
t io nary creep i s no t obvious unl ess the cx pe rilue nt l s IIIrli nt:l lncd for ~l vcr y IOl1g
tiJlle. For the t est s at _5C the secondary cre ep ran ge is ver y s hort " nu poor l y
d efin ed for stresses be twe e n 45 to 85\ of ~ [ n fact No rton' s l.:l\'; is not appl i c.:J
ble since tc rtia ry cr e ep in th is cas e is qllit e inlport ~lllt.

b) f or t e mp era tures be l oh - 23 C brittl e f~i lur e is e.pec t ed fo r a fairly wiuc range


of str e s s . Howeve r, this limit is not ye t ve ry pre c is e l y known.

c) fin a ll y for t es t s a t hi gh temp eratu r es or under 5",,111 st r <:,os es the effec t of the
dead load of the bea m ca n be quite si gnificant.

5, Con c lu s i o n
Flexu ra l s hort ter m test s on beam s a t diffe r ent tCII' p,' r a tur e s t "nu l oad i ng rat es a
y i e l de d th e followillg observation s:

I. Th e flexural s tr en gth Ci, f or 0 > l Oa kPa !s dec r en s es in a 510," lin ear fash io n
for a ll tem pera t ures . For lo ading rates 10 ~ d ~ 100 kP a! s ~ varies co nsiderab ly
with a maximum va lu e a t a s tr es s rAte around SO t o 75 kPa!s. This varia t ion is JIIo re
pronounced at 100;er tempera tur e s .

2. At -23C th ere is a discontinuity in the valu e of 0, an d i t s varia tion with


res pe ct to O.

3. 111 e a verage a pparent e l astic modu lus increases rapidlY with 6 up to 100 kPa !s .
Beyond this val ue the increase is fairl y lin ear. Fo r a give n str ess rat e , th e
value of E is a linear func tion of th e tempera tur e.

As f or l o ng term t es t s th e major con clu sions a re summarized be low :

I. The de flec ti o n para met e r ~h !1' de cr eas e s continuously even after 7 day s of cons

61 7
10- 6 x _5C
8 _20C
C:. - 25 C
CD _40C

Salinity

.k *
Area of instantaneous
or premature failure

100 1000 (j ( k Po)

Fig. 11 - Non-dimensional creep deflection ra te \~h/ R.2 vs


applied s tress a~ different temperatures.

t;)llt l O:I<..lillg for t c mperatures of _40, -25 and _ 20C. This ob serva tion creates
:'010(.' difCicllitics for d etC'rmina tion of the secondary creep zone.

Tl'rf i; lr ~' creep is im porta nt at temperature s of -SoC I.. h e n the stres!=> level in
hC:IIll:-' c'cl'c1s -l~ .... of the inst:lnt clneOU:-i fl exura l s tren g th O

.), S~'srl' m; ltic prcfIlature bri ttl e f;li lures, \"i thout signif icant amount of deformation,
;1}"C ol):,crvcd for t Cm p cr~ltllr ('~ of _25
0
n nd _40o( when the stre::-;s level in the beams

1' .\<':l'l' US 7~', ;I"d 4S~, o f or rcsrc cti vc l y.

";:1 : n::'C;\fl'h \":;1:' ~ poll:-:.or('d by .\S!: RC grrtnts 110 i\ -8~l S 8 an d RO-823 which the aut hors
\\'ot lld I i~(' to gr:1 r erll ll ~ ~c.:,,"no\<Jlcdgc.

618
References
[I] Weeks Iv.F., Assur, A. (1967) "The mechanical properties of Sea-Ice" CRREL.
Co I d regions Science and Eng. ~Ionograph II -C3.
[2] Tinawi, R., Murat, J .R. (1978) "Sea Ice testing in Flexure". Pr oceedings of
the 4th Int. Co nf. POAC Memorial Univ. of NewfoW1dland. p. 638-653.
[3] Butkowich, T.R ., (1958) "Recommended standa rd s for Small-scale Ice Str ength
Tests". Trans. Eng. Inst. Can. 2 p. 112-115.
[4 ] Tabata, T. (1966) "Studies of the Mec hanical Properties of Sea-Ice - X".
The flexural strength of s mall sea-ice beams". Contribution no 793 from the
Inst. of Lo w Temp. Science. Sapporo, Japan.
[5] Michel. B. Ra"'Seier, R.O. (1971) "Classification of River and Lake lee" Can.
Geotech. J. Vol. 8, pp. 36-45.
[6] Cox G.F., Weeks, \'i.F. (1974) "Salinity Variations in Sea-l ee ". J. of Glac.
Vol. 13, no 167-1974, pp. 109-120.
[7] Frederking R.M.W., Timco, G.W. (1980), "NRC Ice Property Measurements During
the Canmar Kigoriak. Trial s in the Beaufort Sea. Ivinter 1979-80". NRC-DBR
Pa per no 947.
[8] Biggs, J .M. (1964), "Introduction to Structural Dynamics . ~1cGra"-Hill, pp.
205-209.
[9] Butkowich, T .R. (1956), "Strength Studies of Sea-Ice", USA, SIPRE, Research
Report RR 20.
[10] Murat, J.R . (1978), "La Capacite Portante de la Glace de Mer", Ph.D. thesis.
Ecole Polytechnique, Montreal. Report No . EP78-R-49.

[Il] Hult. J .A.H., (1966), "Cr eep in Engineering StructJres". Blaindell Publ i s hing
Company.
[1 2] ~1ellor, M., (1980), "Mechanical Propertie s of Po lycrysta lline Ice". Ph ysics.
and Mechanics of Ic e. IUTAM Symposium. Copenhagen 1979. Springer-Verlag,
pp. 217-245.

(13] Lainey, L., (1981) "Etude param!trique des propriet!s mlcaniques de la glace
de mer - Resultats d'essais". Centre d'Ingenierie Nordique de l'Ecole
Po l y technique. Research Report No CINEP-665 -218 .

619
PARAMETRIC STUDIES OF SEA-ICE BEAMS
UNDER SHORT AND LONG TERM LOADINGS
by L. Lainey and R. Tinawi
DISCUSS ION S

Discussion by N.K. Sinha, Nation a l Research Counci l o f C~nada

The authors included too many test reports and therefore co uld not elaborate in detail.
It is not c l ear how the specimens were machined or whether there was any brine drain
age during this process a nd during the "four hours" of co ld room storage, espec i ally
a t -SaC. Severa l ques tio ns come to mind:

a) what were th e orientation s o f the gr a in s with respect t o the h ca ~

and the s tress axis?


b) how was th e s urface finish of th e s peCi mens?
c) how was the l oading rate controlled'
d) how linear was the loading rate?
e) how was the load maintained constant during creep test?
f) what is meant by "dyn amic amp l ification"?
g) what is meant by "the tensi I e s t ress is knolm to be 2 to 3 times
lower than the compressive stress " i n the second paragraph, page 4?

While linear elastic beam theor y can be a ppl ied for very short-term loading conditi on s ,
there are ample discussions in the lit erature that the stress distribution in a beam
is far from linear under loadin g condition s for which the highly non-linear viscous
flo h' takes the dominant role (e . g ., Finnie, I ., and He ller, W.R., "Creep of Enginee
ring Materials", 1959, ~lcGra,,-Hill Book Co., N.Y.). Norton's law for minimum creep
rate ca n not be e xam in ed with beam bend in g experiments withou t an apriori assumption
of th e s tress ex ponent and he nc e the stress distribution during th e required creep
period. How did th e authors est ima te 0 in the creep equa ti on E = BO"?

In discussing th e experimental procedures for Fig. 2, it was pointed out (fi r s t para
graph, p. 4) that " the s t ress in the beam never exceeded 2/3 of the c la s ti c fl exu ra l
stre ngt h at 100 kPa/s". Th e same Fig. 2 was used to es tima te e la s tic modu lu s for
different stresses which "had a lower limit of 4% and an upper limit of 67% of the
flexural strength Of obtained ear li er at 0 = 100 kPa/s" (las t paragraph, p. 4). This
makes the paper difficult to read a nd the resul t s d iffi cu lt to analyse in spite of
th e fact that the discusser wou ld be in te res t ed t o ap ply his theory (Sinha, N.K.,
POAC 81) to these results. Why not s impl y s ti ck to a given stress level and say so?

620
Di sc us sio n by Jean-Paul Nadreau, Universit! Laval

Lainey's experiments on sea-ice beams app ea r as a use ful contribution to th e knowledge


of th e me chenical properties of s ea -ice.

Short term loading

As f ar as s hort term loading experi ment s ar e concerned, some comparais ons with fr es h
wat er i ce can be made.

The fir s t point to be noti ce d i s probabl y th e s ame e volution of ~ with re spec t to


s train rat e (or loa ding rate ). The bell shape s hOl,in g an increase of th e flexural
s trength unde r 100 kPa/s seems more s hifted t oward s the ductile regions th an in the
case of fres h water ice.

For s now ice (L a fl e ur , 197 1) the bell s hape s pan s from 10- 3 S-l to 10- 6 S -l ".he re e s
5 - I
it seems to s tart und e r 10- S for s ea-ic e beams test e d by Lainey.

Th e oth e r important remark to be made concern s th e evolution of the flexural strength


with temperature. Since we know that t ensile stre ngth is quasi ind ep endant f r om t emp
erat ure variations it is intere s tin g to see th e increas e of 0, wit h th e t e mperature.
The given e xpl a nation dealin g with sodium chloride prec ipitation is proba bly the mO s t
import an t on e : it would leads us to con si der the ic e under 22 ,9 "C as a totally diffe
rent one than that ove r 22 ,9 "C. Brine pock e ts must be loo ked a t as nest s for fatal
cr acks. Gra phica lly, a di scont i nuity in the Curve s would have bee n quite aCcel)table .

Fl exur a l creep' t es ts

From fl exural creep t es ts , the Curve s obtained ar e qu i te coll ere nt with s inli l ar previ ous
work s on fresh a nd salin e i ce beams .

The terti a ry c reep obtain ed a t high t e mperature s ilre to be noticed. It s eems from the
gril phs, that th ese beams did not show noticea ble se cond a ry c r eep end that the s train
softening proces s t oo k ove r rother rapidly. If we r e fer to th e typ ic al c reep cur ves
desc rib ed by B. Mich e l ( 1978) th e tertia ry c reep in vo lv e d he re ca n be of th e a or 6
typ es. Sin c e the s train rat e increase is not fo llow pd b~ a sec ond s tag e of pc rmal1ent
creep (c har ac teri s ing s Yntectonic recrys talli za tion of the 6 t ype creep ), it could be
con si dered 85 a tertiarY cree p du e to the multi p lication o f the di s location (a typ e ).

621
Finally, S. Michel' s equation (1981) has bee n app li ed to the se result s with its sim
plif ied form; i.e.:

The fittin g of t his equa ti on to 9 of the curv e> leods to values of th e coefficient 0
of mllltipli c;l t i oll of the dislocatiol1S between 3 and 20 which is til e sa me order of
rna~~n itud e than these obtained on f r es h h'atcr icc bea" :.::. ( \!adrcau rind \iichcl, 1981).

Hc f c r c nc c$
._- ----

Lafleur, P. (197 1) "P ror ri ctes nlfcaniqucs de 13 g l ace de ncig c en flexion"


'Il)sc :'I.Sc. Uni verstt" La va l, Queb ec, pp . 36-48.
'Ji che l, R. (19"~ 1 "Icc 'l l'c il'IO ics" I. e' Pr c'ses de l'Univ ers ite Laval, 499 p .
.'. Iichc l, R. ( 19 XI J "A dva nces in Icc .\1 Ch:lll ics" Proc eed in gs POAC-Sl, SC's . ion /\2, Queb ec.
,aJrC',w, .1.1" . and , Ii eh e l, B. (19S 1.1 "Crc'(' p of 5, f cc Scam, and Plates" Proceedin gs
!'OAC -. I, Se-;. i on A4, Qu(be e.

~" t i on a l Re sc or ch Co unc i l of Cnn ad a

This n:lpc r is an o ther c ont ri h lHion in an on-goin s o;r ies of inv e stiga tions of flex ur a l
bch3vlou r of sc~ ic c bt.::t ll ~ carried out at Ecole Pu l y tcchnique. A \\'e.11th of data Clre
p r c~cn t l' LI on ';l rc flll l y perfon cd b cn m t c :; ~ .

Strr.~ l'atC' is used ~s an in dep e nd ent v3 ri ~ble in pre se ntin g s trength and modulus
r c~u ... . Co ulJ th e .1 u rh o rs desc ribe hm. stress rate \.,' a5 defined (i . e., hOh' \"as r
csr, bl i<hcJ )? Al sO was s tress rate sp ecifically controll e d? The st rength l oading
rate behavi our shown in F ig ure 3 is s imi lar at l C<.J :-:., t a t 10\\' temperature to that
oh~c l' 'cJ by Drouin and "I i ci",l [II. They a ttribllted the variation to the deve l opment
of ;l r1a -; ti t: mo ment over 3 certaln ran ge of IO:1di n,~: rates. Old yo u se e any evidence
n This in the sho rt term loa ding tests ?

I'll<> tr e ncl s in ~pp a rcnt c la s ti c modulu s ( hg. 5,1i) ar e cons ist en t , at l eas t in a quali
t"tive ;-" , h ion, with predictions of Si nha [21. Could the authors comme nt on this.

622
ThL' author'S ;lrL' to hL' I.:OIIIIII('11U (.'U for th is cOlltrihutiorl. It provides ;1 vcr.v u -.:.e ful
cOlltrih {l r iOil 0 11 ;ll'pre ciatioll of sea ic c hL'h ~vio llr ullucr flcxlIru l loading ;Inu \.,. i I I he

;I v.:.lluuhl c s ource for future ~ L)lalytic.:.Jl wurk.

III Ilrou in , ~i. , ,,,,ei ~Ii chcl. II . (1 972) - I.a resi s t a nce e n flexion de 10 glace Ju
Saint-Laurent Jetcrminec en nu t ure. Happort ~CT- 72 -09-26, DEp. de Gpni e Civi I,
Univcrsit c Lavu l , Quebe c.
121 S inl,", N.K. (1979) - effec tive elasti c ity of icc. In Proceedi ngs of Wo rks ho n
o n Bear ing Capacit y of I c c Covers, 16 -1 7 Octobe r, 1978, Ivinnireg. National
Hesearc h Counci 1 of Ca nada , Associate Committee on Geo t echnica l Resea r c h Tech.
Memo . No. 123, rr. 112-123 .
[ .> 1 Nadreau, .J.P . ( 1976) - "Et ude du Fluage de poutr es de glace columnaire". Univ er
s itc Laval , r.c ni e Civil. Re s earch Report No (;CS -76-04.

Discu.s ion by P .R. Iry, Re ~ea rch Department, Esso Resources Ca nada Li mited

!'igure 3 in th e raper e xpre sses be nding mome nt a s a functi on of s tress rate and s how s
that the b e nding moment in c r eas e for cold t e mpera ture and s tre ss rates near 0.1 MPa
S-l Howeve r the bendin g mome nt is expre sse d as an extreme fibre s tre ss, calcu lated
from an e l as ti c analy sis . Since thi s is near a range wh ere delayed el as ti ci ty or
V1 S COUS behavi our or rla s tic deforma tion is occurring it seems unlikely that thi s is

an acc urate represe nt at i on of the ex tre me fibre stress. Taking account of stres s
distr ibuti on throu ghou t th e be am in non-ela stic bend ing would perhaps eliminate the
peaks in th e ex treme fibr e s tres s represented in Fi gure 3.

623
I~L'pl y t o t he di~t'u~:-:.i o ll h y N . Sill ]);l

th e ll :-; L' o f the c,\per i mcllt: J! !'L';o;ult:-.. ~l:Jlly of th e points ui :-;l"u~ ~cd have hcc lI omi tted

due to 1:ICl-. of SIWL"c. If th e very (':o ndL' lhL'd for m of rhe P~IIH.,.'r did cOlltribut e to ;Iny
conftl:-;iOll l t ilL' al.ltllo r~ woul d lik e t o ~ 1I)Ologi:-:l'.

TilL' cO lllpl e t c L'>q)erillll'Ilt;1 1 p roccdurc :l~ ~\rL'll;l ~ qllc:-,tion:-; (:1). (b), (e), Cd ) and reJ
C; II' be found in " ver y dl't~i l ed rerort jUq pub I i,ilcd (I ,oin ey 198 1), 1\, for questio n
(f), th e dyn:lmic aJllpl i fjl' ~ltion i :-; UL'fiIlCd , i n any :-;t ..llld;JrJ te x t on strl.lc r ur<11 u y Jl a

lUi es, ,1:-:' thL' :lmpl ifi catioll of th e ,..:.ta tic response in oruer to sim ulate th e dyn;)m i c

beh:lviollr o f t h e bC:lnlS under rapid ! oa uing ( 11iggs 19(,4). ()ue:-;t i o ll (g): thi s j"

un fortU!l~l tI)' ~l t yp ing er ror where cO lllprcs:-;ivc "rr Cll g th, f":lthcr than stress. is 2-3

times ,l.!,I'eatcl' th :JIl t ensi I e strc ll !.! t h.

ThL' authors arL' full y ,,1\":lre o f the linL'ar clastic bcom theory \"thi ch is on ly ;JJ1plica

ble to in s t :lflt anL'oUs l oa uin g. 1\:-:' for rhe ~ t J'(.' ~~ tli~trihut ion throll,Shollt the b ea m
un uer cn,'cp ! oaui ng, they do vary first bL'cJUSC of stres s r etl i stribu tion in th e ini

ti ~ ll l)Il;ISeS of r r c('p ! oaJi ng ~]nu ;ll~o bccltIsC' of the scronu<.Jry crec r. lienee a ny
l"e fCrCncL' t o the "I in c;)r ex treme fiber stress " h as bec- n mo de by th e ~lutho r .:.; as an
indiclt or of th e l oad intcn :-; ity ill the bC~ lm s. Furthermore , nowh ere was it impl ieu
in rhe paper t h a t Nort on' s 1[1 \.,. {. c.. Ha" is the o nl y 1 ~lh' :l vu i lubl e or the on ly o ne to
b e u,cd.

Finall y regardin g the s t ress level useu in th e cx pcriJnent:J1 t ests. it \"tus f elt that
t hese s tresse s !lou to be exp ressed as :1 pe rcentage of the in ~; t:lnt a n co us strength

v.1luC' ~ imply bec;Juse of it s variation with t em perature. For insto.nt:1I1L'OllS t es ts the

11 and f17"" limits were us eu for computing th e seC;:lnt mo uulus. For crc e r t csts, ui ffe

rent s tr (':-:.s c s were upplieu on th e beam ~lS a pL' l"Ce nta ge of the fle xura l s t rength value.

All th e dct :,l1 , rl'l oted to th c >trc:;s voluc uscd for cilch exrerime nt con bc found in
tl,c rc' port (Loillcy 1981).

624
I

Reply to the discus sio n by J.P. Nadreau

The authors wi sh to thank the di sc uss er for hi s interest and comme nt s .

Short-term l oadi ng

The flexlJral stre ngth varies wi th the loadin g rate and the be1l- shaped variatio n is
par t i cllLil' ly sig nif i c ~ nt for temper atures 100,er th an _2 31. Furthermore, a different
c ry s talline s tru ct ure for the ic e t es ted as well as the t ype of t ests do inf lue nce the
variation of flexural stren gth with loading r a te. The remark conc erning t he qua s i
inde!,endancc of th e tensile strength \; ith temperature is valid for fre s h wat er ice and
for t emperature s higher than _20 "( (Michel 1978). This can be exp lain ed by the fact
that the crystallographic structure of fresh "ater icc is independant o f temperature.
Therefore a constant ten si Ie strength i s not s urprising \"ithin the range of tempera
tures tested. An extrapolation of the re s ult s to values lower than -20"C has not yet
been demonstrat ed ex perimentally in th e above reference. For sea-icc, the fle xura l
strength increases with temperature s be low _2 3"(. An approximation of the curves
into two s trai ght line s with a change in slo!,e around _23"[ would be more acceptable
than a di s( Ol r inuit v implying constant va lues for the flexural s trength.

, I ~ XU r: l i ,orc \.'j1 t est s

Beams t es t ed a t low temperatures with hi gh s tre sses did not exhibit an y se condary
creep. The tertiary creep developed i s Kv type (~Ii che l 1978). For beams tested
under l o,,'er stresses, the creep dev e loped appear to qualify for the c!.v type ('hchel
19 78) . As for the Michel's equation (1981) where the va lue of &varies between 3
20, it prov id es a good approximation to our experimental data. Howe ve r, thi s equa
tion doe s not constitute a ge neral law.

Reply to the discu ss i on by R. Frederki ng

The author s are thankful to the discu ss er for th e ver), useful comment s pro vided:

First of all the stress rate has been defined on the third l i ne of paragraph 3.2.
This stres s rat e was controlled by controlling the loading rate and as s uming a
line ar relation between ~ and a as given in the paper.

625
With respect to the plastic moment development (Miche l & Drouin 1972) the following
points can be made:

1) a very simplistic "ideal" plastic moment is probably very difficult to apply to


sea-ice beams because of th e brine pockets, the non-homogenity, the anisotropy
of the material and its diffe rent strength properties in tension and compression.
2) the neutral axis in ice beams is fairly c lo se to the middle of the beam (Nadreau
1976). Hence the development of a plastic moment is not very probable simply
hecause the tensile strength once obtained lead s directly to the failure of the
beam.
3) The stress diagram is fairly complex throughout the beam height. First becau se
of the varia tion of the e la stic modulu s "i th s tre sses. (These are much lower
near the neutral axis). Seco ndl y the s tre sses are in turn dependant on the
loading rate.
4) other factors do also influence the formation of a plastic moment namely the size
of s pecimen s tested and the cry s t a l size.

