Includes bibliographical references. / This dissertation reports on the results of an experimental and numerical investigation into the response of different materials to air-blast loading. Mild steel, armour steel (Armox 370T and 440T), Aluminium alloy 5083-H116, Twintex and Dyneema square plates were blast loaded on a horizontal pendulum at the Blast Impact and Survivability Research Unit (BISRU), University of Cape Town. The blasts were generated by detonating disc-shaped PE4 explosives of various diameters and standoff distances. The chosen plates are of side length 500mm (4mm thick mild steel and armour steel plates) and side length of 400mm (aluminium, Twintex and Dyneema panels). The charge mass was varied between 9g and 60g for two charge diameters, namely: 50mm and 75mm, and stand-off distances of 25mm, 38mm and 50mm. A polystyrene bridge was used to position the charges at the centre of plates, without any polystyrene between the charge and the plate in order to minimise any effects the polystyrene may have had on the plate deformation. The transient response of the 500mm square plates (mild steel and Armox 370T) was measured with the use of Light Interference Equipment (LIE) and numerical simulations performed in ANSYS AUTODYN, with the aim of gaining greater insight into the response of the two different materials. The details of the experimental setup and method used for the LIE as well as the development of the AUTODYN computational model are presented. The air and explosive were modelled as Arbitrary Langrange-Euler (ALE) elements while the test plates were modelled as Langrangian shell elements. Since the geometry of the plates was square, the simulations had to be performed in 3D quarter-symmetry. The transient response, permanent final displacement and maximum transient displacement of the numerical simulations were compared to the corresponding experimental results. The mild steel plates all exhibited good correlation between experimental and simulated results. However, the Armox 370T simulated results showed an under-prediction of the displacement magnitude and impulse compared to the experimental results. Experimentally, both the mild steel and armour steel exhibited a linear increase in deformation with increasing charge mass. Blast tests were also performed on 3mm thick mild steel, aluminium, Twintex and Dyneema square plates of 400mm side length. The aim was to gain a greater understanding and compare of the response of different material types (ferrous, non-ferrous, Glass Fibre Polypropylene and Ultra High Molecular Weight Polyethylene) under blast loading. The aluminium plates performed better than the mild steel, on an equivalent mass basis, in terms of permanent displacements and failure threshold impulse. The aluminium plates were significantly thicker (10.5mm compared to 3mm) than the mild steel plates, which may have contributed to its response under blast. The Twintex panels mostly exhibited failure in the form of fibre fracture and matrix failure whereas the Dyneema panels only exhibit large inelastic deformation, although the Dyneema were clamped differently to the other panels. Dimensionless analysis was performed on all of the materials except for Dyneema. Initially a scaling factor was used to account for the varying stand-off distances but proved to be unnecessary due to the type of confinement used (unconfined free air-blasts versus partially confined tube). Once the scaling factor was removed, the dimensionless impulse values showed relatively good linear correlation with the predicted trend.

The impact of improvised explosive devices (IEDs) on the safety of civilians can be devastating, especially when solid objects are inserted into the explosives. These inserts are propelled at high speed and increase the lethality of an IED detonation. Due to the wide range of possible IED configurations, a fundamental understanding of momentum transfer from explosives to the solid inserts is required. This project investigated the influence of charge geometry on the velocity of a 5 mm diameter stainless steel ball bearing. The ball bearing was half-buried and centrally placed on the at face of a cylindrical charge which was detonated centrally on the opposite face. The geometric parameters of interest were the charge diameter and the charge aspect ratio (length/diameter). Investigations were carried out in the project through blast and impact experiments as well as numerical simulations. The impact velocity of the explosively driven ball bearing was inferred using the impact crater depth on a witness plate. The correlation between crater depth and the impact velocity was determined using impact experiments which was performed using a gas gun. The average velocity (between detonation and impact) was captured by tracking the time of detonation and impact. The time of impact was recorded through a Hopkinson Pressure Bar (HPB) behind the witness plate. Additionally, the total axial impulse and the localised impulse, over the face of the HPB, were recorded by a ballistic pendulum and the HPB. Numerical simulations were conducted using a commercial software, Ansys Autodyn 18.0. The blast arrangement was simulated using a two-dimensional, axisymmetric model. The maximum velocity, average velocity, impact velocity, total axial impulse and localised impulse were 'extracted' from the simulations. The simulated velocities agreed well with experimental measurements, showing less than 2% variation. The deformed shape of the blasted ball bearings displayed similar characteristics to the model predictions. There were differences in the simulated impulse, with the numerical model predicting higher magnitudes but a less localised distribution. For a constant charge diameter, the bearing velocity increased in a nearly logarithmic manner with the increase in aspect ratio until a critical aspect ratio of <math><msqrt><mi>3</mi></msqrt></math>/2 was reached. At a constant charge mass, the bearing velocity decreased with the increase in charge diameter. The numerical model suggested that the influence of charge geometry on the bearing velocity was likely caused by the shape of the detonation pressure waves. The detonation pressure profile is sensitive to the charge aspect ratio and the diameter.

