Permafrost loss Impact on Dam Safety at Snare Hydro
Perry Mitchelmore, Meco Dartmouth, NS, Canada
Alex Campbell, Meco, Dartmouth, NS, Canada
Gamini Hettiarachchige, Northwest Territories Power Corporation, Yellowknife, NT, Canada
Bradley Harrison, Northwest Territories Power Corporation, Yellowknife, NT, Canada
Permafrost loss continues in northern areas of Canada, both in areas of continuous and semi-continuous permafrost. Climate change predictions of temperature rise are particularly sensitive in northern environments where increases in water temperature will drive permafrost deeper into the ground. Past practices of building dams on permafrost have resulting in vertical displacement of structures and loss of freeboard, and potentially failure.
The Northwest Territories Power Corporation operate several low head dams at the Snare Hydroelectric System that were constructed on permafrost environments in the 1970’s. NTPC recently completed a PFMA of all dams at Share Hydro that identified permafrost loss resulting in dam failure as having a high potential for occurrence. These structures experience vertical displacement and rePermafrost and Damsquire topping up to achieve freeboard. The objective of the current paper will be to further explore the failure modes through thermal models to estimate the rate of permafrost loss and potential for increasing the risk of a failure mode, either through overtopping or through internal erosion.
La diminution du pergélisol se poursuit dans les territoires Nordiques du Canada, tant dans les zones de pergélisol continu que dans les zones de pergélisol discontinu. Les prédictions d’augmentation de la température entraînée par les changements climatiques ont des impacts importants pour les climats Nordiques où le réchauffement de l’eau provoque la fonte du pergélisol en surface. Les barrages antérieurement construits sur le pergélisol ont subis des déplacements verticaux, entraînant une diminution de la revanche de sécurité et augmentant les risques de brèche.
Northwest Territories Power Corporation (NTPC) gère plusieurs petits barrages hydroélectriques situés sur la Rivière Snare. Ces barrages ont été construits sur le pergélisol dans les années 1970. L’étude des scénarios de brèches récemment complétée par NTPC a défini que les probabilités de brèches associées à la fonte du pergélisol sont importantes. En effet, ces structures subissent présentement des déplacements verticaux et l’élévation de leurs crêtes doit maintenant être augmentée afin d’assurer la revanche minimale de sécurité. Cette étude vise à explorer différents scénarios de brèches à l’aide d’un modèle de température. Ce modèle permet d’analyser la vitesse de fonte du pergélisol et l’augmentation potentielle des risques de brèche, soit par débordement, soit par érosion interne des barrages.
To create and maintain safe dams in cryolitic environments, professional engineers need to apply best practices and technologies today with consideration of the future, and design monitoring programs for performance. Canada is a vast country with much of our terrain underlain by continuous, or discontinuous permafrost, as illustrated in Figure 1. Themodern definition of permafrost is ground (soil or rock) that remains below 0 °C for two or more consecutive years.Discontinuous permafrost is subdivided into a more northerly, extensive discontinuous permafrost zone, where permafrost underlies more than 50% of the land area, and the southerly sporadic discontinuous permafrost zone, where 10 to 50% of the land area is underlain by permafrost (CSA 4011).
The potential environmental consequences of climate change are having an impact on all areas, but the impacts are particularly acute on Canada’s northern infrastructure. Northern communities are addressing the challenges by designing more resilient infrastructure, adaptable and adjustable foundation systems, and flexible systems. Dams are somewhat unique in that they are not typically designed to move, and once constructed, are not easily adapted. As well, there are many structures that were constructed before climate change awareness necessitated the need for resilient and flexible designs.
Figure 1: Ground Thermal Regime in Canada (GSC, 2010) pp 19 CSA 4011-10
2 DAMS ON PERMAFROST – GUIDELINES AND STANDARDS
Infrastructure in northern environments will typically be designed to minimize disturbance of the permafrost. In buildings, this is accomplished with thermal separation, cooling and berms. If separation is not possible, the structure will be founded on pile foundations. It is not practical to build a dam on piles, and separation is not always possible. The Canadian Dam Association and the International Commission of Large Dams offer some guidance on how to manage dams on permafrost. The Canadian Standards Association also provides guidance on assessment of infrastructure in northern environments, but does not specifically address dams. Together, these associations provide the known guidance on dams (and infrastructure) in northern environments.
