Geo-Risk 2023: Developments in Reliability, Risk, and Resilience

This paper presents an unsupervised anomaly detection method to infer changing ground conditions in real time during tunneling. The proposed method combines dimensionality reduction and density-based outlier detection methods. The method uses features related to the cutter, thrust, and ground conditioning systems of an earth pressure balance tunnel boring machine (EPBM) as the input. The principal component analysis (PCA) was used to extract information on the ground conditions from selected EPBM features, reduce the dimensionality, and embed the data points in a geometrical space. The local outlier factor (LOF) was used to measure the degree of the anomaly of the projected data points. This study used a data set from the State Route 99 (SR 99) tunnel construction in Seattle, WA. Based on this study, it can be concluded that with appropriately selected EPBM features, PCA can dynamically cluster EPBM data according to the ground conditions. Interestingly, the dynamic behavior of the projected data points does not substantially affect the soil clusters as they were always grouped. This study also demonstrates that LOF can be a sensible measure to detect changing ground conditions.

Interpreting stratigraphic condition is an essential task in geotechnical projects. However, algorithmically delineating and simulating stratigraphic profiles of non-stationary field, defined as heterogeneous stratum such as tectonically distorted or irregularly deposited strata, is still an open question and a challenging task in engineering practices. In this study, a novel approach that applies the image warping technique in computer vision to a non-stationary random field is developed, and it is further combined with an advanced stratigraphic stochastic simulation model. The image warping technique is effective for transforming non-stationary field into stationary field. Subsequently, an in-house developed stratigraphic stochastic simulation model integrates the Markov random field (MRF) model and the discriminant adaptive nearest neighbor-based k-harmonic mean distance (DANN-KHMD) classifier into a Bayesian framework and is applied to the transferred stationary field to efficiently estimate the stratigraphic uncertainty given sparse site exploration results. To demonstrate the effectiveness of the developed approach, a synthetic case is studied. We envision this approach can be further promoted in industry practices for an improved risk control in geotechnical engineering.

Quantitative tunnel face risk assessment is the characteristic challenge of rock tunnel excavation projects. This study establishes a multi-source database and proposes a stacked deep learning method for the quantitative tunnel face risk assessment. Contact and non-contact methods were used to collect multisource data of the tunnel (e.g., face images, site geological information, and rock mass properties). Thirteen multi-source variables describing the rock tunnel faces are considered inputs, and the rock tunnel face deformations computed by the numerical simulation are the target outputs. A staking model architecture is proposed to combine a range of well-performing models, which can make accurate predictions of tunnel face rock mass quality. The Tree-structured Parzen Estimator (TPE) algorithm is applied to determine the optimized model hyper-parameters automatically. The experimental results of a tunnel project in China suggest that the proposed stacking deep learning model performs well at assessing rock tunnel face stability.

This paper presents a framework to enable probabilistic assessment for braced excavation-induced structural damage on a regional scale. Random filed models are created to describe the uncertainties of spatially variable ground displacements induced by excavation, and random variables are adopted to model the uncertainties of soil stiffnesses, structural stiffnesses, and building weights. The uncertainties are propagated to the probability distributions of building characteristic tensile strains (ε_c) through a Monte-Carlo method, in which a 3-dimensional (3D) soil-structure interaction (SSI) model is evaluated in each simulation. With limit state functions defined based on ε_c, damage probabilities were quantified for all buildings in the region impacted by the excavation. Fragility heatmaps that can be used for estimating the probabilities of each possible damage state from impact level were also generated for each building. The framework is demonstrated with an excavation case history executed in Oslo, Norway. The 3D SSI model adopted in the framework can provide more accurate building response prediction than conventional 2D analysis. Meanwhile, the probabilistic assessment method provides a tool to quantify the uncertainty effect in the building assessment of large excavation construction.

In pipe jacking operations, the prediction of jacking forces has been the subject of several studies. Many established jacking force models are based on the strength properties of the negotiated soils or rocks during jacking. The application of these models is useful during the planning for pipe jacking construction. However, such models have limited use during the construction stage when decisions are made based on pipe jacking operation parameters, that is, jacking speed, lubricant, slurry pressure, etc. The influence of pipe jacking operation parameters on jacking forces is not well quantified. This gap could present adverse risks to construction progress, particularly in the highly weathered “soft rocks” encountered in the Tuang Formation underlying the central business district of Kuching, Malaysia. To assist in modelling pipe jacking operation, deep learning approaches have been widely deployed due to its powerful capability of feature learning. Therefore, an interesting question is raised: are we able to absorb the knowledge gained from neural networks to help reduce the decision-making gap in pipe jacking operations between the construction and planning phases? To this end, this paper proposes the use of gated recurrent units (GRUs) with an attention-based mechanism for providing insight into the decisions made on operation parameters during pipe jacking. A case study will be presented to demonstrate the visualization of attention utilized by the deep learning model towards the operation parameters. The findings showed that the input features (i.e., operation parameters) shifted throughout the drive. At the initial sections of the drive, more attention was directed towards operation parameters related to the micro-tunnel boring machine (mTBM) face. Subsequently, the attention was shifted to the cumulative operation parameters with progressive pipe jacking. The findings from this paper have the potential to help pipe jacking specialists identify and quantify the key operation parameters and assess their risks throughout a pipe jacking drive.

