Figure - available from: Multiscale and Multidisciplinary Modeling, Experiments and Design
This content is subject to copyright. Terms and conditions apply.
Source publication
At present, there exist many methods for liquefaction analysis of a soil deposit. Some of them are suitable for only coarse-grained soils, while a few others can also evaluate the liquefaction potential of both fine- and coarse-grained soils. It is important to identify the most suitable method for liquefaction analysis. The current study looks at...
Citations
... Due to its extensive use, there are large SPT databases from across the world. Therefore, correlations with SPT N value are available for virtually any geotechnical parameter, such as consistency [6], index properties [7], shear strength [8][9][10][11], deformation properties [12], and liquefaction potentials of cohesive soils [13][14][15][16][17][18]. ...
This study aims to assess the accuracy and consistency of existing correlations between undrained shear strength (c u) and modulus of volumetric compressibility (m v) with SPT N value, and compression index (C c) with index parameters for Turkish clays, with a specific focus on Ankara clay using laboratory testing of over 2500 high-quality undisturbed samples from 42 sites. A comparison of correlations for c u using five different statistical variables indicates that no single method stands out as superior in terms of both accuracy and conservatism and that the quality of the tests in the database has a significant impact on the correlation performance. Even the most effective correlation for c u appear insufficient for direct application in geotechnical design processes, given its low consistency characterized by coefficient of variations (COV) ranging between 50% and more than 100%. The widely used m v estimates proposed by Stroud (1974) were found to be notably unconservative when applied to Turkish clays. Consequently, a new trendline, leveraging both SPT N value and Plasticity Index (PI), was introduced to address this discrepancy. Meanwhile, an examination of 283 high-quality undisturbed samples from Ankara Clay revealed existing correlations between initial void ratio (e 0) and compression index (C c) to be conservative, yet consistently reliable. This underscores their suitability for routine or preliminary design applications. Notably, a newly derived correlation was proposed, which demonstrates equal applicability to the well-established Bowles (1979) equation.
... Therefore, researchers and scientists applied soft computing techniques to assess the liquefaction potential using the in situ test database (Samui and Hariharan 2015). In recent decades, soft computing techniques have increased because of the availability of in situ databases (Kumar et al. 2023c). These techniques are more accurate and frequent than traditional methods in assessing the liquefaction potential. ...
In soil mechanics, liquefaction is the phenomenon that occurs when saturated, cohesionless soils temporarily lose their strength and stiffness under cyclic loading shaking or earthquake. The present work introduces an optimal performance model by comparing two baselines, thirty tree-based, thirty support vector classifier-based, and fifteen neural network-based models in assessing the liquefaction potential. One hundred and seventy cone penetration test results (liquefied and non-liquefied) have been compiled from the literature for this aim. Earthquake magnitude, vertical-effective stress, mean grain size, cone tip resistance, and peak ground acceleration parameters have been used as input parameters to predict the soil liquefaction potential for the first time. Performance metrics, accuracy, an area under the curve (AUC), precision, recall, and F1 score have measured the training and testing performances. The comparison of performance metrics reveals that the model Runge–Kutta optimized extreme gradient boosting (RUN_XGB) has assessed the liquefaction potential with an overall accuracy of 99%, AUC of 0.99, precision of 0.99, recall value of 1, and F1 score of 1. Moreover, model RUN_XGB has a true negative rate of 0.98, negative predictive value of 1, Matthews correlation coefficient of 0.98, and average classification accuracy of 0.99, close to the ideal values and presents the robustness of the RUN_XGB model. Finally, the RUN_XGB model has been recognized as an optimal performance model for predicting the liquefaction potential. It has been noted that a low multicollinearity level affects the prediction accuracy of models based on conventional soft computing techniques, i.e., logistic regression. This research will help researchers choose suitable hybrid algorithms and enhance the accuracy of seismic soil liquefaction potential models.
... other studies also aim to evaluate liquefaction potential, employing methods such as relevance vector machine (RVM), support vector machine (SVM), multigene genetic programming (MGGP), extreme gradient boosting (XGBoost), random forest (RF), and hybrid artificial neural network (Das & Samui, 2008;D. R. Kumar et al., 2022bD. R. Kumar et al., , 2022aD. R. Kumar, Samui, & Burman, 2023;Muduli & Das, 2014;Samui, 2007;Samui & Karthikeyan, 2013). ...
... The mathematical equation used for the normalization is presented in Eq. (17) (D. R.Kumar, Samui, & Burman, 2023). ...
