Publications
Results
- Showing results for:
- Reset all filters
Search results
-
Journal articleMa S, Taborda DMG, Kontoe S, 2026,
Numerical framework for calculating stiffness of liquefied sands during post-shaking re-solidification
, Computers and Geotechnics, Vol: 198, Pages: 108262-108262, ISSN: 0266-352X -
Journal articleTaborda DMG, Pedro AMG, Matos Fernandes M, 2026,
A nonlinear model for strut behaviour in braced excavations
, Computers and Geotechnics, Vol: 197, ISSN: 0266-352XDeep braced excavations are widely employed to support urban construction, with the axial stiffness of strut systems playing a crucial role in their structural performance. However, field observations have consistently shown that the effective axial stiffness of struts is often significantly lower than theoretical values, frequently attributed to slack arising from imperfections during assembly, gaps between structural elements, and initial curvatures. This paper presents a nonlinear model that captures the deformation-dependent stiffness behaviour arising from those initial imperfections. The model is governed by a single physically interpretable parameter, which controls the offset between effective and theoretical force–deformation curves and serves as an indicator of construction quality. A consistent unloading–reloading response and pre-stressing capability are incorporated within the same framework. The model is applied to a single-propped embedded wall to demonstrate its performance. The results confirm that the range of effective stiffness values reproduced by the model is consistent with instrumented case studies reported in the literature, and that pre-stressing improves average stiffness and reduces wall displacements. A straightforward calibration procedure based on monitored force–displacement data is proposed, enabling application within the Observational Method to improve predictions and inform installation practice.
-
Journal articleSanchez Fernandez J, Ruiz Lopez A, Taborda D,
Data-driven surrogates for predicting thermal performance and response functions of thermo-active piles
, Machine Learning and Data Science in Geotechnics, ISSN: 3029-0422Purpose This paper develops fast, accurate data-driven surrogate models to predict the thermal performance and thermal response functions of single thermo-active piles. These replace computationally expensive finite element simulations and geometry-specific g-function calculations with generalisable machinelearning models suitable for preliminary design and performance assessment. The work aims to improve accessibility, reduce computational cost, and support wider adoption of thermo-active piles within lowcarbon and net-zero infrastructure strategies. Design/methodology/approach Two Artificial Neural Network (ANN) surrogates were trained on databases generated from 3D transient finite element simulations of thermo-active piles. One surrogate predicts transient power output per unit length under a prescribed inlet fluid temperature, while the second predicts normalised pile wall and outlet thermal responses under constant heat flux. Input parameters were sampled using Latin Hypercube sampling. Model training employed feature normalisation, cross-validation, and regularisation, with performance evaluated using standard regression metrics and SHAP-based interpretability analysis. Findings Both surrogate models demonstrate excellent predictive accuracy and strong generalisation. The power output surrogate achieves R² values exceeding 0.99 with mean absolute errors typically below 2 W/m for most of the operational period. The thermal response surrogate reproduces pile wall and outlet gfunctions over ten years with global per-point R² values of 0.994–0.997 and per-timestep averages of 0.971 (wall) and 0.959 (outlet). Validation against a field thermal response test confirms reliable extrapolation beyond the trained diameter range. Computational time is reduced by several orders of magnitude compared with finite element analysis. Originality/value This study presents the first generalisable surrogate framework capable of predicting both power outputper un
-
Journal articleSanchez Fernandez J, Ruiz Lopez A, Taborda D, 2026,
Thermal integrity profiling of concrete piles incorporating three-dimensional defects using deep convolutional neural networks
, Geomechanics and Geoengineering, ISSN: 1748-6025Thermal Integrity Profiling (TIP) has emerged as a powerful non-destructive method for evaluating cast-in-place concrete piles by monitoring heat of hydration (HoH) temperature profiles using temperature probes embedded in the piles. However, conventional TIP interpretation relies either on manual inspection of a series of temperature-depth profiles over time, which is inherently prone to subjective bias, or on back-analysis using simplified modelling approaches. This study presents, for the first time, a fully three-dimensional TIP framework combining Finite Element (FE) simulations with two deep neural networks (NN). A multi-label classifier detects up to three defects per pile, while a regressor quantifies defect volume, location, reinforcement-cage displacement, and hydration parameters. A database with hydration temperature profiles from 19,445 simulations encompassing randomised pile lengths, soil stratigraphies, cage misalignments and multi-defect configurations, is used to train the two NNs. The classifier achieves a per-defect accuracy of 93.6% (96.8% per sample). The regression model achieves near-perfect accuracy for cage displacement (R² = 0.992) and defect volume (mean absolute error of 0.03 m³), and an adjusted R² of 0.92 for defect location. This novel automated approach significantly enhances the objectivity, test speed, and reliability of TIP-based defect detection across a wide range of deep pile foundations.
