Search

The R.V. POLARSTERN carried out ice breaking trials in which ice forces were measured with special panels at two locations in the bow. Time series records of the ice forces for a 13-minute period were compared for differences in terms of magnitude and duration of loading for the two locations. Maximum force of 2800 kN was measured on a 1 m² panel. The nature of ice forces measured on the panels were very different in terms of frequency and duration compared to those measured with strain-gauged frames on the R.V. Polarstern and the CCGS Louis S. St-Laurent., 33rd International Ocean and Polar Engineering Conference (ISOPE 2023), June 19-23, Ottawa, Ontario, Canada, Series: International Ocean and Polar Engineering Conference

The wharf at Nanisivik was constructed over the winter 1974-75. It was the first deep-water wharf in the Canadian Arctic and there were questions on ice-structure interaction processes, ice loading and performance. This provided the impetus for a series of investigations at the site extending up to the present time. The wharf is located on Strathcona Sound, which is characterized by relatively land fast ice subject to vertical motions under tidal action. The nearby community and mine provided access and logistics which were very amenable for extensive on-site investigation. Early investigations extending over several winters showed the build-up of a zone of ice between the vertical wharf face and the adjacent level ice sheet. This ice behavior extends to the present. Various schemes of ice force measurement, directly on the wharf face and in the adjacent ice cover have been conducted and reported on at various conferences. Because of the mine, which the wharf was built to serve, extended season marine shipping occurred and also played a part in the ice investigations. This paper will provide an overview of ice investigations conducted at and around the Nanisivik wharf over an almost 50-year period and a summary of what has been learned., 26th IAHR International Symposium on Ice, June 19-23, 2022, Montréal, Québec, Canada

A project was carried out over the winter of 1985/86 to measure ice forces on the Nanisivik wharf and associated environmental factors. The ice regime was land-fast first year sea ice which attained a maximum thickness of 1.6 m in May. Five panel-type transducers 1-m wide and divided into 0.5-m vertical sections were frozen into the ice cover 40-m out from the vertical wharf face. Measured ice pressures in the panels indicated that the majority of the ice pressure was in the top 0.5 m of the ice cover and all panels responded in a similar/synchronous manner to temperature and tidal inputs. Major loading events were systematic associated with a large or rapid temperature increase. Superimposed on these events were ice pressure oscillations having a frequency twice the tidal frequency. Maximum measured ice pressure on one panel section was 450 kPa. Averaged ice pressure measurements reached a maximum of 155 kN/m. A conceptual model relating ice pressure to tide is proposed to guide interpretation., 26th International Conference on Port and Ocean Engineering under Arctic Conditions, June 15-18, 2021, Moscow, Russia

Using the ISO19906 local pressure-area relationship p = 7.4A⁻⁰.⁷ as a process pressure-area curve to validate numerical ice models and reconstruct force-displacement development in damage assessment studies may lead to overestimation of the energy absorption capacity of ice found in nature. This is because the way this relationship is derived is different from how it is used in structural assessments of ice impacts. In the following paragraphs, we explain (with examples) how to reconstruct realistic ice force-displacement relationships in the absence of empirical force histories while focusing on freshwater, polycrystalline ice crushing, within the brittle regime, against a rigid structure.

Indentation testing at medium scale has been carried out on four occasions, between 1984 and 1990, in the Canadian Arctic. Iceberg ice and multi-year ice were tested. Forces up to 16 MN were exerted on contact areas up to 3 m2 at rates up to 0.4 m/s. All testing was done with the same servo-hydraulic controlled system. Results of the test programs have been previously reported individually. Here the results have been compared in terms of force versus penetration, global pressure versus nominal contact area and volumetric specific crushing energy versus crushed volume of ice. Various geometries of indenter and ice face were used but the volumetric specific crushing energy was relatively independent of geometry of ice or the indenter for the cases examined. Specific energy is lower for lower penetration rates. There was a strong numerical similarity between volumetric specific energy (MJ/m3) and pressure (MPa) for the same test, whether plotted against crushed ice volume or nominal contact area. This paper explores whether volumetric specific crushing energy can provide helpful insights to defining global ice pressures on relatively small global areas, those less than 3 m2. While this linkage may provide an additional data source for global pressure-area relations needed for structural design, further understandings are needed before their application., 25th IAHR International Symposium on Ice, November 23-25, 2020, Trondheim, Virtual Conference