In this paper, the impact of major road tunnels’ failure on the operational efficiency of highway traffic network system is presented. A road traffic topology network based on the specific global positioning systems and geographic information of the main highway network is established. The road tunnels in this road network are considered as important nodes. A double-weighted network in terms of the traffic flow and physical length of the road is thus proposed to evaluate the efficiency of the network. Then, the topological property of this road network can be analyzed by using the network theory. The road network for Suzhou city in Jiangsu Province is adopted for the applicable example. The vulnerability of Suzhou road tunnel traffic network is carried out under two failure scenarios, for example, the sequential failure and random failure of tunnel nodes. Then in the specific case of extremely heavy rainfall weather, this paper analyzes the flood damage of each important tunnel node and the whole highway network’s topological structure with the rise of flood water level in ArcGIS.

The North Bristol Relief Sewer tunnel project involves a 6.5 km tunnel constructed primarily in the complex Mercia Mudstone Group formation. Early in the project, unexpected challenges with cutter tool wear and groundwater control were encountered, raising the need for further evaluation of the ground investigation data and a detailed assessment of the ground risks. This paper summarizes the geostatistical modelling approach used to characterize the lithology and geotechnical parameters within the Mercia Mudstone Group formation, and the added value of this approach to quantify uncertainties and risks. The developed geological model is validated by muck observations recorded during construction. In addition, discussion on recommendations for improved classification of the Mercia Mudstone as it pertains to tunnelling risks is presented.

Evaluation of the failure probability of large rockslides is crucial for quantitative risk assessment of rock slopes. An adequate failure probability evaluation of rock slopes requires at least two components: a reliable set of uncertain input parameters and a reasonable stability evaluation model. In this study, we propose a novel framework combining probabilistic back analysis and probabilistic failure prediction to improve the reliability of failure probability prediction of rockslides. A MATLAB implementation of the three-dimensional (3D) Spencer limit equilibrium method (LEM)—considering a two-layer lithology and irregular slip surfaces—was developed, and Monte Carlo simulation (MCS) was used to assess the failure probability efficiently based on the constructed stability model. The Baige rockslide that occurred twice in 2018 was taken as a study case. The input parameters representing the shear strength of rock masses were sequentially calibrated using the Bayesian updating method. Finally, the failure probabilities of two potentially unstable rock masses were estimated. The computed results can be employed to guide the quantitative risk assessment of the potentially unstable rock masses and provide a reference for similar rockslide stability assessments.

In drill and blasting tunneling method, a quantitative evaluation of geological conditions by drilling energy obtained from computer jumbos is being applied. In this study, a multivariate normal distribution model was developed to estimate geotechnical properties from drilling energy, and tunnel excavation analysis was conducted using the developed multivariate normal distribution model. The results confirmed that the confidence zone of the predicted values for both settlement and convergence covered the measured values. Furthermore, results of an attempt to update the multivariate normal distribution model using a data assimilation technique suggested the possibility of improving the prediction accuracy.

This paper explores the use of probabilistic design for mining activities and how to integrate probabilistic design into decision making in mining engineering. Mining design differs from civil rock engineering design with respect to exhibiting more demanding, and continuously changing, stress conditions, different excavation/extraction methods, and a larger scale of the operations. Relevant examples on the use of probabilistic-based design methods in civil and mining engineering are presented together with commonalities and case-specific characteristics. The acceptance level for the probability of failure is a part of the risk-based design process and needs to be selected together with decision-makers from other disciplines in the mine. Based on experience from, and commonalities with, civil engineering, the integration of probabilistic design in underground mining should focus on the uncertainty of material properties (e.g., effect of spatial variability, parameter distribution, etc.) and on how to implement theoretical findings into engineering practice. The probabilistic design assessment methods should be further evaluated by applying them to real mining cases.

Common machine learning has limited application in rockfall susceptibility mapping by a lack of interpretability. Represented by SHAP, explainable machine learning has been developed recently. This study proposed a novel interpretable hybrid-optimized model based on SHAP and XGBoost to interpret rockfall susceptibility evaluation results at both global and local levels. After hybrid-optimized by grid searching hyperparameters and recursive feature elimination (RFE) screening factors, only nine main factors selected by RFE from the 23 initial conditioning factors influence the occurrence of rockfall in the study area. The developed rockfall susceptibility evaluation model provided the accuracy, precision, and AUC value improved by 0.0846, 0.0809, and 0.0616, respectively, for the test data sets. The hybrid-optimized model has a good performance for rockfall susceptibility. The SHAP-based global and local interpretation could further insight into the rockfall occurrence mechanics.

