Turbulent combustion modeling of coal:biomass blends in a swirl burner.

摘要

In this work an existing numerical model for turbulent combustion (PCGC-2) is analyzed and improved with the primary goal of modeling coal:biomass blend combustion. The viability of blends as fuels is explored with the aim of reducing fuel costs associated with low sulfur coal. The existing model is designed for coal combustion and hence during the simulation of coal:manure combustion an equivalent fuel composition is used. The model is improved by converting the two mixture fraction model to a three mixture fraction model, solving additional turbulent scalar transport equations and appropriately modifying the particle phase routines.; The turbulent combustion model makes use of mixture fraction-probability density function approach which is computationally intensive. To improve the execution speed of the computer program it has been converted to enable simultaneous execution on multiple processors. The parallel execution results in a execution speedup of 70% of the maximum possible speedup. The computer program of the improved model has been structured in such a way as to facilitate future improvements.; For the case of coal combustion the old model assumes that all the mass evolved from coal particles is the same composition as that of parent coal. But the new model distinguishes between volatiles and char and predicts higher temperature spots in the near burner region when compared to the two mixture fraction model predictions.; It was found that for the case of coal:biomass blend combustion even if the amount of manure present in the blend is very small, in localized regions inside the combustor close to the inlet the mass fractions of manure and coal off gases are comparable. In these regions the old and new model predictions differ substantially. This result is important since ignition and flame stability are near burner regime phenomena.; Due to the complex nature of the processes involved in turbulent combustion this research represents only a modest attempt in modeling the process and hence the model predictions are compared with experiments only for qualitative trends.