1991 & 1992 Higher Degree Theses

Click on the thesis title to read the summary

Al-Douri, R.H., Behaviour Of Single Piles And Pile Groups In Calcareous Sediments

Alehossein, H., Numerical Analysis Of Jointed Rock Masses

Chin, J.T., Axial Pile Response In Calcareous Sediment

Clarke, MJ., The Behaviour of Stressed-Arch Frames

Ding, X-L, Precise Engineering Surveying - Application to the Measurement of Large Scale Steel Silos

Fu, D., Damage And Fracture Of Plain Concretes

Hsi, J.P., Analysis of Excavation Involving Drawdown of the Water Table

Huang, X, Two- And Three-Dimensional Elastic Crack Interaction Problems

Kim, D.-H.., Snowdrificing Around and Wind Loads on Buildings for Antarctic Environment

Kwon, Y.B., Post-Buckling Behaviour Of Thin-Walled Channel Sections

Wong, H.K.W., The Effect Of Bridge Approach Slabs On Pavement Deformation

Xu, L.Y., Vibration Control Of Wind-Excited Tall / Slender Structures

Zhang, B.Q., Finite Layer Analysis Of Foundation Behaviour

Zhang, Q., Stability Behaviour Of Steel Silos Filled With Bulk Solids

Zhao, X.-L., The Behaviour of Cold-Formed Rectangular Hollow Section Beams Under Combined Actions

Al-Douri, R.H.
Behaviour Of Single Piles And Pile Groups In Calcareous Sediments
PhD Thesis Supervisors: Professor H.G. Poulos and Associate Professor JC. Small

This thesis presents the results and analysis of an extensive study of the behaviour of model piles and pile groups in New North Rankin calcareous sand under static and cyclic loading, using different test boundary conditions. The results and experience gained from this study may be utilised in the theoretical analysis and design of offshore pile foundations.

To investigate soil strength and soil-metal interface behaviour, monotonic and cyclic shear box tests have been performed on different types of calcareous sands and one silica sand under constant normal load (CNL) and constant normal stiffness (CNS) conditions. The reduction in shear stress in the CNS tests may be predicted from the results of CNI, tests. Thus CNI, tests may be used to estimate the degradation of skin friction during cyclic pile loading.

Model pile tests on a single pile jacked into medium-dense and dense calcareous sand have been performed under various overburden pressures. Two different sizes of test vessel with different vessel base conditions have been used in these tests. There were two major aspects of these tests:

a) 'Ihe effects of initial density, overburden pressure, vessel diameter, and vessel base conditions on the pile shaft friction under static loading, and the degradation factor for skin friction (D_T) under cyclic loading have been investigated. The influence of cyclic displacement amplitude and number of cycles on D_T was also studied. The results showed that the initial density and overburden pressure have a significant effect on the jacking force required to install the pile, and on the static skin friction and the soil modulus obtained for static loading. It has also been found that the initial density, overburden pressure, cyclic displacement amplitude, and number of cycles significantly influence the degradation factor for skin friction (D_T under cyclic loading. The distribution of interface stress along the pile jacked into a sand consolidated by applying overburden pressure on both top and bottom ends of the sand was more uniform than that from applying overburden pressure only at the top surface of the sample.

b) The influences of different cyclic load levels, number of cycles and different types of cyclic loading (uniform and "storm") on the accumulation of cyclic displacement have been investigated. The results have shown that the accumulated displacement of the pile increases as the cyclic load level and number of cycles increase, but the increase depends on load level more than on number of cycles. For storm loading, the accumulated displacement in the first cycle appeared to be higher than for subsequent cycles in each parcel of load.

Model tests on pile groups jacked into calcareous sand consolidated in the large test vessel under various overburden pressures have been performed. The results of these tests have shown that, for medium-dense sand, the jacking force increases as the number of piles increases. The group efficiency (n) also increases as the number of piles increases. The results have also shown that the degradation factor increases with increasing cyclic displacement and number of cycles. There is only a small effect of the number of piles on the skin friction degradation factor " D_T ".

The accumulated displacement behaviour of a pile group was found to be similar to that of a single pile, with increases in displacement caused by increasing load level being more significant than increases arising from increasing numbers of cycles. The results showed a small influence of the number of piles on pile capacity and group settlement for both uniform and non-uniform cyclic loadings.

From the results of interaction tests between two piles, it has been found that there is some interaction between two piles for small spacings and that sand density and overburden pressure both have an effect on the direction and magnitude of the deflection of an unloaded pile caused by the loaded pile.

The predicted and measured results for the residual load in a single pile and a pile in a group were found to be similar. The predicted and measured results for load-deflection response for a single pile and a group also showed good agreement.

The boundary element program "SCARP' has been used to analyse the response of a single pile and pile groups to static and cyclic loading. It was found that it was possible to predict, reasonably well, the residual load after jacking, and the static load-deflection response of the piles.

Provided that appropriate values of the input parameters are used, (especially Young's modulus (ES)), the accumulated displacement for both uniform and non-uniform cyclic loading predicted by SCARP has been found to be in good agreement with the single pile results, although agreement for pile groups was less satisfactory. The differences between predicted and measured results increase with the number of piles in a group. This behaviour may be attributed to the assumption of soil homogeneity between the piles in the group, which is used in the program "SCARP'. On the basis of these comparisons, suggestions are made for modifications of the SCARP program and also for the determination of the appropriate input parameters.