Syntactic parsing using dependency structures has become a standard technique in natural language processing with many different parsing models, in particular data-driven models that can be trained on syntactically annotated corpora. In this paper, we tackle transition-based dependency parsing using a Perceptron Learner. Our proposed model, which adds more relevant features to the Perceptron Learner, outperforms a baseline arc-standard parser. We beat the UAS of the MALT and LSTM parsers. We also give possible ways to address parsing of non-projective trees., This was part of an assignment at my graduate course at LTI. This does not offer any major novelties

Many engineering problems require an estimate of ice thickness, either the maximum likely thickness or the thickness at some time during the winter. This can be for estimating ice forces on an offshore structure in the sea or a bridge pier in a river. Operation of icebreaking ships requires knowledge of ice thickness to establish the viability of transit to northern ports. Similarly, over-ice transportation on seas, lakes or rives depends on a knowledge of ice thickness. At some locations historical records can be used to estimate ice thicknesses, but with changing climate historical records have limited applicability. Having means for predicting ice growth during a winter is a helpful tool. In high Arctic regions it is assumed the primary factors affecting ice growth are air temperature and snow depth. Assessment of the equations against data from Arctic weather stations indicates that the incorporation of snow depth in terms of a mean annual snow depth is a simple means for improving their predictive capability. Modified equations including mean annual snow depth and freezing degree days as input parameters are proposed and tested against available data at two locations in the Canadian High Arctic; Resolute Bay in a marine coastal environment and Baker lake in an inland freshwater lake. These modified prediction equations are proposed for general application in the Arctic., 24th IAHR International Symposium on Ice, June 4-9, 2018, Vladivostok, Russia

The 25th International Conference on Port and Ocean Engineering under Arctic Conditions, June 9-13, 2019, Delft, The Netherlands

The story of offshore Arctic engineering. Dan Masterson. Cambridge Scholars Publishing; Unabridged edition (112015). (Hard cover), 195 pp. CDN$204.07. ISBN 978-1527518162. Dan Masterson enjoyed a 45-year career in which he was engaged in ice engineering aspects of a number of the major offshore petroleum exploration programs in the Canadian High Arctic Islands and the Canadian and Alaskan Beaufort Sea. This generously illustrated book recounts Dan’s own personal experiences in many of these projects. He writes well and his narration captures the spirit, enthusiasm, and optimism of the times. The book contains high level background information on the political and economic factors that set the scene or provide context for some of the projects he was involved in. This background information is factual and well researched, and clearly separated from his personal observations and opinions. The various problems and technical solutions described in the book are interspersed with some of the colourful experiences and characters he encountered.

Node embeddings have become an ubiquitous technique for representing graph data in a low dimensional space. Graph autoencoders, as one of the widely adapted deep models, have been proposed to learn graph embeddings in an unsupervised way by minimizing the reconstruction error for the graph data. However, its reconstruction loss ignores the distribution of the latent representation, and thus leading to inferior embeddings. To mitigate this problem, we propose a random walk based method to regularize the representations learnt by the encoder. We show that the proposed novel enhancement beats the existing state-of-the-art models by a large margin (upto 7.5\%) for node clustering task, and achieves state-of-the-art accuracy on the link prediction task for three standard datasets, cora, citeseer and pubmed. Code available at https://github.com/MysteryVaibhav/DW-GAE., Comment: 6 pages, Empirical paper on improving Graph Embeddings using Random Walk

International Conference on Performance of Ships and Structures in Ice (ICETECH), 28-31 July, 2014, Banff, Alberta

The lighthouse Norströmsgrund in the northern Baltic Sea was well instrumented to measure ice forces and ice thickness in the years 1999 to 2003. Load panel and tilt meter data have been analysed to assess global force on the lighthouse during several ice ridge or hummock field interaction events in 2002 and 2003. The size of the ridges, type of interaction, and level ice load are examined, and an attempt is made to assess keel load. The methods of the ISO 19906 Standard are applied to determine the range of ridge forces that may be predicted at the site., 11th International Conference and Exhibition on Performance of Ships and Structures in Ice (ICETECH), 28-31 July, 2014, Banff, Alberta

