Search

Pressure vessels present a serious hazard when exposed to fires. Boiling liquid expanding vapour explosions (BLEVEs) and other associated vessel failure events are known to have significant consequences, and be key propagators of industrial accident scenarios. Understanding the response of pressure vessels to fire, and especially the rate of pressurisation, remains a significant challenge. This study reviews the ability of existing models to capture the physical processes that drive vessel pressurisation, and the existing fire test evidence used to validate such models. The scope of existing test evidence is found to be inadequate to validate complex numerical models. This study defines and describes a set of test conditions and a novel piece of experimental apparatus that can provide detailed and reproducible test evidence in an economic manner for the purpose of numerical model validation. The equipment included a 2.6 m3 vessel with a full cross-section (Ø1 m) glass window. A pressure compensation system maintained the window integrity, allowing combined temperature and velocity field measurements, using thermocouples and particle imaging velocimetry (PIV), to be made during fire exposure. Initial studies using ANSYS CFX indicate that the Eulerian-Eulerian multiphase model and the RPI wall boiling model are capable, when used together, of providing a good basis for simulation of the pressurisation rate, given the use of appropriate bubble-related parameters obtained by experiment.

The reactor pressure vessel (RPV) is a key component within a nuclear reactor, as it contains fission products and acts as a primary pressure boundary. The hostile conditions these components are subjected to means that the initial mechanical properties and microstructure at the start of life need to be well understood to enable design and safety allowances for their evolution in service to be made. RPVs have historically been manufactured from large thick section forgings, produced from ASME SA508 type steels and fabricated through submerged arc welding. Where the welded joints are paramount to maintaining the integrity and safe operation of the nuclear plant. However, through the collaboration of the Generation IV initiative [1] new materials and joining processes are being researched and trialled to prolong the reactor life. The effect of forging prior to heat treatment is a key feature to understand and it is important to determine what happens to the microstructure following the forming process prior to heat treatment. Through the use of a trial forging it was possible to establish that the effect of forging was not uniform across the whole component, with areas of high and low stress and strain identified. However, the effect of location within the forging was found to have less of an effect on the final properties achieved, compared to the effect of heat treatment. Heat treatment is carried out following the forging process to achieve the mechanical properties required for the application of reactor pressure vessels. However, to determine what happens to both the mechanical properties and microstructure and to investigate the relationship between the two, sections of the trial forging were subjected to different cooling rates and tempering times. It was found that air cooling leads to the formation of ferrite throughout the microstructure and this has a detrimental effect on the mechanical properties achieved. However, water quenching the component does not give the material time for ferrite grains to form and instead results in a lath structure of bainite. Water quenching combined with a tempering heat treatment that results in a Larson Miller Parameter of 19 was found to give preferred properties that are substantially higher than that of the other thermal treatment conditions examined. Narrow gap tungsten inert gas (NG-TIG) welding has also been investigated as a replacement for submerged arc welding of reactor pressure vessels. It was found that the NG-TIG weld provided a more refined microstructure in the heat affected zone and fusion regions of the samples compared to the heterogeneous microstructure achieved with submerged arc welding. This also contributed to the substantially improved mechanical properties, including increased ultimate tensile strength and hardness in each region of the weld microstructure. This research has established an understanding of an SA508 Grade 3 forging and the effect heat treatment and cooling rates have on the microstructure and mechanical properties that can be achieved with this grade of material. A new welding technique for the fabrication of reactor pressure vessels has also been determined to be an improved process, in which the microstructural refinement and enhanced properties achieved will potentially help to prolong the life of the RPV in the next generation of nuclear reactors.

