OCR Text |
Show vaporis~t.ion~ which takes ~la~e ~hen droplet temferature is between T vap and T boiling' (iii) droplet boiling and devolattlIsatIOn. The vapOrISatIOn IS based on the d law using experimental data from n-heptane. Formation of thermal NO was modelled via the extended Zeldovich mechanism using the three principle reactions governing the formation of thermal NO from molecular nitrogen: N2 + 0 <=> N + NO (5) N + 02 <=> 0 + NO (6) N +OH <=>H +NO (7) The rate constants used for these reactions are given in previous publications by the authors (4). The rates of formation of thermal NO depend on combustion temperatures and concentrations of 0 atom and OH radicals. During combustion process, 0 atom and OH radical concentrations overshoot equilibrium values (ie super-equilibrium). Therefore, the use of equilibrium concentrations for unstable species such as 0 and OH for thermal-NO predictions with reactions 5-7 can lead to significant errors. Here, the partial-equilibrium hypothesis was used to determine OH and H and 0 radical concentrations. A full explanation of the partial-equilibrium assumption and their rate constant are given elsewhere (5). Prompt NO is produced by reactions of N2 with fuel-derived radicals in regions near the flame zone of hydrocarbon fuels. In general, contributions of prompt NO to total NO emissions in high-temperature combustion processes are relatively small. Hence a global reaction rates obtained from experimental work was used to predict prompt-NO formation rates. Details are given elsewhere(5). NOx formation has neglible effects on flame chemistry because of the relatively small concentrations in which it is formed, and therefore the NO emissions were calculated using post-processing techniques (1, 5). In most practical combustion systems the flow is highly turbulent, suggesting temporal variations in species concentration and temperature influencing flame properties and species concentrations. Therefore, to prevent considerable errors which results from a laminar treatment of the NO formation rate, turbulence/chemistry interactions were modelled with a joint 2-variable probability density function (pdf) formalism. This technique has proved successful in previous studies(5-7) and uses an assumed shape pdf closure assumption. A more complete description of the code is given elsewhere(5-7). Fig. 2 show typical numerical solutions. Performance of the burner was investigated theoretically using the predicted flow patterns in conjunction with predicted temperatures and chemical compositions. The predicted results indicate that ':l recirculation area can be found at the comers of the burner tip and around the injected fuel , and as the flow progresses forward into the furnace this recirculation area increases as the fuel spray widens, underscoring the three dimensionality of the flow. The predicted temperature distribution inside the flame as shown in Fig. 3a widens at first with distance from the injection nozzle and decreases thereafter. The predicted flame shape indicates a flat and uniform flame with a maximum temperature of 2,680 K produced from the burner. The predicted thermal and prompt NO indicates that thermal NO is the major contributor (approximately 900/0 of the total NO formed). High concentrations of thermal NO were predicted in the vicinity of the mean stochiometric mixture-fraction contours ("flame front"), while prompt NO formation rates peaked on the fuel-rich side of the reaction zone near the burner tip. In general, the trends for predicted NO concentrations correspond to experimental fmdings for different injectors and injector settings. These results allowed optimisation of the fuel spray angles, fuel and oxygen nozzle-exit velocities, etc. A burner with a 100 - 250 kW output was built based on the numerical analysis and was examined experimentally. |