2.1 Canadian Dam Association (CDA)
The CDA Dam Safety Guidelines discuss dams on permafrost in the “Technical Bulletin: Geotechnical Considerations for Dam Safety”. Two types of dams are presented; frozen dams and unfrozen dams. Frozen dams are required to be designed to stay frozen, and should be okay as long as that condition prevails. The vast majority of dams will be unfrozen, and for those the CDA requires the dam be designed with the assumption that the permafrost conditions will thaw due to the presence of water in the reservoir.
The CDA requires unfrozen dams on permafrost be constructed of self-healing granular materials. In this way, deformation can be accommodated by expanding the fill zones upwards. There is a recognition of the potential problems with excess pore water pressures within the thawing zone, but little guidance on how to account for the effect in stability analysis.
2.2 International Commission on Large Dams (ICOLD)
ICOLD has published two bulletins on the topic; Bulletin 105, “Dams and Related Structures in Cold Climate” (1996) and Bulletin 133, “Embankment Dams on Permafrost” (2007).
Bulletin 105 provides guidance on designing for ice, gate systems and a short section on ice and permafrost. The Bulletin is scientifically robust and identifies the potential problem of thawing of ice or permafrost below the dam foundation as causing settlement and increased seepage at dams. There is further discussion on dams in frozen environments and types of dams, but no substantive discussion on how to manage the risk of loss of foundation strength through thawing.
Bulletin 133 focuses on hydropower dams in cryolitic regions, primarily Russia but draws upon experience of other countries as well. The Bulletin discusses features of dam in permafrost areas, trends in design and foundation treatment. As with the CDA, two types of embankment dams identified; “frozen” dams rely on the frozen condition to achieve their imperviousness while “thawed” dams have an impervious area with a positive temperature, at least at times, and allows seepage within design limits. The Bulletin provides more in-depth discussion on the complicated hydrodynamic heat transfer conditions; freeze/thaw/re-freeze conditions that when interfacing with a warm reservoir introduce dynamic stresses on a dam system.
The Bulletin does not specifically discuss climate change, instead provides a high level discussion on environmental impacts of dam construction and operation. It correctly identifies the reservoir as the main source of heat that results in thawing. It also reiterates that dams in a thawing environment will undergo similar failure patterns to those in non-cryolitic zones, although the failure patterns in cryolitic zones are faster and regular. With respect to failure modes, seepage is noted as the most common reason for damage and failure, followed by settlement (23%), crack formation (15%) and stability (7%).
2.3 Canadian Standards Association (CSA 4011)
CSA 4011,“Infrastructure in Permafrost: A Guideline for Climate Change” provides guidance and practical advice on design of infrastructure in northern environments. The document describes the nature of permafrost, trends in climate change, foundation systems for community systems, and presents a process for ensuring climate change is incorporated into decisions. The focus is on northern infrastructure, primarily buildings and civil structure, but not necessarily dams. The principal failure mode for community infrastructure is settlement and loss of strength, resulting in shear failure and sliding. These failure modes are also common to dams, as are the proposed foundation systems.
CSA 4011 provides a two-stage screening process for incorporating climate change into planning for community infrastructure. The screening tool uses a risk based approach to determine the extent of site investigation and design services required for a project or site. The process has a strong bias to traditional community infrastructure, but is adaptable to dam safety decision making. The process is developed in two stages; Stage 1 adopts a risk-management framework to evaluate climate change sensitivity of permafrost to consequences of a climate change-induced failure and Stage 2 defines the detailed analysis identified in Stage 1. The decision making framework is outlined in Figure 2.
Figure 2: Climate Change Sensitivity Decision Framework (CSA 4011)
Stage 1 uses a risk screening matrix of sensitivity to climate change versus consequences of failure. The consequence categories have four levels, as outlined in Table 1. The CSA prepared the categories based on assessing traditional infrastructure, but inferences can be applied to dams. In applying the screening tool to dams, it is assumed that in traditional infrastructure a failure that causes loss of life would be considered catastrophic and would apply to any dam with a consequence classification of High or greater. A dam with a consequence classification of significant would have major consequences and Low consequence category dams would have negligible or minor consequences in the CSA screening tool. The process is familiar to risk based decision models common in dam safety.