Composite materials are increasingly being used in a wide range of structural applications. These applications range from bicycle frames and building facades to hulls of marine ships. Their popularity is due to the high specific strength and stiffness properties, corrosion resistance, and the ability to tailor their properties to a required application. With the increasing use of composites, there is a need to better understand the material and damage behaviour of these structures. In recent years, the increased frequency of wars and terror attacks have prompted investigations into composite failure processes resulting from air-blast. Most of the research has been focused on flat panels, whereas there is relatively little on curved structures. This dissertation reports on the effect of air-blast loading on concave, singly curved fibre reinforced sandwich and composite panels. Sandwich panels and equivalent mass glass fibre laminates were manufactured and tested. Three types of curvature namely a flat panel (with infinite curvature), a curvature of 1000 mm radius and a curvature of 500 mm radius were produced, to determine the influence of curvature on panel response. The laminates were made from 16 layers of 400 g/m² plain weave glass fibre infused with Prime 20 LV epoxy resin. The sandwich panels consisted of a 15 mm thick Airex C70:75 core sandwiched between the 12 layers of 400 g/m² plain weave glass fibre and infused with Prime 20 LV epoxy resin. This arrangement produced a balanced sandwich panel with 6 layers of glass fibre on the front and back respectively. For all panels, vacuum infusion was used to manufacture in a single shot process. Mechanical properties of samples were tested for consistency in manufacturing. It was found that mechanical properties of the samples tested were consistent with low standard deviations on tensile and flexural strength. The panels were tested in the blast chamber flat the University of Cape Town. Blast specimens were clamped onto a pendulum to facilitate impulse measurement. Discs of plastic explosive, with charge masses ranging from 10 g to 25 g, were detonated. After blast testing, a post-mortem analysis of the damaged panels was conducted. Post-mortem analysis revealed that the failure progression was the same irrespective of curvature for both the sandwich panels and the laminates. Sandwich panels exhibited the following failure progression: delamination, matrix failure, core crushing, core shear, core fragmentation, core penetration and fibre fracture. The laminates displayed the following progression: delamination, matrix failure and fibre fracture. Curved panels exhibited failure initiation at lower charge masses than the flat panels. As the curvature increased, the failure modes initiated at lower charge masses. For example, as the charge mass was increased to 12.5 g the front face sheets of the flat and the 1000 mm radius sandwich panels exhibited fibre fracture, but the 500 mm radius sandwich panel exhibited fibre fracture and rupture through the thickness of the front face sheet. The 500 mm radius laminate exhibited front face failure earlier (15 g) than the 1000 mm radius (22.5 g) and flat panel (20 g). Curved laminates exhibited a favoured delamination pattern along the curved edges of the panel for both 1000 mm and 500 mm radii laminates. As the curvature increased, more delamination was evident on the curved edges. The curved panels displayed more severe damage than flat panels at identical charge masses. Curved sandwich panels experienced through thickness rupture at 20 g charge mass whereas the curved laminates did not exhibit rupture at 25 g charge mass. The flat laminates were the most blast resistant, showing no through-thickness penetration at 25 g (the highest charge mass tested) and initiated failure modes at higher charge masses when compared to the other configurations.