As for the comparison of the results of the apparent elastic modulus with those of
S inha (1979), it is quite difficult. Fir st because of the influence of the crystal
.ize and secondly the graphs shown by Sinha are pre sented with reference to time and
.tresses whi Ie the results obtained i n the pap er are with respect to loading rate.
Oespite these differences, the results appear to be qualitatively consistent even
though the use of the logarithmic scale by Sinha tends to flatten the curves at
h~gll stress Y"ates.

Reply to the discussion by R. Kry

The author. appreciate the interest of th e discusser relatin g to the interpretation


of our experimental results. It s hould be borne in mind that the main intent of the
paper "as th e pre se ntation of the experi me ntal results. The theoretical investiga
tioll i s currently undeTl,a y and "ill be fully r e ported in a Ph.D. dissertation.

A numher of points have been raised by R. Fred erking in his discussion. In addition,
however. the authors would like to add the follo"ing comments:

I) \~C agree that the stress diagrams i s non-l inear due to viscous behaviour of
the nl;;terial.

626
2) the extre me fiber ~tre :-;:-;, o ht ~ ine d u...;jng (' I a~ti ~ I i n c:!r theol' Y , is o nl y an indi

c:Jtor for the str('~s V<.l l ul' ra th er th ~ in all i nJor~em cn t of lin C'~ r ela s tic thcory.
It provides in any case a simple pra ct ic a l mea n of evaluating the apparent

clas t i c moJuJll s in fle xural tl' '-> t :'-.


3) the non- l i1l ca r tr C'a tmcllt o f thc :-:.tJ'(... :':-:c~ throu gh the heam l.,ri II proh a bly tend to

fl:ltt c n out th c peaks in I"i gurl' .). Iloh' cve r, it Oliist h e llotC'J fr om the same
ri gurc th~lt thes e pcaks occ ur :I t I OI\" t <:> mpcratllres o nl y. On e hOllld th erefore

expect :-:. a more pronounc eJ visco - c l,l Stic hclwvlour at th e h ighe r t cmpera tu res.

In fact the reverse i s Iwppcning!

627
FRICTION AND ADHESION OF ICE

Pekka Oksanen Technical Research


Research Officer Centre of Finland Finland

ABSTRACT

Adhesion between ice and several materials has been measured. These materials were
steel, concrete, wood, some types of plastics, and some types of coatings for marine
purposes. The kinetic friction between ice and ice ,las also measured. The te1nrerature
range in the friction measurements was -l"C ... -lS"C and in the adhesion measurements
about -2"C ... -20"C.

In this paper the mechanism of ~inetic friction has been considered by developing the
theory based on a water film on the contact surface. According to the theoretical
treatment presented the coefficient of kinetic friction can be calculated, if
hardness of ice, and thermal properties of ice and the material concerned are knovm.
In practi ce, however, al so other phenomena than the water fi 1m mechani Sin may be
involved.

628
1. INTRODUCTION

The ice forces are transferred to structures through their contact surface at which
the transfer mechanisms are pressure, adhesion, and friction. Because the ice loads
on structures are determined by these factors, it is important to know their
magnitude and how they depend on other factors. In this study, friction of ice and
adhesion of ice to various materials have been investigated.

2. EXPERIME~TAL

2.1 Friction measurements

The apparatus for measuring kinetic friction consisted of a turntable on which the
ice was moved, and a fixed and a moving arm [6,7]. The test specimen was nttach~d to
the moving arm and it was connected with a thread to a spring on the fixed arm so
that the spring compensated the frictional force. The movable end of the spring I,as
fi xed to an inductive displacement transducer which gave a signal to a recorder.

Test were carried out on several materials (ice, steel, concrete, wood, and some
plastics and coatings), but because ice is the most interesting one from a
theoretical viewpoint, only the results for it are presented here, the results for
the other materials have heen presented elsewhere l7 J.

The test specimens were 225 11111 long, 51 ITTn wide and 17 nrn thick pieces of wood, on
which a layer of ice, about 2 1Tl11 thick, was frozen. I)uring tile tests the ice side "as
down. The ice used in the experiments was produced from tap water.

The resul ts of the tests have been presented in Table 1 and some of them in Fig. 1.

2.2 Adhesion measurements

The test materials were the same as in the friction tests. The test specimens I'Jere
made by allowing a plate of the I11dterial in nuestio~ to freeze in a c.,vity, \-/hich 'las
made in a block of oolystrene foam and filled witll water. In this I~anner a specilnen
(size ca . 70 rrrl x ?Ol) rrrl, thic~ness ca. to ~1'1) accordirg to Fig. ? I,as obtained,

The tests were carried out in a test machine by pulling the test plate up>lards awl
holding the ice in a special frame so that at the contact surface ~ain1y shear stress
was generated. The loading rate !;as approxil"ately 0,01 ~ 1~ /m 2 s.

629
For each materi al 6 28 tests were performed at tel'1peratures -2 ... _20C. The
coatings were appl ied on steel speci (lens.

Table 1. Coefficient of kinetic friction, ice aga i ns t ice.

Coefficient of friction x 10 3

Temperature _15C _5C _1C


'Jormal load 10 N 30 ~ 45 tj 10 N 30 ~I 45 ~J 10 fl 30 fl 45 N
52
31
28
26
18
17
15
8
6

0,5 m/s 2B 18
15
25
13
15
10
12
10

33
20
15
19
20
16
15
11
10

3fl 25
20
21
15
13
20
8
6

1 m/s 20
16
15
21
IS 10
22
13
12

24
18
15
19
18
13
22
14
11

u
27
20
18
17
13
14
23
13
11

0
(lJ
2 m/ s 14
14
15
22
13
III
25
18
16

>
18
18
16
23
17
14
26
17
14

24
19
13
22
18
16
26
16
12

3 el / s 12
17
16
22
17
16
30
21
19

11
21
18 27
19
17
30
21
18

--

~
0060 f--- - +- - - t - - - - f - - -t - - ----i

0.050 f--- --f-


1

-15C -I/V.
0040 _\ / -t--- - - i - - -+---- - I
0.030 _5 C yeqn (12) ~
0020f--~;:=:::~:~~..,.-----:~~==--.+...--__..j-----J
-l~_Vv ~
0.010 f---'----t--'--- --t------1r -- - t -- - - j
O~--~----~----~---~--~
1 2 3 Vel 0 cit y m/s

Fig. I. Coefficient of friction of ice at different temperatures, normal load 10 N.


(The same ice specimen was not used at all the different tenperatures).
630

Test ~
specimen ~

35mm
Ice

>r /' /1)35 45mm

1/ / I. :/70mm

Fig. 2. Test speCimen in adhes ion tests.

The results are presented in Table 2, where the numbers have been obtained by
cal culating from the mean square fit thr ough the adhesion-temperature observations.

Tabl e 2. Adhesion in MN/m2 at different temperatures.


11ateri aI Temperature
-5'C -10'C -15'C
Steel 0,46 0,46 0,47
Con crete 0,35 0,42 0,48
Wood (pine) 0,30 0,29 0,28
Nylon 0,09 0,13 0,17
Polyethylene 0,09 0,13 0,17
Polyvinylchl oride (PVC) 0,11 0,16 0,20
Polytetrafluorethylene (PTFE ) 0,07 0,10 0,13
Coatings
Inerta 160 0,21 0,28 0,36
Intertuf Epoxy 0,13 0,25 0,38
Intertuf HB Vinyl 0,15 0,27 0,38
Interchlor HB Primer 0,18 0, 31 0,44

3. CALCULATlmJ OF FRICTION

3.1 Friction due to a water layer

The very low fri ction between ice and different materials has been shown to be due to
a water layer, which is formed by frictional heating f2, 3,4) .

631
In the follo,lin9 the coefficient of friction will be calculJted assuming that the

water layer is the only cause of fri c tion .

AsslJmin'l that the frictional force FII is caused by viscous shear in a ..ater layer
betwee n two materials, it is ohtained from the equation

F dv A v
II
T A 110 OJ = no' d . A (i)

whe r2 T i s shear stress, A is contact area, "0 is vis cosity of water, v is velocity
of the slider and d is thic~ness of the water layer.

The coeftlcient JI I,'iction: , = Fw/FN, where cN is the normal force, can be

calculated by finding aporopriate values for the vari a bles in equation (l).

The thic ~ ne ss of the water layer is calculated on the hasis of the energy available
for melting ice under the slider. The energy produced by friction is consumed not
only for melting ice hut also for conduction of heat into the slider and the ice.
These energies can be calculated approximately as fo llows.

It is assumed that a rectangular slider i s moving on sm0th ice at a velocity v (Fig.


3). The temperature of the contact surface i s at the melting point of water (which is
slightly below O'C because of pressure). nuring a time interval b/ v the frictional
energy produced is

(2 )

Heat transfer from the co ntact surface hoth to the ice and the slider i s considered

to he transient. For ice this is obvi0US but for the slider it is true

~ / V

Sol i d 2
~
Solid 1 j j j j /'

( ice)

Fig. 3. Schematic drawing of the point of contact.

632
I

only when the time of measurement is short cOl1pared with the time to reach therf'lal
equilibrium in the slider. In this study the slider was relatively hi9 ann till1c of
measurement short so that this assumption is justified for this case.

The heat conducten into the two solids can be estimaten by assuming t.he transient
temperature distribution to be linear in the nirection perpendiciJlar to the contact
surface. The amount of heat conducted into the solin I (ice) during the time interval
b/v is then

b ( 3)
6
v
where Al is heat conductivity of ice, fi T is temperature di fference hetween the
contact surface and the bulk sol id and ,\ is thickness of the layer into which heat is
conduc ted.

Temperature

I Contact Rea l temperature


surface j, distribution
I / \ I
I / I
I // \. -----rApproximated temperature
I / '\ 'I bistribution
I / \
/ \
I /

I 02 I
I
Solid 1 I SoLid 2

Fig. 4. Schematic drawing of the tef'lperature distributions at the contact surface.

On the other hand this amount of heat is equal to the energy stored in the heaten
layer

llT
PJ . abO . c l . --z (4 )

where ~l is density of ice and c is specific heat capacity of ice.


l

From Eqs. (3) and (4) the thickness can be elill1inated 9ivinq

ab . liT . ~
Tv (5 )

6JJ
The equation (5) is in agreement with the exact calculation of Evans et al. r 41 for
an elliptical contact area, except a rather small difference in the constant term
(0,98 instead of 1/12 = 0,71 in Eq. (5)).

A corresponding equation is obtained for the slider, and using suhscripts 2 for it
the total heat flow by conduction is

(6 )

~1elting of an ice layer of thickness d requires energy Om

(7)

where h is latent heat of melting for ice and Po is density of water.

Combining Eqs. (2), (6) and (7) yields

wFNv 0 ~ = ab . li T 0 /~o (iAlclPl + /A c P2)+abdhp


2 2 o
(8)

from which the water layer thickness d can be solved

d = h~o [W:~i - liT If (/AlclPl + I Ah P2J (9)

Taking into account that F w = wFN and putting d from Eq. (9) into Eq. (1) an equation
for w is o:,ta ined

(10)
o
from which

(11 )

/ , .c-A- ) 2 + n vAh p a'


'2 2, 02 0 0
2
Fir

This equation may be written into a l'Xlr" useful form by making some assumptions about
the contact surface. The total area of contact consists of a number of small contacts
at asperities, but for thi s calculation it can as \~e ll be treated as one

634
continuOtlS su r face . Firstly, it may be noted that the surface area A is e~ual to abo
Second l y, it can be assumed (even if it is orobab l y not 'luite eXAct), that tile form
fN
of the contact surface is s'luare, i.e. a = b. Using notation '-1= A and conse'lue ntl y
a = b = r~' , equation (11) ca n be written in the form
ljl

.,
LIT lTv (/ '~ l ~~ + ,I )'2cZ P2') +

(12 )

FN
T~e oaral'leter Hl "A- can be intcrpreted as th2 indentation ilardness of ice in cases
wilen it determineS the contact area A. 4o\vever, ~lllen t~e temrerature difference is
s,Olall (less t~ an about IO C) an 1 the ~/dter layer correspondinaly thic~, the contact
area is enlar g d and t ~e thic~ness of the wale r lay er decreased because of spreading
out o f ',ater as sho wn in Ref. 7.

3.2 DISC USS /OtJ

Two teMperature rcgions I'/here the be1avior of Friction is ~I)ite different can he
distinguished. lihe n the tePlperiiture di fference ro T is l ar ge (g reater than ahout 10C)
the co effi ci ent of (term due to he at conduction) in E ~. ( 10 ) is rlominant over the
constant term (due to viscous sh ea r), an~ the coefficient of friction is simpl y

(13 )

T~e saMe reS'Jl t is oht ained by Dutting the wate r layer thickness d = 0 in Eq. (9).
This mea ns that on the contact surface th ere is a very thin "late r l ay er ,o/hich is
self -balan ce d, i.e. if the thic~ness of it >!ere decreased, the increasing frictional
he a t ",oul d ~nel t 1Il0re \o/ater, and on the other hand if the water thic~ness '../ere
increase d tile reduction in the frictional heat ,lQu ld cause a temperature drop at the
cont act be lo>! the melting point of '/ilter . Thus at equi librium, temperature at t'le
co nta ct surface is at the mel ting point of water, an r1 the heat produced by friction
i s equal to the ~eat conrlucted into the two solids_ The friction i s thus r1etermined
hy the conduction of heat.

When the te:1perature difference ,'\T is sma ll (amb i ent temperature near ~ 'C) the

co nstant te r m in Eq. (10) beco~es dom inant and the coeffic ient of friction is

635
I

whi ch shows tha t the fri c ti on is determi ned by the vi scous shea r in the wa ter layer
and by melting ice into water. The main difference between E~s. (1 3) and (14) is
that at low temperatures the coefficient of friction decreases linearly wi th { V but
near O'C it increases linearl y with /v (Fig . I).

As a numerical example the coefficient of friction at -15C will be calculated using


Eq. (12) . The following values for the variables can be used [ 1, 5]
3
FN = 10 N 01 = P2 = 916 kg/m
v = l!!!. n 3
= 1,76 . 10- kg/ms

s
o

" 1 = "2 = 2,2 W/m K


h = 330 kJ/kg

Cl = c = 2,09 kJ/kg K 3
2 00 = 1000 kg/m

The most uncertain value is that of the indentation hardness H . It depends on


l
loading time (Fig. 5), which in this case may be of the order of 10- 3 . . . 10- 4 s giving
6
a hardness of the order of 40 ... 80 '10 N/ m2 Using a value of 60'10 111m2 and a
pressure mel ting temperature _5C correspondin3 to this pressure or a temperature
difference LIT = 10 0 C the coefficient of friction from E~. (12) is u = 0,024, which
is in agreement with the experimental results.

E
L
--
z 100
CO
-4
10 5

~ 50
C1J
c 155
'E
o 20
~----105
.c. 3
c 10 _105
o 7
..... 5 10\

f
o
.....
c
C1J
\J 2
C

-1 -2 -5 -10
Temperature

Fig. 5. Indentation hardness and pressure melting (broken line) of ice [5 J.

636
In practice the ice surface is not quite even; rather, there may be scratches and
loose ice powder on it. These tend to decrease the dependence of friction on velocity
and increase the friction at higher velocities, which can be seen in the results with
higher normal loads where the scratching is more intense . Also the properties of ice
can vary depending on its micro-structure, which may be the reason for the relatively
large difference at -lSoC between run 1 (the first row at each velocity in Table 1)
and runs 2 and 3 (the second and third row at each velocity in Table 1), as the
former was made on di fferent ice than the two latter.

In spite of the effect of these other factors the mathematical model gives, however,
the right order of magnitude and some idea of mechanism of the friction of ice.

REFERENCES

1. Bayley, F.J., Owen, J.M. and Turner, A.B., Heat transfer. London , Nelson 1972.

2. Bowden, F.P., Friction on snow and ice. Proc. R. Soc. London A. ~ (1953),
462 ... 478.

3. Bowden, F.P. and Hughes, T.P., The mechanism of sliding on ice and snow.
Proc. R. Soc. London A. l.Zl (1939), 280 ... 297.

4. Evans, D.C.B., l'ly e, J.F. and Cheeseman, K.J., The kinetic friction of ice.
Proc. R. Soc. London fl. 347 (1976), 493 ... S12.

5. Hobbs, P.V ., Ice Physics. London, Oxford University Press 1974 .

6. Keinonen, ,)., An experimental device for measuring friction between ski and
snow. Helsinki 1978. Acta Polytechnica Scandinavica Ph In.

7. Oksanen P., Coefficient of friction between ice and some construction materials,
plastics and coatings. Espoo 1980. Technical Research Centre of Finland.
Laboratory of Structural Engi neeri ng. Report 7.

637
DISCUSS ION

J. t~o lgaard U~emoria1 University of )!el'lfound 1and): The author ma~es interesti ng
suggestin9 on t.'le neC1ani s!'1 (If the fricti on Of ice, base~ on t~e !)resence of a .!ar.:or
fil!'1 at the interface. The problem faceri i s the elucidation of "hy the coefficient of
friction ,las hunn to increase 4J it1 increasin0 sliding ve l oc ity if the te' ~geratclre

was -1 'c , '1hi1e the coefficient rlecreaseo '. ,ith vel ocity at ~ te'~reratl1re of -ISC. At
S'C relatively li ttle change "as observerl,

The arg'.l ment hinges on the interface te,~pe ratures ac'lieved. [larnes et a1 L1,
ca 1culaterl the teMperature rise at '1.01 ~l/S at the contact ~Q ints to be about 15'C
fo r ice s li~in') on granite , but only O,S ' C For ice on brass, in hoth cases for a "v
of iI.3 . In t1e aut1or's eXDeril~ents ",. "as an order of '~a:;nitu~e less, so lIsin,) the
S'l lne a~nroac'1, sim ilar te'~0erat'lre rises \'/O I/].j he cxrccteri at vou n~ 0 .1 m/s. A
ri ~ta ilerl general revie,.<! of interface tC'Glper dtures at sliding interfaces ~as been
Jive r hy ~rchard 2 For ice against ice at ., load of 11) I" at 1 m/s this treat '~(' nt

'lredicts a t empe rature rise of only ar,out 0.5'C , usin) the physical pro r erties quote~

by D. I)ksane n an~ t~f' coeffic ient of fricti on rerorteri here.

The author 11S assumeri that all the water rroriuced by mel ting stays at the interfac e,
3S "" 1/ tile h~l~ncp. of ~eilt ororltlction anri flml is consi~er2rl. There is also -1 i11i!SS
flo"l hillan r.e t.O consider. ,later an~ t~e interface will be left ~ehind as co nt act
Jo int.s M0V~ ~n an1 50 it is rerhan5 f'10re likely trat tre thic~ness of any wat.er filr.1,
s ho'J ld f1eltin') occur, is aoverned pri/T1 ari l y r,y hydrody nami c effects rather than the
heil~ ~al'ln~e (or the O1ass '>'I 1ance, per se) .

Clearly there may be , "~ltinn 0ccur i n') at -1'C at all sfleed5 re ported and the increase
in" wit~ incre-1sing s~rrrl c0'J l~ the n he riue to Iwdrodynamic effects. However, .Jhy,
ot ,l S, shou 1 rl U, hI' l i' s S at -IS'C t~an at -1 'C? The argument presented by the
aut'1 0r isr.'t sufficiently rle ta ile ri or rigoro,)s to nrovirle an answer. The author does
en~ un sugnest ing that at -is'S the water thickne ss i s zero in hi s e~uations, hut
1r.verthe1~ss a "sel f-'la1anceri" .Iat~r fi 1m is pre sent; an apparent contradiction. Are
"Ie perha ns dealin g with interesting properties of very brief contacts between solid
ice surfaces? Derears the !jresence of ice dehris wou1rl provide a more prosaic answer.

1. Garn es, P., Tahor, ~. and Wa lker ,J . C. F., The fri c tion and creep of po1ycrysta11ine
ice. Droc. R. ~oc . London A. _~2_1!. (1971),127 ... 15<;.

7.. Archard, J.F., The temrerature of ruh~in9 surfaces. Wear ~ (1958/59),438 ... 455.

638
P.Oksanen: The calclJlation of the coefficient of friction is based on thermodynami c
halance ([~. 8), an~ it is in a9reement with the analysis of IIrchard, whi ch gives 6T
= wOe using tile values of the example in the text.

The main reason that the coefficient of friction at -15C can be sma ll er than that at
_1C, is that the in~entation hardness of ice (H t ) decreases very sharp l y when the
temperature approaches DOC (Fig. 5).

Putting d = 0 in Eq. (9) does not mean that there is no water film at all, but it
means t~at the film is so thin that the amount of heat required to produce it is
negligible comparerl with the heat conducted into the solids. In other words, Eq. (9)
is a termorlynamical equation, not a mechanical one.

C.O.Smith (Univ ers ity of Saskatchewan): The authors tests are for kinetic (i.e.
running) friction, which is important for moving i ce.

Hm/ever there are cases where we are concerned wi th s tatic friction, as in the case
where one must calculate the force needed to s tart moving an ice floe that has been
beached or grounded on top of a paved roadway, for example .

Our 1imited te s t s on s tatic fri c tion indicate \J values 2 to 3 times as great as the
authors kinetic f ric tion values. Has the author done any static tests and if so,
~lOuld they confirm these higher ~ values?

P.Oksanen: We have made also tests on static friction and the resul ts are presented
in ref. (7]. They are higher than the kinetic friction values approximately by a
factor 1,5 ... 5 depending on the test material, temperature, velocity at the kinetic
test etc. HOI'iever, in case of ice the coefficient of static friction is not a well
defined materials constant, because it depends on the time of contact before the
test. Irrmediately after the specimen has been placed on the ice, a freezing process
begins at the contact surface, and the force to get the specimen to move i s partly
due to static friction ann partly due to adhesion.

639
f..Palosuo (University of Helsinki): The main result is the equation (!2), ,Ihich is
based on general ohysical principles. This explains the n"echanisf'l of friction in a
rather simple case, i .e. ice against ice. When applied to other f'laterials additional
factors ''lay he taken into account.

Velocities in these experiments have varied from 0,5 to 3,0 m/s. Probahly the model
is not valid for velocitie s less than 0,5 lOlls .

