Y. K. Chow

NEURAL NETWORK: AN ALTERNATIVE TO PILE DRIVING FORMULAS

Abstract Artificial neural networks are capable of learning complex nonlinear relationships from a large amount of accumulated data, and similar to human brains, are noise and fault tolerant. This unique capacity suggests that neural networks would be very useful in certain geotechnical engineering applications. A back-propagation network is set up and trained to predict the pile bearing capacity from dynamic testing data. The trained network produces better results than a pile driving formula approach. The effects of various network parameters on the network results are examined in detail. The general understanding developed is potentially useful for the application of neural networks in other geotechnical engineering problems.

ANALYSIS OF PILES USED FOR SLOPE STABILIZATION

Piles used for the stabilization of slopes have to be adequately designed to resist the induced lateral loads due to the movement of the unstable slope. In this paper, a numerical method is presented for the analysis of this problem. In this approach, the piles are modelled using beam finite elements. The soil response at the individual piles is modelled using the modulus of subgrade reaction and pile–soil–pile interaction considered using the theory of elasticity. Two case histories, one for single pile and the other for pile group, are analysed which show that the numerical model can predict the general characteristics of the piles reasonably well. The study suggests that the design of the piles based on the computed response from single pile analysis, ignoring group effects, may be unduly conservative.

A variational approach for vertical deformation analysis of pile group

A variational approach for the analysis of vertical deformation of pile groups is presented. The method assumes that the deformation of piles can be represented by a finite series. The method applies the principle of minimum potential energy to determine the deformation of piles. Using this method, an analytical solution for pile groups in soil modelled by the theoretical load-transfer curves can be obtained rigorously. Analysis of field tests indicates that the method can predict the performance of pile groups reasonably well.

NUMERICAL ANALYSIS OF AXIALLY LOADED VERTICAL PILES AND PILE GROUPS

Abstract A numerical method, based on a simplified elastic continuum boundary element method, is presented for the settlement analysis of axially loaded vertical piles and pile groups. The soil flexibility coefficients are evaluated using the analytical solutions for a layered elastic half space. Results are presented and compared with existing published solutions for the following cases: (i) piles in homogeneous soil, (ii) piles in finite soil layer, (iii) piles end-bearing on stiffer layer, (iv) piles socketted into stiffer bearing layer, and (v) piles in Gibson soil. Reasonably good agreement is obtained between the present solutions and existing published solutions.

OPTIMISATION OF PILE GROUPS

Abstract The optimisation of pile groups is studied using an optimisation technique in conjunction with a discrete element method for pile group settlement analysis. It is demonstrated that it is possible to optimise the performance of pile groups by minimising the load differentials between piles and/or differential settlements. This could be achieved by apportioning the lengths of the group piles to meet the requirement of their positions in the group. Numerical examples are presented to illustrate this principle.

Behavior of Pile Groups Subject to Excavation-Induced Soil Movement in Very Soft Clay

A series of centrifuge model tests was conducted to investigate the behavior of pile groups of various sizes and configurations behind a retaining wall in very soft clay. With a 1.2-m excavation in front of the wall, which may simulate the initial stage of an excavation prior to strutting, the test results reveal that the induced bending moment on an individual pile in a free-head pile group is always smaller than that on a corresponding single pile located at the same distance behind the wall. This is attributed to the shadowing and reinforcing effects of other piles within the group. The degree of shadowing experienced by a pile depends on its relative position in the pile group. With a capped-head pile group, the individual piles are forced to interact in unison though subjected to different magnitudes of soil movement. Thus, despite being subjected to a larger soil movement, the induced bending moment on the front piles is moderated by the rear piles through the pile cap. A finite element program developed at the National University of Singapore is employed to back-analyze the centrifuge test data. The program gives a reasonably good prediction of the induced pile bending moments provided an appropriate modification factor is applied for the free-field soil movement and the amount of restraint provided by the pile cap is properly accounted for. The modification factor applied to the free-field soil movement accounts the reinforcing effect of the piles on the soil movement.

Numerical modelling of negative skin friction on pile groups

Abstract Simplified methods for the analysis of socketed pile groups subject to negative skin friction are reported. In these methods, the piles are modelled using discrete elements with an axial mode of deformation. The soil behaviour is modelled using a hybrid approach in which the soil response at the individual piles is modelled using the subgrade reaction method while pile-soil-pile interaction is determined using elastic theory. The main difference between the methods lies in the manner in which pile-soil-pile interaction is determined. The accuracy of the methods is assessed by comprehensive comparisons with more rigorous theoretical solutions available. The study shows that: (a) the solutions obtained by the continuum model using Chan et al. s solution are in close agreement with the rigorous solutions, (b) for practical pile groups, the layer model can give reasonable solutions provided the socketing is not too deep and (c) the continuum model using Mindlins solution predicts reasonable downdrag forces but seriously under predict the pile head settlements.

Analysis of rectangular thick rafts on an elastic half-space

Abstract This paper presents the Ritz method for the settlement analysis of rectangular thick rafts resting on a homogeneous, elastic half-space. As the considered raft is thick, the Mindlin plate theory has to be used to model the raft in order to allow for the effect of transverse shear deformation in the raft as it bends under transverse loading. A penalty functional is introduced in the Ritz formulation to satisfy the natural boundary conditions of free edges. The correctness of the Ritz formulation and the software developed was established by the close agreement of bending results of the raft foundation obtained by the present Ritz method and those determined by previous researchers. The effect of transverse shear deformation on the bending results is examined for various raft thicknesses and loading conditions.

Monitoring of dynamic compaction by deceleration measurements

Abstract A method of estimating the degree and depth of improvement in the dynamic compaction of loose granular soil by matching the computed decelerations of the pounder from a numerical model to the measured decelerations is presented. This method uses a simple one-dimensional wave equation model in which the soil beneath the pounder is represented by an elastic laterally confined soil column while the surrounding soil is represented by a series of springs and dashpots. The spring simulates the dynamic soil stiffness while the dashpot accounts for the radiation damping effect. Despite the simplifying assumptions in the model, the computed results for the degree and depth of improvement show an encouraging measure of agreement with available data from a laboratory test and field measurements from a dynamic compaction site. The proposed method is potentially useful for the monitoring of dynamic compaction of loose granular soil. And like any new technique, further verifications of the approach are necessary to develop it into a reliable method.

Severe Damage of a Pile Group due to Slope Failure

AbstractA pile group consisting of four cast-in-place concrete piles was instrumented to measure the induced bending moment along the piles due to excavation of an adjacent slope. The green field lateral soil movement profiles and the lateral pile deflection profiles were monitored by in-soil and in-pile inclinometers, respectively. The unexpected early arrival of a rain storm prior to the year-end monsoon caused the failure of the slope, which resulted in the severe damage of this pile group located at the slope crest. This paper examines the prefailure and postfailure behavior of the pile group and it is demonstrated that considering the uncracked and cracked bending stiffness of the piles is vital when evaluating the progressive damage of the pile group due to slope failure.