Three ridged ice loading events have been analyzed from the Norströmsgrund lighthouse, off the Swedish coast. Ice load data from 1999 and 2000 were collected as part of the LOLEIF project. The three events described in this paper occurred in March and April 2000. The ridges were relatively small, with maximum keels of less than 5 m depth and sails of less than 0.5 m height., International Conference on Performance of Ships and Structures in Ice (ICETECH), 28-31 July, 2014, Banff, Alberta

Although many studies have been carried out using field, laboratory and analytical methods, uncertainty remains in the magnitude of forces generated by ice ridges on offshore structures. Norstro msgrund lighthouse, located in the Bay of Bothnia in the Baltic Sea, is a cylindrical concrete structure with a sand filled caisson foundation and a waterline diameter of about 7.5 m. It was instrumented to measure ice forces from 1999-2000 for the LOLEIF project, and from 2001-2003 for the STRICE project. Instrumentation included load panels around one half of the structure for direct measurement of ice force, a tiltmeter for assessment of lighthouse tilt motion, and devices to record ice thickness at the site. Load panel and tilt meter data have been analysed to assess global force on the lighthouse during several first-year ice ridge or hummock field interaction events in late winter and early spring of the years 2000, 2002 and 2003. Ridges were encountered with maximum consolidated layer thicknesses of 1.1 to 1.5 m, and unconsolidated keels potentially of 8 m or greater draft. Global loads are calculated from the forces recorded by load panels, and are also interpreted from tilt data. The vertical and horizontal distribution of ice pressures is examined. Most previous publications have focused on the load panels, whose measurement area is limited by their height of 1.6 m. In the present analysis, an attempt was made to assess whether any keel load was measured by the tiltmeter. The forces generated by ridged ice are also compared to those from the surrounding level ice. The methods of the ISO 19906 Standard are applied to predict ice force using the geometry and assumed properties of ridges at Norstro msgrund. These force predictions are compared to the actual measurements, and reasons for discrepancies discussed. The variability in force estimates is examined with reference to the range of possible values for the assumed parameters, including thickness and strength of the consolidated layer, and porosity and cohesion of the keel rubble., International Conference on Cold Climate Technology, 26-28 May 2014, Narvik, Norway

Ice of many forms grows annually on rivers, lakes and the sea, and as a consequence can interact with hydrotechnical or coastal structures, generating forces on them. These forces must be appropriately assessed and accounted for in the design of structures in water bodies affected by ice. Ice conditions such as thickness, strength, morphology and floe size influence ice forces and must be characterized and suitable extreme values of these features determined. The environmental driving forces of wind, current and temperature must also be understood. Information on ice conditions and environmental factors are combined to provide guidance on calculating the magnitude of ice forces that a structure might be expected to withstand over its lifetime. Various national and international standards and codes that provide guidance on ice forces have been introduced and a high level overview of determining ice loads provided. Areas that would benefit from further research include; investigation of compressive strength of ice for rates greater than 10-3 s-1, the relation between rate effects in laboratory tests and rate in ice loading of structures, and analytical work to implement ice constitutive relations into the nonuniform stress states of ice indentation., 22nd IAHR International Symposium on Ice, August 11 to 15, 2014, Singapore

Fast and effective automated indexing is critical for search and personalized services. Key phrases that consist of one or more words and represent the main concepts of the document are often used for the purpose of indexing. In this paper, we investigate the use of additional semantic features and pre-processing steps to improve automatic key phrase extraction. These features include the use of signal words and freebase categories. Some of these features lead to significant improvements in the accuracy of the results. We also experimented with 2 forms of document pre-processing that we call light filtering and co-reference normalization. Light filtering removes sentences from the document, which are judged peripheral to its main content. Co-reference normalization unifies several written forms of the same named entity into a unique form. We also needed a "Gold Standard" - a set of labeled documents for training and evaluation. While the subjective nature of key phrase selection precludes a true "Gold Standard", we used Amazon's Mechanical Turk service to obtain a useful approximation. Our data indicates that the biggest improvements in performance were due to shallow semantic features, news categories, and rhetorical signals (nDCG 78.47% vs. 68.93%). The inclusion of deeper semantic features such as Freebase sub-categories was not beneficial by itself, but in combination with pre-processing, did cause slight improvements in the nDCG scores., In 8th International Conference on Language Resources and Evaluation (LREC 2012)