The construction of rock tunnels is inherently associated with large geotechnical uncertainty, which not only affects the structural safety and other design requirements but also causes challenges in meeting time plans and budgets. Time and cost estimations of tunnel projects are however still mainly deterministic. The introduction of the contractual document called Geotechnical Baseline Report highlights, however, a need for clients and contractors to address how geotechnical uncertainty affects time and cost estimations. Using probabilistic time and cost estimation methods, a key parameter is the proportion of different construction classes along the tunnel. This paper introduces the Dirichlet distribution (a.k.a. multivariate beta distribution) as a suitable probabilistic model for this epistemic uncertainty and discusses its potential for Bayesian updating and risk assessments of Geotechnical Baseline Reports.

Smart Rock (SR) sensors designed at the University of New Hampshire have been used to instrument laboratory and field rockfall experiments. This paper summarizes the results of instrumented small- and medium-scale tests using SRs embedded in the test blocks, performed with the aim to enhance input parameters in rockfall models and to evaluate the contribution of block rotation, often disregarded in energy computations. Fabricated concrete blocks weighing 1 and 13.5 kg were released under controlled conditions onto a poorly graded sand from a height of 1 to 10 m. The impacted surface was also varied at angles between 0° and 42°. High-speed video and SR data were used to determine block translational and rotational velocities of 45 tests and to calculate their energy restitution after each drop. The results demonstrate that different test conditions produce distinct responses in terms of energy restitution, highlighting the difficulty of using fixed coefficients of restitution to realistically model rockfall trajectories.

The generalized Hoek-Brown failure criterion is widely used in rock mechanics to predict the failure of rock masses. It was introduced to the industry in its entirety in 2002. With the help of this criterion, the difficult to obtain rock mass material constants m_b, s, and a (herein termed the “derived parameters”) can be estimated with the use of the more easily obtained field parameters, GSI, m_i, and D (herein termed the “primary parameters”). As a result, the primary parameters and the generalized Hoek-Brown criterion have gained wide acceptance in the industry. At the same time, probabilistic analysis has become more popular in recent years. This paper considers the probabilistic analysis of a rock slope. Two cases are considered: (1) the random variables are assigned to the primary parameters; and (2) the random variables are assigned to the equivalent derived parameters. It was found that converting from primary to derived parameters resulted in a very significant difference in results if the correlation of the derived parameters is not considered. The importance of considering cross-correlation coefficients when performing a probabilistic analysis with derived parameters is highlighted.

With the emerging application of geotechnical asset management (GAM), there is still a need to further develop tools and methods to assess benefits of intervention actions and evaluate trade-off analysis to alternative projects under uncertainty. GAM has been proposed as a practice to reduce life-cycle costs, reduce performance and operational disruptions, and lead to fewer emergency stabilization projects, and therefore reduce the overall risks related to geotechnical assets. In this paper, we will discuss the use of quantitative risk management and return analysis as an approach to evaluate benefits and perform trade-off analysis between competing alternatives and projects. The study will focus on formularizing the concepts of uncertainty propagation of performance for groups of homogenous and heterogeneous geotechnical assets. Next based on the uncertainty analysis and return, procedures to implement a network level quantitative risk and benefit return from alternative short-term and long-term asset management plans will be discussed and illustrated using simplified examples.

Tunnel construction using tunnel boring machines (TBM) is associated with unforeseen ground conditions. There have been several attempts to use data recorded by the TBM to predict ground conditions ahead of the tunnel face and assist in the automation of the tunneling process. One of the main issues associated with such data, in particular for closed face machines, is that the operator has no view of the ground ahead, compromising ground labelling, which is often estimated from face mappings done at spaced intervals. The use of these “noisy” labels to train ground forecast models affects their performance. In this paper, confident learning (CL) is applied to investigate and detect label errors in a data set from a TBM tunnel at the Porto Metro Project, in Portugal. The detected mislabeled points are further investigated to determine the possible reasons for their mislabeling. Limitations, challenges, and directions for future research are discussed at the end.

Levees face the hazard of water side slope instability following water rise and subsequent rapid drawdown. Levee structures tend to be highly variable with sparse data to define soil properties. In addition, the hydraulic loading conditions for levees are largely controlled by natural processes, such as floods and storm surge, resulting in additional uncertainty. In this probabilistic parametric study, a hypothetical levee structure is examined using a variety of soil properties, including silty and clayey sands (SM and SC), silt (ML), and low and high plasticity clays (CL and CH). Typical properties (unit weight, coefficient of compressibility, and shear strength) will be assigned to each soil type along with high coefficients of variation, modeling highly variable levee materials. Four loading conditions were considered, combining long (three month) or short duration (two week) water rise with high (90% full) or moderate (50% full) water level. The start-of-drawdown seepage conditions and undrained shear strength properties for rapid drawdown analysis were modeled with simplified approaches appropriate to levees. Levees built from SM and SC soils were more sensitive to rapid drawdown due to greater progression of seepage and require careful characterization of undrained behavior. Levees built from clayey soils are unlikely to experience substantial seepage during flooding, making rapid drawdown much less important. Due to intermediate properties, the ML levee showed the most influence of the flood duration.

