�ԹϺ���

Numerical Framework for Stiffness Characterisation of Liquefied Sands During Post-Shaking Re-solidification

29/05/2026

A new paper has been published in Computers and Geotechnics (Volume 198, 2026).

The paper addresses a long-standing challenge in geotechnical earthquake engineering: accurately characterising the compression stiffness and hydraulic conductivity of liquefied sand deposits during the post-shaking re-solidification phase. Conventional laboratory-derived stiffness parameters are shown to systematically overestimate in-situ stiffness under the partially drained conditions that prevail during this stage, leading to overprediction of excess pore pressure dissipation rates and underestimation of surface settlements.

To address this, the paper introduces a hybrid explicit–implicit finite difference method (HEI-FDM) to solve the nonlinear 1D consolidation equation governing post-shaking pore pressure dissipation and settlement. Building on this forward solver, a joint two-parameter inversion framework is developed to back-calculate the constrained modulus and hydraulic conductivity simultaneously, by matching excess pore pressure time histories at multiple depths alongside surface settlement records from dynamic centrifuge tests.

The framework is validated against four well-documented centrifuge experiments, demonstrating consistently high predictive accuracy. The study also examines the influence of co-seismic settlement on initial hydraulic conductivity and the role of void ratio evolution, identifying which factors must be accounted for and which can be safely simplified in practice.

The paper is available open access .

Congratulations, Dr Ma!

21/05/2026

Shengjie Ma successfully defended his thesis today, titled "Numerical modelling of earthquake-induced liquefaction and its associated ground settlements"! Many congratulations!

(Supervisor), Shengjie Ma (Candidate), (External Examiner) & (Internal Examiner)

Impact of Temporal and Spatial Resolution in Slope-Plant-Atmosphere Interaction Modelling

20/05/2026 

New publication in Geotechnical and Geological Engineering

A new open‑access article has been published in Geotechnical and Geological Engineering: “Impact of Temporal and Spatial Resolution in Slope–Plant–Atmosphere Interaction Modelling”
🔗

The study examines how different representations of Soil–Plant–Atmosphere Interaction (SPAI) influence long‑term predictions of slope behaviour. A well‑instrumented London Clay cut slope is used as a benchmark to evaluate the sensitivity of pore water pressures, displacements, and stability metrics to several modelling choices.

The analysis compares:

• Daily vs. monthly atmospheric boundary conditions, highlighting how temporal aggregation can smooth short‑term extremes.
• Dynamic vs. static vegetation (LAI), showing that a carefully selected static LAI can approximate dynamic vegetation for many engineering outcomes.
• Surface‑only vs. depth‑distributed transpiration, demonstrating notable differences in suction profiles and Factors of Safety when root water uptake is represented with depth and how differences in displacements can be exacerbated under future climate conditions.

Relevance
SPAI processes strongly influence seasonal and long‑term slope response, yet they are often simplified due to data or computational constraints. The findings provide guidance on when simplified approaches remain reliable and when higher‑resolution or more detailed representations become necessary—particularly under climate‑change‑driven extremes.

Key message
Modelling decisions regarding time resolution, vegetation representation, and transpiration depth can meaningfully alter predicted pore pressures, displacements, and stability outcomes. Careful selection of these assumptions is essential for robust geotechnical assessment.

A nonlinear model for strut behaviour in braced excavations

16/05/2026

New publication in Computers & Geotechnics

A new paper, "A nonlinear model for strut behaviour in braced excavations", co-authored with António Pedro (University of Coimbra) and Manuel Matos Fernandes (University of Porto), is now available online in Computers and Geotechnics (Vol. 197, 2026, 108235, doi: ).

The paper addresses a well-known but often poorly handled problem in the numerical modelling of deep excavations: the effective axial stiffness of struts is consistently much lower than their theoretical stiffness, sometimes by as much as 90%, due to assembly imperfections, gaps between structural elements, and initial curvature.

The proposed nonlinear model is governed by a single physically interpretable parameter, δₐ, which controls the offset between the effective and theoretical force–deformation curves and serves as a direct indicator of construction quality. A consistent unloading–reloading response and pre-stressing capability are incorporated within the same framework. The model is implemented in PLAXIS as a and validated against field data from instrumented excavations, as illustrated in the paper.

A straightforward calibration procedure based on monitored force–displacement data is also outlined, enabling application within the Observational Method to improve stage-by-stage predictions and inform installation practice.

The paper is open access and can be read .

A novel adaptive sampling approach with batch selection for the automatic generation of surrogate models in geotechnical engineering

28/01/2026

  🚀 New Publication Alert in Data‑Centric Engineering

📄 “A novel adaptive sampling approach with batch selection for the automatic generation of surrogate models in geotechnical engineering.”
Yunxiang Yang, Agustín Ruiz López, Aikaterini Tsiampousi & David M. G. Taborda
🔗

In this work, we introduce CV‑BASHES, a new adaptive sampling framework with batch selection designed to build surrogate models more efficiently and more accurately—especially for computationally expensive engineering problems.