Alehossein, H.
Numerical Analysis Of Jointed Rock Masses
PhD Thesis - Supervisor: Professor J.P. Carter

When analysing the mechanical behaviour of a jointed rock mass, a suitable approach has to be found for dealing with the influence of the joint system on the overall mechanical behaviour of the rock mass. In attempting to model this behaviour at least three approaches are possible: continuum, semi-continuum and discontinuum methods of analysis, e.g. implicit (also called "ubiquitous") and explicit methods of including the joints and the distinct element method may be regarded respectively as examples of the above-mentioned approaches.

In this thesis a variety of numerical approaches for the analysis of jointed rock masses has been considered. In particular a detailed examination is made of the application of the boundary element method to perfectly elastic joints and of the finite element method to both perfectly elastic and elastoplastic joints.

The work may conveniently be divided into the following sections:

(a) The presentation of a general implicit mathematical model for both jointed rock masses and reinforced jointed rock masses. This investigation includes the development of the constitutive behaviour for a jointed rock mass containing any finite number of joint sets in which any of the joint sets and/or the surrounding intact rock may undergo plastic deformation. The plastic failure mechanism is described by a general yield condition and plastic deformation by an arbitrary flow rule. This implicit model can be applied conveniently to problems in reinforced jointed rocks containing any finite number of reinforcement sets. The model can be used for both two and three dimensional boundary value problems in jointed rocks.

(b) The proposal of numerical techniques such as the finite and boundary element methods to be used to obtain approximate solutions to boundary value problems. The numerical methods are verified by means of comparisons between exact and numerical solutions. In particular, the finite element method has been examined for the analysis of not only implicit (ubiquitous) joints but also explicit joint systems. Finite element predictions of collapse load for foundations on jointed rocks have been made and compared with analytical predictions. In order to reduce one spatial dimension of three dimensional problems in prismatic jointed rocks, complex Fourier integral transforms have been introduced and incorporated into the finite element technique applied to elastic rock masses.

(c) The numerical methods referred to above are then used to examine several complex boundary value problems of interest in geotechnical engineering. These include both two and three dimensional surface loading, rigid strip footings, various long, deep and shallow underground tunnels, long rectangular and trapezoidal open excavations and long open excavations near existing tunnels in jointed rock masses with a wide variety of jointing conditions.

(d) Proposal of general elastic solutions for complex boundary value problems in jointed rocks. Using the above-mentioned numerical methods, general elastic solutions, usually in terms of influence coefficients, have been presented for several practical footing and excavation problems in jointed rock masses. These elastic solutions can conveniently be applied to similar problems in jointed rocks having any arbitrary material properties and initial ground stress states.

(e) Comparison of continuum and discontinuum approaches. Using the finite element method, implicit and explicit joint models are compared for several footing and excavation problems. Also, for some excavation problems the continuum and discontimmin finite element models and the discrete element method have been compared. Also, for a few excavation problems, the solutions prepared in this work are compared with available field data.

It was found that reasonable agreement can be obtained between various predictions of the behaviour of jointed rocks, even when quite different numerical techniques are employed. Although the implicit joint model cannot correctly model the localisation of deformations that occurs at the joints, it is still able to predict the overall pattern of deformation satisfactorily. As expected, predictions from the implicit model will improve as joint spacings reduce. However, because the implicit model is so efficient in terms of computing time, it is suitable for the preliminary or comparative analysis of problems in both jointed and reinforced jointed rock masses.

Chin, J.T.
Axial Pile Response In Calcareous Sediment
PhD Thesis - Supervisor: Professor H.G. Poulos

Some studies of the axial response of piles in calcareous sediment have been conducted in this thesis. The experimental part of the work described in the present thesis is aimed towards further understanding the mechanisms governing the cyclic response of piles in calcareous sediments. For the present study, the axial cyclic response of piles in 'Bass Strait' calcareous sediment is studied using a large model pile test facility. Model instrumented single piles of 50 mm and 100 mm diameters are utilised for the present test program. 'Me cyclic response of the model piles when subjected to two-way displacement-controlled cyclic loading is investigated. The results of this series of tests are assessed in order to shed further light on the most appropriate model governing the degradation of skin friction capacity under cyclic loading conditions. It is found that the degradation of skin friction is governed by the normalised (to pile diameter) value of the "cyclic slip displacement".

For cyclic loading analysis, where repeated iterations are required, the use of a simple and computationally efficient approach is highly essential and desirable. To this end, the development of a simple and efficient "t-z" static and cyclic pile analysis approach is presented in the numerical part of the work undertaken in the present study. The utility of the analysis approach developed is demonstrated through some parametric solutions. In particular, the important parameters affecting the computed cyclic pile response are identified. The results of a comparative study of a few cyclic axial pile loading analyses further highlight the inconsistent predicted responses that can be obtained form the different analysis approaches. Overall, it can be concluded that our predictive ability for the static axial response of both single piles and pile groups is satisfactory. For cyclic loading analysis, it is shown that our predictive ability is far less satisfactory.

