No abstract

The establishment of confined concrete strength is an important issue in fiber reinforced polymer (FRP)-confined concrete column. This paper explores the use of Radial Basis Function Neural Network(RBFNN) in predicting the confinedment efficiency of FRP-confined concrete. Based on 362 experimental datas, the RBFNN model with highly non-linear reflection relationship was found and tested by the experimental data. A comparison study between the RBFNN model and four well-known models is carried out, it was found that the RBFNN model could reasonably capture the underlying behavior of FRP-confined concrete and provide better results than other models. The sensitivity analysis of the influential factor is also discussed, it shows that RBFNN-based modeling is a practical method for predicting the confinement efficiency of FRP-confined concrete.

Prestressed high-strength-concrete (PHC) tube-shaped pile is one of the recently used foundations for soft soil. The research on uplift resistance of PHC pile is helpful to the design of pile foundations. A field-scale test program was conducted to study the uplift behavior and load transfer mechanism of PHC piles in soft soil. The pullout load tests were divided into two groups with different diameters, and there were three piles in each group. A detailed discussion of the axial load transfer and pile skin resistance distribution was also included. It is found from the tests that the uplift capacity increases with increasing the diameter of pile. When the diameter of piles increases from 500 to 600 mm, the uplift load is increased by 51.2%. According to the load−displacement (Q−S) curves, all the piles do not reach the ultimate state at the maximum load. The experimental results show that the piles still have uplift bearing capacity.