Ayele Chala

58189127500

Publications - 8

Uncertainty Quantification in Shear Wave Velocity Predictions: Integrating Explainable Machine Learning and Bayesian Inference

Ayele Chala R. P. Ray

Publication Name: Applied Sciences Switzerland

Publication Date: 2025-02-01

Volume: 15

Issue: 3

Page Range: Unknown

Description:

The accurate prediction of shear wave velocity (Vs) is critical for earthquake engineering applications. However, the prediction is inevitably influenced by geotechnical variability and various sources of uncertainty. This paper investigates the effectiveness of integrating explainable machine learning (ML) model and Bayesian generalized linear model (GLM) to enhance both predictive accuracy and uncertainty quantification in Vs prediction. The study utilizes an Extreme Gradient Boosting (XGBoost) algorithm coupled with Shapley Additive Explanations (SHAPs) and partial dependency analysis to identify key geotechnical parameters influencing Vs predictions. Additionally, a Bayesian GLM is developed to explicitly account for uncertainties arising from geotechnical variability. The effectiveness and predictive performance of the proposed models were validated through comparison with real case scenarios. The results highlight the unique advantages of each model. The XGBoost model demonstrates good predictive performance, achieving high coefficient of determination ((Formula presented.)), index of agreement (IA), Kling–Gupta efficiency (KGE) values, and low error values while effectively explaining the impact of input parameters on Vs. In contrast, the Bayesian GLM provides probabilistic predictions with 95% credible intervals, capturing the uncertainty associated with the predictions. The integration of these two approaches creates a comprehensive framework that combines the strengths of high-accuracy ML predictions with the uncertainty quantification of Bayesian inference. This hybrid methodology offers a powerful and interpretable tool for Vs prediction, providing engineers with the confidence to make informed decisions.

Open Access: Yes

DOI: 10.3390/app15031409

Efficient Uncertainty Quantification in Seismic Site Response via Random Field Modeling

Ayele Chala R. P. Ray

Publication Name: Geotechnical Special Publication

Publication Date: 2025-01-01

Volume: 2025-March

Issue: GSP 366

Page Range: 158-169

Description:

Assessing seismic site response with absolute certainty remains challenging due to inherent soil variabilities. Soil variabilities primarily stem from natural geologic processes and lie beyond human control. Soil characteristics, including shear wave velocity, shear modulus, unit weight, and plasticity, exhibit inherent randomness and variability. These factors significantly influence seismic site responses, making it crucial to account for parameter uncertainty in seismic behavior characterization. Precisely characterizing this uncertainty is essential for reliable seismic hazard assessment and the design of earthquake-resistant structures. However, this task faces challenges due to the computational expense associated with many model realizations. In this study, a computationally efficient and user-friendly model for estimating peak ground acceleration (PGA), displacement (PGD), and quantifying uncertainty was developed. The methodology comprises two main steps: (1) 2D equivalent linear seismic site response analyses were simulated using MIDAS GTS NX commercial software. These simulations incorporated randomly generated properties of clay soil, including maximum shear modulus (G0), unit weight (γ), and plasticity index (PI). (2) Leveraging data from the site response analyses, a Bayesian regression model was developed using the R programming language. The accuracy and reliability of the developed model were validated using a new data set, and the results closely aligned with finite element method (FEM) outcomes. By accounting for soil inherent variabilities, the model effectively characterizes the uncertainty of PGA and PGD using mean and coefficient of variation (CoV). Remarkably, the Bayesian approach yielded CoV of response parameters up to 2.33%, a substantial 94.37% relative difference compared to the FEM. Notably, this improvement in uncertainty was achieved while maintaining computational efficiency.

Open Access: Yes

DOI: 10.1061/9780784485996.016

Quantifying Geotechnical Uncertainty in Ground Motion Predictions: Bayesian Generalized Linear Model Framework

Clara Beatrice Vilceanu Ayele Chala Mais Mayassah R. P. Ray

Publication Name: Advances in Civil Engineering

Publication Date: 2025-01-01

Volume: 2025

Issue: 1

Page Range: Unknown

Description:

Accurate prediction of peak ground intensity measures is inevitably influenced by geotechnical variability. Variations in soil properties, subsurface conditions, and seismic inputs introduce complexities that challenge the reliability of predictions. This study introduces a Bayesian generalized linear model (GLM) to probabilistically predict peak ground acceleration (PGA) while accounting for uncertainties associated with geotechnical variability. Latin hypercube sampling (LHS) was employed to generate synthetic datasets of key geotechnical parameters, including plasticity index, shear wave velocity, soil thickness, input motion intensity, and unit weight of soil for hypothetical sites. Subsequently, a series of one-dimensional equivalent linear (1D-EQL) seismic site response analyses were performed, and PGA value at ground surface level were recorded for each analysis. The Bayesian GLM was then developed using these comprehensive datasets to probabilistically predict PGA. The performance and reliability of the developed model were evaluated on a separate test dataset. To benchmark its performance, a Bayesian neural network (BNN) was also developed and compared. In addition, a Shiny-based graphical user interface (GUI), named Bayes-PGA-predictor, was implemented to facilitate practical application. The findings demonstrate that the Bayesian GLM offers a robust and interpretable approach to predicting PGA while effectively quantifying uncertainty associated with geotechnical variability.

Open Access: Yes

DOI: 10.1155/adce/6678669

Effects of Local Soil Profiles on Seismic Site Response Analysis

Ayele Chala R. P. Ray

Publication Name: Periodica Polytechnica Civil Engineering

Publication Date: 2024-03-18

Volume: 68

Issue: 2

Page Range: 403-410

Description:

Local soil conditions play a significant role in the intensity variations of seismic waves during earthquakes. These variations can be either amplified or de-amplified depending on the specific soil conditions. This study aimed to assess the impact of different soil profiles on seismic site responses. The study considered four types of site profiles: sand (Sa), clay (Cl), sand overlying clay (SaCl), and clay overlying sand (ClSa) profiles. To simulate the ground motion, we selected seven sets of strong earthquake records from the European Strong-Motion Database. These records were selected according to Eurocode-8 with a peak ground acceleration (PGA) of 0.24 g, site class A using REXEL computer program. The records were then applied to the bedrock at a depth of 30 meters. Subsequently, a series of 1-D equivalent linear (EQL) response analyses were performed using the STRATA. Amplification factors (AFs) and surface acceleration time histories provided quantitative evaluations for our analysis results. The results demonstrated that site profiles with clay overlying bedrock (SaCl and Cl profiles) exhibited higher seismic amplification and peak ground acceleration in comparison to site profiles with sand overlying bedrock (Sa and ClSa profiles). The maximum median AF is calculated from the SaCl site profile, while the minimum median AF was calculated from the ClSa profile. The relative difference between the maximum and the minimum median AFs was about 33.7%. Based on these results, we can conclude that soft local soils have a pronounced effect on the amplification of seismic waves compared to stiff local soils.

Open Access: Yes

DOI: 10.3311/PPci.22148

Impact of Randomized Soil Properties and Rock Motion Intensities on Ground Motion

Ayele Chala R. P. Ray

Publication Name: Advances in Civil Engineering

Publication Date: 2024-01-01

Volume: 2024

Issue: Unknown

Page Range: Unknown

Description:

Seismic site response is inevitably influenced by natural variability of soil properties and anticipated earthquake intensity. This study presents the influence of variability in shear wave velocity (Vs) and amplitude of input rock motion on seismic site response analysis. Monte Carlo simulations were employed to randomize the Vs profile for different scenarios. A series of 1-D equivalent linear (EQL) seismic site response analyses were conducted by combining the randomized Vs profile with different levels of rock motion intensities. The results of the analyses are presented in terms of surface spectral acceleration, amplification factors (AFs), and peak ground acceleration (PGA). The mean and standard deviation of these parameters are thoroughly discussed for a wide range of randomized Vs profile, number of Vs randomizations, and intensities of input rock motions. The results demonstrate that both the median PGA and its standard deviations across different number of Vs profile realization exhibit a slight variation. As few as twenty Vs profile realizations are sufficient to compute reliable response parameters. Both rock motion intensity and standard deviation of Vs variability cause significant variation in computed surface parameters. However, the variability in the number of records used to conduct site response has no significant impact on ground response if the records closely match the target spectrum. Incorporating the multiple sources of variabilities can reduce uncertainty when conducting ground response simulations.