Clarke, MJ
The Behaviour Of Stressed-Arch Frames
PhD Thesis - Supervisor: Professor GJ. Hancock

Stressed-arch frames are an innovative development in steel structures involving the erection of prefabricated truss frames by a post-tensioning stressing procedure. The erection process results in substantial bending of the top chord, often to the extent that plastic deformation occurs. If the top chord is plastically deformed, its strength in the frame under the action of Eve load is difficult to quantify rationally using the member design rules of specifications for the design of steel structures.

The objective of this thesis is to investigate the behaviour of stressed-arch frames, particularly the top chord. Both theoretical and experimental studies are described. These studies enable the strength of the top chord, as a function of the degree of initial curvature and the geometrical slenderness, to be quantified reliably for application to design.

A finite element nonlinear analysis is developed to simulate the behaviour of stressed-arch frames. Within the assumptions applicable to the classical analysis of in-plane beam-column behaviour, the finite element nonlinear analysis described in this thesis performs a refined analysis of arch and frame behaviour and satisfies the requirements of the Australian limit states Steel Structures Standard, AS 4100-1990, for an advanced structural analysis. The element used in the analysis is applicable to both straight and arched members. Formulations of geometric non-linearity based on the total Lagrangian, updated Lagrangian and co-rotational approaches are included in the analysis. Material inelasticity is implemented through a distributed plasticity model, involving subdivision of the cross-section into a grid of monitoring points. The nonlinear equilibrium equations furnished by the principle of virtual displacements are solved using an incremental-iterative approach. Several different incremental-iterative strategies that enable limit points to be overcome are described.

Two experimental investigations into the behaviour of the top chord of stressed-arch frames are presented. The test frames for both test programmes were provided by Strarch International Limited, of Melbourne, Australia. The first investigation involved the testing of frame subassemblages comprising two top chord segments and one bottom chord segment. Tests were performed for different levels of top chord initial curvature and geometrical slenderness, and provided lower bound estimates of top chord strength compared to the corresponding strengths in prototype frames. The second test programme was performed on small-scale (15.24 m span) stressed-arch frames subject to simulated uniform gravity loading, and therefore enabled the structural behaviour of complete systems, as opposed to isolated subassemblages, to be investigated.

The structural behaviour and ultimate strengths observed experimentally are compared with the theoretical predictions of the finite element nonlinear analysis. For the tests on complete frames, the analysis simulated the complete load-deformation history of the test frames, commencing with the top chord in the assembled (horizontal) position, proceeding through the erection (stressing) stage, and concluding with the structure in its ultimate and post-ultimate response under simulated gravity loading.

The appropriateness and applicability of steel structures standards and specifications worldwide for the design of the top chord of stressed-arch frames is discussed. An extensive parametric study of the strength of the top chord in stressed-arch frames as a function of the initial curvature, the geometrical slenderness and the material yield stress is presented. The results of this theoretical investigation constitute the basis of a simple and economical design procedure for the top chord of stressed arch frames based on the rationale of advanced analysis.

Ding, X-L
Precise Engineering Surveying - Application to the Measurement of Large Scale Steel Silos
PhD Thesis - Supervisor: Dr R. Coleman

The buckling strength of steel silos under axial compression is very sensitive to the small geometric imperfections in the structures. The information on the amplitude and the form of the imperfections in steel silos is required for both the assessment and the prediction of the buckling strength of existing and future designed silo structures. The only way to obtain information on the imperfections in full scale steel silos is to carry out comprehensive measurements of these structures.

Measurement of imperfections in full scale steel silos has never been conducted comprehensively, in terms of the measurement precision and the amount of detail measured of the shapes of a silo wall, due to the substantial difficulties associated with the task and the lack of suitable instrumentations and techniques for the measurement of such structures. Ibis task has to date been considered by civil engineers as too difficult and too expensive.

A new measurement technique is developed which overcomes most of the difficulties in the measurement of the silo structures. The technique employs a newly designed profile measurement system together with instrumentations such as theodolites and computers, to gather measurement data on site. The imperfection data is then reduced and analysed using the developed data reduction theories associated with the measurement system.

Two kinds of observations, measured silo wall profiles and theodolite readings, are combined in a rigorous least squares adjustment model to produce the best estimate of the three-dimensional coordinates of a great number of sample points on a silo wall.

Corrections are applied to all the geodetic observations in the network for the effect of the curvature of the earth, based on a model that simplifies a local potential surface in the survey area with a best-fit spherical surface.

A new concept called the effectiveness of observations is proposed to quantitatively assess the effect of one or a group of observations in a network. A new outlier detection technique, based on this concept, is presented which enables more than one outlier to be identified in a single iteration of adjustment, applying the standard r test designed for the detection of single outliers.

Field measurement is carried out for the measurement of three full scale steel silos employing the measurement system developed in this research. The measurement procedures, the results as well as the problems encountered are discussed in detail.

The shapes of the measured silo walls from the combined least squares adjustment of profile measurement data and theodolite observations, or from least squares fitting of profile data alone, are finally represented with a double Fourier series. The spectra of the measured wall imperfections are determined in this Fourier analysis to show the dominant forms of imperfections in the measured silo walls. These results represent the first data set available for quantifying modes of full-scale silo imperfections.

