LES PREDICTIONS OF COMBUSTOR EMISSIONS IN AN AERO GAS TURBINE ENGINE

In this study, a Rolls-Royce production gas turbine combustor was analyzed using 3D Large Eddy Simulation (LES). The objective of the study was to evaluate the LES modeling approach for predicting emissions (CO and NOx) and pattern factor in liquidfueled combustors at practical engine conditions. Experimental data from Rolls-Royce, at high and low power conditions, were used to evaluate the model. Combustion LES and Reynolds Averaged NavierStokes (RANS) calculations were performed using a compressible, pressure-based, unstructured-grid flow solver within the CFD-ACE+ commercial software. LES solves the general transport equations for mass, momentum, energy, and chemical species at the gridand time-resolved scales of the flow. The Localized Dynamic subgrid Kinetic energy Model (LDKM) was used to model the unresolved (subgrid) turbulence and a 2-variable (mixture fraction and progress variable) assumed probability density function (PDF) method, with decoupled NOx, was used to model the unresolved turbulence-chemistry interactions. A two-step chemistry model (JETA → CO → equilibrium products) was utilized. Lagrangian tracking of spray parcels with source/sink terms for the Eulerian gasphase was included. LES produced unsteady turbulent structures that enhanced mixing compared to RANS. LES, at high power conditions, produced lower CO and NOx and better agreement with exit emissions data compared to RANS. At low power, the LES emissions of CO were improved over RANS, but were still poor due to simplifications in the CO mechanism. Improvements in the filtered (mean) CO reaction rate could be achieved by including turbulent fluctuations of CO. LES computational times on a 15 Personal Computer (PC) cluster were ~10 days. Application of the software to high fuel-air ratio combustors will be conducted in the future to investigate methods for reducing emissions.