Spray Model Evaluations for a Reverse Flow Combustor

Abstract

A previously studied computational fluid dynamic (CFD) fuel spray model used to characterize a pressure swirl atomizer under non-reacting conditions is implemented into a reverse flow combustor simulation. This spray model relies upon a stochastic secondary breakup model to predict the diameter distribution downstream of the atomizer and provides a robust method of simulating the fuel spray in CFD simulations. For 3 fuel nozzle shroud geometries, combustor simulations are performed using the aforementioned spray model and also a traditional model which is based upon inputting a diameter distribution measured experimentally. Reynolds averaged Navier-Stokes (RANS) and Large Eddy Simulation (LES) turbulence models are both used and comparisons of results are made amongst the two turbulence models. Rig test data was also available to compare with the two sets of CFD simulations. N O x emissions and combustor exit temperature metrics were used to compare CFD results to rig test data. The new spray model matches the performance of the current best practice approach at minimum and in some configurations yields more accurate results. Additionally, a blind validation study was performed with the new spray model and compared with rig data from 8 combustor configurations. In the first study, trends in N O x were correctly predicted by the LES/flamelet method with the new spray model however values were significantly under predicted. In both studies, the RANS/flamelet method with the new spray model is close in value of N O x with those seen in the rig data however the model misses the trend at some configurations. Overall, temperature metrics at the exit of the combustor are seen to be better predicted with LES.