Spray combustion in moderate and intense low-oxygen conditions: An experimental study

Abstract

Nitric oxides (NOx) are formed as a byproduct of combustion and contribute to ground-level ozone formation and the creation of conditions harmful for human health. In response to air pollutant emissions regulations, technologies for controlling NOx formation for a entire spectrum of fuels ranging from gaseous, liquid, coals, biomass and residue-derived fuel are of great importance. The technology of Moderate or Intense-Low oxygen Dilution (MILD) combustion (also known as HiTAC or FLOX® ) holds the potential to maximize efficiency and reduce pollution emissions for liquid fuels. It relies on the dilution of recirculated combustion products in pre-heated air, yielding low temperature gradients within combustion chamber and NOx emissions. This study concerns an experimental investigation of spray flames generated in a laboratory-scale burner designed for the combustion of liquid fuels in MILD conditions. The objective was twofold: 1) reach fundamental understanding of the process and, 2) create a database useful for validation of computational models. The successful validation of these models will aid the design and optimization of industrial combustion systems. The configuration used in this study consists of a spray jet injected into a coaxial upward flow of either air or hot-diluted combustion products. The latter case resembles an environment with temperature and oxygen dilution typical for advanced MILD combustors. An important characteristic of the developed system is that the properties of the fluid entrained in the near burner zone are controlled independently of the progress of combustion in the spray flame or the flue gas composition downstream of the flame. By contrast, in a typical furnace or boiler configuration, the properties of the entrained flow are determined by in-furnace aerodynamics and cannot be varied independently. The greater flexibility of this 'spray-in-hot-coflow' burner compared to a furnace or a boiler configuration, makes possible to study systematically a range of different conditions. The spray was created using a commercial pressure-swirl atomizer. Ethanol and acetone were used as fuel because of their well-known physical properties and the availability of detailed and reduced chemical mechanisms for the combustion process. Flames with several combinations of spray and coflow conditions were studied in detail, namely 1) ethanol reacting sprays in air and hot-diluted coflow, 2) ethanol reacting sprays in different hot-diluted coflow conditions, and 3) acetone and ethanol sprays in an identical hot-diluted coflow. High-speed visualizations of liquid breakup were performed to provide insight on the atomization mechanisms. Complementary laserbased diagnostic techniques, PDA and CARS were employed to characterise the properties of gas and liquid phase in the spray region. PDA provided simultaneous measurements of droplet velocity and size statistics and CARS the gas-phase temperature statistics. For the first time, CARS has been applied to determine gas-phase temperature statistics in regions with high droplet density. The velocity and temperature statistics of the coflow were measured using respectively LDA and CARS. The composition of the coflow was measured using a flue gas analyzer. The coflow measurements together with the measurements in the spray region as close as possible to the atomizer, provides a dataset of inflow boundary conditions useful for numerical simulations. First, a comparative study of an ethanol spray in air and an ethanol spray in hot-diluted coflow was performed. These two test-cases serve as basis for comparison between conventional and MILD combustion of liquid fuels. The high-speed visualizations show that a liquid sheet cone emerges from the atomizer nozzle and disturbances, initiating from the nozzle tip, grow in space causing a local thinning of the liquid sheet and disruption. In the case of hot coflow, the presence of strong vaporization promotes the thinning of the liquid sheet and the onset of disruption takes place earlier. Although, in air and hot diluted coflow differences are observed on the onset of disruption, a similar droplet size distribution in the near atomizer region is observed for both cases and the resulting differences in the spray flame structure farther downstream stem from the different properties of the entrained coflow, i.e. temperature and oxygen dilution. In the case of air coflow, an inner and an outer flame-front is observed. In the case of hotdiluted coflow, the heat-release in the inner flamefront is substantially smaller. A significant reduction of temperature samples above 2000 K is observed in the outer flame-front for the hot-diluted case. Secondly, three ethanol sprays issuing in coflows with different temperature and oxygen dilutions were studied. The Weber number remains identical among the three cases and the temperature and the oxygen dilution were varied together. The results provide insights in the gas flow and the droplet distribution in the near atomizer field as well as the subsequent droplet dispersion. Complementary to this set of measurements, a parametric study was made concerning the sensitivity of ethanol spray flames with the injection pressure (and, therefore, initial liquid jet velocity) as well as the coflow conditions. A linear relationship was found between the lift-off height and the injection pressure which is different from what has been observed in a similar burner for a gaseous fuel jet in hot coflow. The lift-off height was found to depend on the droplet convective, vaporization and chemical time scales prior to ignition. The spray flame structure does not change with the different coflow conditions. At the outer flame-front, the peak temperatures above 2000 K are found to correlate with the coflow conditions. Higher coflow temperatures (and, therefore, higher droplet vaporization rates) together with the lower oxygen dilution lead to an increase of the peak temperatures. Thirdly, a comparative study was made of the structure of an ethanol and an acetone spray flame in identical coflow conditions. This study provides insight on the atomization mechanics, the stabilization processes and the resulting peak temperatures of liquid fuels with different physical properties burning in MILD conditions. Both flames are lifted but their visual characteristics are very different. The acetone case is found to have a higher lift-off height contrary to what would be expected based on comparison of the physical properties of the liquid fuels. High-speed visualizations show that for acetone cavitation occurs inside the atomizer. As consequence, the initial droplet distribution, turbulent dispersion of droplets and turbulence modulation of the gas-phase by droplets changes drastically and this is also at the basis of the larger lift-off height of the acetone flame. The two flames also drastically differ in the nature of the spray in the outer region. In the ethanol flame the outer region is predominantly filled by large droplets. In the acetone flame small droplets with low inertia are present and quickly follow the mixing structures contributing to vaporization and the formation of an ignitable mixture. The present study provides fundamental scientific knowledge and insight on the nature of MILD combustion for liquid fuels. In general it can be concluded that differences in properties of the recirculated hot combustion products yield different atomization characteristics and, subsequently, different turbulent dispersion and turbulence modulation of the gas-phase flow due to the presence of droplets. Although combustion in hot-diluted coflow conditions in general leads to absence of high peak temperatures, the degree in which peak temperatures are avoided depends on the interaction between coflow conditions and the characteristics of the atomization process. Additionally, the dataset offers new opportunities for model validation. Its value comes from the combination of several features such as: challenging complexity, degree of completeness of the dataset for each single case and availability of several cases with different flame structure and liquid fuels.