Search

Finding the location of underground utilities, essentially plastic water pipes, can be difficult and disruptive, which is due to the lack of efficient techniques that can be used to map these buried network services. Vibro-acoustic techniques to determine the location of buried water pipes have been studied within the Institute of Sound and Vibration Research within (ISVR) in the past 10 years. One technique involves excitation of the water pipe where it is accessible, normally from hydrant, followed by measuring the ground surface vibration, at the vicinity of suspecting pipe location. From an earlier feasibility study it was indicated that the emitted ground borne waves from the waves that propagate along the pipe and its contained fluid can help manifest the pipe location. Because the attenuation of the waves is generally large for the plastic water pipe, the pipe ground borne waves cannot be sensed at large distances away from the exciter's location, of order of a few tenths of a metre, or at high frequencies, above 100 Hz. This thesis aims at using an in-pipe excitation source, which can be deployed at any desired location along the pipe length, to overcome the attenuation problem. Although in-pipe sources can transfer energy to the pipe contained fluid and might allow tracking of the pipe at larger ranges, current acoustic exciters are not always appropriate, being cumbersome and too large to fit into a typical buried water pipe. In this thesis, two pneumatic devices were designed with the aim of generating high amplitude signals at low frequencies and with the ability of accessing pipes with a wide range of diameters, down to 1 cm. The devices are experimentally characterised by a series of laboratory tests in a water-filled plastic pipe section. A comparison of the acoustic pressure wave transmitted to a fluid filled pipe between a standard electroacoustic device, an electromagnetic shaker, and the pneumatic ones is made. From the previous work on characterization of wave propagation along plastic water filled pipes, it is known that at below 100 Hz, among all kinds of waves, the axisymmetric fluid borne wave can predominately drive the ground borne wave. Owing to iii the good coupling of the axisymmetric fluid and the shell borne wave they should be considered altogether. The wave speed and attenuation of both waves were measured experimentally, and the results were checked for consistency with the theory. An analytical simulation was developed to explain the reason of high wave attenuation variances obtained from the experimental measurement. Due to the dependency of the axisymmetric fluid and structural borne waves' amplitude to the elastic properties of the pipe a simple experimental method was proposed to distinguish between the pipe wall displacement that arises from the two axisymmetric waves. Such a technique might help assess the condition of pipes through indicating the reduction in their elastic properties due to ageing. From the earlier work on detecting buried water pipes, it was indicated that in addition to the axisymmetric fluid borne wave, the exciter applied to the pipe directly/indirectly drives the ground borne waves with substantial amplitudes. To have a better understanding about the overall response of the ground surface, a simple analytical model was developed, taking into account a ground borne wave emitted from the pipe and the source respectively. Like numerous other digital image processing, in the vibro-acoustic technique, an arctangent operator is used to extract the phase or time delay information, as the assumed ground borne waves are emitted and travel to the measurement point. By virtue of the fact that the arctangent operator produces phase images wrapped between −π to π, an unwrapping operator is required to remove the phase discontinuities embedded within the image. Therefore, a novel phase unwrapping algorithm is developed with a low-cost computational requirement. Furthermore, different state-of-the-art two-dimensional unwrapping algorithms are reviewed and compared for their ability to remove the phase discontinuities, produced by the model. In addition, a drawback of applying one-dimensional unwrapping to the two-dimensional wrapped phase image, used in the previous study, is discussed. To benchmark the effectiveness of the developed pneumatic devices, different exciters such as standard mechanical and electroacoustical exciter are utilised to map a buried pipe. Applying mechanical excitation to the pipe is associated to the earlier feasibility study which repeated for the sake of benchmarking the in-pipe exciters' results against it. Of the exciters, one of the designed pneumatic sources successfully mapped the total length (18 m) of the pipe. Because the aim of this thesis is deploying an in-pipe source to overcome the attenuation problem, the utilised acoustical exciters: the underwater speaker and the designed pneumatic sources are deployed at two more locations along the pipe. The results from each exciter are analysed in detail and compared to each other.

Arising in the far-field of large explosions, long duration blast events are typically defined as having a positive phase duration in excess of 40ms. Capable of engulfing entire buildings, long duration blast waves can result in significant damage to a multitude of compression elements, potentially leading to total structural collapse. Current guidelines provide simple empirical methods for estimating peak response and damage state of a column to blast loading. These methods neglect more complex situations such as non-orthogonal structural orientation, combined blast and thermal loading or column position within a frame: all requiring extensive experimentation or numerical modelling to determine a damage state. In this thesis, multi-axis, thermal and frame interaction effects on the structural response of a column to long duration blast were explored. One phase of experimental testing improved the understanding of explosive thermal energy application to I-section surfaces. Another phase of testing explored the response of partially-clad frames to long duration blast utilising experimental findings, an extensive numerical study was undertaken, exploring the response of a column to large scale blast events including combinations of all three effects. Results were quantified and consolidated into a series of end-user charts, akin to those found in the current design standards. These charts provide a quick and simple means of estimating the maximum displacement and corresponding damage state of a column to long duration blast with multi-axis, thermal and frame interaction effects: reducing future requirement for experimental testing and numerical modelling. Findings indicated that a combination of all effects can have a detrimental impact on response, in some cases increasing damage state from minor to total destruction and collapse for the same column to the same blast event.

Slope failures may cause substantial loss of life and damage to infrastructure. In UK much of the rail network was constructed over 150 years ago, and the earth structures were not built to modern standards. As these structures get older, they have become a potential threat to the safety of transport operations. Climate conditions directly influence the behaviour and failure of slopes. The presence of trees increases the depth and extent of desiccation and cracking and may increase the permeability to greater depths, leading to greater changes in seasonal pore water pressures. Seasonal pore water pressure changes lead to corresponding cyclic changes in effective stress. The fatigue of the clay brought about by seasonal effective stress cycles is a possibility though not well established. Climate change is expected to bring extreme weather patterns to the UK in which wetter winters and drier summers will prevail, potentially giving larger cycles of stress. Field investigations of cracks and other macro pores are carried out as part of the current study. To establish appropriate values for the change in near surface permeability of infrastructure cut slopes caused by opening and closing of cracks, field permeability experiments on a cut slope in Newbury were carried out during different seasons of the year. Appropriate near surface permeability was established based on this. A mathematical and numerical study of the influence of a single crack on permeability is presented. To investigate the influence of cycles of effective stress associated with pore pressure changes brought about by seasonal variations in climate on the development of accumulated strain and on the strength of stiff over consolidated clay materials, a series of cyclic triaxial experiments were carried out on soil samples from railway embankments. The results from cyclic triaxial tests on undisturbed and reconstituted Gault and Lias Clay embankment fill materials are presented and analysed. The near surface vertical permeability of the cut slope in Newbury (10-8 m/s - 10-6 m/s) is found to be at least two orders of magnitude higher than that of intact London clay (10-10 m/s) and the end of summer vertical permeability (~10-6 m/s) two orders of magnitude higher than that of end of winter (~10-8 m/s). Near surface vertical permeability varies between summer and winter possibly mainly due to cracks opening and closing. The undrainded shear strength of the reconstituted Lias Clay is consistent between 42.5 kPa to 47.5 kPa. Critical friction angle of reconstituted Lias Clay varies between 24° and 27°. For Lias Clay triaxial samples, permanent axial strains are accumulated with cycles of pore water pressure. The rate of accumulated strain decreases with the number of cycles when the effective stress ratio is below that corresponding to critical state line. The stress conditions under which the sample failed (ϕ'mob = 340) when pore pressure is cycled is with in the range of strength obtained for monotonic tests on the undisturbed fill material (which was between 28° and 37°).

Multi-pile foundation systems are used to support offshore infrastructure in areas of Northern Europe underlain by Chalk. During installation, the tip of the tubular open-ended ‘small displacement’ pile crushes and destructures the Chalk forming a ‘putty’ interface or ‘annulus’ around it. Pile resistance to uplift (i.e. shaft capacity) results from sliding friction between the pile shaft and this putty chalk annulus. The limited knowledge of the characteristics of this interface is associated with pile design guidelines that may be overconservative, which could result in unnecessary increases in economic costs for the offshore renewable energy industry. The present Thesis investigates the prospect of using soil mechanics to describe shaft friction in small displacement piles in Chalk as function of the critical state strength of the chalk putty that composes the annulus. To this end, oedometer and triaxial tests were conducted on low and medium density destructured White Chalk samples to determine the suitability of a critical state framework to characterise the mechanical behaviour of the material. Monotonic and cyclic simple shear tests were then used to simulate load transfer through the chalk-pile interface and examine the compatibility of interface performance with the prospective critical state framework. Thereafter, model piles of different geometries were installed in intact chalk cores and micro-focus X-ray computed tomography (XCT) was applied to measure the density of the chalk putty annulus. The piles were then tested in tension. Oedometer and triaxial results demonstrate that a unique one-dimensional normal compression line and a unique critical state line (CSL) exist for the tested materials, regardless of their origin and the varied preparation methods used. This ‘uniqueness’ is associated with comparable grain shapes and size distributions, and to grain breakage processes that resemble those reported for sands. Simple shear experiments show that interface failure states comply with the CSL, although strain localisation precludes the attainment of critical state strengths when shearing initiates in significantly dense of critical conditions. A CSL-based framework defining interface strength as a function of void ratio was produced and found to compare well with the small number of early-life shaft capacity measurements of piles available in the literature. Yet, this framework could not predict the shaft capacity of the model pile experiments, due to heavily dense of critical states at the end of installation. XCT further revealed that pile penetration processes in chalk and sand are generally comparable, and that design approaches based on the cone penetration test might successfully estimate early-life shaft capacity. Though pile set-up was found to possibly result from annulus consolidation, its prediction may need to rely on empiricism for the foreseeable future.