Fu, D.
Damage And Fracture Of Plain Concretes
PhD Thesis - Supervisor: Professor B.L Karihaloo

The thesis deals with the application of damage and fracture mechanics to the description of the complete load-deformation response of plain concretes under uniaxial and biaxial tension. The main features of the work are:

(i) the establishment of the threshold strain (stress) describing the transition from linear to non-linear response. Expressions for the uniaxial/biaxial threshold strain, flexural threshold strain, volumetric threshold strain (stress), and effective threshold strain (stress) are given;

(ii) the description of the non-linear response which begins at the threshold strain (stress) and ends at the ultimate tensile load. An anisotropic damage model, three isotropic damage models (differing only by the shape and distribution of the damage), an orthotropic damage model and three isotropic damage models with damage nucleation are presented;

(iii) the description of the tension softening response which begins at the ultimate tensile load and ends at the complete rupture. Ibis response is described via a combination of damage and fracture mechanics by modelling the localised deformation along the eventual fracture planes with an array of circular holes with edge cracks;

(iv) an experimental investigation of the uniaxial tensile behaviour of two concrete mixes.

It is found that the models developed in the thesis can satisfactorily describe the pre-peak stress-strain behaviour of these mixes, as well as predict their probable post-peak tension softening response.

Hsi, J..P.
Analysis of Excavation Involving Drawdown of the Water Table
PhD Thesis - Supervisor: Associate Professor J.C. Small

In recent years, demands for high rise buildings in densely populated areas have increased. Construction of such buildings often begins with excavation for the basement level. Ground movements are usually generated accompanying an excavation, and so it is necessary to predict these movements correctly before commencement of the work so that safety of the excavation can be ensured.

In practice, excavations are often carried out in saturated soils. In such cases, groundwater flow can be induced towards the cut as the excavation boundary becomes free draining and then the water table begins to fall. Due to the lowering of the water table, the effective stress in the soil will increase and then an additional ground settlement will be generated. Current methods for the analysis of excavation assume that the soil is either undrained or drained. For both cases, the soil is considered as being saturated and it is assumed that the groundwater table remains unchanged. Therefore the effect of drawdown of the water table accompanying an excavation cannot be accounted for by these methods and so the behaviour of the soil surrounding the excavation cannot be simulated correctly.

The objective of this thesis is to develop a fully coupled numerical method which can be used for the analysis of excavation taking into account the drawdown of the water table.

The work involved in this study includes the following:

A finite element formulation for problems of consolidation, transient free surface seepage and excavation.
Development of the fully coupled governing equations which allow for the calculation of the displacement and the pore pressure simultaneously where the groundwater table is allowed to fall.
Implementation of the Modified Cam Clay model for simulating the elasto-plastic behaviour of the soil.
Programming of a finite element program EXCA2 which is based on the 8-noded isoparametric element and which is able to perform plane strain and axi-symmetric analyses in ether elastic or elasto-plastic materials.
Verifications of program EXCA2 for problems of consolidation, excavation, drawdown of the water table and the Modified Cam Clay model.
Parametric studies in which the factors that influence the drawdown. behaviour of the water table are investigated.
Investigation of excavation behaviour involving drawdown of the water in elastic and elasto-plastic materials.
Application of the fully coupled method to the back-analysis of full-scale instrumented excavations where drawdown of the water table was observed to have occurred.

In this thesis, the fully coupled numerical method for the analysis of excavation is firstly developed. The drawdown of the water table during excavation can be accounted for through the use of the concepts of the Residual Flow Procedure. The stress release due to excavation can be correctly simulated by computing and applying nodal forces which can be derived by use of a virtual work formulation. A finite element program EXCA2 is then developed for implementing the coupled method. Modified Cam Clay theory is incorporated for simulating elasto-plastic soil behaviour. Verifications are carried out to prove that the program can yield correct solutions and parametric studies are carried out to investigate the factors which affect the drawdown of the water table. Two illustrative excavation examples are then used to demonstrate the capability of the program. Finally, case studies of two excavation histories are carried out and the practicability of this method is fully demonstrated.

Huang, X
Two- And Three-Dimensional Elastic Crack Interaction Problems
PhD Thesis - Supervisor: Professor B.L Karihaloo

This thesis is concerned with the toughening of quasi-brittle materials such as concrete, rock and ceramics. The toughening in these materials is a result of the interaction of cracks with other sources of internal strain. Among these sources are phase transformation, as in zirconia based ceramics, microcracks resulting from the release of residual stresses as in ceramics, rocks and concrete.

The interaction problems are studied both in two- and three-dimensional approximations. In the two-dimensional approximation, the problems of interaction between a semi-infinite crack and sources of internal strain, as well as of interaction between the sources themselves are considered. In the three-dimensional approximation problems of interactions between a half-plane crack and sources of internal strain, as well as of interaction between the internal sources themselves are studied.

For the mathematical description of the above problems recourse is taken to two- and three-dimensional weight functions for the cracked body, as well as singular integral equations. Moreover, in one instance a perturbation technique is also employed. In several instances, besides the full numerical solutions, simple approximate solutions are also given. Whenever possible, the analytical/numerical solutions are compared with experimental results, and are shown to be in good agreement.