Open Access: Yes

DOI: 10.1155/2024/9578058

Machine Learning Techniques for Soil Characterization Using Cone Penetration Test Data

Ayele Chala R. P. Ray

Publication Name: Applied Sciences Switzerland

Publication Date: 2023-07-01

Volume: 13

Issue: 14

Page Range: Unknown

Description:

Seismic response assessment requires reliable information about subsurface conditions, including soil shear wave velocity (Formula presented.). To properly assess seismic response, engineers need accurate information about (Formula presented.), an essential parameter for evaluating the propagation of seismic waves. However, measuring (Formula presented.) is generally challenging due to the complex and time-consuming nature of field and laboratory tests. This study aims to predict (Formula presented.) using machine learning (ML) algorithms from cone penetration test (CPT) data. The study utilized four ML algorithms, namely Random Forests (RFs), Support Vector Machine (SVM), Decision Trees (DT), and eXtreme Gradient Boosting (XGBoost), to predict (Formula presented.). These ML models were trained on 70% of the datasets, while their efficiency and generalization ability were assessed on the remaining 30%. The hyperparameters for each ML model were fine-tuned through Bayesian optimization with k-fold cross-validation techniques. The performance of each ML model was evaluated using eight different metrics, including root mean squared error (RMSE), mean absolute error (MAE), mean absolute percentage error (MAPE), coefficient of determination ((Formula presented.)), performance index ((Formula presented.)), scatter index ((Formula presented.)), (Formula presented.), and (Formula presented.). The results demonstrated that the RF model consistently performed well across all metrics. It achieved high accuracy and the lowest level of errors, indicating superior accuracy and precision in predicting (Formula presented.). The SVM and XGBoost models also exhibited strong performance, with slightly higher error metrics compared with the RF model. However, the DT model performed poorly, with higher error rates and uncertainty in predicting (Formula presented.). Based on these results, we can conclude that the RF model is highly effective at accurately predicting (Formula presented.) using CPT data with minimal input features.

Open Access: Yes

DOI: 10.3390/app13148286

Assessing the Performance of Machine Learning Algorithms for Soil Classification Using Cone Penetration Test Data

Ayele Chala R. P. Ray

Publication Name: Applied Sciences Switzerland

Publication Date: 2023-05-01

Volume: 13

Issue: 9

Page Range: Unknown

Description:

Conventional soil classification methods are expensive and demand extensive field and laboratory work. This research evaluates the efficiency of various machine learning (ML) algorithms in classifying soils based on Robertson’s soil behavioral types. This study employs 4 ML algorithms, including artificial neural network (ANN), random forest (RF), support vector machine (SVM), and decision trees (DT), to classify soils from 232 cone penetration test (CPT) datasets. The datasets were randomly split into training and testing datasets to train and test the ML models. Metrics such as overall accuracy, sensitivity, precision, F1_score, and confusion matrices provided quantitative evaluations of each model. Our analysis showed that all the ML models accurately classified most soils. The SVM model achieved the highest accuracy of 99.84%, while the ANN model achieved an overall accuracy of 98.82%. The RF and DT models achieved overall accuracy scores of 99.23% and 95.67%, respectively. Additionally, most of the evaluation metrics indicated high scores, demonstrating that the ML models performed well. The SVM and RF models exhibited outstanding performance on both majority and minority soil classes, while the ANN model achieved lower sensitivity and F1_score for minority soil class. Based on these results, we conclude that the SVM and RF algorithms can be integrated into software programs for rapid and accurate soil classification.

Open Access: Yes

DOI: 10.3390/app13095758

Generation and Evaluation of CPT Data Using Kriging Interpolation Technique

Ayele Chala R. P. Ray

Publication Name: Periodica Polytechnica Civil Engineering

Publication Date: 2023-01-01

Volume: 67

Issue: 2

Page Range: 545-551

Description:

The cone penetration test (CPT) has been the de facto field exploration method in geotechnical engineering for decades. Variations of CPT can measure parameters for seismic, environmental, and hydrological applications. Analyzing response often requires properties in areas that have little or no data. Therefore, given the limited CPT data, it is critical to understand how to accurately estimate the soil properties at unsampled locations. In this paper, we generated soil shear wave velocity profiles using the kriging interpolation technique and assessed their performance using site response analysis. Four kriging interpolation-based shear wave velocity profiles and four additional CPT-based shear wave velocity profiles defined site conditions for response analysis. We performed a series of 1-D equivalent linear site response analyses using STRATA software. The site response analysis results are presented as amplification factors (AF), peak ground acceleration (PGA) profiles, surface spectral acceleration, and surface acceleration time histories. Compared to CPT-based profiles, the results of kriging interpolation-based profiles were evaluated and discussed. The results confirmed the reliability of the kriging interpolation technique in predicting soil parameters at unsampled locations.

Open Access: Yes

DOI: 10.3311/PPci.21865