Computers and Geotechnics最新文献_第3页

Numerical investigation of pore pressure evolution mechanisms in saturated sand during liquefaction using coupled discrete element method 用耦合离散元法研究饱和砂土液化过程中孔隙压力演化机制

IF 6.2 1区工程技术 Q1 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Computers and Geotechnics

Pub Date : 2026-05-01 Epub Date: 2026-02-06 DOI: 10.1016/j.compgeo.2026.107961

Zhuolin Su , Jialin Xu , Chengshun Xu , Kemin Jia , Chunyi Cui

This study develops a three-dimensional fluid-particle coupling numerical model based on the discrete element method (DEM), incorporating point cloud volume sampling technology to achieve high-precision dynamic calculation of particle porosity. The model comprehensively considers the coupling effects of pore structure evolution on pore water pressure fields, establishing governing equations that couple porosity-change-induced (PI) and diffusion-induced (DI) pressurization/depressurization mechanisms. The accuracy of the proposed method is validated through three classical benchmark problems: Terzaghi’s one-dimensional consolidation, undrained triaxial tests, and the Mandel-Cryer effect. Using this approach, the complete process from liquefaction instability to reconsolidation densification in saturated loose sand is successfully simulated, accurately reproducing key liquefaction phenomena including excess pore water pressure accumulation and dissipation as well as microscopic pore structure reorganization. The study achieves quantitative separation of the relative contributions of PI and DI mechanisms during liquefaction, revealing that they synergistically constitute the fundamental control system of the entire process: the liquefaction triggering stage is primarily dominated by the PI mechanism, the development stage shows gradually increasing influence of the DI mechanism, and the reconsolidation stage is entirely controlled by the DI mechanism. This numerical framework provides a powerful tool for in-depth understanding of the fundamental physical mechanisms of soil liquefaction, offering significant theoretical and practical value for prediction and risk assessment of seismic liquefaction hazards.

本研究基于离散元法（DEM）建立了三维流体-颗粒耦合数值模型，结合点云体积采样技术实现了颗粒孔隙度的高精度动态计算。该模型综合考虑了孔隙结构演化对孔隙水压力场的耦合效应，建立了耦合孔隙变化诱导（PI）和扩散诱导（DI）增压/降压机制的控制方程。通过Terzaghi一维固结、不排水三轴试验和mandelr - cryer效应三个经典基准问题验证了该方法的准确性。利用该方法，成功模拟了饱和松散砂从液化失稳到再固结致密化的全过程，准确再现了超孔隙水压力积累消散以及微观孔隙结构重组等关键液化现象。定量分离了PI机制和DI机制在液化过程中的相对贡献，揭示了它们协同构成了整个过程的基本控制系统：液化触发阶段主要由PI机制主导，发展阶段DI机制的影响逐渐增强，再固结阶段完全由DI机制控制。该数值框架为深入认识土壤液化的基本物理机制提供了有力的工具，为地震液化灾害的预测和风险评估提供了重要的理论和实用价值。

{"title":"Numerical investigation of pore pressure evolution mechanisms in saturated sand during liquefaction using coupled discrete element method","authors":"Zhuolin Su , Jialin Xu , Chengshun Xu , Kemin Jia , Chunyi Cui","doi":"10.1016/j.compgeo.2026.107961","DOIUrl":"10.1016/j.compgeo.2026.107961","url":null,"abstract":"<div><div>This study develops a three-dimensional fluid-particle coupling numerical model based on the discrete element method (DEM), incorporating point cloud volume sampling technology to achieve high-precision dynamic calculation of particle porosity. The model comprehensively considers the coupling effects of pore structure evolution on pore water pressure fields, establishing governing equations that couple porosity-change-induced (PI) and diffusion-induced (DI) pressurization/depressurization mechanisms. The accuracy of the proposed method is validated through three classical benchmark problems: Terzaghi’s one-dimensional consolidation, undrained triaxial tests, and the Mandel-Cryer effect. Using this approach, the complete process from liquefaction instability to reconsolidation densification in saturated loose sand is successfully simulated, accurately reproducing key liquefaction phenomena including excess pore water pressure accumulation and dissipation as well as microscopic pore structure reorganization. The study achieves quantitative separation of the relative contributions of PI and DI mechanisms during liquefaction, revealing that they synergistically constitute the fundamental control system of the entire process: the liquefaction triggering stage is primarily dominated by the PI mechanism, the development stage shows gradually increasing influence of the DI mechanism, and the reconsolidation stage is entirely controlled by the DI mechanism. This numerical framework provides a powerful tool for in-depth understanding of the fundamental physical mechanisms of soil liquefaction, offering significant theoretical and practical value for prediction and risk assessment of seismic liquefaction hazards.</div></div>","PeriodicalId":55217,"journal":{"name":"Computers and Geotechnics","volume":"193 ","pages":"Article 107961"},"PeriodicalIF":6.2,"publicationDate":"2026-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146174293","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Evaluating statistical learning approaches to predict suction bucket displacement due to vertical cyclic loading in sand 评估统计学习方法预测吸力桶在垂直循环加载下的位移

IF 6.2 1区工程技术 Q1 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Computers and Geotechnics

Pub Date : 2026-05-01 Epub Date: 2026-02-13 DOI: 10.1016/j.compgeo.2026.107970

Francisco da Silva Pereira , Conleth D. O’Loughlin , Britta Bienen , Bruno Stuyts

Bottom-fixed offshore wind turbines (OWTs) are sensitive to tilting, and therefore strict out of verticality limits apply to the foundation serviceability limit state design. In the case of OWTs supported by suction bucket jackets (SBJs), tilting of the turbine arises from differential displacement between the windward and leeward suction buckets, such that predicting the foundation displacement due to vertical cyclic loading is critical. Assessing displacement accumulation due to cyclic loading for all the suction buckets in an offshore wind farm would require significant time, and detailed soil and loading information may not be available at the early stages of the design. Statistical learning models can map complex non-linear interactions between features and target variables by performing regression techniques in a given dataset (training data). Once the patterns in the training dataset have been learned, predictions can be performed orders of magnitude quicker than numerical models, making them well suited to design practice, particularly during the feasibility stage of a design. This paper investigates the performance of three non-linear statistical learning models (General Additive Model, Random Forest and eXtreme Gradient Boosting) in predicting the accumulated displacement of suction buckets due to vertical cyclic loading. The research data are taken from centrifuge model tests that feature over 80,000 load cycles with varying mean stress, stress amplitude and drainage conditions. The model performance was assessed using statistical metrics (coefficient of determination and mean squared error) and by comparing the measured and calculated displacements for storm loading, with the ensemble models providing encouraging results. The prediction making process of the best performing model (eXtreme Gradient Boosting) was investigated using a game theory approach (SHappley Additive exPlanations) and was shown to be consistent with current engineering knowledge. Notably, the best performing model was able to capture the effects of changing stress amplitude during storm loading, offering a more realistic representation than the cyclic load magnitude ordering approach that is typically adopted in engineering practice.

