Realistic simulation of in-situ stresses is an essential consideration in evaluating the bearing capacities of spread foundations. Therefore, this study aimed to investigate the uplift behavior of large spread foundations by performing a series of centrifuge tests. A total of 12 centrifuge test models were constructed considering the variation in the foundation width (3.5–6.5 m), embedment depth ratio (0.67–1.39), and relative density of dry silica sandy soil (40%, 80%). The measured load–displacement curves indicated that the uplift bearing capacity and the corresponding uplift displacement increased with the foundation width at a certain embedment depth ratio. Two sets of empirical equations were proposed by adopting genetic programming to estimate the uplift resistance factor and uplift displacement with consideration of field stress conditions. The proposed equations reasonably matched these parameters under various conditions reported in the literature. Furthermore, a simplified empirical influence zone was proposed on the basis of the measurement of the ground surface displacement. The empirical influence zone, which was a straight line with an inclination angle of ~0.9 times the soil friction angle measured from vertical, could be considered a significant reference for predicting the failure surface of the shallow foundation under uplift loading.

This study utilized advanced deep learning algorithms to predict consolidation settlement in deep soft clay deposits, with a specific focus on the construction design phase of port facilities. Two innovative hybrid models, namely, the sequence long short-term memory (LSTM)-transformer (SLT) and the parallel LSTM-transformer (PLT) models, were introduced to generate accurate time-settlement predictions by incorporating geotechnical and construction information from sites where preloading was applied. The models were developed and tested using a dataset from study sites in Busan Newport, South Korea. This dataset was constructed through 3D interpolation, which provided a detailed and accurate representation of subsurface conditions. A case study was conducted to evaluate the performance of the model in real-world scenarios. The accuracy of the proposed models was compared with that of traditional methods, including the Hansbo method and a basic transformer model. Results indicated that the proposed models outperformed these traditional methods by producing more accurate predictions. In addition, a parametric study highlighted the effectiveness of the model in capturing the effects of critical factors, such as step loading period, maximum fill height, and clay layer thickness. The SLT and PLT models demonstrated significant potential for enhancing settlement prediction accuracy during the design phase. This improvement in accuracy aids in planning and increases cost effectiveness in projects involving soft deposits.

Aggregate piers have been widely used to increase bearing pressure and reduce settlement under structural footings. The ultimate bearing capacity of aggregate pier-reinforced ground is affected by the soil strength, replacement ratio of piles, and construction conditions. Various prediction models have been proposed to predict the ultimate bearing capacity. However, existing models have shown a broad range of bias, variation, and error, and they are at times unsuitable for practical design. In this study, multiple regression analysis was performed using field loading test results to predict the ultimate bearing capacity of ground reinforced by aggregate piers, and the number and type of the most efficient input variables were evaluated to build a robust predictive model. Accordingly, a multiple regression equation for predicting the ultimate bearing capacity was proposed, and a sensitivity analysis was conducted to identify the effect of input variables. In addition, a deep neural network was applied to estimate the ultimate bearing capacity. The optimal structure was selected on the basis of cross-validation results to prevent overtraining. Prediction errors for two approaches were evaluated and then compared with those of existing models.