Assessing the potential for liquefaction using traditional experimental or empirical analysis procedures is both time-consuming and arduous. Employing a machine learning model that can accurately predict liquefaction potential for a specific site can reduce the time, effort, and associated costs. This study proposes several empirical machine learning models, including deep neural network (DNN), convolutional neural network (CNN), recurrent neural network (RNN), long short-term memory (LSTM), and bi-directional long short-term memory (BILSTM), to assess the liquefaction potential of soil deposits based on SPT-based post liquefaction datasets. To train the proposed models, a dataset comprising 834 liquefied and non-liquefied cases was collected to perform the liquefaction analysis. A Pearson correlation matrix was also conducted to examine the correlation between soil and seismic parameters and the probability of liquefaction. Furthermore, a sensitivity analysis was adopted to assess the impact of soil and seismic parameters on the probability of liquefaction. The proposed model's prediction capability was assessed using several performance indices, including rank analysis, accuracy matrix, and AIC criteria. The comparative analysis of the proposed models' predictive ability to determine liquefaction probability revealed that the RNN model outperformed the others, displaying the highest accuracy and lowest error index values. Subsequently, the RNN model achieved the first rank with a total score value of 70, followed by the CNN (55), DNN (52), BILSTM (47), and LSTM (16) models. The parametric analysis, rank analysis, accuracy matrix, and AIC criteria collectively demonstrate the proposed models' ability to predict liquefaction probability. Furthermore, the robustness of these models was assessed through external validation and comparative analysis.
... The correlation statistical parameters, including R 2 , NS, PI, and VAF, were computed to assess the linear relationship between the actual and model-obtained values of the bearing capacity factor ( N ). Conversely, the metrics, namely, RMSE, WI, WMAPE, and MAE, were calculated to quantify the degree of error associated with the proposed models when predicting the bearing capacity factor ( N ), as per the recommendations provided by previous studies (Kumar et al. 2021(Kumar et al. , 2022a(Kumar et al. , 2023a(Kumar et al. , 2023b(Kumar et al. , 2023c. The mathematical expressions of these metrics are presented as follows: ...
Determining the bearing capacity of ring foundations on rock masses holds utmost importance within the framework of foundation design methodology. To examine the failure mechanism of ring foundations situated on Hock-Brown rock masses, the crucial bearing capacity factor \({(N}_{\sigma })\) is analyzed. This analysis considered three dimensionless input parameters: the geological strength index (GSI), the yield parameter (mi), and the ratio of the internal and external radii (ri/ro). This study focuses on the development of a precise hybrid extreme learning machine (ELM) and least-square support vector machine (LSSVM) based on two swarm-based intelligence optimization algorithms, utilizing Harris hawks optimization (HHO) and particle swarm optimization (PSO). The primary objective of this study is to provide accurate predictions of the bearing capacity factor (\({N}_{\sigma }\)) for a ring foundation. Furthermore, the accuracy of the developed hybrid ELM-PSO, ELM-HHO, LSSVM-PSO, and LSSVM-HHO models was assessed through a comparison between the actual and predicted values of \({N}_{\sigma }\) using various performance metrics, uncertainty analysis, and rank analysis. The LSSVM-HHO and ELM-HHO outperformed the LSSVM-PSO and ELM-PSO models in predicting the \({N}_{\sigma }\) value. The proposed models can be used as soft computing tools to predict the \({N}_{\sigma }\) values in practical applications.
... Various researchers predicted the probability of liquefaction successfully using the advanced regression machine learning techniques (Kumar et al. 2022a(Kumar et al. , 2023aGoh 2016a, 2018;Zhang et al. 2015Zhang et al. , 2021. Recently, Kumar et al. (2021) proposed advanced regression soft computing technique to predict the probability of liquefaction. ...
... Later on, Kumar et al. (2022a) proposed the metaheuristic hybrid ANN model with various optimization techniques to predict the probability of liquefaction accurately. They also suggested the best method for assessing the probability of liquefaction based on post liquefaction SPT database (Kumar et al. 2023a). In geotechnical engineering applications, MARS model-based reliability analysis has recently gained substantial attention (Zhang et al. 2015;Goh and Zhang 2014;Metya et al. 2017;Zhang and Goh 2016b). ...
Liquefaction triggering phenomenon during earthquake is one of the most complicated geotechnical problems due to the complex and heterogeneous nature of the soils. In this study, artificial intelligence based predictive models have been developed to assess the probability of liquefaction. In this context, the total 834 field data sets have been taken from the published literatures pertaining to past earthquakes to develop the machine learning models. Artificial intelligence-based regression techniques such as, Relevance Vector Machine (RVM), Genetic Programming (GP), and Multivariate Adaptive Regression Spline (MARS) have been utilized to estimate the liquefaction potential of a soil deposit. The relative efficiencies of the applied machine learning models have been ascertained by comparing various error indices such as Nash–Sutcliffe efficiency (NS), Weighted mean absolute percentage error (WMAPE), root mean square error (RMSE), Coefficient of determination (R²), variance account factor (VAF), mean absolute error (MAE), adjusted determination coefficient (AdjR²), Willmott’s Index of agreement (WI) etc. This study also suggests mathematical equations based on the obtained result for all models to compute the probability of liquefaction. The performance of proposed models was evaluated on the basis of various performance indices, actual versus predicted curve and rank analysis, etc. Additionally, Taylor diagram and regression error characteristic (REC) curve are also presented to check the effectiveness of the proposed models. On the basic of acquired results, it can be concluded that the GP model predicts the probability of liquefaction effectively compared to MARS and RVM models. This study shows the applicability of GP, RVM, and MARS models that offer a useful alternative tool for earthquake engineers to assess liquefaction conditions in liquefaction-prone areas.
