Scipedia: F. Salazar Self-Archive

Validation of Machine Learning Models for Structural Dam Behaviour Interpretation and Prediction

Fernando Salazar — Wed, 03 Nov 2021 09:14:03 +0100

The main aim of structural safety control is the multiple assessments of the expected dam behaviour based on models and the measurements and parameters that characterise the dam’s response and condition. In recent years, there is an increase in the use of data-based models for the analysis and interpretation of the structural behaviour of dams. Multiple Linear Regression is the conventional, widely used approach in dam engineering, although interesting results have been published based on machine learning algorithms such as artificial neural networks, support vector machines, random forest, and boosted regression trees. However, these models need to be carefully developed and properly assessed before their application in practice. This is even more relevant when an increase in users of machine learning models is expected. For this reason, this paper presents extensive work regarding the verification and validation of data-based models for the analysis and interpretation of observed dam’s behaviour. This is presented by means of the development of several machine learning models to interpret horizontal displacements in an arch dam in operation. Several validation techniques are applied, including historical data validation, sensitivity analysis, and predictive validation. The results are discussed and conclusions are drawn regarding the practical application of data-based models.

A review on thermo-mechanical modelling of arch dams during construction and operation. Effect of the reference temperature on the stress field

Fernando Salazar — Thu, 23 Sep 2021 09:21:03 +0200

Double-curvaturedamsareunique structures for several reasons. Their behaviour changes significantly after joint grouting, when they turn from a set of independent cantilevers into a monolithic structure with arch effect. The construction process has a relevant influence on the stress state, due to the way in which self-weight loads are transmitted, and to the effect on the dissipation of the hydration heat. Temperature variations in the dam body with respect to those existing at joint grouting generate thermal stresses that may be important in the stress state of the structure. It is thus essential to have a realistic estimate of this thermal field, also called reference or

An Interactive Tool for Automatic Predimensioning and Numerical Modeling of Arch Dams

María Jesús Samper — Tue, 21 Apr 2020 17:59:31 +0200

The construction of double-curvature arch dams is an attractive solution from an economic viewpoint due to the reduced volume of concrete necessary for their construction as compared to conventional gravity dams. Due to their complex geometry, many criteria have arisen for their design. However, the most widespread methods are based on recommendations of traditional technical documents without taking into account the possibilities of computer-aided design. In this paper, an innovative software tool to design FEM models of double-curvature arch dams is presented. Several capabilities are allowed: simplified geometry creation (interesting for academic purposes), preliminary geometrical design, high-detailed model construction, and stochastic calculation performance (introducing uncertainty associated with material properties and other parameters). This paper specially focuses on geometrical issues describing the functionalities of the tool and the fundamentals of the design procedure with regard to the following aspects: topography, reference cylinder, excavation depth, crown cantilever thickness and curvature, horizontal arch curvature, excavation and concrete mass volume, and additional elements such as joints or spillways. Examples of application on two Spanish dams are presented and the results obtained analyzed.

Development and validation of a multivariate predictive model for rheumatoid arthritis mortality using a machine learning approach

María Jesús Samper — Tue, 21 Apr 2020 17:40:26 +0200

We developed and independently validated a rheumatoid arthritis (RA) mortality prediction model using the machine learning method Random Survival Forests (RSF). Two independent cohorts from Madrid (Spain) were used: the Hospital Clínico San Carlos RA Cohort (HCSC-RAC; training; 1,461 patients), and the Hospital Universitario de La Princesa Early Arthritis Register Longitudinal study (PEARL; validation; 280 patients). Demographic and clinical-related variables collected during the first two years after disease diagnosis were used. 148 and 21 patients from HCSC-RAC and PEARL died during a median follow-up time of 4.3 and 5.0 years, respectively. Age at diagnosis, median erythrocyte sedimentation rate, and number of hospital admissions showed the higher predictive capacity. Prediction errors in the training and validation cohorts were 0.187 and 0.233, respectively. A survival tree identified five mortality risk groups using the predicted ensemble mortality. After 1 and 7 years of follow-up, time-dependent specificity and sensitivity in the validation cohort were 0.79–0.80 and 0.43–0.48, respectively, using the cut-off value dividing the two lower risk categories. Calibration curves showed overestimation of the mortality risk in the validation cohort. In conclusion, we were able to develop a clinical prediction model for RA mortality using RSF, providing evidence for further work on external validation.