Ice interactions with the Confederation Bridge for the winters 2008, 2009 and 2010 have been studied. Ice conditions for these winters were characterized by two average and one very mild. Tilt data from piers P23 and P24 in response to ice and wind actions were recorded by a data acquisition system. Video records of ice interactions with pier P23 and overall ice conditions adjacent to piers P23 and P24 allowed ice conditions and interaction processes to be related to tilt measurements. Wind data were analysed to correct for the tilt caused by wind actions on the bridge. Limit stress and limit force events were identified. An ice loading scenario that transitioned from limit force to limit stress resulted in the highest ice forces measured over the three years of monitoring, with forces of 3.1 MN and 4.3 MN at P23 and P24, respectively. This loading event corresponded to a likwly adfreeze condition., 21st International Offshore and Polar Engineering Conference, ISOPE-2011, June 19-24, 2011, Maui, Hawaii

Bureau Veritas has an interest in expanding the Guidelines for ‘Ice Reinforcement Selection in Different World Navigation Areas’ to include a chapter on Greenland Waters. This report contains the chapter for the guidelines and includes: 1) typical ice conditions, 2) selection of appropriate ice class, 3) legislation to be followed and 4) additional interesting meteorological parameters. A short supplement is included that provides further detail on source data and methods.

Ice force predictions using the ISO 19906 Arctic offshore structures standard and three national standards have been compared for structures offshore Sakhalin. A multi-leg structure with leg diameters 18 and 24 m subjected to several design ice conditions, level ice 1.2 m, rafted ice 3.5 m and a first year ridge keel 21 m deep was evaluated. ISO 19906 is the most complete and transparent standard in providing guidance. It incorporates guidance on ice strength from the Russian standard, and ice thickness and geometry effects from the Canadian standard. Further guidance is still needed for multi-leg structures., The Tenth (2012) ISOPE Pacific/Asia Offshore Mechanics Symposium, October 3-5, 2012, Vladivostok, Russia, Series: Proceedings of the Pacific/Asia Offshore Mechanics Symposiums(ISOPE PACOMS)

The CCGS Terry Fox field trials produced impacts with growlers, bergy bits and a small iceberg over a range of velocities. Ice mass ranged from 30 t to 22,000 t and velocity from 0.1 m/s to 6.6 m/s. Results have been analysed looking at the effect of speed and ice mass on the measured impact force. There was considerable scatter in the results until the rate of the rise time of the measured impact force was taken into consideration. A trend of increasing impact force with loading rate was observed. To further examine this aspect of rate effect on impact forces, medium scale field indentation tests were also studied, since they were done to simulate full scale impact loading. The tests were done at a number of locations in the Arctic on iceberg ice and multi-year ice, and also showed a rate effect. For the CCGS Terry Fox the maximum force related to the loading rate with the following expression where F is force in MN and is loading rate in MN/s. The expression is supported by data up to a maximum force of 6 MN. Higher forces at comparable loading rates were achieved in the field indentation tests due to higher loading system capacity and better ice-indentor contact.

Ice interactions with the Confederation Bridge for the winters 2008, 2009 and 2010 have been studied. Tilt data from piers P23 and P24 in response to ice and wind actions were recorded by a data acquisition system. Video records of ice interactions with pier P23 and overall ice conditions adjacent to piers P23 and P24 allowed ice conditions and interaction processes to be related to tilt measurements. Wind data were also analysed to correct for the tilt caused by wind actions on the bridge. An ice loading scenario of slow build up of ice force on a pier due to the action of current and wind on a large ice floe was identified. In one instance this resulted in a maximum force of about 4 MN being generated on simultaneously on each of piers P23 and P24.

A Joint Industry Project has been carried out to develop a "Catalogue of Local Ice Pressures". The focus of the project was to determine average local ice pressures on various sizes and shapes of design areas, and to present the data in a form to facilitate probabilistic estimation of local ice pressures. The data are organized in a database to assist extraction of local ice pressures as a function of operational factors and ice conditions. The database includes four digital data sets on ship trials (1994 Louis S. St. Laurent, 1991 and 1996 Oden, and 2001 Terry Fox), two offshore structures (1985-86 Molikpaq caisson and 1990 JZ-20-2-1 jacket structure) and one field test (1995 Grappling Island). An example of the method used to process data from bow measurements during the 1994 Polar transit of the Louis St. St. Laurent is presented.