The process of dam failure is affected by various uncertain and interrelated risk factors. In order to quantitatively analyze the risk of dam failure, a reasonable method should be used for risk analysis. The Bayesian networks (BNs) has an excellent ability to inferencing event-related potentials. Most previous studies show this method has an inefficient analysis process and a subjective result due to the limitations of relying on domain knowledge. Therefore, this paper develops a Bayesian risk analysis model based on historical data statistics. The network structure was established through causal loop analysis, and the probability parameters were obtained by Bayesian learning, which can effectively improve the reliability of the model. Herein, a risk analysis model has been constructed based on the calculation of correlation factors in historical dam failure events from 1954 to 2021. Based on this model, the probability parameters of different dam failure modes caused by extreme weather have been deduced. According to the results, overtopping and structural instability are highly affected by extreme flood factors.

Climate change is a global risk and increases the frequency and intensity of extreme events. At the end of 2020, 98,566 dams were built in China, which are not only essential parts of the national flood defense system but also are key engineering measures to optimize the allocation of water resources. To ensure sustainable safety, climate change’s impact on dams should be fully considered during the design, construction, and operation periods. This paper analyzes the main causes of dam failure cases based on 3,558 dam failure cases in China from 1954 to 2021. Further attention focuses on the cases caused by extreme weather since the 21st century. The detailed case study is carried out on four reservoirs in the Usutule River basin, one of which collapsed by the influence of heavy rain in 2018. Finally, we discuss the climate change impacts on flood control safety and propose adaptation and countermeasure suggestions to strengthen dam safety management in the face of climate change.

Reliability analysis of geo-structures using random field method is getting more and more attention due to soil’s spatial variability. KL–FORM is a computationally efficient and accurate method for the reliability analysis of soil slope with spatial variability. A major concern of the KL–FORM is the determination of the number of KL expansion terms, which controls both the computation cost and accuracy of reliability analysis. The proper number of KL expansion terms can be selected by comparing the discretization error of the random field with a predefined allowable discretization error. To guide the selection of allowable discretization error of soil slope, a thorough analysis of the relationship among the allowable discretization error, size of the random field, auto-correlation distance, and reliability index of soil slopes with spatial variability is carried out, and a criterion for determining the proper allowable discretization error is proposed.

Double-row steel sheet pile (DSSP) cofferdam has been widely used in coastal zone engineering because of its good dynamic resistance, waterproof performance, and reusability. Due to the increasing frequency and intensity of extreme storms, DSSP cofferdam is facing more and more stress. In order to investigate the dynamic stability and failure mechanism of DSSP cofferdam under different foundation conditions, this paper carried out large-scale flume tests with different water levels and wave heights. It is found that DSSP cofferdams in sandy foundation have good stability due to large shear strength. The displacement is small due to the integrity of the DSSP. However, the seepage flow beneath the piles may make the landside seabed slope fail in the dock. DSSP cofferdams in clay foundation had good seepage resistance. However, the large settlement of the seaside pile would decrease the height of the anti-wave structure, resulting in large overflow waves and serious erosion on the landside seabed. Thus, the DSSP cofferdam in clay foundation should have a higher anti-wave wall due to the large settlement caused by the soft clay. The DSSP cofferdam built on the cohesive foundation has a longer buffer time during failure caused by the wave actions and offers more evacuation time for people behind the cofferdam. Besides, despite the failure, the residual height of the cofferdam was relatively high.

Levees can be subjected to excessive underseepage through their alluvial foundations during high water periods, which is abated by the installation of relief wells. The USACE uses a modified version of the blanket theory to design relief wells. However, there are significant uncertainties in the performance of relief wells when compared to field observations. Therefore, this study presents a reliability analysis to examine the probability of failure for relief well systems and examine the controlling parameters. This was accomplished using a generic 3D MODFLOW-USG model and a random forest regression model. The results from MODFLOW simulations were used for sensitivity analysis and for training a random forest model, which was used to perform Monte Carlo simulations to develop fragility curves. The sensitivity analysis results showed that the entry and exit distance and the aquifer’s hydraulic conductivity exhibit the greatest impact on total head. Fragility curves based on uniform random values of the design parameters showed that the probability of failure does not vary significantly as the well spacing increased. These findings show that there is a potential to decrease the economic cost since the well spacing does not affect the performance of relief wells significantly.