Eigenvalue problems, that depend on a parameter, are frequently encountered in structural engineering. The most common contexts are free vibration and structural stability, where the calculations of natural frequency, or the critical buckling load, for a large number of design parameters is of practical interest. In the present work, approximations suitable for early stages of design exploration are presented, since reanalysis is an expensive process. A novel approach that leads to approximations for several classes of problems is presented. The essence of the proposed approximations is to obtain a trial vector, that is rich in the components of the actual eigenvector, first. This is achieved economically via an interpolation of eigenvectors as proposed here. Following this, Rayleigh quotient approximation with these trial vectors is carried out, which is found to provide excellent approximations for a range of problems economically. The method is then applied to illustrative examples arising in structural vibration. The computational gain is found to be relatively more significant as the size of the problem increases. The computational complexity of the proposed method is assessed. The proposed approximation requires adaptations depending upon whether the matrices involved are symmetric, skew-symmetric, or general asymmetric, and also if the parameter-dependent eigenvalue problem is standard, or generalised i.e. in terms of a single matrix, or a matrix pencil. There are three primary reasons for separately dealing with different classes of problems associated with different symmetries of the matrices involved: (i) because of different orthogonality relations, the Rayleigh quotient for each case is different, and (ii) the nature of eigensolutions, e.g. real vs pure imaginary vs complex could depend upon the class or problem at hand, and (iii) the practical contexts in which each of these arise are very different: e.g. the presence of gyroscopy, aeroelastic effects, dissipation, follower forces, or a combination of these. Excellent approximations are obtained for various classes of problems considered here, while providing considerable computational economy.

In this thesis, a one-dimensional (beam-like) repetitive pin-jointed structure with point masses located at nodal cross-sections is analysed with the aim to provide a simple physical explanation for the existence of frequency-propagation zones and decay zones. Two forms of dynamic transfer matrix are derived for the model structure: the displacement-force transfer matrix G, and the displacement-displacement transfer matrix H; the focus is on the relationships between the two, including their respective (dis)advantages. Similarity matrices are introduced to relate G and H, together with their respective metrics. Symplectic orthogonality relationships for the eigenvectors of both G and H are derived, together with relationships between their respective sets of eigenvectors. New expressions for the group velocity are derived. For repetitive structures of finite length, natural frequency equations are derived employing both G and H, including phase-closure and the direct application of boundary (end) conditions. The wave propagation and decay characteristics of the model structure are described in depth. The existence of propagation zones is explained in terms of phase-closure (implying a natural frequency) over the cross-section at the cut-on frequency, and phase-closure over the smallest axial unit - the repeating cell - at the cut-off frequency; these zones can therefore be regarded as extended resonances. Wave interaction between branches displaying normal and anomalous dispersion is explained in terms of the Krein signature, and leads to attenuating waves. At frequencies below cut-on, waves are generally monotonic evanescent, while above cut-off they are generally oscillatory evanescent. Energetics of different wave types under the new taxonomy is investigated. Equations for energy and power are derived in terms of eigenvectors of G and H. Numerical results for axial phase velocity and group velocity of the different modes show some familiar and some peculiar phenomena. Condition for maximum group velocity is proposed. Numerical study on anomalous dispersion reveals that pin-jointed structure which allows negative eigenvalue to occur under static self-equilibrating load will presage anomalous dispersion under dynamic condition. The solution of a two point boundary value problem is numerically stable when one employs the Riccati transfer matrix method. An alternative numerically stable transfer matrix method, which is more direct and transparent, is developed for a repetitive structure fixed at one end, and subject to point-wise distributed loading, with and without an intermediate support. This is achieved by constructing mixed column vectors of participation coefficients, so that spatial evolution involves multiplication by powers of the eigenvalues which are less than or equal to unity.

This research study seeks to evaluate the load response and failure prediction of glass/steel adhesive joints. The need for sustainable construction materials along with recent architectural trends and technological developments have made glass more accessible than ever before in the construction industry. Limited attempts have been made to compare the performance of bolted and adhesive connections for glass/steel structures, while interface characterisation studies are also lacking. Damage initiation and propagation in the adhesive layer is rarely modelled numerically, and the development of cohesive zone models has been restricted to reference values as hybrid coupon tests are difficult to test successfully. Lastly, while the degradation of glass/steel adhesive joints has been examined experimentally, a numerical tool for the prediction of the performance of the joints after exposure is currently lacking. Benchmark designs of glass/steel bolted and adhesive joints were introduced and tested experimentally in four different load cases. Adhesive joints were found to be stronger and stiffer for all load cases examined. It was also observed that lower strength ductile adhesive (in terms of bulk properties) produced joints with higher failure loads. Numerical analyses showed that ductile adhesives developed a large plastic zone and redistributed the stress concentrations more effectively from the corners of the joints. Therefore, a larger adhesive area was resisting the loading. This understanding of the synergistic property combinations of strength and ductility of the adhesives led to the development of a numerical tool for the optimum selection of adhesives based on the joint design. The identified adhesive led to a significant strength increase for every load case examined. The long term performance of glass/steel adhesive joints was evaluated by exposing the joints to conditions of high temperatures and humidity, and the degradation of the bulk properties and the interfaces was recorded. It was shown that the bulk properties and the interface properties degrade at different rates. The glass/adhesive interface degradation was shown to be more significant and controlled the failure performance of the joints. Numerically, a continuum mechanics and a cohesive zone modelling (CZM) approach were evaluated for their suitability in predicting the failure load of glass/steel adhesive joints before and after environmental exposure. Input parameters for continuum mechanics approaches are based on bulk properties only and are easier to evaluate than CZM interface parameters. An in-house heat strengthening methodology development was necessary to increase the strength of small coupon sized glass substrates for accurate interface characterisation. It was shown in this work that both numerical methods were accurate in predicting the performance of the unaged joints. After environmental exposure, the CZM approach, which allows to account for the more severe interface degradation, performed significantly better. This finding highlights the need for reliable enhanced experimental testing procedures for interface characterization for hybrid glass/steel joints.

Glass windows are vulnerable building components and, in the event of a blast, give rise to a large damage radius. Many older buildings contain annealed glass which shatters instantly into angular shards, lacerating the skin on impact and causing significant injury. Up to 80% of injuries in an urban blast are caused by glass failure. In the far-field of large explosions, long-duration blast environments occur, typically defined by a positive phase duration greater than 40ms. Resulting impulses can cause window failure several kilometres away from the point of detonation. Data from long-duration blast events indicate window failure is highly dependent on glazing aspect ratio, area and other structural parameters. Inconsistency in window response has been attributed to material strength variation which can be in excess of ±25%. The Glazing Hazard Guide, the current UK design standard, indicates windows with matching dimensions respond consistently to blast loading, incorporates no extrapolation method to other glazing dimensions and has been demonstrated to be overly conservative. In this thesis, influence of glazing aspect ratio, support conditions, material strength and the blast environment on window failure was investigated. Four phases of experimental testing and a comprehensively benchmarked parametric study enabled quantification of the influence of these parameters. Full-scale experimental testing of glazing response to long-duration blast was undertaken in the air blast tunnel (ABT) at MOD Shoeburyness, UK. Glazing response was dependent on complex interactions between structural and blast parameters. Experimental repeats importantly indicated significant data spread, demonstrating that glazing response should be measured as a probability of failure. Damage probability curves and pressure-impulse charts with contours of failure probability were produced for each parameter and were proposed as a new design method. These charts capture true glazing behaviour, are easy to interpret and highlight to the user that window response should be treated statistically. Findings indicated variation in material strength due to distribution of flaws produced the largest spread in failure pressure.