Engaging soft computing in material and modeling uncertainty quantification of dam engineering problems

María Jesús Samper — Tue, 21 Apr 2020 17:04:35 +0200

Due to complex nature of nearly all infrastructures (and more specifically concrete dams), the uncertainty quantification is an inseparable part of risk assessment. Uncertainties might be propagated in different aspects depending on their relative importance such as epistemic and aleatory, or spatial and temporal. The objective of this paper is to focus on the material and modeling uncertainties, and to couple them with soft computing techniques aiming to reduce the computational burden of the conventional Monte Carlo-based finite element simulations. Several scenarios are considered in which the concrete and foundation material properties, the water level, and the dam geometry are assumed as random variables. Five soft computing techniques (i.e., random forest, boosted regression trees, multi-adaptive regression splines, artificial neural networks, and support vector machines) are employed to predict various quantities of interest based on different training sizes. It is argued that the artificial neural network is the most accurate algorithm in majority of cases, with enough accuracy as to be useful in reliability analysis as a complement to numerical models. The results with 200 samples in the training set are enough for reaching useful accuracy in most cases. For the simple prediction tasks, the results were predicted with less than 1% error. It is observed that increasing the number of input parameters increases the prediction error. The partial dependence plots provided most sensitive variables in dam design, which were consistent with the physics of the problem. Finally, several practical recommendations are provided for future applications.

A performance comparison of machine learning algorithms for arced labyrinth spillways

María Jesús Samper — Tue, 21 Apr 2020 16:53:54 +0200

Labyrinth weirs provide an economic option for flow control structures in a variety of applications, including as spillways at dams. The cycles of labyrinth weirs are typically placed in a linear configuration. However, numerous projects place labyrinth cycles along an arc to take advantage of reservoir conditions and dam alignment, and to reduce construction costs such as narrowing the spillway chute. Practitioners must optimize more than 10 geometric variables when developing a head–discharge relationship. This is typically done using the following tools: empirical relationships, numerical modeling, and physical modeling. This study applied a new tool, machine learning, to the analysis of the geometrically complex arced labyrinth weirs. In this work, both neural networks (NN) and random forests (RF) were employed to estimate the discharge coefficient for this specific type of weir with the results of physical modeling experiments used for training. Machine learning results are critiqued in terms of accuracy, robustness, interpolation, applicability, and new insights into the hydraulic performance of arced labyrinth weirs. Results demonstrate that NN and RF algorithms can be used as a unique expression for curve fitting, although neural networks outperformed random forest when interpolating among the tested geometries.

Shockwaves in spillways with the particle finite element method

María Jesús Samper — Tue, 03 Sep 2019 10:48:00 +0200

Changes in direction and cross section in supercritical hydraulic channels generate shockwaves which result in an increase in flow depth with regard to that for uniform regime. These disturbances are propagated downstream and need to be considered in the design of the chute walls. In dam spillways, where flow rates are often high, this phenomenon can have significant implications for the cost and complexity of the solution. It has been traditionally analysed by means of reduced-scale experimental tests, as it has a clear three-dimensional character and therefore cannot be approached with two-dimensional numerical models. In this work, the ability of the particle finite element method (PFEM) to reproduce this phenomenon is analysed. PFEM has been successfully applied in previous works to problems involving high irregularities in free surface. First, simple test cases available in the technical bibliography were selected to be reproduced with PFEM. Subsequently, the method was applied in two spillways of real dams. The results show that PFEM is capable of capturing the shockwave fronts generated both in the contractions and in the expansions that occur behind the spillway piers. This suggests that the method may be useful as a complement to laboratory test campaigns for the design and hydraulic analysis of dam spillways with complex geometries.

Effect of the integration scheme on the rotation of non-spherical particles with the discrete element method

María Jesús Samper — Fri, 14 Jun 2019 12:54:59 +0200

The discrete element method (DEM) is an emerging tool for the calculation of the behaviour of bulk materials. One of the key features of this method is the explicit integration of the motion equations. Explicit methods are rapid, at the cost of a limited time step to achieve numerical stability. First- or second-order integration schemes based on a Taylor series are frequently used in this framework and shown to be accurate for the translational and rotational motion of spherical particles. However, they may lead to relevant inaccuracies when non-spherical particles are used since the orientation implies a modification in the second-order inertia tensor in the inertial reference frame. Specific integration schemes for non-spherical particles have been proposed in the literature, such as the fourth-order Runge–Kutta scheme presented by Munjiza et al. and the predictor–corrector scheme developed by Zhao and van Wachem which applies the direct multiplication algorithm for integrating the orientation. In this work, both methods are adapted to be used together with a velocity Verlet scheme for the translational integration. The performance of the resulting schemes, as well as that of the direct integration method, is assessed, both in benchmark tests with analytical solution and in real-scale problems. The results suggest that the fourth-order Runge–Kutta and the Zhao and van Wachem schemes are clearly more accurate than the direct integration method without increasing the computational time.