Bibliography: pages 68-69. / This work describes what is primarily an investigation into methods for estimating the maximum permanent deformation of a circular cylindrical shell subjected to internal explosive loadings. A complete rigid-plastic analysis of the transient response is performed. Subsequently the effects of material properties are included. Finally the theoretical predictions are compared with the experimental results obtained from a series of tests on aluminium shell specimens.

Explosive threats have become more prevalent in both military and terrorist theatres of conflict, showing up largely in the form of Improvised Explosive Devices (IED) which are often buried in soil to conceal them and increase their effectiveness. The response of a structure subjected to a blast load is influenced by many factors, namely stand off distance, mass of explosive, degrees of confinement and medium surrounding the charge. This study focuses on characterizing the transient deformation of test plates which have been exposed to different explosive loading conditions including free air blasts (AIR), backed charge (VBP) and buried charge (SBP) configurations. In the three loading configurations, four charge masses are considered, utilizing 10g, 15g, 20g and 25g masses of PE4 plastic explosive which were moulded into cylindrical charges of a constant 38mm diameter. The transient deformation of the test plates was captured using high speed Digital Image Correlation (DIC), which utilized two high speed cameras to record the experiments. Extensive modifications to the blast pendulum to incorporate the cameras was necessary to adapt this technique in a different method to that used in previous literature. The mounting method proposed allowed the cameras to record the experiment while capturing the impulse imparted on a test plate using a blast pendulum. The experimental plates exhibited only Mode I failure, which is plastic deformation, enabling the effect of different loading configurations on the transient and final plate deformation profiles to be identified. Numerical simulations of the experiments were developed to further the understanding of the load arising from the three configurations and the deformation mechanisms involved. The experimental results are used to validate the numerical models, which allow for a better understanding of the evolution of the deformation and strains across the plate. The transient data for the numerical simulation and the experiments were found to match closely. This work clearly shows the effect that the different loading conditions have on the tests plates, specifically the impulse distributions and transient strain in the plates. It was observed in this study that the impulse imparted on a test plate increases with the addition of sand while keeping other test conditions constant. The impulse recorded was observed to increase by 490-540% and 19-100% when compared to AIR and VBP 50mm SOD tests respectively. The loading profile acting on the test plate as a result of the specific impulse changes significantly with the inclusion of sand. The midpoint deflection increases with a decrease in stand off distance, increase in charge mass, increase in level of confinement or the inclusion of an overburden of sand. The observed increase in midpoint deflection of between 90-160% and 30-40% when compared to AIR and VBP 50mm SOD tests respectively was reported. The transient plate profile does not match the final deformation profile.

Structures deform at high strain rates and temperatures when exposed to impulsive loads. To accommodate the macro change there are microstructural changes that occur, i.e., grain morphology and shear banding. Most studies report on macroscopic response, i.e., large inelastic deformation and tearing of the structure, while limited studies have reported on microscopic changes that develop in the structure. The microstructure is directly related to the mechanical properties and performance of the material. Therefore, understanding the effect of high strain rate loadings on the microstructural evolution and subsequent mechanical properties of metals and alloys is necessary for mechanical design. The main objective of this research was to investigate microstructural changes to characterise the strain distribution and plastic deformation, owing to impulsive loading. Features within the microstructure that could be used to characterise deformation included grain size morphology changes, the presence of shear bands and sub-grain networks. The electron backscatter diffraction (EBSD) technique, which used Kikuchi patterns to characterise the strain distribution in the crystal of the deformed material, was also used as a characterisation tool. The first step in the experimental procedure was to select the appropriate material to investigate these microstructural changes. There was also the systematic investigation into the use of single and double heat treatments. These were used to achieve a large equiaxed grain structure, which was desirable from a microstructural point of view but was not desirable for blast-resistant material selection. The two-step heat treatment was concluded to be