Base resistance of super-large and long piles in soft soil: performance of artificial neural network model and field implications

Abstract

This study aims to examine the performance of artificial neural network (ANN) model based on 1137 datasets of super-large (1.0–2.5 m in equivalent diameter) and long (40.2–99 m) piles collected over 37 real projects in the past 10 years in Mekong Delta. Five key input parameters including the load, the displacement, the Standard Penetration Test value of the base soil, the distance between the loading point and pile toe, and the axial stiffness are identified via assessing the results of field load tests. Key innovations of this study are (i) use of large database to evaluate the effect that random selection of training and testing datasets can have on the predicted outcomes of ANN modelling, (ii) a simple approach using multiple learning rates to enhance training process, (iii) clarification of the role that the selected input factors can play in the base resistance, and (iv) new empirical relationships between the pile load and settlement. The results show that the random selection of training and testing datasets can affect significantly the predicted results, for example, the confidence of prediction can drop under 80% when an average R2 > 0.85 is required. The analysis indicates predominant role of the displacement in governing the base resistance of piles, providing significant implication to practical designs.

Influence of Settlement on Base Resistance of Long Piles in Soft Soil—Field and Machine Learning Assessments

“…On the other hand, in Case 2, the base resistance is negligible as the soil underneath the pile tip is soft and gives insignificant resistance to the applied load (Figure 1b). This case can be observed in many coastal regions such as the Mekong Delta, Bangkok and Jakarta, where soft soil layers (SPT < 10) can reach up to 40 m [9,10]. In this context, despite using very long piles, the pile tip cannot reach a rigid layer, so there is a minimal contribution from the base to the pile bearing capacity.…”

“…For example, when the base soil is medium and fine sand, the base resistance can slowly increase due to the high compressibility of the soil underneath before reaching a larger magnitude at later stages. In fact, Case 3 is very common in practice, especially when piles, even when very long, cannot reach the rigid layer [9,13]. In this process, the development of settlement (i.e., the displacement of pile head) is the key indicator for the mobilization of the load bearing mechanism.…”

“…Further, the database collected for establishing ML models was relatively limited in past investigations, e.g., only short and medium piles were considered, while mobilized base resistance with increasing load was still overlooked due to the lack of appropriate field tests. Recent efforts have been made to predict base resistance of long piles based on more than 1100 datapoints collected from different real-life projects [9]; however, this study used a conventional basic ANN model, resulting in a relatively medium level of prediction quality (i.e., R 2 = 0.8-0.87). More importantly, the past applications of ML for pile foundation were often criticized for their black-box nature regarding, e.g., the unseen interactions among different features and the lack of validation with existing methods.…”

Predicting base resistance of super-long piles using a random forest model: A case study from Ho Chi Minh city

“…All the pile foundation sample data in this study were obtained from Ho Chi Minh City, Vietnam, as detailed in reference [9]. The project is located at the edge of the Meizhou River Delta, downstream of the Saigon River.…”

Analyzing Laterally Loaded Piles in Multi-layered Cohesive Soils: a Hybrid Bnwf Approach With Validation and Parametric Study

“…To address these challenges, Ouyang et al [59] proposed a flexible solution for the analysis of laterally loaded piles, leveraging the power of machine learning. Machine learning, an emerging artificial intelligence technology, offers distinct advantages in intensive computation and universal approximation capabilities (He et al [37], Huynh et al [42], Hsiao et al [40], Zhang et al [83], Zhang et al [84], Nguen et al [57], [58]). It commonly utilizes neural networks, which are mathematical models with layered structures comprising linear transformations and nonlinear activation functions.…”