A methodology for the design of dam spillways with wedge shaped blocks based on numerical modeling

Javier San Mauro — Mon, 04 Jun 2018 17:41:06 +0200

Wedge shaped blocks spillways are an innovative solution that allows spilling over the downstream shoulder of earth and rock-fill dams in a safe way. However, they have been barely used as main spillway due to the lack of practical design criteria. An innovative procedure for computer-aided design of wedge shaped blocks spillways is presented in this paper. It includes the design of the drainage and supporting layer considering its seepage capacity and stability. The leakage flow through the joints between blocks is estimated by means of a numerical model calibrated and validated from experimental results. Stability against sliding of the downstream shoulder or the drainage layer is ensured, considering the properties of the granular material selected by the designer and non-linear resistance laws. The procedure will also suggest the shape of a toe protection.

Early detection of anomalies in dam performance: A methodology based on boosted regression trees

María Jesús Samper — Fri, 26 Apr 2019 11:40:14 +0200

The advances in information and communication technologies led to a general trend towards the availability of more detailed information on dam behaviour. This allows applying advanced data‐based algorithms in its analysis, which has been reflected in an increasing interest in the field. However, most of the related literature is limited to the evaluation of model prediction accuracy, whereas the ulterior objective of data analysis is dam safety assessment. In this work, a machine‐learning algorithm (boosted regression trees) is the core of a methodology for early detection of anomalies. It also includes a criterion to determine whether certain discrepancy between predictions and observations is normal, a procedure to compute a realistic estimate of the model accuracy, and an original approach to identify extraordinary load combinations. The performance of causal and noncausal models is assessed in terms of their ability to detect different types of anomalies, which were artificially introduced on reference time series generated with a numerical model of a 100‐m‐high arch dam. The final approach was implemented in an online application to visualise the results in an intuitive way to support decision making.

Numerical modelling of granular materials with spherical discrete particles and the bounded rolling friction model. Application to railway ballast

María Jesús Samper — Wed, 24 Apr 2019 14:16:45 +0200

The Discrete Element Method (DEM) was found to be an effective numerical method for the calculation of engineering problems involving granular materials. However, the representation of irregular particles using the DEM is a very challenging issue, leading to different geometrical approaches. This document presents a new insight in the application of one of those simplifications known as rolling friction, which avoids excessive rotation when irregular shaped materials are simulated as spheric particles. This new approach, called the Bounded Rolling Friction model, was applied to reproduce a ballast resistance test.

Data-Based Models for the Prediction of Dam Behaviour: A Review and Some Methodological Considerations

María Jesús Samper — Wed, 24 Apr 2019 12:12:27 +0200

Predictive models are an important element in dam safety analysis. They provide an estimate of the dam response faced with a given load combination, which can be compared with the actual measurements to draw conclusions about dam safety. In addition to numerical finite element models, statistical models based on monitoring data have been used for decades for this purpose. In particular, the hydrostatic-season-time method is fully implemented in engineering practice, although some limitations have been pointed out. In other fields of science, powerful tools such as neural networks and support vector machines have been developed, which make use of observed data for interpreting complex systems . This paper contains a review of statistical and machine-learning data-based predictive models, which have been applied to dam safety analysis . Some aspects to take into account when developing analysis of this kind, such as the selection of the input variables, its division into training and validation sets, and the error analysis, are discussed. Most of the papers reviewed deal with one specific output variable of a given dam typology and the majority also lack enough validation data. As a consequence, although results are promising, there is a need for further validation and assessment of generalisation capability. Future research should also focus on the development of criteria for data pre-processing and model application.

Air demand estimation in bottom outlets with the particle finite element method. Susqueda Dam case study.

María Jesús Samper — Fri, 11 Jan 2019 14:06:17 +0100

Dam bottom outlets play a vital role in dam operation and safety, as they allow controlling the water surface elevation below the spillway level. For partial openings, water flows under the gate lip at high velocity and drags the air downstream of the gate, which may cause damages due to cavitation and vibration. The convenience of installing air vents in dam bottom outlets is well known by practitioners. The design of this element depends basically on the maximum air flow through the air vent, which in turn is a function of the specific geometry and the boundary conditions. The intrinsic features of this phenomenon makes it hard to analyse either on site or in full scaled experimental facilities. As a consequence, empirical formulas are frequently employed, which offer a conservative estimate of the maximum air flow. In this work, the particle finite element method was used to model the air–water interaction in Susqueda Dam bottom outlet, with different gate openings. Specific enhancements of the formulation were developed to consider air–water interaction. The results were analysed as compared to the conventional design criteria and to information gathered on site during the gate operation tests. This analysis suggests that numerical modelling with the PFEM can be helpful for the design of this kind of hydraulic works.