the most suitable heat treatment for the annealing and homogenisation of the AISI 430 stainless steel plates. The AISI 430 stainless steel plates used were 244 mm by 244 mm in size and had a circular exposed area of 106 mm. These plates were subjected to repeated explosive blasts, using a plastic explosive (PE4). The charge mass was varied for each test and the stand-off distance was kept constant at 150 mm for uniform loads and 13 mm for localised loads. Two plates were selected to investigate the uniform loading scenario. The first plate, a torn plate, used a charge mass of 30 g and one blast and the second plate, an inelastically-deformed plate, used a charge mass of 10 g and was exposed to three blasts. These two plates offered the same overall charge load with a different strain path. A further two plates were chosen for the investigation into the localised loading scenario. One plate, a petalled plate, used a 6 g charge mass and was exposed to two blasts and the second plate, an inelastically-deformed plate, used a 5 g charge mass and was also exposed to two blasts. The latter two plates offered an investigation into the effect of an increased charge load, where charge load affected the strain rate of the deformation resulting from the blast load. All four plates were sectioned across the midline of the dome and then ground and polished to a mirror finish, using OP-S. The polished samples were analysed, using optical microscopy and EBSD. In addition, Vickers hardness tests were carried out along the midline of the sectional plate profiles, in order to evaluate the extent of strain hardening. All the plates showed either a response of inelastically deforming or of complete or partial tearing failures when subjected to blast loads. For inelastic deformation failures, a global dome was characteristic of the uniform loading condition and an inner dome superimposed by the global dome was characteristic of the localised loading condition. Variation of charge mass and the number of blasts showed an increasing linear relationship between the impulse and midpoint deflection. The macrostructure showed a large variation of failures in the localised condition. The microstructural characterisation results produced micrographs showing regions of long, flat grains with multiple sub-grain networks, indicating deformed microstructures of the blast loaded plates. Parts of the microstructures displayed equiaxed/recrystallised grains characteristic of restoration processes, owing to high temperatures. Vickers hardness tests indicated an increase in material hardness as the number of blasts was increased, with a maximum hardness in the central region of the plates. In the first investigation, into uniform loading, the material characterisation results, combined with the fractography results, indicated brittle failure modes typical of high strain rate failures in strain rate sensitive materials, such as the chosen AISI 430 stainless steel plates. In the second investigation, into localised loading, the material characterisation results, combined with the fractography results, indicated a more ductile failure, owing to a 1 g incremental increase of charge mass, which confirmed the strain rate sensitivity of this material.

Includes bibliographical references (leaves 131-134). / An investigation into the deformation and tearing of stiffened quadrangular plates subjected to a uniform blast load is presented. A series of experimental results and numerical modelling using the finite element package; ABAQUS, on built-in quadrangular mild steel plates of different stiffener configurations and sizes subjected to a uniform blast load are reported. The main objectives of this investigation are to determine the dynamic response of stiffened quadrangular plates subjected to uniform blast loads, to assess the effect of the stiffener configuration and size on plate failure and to use a new approach that uses material properties that include temperature dependency to model the plate response. The experimental procedure consists of creating an impulsive load with the use of plastic explosive and measuring the resulting impulse using a ballistic pendulum. Explosive is centrally laid out in two concentric rectangular annuli on quadrangular plates of thickness 1.6mm with stiffeners of sizes; 3x3mm, 3x7mm, 4x3mm and 4x7mm; and configurations; none, single, double, cross and double cross; to provide the impulse required to give deformations up to plate tearing. In all the tests of Mode I category of large inelastic deformation, the plate profiles are characterised by a uniform global dome. The results of mid-point deflection versus impulse for the various stiffener sizes and configurations for Mode I show a generally linear relationship. In all the experiments, thinning mechanisms at the boundary are observed for all plates despite different stiffener sizes and configurations. Thinning, however, is not consistent all around the boundary. Thinning is also observed at the stiffener side closest to the boundary for double and double cross stiffened plates. There is, furthermore, a reduction in the stiffener width where two stiffeners cross each other perpendicularly.