Kim, D.-H..
Snowdrificing Around and Wind Loads on Buildings for Antarctic Environment
PhD Thesis - Supervisor: Associate Professor K.C.S. Kwok

A closed-circuit Snowdrift Wind Tunnel (S.W.T) for snowdrifting simulation was constructed. A scaled atmospheric boundary layer wind without model snow was generated, corresponding to terrain category type 2 (Australian Standard, AS 1170.2-1989) to represent the flow over a typical coastal area of Australian Antarctic Territory in summer. Using this flow, a number of model snow particles were tested to select a suitable model snow by comparing with field data. Sodium bicarbonate was found to be the best.

A turbulent boundary layer wind with the model snow, representing the winter conditions of coastal sites of Australian Antarctic territory, was simulated. It was found that the introduction of the model snow particles transformed the turbulent flow characteristics from that of terrain category type 2 to type 1. Based on the above results, the similitude parameters of snowdrifting simulation were reviewed and applied to the experimental conditions.

The wind loads (mean and standard deviation of lift and drag) and snowdrift were measured simultaneously in the model snow blowing condition for the seven models with different corner geometries, incident wind angles and heights above ground in the SW.T. The results indicated that the models with radiused corner geometries created less snowdrift and lower wind loads than those with chamfered corner geometries. The larger the chamfer or radius, the less the snowdrift volume and the lower the wind loads.

Wind load measurements were performed without model snow in the No.1 Boundary Layer Wind Tunnel (No.1 B.L.W.T.). A turbulent boundary layer wind, representing the prototype summer conditions, was simulated. Wind loads measurements (mean and standard deviations of lift and drag) on seven models with different corner geometries, incident wind angles and heights above ground were conducted. the results showed that models with chamfered and rounded corner geometries gave a better reduction of force coefficients than the reference model, which was of a simple rectangular shape. It was also observed that the buildings with radiused corner geometries gave lower force coefficients than those with chamfered corners.

Kwon, Y.B.
Post-Buckling Behaviour Of Thin-Walled Channel Sections
PhD Thesis - Supervisor: Professor GJ. Hancock

Thin-walled channel sections may undergo a mode of buckling called distortional in which a lip stiffened flange of the section rotates about the flange-web junction. If the sections are composed of high strength steel, then there may be a significant post-buckling reserve of strength beyond the elastic distortional buckling stress in a similar manner to that which normally occurs for local buckling.

A nonlinear elastic analysis based on the spline finite strip method has been developed for studying the post-buckling behaviour of thin-walled sections. The method can handle local, distortional and overall buckling modes in the post-buckling range and the interaction between them. It allows for geometric imperfections, arbitrary loading and non-simple boundary conditions. The accuracy is confirmed by studying several plate, section and shell problems for which solutions are known.

The results of compression tests of thin-walled channel section columns formed by brake-pressing are described. Two different section geometries, a simple lipped channel and a lipped channel with an intermediate stiffener in the web, were tested between fixed-ended boundary conditions. The geometries and yield strength were chosen to ensure that a substantial post-buckling strength reserve occurred in the distortional mode and that local buckling might occur simultaneously at a shorter wavelength. The nonlinear spline finite strip method is applied to the prediction of the post-buckling behaviour and the influence of the interaction between buckling modes for the test channel columns.

Design curves for the maximum strength of different columns undergoing distortional or mixed modes of buckling are proposed. Two basic design methods to account for the local and distortional buckling or mixed local-distortional buckling mode are also proposed. The test results are compared with the proposed methods and are also compared with the Australian Standard, the American Specification and European Recommendations for the design of cold-formed steel structures

Wong, H.K.W.
The Effect Of Bridge Approach Slabs On Pavement Deformation
PhD Thesis - Supervisor: Associate Professor JC. Small

Because there is a difference in stiffness between a bridge deck and the materials used in the approach embankment, traffic loading will cause the pavement to deform relative to the bridge deck and form a 'bump' at the end of the bridge. This causes drivers' discomfort and accelerated damage to the pavement. The use of concrete approach slabs constructed so as to slope down from the bridge deck beneath the pavement has been proposed as a means of alleviating this problem.

In this thesis, experimental and theoretical work have been carried out to investigate the behaviour of approach slabs under repeated loading.

The work is broadly divided into six sections:

1. The development of a testing facility for the testing of model approach slabs under controlled conditions.

2. The assessment of the capability and performance of the testing facility.

3. The planning and conduct of the laboratory testing for the investigation of the deformation behaviour of pavements subjected to traffic loading.

4. The formulation of a three-dimensional finite element program using Fourier transforms for the analysis of the problem.

5. The planning and conduct of monotonic and cyclic drained triaxial tests for the investigation of residual strain behaviour.

6. The analysis of the residual strains in the model embankments and the development of a numerical method for the prediction of permanent displacements.

A testing facility has been developed to provide simulated traffic loading on model pavements underlain by approach slabs. Three series of tests have been carried out to investigate the effect of the orientation of the approach slab on the deformation behaviour of the pavements.