Specifically, this thesis is an investigation into the use of Additive Manufacturing (AM) for the construction of a small rocket component whose fabrication by means of AM has not been researched extensively. In general, it is an example of how AM can be used to introduce complexity into simple, common parts for the purpose of innovation. AM is maturing into a disruptive technology allowing for the placement of payloads into Low Earth Orbit (LEO) and deep space with record-breaking cost reductions. Common AM practices allowing these cost reductions are the use of AM to produce whole systems with reduced amounts of material, manufacturing whole systems with fewer components, or constructing exotic geometries that are not obtainable by means of subtractive manufacturing but result in higher performance characteristics or efficiencies. During the preliminary literature review, it became apparent much of the research and development in the field of AM applications for spacecraft are on those systems providing opportunities for improved cost/benefit margins such as the propulsion system and human-spacecraft interfaces. However, I believe there is missed opportunity in terms of innovation by not researching the use of AM in the other systems. In April 2022, NBC News published an article stating that the cost of sending a pound of payload into space on the space shuttle cost nearly 30,000 USD (in 2021 dollars), while the emergence of modern rockets and their systems has allowed the cost to drop down to around 1,296 USD on the Falcon 9 rocket (Chow, 2022). While this 95.68% reduction in cost is significant, our imaginations are coming up with more expensive, capable, and heavier payloads which further drives our need to find additional ways to improve efficiency and reduce weight. One way we may be able to do this is by reevaluating systems that can be manufactured through AM processes and have a relatively uncomplicated function and construction. This lack of complexity implies that a new, and hopefully innovative, design approach will add little risk to performance, reliability, and safety while providing the opportunity to shave a few pounds. With the need to innovate in mind, I will design and manufacture artifacts representative of the traditional method by which modern spherical Pressure Vessels (PVs) are fabricated (essentially two hemispheres joined in the center by a weld) and a new, potentially innovative design only possible via AM and consisting of two unlike pieces that combine to form a spherical PV. The goal of this new design is to produce a PV that can withstand higher pressures as compared to a similar PV constructed through traditional means. Using FEA analysis and minor postprocessing, I will analyze both designs to evaluate if there is value in researching and re-thinking how we design, build, and use simple products. The results show both artifacts can withstand 100%, 125%, and 150% of their expected normal operating pressures. However, the “traditional” artifact will undergo catastrophic failure at a pressure that is about 2% greater than the point at which the “innovative” artifact experiences catastrophic failure.

This thesis presents a combination of numerical and experimental studies performed to assess the influence of the warm prestress effect on the cleavage fracture toughness of two ferritic pressure vessel steels. The aims of the research are to gain a detailed knowledge of the materials low temperature response under uniaxial and fracture conditions; to examine, using the finite element method, crack tip stress fields during warm prestress LUCF load cycles; and provide a clear and consistent method of classifying the warm prestress effect. An experimental programme investigated the room temperature and low temperature response of two candidate steels, A533B and BS1501. These steels were tested uniaxially under monotonic and cyclic conditions, and in the cracked condition in the as-received and warm prestressed conditions. Application of a three parameter statistical model to the experimental data showed that the distribution of data in the as received and warm prestressed conditions can be described accurately. The shift in the cleavage toughness distribution following warm prestressing was predicted by combining the statistical model with a validated analytical model of the warm prestress effect. Repeated proof loading was shown to increase cleavage toughness in A533B steel, providing the loading was load controlled. There were negligible effects of repeated proof loading on BS1501 steel. Some further enhancement of cleavage fracture toughness was observed when sub critical crack extension was introduced following warm prestressing, although the results were highly scattered. The finite element method was employed to simulate experimental fracture events. It was found from these simulations that fracture occurs following warm prestressing, when the reloaded crack tip stress distribution matches the as-received fracture crack tip stress distribution. The stress matching was observed to occur well into the elastic stress field ahead of the crack tip. This fracture criterion was employed to provide predictions of cleavage toughness following varying applied preload levels. The results were compared to experimental data sets and various analytical models. The Chell model of the warm prestress effect was observed to provide the best agreement with the finite element predictions. Crack tip blunting during the preload steps was found to have no influence on the predictions of cleavage fracture toughness. Differences in hardening response of the material was also shown to have little influence of the predictions of cleavage toughness. Simulations incorporating sub critical crack extension prior to reloading to fracture demonstrated that cleavage 'toughness can be enhanced further by limited crack extension. Large increments of crack growth were shown to reduce the warm prestress effect. The finite element predictions were validated against the appropriate analytical solution proposed by Chell and experimental results.