P.O~sanen: The mathematical model is based on the assumption of the water layer. If
the velocity is low, the frictional heat is not sufficient to keep the contact
surface at melting temperature. So, the model is not valid for low velocities.

640
NECHilN r Cil l. PR ilP.!:~R '['f ,:S .F illlHES [O~

STH ENGT H TO P II. E ST RUCT URES

Hiroshi Sae ki
Assoc iate Professor

Toshiyuki Ono Departemnt of Civil Engineering J ap ~lll


I nstructo r Hokkaido University

Akira Ozaki
Professor

ABSTRACT
Systematic tests were carried Out to c l a rify the effects of push-aut speed. stress
rate, i ce thickness, pile diameter, material of pile and ice temper a ture o n adhesion
strength of sea ice to piles by means of push-out test. Steel piles and co nc rete
piles with diameters 3.15, 5.0, 10.0 and 15.0 cm were used in ice thi c kness frum
1.0 t o 1 3.0 em at a moderatel y high ice temperature ranging from -1. So C t o -4.3C.
Adhesion s trength to piles was observed to have a peak value with changes of
pu sh -out speed and stress rate. Adhe s ion strength increased with increas in g ice

thickness and rougilness of pile surface and also increased witll dec reasing pile
diameter and ice temperature.

INTRODUCTION
When the water freezes over a river, a lake, or a sea and the resulting ice plate
adheres to a pile structure, vertical forces are generated due to a change of the
water level. In case the floatin g i ce surrounds a pile structure, a rise of the
water lev e l generates an upward force on the pile. When the water level drops ,
the fl oa tin g ice plate exerts a downward force on the pile. To prevent lift off,
pullout, or buckling of pile s tructures due to such forces, the vertical ice forces
o n piles must be known. The investigations of the vertical ice forces on pile
11 I
structures due to a change of the water level have been carried out by LofqulSt,
[21 [3 I
Never and Kerr.
In general, failure of the ice cover a djacent to piles due to a change of water
level is classified into two modes, the failure by bending of the ice co ver and
the failure by shear in the ice adjacent to the pile. The largest vertical force
to piles occures at the time of the failure of the ice cover. Therefore, in
641
order to clarify the failure mode of ice cover and to evaluate the largest vertical
ice force on piles, elastic constants (Young's modulus and Poisson's ratio),
the thickn e ss of ice cover, the diameter of the pile, the flexural strength of ice
cover and the adhesion strength between ice cover and piles should be known.
The paper presents the experimental results on the mechanical properties of adhesion
strength between sea ice cover and the piles made from five kinds o f materials
(steel, corroded steel, painted steel, concrete and wood). The adhesion strength
of sea ice to piles may be regarded as a function of diameter of pile ~, ice
thickness h, push-out speed VB ' stress rate TB and materials of piles.

TES TING METHOD AND APPARATUS


Experiments were carried out at Toppushi
fishery har vour beside Lake Saroma which
was connected with the Okhotsk Sea by
two channels 300 m and 50 m in width and
had the salinity of 32 - 33% .
The diameters of piles ~ were 3.15, 5.0
10.0 and 15.0 cm for concrete and steel

piles, and 10.0 cm for painted steel,

c orroded sc e el And wood piles. In order Fig. I Preperation of specimens


to make specimens, rectangular holes
(a) (b) (c)
(3.2 m x 1.0 m) were made and 20 or 25
p
model piles were suspended perpendi


cularly to the water surface at ~teel cap~ , "" .
intervals of 25 em as shown in Fig. l. ; ',", I' ' /.

When the ice around the piles had grm""


to the desired thickness, the ice cover pile \
ice cover ice plate
around the piles was cut with a chain
Fig. 2 Schematic of push-out test
saw as shown in Fig. 2 (a), Core was
taken to avoid vibration. The ice surface around the piles was smooth, but the
thickness of the ice was not constant. The thickness of the ice cover in the region
immediatel y ;I d iac ent to the piles was greater than the overall in thickness.
Therefore, in order to obtain accurate adhesion strength, the ice plate was finished
ac c urately to the desired thickness by means of hand saw and plane. In the next
place, the arrenged specimen was turned upside down as in Fig. 2 (b) and the pile
was thrusted in the hole of steel cylinder with flange to the ice plate as in Fig ,
2 (c). The ice plate was in contact with the flange perfectly , The diameter of
hole of the cylinder was 0.2 mm larger than pile diameter. Steel cap was put on
the pile which was settled in the hole of steel cylinder, and hydraulic ram pushed

642
out the pile through load cell
a nd steel cap as in Fig. J (a)
and (b). The adhesion strength
between sea ice and various
piles was measured by nl - a ns of
push-out test. In reinforced
' ycJl aul i c
concrete engineering, bond ram
strength between reinforcing bar dial gaug
and co ncret e is usually m C ~ lsured
load cell
by means of three kinds of te sts
steel cap
i.e . , push-out test, pull-out pi Ie
tes t and both-end pull test. In
the cas e of reinfor ced COnC[vlL:,

the difference in te sting (n) (b)


Fi g . J. Schema tic di ag ram o f experimental setup t o
meathods produces dit le r en c bond
me asu re adhesi oll stre ng th .
strength. As t he Yo un g 's modulus
of pile made (rom steel and concrete is far l a rger than tha t o f sea i ce , it seems
th a t the differen ce ill testing methods has not a n e ff ect on adhesion strength. TIle
both-end pull t es t is not suitable f or i ce. Adhesion strength TB c an be obtained by
Eq. 1.
P
TB -n ' 'J.I - - ( I )
h
adhesion strength Kg /cm ' push-out for ce ( Kg

thickness o f i ce ( cm ) pile diameter ( e m )

The push-ou t for ce and the push-out speed VB w~re measured by means of load ce ll and
dial gauge wi th strain gage respectively. Stress rate a nd push- out speed co uLd be
continuously adjusted by o il f lux control ap pa ra tus which was co nnected with hycraulic
ram through pressure resistillg hoses.

EXPERIMENTAl. RESUI~S

( i ) Push-o ut speed effects


Since sea i ce is visco-elastic subs t ance , the adhesion strength between sed i ce and
piles is influenced by the stress rate TS and push-out s peed VB' Accordi ng to the
Frederking's experimental resu~~k, it may be seen chat the large r the nominal
deflection rat e 6 be came, the In rger the a dlle s ion st rength between fresll water ice
and wood pile be came within a ra nge of 0 < 0.01 mm/s. He conclud ed that t he adhesion
. 0 . 17 5 151
strength 'B W3~ in prnp~rtion to 6 Pa rame swaran experimented on the adhesion
st rengt h o f frozen sand to three different t ypes of model piles and his testin g
method was push-out test. According to Parameswaran's resu lts, the adhesion strength
increased linearly with in c reasing push-ou t speed VB on a log -l og sc a le within a
range of VB < l.6 x lO- 3mm / s . Figure 4(a) anel (b) shOl' the relation ber"p c'n TB and

64J
rush-out speed VB in the STEEL PILE q> =10cm
5
cases of the ste~l pil~ and T =-3"C
h=4S-6.5cm
the concrete pile. The present 4

.,-.--..------.
"Ce
test was carried Out in the
3
3
range of 5 x 10- < Vs < 10 mm
Is. The adhesion strength of
~
...-- .., . . . ----------.J!.
sea ice to steel pile changes
gradually with the change of

VB as in Fig. 4(a) and the 0L---~10~
~~----L-~~1~--~--~~---L--~
peak of TB exsists within a 10- Va mm/s 10 10
Fig. 4(a) Relation between TS and VB for steel pile
range of 0.1 < VB < 1.0 mm/s.
In the case of concrete pile, I I I
5f- COf\.CRETE PILE ..

~'.:.:~--:--_
the change of TB is similar
to the case of steel pile, 4f-
1:8
but the value of TB increases
rapidly within a range of
3f _
~ 2/ .
VB < 0.2 mID/so The fact that 5 r <?=10cm
there is an increase in ]> T = -3C
adhesion strength with
1- h = 4.5-6.5cm

increasing push-out speed O~--~I~____~_~I____~~I =-__- L_ _~


2 1
within a range of very slow 10- 10- Va rrmls 10 10
push-out speed, coincides [ Fig. 4(b) Relation between TS and Vs for concrete pile
with the results of ParamesJJran and Frederking. [4]
(ii Stress rate effects
Figure 5(a) and (b) show the relation between TS and stress rate ts for the steel and
the concrete pile. The push-out speed of pile was kept at constant speed and the
relation between load P and loading time could be regarded approximately linear.
Therefore, mean stress rate TS is calculated from Eq. 2 .
2 )
2
TS : mean stress rate ( Kg/cm s tD : duration of loading time ( S

The results shown in Fig. 5 suggest that the adhesion strength TS has its maximum
2
within a range of 0.5 < TS < 2.0Kg/cm s for both piles and the change of TS to
concrete pile with change of TS is remarkable in comparison with the change of TS to
steel pile. Frederking experimented on the relation between TB to wood pile and TS
2
within a range of TS < 0.1 Kg/cm .s and the trend of his result is in good agreement
with authors' results. Judging from the abo ve mentioned results, as the maximum of
adhesion strength of sea ice to the steel pile and the concrete pile exists wi r hin a
2
range of 0.5 < TB < 2.0 Kg/cm .s, after this, the experiments on adhesion strength
2
are carried out within a range of 0.5 < TS < 2.0 Kg/cm .s.

644

Fig. 5(a) Relation between TB and TB for steel pile

5!- CONCRETE PILE I



. ... ..
I.
ttI!..._
.............---- . ..-
.
I_

Jf-

~2f- --------
.2'
1!-

---------

, fit
. .

4> =lOan
T = -J"e
h=4.5-6.5an

Fig. 5(b) Relation between TB and TB for concrete ~ile


( iii ) Ice thickness effects
It is well kno\>..'TI that the difference in testing methods produces different strength
of material and the size of specimen has an effect on the strength of material. In
order to clarify the size effect in adhesion strength test, it is necessary to
investigate the effects of ice thickness and pile diameter that have effects on
adhesion strength. Figure 6 (a) and (b) show the relation between the adhesion
strength TB and h for the steel pile and the concrete pile. Though the experimental
data show scattering, the thicker the ice plate becomes, the larger the adhesion
strength to both piles becomes, and the relation between TB and h can be regarded
approximately linear on a log-log scale. The adhesion strength of sea ice to the
steel pile and the concrete Jile is proportional to hO. 'and hO' 18respectively.
On the other hand, Frederkl~g show that the adhesion strength of fresh water ice to
wood piles increases with approximately the square root of the ice thickness. His
results ere in good agreement with the shear stress on the pile surface calculated
by reJiJs theory and are different from authors' results. This difference is caused
by the differences in testing me ~ hod for adhesion strength and in the pile materials.

645
5
STEEL PILE
4 > =10em 0
o
La T =-3C 0 0
3 La=05-2Iq1em 2/s ~
Fig. 6(a) Relation between
TB and h for steel - 0 ~o.osx> 00 0 o
pile NE2~ 0 0
~
.2'
o

4 5
hem

10 ~---~---.,--.-,--""-r",,nn"Tr----~
9
8 CONCRETE PILE
7 =10em
6 T -3C
=;

Fi g . ne b) Relation betwee n
TS and h Ear co ncrete

pile

B5
.
( Cg

. ..--.
l4 ~ ... '
=05-2kg/cm21s ....
-~~
__

.2'3

( iv ) Pile diameter effects


Tes ts for s teel and concrete piles were carrie d out to investigate th e nature of the
pile d iameter influen ce on ad hesion strength. The dia mete rs ~ o f piles use d in th i s
e xper i men t we re 3. 15, 5.0, 10.0 and 15 . 0 cm . The experimental r esu lts are shown in
Fig . 7. The notation 0gr i s the mean d iamete r of grain in sea ice . The horizontal
axis, n~ / Dgr means the ratio of the circumference of t he pile to the mea n diame ter
of grain, a nd the mean d iame ter o f g rai n was 0.8 cm. The experimental results show
that the adhesion strength decrea s e s with incr en si ng pile diameter or relative
ci rcumference /O gr . As the relation between t B and rO/O gr ca n be regarded
- 0 8 0 8
ap pro ximately li n ar on a l og- l og scale, TB i s proportional t o ( .~ / Ogr ) . or ~- .
"ithin a range of 12 , nt /ogr < 60. This re l at i on is in good agreemen t with
Frcderk ing 's reLtIc) T8 ~ -0.79. From the a bov e mentioned results, the adhesion
s t reng th o f ic e t o pile
10

s tructures is pro portio na l to
1 I CCN:RETE Pli
4- 0 ' "'9-- -o'OOalld th is relation 0 STEEL F1
mi ght be app li cabl e t o piles maJe

from various materials .

( v ) Effects of pi le materials
5 0
0

Autho rs experimellted on th e 8

0
ad he sion st reng th between sea ic e

~
and va rious mater ia l s which were 0 0
usually used in of f- sho~e
l
~ ~ CD
0
0
s tru ctures: i .e., stee l, painted .2' 0
T =-35--45"(
st ee l, corroded st eel, concre t e ts =05-2.0kglcm 2
and wood . The diamete~ of the h =4.5-6.5cm
0
Dg,=O.Bcm
piles used Eo~ the expe~ime nt s

was 10.0 cm. Fi gu~e 8 s hows t he


1
~
e x pe~imental ~e s ults of adhesion
10
s trength of sea ice to various
pil es. The co~~oded steel pi les
Fi g. 7 Relat i on between '8 and IT <I>/ Dg~
we~e made by means of soaking
stee l pi le s in sea wate~ fo~ a month. The painted ste el pi les we~e the steel pi l es
coa ted with a laye~ of abo ut 200 ~ thic k ma~ine paint. The adhesion st~ength o f sea
ice to the painte d steel pil es is very smal l and i s one tenth of that to the stee l
piles. F ~ om the exper ime ntal results men tioned above , the adhesion s tre ngth of sea
ice to pile s can be decided by the materi a l and the ~ou ghness of piles und e~ same
exper imenta l conditions.
6,------------------------------------.
( vi ) Effects of s ea ice
<I> =10cm
5f- T =-2.2C 468:

ts=0.5-2.0k:;l1cm2 s 0
tempe ratu re 1:
~I- h =7.0-9.5cm
Generally, the st ren gt h of o ; /IAEAN VALLE 103 80
sea i ce , uniaxial
:0312
c ompress ive strength,
fleKu~al strength and

01.92
tensile str ength , depe nd s
I
on the i ce tempe rature and
t he brine volume. The shear o aoO.16 J
st r engt h of se a i ce and thE
PAMED STEEL WOCfJ CDffiCIE)
STEEL STEEL
coefficients of fricti o n
Fig . 8 Adhesion strength '8 for vario us mat er ials
between sea i ce and various
mater i a ls which can be

647
I

5,r---'---'---~--~1---'---''-~
regarded to be c l ose l y relnted t o STEEL PILE
l, \-- 4> =IOcm

...
th e adhesion strength. de pe nd ('I ll the
temperatJ~J.

---
i ce Th e relation between
t.H"O.5-2.0kgJcm' s

.,, ___.-_,1-_1.
h =1,-6cm

.
..
adllesion strength and ice temper a ture
~
T is shuwn in Fig. 9(n) and ( b ) . The
l o we r the ice temperature T be co mes, -5
.2'
2\ - - .
~
______
.-

--..

the larger the adhe s i on stren g th to


I f
the steel pi le atld th e co ncrete pile OL-__~__L -__~__~__J -_ _~_ _~

beco mes. The relation between TB .:Jt1d I 2 3 l,

'I' ca ll be regarded approximately Fig. 9(a) Relation between TS and T


line a r within a ran ge of -4.5 < T < fo r steel pile
IO r---~---r----.--~---.---'.-~
-l . S" C. The relatio n between ad hesion
f- CONCRETE PILE I I
strength TS to steel pile and i c e
> =IOcm
temperatur e T, is expressed by Eq. 3 . ~=0.5-2 .0kg/cm's

. -.
. .
h =l,-ocm
TS ~ 0.80 + 0. 4S T - - ( 3 )
;p = 10 cm, h = 4 -6 cm
~ 5f-
~ f ~
~...
~.

TS = O.S~ 2. 0 Kg/ c m's .2'f- ---.--- 1


-4 .5 < T < -1. 5 C f
\-

The rel ation between TB to concrete OL-__~__~I____~__L-I__~__- L__~
pile and i ce temperature is expressed I 2 - PC 3 4
by Eq. 4.
Fig. 9(b) Relation between TB and T
for concrete pi le
TB = 1.80 - 0.7ST - - ( 4 )

CONCLUS TON

The res ults o f thi s study may be s tat e d as f o l lows :

1. The maximunl va lu e o f adhesion strength TS of sea i ce t o various piles exists


wit hin d range of 0.1 < VB < 1.0 mm /s and within a range of 0 . 5 < TB < 2 .0
Kg/cm' S .
2. The a dhesi 0 n stren g th TB t " th e s reel pile an d the co nc rete pi l e is pro portiona l
2 le
t o hO and h O respect ivel y.
3 . The adhesio n srrength TS tl) various piles is pro porrional tn <p-0.7~ --- -o ov
4 . The adhe s ion stren g th 'S t tl piles can be decid e d by the material and the Tough ness
o f pile structures unde r sam e experime nr al c0ndiri () ns.
5. The relati o n between TS and T ca n be regarded appro x imately linear within a rangp
o f - 4 .5 < T < -I.S 'C

ACKNOW1.EllGEMEN T
We wish t o express our pr o found g ratirud e f o r the e x tensive cno peration g i ven by
Mr. Yoshibumi Sakakura o f the Sappo rp Muni c ipal p ffi c e and Mit su r u Yamada of Mit sui
Sh ip Building an d Engineering Co., U l ~, former s tudent. ill t he experim(;, llts (-lild
arrangement u f the ddta. This wor k \vilS pa rtl y suppurted by the Sc ientir h : l1esean:h
Funds f r om the Ministry of Educ a ti on, Jdpiln.

REFER ENCES
[1 J Lofquist . B., "Lifting force and bearing c apacity of all ice s heer." Natl. I.{es,
Council Can., Ottawa, Canada. Tech. Rep. R-700, 1970.

{ 2 } Nevel, D. E., "Lifting forces exer ted by iee on struc t ures." Proe. Co nference
on Ice Pressure Against Structure~: Laval Universit y , Quebec, 1966.

[3) Ke r r , A. D., "Iee forc e s on st ru Ctu r e s due to a change of the water level."
Proc. 3rd . Int. Symposium on lee Problems, Hanover, 1975.

[4] Frederking, R. ~1. I,., "Laboratory t e st s on downdr a g loads d ev eloped by


floa ting i ce cove rs on vertical piles." Proe. Port and Ocea n Engineering
under Arctic Conditions, Norwegian Institute Technology, 1979.

[5] Parameswaran, V. R., "Adfreeze strength of frozen sand to model piles,"


Ca n . Geotec . J. Vo l . 15, 197 8 .

[6] Saeki, H., T. Ono and A. Ozaki, "Experiment al stud y o n ice forces on a cone
shaped and an inclined pile structures." Proc. Port and Ocea n Enginee ring
under Arctic Conditio ns , Norwegian Institute Technology, 1979.

649
FOlUlATlON OF SHORE CRACKS IN ICE COVERS DUE
TO CHA.'lGES IN THE HATER LEVEL

Lennart Billfalk Swedish State Power Board )llvkarleby


The Hydraulics Laboratory Sweden

Abstract
The process of breaking up a solid ice cover on a river due to rapid
fluctuations in water level and flow rate starts according to Donchenko
by the formation of shore cracks.

Formulao for predicting the change in water level that creates shore cracks
along a straight river are derived using the theory for beams on elastic
foundation and assuming elastic deformation of the ice . Application to
the case of a hinged support at the shore appears to give results that
agree with these referenced by Donchenko. The analysis also gives
expressions for the distance from the shore to the crack in the cover and
the vertical reaction force against the shore or a structure.

Simple laboratory and field experiments have been performed and the
results of measurements agree fairly well with the theory. Finally some
observations from a regulated river are presented which show how the ice
cover thickens close to the shore due to repeated cracking and subsequent
flooding of water in this area.

650
1. Introduction

Freeze-up of rivers and lakes at constant ,<ater level results in solid ice
co vers that a ttach more or less rigidly to the shores. Pre diction of the change
in water level, that gives rise t o shore cracks, for example at places ,<ere
winter roads cross, could be of importance. The problem that has initiated
the present study is, however , the break -up of solid ice covers on r ivers due
to rapid changes in wate r level s and flow rates associated wi th regulation
of hyd ro power stations.

In Sweden the tendency is to increase the degree of short-time regulation of


hyd r o- power stations. Experience 11as Si101'1n that in some cases this lead s to ire
problems durin g the freeze-up period and t hat th e degree of regulation
preferrably should be re dlJ ced un t i l a s uff ic ientl y strong ice cover has for med.
One interesti ng point is then to determine permissible regulation, according
to actual ice thickness and river co nfigurati on.

Until 1975, research on the stability of ice covers had been concentrated on
broken-up i ce acc umulations [1]. After 19 75 one ma j or contribution regarding
the stabi lity of solid ice covers on rivers can be found namely a study
by Don chenko [2].

Acco"ding to Donchenko the destruction of the edge of a so lid ice cover down
stream of a hydro power station starts with the formation of shore cracks [2J.
ft is believed that the forma tion of shore cracks is the f irs t step i n the
process of bre ak ing up a so lid ice cover also at mos t other location s al ong
a river .

In the paper by Donchenk o a formula by Kachanov is presented for calculating


critical magnitude of water level rise. That is, t he lea s t change in level at
wh ich a s hore crack will occ ur. Unfortunatel y some parameters and two so called
function s of relaxation in thi s formula are not defined which makes the use of
it uncertain. (No translat ion of Kac hanov's original paper, written in Russ ian,
seems to be available).