-
Conference paperLIU J, Tsiampousi K, Ruiz Lopez A, et al., 2026,
A 3D numerical case study of sprayed concrete Lining tunnel construction beneath an existing building
, 21st international conference on soil mechanics and geotechnical engineering, Publisher: ÖGG, Austrian Society for Geomechanics, Pages: 2879-2884The potential impact of tunnelling on the serviceability of existing infrastructures must be carefully considered before construction. 3D Finite Element (FE) analysis has proved to be a useful method when predicting the impact of tunnelling superstructures. Compared with 2D FE analysis, it enables the step-by-step tunnel construction process to be simulated, as well as geometries along the direction of tunnelling to be considered realistically. To minimize ground movements, large diameter SprayedConcrete Lining (SCL) tunnels are commonly constructed through a sequence where a smaller pilot tunnel is excavated first, followed by the enlargement of the pilot tunnel into the full-size tunnel. Such construction process, not typically investigated with 3D FE analysis, is considered in this paper with the simulation of the construction of the Elizabeth Line through a site near Whitechapel station in London, UK. This location provides an interesting case study because the surface settlements of both a greenfield section and of a three-storey building were monitored during construction. The greenfield section was simulated first to gain an understanding of how the construction sequence, various soil properties and tunnel lining stiffness affected the obtained ground movements. The section including the building was investigated subsequently with an emphasis on the evolution of the ground surface settlement with theadvancement of tunnel excavation, as well as on the influence of the pilot-enlargement tunnel construction sequence. The results are in good agreement with the field measurements. Moreover, it is highlighted that during enlargement, bending of the existing pilot tunnel lining in sections yet to be enlarged occurs, which in turn affects how the ground surface settlement develops. This investigation demonstrates that the interaction mechanisms between pilot and enlargement tunnels are complex and that considering the detailed construction sequence is essential to ca
-
Conference paperGhalandari T, Vuye C, Taborda D, 2026,
Data-driven surrogate modeling for thermo-active road design
, Proceedings of the 21st ICSMGE, Publisher: ÖGG, Austrian Society for Geomechanics, Pages: 2077-2080The traditional iterative design process for engineering problems involves evaluating the performance of initial solutions based on prior experience and expertise, comparing them against specified project criteria, and refining design variables to achieve the desired system performance. This cycle continues until a satisfactory solution is reached. However, for more complex problems, the computational demands of simulation modelling can become prohibitively high, especially in design optimization tasks. Surrogate models are machine learning-based representations of complex physics-driven simulations, enabling significantly faster computations. They provide a rapid and accurate means to evaluate responses across multiple design scenarios, making them invaluable for efficient design exploration and optimization. This paper investigates the application of surrogate modelling for the design and thermal performance assessment of thermo-active roads, with a particular focus on systems that integrate a heat exchange layer with embedded pipes in the asphalt pavement, known as Pavement Solar Collectors (PSCs). The surrogate model developed in this study is based on an artificial neural network (ANN) trained on a comprehensive database generated from finite element simulations. These simulations account for variations in geometric configurations and thermophysical properties of materials. The ANN model is designed to predict the outlet water temperature of PSC systems, which is then used to calculate their heat harvesting capacity. Following hyperparameter optimization to enhance the performance of the surrogate model, the proposed framework demonstrated its effectiveness in optimizing the design of a PSC system in a case study. This highlights the model's potential to simplify and improve the efficiency of the design process for thermo-active road systems.