Loss evaluation of dike-break-induced flood disaster has become a worldwide concern due to its enormous economic, environmental, and societal importance. In this paper, the Sanjiao dike in the Poyang Lake district in China is investigated as a case study. An integrated dike-break-induced flood modeling and losses assessment framework is proposed. The MIKE 21-based numerical approach is developed to model the flood routing process in the flood protection region of dike. A flood routing model that can take into account the influence of land utilization on the roughness coefficient of underlying surface is established for the Sanjiao dike. By this means, the flood water depth, flow velocity, flood peak appearance time, and other spatiotemporally distributed flood inundation data can be obtained. With the behavior of the flood well understood, a physics-informed quantitative approach is further developed on ArcGIS and Python platform to assess the dike-break-induced flood losses, including life, economic, and ecological environment losses. The research outcomes provide an integrated decision-making process for flood warning and risk mitigation in the protection region of dike.

Relief well systems are commonly constructed to manage the risk of underseepage pressure induced distress, typically heave or uplift that leads to internal erosion, for levee systems across the United States. Typically installed near the landside levee toe, and extending parallel to the levee alignment, relief well arrays must be evaluated both individually and as a collective system. The evaluation of relief wells is commonly achieved through the application of pumping tests. Pumping tests allow for the assessment of relative performance degradation as relief wells advance through their life cycle. A common approach to the evaluation and assessment of relief wells has been to use the specific capacity ratio (SCR) as a measure of assumed incremental performance loss, as determined from constant rate pumping tests. Utilizing SCR, it is typical to establish a threshold of 80% relative to installation pumping test results as a means for determining potential need for corrective action. However, with advancing technological capabilities, in terms of commercially available software and real-time site geotechnical instrumentation, the evaluation of a wide range of conditions has allowed the practicing engineer to further consider what alternative thresholds may be critical to each levee system. Additionally, advances in risk-based assessment of levee systems have furthered the approach of examining observed performance and site-specific considerations when evaluating relief wells. This proposed method takes a systematic approach to the evaluation of relief well systems through the consideration of multiple critical factors, with the goal of providing a risk-informed approach to levee system relief well evaluation.

Inverse analysis through the finite element model simulation based on measured data is an economical and effective approach to calibrate the constitutive model of rockfill dams. However, this method has some limitations in efficiency and accuracy. This paper proposed a Gaussian process regression (GPR)-based response surface model combined with the Jaya optimization algorithm for parameter inverse analysis of rockfill dams. Gaussian process regression can replace finite element calculation to realize the nonlinear mapping relationship between model parameters and dam displacement. The effectiveness and engineering practicability of the proposed approach in the inverse analysis is tested in two constitutive models on a rockfill dam. Results show that the GPR response surface-assisted global optimization can provide a convenient and high-precision approach for the inverse analysis of rockfill dams at a low computation cost.

Fiber reinforcement is an effective technique for the prevention of shallow slope failures and repair of slope failures. An appealing application for fiber-reinforced soil is the stabilization of infinite slopes. In this study, a reliability-based analysis is conducted to recommend a set of safety factors for practical design scenarios of fiber-reinforced frictional infinite slope. Uncertainties entail the model uncertainty inherent in the prediction of the friction angle of the fiber-reinforced sand and the conventional uncertainties in the input parameters (internal friction angle of the unreinforced sand, unit weight of sand, and water level). Design charts are developed for an illustrative reinforcement scheme to allow the designer to select the suitable slope angle as a function of the friction angle of the unreinforced sand given a target reliability level.

In the United States, landslides exceed $2 billion in loss each year and account for numerous deaths. With thousands of events happening around the country, it is difficult for dispersed stakeholders to keep an eye on every single occurrence. With many factors contributing to slope failures such as prolonged precipitation, vegetation removal, and earthquakes in combination with population increase in rural areas, understanding geologic hazards can greatly reduce the impact on infrastructure, property, and loss of life. The risk is by definition the probability of the occurrence of an event that is not supposed to happen. When applied to the geological field, events such as slope failures appear to be unpredictable. However, with in-depth observations, many events announce themselves by showing trends and when captured in a timely manner can lead to proper mitigation measures. Due to the unexpected nature of geologic hazards, mitigations have historically been reactive. For decades, visual observations and manual monitoring techniques were used evolving later on into geotechnical and structural automated systems that allowed the development of proactive strategies based on data collected regularly from the field. From there, critical knowledge was built around sensor type, spatial deployment, data acquisition frequency, and in recent years, further developments have been allowed by the rise of computer science bringing wireless technology and intelligent monitoring solutions to the forefront. These so-called “intelligent monitoring systems,” like Senceive’s InfraGuard solution, are based on the use of wireless “MEMS” tilt sensors installed on stakes driven into the ground allowing the detection of relative movement of a slope. The sensors take readings in a defined interval and are triggered by a sudden event leading to additional readings, while also activating adjacent sensors and increasing the corresponding monitoring frequencies. The system can also be combined with a camera, which is also triggered by this event and allows visual inspection remotely. As developments and infrastructure projects increase, the need for automation and intelligent monitoring systems becomes a key factor to properly safeguard populations. This paper will provide an overview of such systems, covering various applications, as well as comparing the results with data from other monitoring techniques used in parallel for verification.

