Abstract:
Although Large Eddy Simulation (LES) has demonstrated its potential for
modelling the reaction in simple academic combustors, it is more
computationally expensive than Reynolds Averaged Navier-Stokes (RANS)
which has been used widely for industrial cases. The aim of this research is to
employ LES at minimal grid resolution and computational resource
requirements to capture the main characteristics of the reacting flows in a
helicopter combustor and exhaust plume with the focus on NOx emissions.
Test cases have been carried out to validate the current LES code for non-
reacting jet, non-premixed combustion and unstructured grids. Despite the
moderate grid refinement and simple chemistry models employed, the findings
from these test cases have demonstrated good capabilities of the current LES
to capture the mixing, flame and flow characteristics. In a farther test case, a
key gas-phase chemical reaction selected for the helicopter exhaust plume
modelling has also been tested.
The validated LES code is then employed in the numerical study of the reaction
in the helicopter combustor. The LES predictions in terms of the temperature
and EINOx agree generally well with the combustor design, analytical solutions,
previous LES and test measurements. Subsequently, the potential application
of LES for the calibration of simpler models has been assessed for the generic
and helicopter combustors. The results obtained from LES are compared with
those from a one-dimensional combustor performance and emissions code,
HEPHAESTUS, developed within the Cranfield University Power and
Propulsion Department. The discrepancies between the results are found to be
primarily due to specific simplification and assumptions established in the
HEPHAESTUS model which can be addressed.
Finally, LES has been employed to model the transformation of NO to NO2 in
the helicopter exhaust plume. The findings from this research have
demonstrated that, even without the implementation of highly dense mesh or
advanced reaction model, LES is able to provide results with an acceptable
level of fidelity at relatively low computational costs. These advantages make it
a powerful predictive tool for future design and emissions optimisation
investigations, and calibration of other simpler modelling approaches.