-
Conference paperSanchez Fernandez J, Taborda DMG, Ruiz Lopez A, 2026,
Application of machine learning algorithms for power output prediction in thermo-active diaphragm walls
, 21st International Conference on Soil Mechanics and Geotechnical Engineering (ICSMGE), Publisher: ÖGG, Austrian Society for Geomechanics, Pages: 6033-6036Thermo-active diaphragm walls offer an efficient route to low carbon heating and cooling by embedding heat-exchanging loops within deep retaining structures. In this study, a surrogate model to predict the power output per unit area of a diaphragm wall section is presented and evaluated. A parametrised 3D finite element (FE) model was created in COMSOL Multiphysics and used to assess 500 combinations of wall width and depth, concrete and soil conductivities, inlet–ground temperature differential, convective heat transfer coefficient at the exposed boundary, and fluid velocity. This was used as a numerical database for the training and testing of the surrogate model. The dataset covers 16 time instants retrieved from the transient analysis, in addition to the steady-state solution. The artificial neural network (ANN) regressor subsequently trained achieved an average coefficient of determination (R²) of 0.987. The average error in power prediction is limited to a few W/m², with the model being able to capture the correct trends even in the less accurate samples. Feature importance was assessed using SHAP analysis. This revealed that, as expected, early-time performance is dominated by the temperature differential and concrete conductivity, while the convective heat transfer coefficient at the exposed boundary governs long-term output. By replacing computationally expensive FE runs with fast ANN predictions, the proposed surrogate enables rapid design optimisation, sensitivity studies, and feasibility assessments for a large set of configurations.
-
Conference paperTantivangphaisal P, Taborda D, Kontoe S, 2026,
Development of a lifetime cyclic design method for offshore foundations
, Proceedings of the 21st ICSMGE -
Conference paperLivanidou A-C, Georgiadis K, Chaloulos YK, et al., 2026,
Numerical analysis of helical pile foundations for offshore wind turbines
, Proceedings of the 21st ICSMGE -
Conference paperYang Y, Taborda DMG, Tsiampousi A, et al., 2026,
Back analysis of the Old Oak Common box using surrogate models
, 21st International Conference on Soil Mechanics and Geotechnical Engineering (ICSMGE), Publisher: ÖGG, Austrian Society for Geomechanic, Pages: 3115-3118The observational method (OM) has been adopted in urban excavation projects to help manage uncertainties associated with design soil parameters. By deploying comprehensive ground and structural monitoring schemes, the OM can provide a continuous assessment of the progress of the project, with deviations between observed and predicted responses potentially leading to an interruption of construction operations and/or a review of the original design. As part of this process, the OM requires the frequent back analysis of soil parameters using field monitoring data, with the obtained updated soil parameters being subsequently used to simulate future construction stages. This enables the implementation of remedial measures to be evaluated, as well as potential cost saving options to be explored. In this context, the numerical model must reliably capture the actual soil behaviour while remaining computationally efficient, as the back analysis process typically requires many simulations to identify optimal parameter sets. However, such high computational costs are generally incompatible with the constraints imposed by construction timelines and costconsiderations, thereby restricting the feasibility of employing high-fidelity numerical models. To address these challenges, this paper utilises an Artificial Neural Network (ANN) as a surrogate model, capable of predicting results at different construction stagescomparable to those generated by high-fidelity numerical simulations with significantly reduced computational effort, in conjunction with Genetic Algorithms (GA) for automatic back analysis of soil parameters based on field data. The proposed approach for automatic back-analysis is demonstrated through a retrospective case study of the Old Oak Common station in west London (United Kingdom) which is part of the High Speed 2 (HS2) project, highlighting its ability to identify model parameters based on field measurements.
-
Conference paperMA S, Taborda DMG, Kontoe S, 2026,
Effect of the adopted constitutive relationship on the modelling of post-liquefaction reconsolidation
, Proceedings of the 21st International Conference on Soil Mechanics and Geotechnical Engineering, Vienna, Publisher: ÖGG, Austrian Society for GeomechanicsGround settlement is one of the main consequences of liquefaction and can develop during seismic shaking, as well as for a period after the end of the strong motion. In the post-shaking phase, once the seismic shaking ceases, water continues to flow upwards towards the surface, with ground settlement evolving until the excess pore water pressure is completely dissipated within the soil deposit. This process, known as post-liquefaction reconsolidation or simply reconsolidation, results in volumetric strains and settlements, which are challenging to model numerically using conventional constitutive frameworks for sands as loading under constant stress-ratio is often not captured accurately. In this study, a modified bounding surface plasticity model employing fourdifferent elastic stiffness constitutive relationships is proposed to improve the accuracy of simulating post-liquefaction reconsolidation.Fully coupled dynamic consolidation finite element analyses of level ground deposits subjected to seismic loading are conducted,illustrating the impact of the new formulation in terms of excess pore pressure dissipation and ground settlement evolution.