海底固定式海上风力发电机对倾斜比较敏感，因此在基础使用极限状态设计中需要严格的超垂直度限制。对于由吸力桶夹套（sbj）支撑的水轮机，由于迎风和下风吸力桶之间的位移差异，导致水轮机倾斜，因此预测垂直循环荷载引起的基础位移是至关重要的。评估海上风电场中所有吸风桶的循环加载引起的位移积累需要大量时间，并且在设计的早期阶段可能无法获得详细的土壤和加载信息。统计学习模型可以通过在给定的数据集（训练数据）中执行回归技术来映射特征和目标变量之间复杂的非线性相互作用。一旦学习了训练数据集中的模式，预测可以比数值模型快几个数量级，使它们非常适合设计实践，特别是在设计的可行性阶段。本文研究了三种非线性统计学习模型（通用加性模型、随机森林模型和极端梯度增强模型）在预测垂直循环荷载作用下吸油斗累积位移中的性能。研究数据来自离心机模型试验，该试验具有超过80,000个荷载循环，具有不同的平均应力，应力幅值和排水条件。模型的性能通过统计指标（决定系数和均方误差）进行评估，并通过比较风暴荷载的测量和计算位移，与集合模型提供了令人鼓舞的结果。使用博弈论方法（SHappley Additive exPlanations）研究了表现最佳的模型（eXtreme Gradient Boosting）的预测过程，并证明与当前的工程知识一致。值得注意的是，表现最好的模型能够捕捉风暴荷载期间应力幅值变化的影响，比工程实践中通常采用的循环荷载数量级排序方法提供更真实的表示。

{"title":"Evaluating statistical learning approaches to predict suction bucket displacement due to vertical cyclic loading in sand","authors":"Francisco da Silva Pereira , Conleth D. O’Loughlin , Britta Bienen , Bruno Stuyts","doi":"10.1016/j.compgeo.2026.107970","DOIUrl":"10.1016/j.compgeo.2026.107970","url":null,"abstract":"<div><div>Bottom-fixed offshore wind turbines (OWTs) are sensitive to tilting, and therefore strict out of verticality limits apply to the foundation serviceability limit state design. In the case of OWTs supported by suction bucket jackets (SBJs), tilting of the turbine arises from differential displacement between the windward and leeward suction buckets, such that predicting the foundation displacement due to vertical cyclic loading is critical. Assessing displacement accumulation due to cyclic loading for all the suction buckets in an offshore wind farm would require significant time, and detailed soil and loading information may not be available at the early stages of the design. Statistical learning models can map complex non-linear interactions between features and target variables by performing regression techniques in a given dataset (training data). Once the patterns in the training dataset have been learned, predictions can be performed orders of magnitude quicker than numerical models, making them well suited to design practice, particularly during the feasibility stage of a design. This paper investigates the performance of three non-linear statistical learning models (General Additive Model, Random Forest and eXtreme Gradient Boosting) in predicting the accumulated displacement of suction buckets due to vertical cyclic loading. The research data are taken from centrifuge model tests that feature over 80,000 load cycles with varying mean stress, stress amplitude and drainage conditions. The model performance was assessed using statistical metrics (coefficient of determination and mean squared error) and by comparing the measured and calculated displacements for storm loading, with the ensemble models providing encouraging results. The prediction making process of the best performing model (eXtreme Gradient Boosting) was investigated using a game theory approach (SHappley Additive exPlanations) and was shown to be consistent with current engineering knowledge. Notably, the best performing model was able to capture the effects of changing stress amplitude during storm loading, offering a more realistic representation than the cyclic load magnitude ordering approach that is typically adopted in engineering practice.</div></div>","PeriodicalId":55217,"journal":{"name":"Computers and Geotechnics","volume":"193 ","pages":"Article 107970"},"PeriodicalIF":6.2,"publicationDate":"2026-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146174288","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

A unified contact model incorporating surface abrasion for DEM simulation of granular soil behavior across small to large strains 结合表面磨损的统一接触模型，用于颗粒土小应变到大应变的DEM模拟

IF 6.2 1区工程技术 Q1 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Computers and Geotechnics

Pub Date : 2026-05-01 Epub Date: 2026-02-05 DOI: 10.1016/j.compgeo.2026.107965

Jiuyang Zhou , Xiaoqiang Gu , Hongwei Wu , Jing Hu

The discrete element method (DEM) is a valuable tool for simulating the mechanical behavior of granular soils at both large and small-to-medium strains. However, the use of different contact models for varying strain levels has led to inconsistencies in previous simulations. This study highlights that while factors such as particle shape and fragmentation are critical, the omission of surface degradation mechanisms in current models may significantly contribute to their unsatisfactory performance in conventional undrained simulations. To address this, a unified contact model is proposed to account for the influence of surface degradation (specifically abrasion) on the soil behaviors across small to large strains. The proposed model is formulated such that the normal force–displacement law transitions naturally from Hertzian contact mechanics at small strains to a linear-bounded constitutive relation at large strains, capturing the accumulation of surface degradation as a primary driver of the soil skeleton’s increased compressibility. Simulations confirm that the proposed model effectively captures the macroscopic mechanical behavior of granular soils at the considered strain levels, showing good agreement with experimental data. In addition, the comparative analyses of macroscopic and microscopic properties are revealing that small strain modulus G₀ and macro state parameter ψ_e0 can be well correlated with mechanical coordination number MCN and micro state parameter ψ_MCN0, respectively.

