Cyclic resistance of sand: Experimental & ANN Modeling
Open Access DepositedThis study explores the stress-strain-strength behavior of Ottawa F-65 sand, with a focus on assessing its cyclic strength against liquefaction. Liquefaction is a phenomenon where saturated sands significantly lose strength and stiffness under applied stress, often caused by earthquake shaking. This issue is a critical challenge in geotechnical engineering and earthquake hazard mitigation, making it essential to understand the conditions leading to liquefaction and the factors influencing its severity for the safe design and construction of structures in seismically active regions. A core aspect of this research involves conducting constant volume (undrained) stress-controlled cyclic direct simple shear tests (CDSS) on Ottawa F65 sand samples at relative densities of 50%, 60%, and 90%. Additional CDSS tests with initial shear stresses of 3, 6, 10, and 15 kPa further investigate the soil response to different pre-existing shear stress levels. These experiments aim to elucidate the effects of relative density and static (initial) shear stress on the soil cyclic resistance. Findings indicate that increasing the initial static shear stress enhances the cyclic resistance of both loose and dense sands. Additionally, this study includes a series of CDSS tests on Silica sand at a targeted relative density of 60%, allowing for a direct comparison between the cyclic resistances of Ottawa F-65 and Silica sand. The findings from this comparative analysis highlight the superior cyclic resistance of Ottawa F-65 sand. To supplement experimental findings, an Artificial Neural Network (ANN) model is developed to predict the stress-strain behavior of Ottawa F-65 sand under the tested conditions. This model offers an approach to predicting soil cyclic resistance, leveraging the power of machine learning to interpret complex patterns in the experimental data. The model underwent initial training using an existing dataset derived from previously available experiments on Ottawa F-65 sand and was used to perform blind predictions on a new set of experimental data collected in this research. Following this, a systematic and incremental training method was employed to further refine the model accuracy and reliability. This advanced training strategy was executed in two distinct phases: the first phase focused on incrementally retraining the model based on variations in relative density, while the second phase concentrated on retraining the model incrementally based on different levels of initial shear stresses. Finally, the ANN predictions were compared with the experimental results to evaluate their accuracy and potential for future applications. Keywords: Soil Liquefaction, Cyclic resistance, Relative density, Initial (static) shear stress, Cyclic Direct Simple Shear (CDSS) tests, Ottawa F-65 sand, Silica Sand, Artificial Neural Network
- All rights reserved
Notice to Authors
If you are the author of this work and you have any questions about the information on this page, please use the Contact form to get in touch with us.
| Thumbnail | Title | Date Uploaded | Visibility | Actions |
|---|---|---|---|---|
|
|
Miremadi_gwu_0075M_16853.pdf | 2024-10-02 | Open Access |
|