-
Conference paperPalmieri F, Taborda D, 2026,
Energy dissipation of clays during undrained cyclic triaxial tests
, Proceedings of the 21st ICSMGE -
Journal articleKontoe S, Pedone G, Bellumat E, et al., 2026,
Finite element analysis of laterally loaded open-ended steel piles driven in chalk
, Geomechanics for Energy and the Environment, Vol: 45, ISSN: 2352-3808Open-ended steel piles are commonly driven to support offshore wind energy structures. Their design poses significant challenges in chalk, a very weak brittle limestone found in several regions worldwide. Impact driving causes chalk de-structuration and fracturing around the piles, greatly affecting their lateral load-bearing performance. This was observed in recent field tests undertaken in the UK on piles with different lengths, diameters and thicknesses, exhibiting both geotechnical and structural failures. Most of these lateral loading tests, including those conducted on larger and longer monopiles, were completed recently and were never analysed numerically. This paper presents results of 3D Finite Element analyses conducted on open-ended steel piles with different diameters (up to 1.22 m), embedded lengths (up to 10.16 m) and wall thicknesses (up to 44.5 mm), allowing to explore the marked scale effects observed on site. The newly available field tests also showed that steel yielding can occur before geotechnical failure is reached in chalk when testing piles with practical dimensions. However, steel yielding is usually neglected when modelling soil-pile interaction in geotechnical applications. The paper also aims at covering this gap by introducing a simplified modelling approach to account for elasto-plastic pile behaviour. The analyses delivered generally good matches with field behaviour and allowed to explore the main geotechnical uncertainties affecting accurate pile-chalk interaction predictions, mainly including the extent of the chalk fracturing induced by pile driving and its impact on chalk mechanical properties. The studies provide new and vital guidance for those involved in designing large driven piles for chalk sites.
-
Journal articleSheil B, Anagnostopoulos C, Buckley R, et al., 2026,
Artificial intelligence transformations in geotechnics: progress, challenges and future enablers
, Computers and Geotechnics, Vol: 189, ISSN: 0266-352XOur reliance on the underground space to deliver critical civil engineering infrastructure is growing: to accommodate utility and transport infrastructure in urban environments, to provide innovative housing and commercial solutions, and to support proliferating renewable energy infrastructure, particularly offshore. Artificial intelligence (AI) is arguably the most promising enabler to transform geotechnical engineering by extracting knowledge from data to achieve step-change increases in efficiency, sustainability, reliability and safety. This paper seeks to develop a shared understanding of the state of the art of AI in geotechnics and to explore future developments. By way of example, specific popular use cases in geotechnics are considered to highlight current progress in AI applications including intelligent site investigation, predictive modelling for soil behaviour, and optimisation of design and construction processes. The paper then addresses key research challenges, such as data scarcity and interpretability, and discusses the opportunities that lie ahead in the integration of AI with geotechnical engineering. Finally, priority technological enablers are identified for future transformations.
-
Journal articleYang Y, Ruiz Lopez A, Tsiampousi K, et al., 2026,
A novel adaptive sampling approach with batch selection for the automatic generation of surrogate models in geotechnical engineering
, Data-Centric Engineering, Vol: 7, ISSN: 2632-6736Surrogate models have gained widespread popularity for their effectiveness in replacing computationally expensive numerical analyses, particularly in scenarios, such as design optimisation procedures, requiring hundreds or thousands of simulations. While one-shot sampling methods – where all samples are generated in a single stage without prior knowledge of the required sample size – are commonly adopted in the creation of surrogate models, these methods face significant limitations. Given that the characteristics of the underlying system are generally unknown prior to training, the adoption of one-shot sampling can lead to suboptimal model performance or unnecessary computational costs, especially in complex or high-dimensional problems. This paper addresses these challenges by proposing a novel, model-independent adaptive sampling approach with batch selection, termed CV-BASHES (Cross-Validation Batch Adaptive Sampling for High Efficiency Surrogates). CV-BASHES is first validated using two analytical functions to explore its flexibility and accuracy under different configurations, confirming its robustness. Comparative studies on the same functions with two state-of-the-art methods, Maxpro and SAS, demonstrate the superior accuracy and robustness of CV-BASHES. Its applicability is further demonstrated through a geotechnical application, where CV-BASHES is used to develop a surrogate model to predict the horizontal deformation of a diaphragm wall supporting a deep excavation. Results show that CV-BASHES efficiently selects training samples, reducing 29 the dataset size while maintaining high surrogate accuracy. By offering more efficient sampling strategies, CV-BASHES streamlines and enhances the process of creating machine learning models as surrogates for tackling complex problems in general engineering disciplines
This data is extracted from the Web of Science and reproduced under a licence from Thomson Reuters. You may not copy or re-distribute this data in whole or in part without the written consent of the Science business of Thomson Reuters.