The CSA provides guidance on sensitivity to permafrost to the type of soil in the foundation. Soil types are described based largely on origin and temperature of the permafrost. The sensitivity is to be assessed using the final year of the design life as the basis for defining the temperature zone, using projections for temperature increase. Dams may not be overly sensitive to climate change because they are unheated, and reduce the impact of increases in air temperature on the permafrost. However, the presence of a water body and increases in water temperaturecan have significant impacts on permafrost loss. There is little research to document to impact of seepage on permafrost losses.
The CSA prescribes four (4) levels of analysis based on the risk level identified in Table 1; A through D with A being the most onerous. The Guideline loosely suggests the level of climate change related analysis for each risk category as outlined in Table 2. If the structure is a Level A Risk project the project migrates to a stage 2 analysis. For Level B risk, the project can be screened as a Stage 1 project, but there must be sufficient data to be confident in the outcome. Level A and B projects can be completed as Stage 1 projects.
3. RISK ASSESSMENT METHODOLOGY
Sensitivity to climate change was determined for the Snare Forks Dykes using the methodology prescribed by CSA 4011.
3.1 Step 1 - Site Description and Background
The Snare Hydro System is located about 144kilometres north-northwest of Yellowknife, NT on the Upper Snare River. The Snare Hydro System is located in an area of extensive discontinuous permafrost, as shown in Figure 1. The system is comprised of a cascade of four (4) hydroelectric power generation facilities; Snare Rapids, Snare Falls, Snare Cascades and Snare Forks, each with a powerhouse and several water control structures. The combined generation capacity of the four (4) plants is 29.4 MW with rated plant capacities as follows:
Snare Rapids 8.5 MW
Snare Falls 7.4 MW
Snare Cascades 4.3 MW
Snare Forks 9.2 MW
There are fourteen (14) earthfill dams in the system and four (4) concrete spillway structures. Nine (9) of the embankments and all the concrete structures are founded on bedrock. Of the five (5) embankment dams on soil, three (3) are identified as being founded on lacustrine soil with permafrost. All three (3) are low head saddle dams that form the reservoir rim of Snare Forks reservoir, the last downstream hydro development.
June 15 2006, one of the saddle dams, Dyke 1, breached and resulted in a dam failure. The overtopping was attributed to undetected settlement of the dam and an elevated reservoir stage as a result of upstream discharge. The breach revealed a dam section consisting of sand fill over rock fill founded on clay till.
A cross section of the dam taken shortly after the failure is presented in Figure 3. There was no visible excess ice in the breach section but ice is suspected in the foundation soil.
Rock fill in the foundation of a dam is a problem. The consensus is that the rock fill formed the construction access road and was entombed, rather than removed, when the low areas were converted to saddle dams. The rock fill at Dyke 1 sloped from upstream to downstream and appeared to be continuous, which supports the idea of the rock providing support for a road. Test pits were excavated at Dyke 2 and Dyke 3 that verified a horizontal rock fill layer at those structures as well.
Figure 3: Breach Section, Dyke 1, Snare Forks, June 2006 (Lloyd Courage 2008)
Following the 2006 failure, Dyke 1 was rebuilt immediately with an upstream glacial till core and riprap for wave protection. A similar upstream glacial till impervious zone was constructed at Dyke 2 with riprap protection. A downstream sand fill filter was added to the downstream toe at Dyke 2 and Dyke 3. There is no impervious core added to Dyke 3.
The failure analysis report (CPL 2008) for Dyke 1 identified two causes of vertical deformation, (1) migration of fines from the embankment into the rock fill as a result of seepage, and (2) vertical deformation as a result of melting permafrost. Similarly, a potential failure modes analysis in 2015 identified two (2) high likelihood events, (1) permafrost melt causing settlement and loss of freeboard, and (2) increased seepage in the rock fill due to vertical settlement.
During the winter of 2014, the NTPC contracted Maskwa Engineering Ltd. to carry out a geotechnical investigation of the Snare dykes to characterize the subsurface conditions, including measurement of ground temperatures and qualitative descriptions of ice/permafrost conditions within the foundations of the dykes. The geotechnical investigation confirmed the internal geometry of the embankment as sand fill over rock fill over clay till as the foundation material. The ground was continuously frozen, as expected, but with thick lenses of excess ice (i.e. trace amounts of soil embedded in an ice matrix) measuring about 2.1 to 3.3 m-thick was encountered under the crests of Dykes 2 and 3, respectively. No ice was encountered in the borehole drilled at Dyke 1. Ground temperatures measured during the investigation are given in Figure 4.