Numerical modelling of landslide‐generated waves with the particle finite element method (PFEM) and a non‐Newtonian flow model

María Jesús Samper — Thu, 25 Apr 2019 10:52:18 +0200

Landslide‐generated impulse waves may have catastrophic consequences. The physical phenomenon is difficult to model because of the uncertainties in the kinematics of the mobilised material and to the intrinsic complexity of the fluid–soil interaction. The particle finite element method (PFEM) is a numerical scheme that has successfully been applied to fluid–structure interaction problems. It uses a Lagrangian description to model the motion of nodes (particles) in both the fluid and the solid domains (the latter including soil/rock and structures). A mesh connecting the particles (nodes) is re‐generated at every time step, where the governing equations are solved. Various constitutive laws are used for the sliding mass, including rigid solid and the Newtonian and non‐Newtonian fluids. Several examples of application are presented, corresponding both to experimental tests and to actual full‐scale case studies. The results show that the PFEM can be a useful tool for analysing the risks associated with landslide phenomena, providing a good estimate to the potential hazards even for full‐scale events.

Interpretation of dam deformation and leakage with boosted regression trees

María Jesús Samper — Thu, 25 Apr 2019 09:50:09 +0200

Predictive models are essential in dam safety assessment. They have been traditionally based on simple statistical tools such as the hydrostatic-season-time (HST) model. These tools are well known to have limitations in terms of accuracy and reliability. In the recent years, the examples of application of machine learning and related techniques are becoming more frequent as an alternative to HST. While they proved to feature higher flexibility and prediction accuracy, they are also more difficult to interpret. As a consequence, the vast majority of the research is limited to prediction accuracy estimation. In this work, one of the most popular machine learning techniques (boosted regression trees), was applied to model 8 radial displacements and 4 leakage flows at La Baells Dam. The possibilities of model interpretation were explored: the relative influence of each predictor was computed, and the partial dependence plots were obtained. Both results were analysed to draw conclusions on dam response to environmental variables, and its evolution over time. The results show that this technique can efficiently identify dam performance changes with higher flexibility and reliability than simple regression models.

An empirical comparison of machine learning techniques for dam behaviour modelling

María Jesús Samper — Fri, 26 Apr 2019 12:02:45 +0200

Predictive models are essential in dam safety assessment. Both deterministic and statistical models applied in the day-to-day practice have demonstrated to be useful, although they show relevant limitations at the same time. On another note, powerful learning algorithms have been developed in the field of machine learning (ML), which have been applied to solve practical problems. The work aims at testing the prediction capability of some state-of-the-art algorithms to model dam behaviour, in terms of displacements and leakage. Models based on random forests (RF), boosted regression trees (BRT), neural networks (NN), support vector machines (SVM) and multivariate adaptive regression splines (MARS) are fitted to predict 14 target variables. Prediction accuracy is compared with the conventional statistical model, which shows poorer performance on average. BRT models stand out as the most accurate overall, followed by NN and RF. It was also verified that the model fit can be improved by removing the records of the first years of dam functioning from the training set.

Analysis of the discharge capacity of radial-gated spillways using CFD and ANN – Oliana Dam case study

María Jesús Samper — Tue, 19 Nov 2019 13:21:35 +0100

The paper focuses on the analysis of radial-gated spillways, which is carried out by the solution of a numerical model based on the finite element method (FEM). The Oliana Dam is considered as a case study and the discharge capacity is predicted both by the application of a level-set-based free-surface solver and by the use of traditional empirical formulations. The results of the analysis are then used for training an artificial neural network to allow real-time predictions of the discharge in any situation of energy head and gate opening within the operation range of the reservoir. The comparison of the results obtained with the different methods shows that numerical models such as the FEM can be useful as a predictive tool for the analysis of the hydraulic performance of radial-gated spillways.

Possibilities of the particle finite element method for fluid–soil–structure interaction problems

María Jesús Samper — Fri, 21 Dec 2018 11:33:46 +0100

We present some developments in the particle finite element method (PFEM) for analysis of complex coupled problems in mechanics involving fluid–soil–structure interaction (FSSI). The PFEM uses an updated Lagrangian description to model the motion of nodes (particles) in both the fluid and the solid domains (the later including soil/rock and structures). A mesh connects the particles (nodes) defining the discretized domain where the governing equations for each of the constituent materials are solved as in the standard FEM. The stabilization for dealing with an incompressibility continuum is introduced via the finite calculus method. An incremental iterative scheme for the solution of the non linear transient coupled FSSI problem is described. The procedure to model frictional contact conditions and material erosion at fluid–solid and solid–solid interfaces is described. We present several examples of application of the PFEM to solve FSSI problems such as the motion of rocks by water streams, the erosion of a river bed adjacent to a bridge foundation, the stability of breakwaters and constructions sea waves and the study of landslides.