🌟 What’s new?
✔️ A model‑independent approach using K‑fold cross‑validation to guide sampling
✔️ Novel distance and projection constraints to prevent clustering and maintain space‑filling properties
✔️ Batch selection, enabling scalable training for complex machine‑learning surrogates
✔️ Demonstrated superior performance over state‑of‑the‑art methods such as MaxPro, SAS, and SFCVT

🏗️ Why it matters for geotechnical engineering
We show how CV‑BASHES can automatically and efficiently generate high‑accuracy surrogate models for deep urban excavations, enabling rapid prediction of diaphragm‑wall deformations and streamlining early‑stage design workflows. 

🤖 Impact beyond geotechnics
Although demonstrated through a geotechnical application, the methodology is broadly applicable to any computationally intensive engineering simulations, making surrogate modelling faster, smarter, and more adaptive.

2025 Christmas party

09/12/2025

The annual MAGE Christmas party took place today to celebrate the achievements of 2025 and look ahead at what 2026 is bringing to the group.

From left to right: Ethelbert, Jiaming, David, Shengjie, Yunxiang, Pishun, Pedrag, Francesca, Fabian, Alison, Javier, Kika, Katerina, Hao and Eric

Congratulations to Dr Tantivangphaisal

04/12/2025

Pishun Tantivangphaisal successfully defended his thesis today, titled "Development of a constitutive framework to model the long-term cyclic performance of offshore foundations"! Many congratulations!

 Pishun with one of his supervisors, David, and PhD examiners Prof Burd (Oxford) and Prof Macorini (�ԹϺ���)

Three-Dimensional Interaction of Twin Tunnels: Numerical Analysis of the Waterloo International Terminal Case Study

08/09/2025

🚇 This study, published in the ASCE Journal of Geotechnical and Geoenvironmental Engineering, tackles one of the most challenging questions in urban tunnelling: how do closely spaced tunnels influence each other’s behaviour in three dimensions over time?

Using the Jubilee Line Extension beneath the Northern Line at Waterloo as a benchmark, the following was combined:

• 📊 Extensive field monitoring during excavation and over two years of consolidation
• 🖥 High‑fidelity PLAXIS 3D modelling to replicate real‑world conditions
• 🔍 Detailed investigation of lining stiffness reduction factors, tunnel typology (station vs. running), and hydraulic boundary conditions

The obtained modelling results not only match observed ground movements with high accuracy, but also shed light on the subtle interplay between short‑term excavation effects and long‑term consolidation in London Clay.

Key takeaways:

• Station and running tunnels interact differently due to geometry and stiffness
• Accurate modelling of lining stiffness is critical for predicting settlements
• Long‑term consolidation effects can be as significant as immediate excavation impacts

This work serves as a reference for engineers, researchers, and infrastructure planners tackling complex underground projects in dense urban environments.

📄 Read the full paper by following this

Numerical modelling of the long-term cyclic response of laterally loaded piles driven in sands using the high-cycle accumulation framework

03/07/2025

📘 New Research Published in Géotechnique

A new paper has been published from Pishun Tantivangphaisal's PhD research, titled "Numerical modelling of the long-term cyclic response of laterally loaded piles driven in sands using the high-cycleaccumulation framework". This paper introduces a fully site-specific finite element modelling strategy for predicting the long-term cyclic response of piles driven in sand—central to offshore wind turbine foundation design. By integrating a modified High-Cycle Accumulation (HCA) framework with a calibrated low-cycle sand model, the study achieves accurate simulations of lateral pile behaviour across tens of thousands of load cycles, validated against full-scale field data from the PISA trials at Dunkirk.

🔬 Research Impact:

• This work enables the design of offshore piles under realistic service conditions
• Improves confidence in serviceability and fatigue limit state predictions
• Reduces reliance on generic databases by using exclusively site-specific calibration
• Bridges the gap between element-level soil behaviour and field-scale performance
• Provides a scalable methodology compatible with industry-standard FE tools (e.g., Plaxis)

This work supports more sustainable and cost-effective offshore infrastructure through improved cyclic design methods, informing both academic research and engineering practice.

🔗

A novel machine learning-based approach to thermal integrity profiling of concrete pile foundations

30/06/2025

📘 New Research Publication

A new paper has been published from Javier Sanchez Fernandez's PhD research, titled "A Novel Machine Learning–Based Approach to Thermal Integrity Profiling of Concrete Pile Foundations" in the journal Data-Centric Engineering.

 
📄 Abstract

This study presents a novel methodology for interpreting Thermal Integrity Profiling (TIP) data using machine learning techniques. TIP is a nondestructive testing method that leverages the heat of hydration in concrete to assess the structural integrity of pile foundations. The authors developed a two-stage artificial neural network (ANN) framework trained on a comprehensive dataset generated via finite element simulations.