An important distinction between some of the different types of risk analyses performed in the dam safety setting involves the time frame over which the inputs and results are considered meaningful. A risk analysis completed at the periodic review or issue evaluation level is typically focused on long-term conditions; that is, on the risk posed by the facility if it continues to be operated in its current configuration. In contrast, a construction risk analysis would be focused on temporary or short-term conditions. Long-term risks are often viewed as “involuntary” in the sense that they arose from decisions made by previous generations of designers and decision makers. However, when a facility is modified, we are voluntarily changing some element of the inherited design or mode of operation, and temporarily assuming control over the risks posed by the facility. In this context, the risk portrayal chart visual guidelines are not necessarily informative, and the typical approach used for the evaluation of long-term risks does not apply. This paper describes Reclamation’s approach to construction risk assessment, including how our views on the topic have evolved, what we believe is essential to consider, and examples from recent construction risk analyses. Specifically, it explains why Reclamation is no longer using quantitative risk analysis in the temporary risk context, and discusses the advantages of a methodology focused on the identification and mitigation of key threats.

Understanding the mechanisms of side slope failure is of great significance in the study of landslide dam breaching. Smoothed particle hydrodynamics (SPH) is used to simulate the side slope failure of a landslide dam. The Drucker–Prager elastic–plastic model is implemented into the SPH to capture dam soil behavior. An artificial stress algorithm is applied to overcome the numerical instability in cohesive soils due to tensile forces between SPH particles. Uniform particle distribution method is implemented to improve the stability of stress calculations of adjacent solid boundary particles. The influence of the distribution and composition of soil materials as well as the initial erosion depth is investigated in landslide dams. It is found that the initial internal angle and soil cohesion strongly affect the plastic zone formation, the volume of subsequent dam failure, and the final geometry of the dam breach. The surface sliding and dam instability of dam soils are observed simultaneously in the model. The amount of slope sliding coming from the side slope failure increases with the initial erosion depth.

Slope failure occurs frequently in the loess and loess-derived deposits across the Loess Plateau, northwest China, and unsaturated scenarios occupy quite a large proportion of total failure amount. For reliability analysis of unsaturated slope stability, the uncertainty quantification of soil-water characteristic curve (SWCC) is unavoidable because SWCC plays a vital role in constructing the linkup between saturated and unsaturated soil mechanics. The uncertainties in estimated SWCC arise from not only the determination of model parameters but also the selection of favorable parametrical models. Considering the complex soil water characteristics of loess, the SWCC model of loess is not unique, such as unimodal or bimodal SWCCs. Until now, the influence of SWCC models on reliability analysis of unsaturated loess slope remains unknown. This paper proposes a reliability analysis approach for unsaturated loess slope with the consideration of the bimodal feature of SWCC. This method can simultaneously incorporate the SWCC model selection uncertainty and parameter uncertainty into reliability analysis under the framework of Bayesian model averaging. The proposed approach is illustrated using the example of irrigation-induced landslide in Miaodian village of Jingyang southern tableland. Results show that uncertainty in the selection of SWCC models (e.g., unimodal or bimodal) significantly affects the reliability of unsaturated loess slope stability, and the failure probability may be overestimated and underestimated if only considering unimodal SWCC. It is necessary to consider the model selection uncertainty of SWCC in unsaturated loess slope stability analysis and risk assessment.

A risk-based design approach for earthen dams (existing or proposed, for water or waste) has been developed and illustrated by a simple example. It is comprised of (1) a well-established “tolerability limit” for safety risk (maximum allowable annual probability of exceeding any particular number of equivalent fatalities), which must be met regardless of other aspects of performance; and (2) “as low as reasonably practicable” (ALARP) for all stewardship (environmental, safety, and social) risks, which is of primary interest to public/regulators, as well as for business (local financial and global corporate) risks, which is also of interest to the dam owner/corporation. The assessment of risks associated with any particular dam system option and their subsequent evaluation with respect to these criteria, if appropriately done as discussed in this paper, will lead to defensibly acceptable and best appropriate dam design.

While there is consensus that a quantitative risk assessment (QRA) is the highest level of risk analysis for dam safety, a detailed definition and framework for a QRA is not described in available guidance documents. We propose a definition for a QRA and describe the attributes that distinguish it from other types of risk analysis. In particular, we believe that a QRA should (1) prioritize the use of engineering models and methods over estimates that are largely based on subjective judgements of experts to develop inputs to the risk analysis; (2) employ structured interaction among experts to achieve a well-documented risk analysis that captures the center, body, and range of technically defensible interpretations of data, models, and methods used in the risk analysis; and (3) have a clearly defined framework for identifying and evaluating sources of randomness and uncertainty in data, methods, and models used in the risk analysis.

