ANALYSIS OF EXCAVATION SUPPORT SYSTEM WITH SOIL NAIL OF DIFFERENT PROFILES

Abstract

Stabilizing excavations in various soil types involved the implementation of soil nailing, which entailed inserting reinforcement elements, such as nails, into the soil. Conventional soil nailing systems used straight nails, but newer systems utilized helical nails, featuring a twisted shape that offered enhanced stability and load-bearing capacity. Understanding the effectiveness of soil nailing systems required a thorough analysis of their behavior under different conditions. This study's primary aim was to compare and analyze conventional soil nail (CN) and helical soil nailing (HN) systems. Both finite element analysis (FEA) and limit equilibrium methods (LEM) were employed to study the behavior of these systems. The goal was to optimize the performance of helical soil nailing using Response Surface Methodology (RSM) and a Hybrid Deep Belief Network (DBN)-Coot optimization algorithm. The study included conducting pullout tests and analytical methods to compare the pullout behavior of CN and HN in cohesive soil. Initially, stability comparison was achieved by FEA with PLAXIS-2D and theoretical calculations. CN and HN were assessed for their factor of safety using both FEA and LEM methods. Under comparable soil and loading conditions, the findings demonstrated that HN exhibited reduced deformation and a higher safety factor compared to CN. The study was then extended to optimize soil nailing parameters like inclination angle, surcharge pressure, helical pitch, and shaft diameters for HN. The optimization study used an RSM-based box behnken design (BBD) with 40 experimental runs obtained from RSM-BBD. Additionally, a hybrid DBN-COOT machine learning model was developed and trained to predict the pullout characteristics of HN used in this study. The RSM BBD was performed using Design Expert software, whereas DBN-CO was developed using MATLAB. While validating both RSM-BBD and DBN-CO models, 3% more optimization accuracy was achieved from DBN-CO than RSM-BBD due to the use of coot optimization in DBN's weight optimization process. Overall, the study offered significant insights into the behavior of soil nailing systems and underscored the potential of utilizing advanced modeling and optimization techniques to enhance their performance.