It is found that with the use of inclined slabs, the deformation in the pavement surface is more gradual than for the case where a horizontal slab is used. In other words, the bump usually formed at the end of a horizontal slab can be eliminated.

A finite layer method is developed in this thesis, which provides an efficient means for investigating the cyclic stressed generated in the road embankment by the moving wheel. The effect of the approach slab orientation and the pavement stiffness on the soil response can be examined.

Cyclic triaxial tests have been conducted using a fully automated GDS triaxial testing system. The aim of these tests was to investigate the soil residual strain behaviour. The cyclic stress paths have been estimated from the finite layer analyses.

The triaxial test results showed that the residual axial strain after the first loading cycle increased linearly with the logarithm of the number of cycles (for the 50 loading cycles carried out for each test). This relationship was found to be dependent on the stress level and the gradient of the stress path.

The residual stains in the model pavements have been backfigured from two of the tests. The residual strains after the first loading cycle were found to vary non-linearly with the logarithm of the number of loading cycles. However, the non-linearity was less pronounced for the initial 50 cycles or so.

A stress path method was developed to utilise the cyclic test data for the prediction of residual axial strains in the pavement layers. Based on this method, the residual strains in the pavement layers after the first loading cycle can be computed. The residual deformations are then obtained by integrating the strains obtained for all the layers. This method was applied to one of the model pavement tests and the results are found to agree reasonably well with the observed values.

There are a number of improvements that can be made to this method of prediction, for instance, the incorporation of a non-linear stress-strain law to predict the first cycle deformation. Also, the measurement of the actual stresses acting in the model pavements would be useful in determining the appropriate stress paths for subsequent cyclic tests.

Xu, L.Y.
Vibration Control Of Wind-Excited Tall / Slender Structures
PhD Thesis - Supervisor: Associate Professor K.C.S. Kwok

The results of a theoretical and experimental investigation of the vibration control of wind-excited structures axe presented in this thesis. ]be investigation was divided into two parts. The first part was based on aeroelastic model tests of tall buildings in the wind tunnel. It included experiments and analyses of alongwind and crosswind vibration control of tall buildings by passive mass dampers (TMDs); torsional vibration of tall buildings and its control by TMDs; a semi-analytical method of performing parametric study of TMDs; alongwind, crosswind and torsional mode shape correction factors; and a prediction procedure for assessment of the effectiveness of an active mass damper control system. The second part was a theoretical study of vibration control of wind-excited slender structures by using computation techniques. It included the possible application of tuned liquid column dampers in reducing the response of wind-sensitive structures; and soil-structure-mass damper interaction under wind loading.

The aeroelastic test of alongwind and crosswind vibration control was carried out on a CAARC model in a suburban boundary layer wind model, by using a conventional aeroelastic test rig which simulated two fundamental sway modes. The torsional vibration control experiments were performed on a rectangular tall building model in an open country boundary layer wind model, by using an aeroelastic test rig designed for torsional vibration only. The results obtained from the aeroelastic test programs demonstrated the effectiveness of the TMD systems in suppressing the wind-induced dynamic responses of tall buildings. lle TMD system reduced the vibration caused by alongwind turbulence excitation, crosswind wake excitation or torsional excitation by 20%-45% provided that the parameters of the TMDs were properly selected. The TMD system was found to be even more effective in reducing the vibration caused by lock-in excitation by a factor of 2 or more.

A series of wind tunnel model tests were also conducted to investigate the mechanism of torsion excitation and torsional response of tall buildings, and the sensitivity of the torsional response to eccentricity between centres of twist and building geometry. With the angle of wind incidence normal to the wide face of the building, vortex shedding is the dominant mechanism of torsional excitation. With the angle of wind incidence normal to the narrow face, the incident turbulence and the shear layer re-attachment intermittencies are two important excitation mechanisms. At a reduced wind velocity of 8, the maximum dynamic torque for the eccentric model increased by 30% and the maximum mean torque increased by a factor of 2.8, compared with the values of the basic model.

Based on direct measurement in the wind tunnel of wind-induced response or excitation spectra of the plain building without TMDs, a semi-analytical method of performing parametric study of the TMD was proposed. The results obtained by this method were in good agreement with the corresponding experimental results. In contrast, the conventional parametric study method, which is based on a white noise excitation model, usually overestimates the effectiveness of TMDs for most real situations. The effectiveness of the TMD was also found to be dependent on the type of external wind excitation.

The semi-analytical method was also used to investigate the feasibility of a suboptimal active mass damper vibration control system. Analytical results showed that the effectiveness of passive tuned mass dampers can be considerably enhanced by the inclusion of the suboptimal active control system. The analytical procedure provides a method of selecting the most beneficial control parameters which result in a larger reduction of the building and damper responses by using a small control force or moment.

Sources of error in the aeroelastic modelling technique, caused by the discrepancy between the building model and prototype mode shapes, were discussed. Three mode shape correction factors, for alongwind, crosswind, and torsional responses respectively, were suggested to adjust the experimental response results to the corresponding prototype values. The results obtained by the proposed expressions were in reasonable agreement with the available experimental results.