A probabilistic analysis is made to predict the seasonal evolution of first-year ice loads on the Molikpaq offshore structure throughout a winter in the Canadian Beaufort Sea. The analysis is based solely on the failure modes of the ice, the ice thickness, and the number of weekly ice loading events. The paper illustrates the use of the probabilistic approach using these basic input values. The results of the analysis show good agreement with the measured ice loads on the Molikpaq in the Beaufort Sea.

A large piece of shore fast ice approximately 5 by 10 km broke free and drifted through Northumberland Strait on April 4 2003, failing against the piers of the Confederation Bridge. The ice had been surveyed shortly before its break-up and its thickness was about 1 m with many ridges of about 4 m thick when averaged over a 20-m diameter. While the ice was quite thick, it was late in the season and thus soft resulting in the maximum force determined on the two piers only in the order of 2 MN. Hand held videos of the episode provided an overview of the ice failure processes observed.

Information on the morphology and properties of first-year ice ridges in the Sea of Okhotsk is required for design of offshore structures. Data have been compiled to provide a picture of Okhotsk sea ice ridges. The data include sail heights, keel depths, consolidated layer thickness, porosity and ridge width. The data from the Sea of Okhotsk are compared with data for ridges from the Beaufort Sea. Okhotsk ice ridges were found to have similar proportions to Beaufort ridges, i.e. ratio of keel depth to sail height is on average 5:1, average sail porosity 35%, and average keel porosity 20%. Sakhalin ridges are smaller than Beaufort ridges, however they are proportionately 25 % wider than Beaufort ridges.

Producing machine translation (MT) for the many minority languages in the world is a serious challenge. Minority languages typically have few resources for building MT systems. For many minor languages there is little machine readable text, few knowledgeable linguists, and little money available for MT development. For these reasons, our research programs on minority language MT have focused on leveraging to the maximum extent two resources that are available for minority languages: linguistic structure and bilingual informants. All natural languages contain linguistic structure. And although the details of that linguistic structure vary from language to language, language universals such as context-free syntactic structure and the paradigmatic structure of inflectional morphology, allow us to learn the specific details of a minority language. Similarly, most minority languages possess speakers who are bilingual with the major language of the area. This paper discusses our efforts to utilize linguistic structure and the translation information that bilingual informants can provide in three sub-areas of our rapid development MT program: morphology induction, syntactic transfer rule learning, and refinement of imperfect learned rules

Producing machine translation (MT) for the many minority languages in the world is a serious challenge. Minority languages typically have few resources for building MT systems. For many minor languages there is little machine readable text, few knowledgeable linguists, and little money available for mT development. For these reasons, our research programs on minority language MT have focused on leveraging to the maximum extent two resources that are available for minority languages: linguistic structure and bilingual informants. All natural languages contain linguistic structure. And although the details of that linguistic structure vary from language to language, language universals such as context-free syntactic structure and the paradigmatic structure of inflectional morphology, allow us to learn the specific details of a minority language. Similarly, most minority languages possess speakers who are bilingual with the major language of the area. This paper discusses our efforts to utilize linguistic structure and the translation information that bilingual informants can provide in three sub-areas of our rapid development MT program: morphology induction, syntactic transfer rule learning, and refinement of imperfect learned rules.

The Confederation Bridge is a unique structure located in an area with temperate but very dynamic ice conditions. An extensive monitoring and research program including ice force measurements was carried out under the auspices of an NSERC Collaborative R&D Grant. In addition, the National Research Council carried out a complimentary program to measure the response to pack ice of two bridge piers adjacent to the navigation span. There were two objectives of the work. One was on the effect the bridge might have on ice conditions and winter navigation, and the other was on the ice loading on the two piers. The largest force measured over 8 winters was 2.4 MN. Interpreted pack ice pressure was estimated to be in the range of 1 kN/m to 10 kN/m., August 19-22, 2007

This paper mainly focuses on establishing a probabilistic model to explore the relationship between local pressures and the global pressure. Both spatial and temporal correlations among the local pressures are considered in the study. A vector autoregressive model (VAR) is applied in this research to include these correlations. The global contact area is divided into a group of local areas. The model is used to illustrate the probabilistic averaging effect from local pressure to global pressure. The duration effect in which a higher load is likely to result from a longer interaction process is also showed using this model. The model is also used to explore the pressure-area relationship and pressure-aspect ratio relationship. The results imply that using the pressure-area relationship and the pressure-aspect ratio relationship alone may not be sufficient for global load estimation., 2006