... Finally, the total score is calculated by adding the training and testing scores, and rank is allotted based on the total score obtained for each model. [52] The mathematical Eq. (34) used for the total score calculation is 12 | Eng. Sci., 2023, 24 presented as follows: = ∑ =1 + ∑ =1 (34) where and represent the score of the individual statistical parameters for the training and testing phases, respectively. ...
It is a difficult task for practical engineers to calculate the uplift and penetration resistance of two overlapping pipelines that are buried in clay that increases in strength linearly. Hence, in this paper, four regression models, namely long short-term memory (LSTM), bidirectional long short-term memory (BI-LSTM), emotional neural network (ENN), and multivariate adaptive regression spline (MARS) models, are employed to create a data-driven prediction for the uplift and penetration resistance of two overlapping pipelines. For this purpose, a total of 256 samples of uplift conditions and 384 samples of penetration conditions, including three input parameters and one output parameter, are collected from the lower and upper bound finite element limit analysis (FELA) solutions. The predictive strength and robustness of the employed model were evaluated based on various performance metrics, rank analysis, error matrix, Taylor diagram and uncertainty analysis. Sensitivity analysis is also performed to determine the most and least effective parameters. Additionally, the results of the sensitivity analysis indicated that the pipe embedded depth ratio (w/D) was the most significant parameter in both uplift and penetration conditions. The MARS model produces more efficient performance (R2=0.999 and RMSE=0.008 for uplift and R2=0.999 and RMSE=0.009 for penetration condition) for uplift and penetration resistance prediction compared to the BI-LSTM, LSTM, and ENN models. The acquired findings demonstrated that the MARS model predicted the normalized uplift and penetration resistance with reasonable accuracy and yielded superior performance compared to the Bi-LSTM, LSTM, and ENN models. Therefore, it becomes one of the predictive tools practical engineers use in making preliminary decisions about things such as the uplift and penetration resistance of two overlapping pipelines buried in clay, which increases in strength linearly. It also provides a mathematical formulation for easy hand calculations.
The study presents advanced hybrid artificial neural network (ANN) models to enhance the analysis of slope stability and the prediction of the factor of safety (FOS). Conventional methodologies, such as analytical procedures and numerical simulations, encounter difficulties in accurately representing intricate interactions. Consequently, machine learning (ML) simulation methods are being introduced as contemporary alternatives. This paper provides a comparative examination of hybrid artificial neural network (ANN) models utilizing several optimization algorithms (OAs) like Firefly (FF), Ant Lion (ALO), shuffled complex evolution (SCE), and the imperialist colony algorithm (ICA), simulated on a dataset of 349 slope cases. The performance of the ML models is enhanced by hyperparameter tuning and verified using various proven statistical indicators, error matrix, and external validation. It is concluded that all the applied models are robust for practical applications. The comparative analysis is carried out using rank analysis which proposes ANN-FF (rank = 64, RTR2 = 0.983 and RTS2 = 0.943) as the best performing model, followed by ANN-ICA (rank = 42, RTR2 = 0.9772 and RTS2 = 0.937) and ANN-ALO (rank = 37, RTR2 = 0.973 and RTS2 = 0.909). The developed hybrid ANN models demonstrate significant potential as a novel tool to aid engineers in estimating rock strain during the design phase of various engineering projects.
The article presents a state-of-the-art of machine learning (ML)-based reliability analysis for predicting the safety performance of a simply supported beam subjected to a uniformly distributed load. For this purpose, state-of-the-art hybrid artificial neural network (ANN) models are used using four optimization algorithms: ABC (Artificial Bee Colony), ACO (Ant Colony Optimization), ALO (Ant Lion Optimization), and TLBO (Teaching Learning Based Optimization). For the study, 800 datasets of a simply supported beam were generated, where load intensity (w) and modulus of elasticity (E) are the input parameters and deflection (δ) is the predictor parameter. The authenticity of the dataset is established using descriptive statistics, a histogram, and the correlation matrix. The training and testing performances of the model are analysed using various well-established performance parameters, rank analysis, and AIC criteria. The comparative analysis of the performances is provided using rank analysis, while the visual, reader-friendly analysis is provided using Taylor diagrams. Reliability indices (RI) and probability of failure (POF) are calculated for the actual values of deflection using FORM and compared with ML modelling-based reliability indices. The values of RI and POF for ANN-ALO, ANN-ACO and ANN-TLBO are closer to the actual, while ANN-ABC yields over-prediction. Therefore, it is concluded that the hybrid ANNs simulated in the study are robust and reliable tools for reliability-based design of a simply supported beam, however, the performance of ANN-ABC is less satisfactory compared to others.