In the theoretical study of vibration control of wind-excited slender structures, the structure was modelled as an n-degree-of-freedom lumped mass system taking into account both bending and shear. The soil behaviour, including footing embedment effect, was characterised by a known frequency-dependent complaint matrix. A transfer matrix formulation for non-periodic structures was developed to analyse the effects of liquid dampers and soil compliancy on wind-induced response of slender structures. The numerical computer accuracy of direct matrix multiplication was investigated and the results indicated that the accuracy of the computed results can be guaranteed. Numerical examples showed that tuned liquid column damper systems, which have significant practical advantages, can achieve the same level of motion reduction as passive mass dampers. The numerical examples also showed that soil compliancy can significantly affect the structural responses and the effectiveness of the dampers, depending on the properties of the soil, the properties of the structure, the nature of the excitation and the type of structural response.

Zhang, B.Q.
Finite Layer Analysis Of Foundation Behaviour
PhD Thesis - Supervisor: Associate Professor J.C. Small

The behaviour of foundations is of major interest to geotechnical engineers who must be able to predict the amount of settlement which will take place under earthworks or structures. Also of interest is the rate at which consolidation takes place and what effect this has on the forces and bending moments in the structure.

To simplify the analysis of foundation settlements, one and two dimensional theories have been developed. While these methods are suitable in some cases, they are not suitable for many problems which are three-dimensional in nature.

It is the purpose of this thesis, therefore, to extend the range of problems which can be solved by developing methods which can be applied to three-dimensional problems, but which retain the simplicity of the one and two dimensional techniques.

For this purpose, the finite layer method is introduced for the analysis of foundations on elastic soil. The method can be used to reduce the three dimensional problem to one involving only two spatial dimensions through a combination of Fourier transforms and finite element techniques. The approach allows efficient and easy computation of elastic solutions for a soil subjected to surface loadings by reducing computer storage and saving computational time.

Further, the finite layer method is extended to the analysis of consolidating soil where it is used to set up the equations governing three-dimensional consolidation of a poro-elastic medium. The practicality of the finite layer method is demonstrated by determining the consolidation settlements of a clay layer under three-dimensional loadings of general shape.

The finite layer method is then used for the analysis of raft foundations. By considering the compatibility of the vertical displacements between the soil and the foundation, a numerical method is developed for solving soil-foundation interaction problems. The distributions of differential settlements and bending moments in the foundation can be obtained. This theory is extended to the analysis of the time-dependent behaviour of foundations resting on consolidating soil.

However, the stiffness of a foundation may be influenced by the structure that it supports and so the structure has been included in the analysis. For the analysis of the structure, beam elements are used to represent a rigid jointed space frame. The stiffness of the structure is directly added to that of the foundation, which makes the soil-foundation-structure an integral unit. By using this method, elastic solutions are obtained for soil-foundation-structure interaction problems, and example solutions are presented.

Zhang, Q.
Stability Behaviour Of Steel Silos Filled With Bulk Solids
PhD Thesis - Supervisors: Professor JM. Rotter, and A/Professor P. Ansourian

Large modem steel silos are either elevated or supported on the ground. The main structural components are a cylindrical body, stiffeners (rings and stringers etc), a roof, and in the case of elevated silos, a hopper and supporting column . Silo structures are designed chiefly as containers, therefore the load from the contained material comprises the major part of the load which silos experience. The stored bulk solid exerts both internal pressure and downward frictional drag on the wall, the latter causing a potential for buckling of the wall.

In addition to loading the wall by gravity, the bulk solid has a strengthening effect against buckling under axial compression. This is because, unlike fluids, bulk solids have finite strength and stiffness. The effect of imperfections on the buckling strength of cylinders under axial compression has been studied extensively in the past, as has the effect of internal pressure. The stiffening effect due to bulk solids in silos has, however, not been systematically studied.

In this thesis a study of the stability of thin-walled steel silos is presented. The strengthening effect of the bulk solid is modelled theoretically using a Winkler foundation. The theoretical studies are carried out using the finite element technique. The formulation of an isoparametric doubly curved shell element for the linear elastic static analysis of shells of revolution with attached elastic restraints along the three curvilinear directions is described first. Non-symmetric loads are decomposed into Fourier components. The linear elastic buckling analysis is followed by the analysis of the shell-spring system under axisymmetric load, and nonlinear large-displacement and material analysis.

Linear analysis is often not sufficient to describe the real behaviour of shells. Non-linear effects from both the change of geometry and stresses which exceed the elastic limit may need to be considered. This formulation, including the effect of restraints, is given. The analysis includes full non-linearity on both shell geometry and material. The effect of non-linear springs is taken into account in the formulation. Recently-developed non-linear strain-displacement relations for thin shells of revolution are used. An efficient and accurate state determination algorithm is adopted. The arc-length method implemented by other researchers is used to trace both the pre- and post-collapse load deflection curve. Non-symmetric bifurcation including prebuckling large deflections is given, after the primary equilibrium state is reached.

The linear and non-linear analyses are applied to a number of example problems of restrained shells. Results are compared with published theoretical and numerical results, wherever it is possible. Several comprehensive theoretical studies of the buckling and collapse of steel silo shells are performed using the finite element analysis developed in this thesis. A study of silos filled with bulk solids with axisymmetric imperfections caused by weld joint depressions is firstly given. Attempt is made to relate any buckling strength gain of the shell wall by bulk solids to the mechanical properties of bulk solids, which has been a difficult task.

