Dry Low Emissions Premixer CCD Modeling and Validation

Abstract

Computational Combustion Dynamics has been used extensively at General Electric Company for combustion applications. This paper demonstrates an application of Advanced Combustion Code to GE’s lean premixed dry low NOx emissions LM2500 and LM6000 gas turbine combustors. A methodology for anchoring the Double Annular Counter-Rotating Swirler (DACRS) exit conditions to Laser Doppler Velocity data from a reacting single cup experiment is described. The DACRS exit velocity profiles and turbulence parameters are inlet boundary conditions for the annular combustor simulation. Since over 80 per cent of the total air enters the combustor via the premixers, inaccuracies in these boundary conditions have a significant impact on the predicted flame shape, liner temperatures and emissions. The paper shows comparisons between measured and predicted velocity in a rectangular duct equipped with a single DACRS. The k-ε turbulence model and the two-step eddy break up/eddy dissipation combustion models are used to predict the reacting flow field of the natural gas/air flame. The inlet velocity profiles are developed first to match the LV data and the observed flame impingement location at nominal settings of the inlet turbulence parameters. The sum square error between measured and predicted velocity is used as the optimization function. Next, a design of experiment computational study is conducted to determine the inlet turbulence length scale and kinetic energy in order to further improve the data match. The eddy break up model is shown to be more robust than the eddy dissipation model. The eddy dissipation model resulted in slow combustion rates, and high fuel and carbon monoxide emissions.