Load capacity assessment of concrete dams often includes verification of the stability for multiple separate failure modes, such as sliding and overturning. However, in the case of dams, the underlying failure mechanism for these failure modes may be too idealized, and the analysis could yield inaccurate results. Previous research has, for example, shown that regular rigid-body sliding failure analysis provides inaccurate load capacity estimates for dams with uneven interface geometries. This article discusses the behavior of such dams and presents a failure mode that combines the traditional sliding and overturning failures. The failure mode is termed combined sliding and overturning and serves as an intermediate to the traditional failure modes. It allows for the assessment of concrete dams with uneven interface geometries, whose behavior is not expected to be fully represented by only sliding or overturning. To estimate the load capacity for the presented failure mode, an analytical formulation based on simple force and moment equilibrium is provided. The formulation is compared with finite element simulations and previously reported results from experimental scale model tests and is shown to accurately predict the load capacity.

Common analytical assessment methods for concrete dams are unlikely to predict material fracture in the dam body because of the assumption of rigid body behavior and uniform- or linear stress distribution along a predetermined failure surface. Hence, probabilistic non-linear finite element analysis, calibrated from scale model tests, was implemented in this study to investigate the impact of concrete material parameters (modulus of elasticity, tensile strength, compressive strength, fracture energy) on the ultimate capacity of scaled model dams. The investigated dam section has two types of large asperities, located near the downstream and/or upstream end of the rock-concrete interface. These large-scale asperities significantly increased the interface roughness. Post-processing of the numerical simulations showed interlocking between the buttress and the downstream asperity leading to fracture of the buttress with the capacity being determined mainly by the tensile strength of the buttress material. The capacity of a model with an asperity near the upstream side, with lower inclination, was less dependent on the material parameters of the buttress as failure occurred by sliding along the interface, even with inferior material parameters. Results of this study show that material parameters of the concrete in a dam body can govern the load capacity of the dam granted that significant geometrical variations in the rock-concrete interface exists. The material parameters of the dam body and their impact on the capacity with respect to the failure mechanism that developed for some of the studied models are not commonly considered to be decisive for the load capacity. Also, no analytical assessment method for this type of failure exists. This implies that common assessment methods may misjudge the capacity and important parameters for certain failure types that may develop in dams.

Analytical methods of structural stability assessment of concrete dams are often too simple and thus conservative in their predictions. Without the actual foundation geometry, capacity for some rigid body failure modes are underestimated. This is problematic when deciding upon remediation activities of a dam that is considered unstable and may divert the restoration activities from where they are most impactful. In a previous study by Sas et al. 2019 where a section of an existing dam was scaled down and tested experimentally, the model indicated that several areas were experiencing large stresses, potentially leading to failure. This raised the research question whether another type of failure would occur for different material properties. Therefore, this paper delves into a probabilistic numerical approach, through finite element analysis, to evaluate dam stability based on randomization of a number of material properties such as modulus of elasticity, tensile strength, compressive strength, and fracture energy. The variation of the aforementioned material properties did not impact the failure mode, which was consistent among a broad range of material strengths.

This paper presents the development and installation of a prototype ice load panel and measurements of ice load from February 2016 to February 2018 at the Rätan hydropower dam in Sweden. The design of the 1 × 3 m² panel enables direct measurement of ice pressure on the concrete surface is based on previous experience from similar measurements with sea ice. Important features of the design are sufficient height and width to reduce scale effects and to cover the ice thickness and variations in water level. The Rätan dam was chosen based on several criteria so that the ice load is considered to be reasonably idealized against the dam structure.

For the three winters 2016, 2016/2017, 2017/2018, the maximum ice load recorded was 161 kN/m, 164 kN/m and 61 kN/m respectively. There were significant daily fluctuations during the cold winter months, and the daily peak ice loads showed a visual correlation with the daily average temperature and with the daily pattern of operation of the power station with its corresponding water level variations.

De kalla svenska vintrarna innebär att betongdammarna riskerar att utsättas för istryck. I Kraftföretagens riktlinjer för dammsäkerhet (RIDAS) anges ett dimensioneringsvärde som varierar mellan 50 – 200 kN/m från söder till norr. Särskilt vid låga dammar i norra Sverige så kan islasten utgöra en stor andel av den horisontella kraften som verkar på dammen. Trots islastens stora påverkan, är kunskapen om den faktiska islasten som verkar direkt mot dammen relativt låg.Inom det tidigare SVC projektet ”Lastförutsättningar avseende istryck” har en prototyp av en lastpanel utvecklats, vilken är avsedd för att mäta islaster som verkar på betongdammar. Panelen sitter sedan 2016 installerad på Rätans betongdamm och har sedan dess mätt islasten mot dammen.I denna rapport presenteras en utvärdering av olika prediktionsmodeller avsedda för att förutsäga istryck, där dessa prediktioner jämförs mot de uppmätta istrycken vid Rätan. För detta har två befintliga modeller använts:• En halvempirisk modell föreslagen av Comfort, et al. (2003). • En mekanisk modell där istryck modelleras utifrån spänningar i isen, där isen modelleras med Norton-krypning enligt Petrich, et al. (2015).Beräkningarna visar att de modeller som tillämpas i denna rapport ger signifikanta avvikelser jämfört med de islastmätningar som genomförts vid Rätan. Detta gäller både den maximala islasten och den relativa skillnaden mellan de dagliga islasttopparna som sker under säsongen. Det är utifrån dagens kunskapsläge inte möjligt att avgöra om orsaken till denna avvikelse mellan modeller och mätningarna beror på felaktigheter i modellerna eller mätningarna, eller om det är en kombination av båda. Det framgår dock tydligt vid utvärderingen att avvikelsen mellan dessa främst beror på den del av islasten som uppkommer till följd av vattenståndsvariationer.Tidigare studier visar att islaster orsakade av vattennivåvariationer är den mekanism som resulterar i de största islasterna mot dammar. Trots detta, så är denna effekt exkluderad i majoriteten av alla modeller för islaster. Det finns således ett tydligt gap mellan de mekanismer som observerats i mätningar och befintliga prediktionsmodeller. Modellen som utvecklats av Comfort beaktar inverkan från vattenståndsvariationer, men resultatet visar att just denna del av modellen