离散元法（DEM）是模拟颗粒土大应变和中小应变力学行为的一种有价值的工具。然而，对于不同应变水平使用不同的接触模型导致了先前模拟的不一致。该研究强调，虽然颗粒形状和破碎等因素至关重要，但当前模型中遗漏的表面降解机制可能会显著导致它们在常规不排水模拟中表现不理想。为了解决这个问题，提出了一个统一的接触模型，以考虑表面退化（特别是磨损）对小到大应变土壤行为的影响。所提出的模型是这样制定的：法向力-位移定律在小应变下自然地从赫兹接触力学过渡到大应变下的线性有界本构关系，捕捉表面退化的积累作为土壤骨架压缩性增加的主要驱动因素。仿真结果表明，该模型有效地反映了考虑应变水平下颗粒土的宏观力学行为，与实验数据吻合较好。此外，宏观和微观性能对比分析表明，小应变模量G0和宏观状态参数ψe0分别与力学配位数MCN和微观状态参数ψMCN0具有良好的相关性。

{"title":"A unified contact model incorporating surface abrasion for DEM simulation of granular soil behavior across small to large strains","authors":"Jiuyang Zhou , Xiaoqiang Gu , Hongwei Wu , Jing Hu","doi":"10.1016/j.compgeo.2026.107965","DOIUrl":"10.1016/j.compgeo.2026.107965","url":null,"abstract":"<div><div>The discrete element method (DEM) is a valuable tool for simulating the mechanical behavior of granular soils at both large and small-to-medium strains. However, the use of different contact models for varying strain levels has led to inconsistencies in previous simulations. This study highlights that while factors such as particle shape and fragmentation are critical, the omission of surface degradation mechanisms in current models may significantly contribute to their unsatisfactory performance in conventional undrained simulations. To address this, a unified contact model is proposed to account for the influence of surface degradation (specifically abrasion) on the soil behaviors across small to large strains. The proposed model is formulated such that the normal force–displacement law transitions naturally from Hertzian contact mechanics at small strains to a linear-bounded constitutive relation at large strains, capturing the accumulation of surface degradation as a primary driver of the soil skeleton’s increased compressibility. Simulations confirm that the proposed model effectively captures the macroscopic mechanical behavior of granular soils at the considered strain levels, showing good agreement with experimental data. In addition, the comparative analyses of macroscopic and microscopic properties are revealing that small strain modulus <em>G</em><sub>0</sub> and macro state parameter <em>ψ</em><sub>e0</sub> can be well correlated with mechanical coordination number <em>MCN</em> and micro state parameter <em>ψ</em><sub>MCN0</sub>, respectively.</div></div>","PeriodicalId":55217,"journal":{"name":"Computers and Geotechnics","volume":"193 ","pages":"Article 107965"},"PeriodicalIF":6.2,"publicationDate":"2026-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146174383","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Prediction of rock mass energy evolution during deep tunnel construction using static temporal fusion transformer and numerical surrogate model 基于静态时间融合变压器和数值代理模型的深部隧道施工岩体能量演化预测

IF 6.2 1区工程技术 Q1 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Computers and Geotechnics

Pub Date : 2026-05-01 Epub Date: 2026-01-31 DOI: 10.1016/j.compgeo.2026.107930

Yaoru Liu , Songyu Yue , Rujiu Zhang , Yuequn Huang , Muwu Xie , Qiang Yang

During TBM tunneling, timely and effective prediction of energy evolution of surrounding rock is critical for forecasting potential hazards like rockburst, serving as a fundamental safeguard for deep underground construction. So far, most researchers often underestimate the importance of rapid prediction of the energy evolution of tunnel surrounding rock, resulting in the inability to predict specific information such as the location and time of rock bursts. In this study, a surrogate model for predicting the evolution of energy dissipation rate of tunnel surrounding rock based on the static TFT model is proposed to achieve fast time series prediction. Building upon the Temporal Fusion Transformer (TFT) framework, the static TFT model which considers the time invariant nature of tunnel surrounding rock data is proposed. 4373 numerical samples containing 9 surrounding rock energy influencing factors and 12 output features are established and trained on the model guided by the proposed mixed data and physical loss function. The model’s performance is evaluated through sample size impact, and ablation feature experiments, as well as comparing the predictive accuracy and fitting effectiveness with baseline models. It is found that the proposed model achieves superior performance across all metrics in predicting surrounding rock energy evolution without redundant features. Specifically, it attains an MAE of

0.0447 J \cdot m^{- 3} \cdot s^{- 1}

, an R² of 0.9201, and an MSE of

0.0148 J^{2} \cdot m^{- 6} \cdot s^{- 2}

for energy dissipation rate prediction. These outcomes signify a substantive advancement in rapid energy evolution forecasting for tunnel surrounding rock and provide an early-warning basis for related geohazards.

在TBM掘进过程中，及时有效地预测围岩能量演化是预测岩爆等潜在灾害的关键，是深埋地下施工的根本保障。迄今为止，大多数研究人员往往低估了快速预测隧道围岩能量演化的重要性，导致无法预测地压发生的具体位置和时间等信息。为了实现快速的时间序列预测，在静态TFT模型的基础上，提出了一种预测隧道围岩能量耗散率演化的替代模型。在时间融合变压器（TFT）框架的基础上，提出了考虑隧道围岩数据时不变特性的静态TFT模型。在提出的混合数据和物理损失函数指导下，建立并训练了包含9个围岩能量影响因素和12个输出特征的4373个数值样本。通过样本量影响和烧蚀特征实验，以及与基线模型的预测精度和拟合效果进行比较，对模型的性能进行了评估。结果表明，该模型在预测围岩能量演化过程中没有冗余特征，在所有指标上都取得了较好的效果。其中，能量耗散率预测的MAE为0.0447J·m-3·s-1， R2为0.9201，MSE为0.0148 j·m-6·s-2。这些结果标志着隧道围岩能量快速演化预测取得了实质性进展，为相关地质灾害预警提供了依据。

