Simulating Extinction and Blow-off in Kerosene Swirl Spray Flames
Repository URI
Repository DOI
Change log
Authors
Abstract
Alternative jet fuels are being developed for use with existing jet engines, however there are still knowledge gaps concerning how unusual compositions and properties of these fuels will affect combustion performance. Physical and chemical processes leading to problematic engine stability phenomena like flame extinction and lean blow-off (LBO) are still not well-understood for conventional spray flames, but alternative fuels provide additional challenges as they have been observed to have increased variability from expected behaviour at conditions close to LBO. Evaporation is known to be the limiting factor for combustion in spray flames, and experiments have shown both gaseous and spray flames exhibit increased amounts of local extinctions as the equivalence ratio is decreased. The flame structure and transient behaviour of spray flames behave very differently compared to gaseous flames at near-blow-off conditions and during the blowoff transient. Fuel starvation has been proposed in past experiments as a significant reason for why spray flames blow off more quickly and at richer equivalence ratio compared to gaseous flames but has been explored very little in computational studies.
The prediction of fuel starvation and LBO phenomena using numerical simulations with detailed chemistry are the primary focus in this work. Large Eddy Simulations (LES) with the Conditional Moment Closure (CMC) turbulence-combustion model are used, as this methodology has shown good results in simulating extinction and blow-off in both gaseous and spray flames in a lab-scale bluff body swirl spray flame configuration. The jet fuels simulated are the Dagaut Jet-A1 surrogate and the U.S. National Jet Fuels Combustion Program (NJFCP) fuels of interest: A2, C1, and C5. A2 is a conventional Jet-A used as a reference fuel, whereas C1 and C5 are synthetic kerosenes with unusual fuel chemistry or liquid property characteristics. These NFJCP fuels are represented using the Hybrid Chemistry “HyChem” lumped pyrolysis detailed kinetic mechanisms.
Simulations in non-premixed laminar counterflow flamelet configurations are conducted at pressures of 1 atm and 10 atm for stable scalar dissipation value flamelets up to extinction, and during the extinction transient. Species trends in the three HyChem fuels and the Dagaut Jet-A1 surrogate are compared in detail. In comparison with experimental blow-off trends, only C5 deviates from expected behaviour and is the most robust fuel against extinction via high scalar dissipation rate. This highlights the interplay of both chemical and physical forces contributing to a real fuel’s tendency for LBO. Reignition of an extinguishing laminar flamelet using the HyChem A2 mechanism is also achieved through decrease of the scalar dissipation rate, although after a certain time the flamelet is not recoverable due to lack of chain-branching radical species.
A stable condition LES-CMC simulation using the HyChem A2 (Jet-A) chemical mechanism is used as a starting point and reference for lean blow-off simulations. The computational domain is based on the Cambridge bluff body swirl burner, with a structured LES mesh and a coarse structured CMC grid. The simulation is run using an Eulerian-Lagrangian framework for multiphase flow with the Abramzon and Sirignano evaporation model. Overall flame size and shape from the LES are fairly similar to experimental OH* and OH-PLIF with Mie scattering results, however there are significant differences in location of peak heat release rate and further work is required for validation of the simulations against experiments. CH is discussed as a promising experimental marker for local extinction and location of heat release.
Three fuel mass flow rates from the experimental blow-off curve for the Jet-A flame are simulated. The three simulations exhibited LBO at air flows between 5–20% greater than experimental bulk air blow-off velocities. Heat release rate decreased by at least 80% in the flame zone around the stoichiometric mixture fraction, however globally the combustor saw an increase in heat release rate due to the presence of unburnt droplets continuing to vaporise downstream. The asymmetric flame structure and duration of the blow-off transient in the simulations align very well with previous experiments with kerosene and other low-volatility fuels. The LBO transient lasted between 10–30 ms. Fuel starvation is suggested to be a driver of spray flame extinction, through decreased temperature and reduced evaporation caused by increased quantities of cold air in the system. Unburnt vaporised fuel remains in regions of temperature below 1200 K, where the fuel is no longer able to pyrolyse completely, resulting in non-flammable local mixtures. The quantity of local extinctions observed in both conditional and unconditional space is lower than expected compared to gaseous flames, and is linked to low values of the conditional scalar dissipation rate. Changing the model used to close the conditional scalar dissipation rate in the CMC equations is suggested as a potential way to improve the LBO results, as the Amplitude Mapping Closure model does not account for the very lean mixtures experienced at LBO conditions.