The proposed model:

1. Classifies piles as defective or non-defective,
2. Predicts defect size and location,
3. Estimates reinforcement cage misalignment, and
4. Back-calculates key concrete hydration parameters.

The approach demonstrated high predictive accuracy and robustness, even under noisy data conditions, and is compatible with standard TIP workflows.

 
🏗️ Significance

This research contributes to the advancement of nondestructive testing in geotechnical engineering by:

• Enhancing the reliability of TIP for defect detection
• Reducing the risk of undetected structural anomalies
• Supporting safer and more efficient foundation construction practices.

🔓 Open Access and Data Availability

The article is freely accessible via Cambridge University Press. The used for model training is available on our .

 
📖 Read the Full Article

👉

Numerical study of the effect of soil-plant-atmosphere interaction under future climate projections and different vegetation covers

14/06/2025

A new MAGE paper has been published from Maryam Maddah Sadatieh's ongoing PhD project on soil-plant-atmosphere interaction. This paper investigates these interactions under future climate projections and different vegetation covers. It combines ecohydrological and geotechnical modelling to assess the impact on slope stability and serviceability. Using the Tethys-Chloris model, the research calculates actual evapotranspiration and integrates it into geotechnical analysis. Results show vegetation increases slope safety by enhancing soil suction. Climate change effects and vegetation dynamics are analysed, revealing complex interactions between climate, soil hydraulic properties, and slope behaviour. The study emphasizes the importance of realistic boundary conditions and field monitoring for accurate geotechnical assessments.

The new paper is available on , where it is Open Access!

Presentation on Numerical Modelling of Long-term Cyclic Loads on Offshore Monopiles

25/03/2025

Pishun Tantivangphaisal delivered a talk to the SUT ECOSIGG and the British Geotechnical Association Early Careers Group. This talk took place at �ԹϺ��� College on 25th March 2025 and focussed on Pishun's experience with developing, implementing and applying tools for modelling the effects of long-term cyclic loading on offshore monopiles. Pishun also gave us some insights into his career in industry and research.

Read Pishun's paper in and watch the recording in our page dedicated to offshore geotechnics research.

Presentations at the DTE-AICOMAS 2025 conference

27/02/2025

Two of our researchers presented their work at the 3rd IACM Digital Twins in Engineering Conference (DTE 2025) & 1st ECCOMAS Artificial Intelligence and Computational Methods in Applied Science (AICOMAS 2025), in Paris.

Javier presented his research on "Integrating machine learning classification with thermal integrity profiling for concrete pile assessment", while Yunxiang provided an update on new sampling methods for surrogate models in his talk "A model-independent adaptive sampling approach for surrogate design in geotechnical engineering".

Implementation of a practical sand constitutive model coupled with the High Cycle Accumulation framework in PLAXIS

28/01/2025

A new MAGE paper has been published from Pishun Tantivangphaisal's ongoing PhD project that details the implementation of the High Cycle Accumulation framework in PLAXIS, coupled with a state parameter-based constitutive model for sands. It proposes changes to the base formulation and shows its application to the challenging problem of calculating the accumulation of deformations in offshore wind turbine foundations when subjected to thousands of loading cycles.

The new paper is available in , where it is Open Access thanks to the support of �ԹϺ��� Open Access Fund. 

2024 Christmas party

16/12/2024

The annual MAGE Christmas party took place today to celebrate the achievements of 2024 and look ahead at what 2025 is bringing to the group.

Back row: Pishun, Fabian, Maryam, Ethelbert, David & Eric
Front row: Javier, Alison, Katerina, Shengjie & Yunxiang
Not present: Agustin, Dariel, Francesca, Jiaming, Julia & Stavroula

We look forward to the next get together, which is scheduled for the week of the Rankine Lecture in mid-March!

A new method for estimating ground movements due to shaft excavation

06/12/2024 

A new MAGE paper has been published that proposes a new method for determining vertical and horizontal ground movements due to the excavation of shafts in London Clay. This paper is available on the with the accepted copy available in .

As part of this paper, we created a spreadsheet that implements the developed expressions, which is available for download in .

First place in blind prediction contest

06/11/2024 

The results of the organised by TU Darmstadt for pile foundations under monotonic and cyclic lateral loading are out! 

The IC MAGE team comprising Pishun Tantivangphaisal, Fabian Wall and David M. G. Taborda from the Geotechnics section at �ԹϺ��� College ranked first in the monotonic prediction, second in the cyclic prediction and took the top prize for the overall top prediction of the event, among 18 teams from 15 different countries around the world! 

The team combined Seequent PLAXIS 3D with material models and advanced numerical tools developed by the IC MAGE research group and will now get the chance to present their prediction at the 5th International Symposium on Frontiers in Offshore Geotechnics in Nantes, France, in June 2025!