{"title":"Prediction of rock mass energy evolution during deep tunnel construction using static temporal fusion transformer and numerical surrogate model","authors":"Yaoru Liu , Songyu Yue , Rujiu Zhang , Yuequn Huang , Muwu Xie , Qiang Yang","doi":"10.1016/j.compgeo.2026.107930","DOIUrl":"10.1016/j.compgeo.2026.107930","url":null,"abstract":"<div><div>During TBM tunneling, timely and effective prediction of energy evolution of surrounding rock is critical for forecasting potential hazards like rockburst, serving as a fundamental safeguard for deep underground construction. So far, most researchers often underestimate the importance of rapid prediction of the energy evolution of tunnel surrounding rock, resulting in the inability to predict specific information such as the location and time of rock bursts. In this study, a surrogate model for predicting the evolution of energy dissipation rate of tunnel surrounding rock based on the static TFT model is proposed to achieve fast time series prediction. Building upon the Temporal Fusion Transformer (TFT) framework, the static TFT model which considers the time invariant nature of tunnel surrounding rock data is proposed. 4373 numerical samples containing 9 surrounding rock energy influencing factors and 12 output features are established and trained on the model guided by the proposed mixed data and physical loss function. The model’s performance is evaluated through sample size impact, and ablation feature experiments, as well as comparing the predictive accuracy and fitting effectiveness with baseline models. It is found that the proposed model achieves superior performance across all metrics in predicting surrounding rock energy evolution without redundant features. Specifically, it attains an <em>MAE</em> of <span><math><mrow><mn>0.0447</mn><mi>J</mi><mo>·</mo><msup><mi>m</mi><mrow><mo>-</mo><mn>3</mn></mrow></msup><mo>·</mo><msup><mi>s</mi><mrow><mo>-</mo><mn>1</mn></mrow></msup></mrow></math></span>, an <em>R</em><sup>2</sup> of 0.9201, and an <em>MSE</em> of <span><math><mrow><mn>0.0148</mn><msup><mi>J</mi><mn>2</mn></msup><mo>·</mo><msup><mi>m</mi><mrow><mo>-</mo><mn>6</mn></mrow></msup><mo>·</mo><msup><mi>s</mi><mrow><mo>-</mo><mn>2</mn></mrow></msup></mrow></math></span> for energy dissipation rate prediction. These outcomes signify a substantive advancement in rapid energy evolution forecasting for tunnel surrounding rock and provide an early-warning basis for related geohazards.</div></div>","PeriodicalId":55217,"journal":{"name":"Computers and Geotechnics","volume":"193 ","pages":"Article 107930"},"PeriodicalIF":6.2,"publicationDate":"2026-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146174387","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Explainable machine learning and generative diffusion modeling for improved susceptibility mapping of rainfall-induced clustered landslides: A case study from Wuping County, southeastern China 基于可解释性机器学习和生成扩散模型的降雨诱发集群式滑坡易感性制图：以中国东南部武平县为例

IF 6.2 1区工程技术 Q1 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Computers and Geotechnics

Pub Date : 2026-05-01 Epub Date: 2026-02-18 DOI: 10.1016/j.compgeo.2026.107989

Yu Huang , Yinke Li , Lisha Hei , Jialing Zou , Dingyu Chen

Rainfall-induced clustered landslides have become increasingly frequent in southeastern China, characterized by strong multi-factor coupling and posing significant threats to local communities and infrastructure. Taking Wuping County, Fujian Province, as a representative case, this study established a post-event inventory of 6,005 shallow landslides triggered by the extreme rainfall on 15–16 June 2024. An equal number of non-landslide samples were generated, and eleven topographic, geological, hydrological, vegetation, and anthropogenic factors were compiled into a 12.5 m-resolution dataset (training/testing = 7:3). Based on station-observed rainfall data, a structural rainfall analysis revealed that the landslide clustering was jointly triggered by the dual mechanisms of “antecedent rainfall accumulation” and “short-duration high-intensity pulses.” A comprehensive factor quality assessment was performed, including multicollinearity analysis (VIF < 5, TOL > 0.1) and Pearson correlation screening, confirming the independence and reliability of the conditioning factors prior to modeling. Six models—SVC-GridSearch, SVC-Bayes, SVC-GWO, SVC-PSO, Random Forest, and XGBoost—were then developed and compared, with SHAP analysis used to enhance interpretability and cross-validate with IGR results. The XGBoost model achieved the best performance on the test set (AUC ≈ 0.915). To address class boundary ambiguity, a Denoising Diffusion Probabilistic Model (DDPM) was further introduced for controlled data augmentation in the 11-dimensional factor space, generating about 12% of targeted samples within the model’s “confusion zone” (predicted probability 0.45–0.55). After augmentation, the XGBoost AUC increased to ≈ 0.931, with a significant DeLong test result (p < 0.01), improved sensitivity, and narrower confidence intervals. The proposed hybrid framework of explainable machine learning and generative probabilistic modeling effectively enhances susceptibility mapping accuracy under limited-sample conditions and provides technical support for risk assessment, emergency control, and mitigation planning in southeastern mountainous regions.

降雨引发的群体性滑坡在中国东南部地区日益频繁，具有强多因素耦合的特点，对当地社区和基础设施构成重大威胁。以福建省武平县为例，建立了2024年6月15-16日极端降雨引发的6005个浅层滑坡的事后清查。生成等量的非滑坡样本，并将11个地形、地质、水文、植被和人为因素汇编成12.5 m分辨率的数据集（训练/测试= 7:3）。基于台站观测降水资料，进行结构性降水分析，发现滑坡聚集是由“前期降水积累”和“短时间高强度脉冲”双重机制共同引发的。进行综合因子质量评估，包括多重共线性分析（VIF < 5, TOL > 0.1）和Pearson相关性筛选，在建模之前确认条件因素的独立性和可靠性。然后开发并比较了六个模型——svc - gridsearch、SVC-Bayes、SVC-GWO、SVC-PSO、Random Forest和xgboost，并使用SHAP分析来增强可解释性，并与IGR结果交叉验证。XGBoost模型在测试集上取得了最好的性能（AUC≈0.915）。为了解决类别边界模糊问题，进一步引入了去噪扩散概率模型（DDPM），在11维因子空间中进行受控数据增强，在模型的“混淆区”内生成约12%的目标样本（预测概率为0.45-0.55）。增强后，XGBoost AUC增加到≈0.931，DeLong检验结果显著（p < 0.01），灵敏度提高，置信区间更窄。提出的可解释机器学习和生成概率建模的混合框架有效提高了有限样本条件下敏感性映射的准确性，并为东南山区的风险评估、应急控制和减灾规划提供了技术支持。