The present study deals with the f ormation of shore cracks. A formula for
predicting the critical change in water le vel is derived and compared to th at
referenced by Donchenko. The di s tance from the s hore to the crack as well as
the vertical rea ction f orce against the shore are also computed. Furthermore

651
the results of some simple laborat ory and field experiments are shown.

2. Theory

In a general case the deflection of a floating ice cover couid be anal yzed
with the theory for plates on elasti c foundation. For a straight part of a
river, however, the theory for beams on ela s ti c foundation could gi ve an
adequate description of the deflection of the ice cover across the river.
This presumes that the deflection of the ice cover in the direction of the
river can be ignored. LOfquist analyzed already in 1944 in that way the lift
force excerted on structu res due to a rising i ce cover that attaches rigidly
to the structure [3].

Assume an initially plane ice cover. The deflection and the distribution of
the bending moment of the ice cover could then be derived by analyzing a
strip of the ice perpendicular to the shore. The deformation of the ice is
assumed to be elastic and the limitations of that assumption are briefly
discus sed later.

On figure 1 a plan and a cro ss section of an ice covered riverare shown. The
half width b of the river is assumed to be so large that a variation on the
water level gives rise to negli gible deformation of the ice cover at the ce nter
of the ri ver.

shore line

~ie:':,5
strip of unit width

cross section I
Figure 1. Plan view (left) and cross section of ice covered rive r.
Definition sketch.

Consider fi rs t a drop 6h in the water level which is assumed to be less than


the thickness of the ice. The drop in water level acts as a uniformly
di s tributed load q on the ice. The load can be written q = k-6h, where k is
the modulu s of the foundation (approximately 10' N/m' for water).

652
Under the given assumptions the deflection of the strip of ice can be
analyzed as a semi-infinite beam. The solution to that problem can be derived
in the following ,yay according to Hetenyi [4].

Cons ider an infinitely long beam, as shown on figure 2, where pOint A indicates
the support (shore line) at the end of the semi-infinite beam.

M:] Q

~ rl I"r'l'r;~I.~I.~~~:~.:.~

rx y
Figu re 2. The deflection and moment distribution for the semi -infinite st rip
of ice are derived by superposing the solutions for an infinite
beam on elastic foundation under the load s Po' Mo and q.

Because of the load q there is at point A a bending moment MA and a deflection


YA in the infinite beam. Consider now the semi-inf inite beam with a hinged
end at point A. The end conditions for the semi-infinite beam at point A are

y = and M = 0, that is the deflection and the moment respectively should
be zero at this point. These conditions can be fulfilled by applying a moment
Mo and a force Po close to the left of point A at the infinite beam and require
that these forces produce -Mn and -YA respectively. According to Hetenyi
the moment MA and the deflection YA can be written:

M q e - I.b. Ab
A iITz- 'Sln

q - Ab
YA = 2k (1 -e ' cos Ab)

For large values of b the moment and the deflection are approximately:

(1)

653
(2)

where ~ ~ (k/4EI)1/4, E = modulus of elasticity of the ice and I = d 1 /12.


The term l/A is often called the characteristic length of the system.

Furthermore the expressions for the end-conditi oning forces are:

( 3)

(4 )

Equations 1 to 4 give

Po -q/ \ (5 )

(6 )

By the method of superpOSition the moment M (Ax) and the deflection y ( AX)
at a distance 'x from point A of the semi-inf i nite beam are determined by
adding the contribution s from q, Po' and Mo' By dOing so and ignoring te r ms
that are small for large values of b the expressions for M (Ax) and y (Ax)
>Ii 11 be:

(7)

(8)

By equating the derivative of M (Ax) with respect to x to zero it is found


that the maximum moment occurs at Ax =rr / 4.

b. Fixed end
---------

If the ice is rigidly atta ched to the s hore the co ndit ions for point A are
y = 0 and e ~ 0 , >lhere e i s the angular deflecti on. In analogy with the
deri vat ion of equations 7 and 8 it can be shown that the following expressions
will hold, provided that b i s large.

_ 4 hk.e- A~(si n\x-cosAx)


I~ ().x) - 2\ ' (9)

654
Y (Ax) OAh[l-e-Ax(sinAx+cos Ax)] (10)

The largest absolute value of M (Ax) in this case will be for Ax 0, that
is at the shore.

The reaction force per unit lenght that acts against the shore (or on a
structure), will be ~hk/A for an ice cover having either a hinged or a fixed
end at the shore.

In figure 3 the expressions 7 to 10 are shown in dimensionless form. For large


enough water level variations the flexural strength of the ice will be reached
and cracks will develop where the bending moment has it's maximum. Having a
hinged end the ice cover should break at a distance x = TI / 4A from the shore.
If the ice cover is rigidly frozen to the shore the ice cover shou ld first
break close to the shore. Thereby a hinged or a freely supported end is
created. If the water level decreases more the ice cover will also in this
case break at a distance TI/4A from the shore.

o 1.0 2.0 3.0 4.0 5.0

O~~.i
0.2 "" Eq 10
0.4 'Y
0.6 "
4ffl 0.8 ',,_
6
1
1.0 ---~-~-~~-~--------------------i
1.2L-__________________________________________ ~

-1.0,r-----------------------------------------~
-0.8 ,
-0.6 \ Eq 9
-0.4 ~
M(>'x) 2), 2 1-0.2 "
6Iil< 0 ' ...... . +----f-=~=-_='=--.....,.~--__t
0.2~7 .... ~-------
o. 4L~:::::=====___ _ _ _ _ _ _ _ __ __ _ _~

Figure 3. Deflection (above) and moment distribution in dimensionless form


according to equations 7-10 .

655
It could be noted that expressions 7 to 10 apply also if the water level rises
provided that the assumed end conditions are relevant in this situation and
with due regard to the adopted sign conventions.

The cri tical change in water level that creates cracks can now be computed.
Assume that the ice cover will crack if the moment M = 0bd'/6 is reached. Eq.
9 then gives the foll owi ng critical change in water level for a fixed end:

Ah cn't = 0,0058 0b ~ (11)

and eq. gives the critical change for a hinged end:

L\hcn. t = 0.018 0b'id7f ( lZ)

where 0b = fle xural strengt h of the ice.

The values of u and E vary wit h temperature and type of ice. The flexural
b
s trength also varies with the thickness of the ice [5J.

On figure 4 the critical change in water le ve l, 6hcrit acco rding to eq. lZ, is
plotted for 0b = 10 ' N/m ' and E = 6.5 -1 0' N/m'. On this figure also some values
on the cri tical change in water level, given by Donchenko, are shown. As can be
seen, Donchenko's va lues for an air temperature of -ZoC lie close to eq. lZ for
the ch osen va lues of Db and E.

0.4

0.3 o
0
0
0
E 0.2 0
x
...., 0.1 x
I
U

-"

<J 0
0 0.1 0.2 0.3 0.4 0.5
d (m)

Figure 4. Critical change in water level bhcrit for an ice cover of thickness
d that has a hinged end at the shol'e (Eq. lZ for Db = 10' N/m' and
E = 6.5-10' N/m l ). Critical valu es for temperatures -ZoC (x) and
-Z5 0 C (0) given by Don chenko are also shown.

656
I

The presented analysis is valid for r apid changes in the water level when the
ice is deformed elastically and a brittle fracture occurs. For slower changes
in the "ater level the viscous deformation of the ice plays a role and the ice
cover can withstand la rger change in water level before it breaks or reache s
a state of permanent creep. The conditions for having elastic deformation in
this specific case has been briefly analyzed by Billfalk [6]. In that study
it is shown that the limit between pure elastic and visco-elastic behavior
is heavi ly dependent on the type of ice and the reological model that i s used
to evaluate these effects or the experimental data considered.

Hichel -refers some bending experiments of snow ice beams [7]. These experiment s
showed that partial plastification of the cross section but still a brittle
fracture occured for strain rates larger than 10- 5s -1. For an ice thickness on
_ 1
the order of 10 m this strain rate occurs in the s ection of ma xi mum moment in
an ice cover that has a hi nged end, if a change in water 1eve 1 of the same
order as the ice thickness occurs "ithin the order of ten seconds ("critical
time"). Based on other experiments one could find "critical times" on the
order of minutes [6].

3. Expe ri men ts

In order to check the theory pre se nted under section 2, some crude experiments
have been undertaken for an ice cover with a hinged end. The parameter that
easil y could be determined "ith good accuracy was the distance from the
support (shore) to the craCk. This parameter was also judged to be
little influenced by the characteristi cs of the ice.

In a non-heated laboratory hall an ice cover was frozen in a basin made of


steel plates, 5 m long, 0,5 m wide and 0,3 m deep. Water from a nearby river
was used. The ice cover wa s melted free from the sides by Circulating water
at a temoerature iust above DoC in a channel surroundina the test basin.
A drop in the water level Ylas simulated by lifting the ice cover at one end.
This was done by usinq a simple lever to raise an iron har s liS oended benea th
the ice cover. The support was raised evenly at such a
speed tha t the ice cover failed in around 5 seconds. By measuring the
app li ed force on the lever the rea ct ion force from the ice cove r on the
support could be determined. A simple mechanical indicator shm,ed the rise
of the support necessary to create a crack in the ice cover.

657
On the laboratory scale, ice covers of ma ximum 0.05 m thickness could be tested
without si9nificantly violating the assumption of havin'l a semi-infinite ice strip.
In order to have data for thicker ice covers some field experiments were
conducted. Ice strips, 20-30 m long and 0.5 m wide, were cut out from a natural
ice cover by a chainsaw mounted on a sleigh. A drop in the water level was
simulated in a way similar to that adopted for the laboratory exper iments.

The distance from the support to the crack, determined in the laboratory and
in the field, are shown on figure 5. On the figure the theoretical curves for
two different values of the modulus of elasticity are also shown .

2.0 10 o

E 1.5
.0
u
'"u' 1. 0
..,0 o
(l)
u
0.5
c
..,'"
0
0
0 2 3 4 o 10 20 30 40 50
d (cm) d (cm)

Figure 5. Distance from support to crack as measured in the laboratory (left)


and in the field. Theoretical curves for = 6.5-10'N/m'(10wer
curves) and E = 10,oN/m' are also shown.

It is noted that the scatter of the experimental data around the predicted
curves is quite large, especially for the field observations. It could also
be mentioned that in a number of laboratory experiments, not shown on figure 5,
more than one crack (mostly two) developed apparently simultaneously. If two
cracks developed, however, they occurred one at each side of the predicted
location. The ice strips tested in the field might also have been cracked
during the sawing which could explain the large deviation from the predicted
values in some experiments.

The critical change in water level, measured as the vertical displacement of


the support was fairly well predicted for the field test, assuming
0b = 10h N/m' and E = 6.5-10' N/m'. The scatter in the 1eflection results of the
658
laboratory experiments was very large. All 1abora tory
e xperiments , however, required a much larger displacement of the support to
create a crack than predicted by eq. 12, assuming Db = 10' N/ m' and
E = 6.5-10 ' N/ m. (The theory is not strictly valid if a part of the ice
cover is raised above the water level) . One possible reason for thi s could be
that the fle xural strength of thin ice covers is much higher than it is for
thick ice covers. According to Lavrov the theoretical ratio between the
fle xural strength for an ice cover 0.1 m thick to the strength of a 0.5 m
thick ice cover is 1.5 [5J.

5. Discuss ion

The presented theory and experiments regard an ice cover with constant
thickness that has a fi xed or a hinged end at the shore.

In a river where the water level varies, crack s will develop along the shore
perhaps already during the freeze-up. This often means that water gets up
on the ice along the shore and s ubsequently freeze s . In figure 6 it is shown
how this process creates a significant thickening of the ice cover along the
shore in a river where the water level under the actual per i od of time
typically varied 0.3 to 0.5 m during the day with maximum changes on the order
of 1.5 m.
Distance from shore (m)

0 5 10 15 20 25
0

E 0.2
- - -15- Jan. -80

-----
14 Febr. -80

QJ
> 0.4
QJ

....
..,
QJ 0.6
'"3: I indicates
0.8
3:
0 ! a crack
QJ
.Cl
1.0

..,
..c:

a.

QJ
Cl
1.2

Figure 6. Thickening of the ice cover at the shore in a river with a


fluctuating water level. Cracks visible at the uppe r surface of
the ice are also indicated.
659
-

At these observations cracks could be seen at the upper surface of the ice
cover. These cracks followed the shore at a fairly constant distance even
though the s hore line was somewhat irregular.

From the presented field observations it is obvious that the derived


exp ressions for predicting the occu renc e of shore cracks must be used with
caution if applied to an ice cover that ha s been exposed to a changing water
level during its formation. The theorY,however,should apply outside the
thicker part of the i ce cover provided t hat the end conditions can be
established. In any cas~ cracks \Iill appearfurther out from the shore
than predi cte d by the th eory once the ice cover has cracked and thi ckened
along the sho re.

Acknowledgements

The author would li ke to thank Dr Krister Cederwall for valua ble comments
on this paper.

References

1. ~1ichel, B. and Abdelno ur , R., "Break-up of a solid river ice cover".


Proc. IAHR, Third Int. Symp. on Ice problems , p 253-259, Hano ver 197 5.
2. Donchenko, R.V., "Conditions for ice jam formation in tailwaters".
Dr aft Translati on 669, Ma rc h 1978, Cold Regi ons Research and Engineering
Laborat ory, Hanover, NH, USA.
3. Lofquist, B., "Lift force and beat'ing capacity of an ice sheet"
(in Swedish with and English summary). Te knisk Tid skrif t 1944 ,
Vol. 25. Al so available as Technical Tran s lation TT-164, National Research
Council of Canada, Ottawa 195 1.
4. Hetenyi , M., "Beams on elastic foundation". Ann Arbor, Univ. of r~ichigan

Press, 1946.
5. Lavrov, V.V., "Deformation and strengt h of i ce". Gidrometeorologicheskoe
Izdatel' stvo , Leningrad 1969, Tra nslate d from Russian - Israe l Program
for Sc ientifi c Tran s lation s , Jeru sa lem 19 71.
6. Billfalk, L., "On the effect of short time regulation upon freeze-up and
break-up of rivers". Rep ort under preparation (i n Swed ish).
7. ~'ich e l, B., "Ice Me chanics", Les Presses de 1 'Universite Lav a l, Quebec,1978

660
Discussion by D. Nevel
of
FORMATION OF SHORE CRACKS IN ICE COVERS
DUE TO CHANGES IN WATER LEVEL
by
1. Billfalk

In equations 11 and 12, the unit weight of water was assumed to be


9806 N/m 3 . Therefore the dimensions of Newtons and meters must be
used throughout equations 11 and 12. The results in Figure 4 can
be interpreted as a strength test. From Donchenko's plotted data
it appears that the ice was t wice as strong at -25C as at -2C.
The experimental data in Figure 5 can be interpreted as a test to
measure the characteristic length l/ A. Although the maximum stress
occurs at AX ~ n/4, its distribution is rather flat in this region.
The variation of the ice strength controls where the break occurs
is a broad region. Hence, many tests must be performed to obtain
the average and the statistical variation. The measurement of the
characteristic length or Young's modulus by a non-breaking method
would give a more consistent value for the characteristic length.
A viscoelastic solution of this problem was presented by Nevel(ll
to determine the force-water displacement relation, but the position
of the crack was not considered. The creep process, as well as the
ice growth process as discussed by the author, are the features that
produce results which differ from the elastic solutions.

(1) Nevel, D. E., Lifting Forces Exerted by Ice on Structures,


Proceedings of Conference on Ice Pressures Against Structures,
Laval University, Quebec, Canada, November 1966.

Answer to discussion by D. Nevel:


It is correct that Pg, the unit \-,eight of \~at er, is assumed to be about
9806 N/ m' in equations 11 and 12. The author also agrees with Nevel 's
other comments.

661
FOR'~TIO N OF SHORE CRACKS 1M ICE COVERS DUE
TO CHANGES 11 THE WATER LEvEL

Lennart Billfalk Swedish State Power Board Alvkarleby


The Hydraulics Laboratory SljEDEN

Rel"lark:

In Finure 4 Donchenko's results suonest a larner 6hcrit at lower


teMoera tures. Is thi s due to i ncreas i nq fl exura 1 s trennth vii th decreas i no
temoera ture?
Ans YI '?r:

The fornulil aiven bv ~onchenko is 0" the sane farm as e~. 12, i.e. the
6hcrit is ~raportional to 0b and inversely oroDortional to the s~uare root
of E. Furtherrrore, the bendinn stren~th usually increases I'lOre rapidly
than elastic modulus as the temperature decreases. This impl ies that
6h should increase with decreasinq temperature .
cr it

Remark from P.. Frederkinn:

I./hat '~as the loadinn tirrle for your lono beam tests ? How does this compare
"lith the 10ildinCJ tirJe you Mioht expect in nature?

Answer:

The lonn beams tested in the field broke after a loadin~ time of 5 to 10
seconds .

The fastest water level variations that f71i~ht be encountered in a river


would probably be caused by a rapid s hutdown of a hydro Dower plant. This
mioht cause "loadinCJ times" on the sane or-der as in the experinents.
Loadin0 times on the orner of some minutes could be the effect of a start up
of a hydro power plant. The presented theory. assuning elastic deformation
of the ice, i s probably fairl y accurate also in this latter case as far as
10a dina tiMe is concerned.

662
THE ESTIMATION OF CRACK PATTERN ON ICE
BY THE NEW DISCRETE MODEL

N. Yoshimura Tsu Research Lab., Nippon Kokan K.K., Tsu, Mie, Japan

K. Kamesaki Tsu Research Lab., Nippon Kokan K.K., Tsu, Mie, Japan

Abstract
The force exerted on a fixed offshore structure by impingement of
ice sheet consists of several components --- one to break the ice sheet,
another of ride-up of broken ice pieces on the structure surface,
another to move aside broken ice pieces, and so forth. Although the
force to break ice sheet is not as big as the other components, the
analysis of ice breaking pattern is important because the other com
ponents entirely depend on the shape and s ize of broken ice pieces.
A new numerical model named 'Rigid Body Spring Model' (RBSM) has been
ad op ted to analyze the bending failure problem of an ice sheet. A
computer code has been developed and good agreement was obta ined between
the theoretically predicted va lues and the experimental ones.

1. Introduc t ion
The authors must emphasize here that, with a material as brittle as
ice, the stress relief caused by crack initiation has to be considered.
Ice breaking pat tern has been studied by using classical elastic
theory [4], numerical elastic ca lculation method such as FEM[5] and
limit design concept[6]. By these methods, however, it is difficult
to deal with stress relief procedure mentioned above.
RBSM, developed by Prof. Kawai of Toky o Univ., is suitable for
analyzing the ultimate strength and collapse mode of structures[lJ
[2J, and has been successfully applied to failure analyses of beam o r
slab structures. RBS~1 is .also suitabl e for treating stress relief

663
problems, therefore the authors have adopt ed this method to analyze
fracture l'dttern of ice sheet.

2. RBSH model
RBSM has many variations. Among them, authors selected Ka',;ai
Watanabe's plate bendin g model[ 3 ], and generalized it to a plate
membrane-bending model. In this paper, however, formulas are deve
loped only for the bending problem due to space restrictions.
In RBSM, a plate to be analys0j is devided into triunyular r igi d
elements as shown in Fig.5. A typical rigid element is shown in Fi g .l.
The x - y plane coincides with the middle plane of the plate element.

2.1 Differential form of stresses


Displacement component s of an arbitrary point in the rigid
element L are expressed in terms of the
I..W.w
d isplacement components of the element

~
:I XG ;"l.
center of gravity G .
L '..1'1" ./ l
U=z<P (1) . . /I t...., ,,,I . I

o
V=-Z" ( 2) . <1>, . . ~. \I, ~ 0 ",
o ~ n, Un' I th edge
W=Wo+(Y-YG)X o -(x-xG)<p o (3 )
- x .U. u
Xo
where
U,V,W: di spl acement in x , y,z direction Fig.l Rigid element
Wo translation of G in z direction
L

xo'o: rotation of G about x,y direction

L
xG'YG: x,Y coord inate of GL
On the i-th(i=I,2,3) edge of the element, two other c oordinate axes
are defined as shown in Fig.l. The si axis is tangential and the n
i
axi s is outward normal. The direction of the si axis is taken as nixs
i
coincides with z direction. Displacements in n and si direction (U
i ni
and V ) are represented as follows:
si
Uni=liu+miv =Z(li<Po-mi)(o) (4 )

Vsi=-miu+liV=-Z(mi<Po +li)(o) (5 )

where
li=cos(ni,x), mi=cos(ni,y) (6)
Generally on the i-th edge of element L, there is an adjacent el eme nt
R . The center of gravity of the element L is G and that of the
i L
element Ri is G (se e Fig.2a).
Ri
Stresses a c ting on the i-th edge (see Fig.l) are assumed in

664
di ff er n t ial f o rm as follows:

a=E(eU ./en) /(l-v 2 ) = ( lJ Ri. - l L . )/ { H(1-.,)


2 } (7 )
n1 n1 nl n1 1

T . =G( V . len) -G (V Ri. _ Vl, .) /H (8 )


nSl Sl Sl $ 1

(9 )
where
E: Young's modulus
G: Shear modulus
~ : Poisson 1 s ratio
Suffix L a nd Ri mean the
displac eme nt of G and (0) (b)
L
G respectively. Fig. 2 D ~L ln iti on o f H,
Ri 1
H. is the sum of the distance
1
between G and the i-th edge and that beLwccn G and the i-th edge .
L Ri
If there is no adjacent el e ment on the i-th edge, Hi is the distance
between G and the i-th edge (see Fig.2a and 2b).
L
By integrating egs(7)-(9) a l o ng the thickness of the plate(t),
force and moments per unit len g th acting on the i-th edge, Mni' Mnsi'
and Q (see Fig.3) are obta i ne d as follows:
Zi
Qli

~
Mn1. =fhh"
- n1. Zdz
(10 )
ns ,
'J/
(11) , ~n~:

QZ=f hh T .dz ( 12) Fig.3 Force and mome nts


1 - nZl
acting on the i-th edge
where h=t / 2
Substitut ing egs(4) and (5) into eg s ( 7) -(9), and applying to e gs (lO )
(12), force and moments acting on the i-th edge are represented by
the displacements of G and G as f o llows:
L Ri

M ,=Et 3 { l.~ Ri -m.x Ri -l .~ L +m,x L} / { 12(1-v 2 }


)H. (13)
n1 1 0 1 0 1 0 1 0 1

(14 )

Ri Ri Ri Ri Ri L L L L L} (15 )
QZ,=Gt
1 { W0 +(Y-YG)X 0 -( x-XG)~ 0 -w 0 -(Y-YG)X 0
+(x-XG)~
01
/H.