In many countries high speed rail is playing a growing role in improving the capacity, availability and carbon cost of national infrastructure. Higher speeds require straighter railway alignments, which often means crossing areas of soft ground that have historically been avoided. Due to their low ground-borne surface wave speeds, there is a greater likelihood in such areas of the train passage resulting in critical velocity effects - a phenomenon in which excessive ground and track movement and vibration occurs. This can cause extensive damage, and may result in a forced reduction of train speeds. The aim of this research is to assess methods of determining appropriate soil parameters for use in simple elasticity-based models for the prediction of critical velocity effects on railways. After a review of existing knowledge, the research consists of field measurements, laboratory experiments and modelling. This research focuses around two case study sites on the UK rail network. Soil samples were extracted from each site, and tested in resonant column and triaxial equipment, to investigate their strain-dependent stiffness and damping. This involved testing on soil types for which there is little published data, including highly organic silts. The results are compared with measurements taken in situ, including seismic analysis and heavy probe tests. A linear elastic model, MOTIV, of train-induced vibration is used to investigate the key soil and model parameters required for critical velocity analysis, as well as to assess the importance of the non-linearity of soil stiffness and damping with strain. It is clear that the use of strain-degraded soil parameters, for example through the use of an equivalent linear model, is essential. Recommendations are made for the best methods to obtain the relevant data, from which reliable critical velocity predictions can be made using linear soil models.

Long-duration blast waves are typically defined by positive pressure durations in excess of 100ms and are characteristic of very large explosive events. Such blasts generate substantial impulse and dynamic pressures (blast winds) that considerably exceed shorter duration blasts with the same peak overpressures, which are particularly damaging to drag-sensitive structural elements such as steel columns. A number of recent large-scale industrial accidents have highlighted the potential for multiple detonations to occur that can cause long-duration blasts to successively interact with structures, and constituent column elements from different axes. This research investigates and quantifies blast loading, structural dynamic response and resulting damage states of steel column elements subjected to both single and multiple (successive) long-duration blasts. Primarily, this research examines the influence of multi-axis interaction and the effect of prior column damage on subsequent structural response, given the case of multiple detonations. Ten full-scale long-duration blast trials were conducted, testing steel I-section and square hollow section (SHS) columns aligned at different orientations to the blast to examine the influence of multi-axis interaction. Six trials also investigated the effect of prior column damage on subsequent structural response and damage accumulation by subjecting SHS columns to three successive blasts. Analysis of results revealed a clear relationship between blast loading, column mechanical properties and magnitude of structural response as a function of section orientation to the blast, thus demonstrating that multi-axis interaction has significant influence. Columns aligned at oblique orientations to the blast exhibited larger structural responses and higher loading due to increased projected area and drag coefficient providing evidence to question the validity of existing simplified blast loading characterisation methods. Column response to multiple blasts generally exhibited reduced structural resistance, with damage accumulation influenced by local residual stress states, load direction symmetry and sequencing. This research also examined uncoupled computational models of multi-axis blast loading and structural dynamic response of steel columns using computational fluid dynamics (CFD) and finite element analysis (FEA) respectively. Comparisons with experimental data have shown reasonable levels of agreement, indicating potential to provide future predictive capacity.

This research project investigates the thermo-mechanical non-linear response of steel structures to combined long duration ( > 200msec) blast and thermal loads. Depending upon explosive size and standoff distance the thermal load can reach, and degrade a structure prior to the arrival of the blast load. The thermal load produced by explosive events can damage steel structures such that the damage from the combined thermal and subsequent blast load would be greater than or different to, the damage from the blast load alone. Parametric studies presented in this thesis examine the response of structural steel columns to a range of blast and thermal loading regimes and investigate the effect of parameters including connection type, compressive loads and thermal conductivity. Data from previous research is used to develop predictive algorithms calculating the thermal and blast load regimes from explosive events. Transient, non-linear, coupled (thermal and structural) analyses using the finite element analysis (FEA) program, LUSAS are performed on the structural steel columns. The design of, and results from a series of trials recording temperatures, thermal flux and pressures, and observing the response of 2mm thick steel plates, within a 41kg TNT equivalent (eq.) explosive fireball, is presented. The design of, and results from experimental trials investigating the response of structural steel columns to combined thermal, compression and long duration blast loads within an Air Blast Tunnel (ABT) are also presented. Results from both sets of trials are compared to predictive non-linear FEA models of the steel plates and columns, developed to provide verification of computational procedures used throughout the research project. Resistance curves for the vulnerability analysis of whole buildings to explosive events are presented. The curves represent the elastic and plastic failure mechanisms of structural columns to thermal and blast explosive loads. This research project demonstrates that under specific loading regimes steel structures can exhibit a thermo-mechanical, synergistic response to combined thermal and blast loads from explosive events. This has been achieved through detailed investigation into previous research, thorough and comprehensive parametric studies and a series of innovative experimental trials using M.O.D national testing facilities.

Due to the demands of increasing population and environmental concerns, the high-speed train has become an important means of transportation in many countries. However, severe vibration induced by high-speed trains can occur when train speeds approach the speed of waves in the ground and track deflections can then become large. Consequently, the track and ground may no longer behave linearly. The aim of this work is to develop a three-dimensional model of vehicle/track/ground interaction in the time domain that can include the consideration of soil nonlinearity. A three-dimensional time-domain model of a coupled vehicle, track and ground has been developed in the finite element (FE) software ABAQUS. A new analysis approach has been adopted for modelling the moving vehicle without the need for a user-defined subroutine. The results from a track on a hemispherical ground model surrounded by infinite elements show good agreement with the results from a wavenumber finite element/boundary element method at lower speeds. However, significant differences are found for load speeds close to the critical speed for a homogeneous ground. This has been shown to be due to a relatively long transient in the numerical simulation that requires a considerable distance to achieve convergence to steady-state results. As a result, a very long model is required to derive the steady-state results. However, this becomes expensive because of the model geometry. Moreover, the results from the hemispherical model contain incorrect whole-body displacements due to the use of the infinite elements that make the ground model become unconstrained. Therefore, a cuboid model with a fixed boundary at the bottom is used for the further simulations to prevent these incorrect phenomena. Two different cuboid models, with or without the infinite elements at the sides, are compared. Finally, a wider cuboid model with fixed boundaries is used with appropriate Rayleigh damping which shows the best results and efficiency. A very long model, around 150~300 m, is required when the load speed is equal to the critical speed for a homogeneous soft ground. However, a shorter model is sufficient to obtain the steady-state results for a layered half-space. For a load moving close to the critical speed on a layered half-space, the track oscillates with a certain dominant frequency. An investigation is presented into the dependence of the oscillating frequency on the ground properties, taking account of the dispersive surface waves. Three different methods are used to investigate this oscillating frequency for a layered half-space ground and a parametric study is carried out. The oscillating frequency is found to vary with the speed of the moving load and tends to decrease when the load speed increases. It mainly depends on the shear wave speed of the upper soil layer and the depth of this layer. A formula is introduced to estimate this oscillating frequency in a layered half-space. Finally, soil nonlinearity is introduced in the FE model through a user-defined subroutine. The nonlinearity is specified in terms of the shear modulus reduction as a function of octahedral shear strain, based on data obtained from laboratory tests on soil samples. The model is applied to the soft soil site at Ledsgård in Sweden, from which extensive measurements are available from the late 1990s. It is shown that the use of a linear model based on the small-strain soil parameters leads to an underestimation of the track displacements when the train speed approaches the critical speed, whereas the nonlinear model gives improved agreement with the measurements. However, the results are quite sensitive to the choice of nonlinear model. In addition, an equivalent linear model is considered in which the equivalent soil modulus is derived from the laboratory curve of shear modulus reduction using an 'effective' shear strain. It is shown that the predictions are improved by using a value of 20% of the maximum strain as the effective strain rather than the value of 65% commonly used in earthquake studies.

High performance foiling catamarans are one of the fastest growing sectors in the sailing and sport industries, allowing athletes to perform in extremely fast and spectacular boats. These boats fly above the water with the aid of foils that not only provide horizontal side-force to counterbalance the aerodynamic forces from the sails, but also deliver vertical force that supports some or all of the mass of the boat. Exploring the possibility of using Passive Adaptive Composite (PAC) on the hydrofoils to control their pitch angle enables the boats to achieve a stable flight in a wide range of weather conditions. This thesis presents an experimental and numerical evaluation of bend-twist elastic coupling in composite passive-adaptive structures. Due to the lack of experimental validation in Fluid Structure Interaction (FSI) investigations, a full-field deformation of an aerofoil-shaped section under wind loading is measured. Moreover, the influences of structure deflection on flow behaviour are investigated by looking at the changes in flow features evaluated on a transverse plane downstream of the trailing edge. The experimental analysis was carried out at the University of Southampton R. J. Mitchell wind tunnel and involved the use of full-field non-contact measurement techniques such as high speed three dimensional Digital Image Correlation (3-D DIC) and stereoscopic Particle Image Velocimetry (PIV). After assessing the validity and repeatability of the experiments, the research focuses on the development of a numerical FSI investigation that involves the use of a structural and a fluid solver to simulate the aero-elastic behaviour of composite tailored specimens with different internal structures. The numerical analysis is developed as a tool to allow the design of a new structure able to achieve a constant level of lift force (corresponding to the weight of the catamaran) in increased flow speed. During the research project it was proven that the efficiency of the foils can be improved by tailoring the internal structure to induce smart coupled bend-twist toward a wash-out (feather) or wash-in (stall) position under increased loading.