{"title":"Explainable machine learning and generative diffusion modeling for improved susceptibility mapping of rainfall-induced clustered landslides: A case study from Wuping County, southeastern China","authors":"Yu Huang , Yinke Li , Lisha Hei , Jialing Zou , Dingyu Chen","doi":"10.1016/j.compgeo.2026.107989","DOIUrl":"10.1016/j.compgeo.2026.107989","url":null,"abstract":"<div><div>Rainfall-induced clustered landslides have become increasingly frequent in southeastern China, characterized by strong multi-factor coupling and posing significant threats to local communities and infrastructure. Taking Wuping County, Fujian Province, as a representative case, this study established a post-event inventory of 6,005 shallow landslides triggered by the extreme rainfall on 15–16 June 2024. An equal number of non-landslide samples were generated, and eleven topographic, geological, hydrological, vegetation, and anthropogenic factors were compiled into a 12.5 m-resolution dataset (training/testing = 7:3). Based on station-observed rainfall data, a structural rainfall analysis revealed that the landslide clustering was jointly triggered by the dual mechanisms of “antecedent rainfall accumulation” and “short-duration high-intensity pulses.” A comprehensive factor quality assessment was performed, including multicollinearity analysis (VIF < 5, TOL > 0.1) and Pearson correlation screening, confirming the independence and reliability of the conditioning factors prior to modeling. Six models—SVC-GridSearch, SVC-Bayes, SVC-GWO, SVC-PSO, Random Forest, and XGBoost—were then developed and compared, with SHAP analysis used to enhance interpretability and cross-validate with IGR results. The XGBoost model achieved the best performance on the test set (AUC ≈ 0.915). To address class boundary ambiguity, a Denoising Diffusion Probabilistic Model (DDPM) was further introduced for controlled data augmentation in the 11-dimensional factor space, generating about 12% of targeted samples within the model’s “confusion zone” (predicted probability 0.45–0.55). After augmentation, the XGBoost AUC increased to ≈ 0.931, with a significant DeLong test result (p < 0.01), improved sensitivity, and narrower confidence intervals. The proposed hybrid framework of explainable machine learning and generative probabilistic modeling effectively enhances susceptibility mapping accuracy under limited-sample conditions and provides technical support for risk assessment, emergency control, and mitigation planning in southeastern mountainous regions.</div></div>","PeriodicalId":55217,"journal":{"name":"Computers and Geotechnics","volume":"193 ","pages":"Article 107989"},"PeriodicalIF":6.2,"publicationDate":"2026-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147385401","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

A dual mortar method for analyzing the effects of wave-induced instantaneous liquefaction on an immersed tunnel with a liquefaction-associated non-Darcy flow model 基于液化相关非达西流动模型的双砂浆法分析波浪瞬时液化对沉管隧道影响

IF 6.2 1区工程技术 Q1 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Computers and Geotechnics

Pub Date : 2026-05-01 Epub Date: 2026-02-16 DOI: 10.1016/j.compgeo.2026.107988

Shichong Han , Mozhen Zhou , Tielin Chen , Dingli Zhang , Wengang Qi

Engineering practices have illustrated that ocean waves can cause non-ignorable deformations in immersed tunnels and the surrounding seabed. Under extreme wave conditions, instantaneous liquefaction of the seabed can occur, leading to a decrease in the bearing capacity of the seabed and potential damage to subsea structures. The liquefaction-associated non-Darcy flow model, previously proposed to eliminate the nonphysical tensile behavior of the seabed, is incorporated into a wave–seabed–tunnel model, which introduces special interfacial conditions between the seabed and the immersed tunnel. These interfacial conditions are numerically treated by developing a dual mortar method, which also permits using nonconforming meshes between the seabed and the tunnel. This model is applied to investigate the wave-induced response of an immersed tunnel fully buried in a non-cohesive seabed. The representative values of wave parameters are obtained by analyzing in-situ monitoring data from coastal stations, and then used as the input of the wave–seabed–tunnel model. The numerical results show that extreme waves can pose significant influences on the immersed tunnel.

工程实践表明，海浪会引起沉管隧道及其周围海床不可忽视的变形。在极端波浪条件下，海底会发生瞬间液化，导致海底承载能力下降，对海底结构物造成潜在破坏。为了消除海床的非物理拉伸行为，之前提出的液化相关非达西流动模型被纳入波浪-海底-隧道模型，该模型引入了海底和沉管隧道之间的特殊界面条件。通过开发双砂浆方法对这些界面条件进行了数值处理，该方法也允许在海底和隧道之间使用非一致网格。应用该模型研究了非粘性海底全埋隧道的波致响应。通过分析海岸站的现场监测数据，得到波浪参数的代表性值，并将其作为波浪-海底隧道模型的输入。数值计算结果表明，极端波浪对沉管隧道有较大的影响。

{"title":"A dual mortar method for analyzing the effects of wave-induced instantaneous liquefaction on an immersed tunnel with a liquefaction-associated non-Darcy flow model","authors":"Shichong Han , Mozhen Zhou , Tielin Chen , Dingli Zhang , Wengang Qi","doi":"10.1016/j.compgeo.2026.107988","DOIUrl":"10.1016/j.compgeo.2026.107988","url":null,"abstract":"<div><div>Engineering practices have illustrated that ocean waves can cause non-ignorable deformations in immersed tunnels and the surrounding seabed. Under extreme wave conditions, instantaneous liquefaction of the seabed can occur, leading to a decrease in the bearing capacity of the seabed and potential damage to subsea structures. The liquefaction-associated non-Darcy flow model, previously proposed to eliminate the nonphysical tensile behavior of the seabed, is incorporated into a wave–seabed–tunnel model, which introduces special interfacial conditions between the seabed and the immersed tunnel. These interfacial conditions are numerically treated by developing a dual mortar method, which also permits using nonconforming meshes between the seabed and the tunnel. This model is applied to investigate the wave-induced response of an immersed tunnel fully buried in a non-cohesive seabed. The representative values of wave parameters are obtained by analyzing in-situ monitoring data from coastal stations, and then used as the input of the wave–seabed–tunnel model. The numerical results show that extreme waves can pose significant influences on the immersed tunnel.</div></div>","PeriodicalId":55217,"journal":{"name":"Computers and Geotechnics","volume":"193 ","pages":"Article 107988"},"PeriodicalIF":6.2,"publicationDate":"2026-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147385840","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Studying the mechanical behaviour of an anisotropic clay rock around gallery intersections and the effect of the support stiffness 研究了各向异性粘土岩在巷道交叉点周围的力学行为及支护刚度的影响

