The Direct Numerical/Large Eddy Simulation of Fuel Sprays into Combustors and Augmentors
Navy SBIR FY2010.1


Sol No.: Navy SBIR FY2010.1
Topic No.: N101-028
Topic Title: The Direct Numerical/Large Eddy Simulation of Fuel Sprays into Combustors and Augmentors
Proposal No.: N101-028-1348
Firm: TDA Research, Inc.
12345 W. 52nd Ave.
Wheat Ridge, Colorado 80033-1916
Contact: James Nabity
Phone: (303) 940-2313
Web Site: http://www.tda.com
Abstract: A turbine engine combustor can be modeled with computational fluid dynamics, however due to current constraints on computing speeds the engine models do not have sufficient detail for accurate quantitative predictions. The sub models for chemistry, turbulence, and fuel injection tend to be grossly simplified so that solutions can be obtained within reasonable timelines. Typically, spray models simply define the point of injection, the flow rate, the initial vectors for the droplets and the droplet size distribution. The trajectories of the evaporating droplets are then calculated. However, combustor performance largely depends upon the efficacy of fuel injection and subsequent atomization of the spray, yet simple spray models cannot predict the engine performance to be expected with a specific atomizer. Therefore, TDA Research and the University of Colorado propose to directly simulate the primary atomization of liquid sprays; the only numerical approach that can resolve all flow scales. The resulting model will then be coupled with turbulence models. We will use high-speed imaging and a PIV/particle sizer to obtain detailed near-field and far-field measurements of orifice and airblast atomizers to validate the model. In collaboration with GE we will build a CFX compatible model for accurate prediction of augmentor performance.
Benefits: The expense of gas turbine development could be greatly reduced if computational models were available that could accurately predict combustor performance. Currently, stochastic spray models are used in the ANSYS commercial CFD codes such as Fluent and CFX, but they only become accurate after they have been tuned with experimental data (i.e. the answer is known). Thus, a fuel injection model that can accurately predict primary fuel breakup from first principles will have broad application to the engine modeling community.

Return