Devastating damages and failure of tailing dams during the earthquake and heavy rains have been well acknowledged in the past. The factors associated with the tailing dam’s failures are earthquake characteristics such as peak ground acceleration (PGA), frequency content, duration and epicentral distance of earthquake; rainfall characteristics like the amount, the intensity, the duration, the frequency or return period, the seasonal distribution, and velocity of the rain flow. Design criteria of the dam, including the foundation soil below the dam and the topographic nature of the local sites, are also essential factors for the tailing dam performance during extreme events. An extensive review is carried out to investigate the dam failure during the 1962−2022 earthquakes and rainfall. Based on historic frequency of extreme events, the damage intensity scale (DIS) for a tailing dam is proposed in the present study. The earthen tailing dams with a geometry ratio greater than 0.8 are more vulnerable than others. Several co-relations are also examined in the present study to identify the damage governing parameters during extreme events.

Monopile foundation design for supporting offshore wind turbines has largely relied on classical p-y approaches originally developed for longer and more slender pile foundations. The shorter length and greater diameter of monopoles, however, has required empirical modification to the p-y approach to account for shear deformations and coupling between deflections and rotational deformations. Analysis of laterally loaded monopiles is fundamentally 3D, and recent studies have aimed to calibrate p-y springs using full 3D finite element analyses. This paper presents a preliminary study that uses a pseudo-3D approach suitable for modeling axisymmetric solids subjected to non-axisymmetric loading that is ideally suited for this type of geotechnical system. The method uses 2D finite element analysis in radial planes with circumferential variations modeled using a Fourier series. Soil property variation with depth is modeled using random fields. For risk assessment studies requiring Monte-Carlo simulations, the pseudo-3D approach offers greatly reduced computational costs compared with full 3D analysis. While the pseudo-3D approach is not new, it has not, to the authors’ knowledge, been combined with a random field description of soil properties. The results presented in this preliminary study focus on estimation of the probability of excessive lateral deformations at the submerged ground surface.

Design for the widely used gravity foundations supporting wind turbines often requires rigorous rotational stiffness analysis in addition to the conventional foundation design requirements such as overturning stability, bearing capacity, and settlement. The rotational stiffness is related to the shear modulus of the bearing soils and the foundation size. The size may vary depending on the experience and skill of engineers while the stiffness is subject to degradation under cyclic loads. With the trend of using larger wind turbines and reducing the construction cost of the foundation, allowing the foundations to uplift during turbine operation is gradually adopted in the wind energy industry which was not previously allowed according to some practical guidelines. Since an uplifted foundation reduces the foundation’s effective bearing area and increases the potential for soil degradation under cyclic loads, it is imperative to develop a method to judge the risk associated with the uplift ratio. Aiming to this goal, this paper performed a study on the failure probability of rotational stiffness of circular gravity wind turbine foundations in clays. The foundation sizes for given wind loads and site conditions were assumed normal distributions, and the uplifted ratios corresponding to the diameters of the foundations were obtained for different load conditions. Numerical simulations with Plaxis 3D for these uplifted foundations using a soil-hardening with a small strain constitutive model were carried out to obtain excess pore water pressure and degraded stiffness. The stiffness degradation was interpreted with shear-strain and excess pore water pressure ratios, and the probability distributions were fitted. Finally, the failure probability for the rotation stiffness was analyzed utilizing Monte Carlo simulations based on performance functions, which were built based on ACP (2021) and empirical equations recommended by DNV.

Shenhu area is a potential hydrate exploitation zone. However, hydrate exploitation in this area may induce regional debris flows, which can destroy infrastructure. To quantitatively evaluate the areas vulnerable to disaster and the intensity of a potential disaster, the run-out behavior of debris flows should be systematically investigated. Previous studies have investigated the influences of the ambient fluid and the yield strength of the sliding material on the run-out scenarios of debris flows. This paper investigates the influence of the initial location of debris flow and analyzes the extreme disaster scenarios of concurrent multiple debris flows. Results show that a debris flow may damage an infrastructure only when they lie in the same watershed. In the extreme disaster scenarios of concurrent multiple debris flows, the maximum mass thickness and mass flow velocity are significantly greater than those of a single debris flow. The debris flow with a higher velocity can “absorb” the debris flow with a lower velocity to form a larger sliding body, which may amplify the damage to infrastructure.

Fiber-reinforced polymer (FRP) composites coupled with seawater sea-sand concrete (SSC) provide an innovative and sustainable solution by replacing the conventional piling materials for the marine infrastructure. This study investigates the axial behavior of FRP composite SSC model piles subjected to cyclic loading of different amplitudes and mean load levels. The strain along the depth of piles is measured by an advanced distributed optic sensing technique called optical frequency domain reflectometry (OFDR) having a spatial resolution of 1 mm and ±1με sensing accuracy. Three structural configurations, FRP tube confined and FRP rebars cage reinforced and centered FRP rebar SSC piles ended in rock-socket, are investigated in physical models to examine the performance of FRP composites and SSC in pile foundations. The accumulated displacement of model piles under different modes of axial cyclic loading are analyzed and explored in detail. It is found that the accumulation of permanent cyclic displacement increases markedly initially till 30 cycles and then followed a constant trend with increasing cycles passing. Under the same cyclic loading conditions, the FRP tube confined model piles exhibited lower cyclic degradation leading to stable behavior. The FRP tube confined model piles showed higher confinement and axial cyclic capacities compared to those reinforced with FRP rebars. The OFDR sensing technique monitored the localized effects efficiently that how the load is distributed along the length of model piles.