IF 6.2 1区工程技术 Q1 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Computers and Geotechnics

Pub Date : 2026-05-01 Epub Date: 2026-02-13 DOI: 10.1016/j.compgeo.2026.107945

Panteleimon Rapanakis , Benoît Pardoen , Denis Branque , Jan S. Cornet , Gilles Armand

Gallery intersections are frequently encountered in underground construction of deep geological repositories for nuclear waste. Their three-dimensional nature, combined with the excavation sequence and primary support installation, makes them a complex challenge to address. The redistribution of stress and strain that occurs, plays a key role in the response of the surrounding rock. Furthermore, a high in-situ stress state and an anisotropic nature of the material could have a substantial influence on the response of the rock. In this context, the present study investigates the behaviour of perpendicularly intersecting supported galleries excavated at great depth in an anisotropic clay rock through 3D finite element analyses. A conventional step-by-step progressive excavation is modelled as well as the phasing of the intersection excavation. The surrounding rock is a sedimentary indurated claystone, the Callovo-Oxfordian (COx) clay rock, which is assumed to follow an anisotropic Drucker-Prager elastoplastic constitutive law with shear strength hardening. Different stiffnesses for the elastic support are used to account for a flexible, an intermediate, and a more rigid support. The obtained results focus on the impact of each support stiffness on the stress distribution, plastic strain, and the generated plastic zone in the surrounding rock mass.

在核废料深地质处置库的地下建设中，经常遇到坑道交叉。它们的三维性质，再加上开挖顺序和主要的支撑装置，使它们成为一个复杂的挑战。发生的应力和应变的重新分布在围岩的响应中起着关键作用。此外，高地应力状态和材料的各向异性可能对岩石的响应产生重大影响。在这种情况下，本研究通过三维有限元分析，研究了各向异性粘土岩石中垂直相交的大深度支撑巷道的行为。一个传统的循序渐进的挖掘，以及交叉挖掘的阶段性建模。围岩为沉积硬化粘土岩，即Callovo-Oxfordian （COx）粘土岩，被认为遵循各向异性Drucker-Prager弹塑性本构律，具有剪切强度硬化。弹性支撑的不同刚度用于解释柔性、中间和更刚性的支撑。得到的结果集中在各支护刚度对围岩应力分布、塑性应变和塑性区生成的影响上。

{"title":"Studying the mechanical behaviour of an anisotropic clay rock around gallery intersections and the effect of the support stiffness","authors":"Panteleimon Rapanakis , Benoît Pardoen , Denis Branque , Jan S. Cornet , Gilles Armand","doi":"10.1016/j.compgeo.2026.107945","DOIUrl":"10.1016/j.compgeo.2026.107945","url":null,"abstract":"<div><div>Gallery intersections are frequently encountered in underground construction of deep geological repositories for nuclear waste. Their three-dimensional nature, combined with the excavation sequence and primary support installation, makes them a complex challenge to address. The redistribution of stress and strain that occurs, plays a key role in the response of the surrounding rock. Furthermore, a high in-situ stress state and an anisotropic nature of the material could have a substantial influence on the response of the rock. In this context, the present study investigates the behaviour of perpendicularly intersecting supported galleries excavated at great depth in an anisotropic clay rock through 3D finite element analyses. A conventional step-by-step progressive excavation is modelled as well as the phasing of the intersection excavation. The surrounding rock is a sedimentary indurated claystone, the Callovo-Oxfordian (COx) clay rock, which is assumed to follow an anisotropic Drucker-Prager elastoplastic constitutive law with shear strength hardening. Different stiffnesses for the elastic support are used to account for a flexible, an intermediate, and a more rigid support. The obtained results focus on the impact of each support stiffness on the stress distribution, plastic strain, and the generated plastic zone in the surrounding rock mass.</div></div>","PeriodicalId":55217,"journal":{"name":"Computers and Geotechnics","volume":"193 ","pages":"Article 107945"},"PeriodicalIF":6.2,"publicationDate":"2026-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146174290","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Crack initiation and propagation mechanism of borehole surrounding rock subjected to cyclic thermal loading: insights from theoretical solution and DEM simulation 循环热载荷作用下钻孔围岩裂纹萌生与扩展机制：理论解与DEM模拟

IF 6.2 1区工程技术 Q1 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Computers and Geotechnics

Pub Date : 2026-05-01 Epub Date: 2026-02-03 DOI: 10.1016/j.compgeo.2026.107943

Qiuxin Gu , Qiang Zhang , Kai Zhang , Hui Liu , Yihan Du , Bo Huang , Wei Han

The hydraulic fracturing and low-temperature thermal stimulation are commonly adopted for the construction of geothermal reservoirs in hot dry rock (HDR). However, it is hardly involved in the existing research on the variation law of the temperature and stress fields in the surrounding rock and the crack initiation propagation mechanism when the cooling fluid flows through the borehole. In this study, the unsteady temperature and stress field of the borehole surrounding rock during the cooling process was derived and solved firstly using the heat transfer and elasticity mechanics theories. Then, the crack initiation and propagation criteria for the borehole surrounding rock are proposed according to the fracture mechanics theory. Finally, the initiation and propagation laws of thermal cracks in the borehole surrounding rock under cyclic thermal shock are investigated through the discrete element method. The results reveal that when the cooling fluid is injected into the borehole, the temperature of the rock around the borehole drops the fastest. As the distance from the borehole increases, the temperature gradually rises and gets closer to the initial rock temperature. The temperature variation of the surrounding rock is closely related to the duration of thermal shock. During the initial stage of liquid nitrogen injection, the temperature drop is the most obvious. With the increase in thermal shock time, the tangential stress transitions from compressive stress to tensile stress. The tensile stress is the largest at the edge of the borehole, which is the location most prone to cracking. The mesoscopic cracking characteristics of the borehole surrounding rock are influenced by multiple factors, including buried depth, initial temperature, cooling method, thermal cycles, and the mesoscopic composition features of HDR. These research findings provide significant theoretical reference for EGS reservoir construction and high-efficiency stable operation