665
2.2 Equilibrium equations
Equilibrium equations about the force and moments acting on the
element L are given generally as follows:
Isnzds+IvFzdV+~Pz=O (16)
Is (n:z;y-n y z ) ds+ Iv
(F zy-FyZ) dV+1::~x=O (17)
Is (nxz-nzx)ds+Iv(F XZ-F z X)dV+1:M y =O (18)
where
ni: the surface force acting in the i-th axis.
F i : the body force acting in the i axis. In this case F ' Fy are
x
assumed to be zero.
P : the concentrated force acting in the Z axis.
z

Mi : the concentrated moment acting about the i-th axis.

By coordinates' transformation,n ' ny and n become finally as


x Z
nx=lun-mT ns (19)
ny =m U n +1 T ns (20)
nz=T nz (21)
Substituting eqs(19)-(21) into eqs(16)-(18), referring to eqs(lO)-(12),
and the~ expressing the line integral along the whole edge of the
element L by summation of the line integral along eac h edge, finally
the equi librium equations are obtained as follows:

~(IiQZidSi)+IvFzdV+~Pz=O ( 22)

(23)

~(-li QZi xds i + h1 i Mni ds i-Ii mi Mnsi ds i) - Iv F zXd V+1:M =0


y
(24 )

F means body force due to gravity and buoyancy. Body force per unit
Z
area fz is expressed as the function of Z
fz=a+bZ (25)
wh ere a=- Pice t, b=O (in air) (26)
a= pwd-Picet, b=-p (afloat) (27)
w
a=(Pw-Pice)t, b=O (submerged) (28 )
Pw: density of water, Pice: density of ice, d: draft of ice
Eqs (22)-(24) are expressed in terms o f displacements of the
center of gravity of the element itself and the adjacent elements, by
virtue of eqs (13)-(15).

666
I

2.3 Boundary conditions


Stress and displac ement boundary conditions can be applied by
changing eqs(l3)-(l5). For example, when a moment about the si axis
is applied to the i-th edge, M . becomes
* * nl
Mni=M*nis+( Mnie -Mnis)Si / Li (29 )

where M~is and M~ie are the starting and ending value of the given
moment per unit length on the edge. Li is the length of the i-th edge.
when the i-th edge is free
Mni=Mnsi=QZi=O (30)
On the other hand, when a forced displacement is applied to the
i-th edge, the adjacent element to the i-th edge is treated as the
element with a base length Li and a very small height. The displace
ment of the adjacent element is replaced by the forced displacement.
In case when the rotation about the s axis is applied, M . becomes
3{ * *. 1 L L} { 2}n 1
Mni=Et 0nis+(Onie-Onis)Si / Li-lio+mixo / 12(1-v )H i (31)
where O~is and O~ie are starting and ending values of the prescribed
rotation. and Hi is the distance between G and the i-th edge.
L

2.4 Equilibrium of total structure


Compiling the equilibrium equations of all N elements, a set of
simultaneous linear equations in terms of the displacements of center
of gravity of all e lements are obtained in the following form :
Kd = F (32)
d is a vector of displacements of th e center of gravity of all elements.
The number of equations is 3N in a bending problem . By the reciprocal
theory. K is a symmetric matrix.
By so lving the simultaneous linear eqs(32). the displacements of
center of gravity of all elements are obtained. Substituting the
displacements into eqs(13)-(15), force and moments on element edges
are calculated .

3. Application of RBSM to fracture analysis o f the i ce sheet


Kawai et al. (3] applied RBSM to the analysis of ice fracture
problem. Their algorithm is the load increment method with converting
failed edges to free edges . neglecting stress relief. They analyzed
the bending collapse problem of an ice-breaker bow model and compared
their result with experiment, and predicted almost the same collape
pattern as observed in the experiment. Their results, however,

667

-
predicted cracks which were not observed in the experiment. The
authors think this is due to the fact that the assumed stress strain
law in their model was elastic - perfectly plastic and that it neg lec ted
the stress relief accompanied with the initiation of cracks . The
authors have added some features to Kawai - Watanabe ' s algorithm.
The improved algorithm is as follows:
(Task 1) Apply the unit load increment proportional to a given loading
pattern , solve the equations (32) and calculated the stresses on
edges.
(Task 2) Calculate the load multiplication factor for each edge
necessary to fail the edge. Find the minimum of the fa ctors,
and multiply the displacements and stresses for the unit load
increment by the minimum factor, and accumulate them to those
of the previous stage . Then only one edge comes to failure.
(Task 3) Convert the failed edge(s) to free edge(s), and solve the
equations(32) applying the forces and moments, which were
formerly transmitted through the edge(s), to newly defined free
edge(s) in the opposite sign. Add the calculated displacements
and stresses to those of the p revious stage. On the newly d rE ined
free edge(s) , stresses become zero. This task corresponds to
stress relief. In Task 3 loading edges are supported.
(Task 4) Examine whether the stresses on each edge have reached to the
failure condition. If there are such edges, go to Task 3. If
not, go to Task 1. If the desired load value is fin ~ ! li got,
terminate the calculation .

4. Example
The authors analyzed the collapse pattern and load on a laboratory
~xperimental model of a 45 cone, and compared with the experimental
results. A series of tests was conducted in 1980 by the authors, and
the selected model test is one of them. The specimen is shown in Fig.4 .
The mechanical properties of ice sheet used in

~ , ~ , ~Il
the test are given as follows:
thickness of ice 42 mm
flexural strength (push down) 59 .8 kPa
~ \45' Kl1
Young's modulus
(pull up) 23.5
4
kPa
3.9xl0 k Pa
L. 2.~ J :;lIt
UNIT : mm

Fig .4 Test specimen

668
Poisson's ratio 0.333
Specific gravity (ice) 0.9036
(water) 1. 0035
Frictional coefficient (ice-structure) 0.02-0.05
The experiment was conducted in the 2.90m wide and 6m long tank. The
r ecorded l oad showed a transient feature on the initial stage, then a
steady state was attained. In the final stage, the load peaks became
larger due to the end effect of the tank.
During the steady state, a uniform
600mm wide open channel was formed
by the cone. Towing speed was
27.6nun/sec.
The authors generated RBSM mesh as
shown in Fig.5, considering the symmetry
about the cone 's proceeding direction
-'
(y axis in Fig.5). Boundary conditions
are a ls o shown in Fig.5. Fig.6 shows
the loading pattern on an expansion of
the element mesh near point A. N means E
n )(0
compressive load applied on edge. Mn was o
introduced because ._11)
C")
compressive loads acts
on the lower surface
of ice sh e et.
Frictional effect was
neglected in ' this
analysis.
'-"

Fig.6 Loading pattern Fig.5 Element mesh

The fracture criterion assumed was that an element edge will fail
when the fiber stress (membrane+bending) of the upper(or lower) su rface
of the edge comes to push-down(or pull-up) flexural s trength , and
that after failu re the edge wil l not carry any bending and membrane
stress components. with the exception that the membrane stre ss com
ponents arc transmitted through the failed edge when normal membrane
stresses are compressive,

669
The calculated crack pattern is shown in Fig.7, and the numerals
in the figure give the sequence of crack initiation. The typical crack
pattern observed in the experiment is also plotted in bold line.
Calculated and experimental crack pattern coincide with each other to
a great extent.
The ice force measured in the exp e riment is divided into sliding
and fluctuating part as shown in Fig.9. In this paper, the fluctuating
part is compared with the calculated value. The fluctuating part
includes both ice breaking force and the force induced by ice slab
rotation on the slope. The same forces are also counted in the
analysis, except that frictional force is neglected.
The computed and exper imental forces a re shown in Table I. It shows
a good agreement between the estimated and experimental values with
respect to vertical force. There is, however, a rather larger
difference with respect to horizontal force. It is considered that
this is caused by the fact that the frictional effec t was neglected
in this analysis.
Fig.8 shows the estimated cr a ck pattern in case that ice is
assumed to behave like steel and s tress relief does not occur. The
computed necessary loads are also s hown in Table 1.

, . "

~
..
",
~ -
/ I' - --f

,!!,- '-

.,

, ,
''' , \
' "
.- '.'.
, '. ~-.-,/<;'"
, _fr,"

" \ ',\,' ', ',


~. \ j ,I
- ,\ _ . ~ ~- - --- - - ~ ,-

I '

Fig.7 Crack pdttern Fig.8 Crack pattern


(without stress relief)

670
~
. -- .1

Fig.9 Ice load vs Ti me curve


o FLUCTUAIINC

; . -
'IV
,
SliOING

PAAT

I T t ME_

Table 1 Comparison between calculated and experimental val ue

R.B.S.I1.

with without
experiment
stress relief stress relief

horizontal force 96.0 N 73 .6 N 122.0 N

v e rtical force 83.5 N 68.7 N 122.0 N

5. Conclusion
1. In analyzing ice sheet fracture, stress relief cannot be neglected.
2. RBSM with the concep t of stress reli e f has been proved to give good
results by comparing experiments with computational results.

Acknowledgement
The authors are grateful to Prof. Kawai and Dr. Watanabe for their
kind advices about RBSM. The authors also thank Prof. Hirayama (Iwate
Univ.) for his kind advices about laboratory ice model tests.

References
1. Kawai, T.: New Element Models in Discrete Structural Analysis ,
J. of th e Socie ty of Naval Architects of Japan, Vol. 141,

pp. 174-180 (1977)

2 . Kawai,T.: Collapse Load Analysis of Engineering Structures by


Using New Discrete Element Models, IABSE Colloquium, Kopenhagen,
(1979)
3. Kawai ,T. and Watanabe,~I.: Simulation of the Bending Collapse of
Ice Plates Using a New Discrete Model, J. of the Society of Naval
Archit e cts of Japan, Vol.147, pp.306-315, (in Japanese) (1980)

671
4. Nevel,D.E.: The Ultimate Failure of a Floating Ice Sheet: Proc. of
IAHR International Symposium on Ice Leningrad, pp.17-22 (1972)
5. Bercha,F. G. and Danys,J.v.: Prediction of Ice Force s on Conical
Or l shore Structures: ~Iarine Science Communications, 1(5),
pp.365-380 (1975)
6. Ralston,T.D: Ice Force Design Considerations for Conical Offshore
Structures, Proc. of POAC 77 (1977)

672

THE ESTIMATION OF CRACK PATTERN ON ICE


BY THE NEW DISCRETE MODEL

DISCUSSION
by:
Rene Tinawi, Ecole Poly technique, Canada
(1) The membrane e ffect in your mod e l is only r e quired if you are
considering large displacement theory. Are you in fact assuming a
large displacement theor y ? If so, why?
(2) Your failure criterion is based on results of uniaxial stress
in bending of c~ntilevers. How valid this assumption? In fact
your finite element model does yield a multi-state of stress in
your plate.

AUTHOR'S REPLY
by :
N.Yoshimura, Tsu Research Laboratories, Nippon Kokan K.K., Japan
(1) We do not assume the coupling of membrane and bending stresses
by the large displacement theory. We just calculate the total
stresses in ice sheet by adding membrane and bending stresses.
And the total stresses a r e compared with an assumed failu r e
criterion.
(2) We select the failure criterio n for the sake of simplicity.
If the valid failure criterion for multi-state of stresses will
be developed in future, we can easely apply it to our n~merical

model.

673
STUDIES OF ICE ACTION ON PUMPED STORAGE

POWER PLL\NT STRUCTURES

Sokolov I.N. M.Sc.( Eng.) The B.E.Vedeneev VNlIG, Leningrad U.S.S.R.


Gotlib Ya.L., M.Sc.( Eng.) The "Hydroproject'! Institute, Moscow US.S.R.
Dick P.G., Engineer The "Hydroproject" Institute, Volgograd U.S.S.R.
Research Di-vision
Ryabkin G.M., Ellgineer The "Hydroproject,r Institute, Volgograd U.S.S.R.
Research Di'Vision

SUMMARY

It is the purpose of the present paper to report results of observations on


ice thermal regime in basins of a pumped storage pO\l\r plant under conditions of
the moderate-continental climate. 'The vvithdraV\.6l stru.cture of the pO\.r\.er plant is 10

cated on the bank of the reservoir of the volume by t;v'-'O orders larger than the vol

ume of the upper basin. iVleasurements of changes in vl.6ter temperature were made

at the sites of the ",o.ter intake in the upper basin and of the withdrawal structure

in the r e s e rvoir. Ice observations v...ere performed in the upper basin including its

banks.
The studies w~re conducted over the period of four years (1976-1980). Meas

urements "\ere continuously taken starting from the initial ice formation up till the

moment at vvhJch ice cleared from the basins in spring. In the course of observa

tions the records v\ere obtained: on heat excha nge and heat losses in the basins,

ice cover formation in dependence of water temperature variations, specific features

of ice accumulation in the upper basin and ice action ugainst the basin slope pro

tection.

674
The structu res of th e pumped storu g e plant include the uppe r basin, approac h

canal , pens tock s , pov."er h ous e u.nd disc hDrge c ( ~n,31 t o the lOVl..r r e 5 "tvoir. The
' I ::;0+90
area o ( th e upper basin i s 1100 x I] ~) O rn, that of lra.nsiti o n e qucl. l ~ to 485 x - - 2 -- m.

T he dra \,vdo \,.vn height of th e uppe r bas in is 6 m. Th e approach canal '-vith the water
, ~ 90+50 , , '
Intak e is ..)10 X --:c-'- m. SIX tu rbIn e s ore Install ed at th e pO\r\er house, thre e of

them b e in ~~ of conventionul a nd thn: <.' ~ of reveL-c,ibl e type . Tok:.Ll insL:1J led capacity

o f the p l an t i s 240 MWt. Wul e r discharge o f lh ( ' pla nt ope " a ting in the tu rb ine re

gime is 3 6 0 m 3 / s , ilnd in th e pump regirYE' - 150 m3 Js. A v crd !=-'.0 h ead is 70 m. The

dime nsions o f lh e di ~c h<Jrge c imaJ a re J. OO In ( lhe Ir n:4 h ) by 70- ] 00 m (the width ).


The po\.\er plunt oper,"d es (l c cordi t l~ to either o n e - or l:v\O-cycle diurnal sc h e d

ule. The o ne cycle op e tTJti o n include s thC' turbine r e q irnc - J hrs, power plant shut
dO'\'\IT1 - 1 hr, pump n~~ im o - 7 h rs o.nd pov\oer pla.nt s hut- do\\/f1 d fter the upper o dsin
is Glled - 13 hrs. At th e b,,,,, -c yciQ ope rQtion pow0r pla nt s h'At-down periods Qfter
bct .-; in Cilling dre reduc ed . Th e rdtc of \~c.t tor St': l0 ('l v a ria tion i s 0 .84 m/hr at th e pump

regim e and 2 m/hr - il l th e lurb i n e n ' j lt l e .

A ir te mpera.tut'e Llrnpl itud e is u p to J~oC dU I'i n~ ~\.ointe r . C o o l p eriods are fo llo w


ed b y tha w p e riods nith a bo v e - ze r o du ll y a.ir tC' lI lp (ra tut'e . P o ur-cubl ' co nnectors of

r e sisli..lnce the rmo meters e nable d f"'e li':lble conlrol over th e ir pov..er suppl y and o pera

tion to be provid ed. The p o \\,..:! r sur p l y o f lhennoll J( ~t {' rs w a s e nsured H.;.th lhe h e l p of

t\I\.Q switches . The recording e quipmenl CUI j.....:.isted of hig hly s e nsitive electronic poten

ti ome te l 's vvith millivolt scale.

The thermometers d sse rnbled 'vvith ca.ble conn ec t ors t tnd r e cording equipment

\.o"\re calibrated in Geld. '!'he Llnit e no.bled \;\(Itcr (or oL h <?r medium) temperature to
o
be me<.lsured w ithin the runge o( - 0. 3 to +1.0C; +2.0 C etc . d epe nding on C1 numb e r

of VIIUvc -range s\A~tch s ti.-lgC.S . 'I'he rm o m -' ~ ~ I.s w ere m o u nte d ut ~ to 7 points o ve r the

depth a t consta nt sites l oca ted ab 0 ut 60 m fro m the "", te r intake trashrack s in th e

upper b a s in and 1 B m from the vv1 Lhd rLl~ structure tro s h r"Ll cks in the n ~ ( ' rvojr.

Automa tic control units 'v\re switc h e d in before th e ic e C3. p p~d red and svv.itched 0((

a lter the i CE> cleared from l h e basins .

Temperatures vl.re recorded a nd tran::n.l.illcd to d poten tiometer tape a t 6 or

7 minLrtes intervals. Inte rpretCttion of records obl:ujned \I\ i 1.-' pe rforlTE'd once an h our.

Variations of wat er temperatures, air temp en.ltures d nd vind speeds for i ndivid

ual p e riods of time ,t rl' S hO'.\1 1 in fi g un ...; 1, 2, J and 4- f o r comparison. Fig. 1 in

dica tes \lva te r tempera ture chara cte ri s tics in th e upper a nd I O\.o\e r bas in s a t p C' rsist .....

e n t sub - zero a ir te:np e r u tu n""' s ranC!,ing [ro m -4 t o _14C a nd vwn d s pe ed [rom 4 to


0
1 4 m/s. Wilter temperature in b asins clec re Clsc d from 3 . 0" to 0.44 C durill n 4 to 7
December, 1978. Such consid e rClbl e wate r cooling a t th e site of the wilhdnm,,1 struc
a nd
ture",in th e upper basin \!Vd S record e d thr'ee times duritl G. the four years of observa

ti o ns. A l lhe lm;est air temperaturC' "" recorded in lhe early part of the peri od of il

675
perma11ent ice cover arter thaw days with above-zero air tCIJljl :_'rature s up [0 -). "iC

the te mperatures of wu.t(.?r in t h L~ ufJp(::..r basin '-inc! in th e n.. .;( ' rvoir \''Rre 0.17 a n d
o
O.32 C I respectively. Later after the polyny,.~s had decr\":>tlSP d the cort'e spol lrting
\<\Oter temperntuIT' C' became 0.32 and o . 'uoe (Fi' . 2).

In the \>\6.ITnc st days ot the \,'\.i nter tha\v geriod , ii.trr t\,vo v..eeks of l o\-\' tem }( 'ra
0 0
tures dow, to _'2.8 e the water t e mpe rature incre ase d to ').76 dn d 0.84 e in the up

per basin and re s e rvoir. respectively. In thaw days ".Hh air tem perature _3C th e
0
water temperature decrease d dmvn to 0.35 e in the upper \)d~itl and O.1 ,'OC :n the

reservoir (Fig. 3).

Water temperature va riations in the upper basin and in the, ' o rvol," at two

cycle operation 01 the pov\er plant under w inLt'r cOl 1clitions a re shovvn ct S l?ig . 4. .

Diurnal air temp(.? rature variations ranged from -17 to + 7C. At tvvo~cycl e ope ra.lion
o
of the p lant the "vater tempera ture dec reas <!> d dmvn to a minimwn of _O.ll e in the
o
hours o f the plant shut-dov"n after the turbine '"(.'Clime a nd to +0. 2S e a t the onset

of the pump regime.

The investigations shovv tha t \i\eter temperatures have a pronounced cyclic

variation dependent on the operation n ?gime of the pcurt ped stor",g e poV\er p l a "t ,-" ,d

the heat exchan ge in the uppe r basin and the reservoir. At the plant under di s c us
sion it is the lov..er reservoir ,,\!hich mostly contributes to the heat storing regu!c;Jtion,

this r ese rvoir being by many times larger than the upper basin. Slightly higher tem

peratures occurring s ometimes in the upper basin are , ittri butc'd to stron.~ vvind di

rected t OV\6rds the v"thdt"aAaI structure (Fig. 1). In a number of Cil:SL.'S higher tem
peratures in the upper basin \l\re r e corded in th:nv periods ""hen the plant ope r..J.ted

in the pump regime (Fig. 3).

Alongside vvith continuous records of \o\ater te mperatures at the sites of the


and
withdra'A8l structure,/"'o.the water intal-;:e, as well a s at .-,0vC?ral sites tarther out in the

upper basin individu.al series of additional observations and ITI "" 05Uremc nts \.': ( ~re

made during 1978-1980 .

During not very cold winters initial ice for m ation in the uppe r basin a nd ap

proach canal VI6S observed VIohen the summation of sub-zero averag e diurna l air

temp eratures reached 100C. In the upper bas in more or 1 s s stable ice cover 10rms

in the area of open "",ter at the level of the dead sto rage . Due to tilling and empty

ing of the basin every 24 hrs the ice co..rer is cracked lnto larg ice snE',-'ts.

The Zone of basin banks bet;\en the l eve l 01 the dead storag and the re

tention 1A6ter l evel is the zone 01 an un s table ice cover (the b re u k-up zone). An
ice prism form s here as a reS Ult 01 lragmented ice s e tuing on the slope R. nd Lreez

ing toge ther with shorelast ice during the diurnal drav\down of the r e s rvoir. Th e

large s t triang ular prism of ice formed on a 1:4 slope 1A6>', 12 m wid0. and :; m hl ,q h.