The results of an investigation into the characteristics of catenary action and their effects on structures are reported here. There are five aspects presented in this thesis: a literature review on progressive collapse and beam-to-column connections; an exact analysis of catenary action of truss; nonlinear finite element analyses of different post-column removal performance as a function of the original and retrofitted beam-to-column connection geometries; comparison of different behaviours of different beam-to-column connection geometries; and the minimisation of the strengthening plate weight for the retrofitted structures. This thesis demonstrates that when a column is removed, the bending moment will decrease significantly as the catenary tensile force and the plastic deformation increase in the beams connected to the damaged column. This thesis also shows that when a column is destroyed by a blast, the failure strain will be reached at the simple beam-to-column joint. To enhance the survival capability of the steel framed structures subjected to terrorist blast, retrofitting schemes were proposed for strengthening the joints of tall steel framed structures, ensuring the full development of catenary action. This thesis simulates the post-column removal behaviour of the original and the strengthened structures by means of the ABAQUS finite element package. For this purpose, sophisticated two- and three-dimensional models of catenary action are developed. Through comparing different results of the original and the strengthened structures from a column removal, the advantages of the proposed retrofitting schemes have been demonstrated. This thesis investigates the relative advantages and shortcomings between two different retrofitting schemes and compares the proposed retrofitting schemes with other moment beam-to-column connections used for new construction. For the same resisting capacity of the retrofitted structure, the Vertical Plate Scheme will require much more material than the Flange Plate Scheme. But the Vertical Plate Scheme still has its advantages. The Vertical Plate Scheme does not require the removal of the floor slab and easier to conduct than the Flange Plate Scheme. Finally, it is demonstrated that the sizes of strengthening plates can greatly change the overall behaviour of the retrofitted structures; with the reduction of the strengthening plate sizes, the stress and strain will increase and the structure will experience elastic stage and elasto-plastic sequentially. This thesis minimises the thickness and the length of the strengthening plates by the exact analytic method for simple structures and the finite element method for complicated structures.

This thesis investigates the three-dimensional container loading problem. We review all literatures published in this area, and explain our unique problem raised from an industry partner. As constructive heuristic remains uncontrollable to us, we design improvement algorithms both to and not to intervene with how constructive heuristic works, namely iterated local search and beam search. New benchmark data sets for multiple containers problem are generated to fill the shortage of challenging data sets. Computational results for homogeneous containers problem indicate that while both approaches work on our problem, beam search remains a favourable choice. We also extend our algorithms to solve heterogeneous containers problem.

The basic concept of Hybrid Active-Passive Constrained Layer Damping (HAPCLD) treatment was proposed by introducing active control to the concept of passive constrained layer damping configuration in the 1990s to compensate for weak points in active and passive controls by using their respective merits for more robust and stable control. Since then, combinations of various configurations and applicable control strategies have been proposed and studied in many engineering areas. However, there is still a need for a new modelling method to more easily establish models of HAPCLD treatment and its validation through control analysis and experiment with various structures from beams to curved plates. In this thesis, velocity feedback control strategy was applied to cantilever beams with four different configurations of HAPCLD treatment to check their applicability. Moreover, the application was expanded to flat and curved plates. Control results with each configuration for flat and curved plates were analysed by using self-established MATLAB codes based on the Finite Element Method (FEM) with the basic concept of a layer-wise approach for coupling each layer of structures and deriving Equivalent Single Layer (ESL) models. This new numerical modelling method was established by introducing coupling matrices based on a layer-wise approach to combine individual FE mass and stiffness matrices of each layer into one ESL model for a whole structure. Furthermore, these numerical models were supported by experiments in a lab. All measured data was compared with simulation results and they were confirmed in good agreement in general. In addition to this, the relation between mode shapes and control by piezoelectric patches occupying a broader area than an ideal actuator was studied to find the conditions for more stable control of flat and curved plates. In conclusion, as discussed for active control with beams, AC/PSOLD treatment, which consists of a piezoelectric actuator directly attached to a base structure and a stand-off layer with a viscoelastic core and elastic constraining patch laminated on the piezoelectric actuator, was clarified to give the most efficient and robust active control results for plates regardless of the curvature of all HAPCLD treatments dealt within this thesis as well. AC/PSOLD treatment could give similar reductions with smaller control gain in simulation. And, larger reductions were obtained with measured transfer functions in experiments than other configurations.

This thesis presents a review of structural robustness in conventional (non-hardened, nongovernmental and non-military) buildings subjected to malicious actions and has also investigated the capability of tensile membrane (catenary) action as a means of emergency load redistribution in reinforced concrete (RC) framed buildings, designed in accordance with BS 8110 and Eurocode 2. Case studies and forensic data relating to the response of conventional blast damaged buildings have been investigated. Based on the conclusions drawn from this investigation a simple ranking system has been presented that identifies the level of confidence offered by various forms of direct and indirect robust design. Catenary action has been shown to be a fundamental mechanism of emergency load redistribution for which research has been relatively limited, particularly in relation to RC frame constructions. An experimental investigation has been carried out; this involved the testing to failure of a series of half scale RC strip specimens, detailed to conventional detailing practices, to investigate the large displacement behaviour and ultimate collapse resistance of laterally restrained RC floor components following the loss of intermediate column support. The results showed characteristic compressive and tensile membrane responses. However, tensile membrane action was found to provide the largest reserve of strength and to offer the highest potential as a means of emergency load redistribution. By testing specimens to outright collapse the investigation identified factors influencing catenary performance and demonstrated that the total energy absorption by work done in catenary action was influenced by the tensile properties and detailing of the specimens. The current ultimate limit criteria and theoretical approach to ultimate load prediction in catenary response have been critically reviewed using results obtained in testing. The findings have been implemented in an analytical study of collapse resistance by ultimate catenary response. A series of exemplar floor systems, of different size and detailing arrangements have been examined. The conclusions presented indicate that safe load redistribution by catenary action is dependent on the area of the unsupported structural bays and the area of the bottom reinforcement specified at or across structural connections, for which design parameters such as moment redistribution, detailing rules and tie force requirements are governing factors.

Even in simple, quasi-static cases, blast and its interaction with structures is a complex phenomenon. Long duration blast, dened for the purposes of this research as a blast event with a positive phase duration in excess of 100ms, increases this complexity due to not only the persistent eects of the static blast overpressure, but also the dynamic pressure associated with drag wind trailing the shock front. Brittle materials such as concrete and masonry produce large numbers of initial fragments, which when caught in the drag wind produce substantial debris distributions, presenting a number of potential hazards ranging from infrastructure obstruction to personal injury. This research investigates the eects of long duration blast loading of masonry structures, using a modular `Base Panel' approach to descirbe structures as a composition of simple panels. Five experimental blast trials were conducted, testing a total of 22 masonry structures against varying blast parameters. The results showed correlations between both the breakage and debris distributions with respect to panel geometry. Computational Fluid Dynamics and the Applied Element Method were used to model the experimental trials, with the results showing good agreement oering a promising modelling platform for future work. The results, both experimental and computational, demonstrate the `Base Panel' approach to be an eective tool for the prediction of masonry debris distribution.

Current research with respect to the protection of civilian infrastructure against complex blast loading conditions is primarily focused towards the effect of external explosive sources. As a consequence, the general literature on internal building detonations and specifically in the context of protective design and assessment of structures against these loading conditions is incomplete. Existing guidelines developed for comparatively noncomplex external explosive blast remain unconservative when applied to internal building detonations due to blast wave confinement and complex interaction with structural components. In particular, reinforced concrete (RC) columns in internal blast environments are subjected to time-variant uplift forces coupled with lateral pressures leading to destabilisation and a critical loss of structural integrity. Research presented in this thesis provides an original understanding towards: (i) – the influence of transient uplift forces on the vulnerability of RC columns subject to lateral blast pressures and, (ii) – design and assessment of RC columns against the effect of time-variant coupled uplift and lateral blast pressures due to internal building detonations. Research in this thesis is based on advanced uncoupled Euler-Lagrange numerical modelling splitting the structural and flow solvers for maximum integrity and accuracy. High-resolution simulations of complex flow fields are analysed using the hydrocodes Air3D and Autodyn, whilst Extreme Loading for Structures [ELS] based on the Applied Element Method is used for modelling the transient-dynamic structural response of columns. These numerical techniques are comprehensively ratified and underwritten, both qualitatively and quantitatively, using published independent experimental test data. Verified numerical modelling is subsequently used to conduct a set of comprehensive parametric studies covering both vented (frangible perimeter walls causing pressure venting) and contained (non-frangible perimeter walls causing repetitive wave reflections) internal blast environments. Results of these parametric studies are thoroughly analysed with the use of multi-variable nonlinear regression analysis techniques and are presented in the form of separate assessment and design charts. This thesis also presents a set of column hazard charts developed based on the parametric studies. These hazard charts provide threshold combinations of TNT Equivalency and critical radial distance corresponding to different column damage levels ranging from ‘No Damage’ through ‘Low Damage’ and Moderate Damage’ to ‘Imminent Structural Collapse’. The output of this research will be of direct relevance to both practitioners and researchers involved with protective design of civilian and military buildings.