Coastal erosion presents major risks to offshore communities. This paper describes the development of GIS-based remote sensing image processing to evaluate the historical erosion progression along Lake Erie. The historical erosion rates of the coastal line are determined from the GIS data. Empirical and data-driven models are developed to predict the future erosion trend. The study also evaluates the effects of climate change on shoreline erosion. The corresponding risk in property and financial loss are evaluated. The analyses’ results allow the community to make informed decisions and actions for shoreline erosion mitigation based on risk associated with different climate change scenarios.

It may become necessary to perform dredging near existing structures either for channel deepening or for environmental remediation of contaminated sediments. The dredging adjacent to the existing structures is likely to impact the geotechnical as well as the structural stability of the structures and could result in deformation or failure of the structures depending upon the extent and depth of dredging. The risk to the existing structures due to dredging is dependent on geotechnical conditions both on the marine and upland sides of the structure, the condition of the structures, structural details, and the extent and depth of the dredging. Some dredging projects may involve a large number of structures where there is no access to the upland areas nor design and construction records for structures, making individual detailed assessments difficult, costly, and often impractical. In addition, accurate risk assessment can be especially challenging for older structures where as-built construction plans and detailed loading information are not available, and where geotechnical data is limited. Accordingly, a risk-management/avoidance-based approach is needed to assess potential impacts and limit the risk of damage to existing shoreline structures from the impacts of dredging. This paper presents a methodology for evaluating and avoiding risk through a progressive step-by-step process to provide increased resolution in stages as a screening tool to progress from general risk screening to identification and avoidance of specific structure risks.

The main goal of this paper is to present a new tool for students to learn and solve rock slope stability problems in which the failure results from sliding on a single planar surface dipping into the excavation. It consists of an Excel spreadsheet with a graphic interface created with the aid of VBA (Visual Basic for Applications) to help in the data entry and formulation of the problem. With this spreadsheet, students can easily solve 2D problems of plane sliding, with and without reinforcement. This paper also presents a probabilistic approach to deal with this kind of problem using Monte Carlo simulations. Some example problems are presented to demonstrate the capability of the software. Each parameter can be selected to be represented by its probability distribution; therefore, it is possible to examine the effect of variability on slope stability. The spreadsheet consists of a good learning tool for basic theory related to rock mechanics, also how probabilistic analysis can be applied in geotechnical engineering.

A fully integrated mixed reality game system called multiphysics enriched mixed reality for integrated geotechnical education (MERGE) is developed to improve student education in the context of geotechnical engineering. Several minigames are designed for students to enhance their geotechnical knowledge. Players can earn points and update their appearance by playing these minigames, which stimulates their interests in geotechnical engineering. By providing students with visualization, collaboration, and simulation tools, we hope to promote the understanding of geotechnical experiments. Based on the laboratory results, numerical experiments are conducted to help students understand the geotechnical application. The leveraging mixed reality technology offers an opportunity for students to access advanced equipment in geotechnical experiments. This game will be tested by students in next semester. The main contribution of this work is a discussion of the educational technology and processes behind implementing a mixed reality educational game. We provide the educational background to inform researchers who seek to develop similar games.

According to the summary report for TC309/TC304/TC222 Third ML dialogue on “Data Driven Site Characterization (DDSC)”, many researchers and practitioners opined that an insufficient education on probability, statistics, and data analysis in universities is one of the impediments for promoting and enhancing professional activities in geotechnical and geoenvironmental engineering in areas related to reliability-based design, risk-based decision analysis, and probabilistic site characterization. This study investigated the current status of education at the age of Industry 5.0 and georisk education for civil engineering in Japan. In order to investigate the status, we reviewed the education policy, especially mathematical education and recent revisions in the curriculum guideline in Japan. Current curriculums for civil engineering department in some Japanese universities were also investigated.

The teaching for geotechnical reliability analysis to graduate students is difficult because the concept is abstract and the methods are complicated. Reasons for adopting case-based teaching method for geotechnical reliability analysis are explained, and some specific case-based teaching methods are summarized. The case-based teaching is mainly composed of three parts. The first part consists of teaching of the importance of geotechnical reliability analysis. Geotechnical investigation report is taken as an example for explaining the reason and necessity of geotechnical reliability analysis. The second part consists of teaching of several popularly used reliability analysis methods. Three types of examples with increasing complexities are used for the teaching and comparison of reliability analysis methods. The third part consists of teaching of the application of geotechnical reliability analysis in engineering practice. National norms are recommended to students to strengthen their understanding of the practical importance and necessity of leaning geotechnical reliability analysis.