In its front portion cavities are formed due to non-uniform deposition of frd ' m(' nted

676
ice, and melting of the prism during the basin filling. The length of the ice prism
measured along its horizontal area surface W3.S 1 to 4 m. The front side of the
prism can be Sf'< ' n in figures 5 and 6. lee density in [ragments jointings varied from
0,6 t/m 3 in t.h9 cantilever portion of the prism to 0.9 t/m 3 in its upper ",edge. Studies
performed earIler yielded th" density of masS ice on a slope ranged from 0.28 to
0.83 t/m 3 , the density of 0.65-0.74 t/m 3 prevailing.
In the approach canal an ice cover freez(~s in periods of pO\l\.er plant shut-down
0 0
at persistent air temperature _5 C and below. At an air temperature of about _25 C

at pO\'Ir plant shut dOIMl for a holiday the ice cover thickness in the approach ca
nal amounts to 0.05-0.07 m. At dralIDolMls of upper basin ice clears [rom the ap
proach canal. Ice sheels and [ragmenls are accumulated in [ronl of the ,\Gler intake

where they are gradually broken and passed through the turbines. Ice cover [rag

ments are retained in the upper basin by the boom wilh metal floats localed at the
site where llow velocilies equal lo 0.7-0.8 m/s.

Ice action aga inst the slope protection of the upper basin is mainly character

ized by loads induced by the ice prism. They can be subdivided into the prism

dead weight over the contact with lhe slope and the load of the prism cantilever
portion. A protection base is subjected to the lotal pr""sure of the ice prism only
during the periods of basin emtying. According to measurements the resultant of the

pressure of the ice prism 10 m long and 2.5 m high il.ftcr the basin dra\<vd ow-n was

equal to 1.2 t-f/m 2 . The pre ssure was measured with the slring ~,('nso rs embedded
inlo protection plales wilhin the zone of ice freezing.
The action of the cantilever portion of the ice prism logether with the ice
cover frozen lo it on a slope protection a l the basin emptied is suggesled lo be
considered by two stap c' ;.. ; . Ice cover break-up in the zone of its adhesion to the
ice prism occurs a t lhe initial stage of the dra,\down of the wa ter level by 0.0 5
0.1 m during 1.5-2.0 min. The most part of the cantilever portion of the ice prism
during this period continues to be suspe nded and its pn-:.6.6ure on the slope protec
tion is minimum. The rlJd.ximum pre --.-.ure of the cantilever portion occurs only after

the drawdown of lhe w aler below the level of lhe cantilever portion foot. A t lhe

power plant under consideration the maxlmJ.m V'alue is reach ed in a n hour at the

dl~awdoV\rf) rate 2 mph.

The two- s tag" scheme of the evaluation of ice ac tion on lhe slope prolection
at drawdo\>vn of the upper basin complies vvith the physics of the phenomenon observ

ed in field and e . a bies the estimated loads lo be reduced. The a ction of the ice

cover frozen to the cantilever porti o n of the prism al high rale of wate r lev el varia

tion is of rather dynamic than slatic n a ture. Besides, tha acti on "gainsl lhe slope

protection and the break-up of the ice cover occurs not in the contact z one but in

677
the cantilever portion of deformable ice. As a result the torce of the ice thrust is

reduced.
The action of the tront portion of the prism on the slope protection can be

considered using the scheme of cantilever work under the triang ular load 01 the

dead w e i g ht of the ice m a s s.

The data on forecasts, analyses and studi c& of ice thermal conditions obtain

ed at thle Kievskaya pum ped storage power plant are p,;blished by various ,,-uthors

/ 1-10/.

1.A\ th e pumped stora g e power plant operating under the condition s of th'? mod

erate continental climflte the water tem qeratun'.;;; in the lower re s ervoir which is by

two orders la rger than the upper basin appeared to be higher during the w hole w in

ter.
2. Diu r nal va riations of \'va t,er temperature in basins during the lransitio:t and ice

cov er periods have pronounced cyclic nature, vvhich i s con s istent wi. th the o::,eration

sch ed ule of th'? pumped sto re ge po""er plant.

3 . At the ::J:~e cycle> operation m'::Jre equalized heat exchange is observed ben,\en
the upper b s in and the 10l',2r re se rvoir th',,,, at th e tw::J-cycle opcrution . Two-cycle
schedu.le of the po"""r plant operation contribute,; to more intensive water cooling in

the basins.

4. Arp roach canal.s of large length are re s ponsible for increased heal losses

and th..:: fOi.~mation of additional amounts of ice. Elimination of approach canaJs seems

to be preferable.
5 , Ice action on slope protection of the upper basin is mainly characterized by
the impa cts of the ice prism and ice cover frozen to it during shut dovvn.s of the

pumped storage power plant. The deC","' ,l s e in th')s e load effects can be achieved

by l1attening the bank slopes or by providing", berm in the zon e of the ice prism

formation,

RE.FE.!<E.NC E.S

1. Dotse nko T.P. K;evskaya GAE.S na r. Dnepre (The Kjevsk aya pumped stor

og power plant on the Dnieper river). - Gidroter::hnicheskoy e stroitelstvo, 1963,

N 5 , p. 1-8.
2. S,noUn V.I., Sokolov ioN. Ledovye zatrudnenia na Kievskoi GAE.S i meroprl
yatiya po ikh u.slraneniyu (Ice tro'-,bles at the K.ievskaya pumped stora ge power

678
plant a.nd rrcvc-nting mCilsures). - Trudy koordinatsionnykh sovesh chlJ.nij ro gid

rotckhnikc. L., E n cr.~il. 1908, vyp . 12, p. /lOJ-414.

3. S hmulson 0.0. Kievsk aya gidrCD k k umuli ruyshchaya e1ektrostantsiya (The

Kievsku YiJ. pumped storage pOV\.Qr pli.J.nt). - Gid rotekhni ch eskoye stroitel stvo, 1972,

N 4, .,.,. 3-S.
4 . Po wshni k 5.1. N?kotorye vo prosy ekspl u.:ltatsii Ki evskoy GA ES (On th e
rroblems r e l a ted to ",pe ruti on of the K ic v :-; k0Yu ' pumped sto rdg" po"""r plClnt). -
Gidrotckhnicheskoye stroitelstvo, 1976, N 7 , p . 9 -12.

5. Rubanik M .N., Vonokov V.K. GA ES po rroterialam naturnykh nablyudenij

(Ice conditions in the upper b asin of the Kievskaya pumped s torago power p l ant

in accordanc e wi th field observations). - Trudy Gidroproekta, 19 78 , N 5 5 , p.100 -109.

6. Sokolov I.N., KovCllevsky 5.1., D onov A.A. P rognoz ledotermicheskog o re

z.hima bjefov K.:evskoy GAES i ego op ravdyvaemost v pervye g od y ee norrralnoi

eks"I uatatsii (Prediction of the ice thermal conditions in the basins of the Kiev

skaya pumped storu: ;l! pov.er plant and its jus tific ation during SO:Te early years of

th e "lant normal o peration) . - Materialy konferentsij i sov eshchanij po gidrotekh

nike. - Energia, 1979 , p. 23-25.

7 . Dick P.G. l'atu rnye tedotermich eskie is s l edovani ya n0 Kievskoi GAES

(f"ield studies on ice thermal conwtions in th e busins o f the Kievskaya pum ped

storage power p10nt). - Materialy konferentsij i soveshchanij po gidrotekhnike. -

Energia, 19 79, p. 2 3-25.

8. GoUib Ya.L. I f',.c;jed ovaniye teplovykh kharakteristik vodoem ov GAES


(Studies on th e thermal charac teri s tics of th" basins of the p wnped storage powe r
plant). - Sbornik nauchnykh trudov Gictroproekta, M., 1979, p. 85-93.
9 . Sokolov !.N. LChet Ied ovykh yavl e nij pd stroiteIstve GAES v surovykh

klimaticheskikh uslovi yakh (Consideration of ice conditions v.. .h en con structing the

pumped sto r a g e power plant in severe climate). - Izvestia VNJIG, 1980 , v. 143,

p. 83-87.
10. Sokolov l.N. Ice conditions in reservoirs of pum ped s torage power plants.
- LAHR Symposium "lee and its action on hydraulic structures." Reykjavik, Iceland,

7-10 September, 1970, R- 4.9 .

679
9 7 8
c
l:
~~
:t: /'
Ol ______~_L_ _ _ _ _ _~L__L~~~_L~~_L~

_~ I .. 12
8

~ !=--+'-"F----. ,,,,'NI
O~~ __~~~____~-L__~____~~~-L~

250
225

POO
v
_ 175

""ir ISO

05
/ 00
J?S
050

P'\c.. 1. \"'Vatcr lernpc r <.J. l ure v.:ldu li o n D S ;.'1 fun c ti o n o f o ir

tempero. tu re, 'wi nd di l"I ~c ti on dl1d sp( ' ( d W1d e r

vari ous ope r", )ti, I '" condil i ( >r L' o[ th e pumped :~tor-

1978.

1 - wute r level i n th e uppe r ba.sin ; 2 - \,, ,ind

directi o n Llnd s p e ed; J - cu r tempemture ave r

<..l~e d [o r a l O-rn..inutL' s, inl8 rvul ; 1, S - avpra a,.e

te lllpc I\.J.tUI"'t..- in th e uPf J(.;' r' lA-l s i ' l and IT"s c rvoir,

680
~
" "
::2 O ~ __~__~______~~-L~__~-L~L--L~L-______~

I
a\.)- 16

-20
J-N-28
.1 I

m
0"
07S
., ...
. ----- ~~
- - I-. I'-
/1 'o.
-
.... .
-11 -- .-' -\
._ .
. ..'.... .f r ................

,.;: ~~. ~~ .~ ~< . ..-..; S~..


. ' ... - ,:\L - - ~
o . .

Fig. 2 . \Na te r tempe roture varia tion as a [un ction of .:ti r te rn pc n:th..tl"e,

vvin d direc tion a nd !-,", p\. ~ ( ~ d under vd rious operati ng conctiti Ol ls

of the pumped s tora!; e pOl>\ r p l an t for the period of 30 D e _


cembe r 1.9 78 t o 3 Ja nuary 1.979 .

1 - \l\e.ter level in th e uppe r b~ts-in; 2 - "\~nd dire c ti o n and


speed ; 3 - a ir te mpera ture a ve ra ged for a lD -min utes inter

v aJ; 4 , 5 - aven:.J r1 e w,l.tcr temp era ture in the uppe r basin


and rese rvoir, r espec ti vely.

681
I

Pig. 3. Water temperature variations as a function of air


temperature! wnd direction and speed Wlder var

ious operating cone. itions of the pumped storilge


pm\.r plant for the period of 11 to 14 January

1979.
1 - V\Bter l evel in the upper basin; 2 - ""nd

direction and speed; 3 - air temperature aver

aged for a 10-minutes interval; 4, 5 - average

water temperature in the upper basin and reser

voir! ro s pectively.

682
!!T 7
r
Ig 2 21 22 2

V\~I ~ ~
; I Y 0\~~"Ij\~V
~\Jr~
, ! 1
I I
\ 1\;
1'"
. 'I
~ 0
E N ~ NO
~ "",,-1"1""0
"~. NM<!
- ~' .i""'~Jot
I I -, I I ' I
8 _._J_ J- I
, I
, I I
I I ;T 1-~_ I i I
4
~ 0
I---r
- f- l-. I
I
- I, '' '
Ii
I -L VI I . II JIll,