A combined theoretical and experimental study of the criteria governing the failure analysis of pressure vessels with defects has been performed. The fields of fracture mechanics (linear elastic and elastic-plastic behaviour), failure governed by large scale plastic deformation and energy balance methods are critically reviewed. All three approaches are shown to have relevance in the complete failure analysis of a structure with defects. An experimental study of the failure mechanisms of precracked polycarbonate and maraging steel plates and model polycarbonate vessels is presented. Room temperature, static loading tests are performed on 85 mm wide, 5 mm thick compact specimens of polycarbonate (0.013 andle; aandfrasl;W andle; 0.810). For comparison, 51 mm wide maraging steel compact specimens are monotonically loaded at room temperature. The influence of through-thickness constraint on the fracture toughness and slow crack growth characteristics of the steel is investigated using plates of varying thickness (3.1 mm-25.4 mm) and initial crack length (0.386 andle; aandfrasl;W andle; 0.766 for 3.1 mm thick sheet); various ancillary studies (scanning electron microscopy, surface deformation studies) complement the results. The crack growth behaviour of longitudinal through (0.704 ≤ c / andradic;‾DT/2 ≤ 1.434) and deep part-through (dandfrasl;T = 0.700 and 0.878) cracks in 50.8mm and 102mm diameter (5mm and 6mm wall thickness) polycarbonate cylinders is also studied. Bowling and Townley's two-criteria approach to failure i.e. LEFM on the one hand and limit analysis on the other, is shown to provide a useful method for assessing the relative importance of crack initiation, in the presence of limited crack tip plasticity and general yield as failure criteria for a given sized defect. Thus, for crack tip plasticity fully contained by an outer elastic field i.e. not general yield, the LEFM parameter, KIc (with the possibility of a plasticity correction factor for thin sheet) can be used to predict crack initiation. For the low strain-hardening maraging steel, Irwin's plane strain plasticity correction, 1andfrasl;6π (KIc/σy)2 is shown to be applicable to sheet thicknesses comparable to 1andfrasl;π (KIc/σy)2 i.e. twice Irwin's plane stress plastic zone radius.

Over the past decade signi cant interest in heat transfer during lling of pressure vessels has arisen due to overheating and under- lling when utilizing compressed hydrogen gas as a practical storage solution for clean energy sources. The problem is most clearly evident when hydrogen is stored as compressed gas for use in the automotive industry since carbon bre reinforced plastic (CFRP) vessels are used for their low weight. The low thermal conductivity of the plastic component of the vessel wall has an insulating e ect on the stored gas. During lling, the gas undergoes heating due to compression work, particularly during lling rates comparable to that of a traditional vehicle. As the CFRP material can be damaged by overheating, it is important to be able to predict the temperature of the gas, and the temperature distribution in the vessel wall. Key to this is understanding the heat transfer between the gas and the wall during lling. While signi cant progress has been made with larger vessels, few studies have considered the lower Reynolds number regimes. The aim of the PhD is to ll the gap in the literature for lower Reynolds number regimes by providing knowledge and fundamental data required for accurately predicting heat transfer during lling of gas pressure vessels. The methodologies being used include experimental work, numerical simulation and analytical methods. The thesis addresses understanding the heat transfer by experimentally producing benchmark quality temperature measurements for validation. Experiments have been undertaken which provide detailed measurements of thermal elds and heat transfer and the thermal response in the vessel. Computational uid dynamics (CFD) is used to determine local Nusselt numbers and understand the development of boundary layers within the vessel. An analytical solution has been developed by neglecting convection, essentially treating the vessel as a solid with and evenly distributed heat source. This model has provided an upper limit on the temperature rise within the vessel and lower limit on the heat transfer. The developed analytical solution elucidates the e ect of near wall compression heating and provides a theoretical limit for the minimum Nusselt number and maximum gas temperature rise in the vessel. The minimum Nusselt number limit is approached in the case of low Reynolds number. Taking the limit of small Reynolds number for a vessel being lled with compressed gas, the energy equation was found to asymptote to the unsteady heat conduction equation with heat generation and variable density. This equation was solved analytically for cylindrical and spherical geometry. Assuming the density changes linearly with time, a solution is obtained which can become identical in form to the constant density solution if the Fourier number is de ned using a log-mean density rather than the instantaneous density. For a constant energy input rate the Nusselt number was found to approach a constant value after an initial transient period. At such 'steady state' conditions, the Nusselt number based on the diameter for cylinders with aspect ratios larger than 1 rapidly approaches an integer solution of NuD = 8. For cylinders with aspect ratios less than 1, the Nusselt number based on the cylinder length (height) characteristic dimension rapidly approaches NuL = 6. It is shown experimentally and numerically that during compression lling, the heat transfer asymptotically approaches this analytical solution at low Reynolds numbers. Four distinct regimes were identi ed during the lling of the vessel. At the very beginning of lling, heat transfer is dominated by heat conduction and can be described by the analytical solution. Following this there is a period dominated by forced convection from the jet of gas entering the vessel. A signi cant temperature di erence and increased density of the gas causes natural convection currents to develop as the Rayleigh number increases. Both free and forced convection play an important role for a period of time which can be described as a mixed convection regime. As the lling rate and jet penetration into the ow eld decays forced convection e ects diminish leaving only free convection which continues after gas has stopped entering the vessel until thermal equilibrium is reached.