| Summary: | For the past fifty years or so, there has been a great deal of interest in the use of water based explosion suppression systems, designed to mitigate or reduce the impact of thermal explosions and their consequential overpressures, which may be as high as 2MPa in outdoor environments. This level of interest has been heightened in more recent years due to a number of high loss explosion events including, Flixborough, UK (1974), Piper Alpha, North Sea (1998) and Buncefield, UK (2005). All of the previous research has focused on the suppression and mitigation proficiency of existing or new water deluge systems, which deploy sprays containing droplets 200≤D32≤1000μm. Where a high speed flame propagates through a region of spray containing such droplets, the flow ahead of the flame will hydrodynamically break up the droplets into fine mist, which in turn will act as a heat sink in the flame, with a resulting degree of suppression. These studies concluded that in most cases, existing deluge systems contributed to a global reduction in flame speed and thus caused a decrease in the resultant damaging overpressures. This present study however, is focused on the mitigation of slow moving deflagrations with resulting speeds of ≤30m/s. A flame travelling at such low relative speeds will not possess the inertia to inflict secondary atomisation by hydrodynamic break up. Consequently, the droplets within the spray must be small enough to extract heat in the short finite moments that the flame and droplets interact (approximately 0.03ms for a representative 1mm thick flame front). Previous theoretical studies have suggested that droplets, D32, in the order of 10μm - 20μm will be required to successfully mitigate combustion without relying on further droplet break up. To date, there have been no other published experimental studies in this area. An innovative high pressure atomiser known as a Spill Return Atomiser (SRA) was selected, which contained a unique swirl chamber and was originally developed for decontamination and disinfection. The efficient atomisation of the SRA produced fine sprays containing droplets, D32, 15μm - 20μm. A series of „cold trials‟ were conducted to further develop the single SRA, which manifested in the creation of several exclusive single and multiple spray options in counter, parallel and cross flow, with the direction of the propagating flame. These new configurations were supplied with deionised water at a liquid pressure of 13MPa and were qualitatively analysed using High Definition (HD) imagery and quantitatively characterised using non-intrusive laser techniques. During the development stages of this study the SRA spray cone angle was increased from 34.7˚ to 49.2˚and the exit orifice flow rate was raised from 0.295 L/min to 1.36 L/min. The increase in flow rate provided a number of spray options ranging from 17≤D32≤29μm, with liquid volume flux of 0.011 cm3/s/cm2 - 0.047cm3/s/cm2 and mean droplet velocity of 0m/s - 21.4m/s, with the resulting characteristics giving way to complete explosion mitigation qualities. The second phase of this study was to conceive, design and build a suitable apparatus capable of producing slow representative flame speeds within the range of 5 m/s - 30m/s. In excess of 250 mitigation „hot trials‟ were performed using the unique conformations produced during the „cold trials‟, whereby a configuration consisting of 4 x SRA‟s in cross flow (X/F) configuration, successfully and repeatedly, completely mitigated homogeneous methane-air mixtures throughout the whole flammable range E.R. 0.5≤(ϕ)1.0≤ 1.69 (5 - 15%), with flame speeds ranging from 5 - 30m/s. The combined spray configuration consisted of four SRA‟s which were 105mm apart and each opposed by 120˚, thus providing a total spray region of 315mm (spray centre to centre). As the sprays did not overlap or converge, the liquid volume flux remained as 0.047cm3/s/cm2. With droplets, D32, ≤30μm generally requiring impact velocities of approximately ≥142.83m/s to break up further, the flame speeds experienced in these trials of ≤30m/s would not have caused hydrodynamic break up of the droplets in the sprays. Therefore, due to the flame speeds and drop sizes utilised in this study, the droplets entering the flame front would have been in their original form. Although some comparisons were made using the experimental data with Computational Fluid Dynamics (CFD), it proved to be an extremely complicated phenomenon. This was due to the presence and interaction of the complexities of the combustion process and other variables such as water droplet dynamics and heat transfer modes. As such, a set of recommendations have therefore been proposed in pursuing this work in future projects.
|
|---|