In the last few decades much effort has been spent to investigate the installation effects of diaphragm wall panels in an overconsolidated deposit. Conversely, in the literature there is relatively little focus on the effects of bored pile construction. In this thesis a number of three dimensional finite difference analyses using FLAC3D have been carried out to investigate the stress changes and consequent movements due to bored pile and diaphragm wall installation in stiff clay. The three-dimensional numerical models developed to carry out these analyses were used to analyse, interpret and explain the instrumented data of ground stress changes during bored pile wall installation in an overconsolidated clay deposit in connection with the Channel Tunnel Rail Link (CTRL; now HS1) at Ashford. Two different soil models were implemented in the three dimensional finite difference analyses, having linear and non-linear stress strain characteristic. It was found that both analyses could capture the observed pattern of changes in stresses. Furthermore, it was found that the difference in stress changes and ground movements after a single pile or panel installation were primarily a consequence of the different geometry between pile and panel. In the literature, only limited research has been carried out to quantify the three dimensional effects of diaphragm wall installation. For this thesis a number of three dimensional finite difference analyses, using FLAC3D, were carried out, to investigate the installation effects of a group of panels in an overconsolidated clay deposit. One of the key uncertainties associated with the design of in situ embedded retaining walls in an overconsolidated deposit concerns the long-term horizontal stress acting on the wall. There is some concern that the high in situ horizontal stresses in an overconsolidated deposit may become re-established in the long-term, despite the reductions that occur during retaining wall installation and subsequent excavation in front of the wall. This thesis also presents long term case record from an embedded retaining wall that forms part of the Channel Tunnel Rail Link at Ashford, Kent, and a retaining wall at Coventry are also presented. Approximately 13 years of lateral stress monitoring around the embedded retaining wall at Ashford are presented and discussed. The long term field data shows that the in situ total horizontal stresses in the ground close to a retaining wall in an overconsolidated clay deposit did not re-establish in the long term.

An important failure mode of foam cored composite sandwich structures is the debonding between the face sheet and core. The challenges in establishing and improving the damage tolerance of sandwich structures mainly arise from the lack of reliable methods to characterise the fracture behaviour at face sheet/core interface. It has been demonstrated that the inclusion of crack arresting devices can improve damage tolerance. Characterisation of the fracture behaviour at tri-material interfaces is essential to evaluate the efficiency of such crack arresting devices, which are embedded in the foam core. Thereby, the research reported in the thesis seeks to establish optical measurement methods for characterising the stress fields at the crack tip and the crack tip parameters (i.e. fracture toughness) for foam cored composite sandwich structures, to provide a better understanding of the fracture behaviour at bi/tri-material interfaces. Thermoelastic stress analysis (TSA) is used to establish the stress state at interfacial cracks. In the vicinity of growing interfacial cracks large and discontinuous motion occurs that has a deleterious effect on TSA, therefore a new motion compensation (MC) method is required. The thesis describes how the MC was developed and demonstrates the technique is essential for quantitative evaluations in the neighbourhood of a growing crack. Thus, the full-field stress state at an interfacial crack is obtained reliably and with a high spatial resolution. An experimental method based on high speed infrared (IR) thermography is defined that enables the characterisation of fracture toughness by measuring the increase in temperature at the crack front (ΔT) during crack propagation. It is demonstrated that IR thermography with 15 kHz frame rate can be used to conduct a quantitative measurement of the crack front temperature associated with the crack growth. A constant of proportionality, ψ, is derived that links the temperature change per unit area at the crack front to the fracture toughness. It is shown that the ψ values obtained from specimens with the same interface are identical, even though the specimen dimensions and loading mode-mixities are different. It is demonstrated that by determining the values of ψ, the fracture toughness can be determined in any loading configuration by a direct temperature measurement. The efficiency of crack arresting devices or ‘peel stoppers’ is established using TSA. The efficiency is defined as the ability of the peel stopper to deflect the crack from the face sheet/core interface and arrest its growth. To enable effective crack deflection away from the interface, modifications of the initial peel stopper design are proposed. TSA identified the local stress concentrations introduced by the different peel stoppers and established the mechanisms that control crack propagation in the vicinity of the peel stoppers. Thereby defining a configuration that contains the face sheet detachment in mode I dominated loading. The work described in this thesis develops the optical measurement methods, based on imaging, to address the challenges in understanding interfacial fracture mechanisms and the determination of the crack tip parameters. This provides an important contribution to the understanding of the mechanics of interfacial fracture in foam cored composite sandwich structures.

Geotechnical engineering guidelines mandate the use of the most onerous hydraulic criteria for the design of earth retaining structures below the water table. Consequently, favourable local conditions, including the geometry of the structure, are not usually exploited. This means that retaining walls in particular are typically designed to resist hydrostatic pressures below the water table. Investigations have shown however that pore water pressures, axial stresses and bending moments reduce when groundwater seepage is allowed through the segmented linings of shallow tunnels. Contiguous pile retaining walls, by their nature, are also permeable. Allowing for groundwater seepage through the gaps in a retaining wall formed from contiguous piles could result in the pore water pressures on the active side of the wall being less than behind conventional impermeable retaining walls such as diaphragm walls. Numerical simulations, laboratory flow tank experiments and long-term field monitoring were conducted to determine the impact of pile gaps on the hydraulic conditions around contiguous piles. A relationship between the resulting bulk permeability of the equivalent structure and the pile gap was derived from 2D numerical analyses and verified by flow tank experiments. This expression can be used to calculate bulk permeability values for uniform retaining walls representing circular piles in 2D numerical simulations. The permeability relationship was used to calculate and assign equivalent bulk permeability values for a continuous retaining wall of uniform cross-section during the back analysis of the hydraulic conditions around the contiguous pile retaining wall at CTRL, Ashford. Pore water pressures and horizontal total stresses from the back analyses were consistent with those from the field measurements but were much lower than behind retaining walls formed from secant piles in similar conditions. Dimensionless charts were presented to estimate the groundwater level and the increased settlement observed behind contiguous pile retaining walls. The results demonstrated that the economic advantages of allowing through-wall seepage are greater than the perceived disadvantages.

Frames are traditionally designed as being perfectly pinned or perfectly rigid. These connections do not actually exist in real structures, with connection behaviour being between these extremes. Assuming that the connection is pinned leads to an overly conservative design as no moment transfer can occur at the connection, meaning bigger steel sections being used for the beams than are necessary. Site monitoring of the first building to make use of the semi-continuous design approach was conducted. The monitoring covered a primary and secondary beam during the construction and later loading of the structure. After the site monitoring the structure was modelled, using the ANSYS software package for finite element analysis. The results for this were compared to those from the site monitoring, and the results were found to be very close, so the model was used in further analyses. In the next set of analyses the connection stiffness and column base stiffnesses were changed, by making changes to the M-φ curve. It could then be seen how the joint stiffnesses affect the flexibility and strength of the frame. The model was then expanded to model multi-storey construction. Analysis techniques were used to show the trends that exist in the finite element results. These analyses were conducted on the results of frames with different section properties, different connection stiffnesses, and numbers of storeys.

This thesis is concerned with the damage assessment of composite structures under static loading case with geometric non-linearity. Both the finite element method and the boundary element method are studied for this purpose. A finite element based computational damage model is developed for predicting the nonlinear response, first-ply failure and ultimate collapse strength of uni-directional laminated composite plates. The damage model is implemened into the finite element program ABAQUS. It contains theory of large deformation and large strain. The model is then extended to laminated composite structures with woven fabric piles. A simplified model is developed for prediction of stiffness properties of woven fabric composite plates. In both the cases numerical results are compared against test values. It is demonstrated that excellent correlation with experimental results can be achieved. In the context of the boundary element method (BEM) the present research focuses on stress analysis of 2-D orthotropic structures. A novel technique is proposed for accurately computing the singular integrals in the 2-D boundary element method.

The experimental programme investigated the behaviour of 81 stocky concrete-filled E-glass FRP-composite columns subject to nominal eccentric load. The main parameters investigated were concrete strength, orientation of confining fibres, column slenderness and type of longitudinal FRP reinforcement. The experimental study found the greatest enhancement in load capacity was achieved using FRP-composites with fibres oriented in the hoop direction. The failure mode of the columns was governed by the type of longitudinal FRP reinforcement. Failure of columns reinforced with additional carbon FRP bars were initiated by the failure of the FRP. Additional E-glass FRP-composite bars as longitudinal reinforcement were found to increase the post-crushing stiffness and the ultimate load capacity of the column. The compressive failure strain of the E-glass reinforcement with triaxial confinement was comparable to the tensile failure strain. Second-order effects were found to be more significant at lower slenderness ratios in concrete-filled FRP-composite columns than for conventional R.C. columns. Furthermore, the benefits of FRP-confinement are negligible if the column slenderness ratio is greater than 12. The proposed design method for concrete-filled FRP-composite columns uses the CONFINE computer model developed as part of this research. The deflected profile of the column is assumed to be approximated by a part sinusoidal waveform. The CONFINE model gives good correlation with the experimental data for concrete-filled glass FRP-composite columns subjected to small eccentricities of load. The mathematics of the model enables predictions of the deflections, curvature and strains in the columns over the entire load history with a high degree of confidence.