tL
~ -4
.~ -8 I
~I
ll-
k
...... '" ('t'
/\
~
-1i 11. ,\ i~
l' 'I
I..
1M
II
"I
.., -12 h.
""i "\ ..J 1\1\ JY .... I
-16
- - .
~ T I
11.1 L I . - 'I'
..las
~~~y. ~~~~ i~
I
.. r":<I" 1.'-''1 ''
T.'
025 f-
o I ' 'Ii ':tr i ......

Fig . 4 . vVate r temp erature v ariation as i) function of air temlJc r n ture. ~nd

direction and s p ee d under v arious ar K' r u ling conditions of th e

pumped sto rage plant [o r th e period o [ 19 to 23 February 1979 .

1 - wate r lev el in the upper basin; :.> - Vlind direction and speed;

3 - air tempera ture averaged for a lO-minutes interval; 4 , 5

avera~e \l\6te r temperature in the upper basin a nd t"e s c l'Voir,

r espectiv e l y .

683
F'ig. 5. Ic e conditions t th o draVldO\"K\ of the upper b as i n

1 - i c e pri5 m on the ba nk sl op ; 2 - b roc5!< - up zone of lhe ice cover;

3 - s olid ice cover within th e a. r C'c::l a t th lev e l of the d Qwd s tor a g e .

February, 1 9 8 0.

Fig. 6. Ic e prism Vlith caviti e s formed (front view)

1 - ic e pris m on th b a nk slope; 2 - break-up zone; 3 - ice cover

within the area of the level of the dea d 5t01"',I''-' .

684
DESIGNING ICE BRIDGES AND ICE PLATFORMS

L.W. Gold Division of Building Research Canada


Associate Director National Research Council of Canada
Otta"a, Canada

The current basis for design of ice bridges and ice platforms is considered.
Experience and performance observations showing that the moving load problem can
be treated analytically using the theory of thin elastic plates on an elastic
foundation are summarized , and an opinion is expressed as to the information still
required. The basis for the design of ice platforms, for which a validated
analytical method has not yet been established, is revie"ed; and the possibility
of progress on the problem, provided by the capability for measuring strains in
ice covers, is pointed out.

685
One of the earliest descriptions of the construction and use of an ice bridge in
Canada is contained in the records of the parish of Sainte-~larie-~ladeleine [1].*
In 18 78 it was de c ided to construct a new church of stone that was to be obtained from
a quarry on the opposite side of a nearby river. Winter was late and it was not until
19 March 1879 that a road was completed on consolidated ice. About 360 m3 of stone
was taken over the ice by horse and sleigh by the time melting began and water started
to flow over the bridge. This I;ork was completed without accident, although it "as
cont inued right up to the time of rapid deteriora tion of the ice. It probabl y is
representative of the experience-based practice of the da y.
In many northern countries ice covers are used extensively for storing logs in
preparation for floating them to the mills in sprin g. By the 1950's loads in excess
of 50 tonnes were being placed on covers routinel y , with relativel y few incidents
[2,3]. The preparation of the ice for such loads "as based primarily on experience;
the individuals responsible for it generally had no engineering or technical
education.
Th ere has been a grol;ing use of ice covers for non-routine purposes , particularl y
for construction and for transportation and the de velopme nt of re sources in remote
areas. As a result, considerable attenti on ha s been given during the past fel; ye ars
to placing the desi gn of ice bridges and platforms on a proper engineering basis.
T"o approach es have been used in developing this basis. One is the dete rmination
of the allowable load for given ice thickness from records of experience and observa
tions of performance. The second is the specification of the "failure" condition as a
limit s tat e. In the limit s tate approach it is necessary to de ve l op a valid mathema
tical description of the behavi our of i ce covers under load and of the s trains and
stresses induced in them. The "failure" condition must then be defined and the allO\;
able load specified as some fraction of the failure load.
It is very difficult and perhaps impossible to specify all the failure conditions
for an ice cover. There are several reasons for this. A principal one is that ice is
normally at a temperature I;ithin 40 Celsius degrees of its melting point and,
therefore, in a "high temperature" state. Its strength a nd deformation properties are
t emperature- a nd time-dependent in this range.
Ice in i ce bri dges and platforms is subje c ted to a wide variation in structure
and qualit y . although careful control may have been exercised in construction. In
addition. temperature changes cause cracks and water currents cause erosion and thin
areas. Surveys of i ce failures have sholm that most acci dents are due to imperfections
in the ice cove r or to effects that ha ve not yet been properly accounted for in design

The author "'ishes t o thank Profes sor B. Michel for bringing this reference to his
attention.
686
or usc, rather than to the exceedence of the allowable load for the average thickness
of ice present [2,4 ]. The existence of thes e imperfections and non-normal conditions
must be expected and can only be taken into account ~hrough careful observations
during use.
Because of lack of knowledge concerning the deformation be haviour of ice and
because of the variability in factors that de termine the strength of ice covers, field
data are s till the principal basis for the specification of allowable loads. The
th eo retical descriptions of behaviour now evolving with the development of the limit
s tate approach are providing a rational mathematical framework for the analyses and
presentation of this experience. Too often, however, experience is not recorded. In
addition, it is not yet possible to measure in the field all the factors necessary to
describe full y the response of an ice cover to load. It is relatively easy to mea s ure
deflection; but only recently have methods for measuring strain been successfully
demonstrated [5,6] . There is as yet no satisfactory method for measuring stress. If
the use of ice covers for supporting loads is to become more of an engineering science
than an art it will be necessary to develop and demon s trate design methods and
criteria that can ensure safe and sa tisfactory performance. This paper is a brief
review of the current s tate of the development of this knowledge for both moving and
static load problems.

Moving Loads
The starting point for the moving load probl e m has been the theory of a thin
elastic plate on an e lastic foundation [7 ,8 ,9]. Observations indicate that this
theor y should be sufficiently accurate for speeds in excess of 1 km/ h as long as
proper acc ount i s taken of the strain rate dependence of the elastic modulus [10]. At
low speeds the shape of the deflected surface is essentially the same as that for the
s tatic e lastic case [II]. If it is assumed that load, P, acts over an area of
effective radius, a, then the expression relating P, ice thickness h, and maximum
stress am (which occurs under the load) is

11 a bh 2
m
P 8(b) a h 2 (2)
3(l+\!)kei 'b m
where a
b
i
!

~
3

Gl2pg(I-
Eh \! 2~

E elastic modulus
\! Poisson'S ratio
density of water

687
kei ' (b ) i~ the fir st derivative of one of the
mouified 8esse l functions.

FOT the linlit s t a te, the maximum stress i s Jssumed to be the ten s ile strength of
icc. If this propert y of the icc and th e modulu s of elas ti city are kno"n, icc thick
ness required for given load s can he determined u si ng s uperposition (and computer if
necessary ) for co mplex load geo metric s [8,12].

Obser va tion s g ive ~ 16 h" m for fresh "ater icc [3]. This corresponds to a n
elastic modulus of G. ') x 10 3 MPa, a rel a tiv e l y high value. Si nh a [1 3] ,ilo",
that this value is associated "ith a reriod of l oadi ng of about 20 s at -Ioe. It
"ould be ex pected that th e modulus of ela s ticity co uld relax t o about 50 % of that
va lu e for very slow l y moving load s . Suc h a decrea se " ould cause a de crease in ! of
about 16 '6 . If the effective radius of the load ar ea i s 1. 25 m, the corr es ponding
increase in B(b) is less than tbi s a mount.

If the effec tive r adi us of l oading 1S 1.5 m, B has a valu e of about 0.75 for
h ~ 0.25 m; 0.5 for h = 1. 0 m; and 0.42 for h = 2 .0 m. For th e ran ge of ice thickness
and e ffe ctive area of l oading us ual for ve hi c l es travelling on i ce its value is about
0.6.
In vi ew of the uncertainty in th e time dependenc e o f 8(b) and in the va lu e t o be
used for ten s ile s trength, it is often ass umed in the a na l yses of performance data
that
P' (3)
g

where pt is in kg, h in m, and A is a constant.

Observations on the successful use of i ce covers have s ho"n that loads ar e us ua ll y in


the range P' = 3. 5 x 104h2 to 17.5 x IO"h 2 . The usual recommended upper limit f or
lo ads to be pla ce d on ice of thickn ess h is about P' ~ 7 .0 x 104h 2 . P' = 17 .5 x 10"h 2
defines the approximat e uppe r limit for situations for "hich ri s k is acce ptable (e.g.,
tank s in warti me). Durin g a s urve y of "ood pla ced on ice cover s , nin e f ai lures
occurred durin ~ the placin g of 42,500 truck l oads for which loading s r a ng e d from
P' ~ 0. 7 x 10"h " to 10.5 x lO"h 2 . Th e distribution in the number of lo ads "h e n
p l ot t ed again s t P' / h2 "as approximat e l y normal, with th e max imum at P' ~ 4.2 x 10"h 2
[3].
e ' perience ha s sho"n that good quality ice covers no t subject to thermal stress
should suppo rt moving load s sat i s factorily to l oadings of P' ~ 14 x 104h 2. The ice
should not be s ubjected to repe titive loads at this le vel a nd its use mu s t be under
the control o f an individual kno"l edgea ble about ice and the factors th at de termine
the stre ngth of ice covers. For uncontrOll ed si tuations expe ri e nc e indicates th a t
loading s should be res tri c ted to pi = ~.5 x lOlt h2, but ,.'ven the ll f;jj lures call be
ex pected owing t o imper f ec tions in th e cover :lnu th~ effects of therma I stress.
688
[f it i s ~Jssumed th nt B hns n vR Ju e of 0.6, the rnn ge of mn x imuln stre ss nssoc ia
ted h' ith 3.S x 10" < P' / h 2 < 17 . S x IO~ i s ~bollt D.S t o ~ ~lr~. Th e upper v~ llIe
e xceeds the s tr ess thnt \,.,ouJd be ex pec ted t o C<H I ~e c r3cks t o form ;It th e und e r s urf<l cc
of the cover, but exper i ence ha s 5hol,.,n th a t <l c over ca n tolerate suC'h cr.1ch formntl o n
for mo v ing load s \.,.'i thout fni lure. Expe ri e nc e ~lso s hoh' s that the IOh'cr val ue is ~

safe, ~llowable, ten s ile stres s for i ce .

Although field ob s ervati ons ha ve made it possible to delineate in a reasonable


manner the elastic modulu s and a ll o\\I;1h1e m.:1ximum stress to be used for the design of
ice bridges, much sti ll has t o be done. Relatively little attention has been given to
the effect of temper ature and salini ty. Kerr and Palmer [141 h'l\Ie shown that for the
elastic case the elastic or rigidity modulus can be rel,laced by an effective value for
plate bending calculation s, giv e n by
h - lo
D f l E(l)d6 (4 )

- 60
where 60 is the position of the ne utra l plane and E(l) is the value of the modulus a t
distance 6 below the surface.

The deflection of the cover a nd , therefore, the linear strain va riation through
the ice cross-section is determined by th e lo ad and D. As E(6) va rie s throu gh th e
section, the s tress di s tribut i on is no longer lin ear . Since the bottom s urfa ce of the
cover is always at the melting point, th e maximum s tre ss mHy, in fact, occur a t some
position in the int erior. Th e el as tic modulus of i ce , however. becomes le s s t e mpera
ture depend e nt with decreasing pe riod of loadin g [1 3 ], and the value of 6.9 10 MPa
is probably reasona b le for fresh-I,.,Cl t e r i ce in most moving-load situations and the
temperature range 0 to - 20C.

The effective value of 1 and t e ns ile strengt h will increase with increasing
vehicle speed. Increasing I causes B to decrease, and this partly offsets the effect
of the increase in st rength. Increasing the vehicle speed, hO\~'cver -' causes another
effect that must be taken into consideratio n.

When a vehicle travels on ice cover s , hydrodynamic "ave is set up in the under
lying water. This wave trav e ls with a speed that depends on the depth of the "ater,
thickness of ice cover, and modulus of e l as ticity of the ice. If the speed of the
vehicle coincides "i th that of the hydrodynamic "ave, the deflection due to load
reinforces that associated with the wave. This problem has been considered by
Assur [15], Nevel [16] and Eyre [II] .

Figure I presents measurements of the r a tio of actual maximum deflection t o the


elastic deflection at zero speed. Th e deflection is a maximum at R critical s peed,
u c ' that depend s on th e properties of th e ice cover and the thickness of th e ic e
689
I I

oRE F [11]
REF [25]

a
2 -
... . .
0

>:

.:...
>:

t- 0
o. 0
: ..
0

"0
o
o. Figure 1
Dependence of the ratio of
defle ction, w, at speed u to
the elastic deflection, w0'
0 I I
0 on the ratio of the s peed to
U/U the critical speed, u
c c

[15,16]. Plotting measurements in the non-dimensional form of Figure 1 shows that


deflection at the critical speed is about two and one-half times that at low speeds.
The theory presented by Nevel accounts for the features shown in Figure 1, except for
those at the critical s peed where the effect of dissipative processes that limit the
maximum defle ction have not yet been properly described. The dependence of maximum
stress and tensile strength on vehicle speed ha s also still to be established.
Report s of failure have indicated that speed has been a contributing factor in some
cases. It is debatable whether vehicle speed is an important factor for loads of
P' = 3.5 x I04h2 and less.

More precise information is required on the dependence of 1 and tensile strength


on ice type, temperature distribution, and speed. It would be useful to carry out
additional field experiments sim ilar to those of Eyre [ll] in which strains,
deflections and acoustic emi ssion to detect crack formation are measured as a function
of vehicl e load and speed. Such studies are needed, in particular, for sea ice for
which present knowledge of elastic modulus and strength are appreciably less than for
fresh-water ice.

Stationary Loads
A demonstrated limit state design method for determining the safe thickness for
stationary loads has yet to be established. The rea son for this is the lack of know
ledge concerning the relations among load, deflection, deflection rate, strain, strain
690
rate, and stress. The mathematics of the problem are diFficult because of the
non-linear relation between strain rate and stress.

The appropriate criterion for the stationary load problem is probably one of
performance based on allowable deflection or deflection rate rather than allowable
stress, particularly for loads that will be in place for periods of more than one day.
For shorter periods, maximum strain rates are in the range for which it may be neces
sary to limit the maximum stress as well.

Much attention has been given to the stages of failure for ice covers under
stationary loads. The information provided by such experiments may be misleading
because they involve deflections greater than the freeboard. Perhaps the only
situations for which deflections of this magnitude can be tolerated routinely are
those such as the storage on ice covers of wood that is to be floated to mills after
spring thaw. In general, if the ice is to support material that must be retrieved or
activity such as drilling, there are practical reasons for keeping the deflection less
than the freeboard in addition to the limits that this places on stress and strain
[17l
If deflection is limited to the freeboard, the maximum strain induced is less
than 1% for normal ice thickness. This means that deformation is confined to the
primary creep stage and, from the point of view of the deformation behaviour of ice,
is one of small strain. The implications of this have not yet been fully appreciated
or exploited. In dealing with the problem most investigators have assumed a linear
viscous behaviour or a constitutive relation of the form

(5 )

where E is the strain rate for constant $tress o. and 0, 00 and n are constants.
The value for n has usually been taken to be that found for the secondary creep stage.
Work by Gold [18) demonstrated that for the uniaxial constant load condition the
expression relating strain, stress and time has the form

,; (o,t) = A(t)On(t) (6 )

where A and n tend with strain to their constant secondary creep stage values.
It was also found that n depended on the type of ice and on the strain history. For
simple compression at _Iooe and stress between 0.4 and I.S ~IN!m2, n was independent of
the stress at a given time and changed only slowly with time after about ISO min.
In determining n from creep tests, it is usual to use the linear region of the
creep strain/time curve. This region occurS over a time period that depends on stress
and temperature. The value of n to be used in the constitutive equation for bending
problems, however, must be determined from strain rate values for the same time for

691
-~

each stre ss and temperature since "hat is required for ca lculations is the s train rate
dependence of the stress at a given time and temperature.

Murat [19] observed that the maximum deflection vers us time for beams subjected
to a constant four-point load and simpl y-suppo rt ed plates subjected to a constant load
at th eir centre became essentially line a r after an initial transient phase. This
constant rate of deflection developed while th e maximum strain was still o f the or der
of 0.1%. It required about 5 h to establish the constant defl ec tion rate for the
beams at the IQI,'est load level; f or th e plates, it t oo k over 40 h. Taking the se
results and the earlier work of Gold [18] and o f Krausz [20] into consideration, it
seems clear that the transient phase before the apparent steady-state behaviour must
be associated with the transition from the initial e lastic condition to the condition
for whi c h the st ress distribution through the section depends primarily on strain
rate. The strain to do this i s in agreement with the time dependence of n observed by
Gold.
Masterson et al. [5] state that defl ec tion of the platform the y observed var ied
with time raised to the power 0.47. l'/hen th e deflection measurements presented in
their Figure 6 are plotted on a log-log scale, however , thi s dependence is found only
for time in excess of about 10 days . The ratio of their measured maximum st rain to
deflection, assuming that 0 deflection coincides with 0 measured strain, is a lso
co nstant after that time. These observations suggest that for thick platforms
designed to support loads for man y days the initial transient phase ex tends over
periods of da ys in contrast to minutes or hours in labora tory sca le experiments.

These observations are consi s tent with analyses of Sinha [13 ,2 1] indicating that
delayed elastic beha viour dominates immediatel y following the application of lo ad . With
time it becomes progressively less significant "ith respect to the component of the
strain due to viscou s flow. In fact, the evidence suggests that it is the ne glec t of
the delayed ela st ic behaviour that i s re sponsible for the ap parent time (and perh a ps
stre ss) dependence of A and n in equation (6).

Murat [19] found from his experiments on beams and simpl y-supported plates that
at a given time during the period of constant deflection rate

(7)

where P is the constant applied load and L the length of beam or diameter of plate.
He found the av erage value of n to be 2.32 for beams subjected to four-point load and
2.70 for si mpl y- s upported plates. The te s ts were carried out at -10C. The lower
ran ge of the maximum stres s induced in these test s probably overlaps the upper range
for satisfactory performance of i ce platforms support i ng a sta ti ona ry l oad for more
than one day. From the geometry of the te s ts it is po ssible to state (for them) that

692
(8 )

If a general relation exists between 1m and ~ for ice platforms, it should be possible
to develop at least an empirical relation of the form of equation (8) between maximum
strain rate and load.

The measurements of Masterson et al. (5] on an ice platform about 6.5 m thick
confirm that it is reasonable to assume that during deflections under stationary loads
both strain and strain rate increase linearly through the cross-section of the cover
to their maximum values in tension or compression. Observations have indicated also
that for deflections less than the freeboard the shape of the deflected surface for a
reasonable distance away from the load can probably be described by an equation of the
same form as the initial elastic one. If this is correct, one can assume Em ~ ~ ,
where is a function of time and is the characteristic length that fixes the shape
of the surface in the vicinity of the load. Because of the elastic foundation effect
of water, the maximum strain rate is not directly proportional to the deflection rate.

With the capability that now exists for measuring strains directl y in platforms,
it should be possible to establish the expressions relating load, ice thickness,
deflection rate, and maximum strain rate. If this c an be achieved, the performance
criterion can be specified in terms of allo"able deflection, maximum strain rate, or
stress. Under certain conditions it could be stated as allowable maximum (P/ h 2 ), as
given by Murat [19]. If a stress criterion is used, it will be necessary to establish
a relation of the form of equation (5), from which the stress through the cross-section
can be calculated from the strain rates. The deflection rate to be used for the
calculations would be determined from the length of time the load is to be on the ice,
the maximum allowable deflection, and an apparent initial deflection that includes the
delayed elastic component of the strain.

The performance criterion must be chosen so that, when satisfied, cracks \"ill not
propagate during the period when the platform is undergoing the allowed deflection.
Calculations indicate that when deflection is limited to the freeboard for loads that
will be in place for more than one day, the maximum strain rate and stress are
sufficiently 1m, for this to be the case. The limit state conditions, however, have
still to be establi s hed through experiment and performance measurements. As the limit
state values of the critical stress and strain rate are not known, the safety factor
associated with current experience-based practice cannot be spe c ified.

A good example of the current design approach for ice platforms i s that used for
platforms supporting off-shore drilling activity in the high Arctic (5,22,23]. Loads
have been in the range of 500 to 1500 tonnes. The performance criterion applied has
been that total deflection must not exceed the freeboard and that the initial elastic
693
maximum stress must not exceed 345 kPa. The basis for the specification of maximum
stress was th at cracks are not observed to form in fresh-water ice during the first
1% creep strain caused by a constant uniaxial compressive st re ss of less than that
value [24].

Over t ime, the loads placed on these pl atforms ha ve been increased . Platforms
for the heav ier lo ads have been designed on the basis of past experience, primarily by
establishing from performance measurements the time dependence of the apparent value
of 2. Ice thickne ss was chosen to ensure that the allowable amount of maximum defle c
tion would not be exceeded during the period in which the l oa d would be on the ice.
The range of l oa ds placed on the platfor ms is now sufficiently wide to make it
possible to es tablish an empirical e xp re ss ion rel at ing desi gn l oa d , PI, ice thickness,
hi' and deflection, wI , of th e form

(9)

where KO, ho, and Po, are reference deflection, ice th ick nes s and load, and m and n
are constants [22].

Stress le ve ls within the platforms have been checked through computer ca lcul a ti ons
by makin g assumptions that allow strain rates to be s pecif ied and usin g a relation
between stres s and strain r a te of the form of equation (5). It is difficul t to
~pnuralize fr om this experience becau se the load distribution is not simple and the
platforms are tapered rather than of constant thickness. ",ca s urement s now being made
of the time-dependent strains through the cross-se c ti on of the platforms will provide
a bet ter basi s for future use of the obser va ti ons on performance in developing a nd
confirming a more universal de s ign meth od .

Co ncl us ions
- -- - --
Traditionall y , the use of ice br idge s and platforms has be en based on experience .
Observations have shoh'11 that ice can be as s umed to behave elastically for loads movin g
at a sp eed greater than ab out I km/h if proper account is t a ken of the time and
t emp era ture dependence of the elastic modulus. Values of allowable ma ximum tensile
stress detcrmined from fi e ld practice lie in the r a nge of O.S to 3 MPa. The r isk
associa ted with moving loads that induce a maximum stress in this ra nge is determined
primarily by imperfections in the ice co ver and uncontrollable factors such as
temperature changes. Information is required on the time dependence of th e maximum
stres s and tensile strength to establish the dependence of bearing capacity on vehicle
sp eed.
The mathematical basis for the design of ice platforms for stationary loads ha s
not ye t bee n de ve loped. The principal barrier i s the non-linear relation between
strain rate and stress and the difficulties this pos es for ana l ys is. Practicall y

694
.::tIl station:lry l oad problems are as sociated \.Jith maximum strains less than about 1%.
Observations indic a te tll~lt after all in iti a l transi e llt period, probably associated with
the transition of the initial elastic state to the fully viscous state, the stress has
a power law dependence on the strain rat e . The exponent at a given time appears to be
es s entially con s tant for s tre s s greater than about 0.4 MPa; it is not known whether it
remains constant f or strain rates inducin g stress below that va lu e. If the expres
sions relating load, icc thicknes s, deflection, deflection rate, strain, and strain
rate can be established , it should be possible to develop a general design method that
takes into account the initial transient behaviour. The recently demonstrated abi lit y
to measure strains in s itu should allow these relations to be found.

Limit ing deflection of the ice cover to the fre e board is probably a practical
performance cri t erion for load durations of more than one day. This criterion effec
tively ensures that the maximum strain rate sta ys below the critical value required
for propagation of cracks. If deflections are allowed to exceed the freeboard th at
would normally be as sociated with an all -i ce platform of given thickness, or if loads
arc to be stored for s hort periods only, the maximum all ow able stress level and a
va lid method of calculating the maximum stress will have to be establish ed.

This paper is a contributi on from the Division of Building Research, National


Research Council of Canada, and is published with the approval of the Director of the
Division.

Refere nce s
[I] Loranger, M., La merv e ille d'un pont de glace. Apostolet 79 , Ob lat s de Mari e
Immaculce, ~ (2) , IS79, p. 7-S.
[2J GOld, L.lI'., Field study on the load bearing capacity of ice covers. Woodlands
Rev., Pulp 6 Paper Mag. Canada , ~2., 1960, p. 153-154, 15 6-15S.
[3J GOld, L.W., Use of ice covers for transportation. Can. Geotech. J., S (2), 1971,
p. 170-181.
(4) Sundberg-Falkenmark , M., Load bearing capa cit y of ice. Swed. Inst. Meteor. and
Hydrol., Serie s I, Stockholm, 1963 .