热干岩地热储层的开发通常采用水力压裂和低温增产两种方法。然而，对于冷却流体流过井眼时围岩温度场和应力场的变化规律以及裂缝起裂扩展机制的研究，现有的研究很少涉及。本文首先利用传热力学和弹性力学理论，推导并求解了冷却过程中钻孔围岩的非定常温度场和应力场。然后，根据断裂力学理论，提出了钻孔围岩裂纹起裂和扩展准则。最后，通过离散元法研究了循环热冲击作用下钻孔围岩热裂纹的起裂和扩展规律。结果表明，注入冷却液后，井眼周围岩石温度下降最快。随着离钻孔距离的增加，温度逐渐升高，逐渐接近岩石初始温度。围岩的温度变化与热冲击的持续时间密切相关。在注入液氮初期，温度下降最为明显。随着热冲击时间的增加，切向应力由压应力转变为拉应力。钻孔边缘处的拉应力最大，是最容易开裂的位置。井眼围岩的细观开裂特征受埋深、初始温度、冷却方式、热循环、HDR细观组成特征等多种因素的影响。这些研究成果为EGS水库建设和高效稳定运行提供了重要的理论参考

{"title":"Crack initiation and propagation mechanism of borehole surrounding rock subjected to cyclic thermal loading: insights from theoretical solution and DEM simulation","authors":"Qiuxin Gu , Qiang Zhang , Kai Zhang , Hui Liu , Yihan Du , Bo Huang , Wei Han","doi":"10.1016/j.compgeo.2026.107943","DOIUrl":"10.1016/j.compgeo.2026.107943","url":null,"abstract":"<div><div>The hydraulic fracturing and low-temperature thermal stimulation are commonly adopted for the construction of geothermal reservoirs in hot dry rock (HDR). However, it is hardly involved in the existing research on the variation law of the temperature and stress fields in the surrounding rock and the crack initiation propagation mechanism when the cooling fluid flows through the borehole. In this study, the unsteady temperature and stress field of the borehole surrounding rock during the cooling process was derived and solved firstly using the heat transfer and elasticity mechanics theories. Then, the crack initiation and propagation criteria for the borehole surrounding rock are proposed according to the fracture mechanics theory. Finally, the initiation and propagation laws of thermal cracks in the borehole surrounding rock under cyclic thermal shock are investigated through the discrete element method. The results reveal that when the cooling fluid is injected into the borehole, the temperature of the rock around the borehole drops the fastest. As the distance from the borehole increases, the temperature gradually rises and gets closer to the initial rock temperature. The temperature variation of the surrounding rock is closely related to the duration of thermal shock. During the initial stage of liquid nitrogen injection, the temperature drop is the most obvious. With the increase in thermal shock time, the tangential stress transitions from compressive stress to tensile stress. The tensile stress is the largest at the edge of the borehole, which is the location most prone to cracking. The mesoscopic cracking characteristics of the borehole surrounding rock are influenced by multiple factors, including buried depth, initial temperature, cooling method, thermal cycles, and the mesoscopic composition features of HDR. These research findings provide significant theoretical reference for EGS reservoir construction and high-efficiency stable operation</div></div>","PeriodicalId":55217,"journal":{"name":"Computers and Geotechnics","volume":"193 ","pages":"Article 107943"},"PeriodicalIF":6.2,"publicationDate":"2026-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146174472","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Research on catastrophic process of large-scale slope using 3D-DDA with machine learning-based parameter optimization 基于机器学习参数优化的3D-DDA大尺度边坡突变过程研究

IF 6.2 1区工程技术 Q1 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Computers and Geotechnics

Pub Date : 2026-05-01 Epub Date: 2026-02-07 DOI: 10.1016/j.compgeo.2026.107969

Jingyu Kang , Xiaodong Fu , Hao Sheng , Jian Chen , Xu Cheng

The investigations on the entire process of slope instability holds significant implications for disaster prevention. Discontinuous deformation analysis (DDA) serves as a robust numerical tool capable of capturing dynamic interactions among rock blocks, making it particularly suitable for landslide simulation. However, due to the low computational efficiency and complex parameter determination, its application to large-scale landslides remains challenging. In this study, we employ parallel computing and machine learning to enhance three dimensional (3D) DDA for large-scale landslide deduction. Firstly, the OpenMP parallel strategy is adopted in the full stage of 3D explicit DDA, and the accuracy and efficiency are validated by a designed case. Then, a surrogate model based on the Light Gradient Boosting Machine (LightGBM) and Bayesian optimization is proposed to determine the subjective parameters in simulation. The training data involving common interaction modes in landslide are obtained by 3D-DDA and the performance of the surrogate model are evaluated by various indicators. Finally, a large-scale landslide is comprehensively analyzed based on the enhanced 3D DDA. The digital elevation of the landslide is obtained by unmanned aerial vehicle, and the numerical model composed of over ten thousands blocks is established. The time step and contact stiffness is provided by the surrogate model. The catastrophic process of the landslide is reproduced and the kinematic characteristics are comprehensively investigated. The present study enhances the capability of 3D DDA in large-scale simulation and can also be referenced by related discontinuum-based methods.