A comprehensive study of the collapse failure of light gauge steel silos made of thin corrugated sheets is made. One of the common failure modes in thin corrugated silos is that in which a few corrugations are squashed together to form a fold. Two simple collapse mechanisms are presented, each valid for a range of geometry. The proposed theories are verified using the finite element technique.

Another common failure mode for silos built of thin corrugated sheets is buckling of wall under axial or circumferential compression. The wall is idealised as an orthotropic shell. The orthotropic properties are therefore important to the stability analysis of corrugated shells. A number of expressions have been proposed in the past for the orthotropic properties of a flat corrugated sheet. The circumferential curvature influences these properties. A numerical study using the finite element analysis is given in this thesis for the accurate definition of these properties, and the results are compared with existing proposals.

Finally, tests of thin steel model cylinders filled with different bulk materials are conducted under several load conditions. The purpose of the experiments is to measure the strength gains of cylinders filled with bulk solids and compare with predictions of the finite element studies of restrained cylinders.

Three series of tests were conducted: (a) Empty cylinders under axial compression; (b) Cylinders filled with pressurised water under axial compression; and (c) Cylinders filled with bulk solids under axial compression. Two types of bulk solids are selected; one is the relatively softer wheat~ and the other the stiffer Sydney sand. The bulk solids inside the models are stressed to different levels to achieve different ratios of axial and hoop stress in the wall. The results indicate that the buckling strength of the wall can be increased significantly by the presence of bulk solids, and in some cases, it can be well above the classical buckling stress for a perfect cylinder under axial compression.

Information on the quasi-elastic behaviour of bulk solids inside silos is limited. A study is presented here to add to the available data. The mechanical properties of sand and wheat inside silos are given and compared with predictions suggested in the design specifications and in the literature.

Zhao, X.-L.
The Behaviour of Cold-Formed Rectangular Hollow Section Beams Under Combined Actions
PhD Thesis - Supervisor: Professor G.J.. Hancock

The main objective of this thesis is to investigate the behaviour of Square (SHS) and Rectangular (RHS) Hollow Sections subject to pure concentrated force, pure moment and combined moment and concentrated force, where the concentrated force is applied to tubular sections either by welding a square hollow section branch member or by a bearing plate. The following experimental and theoretical studies have been performed.

An extensive test program on a range of cold-formed SHS and RHS subject to pure concentrated force, pure moment and combined moment and concentrated force is described. The concentrated force was applied by either welding a square hollow section branch member (called T-joint tests) or by a bearing plate (called bearing tests). For the T-joint tests, the ratio (β) of the width of the branch member to the chord member varied from 0.5 to 1.0 in order to observe different failure mechanisms. For the bearing tests, the ratio (γ) of the bearing length to the width of the bearing plate varied from 0.5 to 1.0 in order to show the effect of the bearing length on the failure load. For both T-joint tests and bearing tests, the span of the simply supported beam was varied in order to determine the interaction relationship between moment and concentrated force. The slenderness of the member section and the shape of the section (SHS or RHS) were varied to cover a wide range of sections. The results of the bearing tests are compared with the T-joint tests on similar sections to demonstrate the significant reduction in failure loads and the more severe interaction between bending moment and concentrated force for the tests using bearing plates.

Plastic mechanism analysis (yield line method) is adopted in the thesis, since the failure modes of SHS and RHS involve local collapse mechanisms. The final result of a plastic mechanism analysis largely depends on the plastic moment capacity of yield lines and the mechanism model assumed. For the plastic moment capacity of yield lines, existing formulae are reviewed. The lower bound solutions, which are based on the Tresca yield criterion and the von Mises yield criterion, are derived in the theses to determine the reduced plastic moment capacity of an inclined yield line under axial force. A series of experimental tests on plastic hinges under axial force were performed to verify the formulae. Simplified expressions for the verified formulae are given to permit easy application of the theory. The mechanism models developed in the thesis are based on the experimental observations obtained in the thesis.

For T-joints in RHS under concentrated force, a modification of the Kato model is proposed for predicting the yield load. A more refined model is developed to predict the post-yield response and the ultimate load, where plastic hinges in the web, the membrane force in the flange and strain hardening of the material are considered. The virtual work principle is used in the derivation. For RHS under bearing force alone, a mechanism model is developed to predict the web yield load. The eccentric loading of the web where the corner radii meet the flange is considered in the model.

For T-joints in RHS subject to combined bending and concentrated force, the mechanism model is approximated by the modified Kato model. the reduction of the plastic moment capacity of yield lines is considered in the derivation. The action in the chord (bending moment) rather than the chord normal stress is used for interaction curves. A proposed interaction formula is given, which includes the variation of capacity with β. For RHS subject to combined moment and bearing force, interaction formulae are derived based on the test result.

Existing design procedures from Australia, America, Britain and Canada, both for cold-formed members and for hot-rolled members, are reviewed and compared with the test results. Improved design procedures are given for SHS and RHS under concentrated force alone and for SHS and RHS subject to combined moment and concentrated force.

The University of Sydney - Civil Engineering

1991 & 1992 Higher Degree Theses