(5) Masterson, D. M., Ander son, K.G. and St randb erg, A.G., St rain measurements in
floating ice platforms and their app li cation to platform design. Can. J. Civ.
Eng., ~ (3), 1979, p. 394-405.
[6] C-CORE., The measurement of subsurface strai n on Ro che 0-43 artificially
thickened se a ice drilling platform . Contract Report No. 7S -16, Memorial
University, St. John's, Nfld., 1975.

695
[ 71 h!c stcrg;:J:lrd, IL f\L, New forrn ul35 for s tresscs and s t ra i.n s in concrete p: lvc mcnts o f
eirfields. Tre ns. Am . Soc . Civi l Fng. , 2.! .~'.' 1 9~8, 1'. ~ .,::; 44~.

[ Il ] Wyma n, ~I., ile flectio ns o f an inf ini t e pl ate . Can. J . ltc s. A, 28, 1950, p.29~-302.

[ 9] Ke rr, A.n. J The bC<l rjng ca paci t y of flont jng ice plates subjcc ted to s t ~lti c Or
q ua s i-st"tic load s . J. (; l acio l., .!2 (76), 1976 , p . 22 9-2 (,Il.
[lO] Go ld, L.lv. , Icc press ure "nd hearin g capacity, ~ Geotec hn ica l Engin eeri ng for
Co ld l~egio n s , Clla pt. 10. (cd it ors /\nder sland , O.B . and Ande rson, D. ~1.),

McGr"" Hill, 1978, p. 5 0S-SS I.


[I I] [yre, D., The flexurel mo tion s o f a fl oa ting i cc s hee t induced by moving
vehicles. J. GI"c iol. , .!.Q. ( 8 1), 19 77, p. 555 -569.
[L~] Neve l , D.F ., Safe icc l oads comp uted "it h a po ck et calc ul "to r. I'roc. Wor kshop
o n 13carin g C3PLlCity o f I cc Cove r :-;. Ass oc. Ct tec. Ccotcch. Res ., Natlonal
I( csea rch Counc il of Ca nada, T,1 12:1,1979, p. ~ n., - 222.

[13] S inh a , N.K., Rheo l ogy o f colum nar - gra i ned icc. Expe l'. Mec han. , 18 (1 2), 1978,
p. 464 - 470.
[14] Kerr, A. D. and Palmer, Iv . T., The deformati ons and s tresse s in fl o~l ting icc
s heets. Acta ~1ec hilnj ca , I S , 19 72, p. 5 7 -7 ~ .

[1 5 J Ass ur, A., Traffic ovc r froze n or crus ted sur faces . ~lec h. Soil Vehic l e Sy stem s.
Proc. 1st Int. Conf. ~Iech. So il Veh icle Systems , Torin o - Soint Vincent,
12-J6 June 1961.
[16J Nev el, O.E., Mov ing l oods o n 3 flo~ tin g icc sheet. Co l d Regi on s Eng in eeri ng
Loboratory , U. S . Corp s of engin eers , lIanover , N.H., Re searc h Repor t 261, 1970.
[1 7 1 Frederking, R.M.W. and GOld . L.W ., Th e beari ng cap3ci t y of ice c overs under
s t a tic l oads . Can. J. Ci vi I En g . ~ (2), 1976, p. 288-2 93.
[I 8 J Go ld, L.W., The initia l creep behavi o ur of co lumnar-grained icc, Part I:
Observe d Beh avio ur: Pa rt II An,ll ys i s. Can. J . Phys., ~, 1965, p. 1 4 1 ~- 14 34.

[l 9 J ,Iu rat, .J.R., La copa cite port a ntc de l a g l3ce de mer. Ph .D. Th esis, Ec o l e
I'olyt ~c hniqu'~ J~ elontreal , 1978.
[ 20J Kra usz, A.S . The cr eep of ice in benching . Can. J. Phys., 41 (I), 1963,
p . 167- 177 .
[2 1J S inh a , N.K., Short t e rm rh eology of po l ycrys t a llin e icc. J. Glaciol. , 21 (85),
19 78, p . 45 7 - ~73 .

[ 22] Masterson, D.M., And erson, K.C. a nd Stra ndberg, A. C., Reply to Oiscussion:
Strain measureme nts in f l oa ti ng i cc p l atfo rms and their app licati on t o p l a tform
desig n . Ca n. J . Ci v. [ng., ?- (3), 1980, p. 565-568.
[23] Bea udois, D.J., IVatt s, J.S. a nd Masterso n , D.~I., A sys tem fo r offshore dri lling
in the Arct ic Island s. Offshore Tech. Conf. , Houston, Texos, Paper No. TC 2622,
1976.
[24] GO ld , 1.. \\'., The p rocess of f3i l ure of cO l umnar-g r o ine d ice. Phi 1. olog. , 26 (2),
1972, p. 3 11 -328 .

696

[ 25 1 !leltaos, S" he l d s tudi es of the r es pollse of float ing i ce s he e t' to moving


l O(l J s . Proc . \\lor k sh o p o n l3eaJ"i ng Ca pa c it y of Icc Covl'r:-;~ I\::;SOC . Cttc c , Geo t cc h .
Res " Nation" l Res""rch (ounci l o f CanaJa, T~ I L'3, 19 7'1, p . 1- 11,

697
IAHR DISCUSSION SHEET.
Author: Loren Go 1rl
Invited IAHR Paper, "Designing Ice Bridges and Ice Platforn~ ."
Discusser: Phil Johnson, P. E.
1045 Lakeview Terrace, Fairbanks, Alaska, 99700, USA
Discussion:
I am discussing the second Section of Gold's paper titled "Movinq Loads . "
Gold presents what can be called the "Critical Speed" reaction of a floating
ice sheet to a movinn load althouqh he does not develop the theme completely or carry
it to a logical conclusion. It is not a complete description of the reaction of
an ice sheet to a moving load as I hope to show .
The princiral point is that the moving load will generate a hydrodynamic wave
that \'/i 11 travel through the water and deform the ice at the location of the wave . If
the vehicl e is trave1in9 at the saf'le speed as the wave (the critical speed), the
deflection of the ice uue to the wave is joined by that of the vehicle and the total
deflection (depressiun of the ice sheet) under the vehicle is substantially increased.
~evel [16J (Gold's References) showed that theoretically the deflection would go to
infinity but that observations showed that it merely reached a finite maximum. Gold
combines the data of Eyre and Belthaos to show that the deflection increase does
oceul' but that the maximum is around 2-1/2 tif'les the static deflection on moderately
thick fresh-"ater ice. This all seems confirmed by theory and field tests and I
rio not dlsagree.
Continuing further, efforts h2ve been made for many years to find the tensile
stress in a flo ating ice sheet. Precise solutions have been developed which rlepend
upon cal cul ating the the deflection of the ice sheet under the load. However, the
tutal deflecti0n a~ the critical speed depend upon the two factors which deform the
ice sheet - the weight of the moving vehicle and the presence of the movin g wave - so
it is difficult to find the total stress in the ice as a function of deflection.
It is almost certain that increased deflection indicates increased stress but the
rcl~[ionship l S not kno~n more precisely. Stress, particularly tensile stress in
the bottoln of an ice sheet~ i, ~;:1portant because an ice sheet will begin to fail when
the tensile stress at that point reaches the tensile strength of the ice. Although
Gold did not state this precisely. ! agree with his implication that increased deflection
under the movin g vehicle at the critical speed also indicates increased stress in the
ice .
'lot addressed by Gold or most others who have published on the Critical Stress
problem is the prohabilitv of enclJunterino these increased- stresses. -~ 'Jhile not
698
Phil Johnson

sp'cc ific~lly stated. the lmpres'.inll is q iven thilt a vet.icle travelin~ on the ice is
ver'ya!)t to encounter them and t),e l'e,uHs. also nut discussed in derail. are apt to
bE' IJl1pleJSilnt. 11a ny a',th ors ad v ise that vChi c le sPeeds on ice should be strictly (on
tr0'i'1ed and ~"rt low - usually at '10 hlph or s 100;er - ionOI'inq the fact that the crit
i _)l speed F0r In ice ~ he~t on shallow wJter is at or nedr t hat s pe2d.
eased on the dlScussinll ~ bcv~. wr Cln iJentHy two ccnrlithns that mllst be met
hef"re a iUV vinq 'I<: hi cle nn ,II ice shee t 1'11 11 encOlJOter critici1! sneed C(lnclltions. The
first is that tl,0,re lrust l10 ., hydrodyr;ilillic I'/ave ifI the "iltCI' ....hile the second is
,hilt the "e lliclp nilist b? t"J'/e'li ll'~ I/ith lnis ./dve. Be c" u~,r. nf thi s secor;d (o nditicl],
it arnears that the vehicle its,,1 f ,"oust qcn'.' ~Jtf' the I/dve by travel ing ill ~ straight
'~ ine ., t the uit;c,;l ',pe cd .,n j th'; ll tril ,p l I~ the s" .. '~ dirP ~ tion ilt the Sol,le spepJ t o
utilize coo? tile Cl'it i c"l <.)e(, j c r,:itio% that. '~ ,1V2 b2en :1~'Jelope<l. Fi'J 'j re 1 Gf
Gold's sho.,<; th .l t ," ,,, 'Jchicle' "1 '.' e1 1""". ( ~e vcrs close to th? critical speed for the
dellectiO'lS:O l'~ach ,1 ,nax ill.)ffl. It d;>C"ili'S Lt',a t iF the vehicle speen 'Iaries somewhat,
the full criti c nl 5[Jperl e . w ,u ld nc' DI' ccvu lo!)ed.

,~evel* !! den' ",,' +J'is n.,tt:r f n "n ar,l;thel' anole nnd 5ta,.ed "" .O'IE: sholJld not
trdvel lit d co ns l. Jn ;; ', ~ " ed r:e,. tlrf critic ;j] velocity. '4 ~en Dassinq through t he
cI 'i tical \' loc itv, one do so '1'li rl. l .1 in ord~r to insure ~ t"il ns i~nt rather
';' I: " "I'~

than 1 s ;~ady state rondition." ,'level's I'Icrry )bout the ste~jy-stilte conditi o1s end
inJirference to t~e l r an si8 nts ,nak(>') S ~1i s e ar,d confirms ~he conclusi(Jl1s reJched.
lc
It {' ;1 peilrs th,:.; r il ndOlil chl\n '1 r:.~./ ~pt?,~d and the .lirt?ction nf t rav el '.:hat would OCCUf

'''''.'f, a vehi c le is Lr,.I,.,ling fl'ce l v on un 'I~C ~I l <:et ./oul e: ~c adequ ctr tv 0revEflt the
steddy,~tat<: conditions lwces;;,~l'y to S:.'fle1' dte the :.r Hieal s~eed cor,diti cns allt! th il t
su ,;h ~ vehicle \'10'J1. j have littl ~ c;ldnce of encOllnterinn thr:f'l . H0I1('vcr, i:e trovel is
cften constrained to rl qrticu;rlr trock by clearinfJ an icc road. Tn this case, rar,
rio,,,nes s 'in the tra c ~ follo"ed "ill be eli, ,:lnated. Further. trovel ronditior" on ti'at
tri\( ~. ';/ill tC':ld tu be uniform so that a vehicle wO:lld tend t o f i nd a comfortable speed
and hold it. In such a r~se, ~ vehicle 1'I0uld be ap t to encount er crltical spe~d con
~iti o n s . This could be Jvoided hy rrsting the iC E road with the critiCAl 5~eed and
d,i'!isinr, (~rivers to JVOJid the~ : . T~i) ',:e'u1d c;ompletely aV,)i(i cr'tical src:ed conditi01"
excc nt for c cur~l( '_ , driver.

nne rurther point rprr- ' i , .$ . P-C5~iil'dl into ice reaction t o vehicles
tra vf' linCj on liil i,,~ shen ~as I)pen (unf in e d to l;t!:es wh" ,c t hE: exis' .'nee of t;;e cr it
ical :;P'cl0<i ph{)nWlen,( ru; r,e : " demon<tr;;tf:J and (lUdntifi"d. rcr, tll'.' some ef f p. ct be
found on l' iv<'r ice where tile "/a~.er i~ mOr/iny with re~!)t'ct to the ice, where scme of
the flo., miqht be turb,il" ".' where ,';at" l' depth milY cha~(1e rdP id'.'; a nn the 11ater may

*Nevel, Donald E., ';caring CdPdCity of floatinq ie,' siwecs, lJ:lpuol is hed S RRE~ docu
ment, r)eCe ln~Er 196f..
699

Ph i I Johnson

flow in a curved path? It does not appeHr reasonable and S hOU'~ be ch pc k0d.

The above discussion shows that it would be unusual for? vehicle travelinq
somewhat randomly on an ice shee t to encounter cr i ti c al s ped condition:;. The pro
bability would be increased if the travel were restricted to a preDared ice roarl
but could be eliminated by postina the road with the crit ical speed and advi s ing
travelers to avoid that speed. In addition, it milY He ll be irr,possible to meet the
c ritical s peed conditions on ri ve r ice. All in a ll, it seems questionable \,hethe r
the approach to the problem of movin q lo ads s ho uld be bas e n on thE "critical Sfl,:"d"
effect.

700
Discussion by P. Johnson on "Designing Ice Bridges and Ice Platforms"

Volume 1, Page 32

Author's Reply
by:

L.W . Gold, Division of Building Research, National Research Council,


Ot tawa , Canada

Mr. John son has identified important questions that still remain
to be answered concerning the dependence of the safe performance
of ice covers on vehicle speed. It is difficult to measure the
stress that is induced in the ice by moving l oads . As vehicle
speed increases, however, ice becomes more elastic in its behaviour
and elastic theory should become more appropriate for describing
deflections and stresses. It is very important that measurements
be made in the field of the dependence on vehicle speed of the
maximum strain rate induced in the cover. Observations should be
made at the same time of cracks that are indu c~ d by the load,
probably using sonic techniques. This information, along with
laboratory measurements of the strain rate and temperature
dependence of the elastic modulus, should provide the information
that is required on the dependence of safe performance on
vehicle weight and speed.

701
FIN BOO\" ICE GATE FOR ICE CONTROL

Ai':D WINTER NAVIGATION

Gee Tsan g National Water Resear ch Institute Canada


Researc h ,)cientist Canada Centre for Inland Waters
Burlington, Ontario

ABSTRACT
W~en fins are attac hed to the downst rea, n side of a boom, the boom will be steered into the
current. /\ gClte forrned by twO fin booms is proposed for provision to conventional floating ice
booms or other ice retention barriers for river ice control and for winter navigation. The theory
for desi gning the fin boom ice ga te is presented and design cur ves are shown. Laboratory model
st: !' :ics showed that the fin boom ice gate performed well and is a useful apparatus for river ice
control and winter navigation.

INTRODUCTION

In rn.!i1ag ing a natural river in wi nter, it is general practice to acc elerate the formation of
an ice cover on the river to cut down the heat loss from the river and consequently the
production of fra z il ice and the associa ted operational problems . A common method to
accelerate the formation of the ice cover is to place a floating boom across the river. The boom
arrests the drift ice and initiates the formation of the ice cover.
Whil e conventional floating booms prove to be an effective engineering apparatus, they
pr esen t barrier s to the channel and make winter navigation difficult. In addition, since the
removal and reinstallation of a convent i onal floating boom or a part of it is an invo lved and time
co nsuming process, and at times can be even dangerous, it is not an ideal apparatus for
controlling the retent ion and discharge of the ice that had accumulated in front of the boom. It
wo uld be desirable if a ga te could be provided in a conventional boom which could be opened and
closed easily to fac ilitate winter shipping and ice discharge and retention control.
A finned boom has been de veloped by Tsang and Vanderkooy [3], originally for the
purpose of recovering spilled oil from ice infested rivers. Figure I is a photograph
showing the field te st of the prototype of the boom. This ice-oi l boom consisted of a
rigid perforated boom barrier and a number of fins or rudder s behind the boom. The

f i ns were pivoted to hinges and could rotate about the vertical axes of the hinges. The angle
702
between the boom and the current could be varied by controlling the angle between t he fins and \
the boom with cables . When ice floes and oil slicks drifted to the boom, the ice floes were
deflected to the outer side and the oil slicks flowed through the perforated openings to the
downstrearn side of the boom, to be deflected by the fins to the shore for recovery. The ice-oil
boom, or the fin boom as it now may be called, was so stable that workers could remain standing
on the boom while large ice floes were being deflected to the outside. It is proposed here that
fin booms be used to provide the desired gat"s [or conventional floating booms. A gate can be
formed by placing two plane symmetric fin booms in an opening of a conventional floating
boom. The V formed by the two fin booms can be opened and closed by manipulating the fin
angles.

Figure I Ice-Oil Boom Deve loped by Tsang and Vanderkooy [3]

A conventional floating boom has a stress-relieving safety feature. When the ice force
behind the boom becomes excessive, the floating boom is depressed allowing the ice to spill
over and thus avoids excessive stress. A fin boom also has the built-in Stress relief
mechanism. When the ice for c e behind the gate becomes excessive, the ga te will be jarred open
allowing the ice to pass through and thus avoiding excessive stress. After the ice has passed,
the gate will swing close again. Needless to say, when used to construct a gate, a fin boom does
not have to be perforated.

703
DYNAMIC ANAL YSIS OF FIN BOOM

Figure 2 shows" ga te formed by t wo fin boo l ns. For easy engineering con struction and
hand li ng , the fin boo ln s ar e assumed to be co mposed of iden t ica l un its . The re are n+N units in
eac h of the two f in boolns shown in Fi!;;u re 2. For the first n unit s, the [,ns are removed to avo id
interference with the main floating i ce boo m when the ga te is in operation . For the reln ain ing N
units, the (ins :lre attached to the boom. Fo r each finned unit, the fin is hinged to the
downst ream end to gi ve the Inaximum ruddering effect. The followin!;; dynamic ana l ysis is
applied to one side of the gat e (the right hand si de boo m) on l y.
fL OW

-0
ANCHOR

I
I

Fi gure 2 Ice Gate Forln ed by Two Fin ~ ooms

The wa t er dr ag on the boom barrier produ ces a co unterc l OCkw i se mom ent to open the gate .
This mom en t abo u t the ups tream end of boo rn 1\ ca n be shown to be gi ven by (T sang and
Vanderkooy [ 3 1)

vlb -_[ 2"I C o V2,. \ Lb JI2" (n


I) + N) 2.2
son e

whe re CD is the drag coef[ ,cient of the boo'n barrier, Ab and Lb are the projected i mmersed
area and len !;; th of one boom barrier unit res pectively, p is the density of wat er , V i s the velocity
of the flow and 6 is the angl e between the current and the boom. The abo ve InOmen t may be
2
norrnali zed by being div ided by (CD
0 V ' \ L )/2 to give
b

70 4
( Z)

Counteracting the above moment is the moment pruduced by the dru g s on the [ins. It Cdn
be shown that a fter norm ali z ing t his morn ent i s gi ven by

(3)

where Af is the projected immersed area o[ the fin, a is the fin angle, L is the length of the [in
f
and Lg is the width of the gap betw een the fin and the boom. The ga p is nec essa ry for the
rotation of the fins about the hinging points.
At equilibrium, ~n e ha s

(4)

The substitution of Equations 2 and 3 into the abo ve equation and the subsequent rearr angelnent,
noting that Af/Ab=(Hf/Hb).(L/Lb)' wher e H and Hb are the depth of ilnmer sion of the fin and
f
the boom respectively, lead to

-1
e cot (5)

The above equation gives the relationship between a and e under different parametric conditions.
Tsang and Vanderkooy [3] showed that, within practical ranges, n, i'J and Lg/ Lb are not
important parameters in affecting the a-e relationship. The important paramete rs dre Hf / Hb and

Lf / L b
Figures 3a and 3b are plottings of e versus a under the parametric conditions of n=2, i'J=12
and LiLb=0.2. As noted a bove, uSing other n, Nand Lg / Lb values would not slgnlflcdntly alter
the shapes of the Curves. The curves In Figure 3a were plotted by keeping L / L =0.8 and
f b
progressively increasing the values of H/H from 1.0 to Z.O. On the other hand, the curves in
b
Figure 3b were plotted by keeping H/Hb=l and progressively increasing the value of L/L from
b
0.5 to 1.6.
0
It is seen from Figure 3 that when the fin angle a is 180 , the boom angle e will be zero,
meaning that the gate is now open. As the fins are gradually closed, 6 will be progressively
increased, meaning that the gate is now being closed. The maximum boo m angle is rea c hed when
0
the fin angle reaches appro ximately 110 . After thi s point, further c losing th e fin will bring the
adverse result of reducing the boom angle, or opening the boom. It can be shown that when the
fin angle becomes less than the angle shown by the intersection between the curves and the
horizontal axis shown in Figure 3, the boom will even swing to the other side .

705
90
'"
eo eo
a.
Param@!ef3'
Patamel~
n ,,2 O.2 ~b'
.
N..,2

. ~ .. o.2 ~..Q. B
70 70

""
n"2 N-t2
Lb Lb "ru
~
3 5
L
60
.
eu-1 2 3 5
!
60
L, = 0.5 0.8 to t4 r.6

~ ~-
1 12 l4 1.6 1.8 2.0 50
r;;
C
'" c
~
~ '0

<> '"0
a 30
" 30

8 '"
ID
~ 20

'0
"
80 QOIOO ttOr.!O
F" ArwIe a, deg'eeIS

a. b.
Figure 3 Typical Fin Angle-Boom Angle Relationships

The fact that the maximum boom angle is reached at a fin angle of 110 is not accidental.
An examination of a mul ti tude of e-a curves plotted under different parametric conditions
showed that this is always the case. As it is seen from Figure 3 that e is not sensitive to the
change of CJ. when it is near its maximum value, CJ.=1 10 thus may be used as the fin angle producing
the maximum boom deflection. With CJ.=1100, n=2, N=12 and L /L =0.2, Eq. 5 is changed to
g b

60

>< Hf/Hb-3.0
E50
CD
2.0
'"
g> ,5
<4: 40
1.0
.~

Q; 30
0

'O"J
u::
E 20 Hf -Depth of immersion of fins,
:J
E Hb- boom.
;( "
'"
~
10~---L----~-- __L -__~____~__~L-__~
o 1,0 2,0 3,0
Fin Length to Unit Boom Length Ratio, lfllb

Figure 4 Maximum Deflection of Boom for Different Fin Sizes

706
emax = cot- I 3.041
H '-
f
L;-L;' ----------J-[72 .0.364/ (6)

[ H L (2[ - .3.107)
b b b

According to the above equation, the maximum boom deflection angle 1rnax is plotted against
Lf/Lb at different pararnetric Hf/H values as shown in Figure 4. Once again, it is noted that
b
the curves are also valid for other practical values of n, Nand Lg/L . It is seen from Figure 4
b
that if the fins are sufficiently long and deep, the boorn is capable of swinging into the current at
large angles.
RETENTION OF ICE BY GATE
A gate formed by two freely floating fin booms sw ung at their maximum deflection angles
will have no ice containing power becau se for each boom the gate closing moment produced by
the drags on the fins is cOlnpletely balanced by the gate opening moment produced by the drag on
the boom. A functional gate can only be (armed when the boom angle (hal( the gate apex angle)
is less than 9
max
Let 8 be the booln angle of a (unctional gate, from Equation 2 one sees that because the
g
boom deflection angle is reduced from 9 to A , the gate opening moment produced by the
max g
drag on the boom is reduced by

(n N)2 2 2
---2-- (sin 9 - sin 8 ) (7)
max g

Similarly, from Equation 3, one can see that the incredse in the gate closing moment because the
reduction o( the boom angle (rom 8
max
to ag is given by

Hf
11;'
Lf
Lb N
[
(n.
N I
-2) cos 110
a
-
L
(~
L(
2Ib" )
J

[ 2(8 0.2 0] (3)


sin .110 )-sln (9 . 110)
max g

The sum of IIlvl


b
* and IIM f * gives the normalized Inoment for the boom to contain ice.
The ice cover accumul;Jled in front of the gate is subjec t to the shear of the underflow and
this shear force is transmitted to the boolns. For an unconso lidated ice cover, the shear force
that is acting on the ice cover and is transmitted to the booms increases with the length o f the
ice cover in front of the gate. There is, however, an upper limit to this ice force. In studying
the retention of ice by a floating boom across a river, Perham and Racicot [11 observed tha t,
when the length of the ice cover is three ti mes the width of the river, the maximum force on the
boom is reached. Beyond this point, th e additional ice force resul ting from the lengthening of
the ice cover is taken up by the shores_ Applying the above observation to the ice cover in front
of the gate, one may say that only the part of the shear force that is exerted by the current on
the ice cover accumulated in front of th e gate to d length of three times the gate's width has to

707
be supported by the gate. It can be shown that the opening moment on the gate produced by this
maximum ice force is gi ven by

M. (9)
l

where CO ' is the drag coe fficient of the undersurface of the ice cove r. Normalizing the above
l 2
moment with (CO P V A L )/2, one obtains
b b

C OiLb 3 3 3 I
M.
l*
<;; FIb (n + N) sin eg ( 2" + J co t eg) (10)

when M. is less than (I'l Mb + I'l M ), the ga te will remain closed.


1* * f*
From

(II)

one obtains an equation wi th the following parame ters:

(J2)

The relationship between e ' H/H and Lf/Lb is approximately given by Figure 4. If, for
max b
simplicity, some reasonably practical values are assigned to n, Lg/Lb and Lb/H , then the above
b
e qua t i on is reduced to

(13)

The drag coefficient of a rectangul ar submerged body is about 2.0 (Streeter [2J). For the
boom barrier of a fin boom , because it is partially immersed in water only, the drag coe fficient
should be different. The drag coefficient shou ld also depend on the actual design of the boom.
0
For the ice/oil boom developed by Tsang and Vanderkooy [3] with 45 openings for the oil slicks
to pass through, laboratory experiments showed that the drag coefficient was greatly affected by
0
the fin angle (Tsang, 1980, unpublished resul ts). When the fin angle is 110 , the drag coefficien t,
however, is approximately 2.0. In the absence of more reliable da ta, this drag coefficient may be
used in the present study.
In the analysis shown thus far, the drag coefficien t of the fins has been assumed to be the
same as that of the boom . In the experiments of the ice/oil boom mentioned above, however,
this was found not to be the case. Because of the shedding of eddies by the upstream boom
barrier, the drag coe fficient of the fins was higher than that of the boom. The fin drag
coefficient to the boom drag coefficient ratio was found to be dependent of the fin angle. For a
0
fin angle of 110 , this ratio was about 1.75. For conse rvative reasons, however, the fin drag

708
coefficient is assumed to be the same as the drag coefficient of the boom for the remainder of
this study.
The drag coefficient of an ice cover can be obtained from study ing the shear on the ice
cover. According to Perham and Racicot [I J, there are three periods of ice forces against an
ice boom. The first period is when the ice cover is still unconsolida ted following the onset of the
winter. In this period, the shear on the ice cover is transmitted directly onto the boom. As
mentioned earlier, the ice force on the boom increases with the length of the ice cover until the
ice cover length reaches three times the river's width. Thereafter, the addi tional ice for ce is
taken up by the banks.
(n the second period, the ice cover consolidates and is also frozen to the banks. Much of
the ice force, therefore, is transmi tted to the banks and the ice boom experiences a large
reduction of the ice forces.
The third period is the time after spring breakup. In this period, the ice cover thaws away
from the banks and the ice force is once again transmitted to the ice boom. Because the breakup
ice cover usually is quite rough, many times, ice jams may also develop; this means a greater ice
force on the boom than during period l.

1.5 2.0 3 .0
3.0

.0

...J

::;
0
co
c: 2.0
.r::

0,

c
'"
...J
E
0
0
III

~
.r:: H f - Depth of immersion of fins .
rnc Hb - .. boom
'"
...J
C
u..

Half Apex Angle of Gate 9 g. Degrees .

Figure 5 Fin Size for Fin Boom Gate of Desired Apex I".ngle

709
Perham and Racicot [I] measured the shear stress on the ice cover for two early winter
(period I) occasions and two early spring (period 3) occasions. From these shear stresses and the
accompanied Froude number va lue s, the skin drag coefficient of the ice cover C can be
Di
calculated to be 0.052 and 0.075 for the two period 1 occasions and 0.081 and 0.109 for the two
period 3 occasions. Using the value of CDi=O.IIO for conservative reasons, one obtains a CD/CD
ratio of 0.055. With such a constant drag coefficient ratio, Equation 13 is reduced to

([4 )

If suitable values of N are assigned, the above equation gives the relationship between L/L and
b
9 under different para,netric Hf/Hb conditions .
g
Equation 14 can be made explicit, and, with the help of Figure 4, be solved numerically.
The Lf/Lb versus ag curves in Figure 5 were generated by assigning N=12, n=2, L /L b =0.2 and'
g
L /H =15. Other engineeringly practical parametric values may also be used to generate
b b

addi t iona l curves for engi neer i ng design of fin boom gates.

LABORA TOR Y EXPERIMENTS

Laboratory experiments were conducted to test the proposed {in boom ice gate. T he
experiments were carried Ollt in a two-metre wide flume. The parameters of the two model fin
booms forming the gate were iiS follows :

TABLE PARAMETERS OF MODEL FIN BOOMS

Paralneters n L /Lb
___ ._.- __ ___ ~_. _._. ___ ' __._. __..;\J
____ .___ .._-_.__ ._.1:'. ._. ____ .__.- .._Hf/Hb
.__... _ . , .L/ '-b
_ .___._ Lb/Hb
_.___
2 12 0.2 3.0 2.5 15

The model booms and gate were intended to be 1/30 model of a prototype. Following were
the dimensions of the model booms and the geometric and hydraulic parameters of the model
gate. The co rrespond ing dim ensions and parameters of the pr ototype booms and ga te are also
shown in the table.

TABLE 2 DIMENSIONS AND PARAMETERS OF MODEL AND


PROTOTYPE BOOMS AND ICE GATE

Fr W
g
~--""''''''-'-'-~'~'-''----'-'- ' -

Model 1 cm
--- ---_.. -.---"--.-.......---..,..-.-----..--.-_._-
15 cm 4 cm 0.08
--
25 cm
- ,- , .,. .- --.--~--~ ......,...--..~.,..-

20
0