Figure 4: Ground temperatures measured during drilling on March 16-19, 2014
Since the dykes are used as roads, the crests are cleared during the winter, while snow accumulation typically occurs upstream and downstream of the crest.
3.2 Step 2 - Functional Criteria
The intended function of the Dykes is to retain water in Snare Forks forebay up to the Maximum Flood Level (MFL) of 175.8 m. Subsequent to failure of Dyke 1 in 2006, Dyke 1 was reconstructed and Dykes 2 and 3 had their crest grade, elevation 175.9m, reinstated. The NTPC initiated annual crest surveys of all dams in the Snare Hydro System that indicated little change from 2006-2009. During an annual inspection in June 2013, it was noted that Dyke 3 had settled noticeably, while Dyke 2 also showed signs of settlement near the left abutment. Low points surveyed along the crests of Dykes 2 and 3 were about 0.6 m lower than the design crest elevation. Further inspection on August 20th, 2013 indicated a similar amount of settlement (0.6 m) near the right abutment.The NTPC initiated a program to reinstate the design grade. The differences in crest elevation are outlined in Figure 5.
Figure 5: Crest Survey of Snare Forks Dykes, before and after crest reinstatement
3.3 Step 3 - Characterization of Permafrost Sensitivity
Measured temperature trends indicate that the western Canadian Arctic is warming at a rate unprecedented in the last 400 years, while over the past several years, mean annual air temperatures have risen more rapidly in the eastern Arctic than anywhere else in Canada.Meteorological data for temperature from Yellowknife A was assessed for the temporal effects in temperature. Available Environment Canada data comprises monthly record of mean maximum temperature, mean temperature and mean minimum temperature. Both monthly and annual summary statistics were analyzed. The datasets were processed to remove seasonal variation (calculation of means, minimum and maximum) for each year in the study period. The data correlations with time were plotted to have a visual assessment of any trend in the data. Figure 6illustrates the temperature change over the past fifty (50) years.
Figure 6: Trends in air temperature, Yellowknife Airport
Upward trends are evident from the plots in all the three regimes of temperature. The annual minimum mean max, mean and mean min time series at each site have demonstrated the strongest increasing regression slopes. A statistically valid warming trend is observed at Yellowknife equal to 0.055oC/annum, which may produce a 4.8oC temperature rise over the period 1965 to 2050, which is very similar to the values predicted by others.
The initial temperature of the permafrost (i.e. mean annual air temperature) is -4.5 °C. Assuming an air temperature warming rate of 0.05 °C/year, the permafrost temperature at the end of a 50 year and 100 year period would be -2 °C and +0.5 °C, respectively. Using Table 7.3 in the CSA guideline (Canadian Standards Association, 2010), the permafrost sensitivity to climate change is high (H) for both cases due to the presence of layers of excess ice beneath the foundations of Dykes 2 and, 3.
3.4 Step 4 – Identification of relevant failure modes
Two principal failure modes have been identified for the Snare Dykes, 1) overtopping as a result of vertical settlement and, 2) internal erosion. Internal erosion is believed to occur only at elevated reservoir levels. The likelihood of failure of the dykes was described as high in a potential failure modes analysis, primarily an apparent increase in the rate of settlement. The consequence classification for Dykes, 1, 2, or 3 is “significant”, which may reasonably correspond to a “major” consequence inTable 1.
3.5 Step 5 – Climate change analysis resulting from screening
The results of the Stage 1 assessment for a major consequence structure with a sensitivity rating of high requires an A risk level of analysis, or a detailed-quantitative analysis. The project migrated to stage 2.
3.6 Stage 2 – Detailed Analysis
A two dimensional (2D) finite element model was carried out using Geo-Studio 2012 to simulate the thawing of the permafrost in foundation of Dyke 3 in response to warming of mean annual air temperature over a period of 50 years. Combined thermal and seepage effects are accounted for using a coupled convective analysis, in which the thermal modeling program TEMP/W and seepage modeling program SEEP/W communicate with each other at the end of each time step.
The cross section used in the two dimensional (2D) finite element model is shown in Figure 7. The actual lateral extent of the excess ice is not known, but is assumed to exist only directly beneath the dyke. The crests of the dykes are used as access roads and cleared from snow during the winter. Snow accumulates on the other surfacesinsulating them from the minimum air temperatures in the winter season.