The Canadian Standards Association (CSA) developed a comprehensive Offshore Structures Standard in the early 1990s. The Code has had limited use in Canada, but S471 General Requirements, Design Criteria, the Environment, and Loads has seen international application. The Code is performance-based; setting overall targets for reliability and provides specific guidance on achieving these targets. The provisions in the CSA Standard for ice loads have been reconciled with the Russian codes such as SNiP and VSN. A comparison of the codes has shown that they predict similar ice loads., 2006

Compressive ice failure is an important aspect in the design of offshore structures in ice environments. The authors concentrate attention on the crushing failure mode. The scale effect is a phenomenon whereby the average pressure decreases with contact area. In classical elasticity, viscoelasticity and plasticity, there is no scale effect. In the case of ice compressive failure, there is ample evidence of a scale effect. Two factors are dominant in this situation. Ice is prone to fracture, thus reducing the volume of material to be crushed, and aiding in the formation of high pressure zones. The second factor in the scale effect is the averaging process that results from the transition from local pressures on small areas (dominated by high pressure zones) to global pressures, where this effect is averaged. The analysis of probabilistic averaging is described with the example of pressures measured on the Molikpaq structure. The implications for design are significant., The 18th International Symposium on Ice, IAHR'06, August 28-September 1, 2006, Sapporo, Japan

During the 1980s extensive offshore geophysical surveys yielded over 100,000 measured ice thicknesses in one area of the Arctic Archipelago. The date and location of these measurements are known and can be compared with ice regime information on Canadian Ice Service ice charts from the same time period. Air temperature data are available from Arctic Weather Stations at Mould Bay and Isachsen, from which ice strength can be calculated. It has been demonstrated that ice cover strength decreases by a factor of 10 from mid-winter to just before break-up. A methodology has been proposed for using all the aforementioned data to make adjustments to the strength factor p* in the CIS Ice Forecast Model for application in aiding ice forecasting in the Canadian Arctic Archilelago.

In 1991 the ODEN made a voyage to the North Pole. An area in the starboard bow of the ship was instrumented with strain gauges to measure local ice pressures. The ship encountered various ice conditions in its voyage, and whenever a threshold strain was exceeded, a 6 to 12 second-long “event” record of strains was made. Based on a finite element analysis of the ship structure the strains were converted to average ice pressures on a 40-sub-panel area, 10 frames wide (8.5 m) by 4 panels high (3.52 m). The bow is a “landing-craft-form”, sloped 22o from the horizontal. There were 784 events and for each of them average pressures on various area and shapes were determined and related to ship speed and ice conditions. A means of predicting probable ice pressures as a function of a “design” area and exposure is presented., 2005

The STRICE project installed load panels on half the circumference of the Norströmsgrund lighthouse in the Gulf of Bothnia to make direct measurements of ice forces. Near continuous measurements and observations of ice conditions were also made. As an adjunct to the project, a high-resolution tiltmeter capable of simultaneously measuring tilt in two orthogonal directions was installed inside the lighthouse for the 2002 winter season in order to assess an alternative means of measuring global ice force. For quasi-static ice loading, the tiltmeter provided a good representation of ice force acting on the lighthouse regardless of the direction from which the forces acted. In the case of dynamic ice loading with frequency components up to 3 Hz, a 7-second moving average "filter" provided a good representation of force components up to 0.5 Hz., 2005

The Amauligak I-65 deployment of the Molikpaq for the winter of 1985/86 provides a good source of local pressure data and a basis for estimating global loads on a wide structure. Data including Medof panel pressures and hourly ice conditions at Molikpaq have been extracted from the “Catalogue of Local Ice Pressures” (CLIP) database. Pressures as high as 10 MPa were experienced on small areas; on areas greater than 3 m2, peak pressures are less than 2 MPa. A method is presented for the calculation of peak loads on the north and east faces, using local ice loads to obtain estimates of global face loads. Extreme values of global ice loads were of the order of 200 MN, with most ice forces during the season being much lower. Thicker ice generally corresponds to higher ice loads on both the east and north faces. However for the thickest ice, lower than expected forces were calculated for the faces of the structure. This could be due to less ice drift (and lower exposure) for very thick ice., 2005