This thesis is concerned with the development of a non-destructive testing technique, namely the impact-echo method, for the evaluation of tendon ducts in post-tensioned prestressed concrete bridges. The impact-echo method has been shown to be useful for damage detection in concrete structures, particularly plate-like structures. However, when it is applied to beam-like rather than plate-like structures the resulting spectra are complex and difficult to interpret. One source of complications is the excitation of cross-sectional modes in the beam. This thesis discusses an investigation into the nature of these modes, and their potential applications for non-destructive testing. Finite element models have been developed to investigate cross-sectional modes where grouted and ungrouted ducts are present and the results from these are described together with an experimental validation. It is concluded that a shift in the frequency of the dominant cross-sectional mode can be used to identify large ungrouted ducts. In addition, the frequencies of the dominant cross-sectional modes can be predicted for both grouted and ungrouted cases using finite element modelling. Further finite element analysis showed that varying the compression wave speed in grout over a realistic range had minimal effect on the amplitude and frequency of the dominant cross-sectional mode.

Three exact solutions are considered, within the context of the linear mathematical theory of elasticity, pertaining to the decay of self-equilibrated end loading for a class of geometries based upon polar coordinates. For a curved plane beam, an eigen-equation is derived, whose roots determine the rates of decay and degenerate to the Papkovitch-Fadle solution for the plane strain strip when the beam centre-line radius of curvature approaches infinity; this shows that the decay rates are largely insensitive to the beam curvature except for very small inner radius. For the plane and anti-plane elastic wedge, subjected to self-equilibrated loading on the inner or outer arcs, radial variation of stress is affected by a combination of free-edge stress interference and the convergent or divergent geometry. When the load is applied to the inner arc, the two effects act in concert in which case decay is assured; when the load is applied to the outer arc, the two effects act in opposition, and Saint-Venant’s principle (S.V.P) ceases to be applicable for wedge angles 2α > π for symmetric loading and anti-plane deformation, and 2α > 257º for asymmetric loading. It is concluded that the crack tip stress singularity, which is at the heart of Linear Elastic Fracture Mechanics, is attributable to the failure of S.V.P. for just one particular eigenmode for the wedge angle 2α = 2 π. A Finite Element-Transfer Matrix Method is developed for determination of decay rates of self-equilibrated end loading for frameworks and continuum prismatic beam of arbitrary cross-section. Nodal displacements and forces on either side of a repeating cell are considered as state variables and are related by a cell transfer matrix. Assuming consecutive state vectors to be related by a constant multiple λ leads directly to an eigenvalue problem; the decay factors, λ, are the eigenvalues of the symplectic transfer matrix. Eigenvalues occurs as reciprocal pairs (that is, if λi is an eigenvalue then so is 1/ λi) according to whether decay is from left to right or vice-versa. The multiple eigenvalues λi = 1/ λi = 1 are associated with the rigid body eigenvectors and their related principal vectors which describe the force transmission modes. The matrix of eigen-and principal vectors then forms a similarity matrix which transforms the original transfer matrix into Jordan canonical form. Both bi-orthogonality and symplectic adjoint orthogonality properties of the eigenvectors allow modal decomposition of an arbitrary end load. As a by-product of the method, it is possible to determine exact ‘continuum’ beam properties of the framework, which is useful in preliminary design work. The method is applied to the important case of a beam of rectangular cross-section for a wide range of aspect ratios. The Transfer Matrix Method is modified to become the Force or Displacement Transfer Matrix Method, which has the advantage of reducing in size the original transfer matrix by one half, and overcomes numerical ill-conditioning. Accuracies of all the developed methods are found to be very good when compared with available exact solutions.

This thesis reports a managerial investigation into structural engineering and its use of information technology. Structural engineering has developed from traditional professional origins and is increasingly operating in a more commercial and competitive environment. In addition, recent developments in computer software and hardware provide the mechanisms to semi-automate some of the structural engineer's work. Together these factors provide considerable impetus for change, presenting threats and opportunities. The study builds on previous work and is based on two significant questionnaire surveys. The first was sent to all UK based structural engineers and the second (the main research instrument) to UK structural engineering firms. The surveys investigate the commercial operating environment and the strategic use of information technology along with the related issues of education, training and changing work practices. The main survey provided 439 valid replies, which represents a 59% response rate (which is significantly higher than comparable studies). The work has analysed the potential benefits, constraints and threats presented by the available technology and the commercial environment. It has identified significant differences in the strategic use of information technology between the largest and smaller firms and provides evidence of industry restructuring. It analyses factors that impact on the ability of structural engineering firms to win work and issues that pose commercial threats. It provides statistical information and analytical tools that enable organisations to benchmark their operations and thus to maximise their use of information technology in the pursuit and execution of work. The study concludes that information technology can be used to semi-automate and enhance engineering design. It is a powerful strategic weapon but offers threats as well as opportunities and will have far reaching consequences for the structural engineering profession.

This thesis describes numerical investigations of the semi-rigid behaviour of the joints in precast concrete building frames. This research was one of three co-ordinated projects funded by EPSRC and carried out at Nottingham, Southampton and City Universities. It developed standardised finite element analysis techniques for modelling of reinforced concrete structures, improving the method of adding the smeared concrete material to overcome the early crushing phenomenon of the concrete elements. A number of experimental connections and precast concrete joints were numerically analysed to calibrate the finite element analysis techniques. Internal mechanism of the joints, such as cracking and crushing status of concrete components, stress distribution of the connectors and failure modes of the joints, all had to agree for the techniques to be acceptable. This calibration used connections tested in European Union countries through the COST initiative, including a T-stub connection, a bolt in concrete joint, and six billet-welded plate joints tested in the University of Nottingham. Sensitivity analyses defined the mechanical behaviour of the billet-welded plate joints when a wide range of parameters were varied within realistic engineering limits. Simplified calculations for defining the characteristics of the joints were proposed for use in semi-rigid analysis of complete building frames. Finally, the finite element analysis techniques were extended to the modelling of the innovative fibre reinforced plastic tube confined concrete columns tested at the University of Southampton. The smeared concrete technique developed was shown to be effective for modelling the performance of locally cracked, and crushed, concrete components with limited confinement. The technique is also effective for the analysis of completely confined concrete up to approximately double the normal uniaxial compressive strain, after which an analysis based on the crushed concrete's equivalent granular material properties was found to be more suitable.

Secondary or Parasitic internal bending moments in statically indeterminate continuous prestressed concrete beams are produced by the presence of the redundant reactions at the internal supports, assuming that the tendon profile does not coincide with the line of pressure produced in the concrete, i. e. that the tendon is not concordant. Whilst behaving in an elastic manner, these Secondary moments may be separated from the total internal bending moment (Secondary moment plus moment produced by applied loads) using various methods such as equivalent loads, together with a stiffness analysis. Beyond the Serviceability Limit State, the beam will begin to crack, altering the flexural stiffness as well as causing local increases in the tendon force, in turn altering the Secondary Moment. Therefore if the designer wishes to incorporate Moment Redistribution at the Ultimate Limit State, assuming that the critical sections have sufficient ductility, how to include Secondary Moments, if at all, is of much debate. Many different approaches have been adopted to shed light on the problem, the majority using nonlinear analyses of various forms. The advent of Finite Elements has seen an increase in their popularity for application to such nonlinear problems. This study begins with a Finite Element analysis of simply supported prestressed concrete beams, followed by an analysis of a two span example with a single curved tendon profile, fully bonded to the concrete, up to Ultimate Limit State. Although these analyses allowed an investigation of the overall nonlinear behaviour, the Secondary moment could not be separated from the Internal moment in the post cracking stage. As a result of the Finite Element Analyses, a method of applying equivalent loads past the Serviceability Limit State in an attempt to estimate the magnitude Secondary Moments up to the point where the beam is rendered statically determinate, was developed and utilised in a computer program. Two two-span beam examples and a three span beam, both with curved fully bonded parabolic tendon profiles were analysed to highlight the possible effects the secondary moment could have on the behaviour at Ultimate Limit State.

This thesis describes a numerical study and an experimental investigation of the behaviour of semi-rigid beam-to-column composite connections subject to wind loads. Connections in composite frames are normally designed as pin-jointed connections. Diagonal wind bracing is included in such buildings to resist wind loads. The use of semi-rigid connections in the design of such frames can provide substantial savings in material and erection costs. It also avoids the intrusion of bracing into the useful space within the building. The connection was studied numerically using a finite element package (ANSYS). The parameters affecting the behaviour of the connection have been used to develop two-dimensional finite element models for the different configurations. A three-dimensional model was also developed to check the accuracy of the two-dimensional analysis. Material properties, geometrical dimensions, boundary conditions and loading were carefully given as input data to represent, as realistically as possible, those of the actual joint. Material non-linearly was considered for concrete, steel, reinforcing bars and shear connectors. The bolt slip and the construction tolerance were also accounted for. The developed model was then used to determine not only the overall stiffness of the frame but also the stress and deformation distributions within the joint components. Different options for stiffening up the joint laterally were also studied. The models predicted with a very good accuracy the joint lateral stiffness under horizontal loading. Also they provided useful information which could not be obtained experimentally e.g., concrete stresses, tensile stresses in the reinforcing bars. An experimental investigation was carried out to verify the safe behavioural prediction of this type of connection. Because of the high cost and time consumption required to make the sample, a single and representative connection was selected on which the experimental work was carried out.

