Validation of a Stochastic Fuel Injection Model for Gas Turbine Combustors

Abstract

In gas turbine combustors, enhanced atomization through the whole combustor region is essential for satisfactory performance since droplet size and distribution can have direct impact on almost all key aspects of combustion. To predict these flows, KIVA-II, a three-dimensional full Reynolds-averaged Navier-Stokes solver with the capability to handle finite rate chemistry and liquid spray injection is utilized. The Monte-Carlo based spray model in KIVA-II was developed to predict the flows in internal combustion engines and includes submodels for drop injection, breakup, coalescence, and evaporation. To assess the validity of the spray model for gas turbine combustors, numerical flow field predictions have been compared with experimental data provided by University of California, Irvine (UCI) Combustion Laboratory. The predicted spray behavior is in satisfactory agreement between the numerical prediction and the experiment downstream near the fuel injector. However, far downstream of the nozzle exit the deviation between the numerical results and the experimental data increases.