对边坡失稳全过程的研究对防灾减灾具有重要意义。不连续变形分析（DDA）作为一种强大的数值工具，能够捕捉岩石块体之间的动态相互作用，使其特别适用于滑坡模拟。然而，由于计算效率低，参数确定复杂，其在大型滑坡中的应用仍然具有挑战性。在本研究中，我们采用并行计算和机器学习来增强大规模滑坡推理的三维DDA。首先，在三维显式数据处理的全过程中采用了OpenMP并行策略，并通过设计实例验证了该策略的精度和效率。然后，提出了基于光梯度增强机（LightGBM）和贝叶斯优化的代理模型来确定仿真中的主观参数。利用3D-DDA获得滑坡中常见相互作用模式的训练数据，并通过各种指标对代理模型的性能进行评价。最后，基于增强型三维DDA对某大型滑坡进行了综合分析。利用无人机获取滑坡数字高程，建立了由万余块体组成的滑坡数值模型。时间步长和接触刚度由代理模型提供。再现了滑坡的灾变过程，全面研究了滑坡的运动学特征。本文的研究提高了三维DDA在大规模仿真中的应用能力，也可为相关的基于间断体的方法提供参考。

{"title":"Research on catastrophic process of large-scale slope using 3D-DDA with machine learning-based parameter optimization","authors":"Jingyu Kang , Xiaodong Fu , Hao Sheng , Jian Chen , Xu Cheng","doi":"10.1016/j.compgeo.2026.107969","DOIUrl":"10.1016/j.compgeo.2026.107969","url":null,"abstract":"<div><div>The investigations on the entire process of slope instability holds significant implications for disaster prevention. Discontinuous deformation analysis (DDA) serves as a robust numerical tool capable of capturing dynamic interactions among rock blocks, making it particularly suitable for landslide simulation. However, due to the low computational efficiency and complex parameter determination, its application to large-scale landslides remains challenging. In this study, we employ parallel computing and machine learning to enhance three dimensional (3D) DDA for large-scale landslide deduction. Firstly, the OpenMP parallel strategy is adopted in the full stage of 3D explicit DDA, and the accuracy and efficiency are validated by a designed case. Then, a surrogate model based on the Light Gradient Boosting Machine (LightGBM) and Bayesian optimization is proposed to determine the subjective parameters in simulation. The training data involving common interaction modes in landslide are obtained by 3D-DDA and the performance of the surrogate model are evaluated by various indicators. Finally, a large-scale landslide is comprehensively analyzed based on the enhanced 3D DDA. The digital elevation of the landslide is obtained by unmanned aerial vehicle, and the numerical model composed of over ten thousands blocks is established. The time step and contact stiffness is provided by the surrogate model. The catastrophic process of the landslide is reproduced and the kinematic characteristics are comprehensively investigated. The present study enhances the capability of 3D DDA in large-scale simulation and can also be referenced by related discontinuum-based methods.</div></div>","PeriodicalId":55217,"journal":{"name":"Computers and Geotechnics","volume":"193 ","pages":"Article 107969"},"PeriodicalIF":6.2,"publicationDate":"2026-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146174464","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Utilizing physics-informed neural network and geotechnical distance field for solving three-dimensional nonlinear consolidation 利用物理信息神经网络和岩土距离场求解三维非线性固结

IF 6.2 1区工程技术 Q1 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Computers and Geotechnics

Pub Date : 2026-05-01 Epub Date: 2026-01-23 DOI: 10.1016/j.compgeo.2026.107926

Tran-Gia-Khiem Nguyen, Jongmuk Won

Solving three-dimensional (3D) nonlinear consolidation is complex and computationally expensive. This study proposes a framework for solving 3D nonlinear consolidation by utilizing an improved physics-informed neural network with hard constraints coupling with machine learning based geotechnical distance functions for three-dimensional spatial interpolation. The performance of the developed framework was assessed by comparing pore water pressure data between the developed framework and those obtained from COMSOL Multiphysics. In addition, the impact of vertical hydraulic conductibility heterogeneity, compression index, and void ratio on long-term settlement was also evaluated and discussed. It was found that the proposed framework showed a reliable estimation of the 3D distribution of pore water pressure across the 3D domain, achieving results that are comparable to data obtained from COMSOL. In addition, the heterogeneity of hydraulic conductivity can be successfully considered using the developed framework, which enables assessing the long-term settlement of a clay deposit with high uncertainty of hydraulic conductivity. Overall, the developed framework shown in this study can be applied to complex consolidation problems with low computational costs and high accuracy.

求解三维（3D）非线性固结是复杂和计算昂贵的。本研究提出了一个解决三维非线性固结的框架，该框架利用改进的物理信息神经网络与硬约束耦合，基于机器学习的岩土距离函数用于三维空间插值。通过比较开发框架与COMSOL Multiphysics获得的孔隙水压力数据，对开发框架的性能进行了评估。此外，还评价和讨论了竖向导电性非均质性、压缩指数和孔隙比对长期沉降的影响。研究发现，所提出的框架可以可靠地估计孔隙水压力在三维域中的三维分布，其结果与COMSOL获得的数据相当。此外，利用所开发的框架可以成功地考虑水力导电性的非均质性，从而能够评估具有高水力导电性不确定性的粘土沉积物的长期沉降。总的来说，本研究中显示的开发框架可以应用于复杂的固结问题，计算成本低，精度高。

{"title":"Utilizing physics-informed neural network and geotechnical distance field for solving three-dimensional nonlinear consolidation","authors":"Tran-Gia-Khiem Nguyen, Jongmuk Won","doi":"10.1016/j.compgeo.2026.107926","DOIUrl":"10.1016/j.compgeo.2026.107926","url":null,"abstract":"<div><div>Solving three-dimensional (3D) nonlinear consolidation is complex and computationally expensive. This study proposes a framework for solving 3D nonlinear consolidation by utilizing an improved physics-informed neural network with hard constraints coupling with machine learning based geotechnical distance functions for three-dimensional spatial interpolation. The performance of the developed framework was assessed by comparing pore water pressure data between the developed framework and those obtained from COMSOL Multiphysics. In addition, the impact of vertical hydraulic conductibility heterogeneity, compression index, and void ratio on long-term settlement was also evaluated and discussed. It was found that the proposed framework showed a reliable estimation of the 3D distribution of pore water pressure across the 3D domain, achieving results that are comparable to data obtained from COMSOL. In addition, the heterogeneity of hydraulic conductivity can be successfully considered using the developed framework, which enables assessing the long-term settlement of a clay deposit with high uncertainty of hydraulic conductivity. Overall, the developed framework shown in this study can be applied to complex consolidation problems with low computational costs and high accuracy.</div></div>","PeriodicalId":55217,"journal":{"name":"Computers and Geotechnics","volume":"193 ","pages":"Article 107926"},"PeriodicalIF":6.2,"publicationDate":"2026-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146026134","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0