The 3-D equilibrium equations are used to determine the transverse shear stresses of laminated composite plates which are first analysed using a commercial FE package (ANSYS) based on the first-order shear deformation theory (FSDT). The results are compared with 3-D exact elasticity solutions. The application of the 3-D equilibrium equations to FEM produces very good transverse shear stress prediction for thin and medium thick (h/a≤0.1) symmetrically laminated plates. A two-phase procedure is combined with the FEM using re-analysis, in which limited ability of the method in improving the accuracy of the transverse shear stress prediction has been found. A new hierarchical finite element method (HFEM) is developed for laminated composite rectangular plates including geometrical non-linearity. The application of very high order polynomials as displacement functions makes it possible to model a whole rectangular plate as only one element(super-element). The difficulty in integrating the very high order polynomials needed to form the stiffness and mass matrices, is overcome by applying symbolic manipulation which yields the exact values of the integration. The von Karman geometrically nonlinear strain-displacement relationships are used. The transverse shear stresses are neglected for thin plates. In-plane displacements are included in the model. The HFEM is applied to linear free and forced vibration analysis, and to static geometrically non-linear analysis of isotropic and laminated rectangular plates. Very good numerical properties (convergence, stability, numerical efficiency et al) have been shown. In the application of the HFEM to geometrically nonlinear free vibration analysis of isotropic and laminated rectangular plates, the harmonic balance method is employed. The influence of in-plane displacements on static and dynamic nonlinear behaviour of rectangular plates is studied. The applicability of Berger's hypothesis to isotropic and laminated rectangular plates is discussed.

In this thesis, the developed stability analyses of unbraced rectangular steel structures are based on the transfer matrix technique. The elastic stability of rigidly-connected structures was considered first and the analysis, implemented on a microcomputer using FORTRAN-77, is based on the direct iterative technique. Then the formulation was extended to account for the effect of flexibility at beam-column connections, and the finite width of the connection. The connection behaviour was modelled linearly, as well as non-linearly, the latter being a more realistic representation of the physical model. The problem with the linear model was solved using the direct iterative as well as the linear incremental method, while for the non-linear model the problem was solved applying the linear incremental technique. The efficiency, effectiveness and accuracy of the transfer matrix technique in solving the elastic stability problem was confirmed by comparison of the results obtained with other previously found, experimentally and analytically. Comparisons were also made with the results obtained from an available finite element software, ANSYS. Excellent agreement was achieved. The formulation of the inelastic instability problem for a rigidly-connected structure was also developed and implemented on the computer. An approach which differs from any other conventional second-order elastic-plastic analysis was adopted. The plastic hinge in the stanchion was assumed to form a small finite distance from its end, equal to half the girder depth. The solution of the elasto-plastic problem is based on the linear incremental technique. The approach presented is relatively simple and has given very good results for the analyzed frames compared with other predictions.

Recently, there has been an increasing use of thin-walled open-section primary and secondary members in structural frameworks. These members are often connected off-centre, offering eccentric restraints to each other. Such restraints are capable of both causing local deformation to the members meeting at the point of restraint, and influencing the mode of instability of the framework. The object of this study is to investigate the effect of such restraints on the stability of such frameworks. To do so, a methodology for analysing arbitrary frameworks connected off-centre is developed, implemented on a FORTRAN computer system, and investigated experimentally. A 14x14 non-linear element stiffness matrix composed of the twelve standard, and two warping, degrees of freedom is developed. In order to ensure compatibility of displacement at the restraint point, a procedure for transforming this matrix, together with the externally applied forces, to coincide with the point of restraint is developed. This is implemented in a FORTRAN program, accounting for the orientation of the member in space. The program is thoroughly tested by analysing existing examples of beams, columns, and 2- and 3-D frameworks connected at their centroids. The structures are subject to axial and bending loading, supported in various ways, and exhibit a wide variety of buckling characteristics. The eigenvalue eigenvector approach is used in the program, employing the Vianello-Stodola technique, to obtain the stability mode and load. Two methods of evaluating the local flexibility matrix of a section are devised. One method applies a statically zero force-stress system of loading to an ANSYS beam model made from area elements. The resulting displacements obtained from an analysis of such a model are considered as the local flexibility of the section at the point of application of the force. In the second method, the local flexibility matrix is obtained from the displacements obtained when a unit force is applied to the ANSYS model, and those obtained when the model is analysed as having a rigid cross-section. A method of combining this matrix to the 14x14 stiffness matrix is devised and implemented in the program.

The Boundary Element Method is applied to problems involving plates loaded in their own plane. Initially, the membrane stress distribution is obtained through a boundary integral formulation of the plane stress problem in terms of the Airy stress function. Subsequently, a boundary integral formulation of the eigenvalue problem of plate buckling is developed which contains no curvatures and requires the modelling of the domain deflections only. A numerical implementation of the formulation discretizes both domain and boundary with one degree of freedom per domain node and two per boundary node. Various boundary and domain element types are tested. The transverse moments at corners, appearing in jump terms, are evaluated using polynomial approximations of the slope normal to the boundary. The deflection and normal slope of free boundaries are evaluated from the domain deflection model. The dual-reciprocity technique is applied to the domain integral appearing in the formulation for the critical buckling loads. The deflections are modelled using either trigonometric deflection shapes or a distribution of nodal unknowns related by a Fourier series approximation. A Fourier series scheme is then applied to the curvatures expression in the domain integrals and the Rayleigh-Green reciprocity equation is used to transform these integrals to the boundary. Incremental boundary integral equations of the large deflection problem of imperfect plates are derived. The curvatures are eliminated from the equation governing the bending deflection. The curvatures appearing in the equation for the membrane stresses are expressed in terms of the deflections using various nonlinear domain deflection models. Four systems of equations are obtained that are solved iteratively to account for the second-order terms. Computer programs written in FORTRAN-77 and run on a VAX/VMS microcomputer are developed. Convergence tests and parametric optimisation studies are performed. Results obtained from these programs are compared to analytical, numerical and experimental results. The boundary element method yielded very accurate results in almost all examples solved throughout the project.

This study presents an investigation into the possible applications of the transfer matrix method to the analysis of frame-shear wall systems. The objective is to examine the efficiency, simplicity and accuracy of the method in comparison with the most popular of the other techniques that have been used to analyse those structures. Had it been proved to be superior, this method would be promoted as an alternative simple tool for engineers and designers. Linear static and dynamic analyses of both two- and three-dimensional configurations were developed and implemented on computer using FORTRAN. The simplicity of those analyses was apparent, and their accuracy and efficiency were verified by analysing a wide range of geometries for which experimental and other analytical results were available. The validity, versatility and accuracy of the technique were further verified by extending its application to the nonlinear static analysis of two-dimensional reinforced concrete structures. An advanced model for steel-concrete interaction wsa used in determining the elements properties. The one-component model was used to represent the nonlinear behaviour of all elements. The systematic formulation has allowed developing a FORTRAN program capable of performing linear static, linear dynamic and nonlinear static analyses. Finally, a transfer matrix based procedure for the nonlinear dynamic analysis of two-dimensional reinforced concrete frame-shear wall structures was developed. Damping was accounted for in the form of a simple hysteretic model. A collocation scheme was used as the basis for the step by step numerical integration of the equations of motion. Both Giberson's nondegrading model and a simplified version of Takeda's degrading model were used in determining the hysteretic properties of all elements. The one-component beam model was used to represent the nonlinear behaviour of beam and column elements, while the multiple spring model was used for shear walls. The accuracy of the proposed procedure was tested on both linear and nonlinear problems. The levels of accuracy in the computed results were almost excellent for linear or mildly damaged structures, while it was acceptable for structures that experienced severe damage.

When attempting to control the vibration transmitted from a machine into and through the structure upon which it is mounted, it is desirable to be able to identify and quantify the vibration transmission paths in the structure. Knowledge of transmission path characteristics enables procedures to be carried out, for example, to reduce vibration levels at points remote from the source, perhaps with the objective of reducing unwanted radiation of sound. One method for obtaining transmission path information is to use the concept of vibrational power transmission. In any vibrating system there are a variety of mechanisms by which the vibration is transmitted through the substructure. The concept of power transmission allows the magnitude and direction of the various transmission paths to be compared. Simple measurements of vibration amplitude at various points in the structure do not suffice to identify the vibration paths. If stationary waves are present in the structure, large vibration amplitudes will be measured while power is being transmitted through the structure. Measurements are made of the vibrational power associated with flexural and longitudinal waves in beam-like structures carrying both wave types simultaneously. In order to make these measurements the relationship between measured wave amplitude and vibrational power is determined. Much care is taken in identifying the various sources of error in the measurements and the possible methods of eliminating these errors. Predictions are made of the vibrational power transmission in beams containing flexible sections and bent and branched beam systems. Models are developed which determine the wave type which carries most power in each section of the system. The theoretical predictions are confirmed by experimental observation. By establishing the wave type which predominantly causes power transmission it is then possible to apply the most suitable vibration control technique.