Figure 7 Cross section of Dyke 3 assumed for the 2D finite element model
A constant geothermal gradient of 6 kJ/day/m was applied to the bottom boundary of the model. At the surface of the dyke crest, the ground surface temperature was assumed to be equal to the mean daily air temperature. The insulating effect of snow cover was accounted for by applying a factor of up to 0.6 for the coldest winter air temperatures.
For the initial conditions, it is assumed that the ground thermal regime is in equilibrium with the current climate (i.e. neglecting the effects of recent warming of annual air temperatures). This was achieved by modeling a 20-year calibration period in which the annual ground/reservoir surface temperature cycle was based on the Canadian Climate Normals from Environment Canada for Yellowknife, NT. After the initial calibration, the model was performed on a 50-year period of linear warming of the air at a rate of 0.055 °C/year. The resulting 70-year (20 year normal + 50-year warming) air temperature cycle is shown in Figure 8.
Material properties were estimated based on the methods described in the TEMP/W and SEEP/W user manuals(GEO-SLOPE International Ltd., 2012). Some of the relevant material properties are summarized in Table 3. The actual material properties change with time based on the degree of saturation and temperature, which is recalculated by the software for each time step. The warming effect of the forebay was accounted for by modeling it as a “soil” with a volumetric water content of 1.0 m3/m3.
Figure 8: Input air temperature cycle
Results indicate that at the end of a fifty (50) year period, the permafrost table will have lowered by approximately 0.72 m directly beneath the dam crest. The resulting settlement will cause the rock fill layer to progress deeper into the foundation, possible differential settlement, and will likely result in ongoing topping up of the dam crest.
The Snare Hydro System was constructed in the 1970’s and consists of a portfolio of fourteen embankment dams and four concrete dams, the majority of which are founded on bedrock and insensitive to permafrost loss due to climate change. There are three low head saddle Dykes at Snare Forks Hydro that have foundations of permafrost and lacustrine soils. The presence of permafrost in the form of excess ice was confirmed as part of a 2014 geotechnical investigation at two of the three dykes. In 2006, one of the Dykes, Dyke 1, failed and the failure was later attributed to overtopping as a result of crest settlement, in part due to internal erosion and in part due to settlement from permafrost losses.
Climate change models predict air temperature will increase by about 5.5 degrees celsius per century in the Snare region of the Northwest Territories. Temperature increases are having an impact on infrastructure of all sorts, roads, buildings and dams. The Canadian Standards Association developed a screening tool to provide guidance on the appropriate level of technical analysis to assess the impact of climate change on northern infrastructure (CSA 4011). The methodology is adapted slightly to make the toll applicable to dams, merging the different classification systems based on assessment of consequences.
When applied to dams, the guidance in CSA 4011 has a bias towards quantitative analysis because of the typically high consequences of a dam failure, compared to a foundation failure. However, the analysis does provide a logical process to assess the impacts of climate change in an efficient manner and provides guidance on how to incorporate climate change into analysis of dam safety.
Bulletin 105, “Dams and Related Structures in Cold Climate”, International Commission on Large Dams, 1996.
Bulletin 133, “Embankment Dams on Permafrost”, International Commission on Large Dams, 2007.
Canadian Dam Association, “Dam Safety Guidelines”, 2007.
CSA 4011-10, “Infrastructure in Permafrost: A Guideline for Climate Change”, Canadian Standards Association, June 2010.
Courage Projects Limited, “Inspection Report, Forks Dykes 1, 2 & 3 - Subsurface Investigation March 16-19, 2014”, NTPC, March 2014.
Courage Projects Limited, “Inspection Report, Forks Dykes 1, 2 & 3 - Subsurface Investigation March 16-19, 2014”, Northwest Territories Power Corporation, March 2014.
Maskwa Engineering Ltd., “Geotechnical Investigation for Snare Hydro, NT (draft)”, Northwest Territories Power Corporation, May 2014.
Meco, “Climate Change Study-Assessment of Impact on Hydropower Production in the NWT”, Northwest Territories Power Corporation, 2010.
109 Ilsley Avenue, Unit #14
Dartmouth, Nova Scotia, B3B 1S8
Toll Free: 1.877.434.3131