Concrete dams in the cold Swedish climate may be subjected to a pressure load from the ice. The hydropower industries guidelines for dam safety (RIDAS) specify that dams should be designed for an ice load between 50-200 kN/m, varying from south to north. This load can thereby be significant and may constitute a large portion of the total horizontal load, especially for smaller dams in the northern part of Sweden. Despite this, the current understanding of ice loads is limited.In the previous SVC project, "Load conditions for ice pressure," a prototype ice load panel was developed. The panel was installed on Rätan's concrete dam in 2016 and have measured the ice load against the dam since then.In this report, different models intended to predict ice load are presented and these models are applied to predict the measured ice load from Rätan. For this, two models were used:• A semi-empirical model proposed by Comfort et al. (2003). • A mechanical model where ice pressure is modelled based on ice-stresses, where the thermal ice stresses are modelled with Norton creep as proposed by Petrich, et al. (2015).Both models inaccurately predict the total and the relative magnitude of the measured ice load peaks, and underestimate the maximum measured load during the winter. It is from today's knowledge not possible to determine whether the measurements are correct and if the prediction from the models is wrong, vice versa or a combination of both. Especially the part of the model that predicts the ice loads from water level variations shows low explanatory value, and currently no model exists that accurately can predict ice loads caused by water level variations.Two things seem obvious regarding the modelling of ice loads; ice load caused by water level variations is the mechanism that provides the most significant contribution to ice load on the dam, and at the same time almost all prediction models for ice load on dams only consider thermal loads. There is thus a clear gap between the mechanisms found by measurements and the development of models. The model developed by Comfort takes into account the impact of water level variations. Still, the result shows that this particular part of the model gives the least

In this paper, a reliability based framework for the assessment of sliding stability for concrete dams was presented. The framework consists of several parts based on the Probabilistic model code for concrete dams developed by Westberg-Wilde and Johansson and includes guidelines on how reliability based sliding stability assessment should be performed, together with recent work by Krounis et al. how to account for partially bonded interfaces. In the proposed framework, the assessments start with performing preliminary calculations using a priori assumptions on parameters included in the analysis. If cost-benefit analyses show that further analyses could be beneficial, investigations are undertaken on relevant parameters in the failure modes. The results from the investigations are used to update the calculations in the assessment and decisions on stability enhancing measures are undertaken if necessary. In the presented example the preliminary sliding stability analysis of the interface, before testing was performed, showed a reliability index of 4.91, indicating an unacceptable failure probability of the dam without any testing. Taking into account the information obtained from testing the basic friction angle of the interface increased the reliability index from 4.91 to 7.24, clearly showing the gain of including test results in the assessment. When the influence of cohesion was accounted for a reliability index of 6.49 was obtained, which shows that cohesion can give a potential gain to the stability, even though it in this case still is lower than the gain from updating the basic friction angle. When both limit states of the interface were considered as a system the reliability index increased to 8.1.

In the last decade the attention and application of a reliability-based methodology for concrete dams has increased. A recent project aiming at bringing forth a reliability-based methodology for design and assessment of concrete dams founded on rock for conditions applicable in a Nordic climate has resulted in a “Probabilistic model code for concrete dams” (PMCD). The objective of Theme D was to estimate probability of failure of an existing concrete dam for sliding along the concrete/rock interface and sliding along a joint in the rock mass, using the PMCD. The dam analyzed is a 25 m high concrete gravity dam located in the north part of Sweden. Contributions from six authors were received and have been analysed in this summary along with a reference solution by the authors. The first assignment was to estimate the deterministic factor of safety. Although the definitions of the factor of safety were similar there was large differences in the results. For the probabilistic analysis, definition of limit state functions was straight forward and have been defined similarly. Variables in the probabilistic analysis were defined somewhat differently, e.g. for concrete density, friction angle and ice loads. The results of the probabilistic analysis of sliding along the interface for normal water levels were varying, although five of the results were within the range of β = 3.7-5. There was less variability for the flood load case and for sliding along the rock joint. There reason was considered to be mainly due to the different parameter definitions. Identification of the most important parameters was successful; although the exact sensitivity values varied (due to variation in parameters), the most important factors were identified. In the calculation of system reliability, the previously described differences were reflected. Bayesian updating proved to be a tricky task, where especially results of the updated standard deviation varied. One conclusion is, however, that the updating of the friction angle is rewarding in terms of increasing the safety index due to the reduction in epistemic uncertainties.

For a probabilistic methodology to be trustworthy it should produce stable and reproducible results. The conclusion is that the PMCD is successful as a guideline in this process, but that further development and more experience of practical use is necessary. More benchmarks of similar characteristics are thus believed to be a good way forward and a broader discussion among practitioners would also be beneficial in reaching a “consensus” on how to perform reliability-based assessments.