The nonlinear vibration of fully clamped beams and rectangular isotropic and laminated plates at large vibration amplitudes is examined in this work. Complementary theoretical and experimental studies have been carried out which show that the mode shapes and natural frequencies are amplitude dependent with accentuated nonlinear behaviour in the case of CFRP plates. A theoretical model has been developed, based on Hamilton's principle and spectral analysis, which allows all of the above cases to be treated in a unified manner and provides a rapid method for calculating the change in mode shapes and natural frequencies at large amplitudes of vibration. Measurements and analyses of the spatial distribution of the response harmonic distortion of a fully clamped rectangular isotropic plate show that the response second harmonic component is spatially distributed according to the second mode shape of the plate tested. An attempt has been made to develop a theoretical model for nonlinear free vibration of beams and plates taking into account the spatial distribution of harmonic distortion.

Analyses of reinforced concrete coupled shear wall structures are developed based on the transfer matrix technique. Initially an elastic analysis of coupled shear walls under static loads was applied to a parametric study of joint behaviour. The local-flexibility at the wall-beam interface as well as the overall joint flexibility were considered. The latter was modelled by treating the intermediate joint as a rectangular panel analysed as a finite element. Finally, analytical responses were obtained with the combined effect of the local with the overall flexibility as well as with line connection of successive wall segments. The formulation was then extended to a step-by-step incremental analysis of inelastic responses. The static load was divided into suitable finite increments. A simple non-linear one-component model was adopted to represent the elasto-plastic behaviour of beam elements. Emphasis was given to reinforced concrete as structural material. In the case of dynamic analysis, the basic principle lies in evaluating the responses for a series of short time increments, ΔT. Damping was incorporated into the formulation using constitutive law, and the equation of motion solved by adopting Wilson-θ numerical integration scheme. The non-linear behaviour was approximated as a sequence of successively changing linear systems. The method was tested throughout its development by comparison of its predictions with a large number of available experimental as well as analytical results. Very good agreement has been observed in most cases. Its application is thus fully justified in view of its efficiency and accuracy.

This work shows the use of the Direct Boundary Element Method in solving practical problems in three-dimensional, linear, steady-state elastodynamics in the frequency domain. Boundary integral equations for bounded bodies and external problems in infinite domains were reviewed. The boundary integral equations were solved numerically or semi-analytically and the problem boundary was discretized using discontinuous constant and linear triangular elements for both displacements and tractions. The choice of discontinuous triangular elements over continuous triangular ones has been discussed in detail. Integrations on a field element (i.e. an element which does not contain any source node) were performed using Hammer's numerical integration formula for triangular areas. However, integrations on a source element (i.e. an element containing one or more source nodes) were solved using two different approaches: one semi-analytical and the other numerical through element sub-division. The semi-analytical approach involves the use of a numerical procedure originally presented by H.R. Kutt and later modified by P.S. Theocaris for evaluating two-dimensional Cauchy principal value integrals. In the entirely numerical approach, the source element was divided into sub-elements, then the Hammer's numerical integration scheme was applied to approximate the integrals on each sub-element. The above formulations were implemented in computer programs written in Fortran. Numerical solutions, when compared with published analytical results, indicate that the semi-analytical approach produced more accurate solutions. They also show that the Direct Boundary Element Formulation using discontinuous triangular elements can be used for solving practical elastodynamic problems.

In this thesis a new method is presented for the analysis of partially prismatic thin-walled structures containing non-prismatic components. The method, known as the 'Mixed-Mode' method, combines the finite strip and finite element methods by discretising the prismatic part of the structure into finite strips and the non-prismatic parts into finite elements. Three types of non-prismatic component are considered; a transverse component, a longitudinal component and a spring support, the latter allowing multi-span structures to be analysed. The theory of the mixed-mode method is developed for linear static analysis and is thoroughly tested by analysing a series of representative structures. The equilibrium equations produced by the method are shown to possess a block structure corresponding to the harmonics of the finite strip discretisation. Advantage is taken of this in a block iterative solution procedure for the equilibrium equations. The efficiency of the iterative procedure is explored in a series of examples. The method is also used for dynamic analysis, specifically in a study of the free vibration problem. The method is again thoroughly tested by analysing a series of representative structures.

The available methods of improving earthquake response are examined and the conclusion is drawn that two areas show sufficient promise to be worth examining in detail:- 1 Isolation of the structure from the foundation by soft mechanical elements such as rubber pads. 2 Adding mechanical energy absorbing dampers (EADs) to the structure. The study of these two approaches is combined in a general study of the use of EADs in building structures. The isolation method simply becomes a special case of the more general approach adopted. In order to get a general feel for the effectiveness of EADs a series of time domain analyses is made of a ten storey building, using various configurations of EADs, with and without soft elements introduced into the structure. The ground motions used are five simulated events generated for the same length of time and for the same site conditions. These show that useful results can be obtained without introducing soft elements, but that if these are introduced, considerably better results are obtained. It is also shown that the amount of damping used has an important influence on response and, especially where soft elements are used, shows a distinct optimum value. A new configuration shows a performance superior to that of more conventional isolation methods. Because the use of time domain analysis is not generally a viable proposition for design purposes, an alternative approach is developed using a new method of time modelled random vibration analysis, which can be applied to systems with a limited number of discrete non-linearities. This brings the computational resource required to a practical level so that optimisation of the various parameters can be achieved. Design studies are made of alternative types of building using EADs, including cost effectiveness, and the sensitivity of the design to possible variations which may occur in the design parameters used. The final design analysis is compared with results from recorded earthquake acceleration records.

Plastic theory is now widely used for the design of building structures. Consequently, the prediction of the real behaviour of structures up to collapse, has been objective of many studies. The aim of this study is to develop a theoretical approach, based on the matrix stiffness method to the problem of the second-order elastic-plastic analysis of steel structures considering the effects of the spread of plasticity. In doing this, a new analytical approach to the calculation of the instantaneous stiffness and carry-over factors of elastic or inelastic, prismatic and non-prismatic, isolated beam-columuns was developed. In addition, a method was derived for the calculation of the exact moment-curvature-thrust relationship for partially plastic I- and box cross-sections. In order to verify the validity of the approach, eight major tests were carried out on isolated members formed from rectangular hollow sections. Measurements were taken through the elastic and inelastic stages, up to failure. The experimental results obtained confirmed the proposed theoretical method and excellent accuracy was achieved. Non-linear elastic and elastic-plastic analyses of steel structures are described. The computer program based on these is also capable of linear elastic and elastic-plastic analysis. Although there was no opportunity to carry out tests on frames, the method was checked against the experimental results from two frames tested by others. Furthermore a frame with uniformly distributed load members, and fifteen-storey rectangular and irregular frames were analysed. The results obtained agreed closely with the experimental results and with the results of other analyses. It was found that although the effects of spreading plasticity in an isolated member are large, in big structures the effect on the load carrying capacity is small. However, it was also found that deflection can be affected and that the distribution of plastic hinges at collapse is modified. The thesis also draws attention to possible developments in the analysis to include the effects of strain hardening and residual stress and shows how it can be extended to the analysis of reinforced concrete structures.

This study is concerned with the investigation of the effect of joints on the three-dimensional stability analysis of framework structures composed of thin-walled elements. Warping behaviour in the vicinity of framework joints is an important consideration in such an analysis. The main objective is to develop a solution procedure for investigating the warping behaviour of framework joints as well as demonstrating its effect on the three-dimensional stability of framework structures. A finite element technique is proposed for determining the relationships between the warping parameters (warping displacements and warping moments) at a joint. The technique is based on modelling a joint by a number of area elements with suitable stiffness proporties. The stresses and deformations resulting from the analysis of the finite element model are then used to obtain these relationships. The validity and accuracy of the finite element solution have been examined by analysing a number of problems for which exact solutions by other methods are available. The technique is also applied to study the warping behaviour of two joints which are of practical importance. Moreover, the finite element solution is compared with experimental observations taken from a test on a large-scale model of a joint and the comparison shows a good correlation between both results. The relationships of the warping parameters at rigid joints are then incorporated into a general three-dimensional stability analysis of framework structures by modifying an existing finite element stability formulation. Several illustrative examples are solved in order to assess the effect of the joint's warping behaviour on the buckling capacity of framework structures. Finally, an experimental program was developed to verify the theoretical predictions. Tests were conducted on a large-scale steel portal frame model to study its three-dimensional behaviour. Particular attention was focused on evaluating the effect of the warping behaviour of the joints and the stiffness of the secondary members on the buckling strength of the model.