2. School of Aerospace Engineering MITE Numerical Modeling of Compressor and Combustor Flows Suresh Menon, Lakshmi N. Sankar Won Wook Kim S. Pannala, S. Niazi, C. Rivera, A. Stein School of Aerospace Engineering Georgia Tech, Atlanta, GA 30332-0150

3. School of Aerospace Engineering MITE RESEARCH OBJECTIVES Develop first-principles based tools for modeling flow through axial and centrifugal compressors. Develop first-principles based tools for modeling two-phase reacting flow within combustors. Use these tools to explore control strategies for stable operation of compressors and combustors.

4. School of Aerospace Engineering MITE Compressor Modeling: Progress To Date A two-dimensional rotor-stator Navier-Stokes code has been developed, and used to model rotating stall. A reduced order model based on 2-D simulations has been developed, and validated. 3-D Navier-Stokes simulations have been completed for a NASA centrifugal compressor configuration. Stable operation of the 3-D configuration has been achieved at low mass flow rates using passive control devices.

5. School of Aerospace Engineering MITE Two-Dimensional Flow Solver Solves compressible Navier-Stokes equations for Rotor-Stator Configurations. Can model oscillating blades, inflow and downstream disturbances. Has been extensively validated. (Rivera, Ph. D. Dissertation, May 1998.) Some validation studies were presented last year. Forms the basis for the new Reduced Order Model.

6. School of Aerospace Engineering MITE 1 2 3 4 5 6 7 8 9 REDUCED ORDER MODEL Flow Field is divided into Macro-zones. In each zone, there are 4 states - , u, v and T

7. School of Aerospace Engineering MITE Reduced Order Model II Current Zone Neighbor Zone In each zone, the governing equations are applied: A coupled system of ODEs result.

8. School of Aerospace Engineering MITE Reduced Order Model III This system of simultaneous nonlinear ordinary differential equations couples states from all the zones Steady state solution yields performance map. The unsteady solution may be used to analyze the nonlinear dynamics of the system.

9. School of Aerospace Engineering MITE Compressor Performance Map

10. School of Aerospace Engineering MITE 1 2 3 4 5 6 7 8 9 REDUCED ORDER MODEL Incoming Disturbances may be inexpensively modeled. Throttle effects may be inexpensively modeled, and system transients studied.

11. School of Aerospace Engineering MITE NASA Low Speed Centrifugal Compressor 20 Full Blades with 55° Backsweep Inlet Diameter 0.87 m Exit Diameter 1.52 m Design Conditions: –Mass Flow Rate 30 kg/sec –1862 RPM –Total Pressure Ratio 1.14 SIMULATION SETUP

12. School of Aerospace Engineering MITE Single Passage Grid Modeling 3-D SIMULATION SETUP Grid Size: 129x61x41 = 322,629 points

13. School of Aerospace Engineering MITE Inlet: p 0,T 0,v,w specified; Characteristic equation solved to model acoustic waves leaving the domain. Diffuser Exit: p back specified; entropy and vorticity are extrapolated from Interior. Periodic Boundaries: Flow properties are periodic from blade to blade. Blade Surface: no-slip velocity conditions. 3-D SIMULATION SETUP Boundary Conditions

14. School of Aerospace Engineering MITE Surface Pressure Distribution Computations Vs. Measurements

15. School of Aerospace Engineering MITE Surface Pressure Distribution Computations Vs. Measurements

16. School of Aerospace Engineering MITE Compressor Performance Characteristics CFD without bleeding

17. School of Aerospace Engineering MITE Grid Sensitivity Impeller Performance Map for LSCC

18. School of Aerospace Engineering MITE Velocity Field (Colored by Pressure) RESULTS (Design Conditions ) Diffuser Region is Well Behaved No Separation

19. School of Aerospace Engineering MITE Velocity Field (Colored by Pressure) RESULTS (Off-Design Conditions ) Diffuser Region Shows Small Separation Onset of Instabilities

20. School of Aerospace Engineering MITE Effects of Bleeding on Diffuser Performance Without bleed With bleed

21. School of Aerospace Engineering MITE Compressor Simulations: Conclusions A new CFD based reduced order model has been developed and validated. A 3-D unsteady compressible flow solver for modeling centrifugal compressors has been developed and validated. Good agreement with experiments have been obtained for a Low Speed Centrifugal Compressor (LSCC) tested at NASA Lewis Research Center. For the LSCC, flow instabilities were found to originate in the diffuser region. Stall control by the use of bleed valves on the diffuser walls has been computationally demonstrated.

22. School of Aerospace Engineering MITE Combustor Modeling - Progress To Date A stand-alone methodology for droplet convection, vaporization, turbulent mixing and chemical reaction has been developed, and was reported last year. During the current period, this methodology was successfully coupled to gas-phase unsteady flow solvers. Incompressible and compressible versions of the two phase flow solvers have been developed. Ability of the methodology to track particles injected into a vortex has been verified. Validation against Ga Tech experiments are in progress.

23. School of Aerospace Engineering MITE Droplet Trajectory Droplets see local flow properties (Temperature and Velocity). Energy, Mass Transferred to subgrid. Momentum transferred to the supergrid. Droplets below a cut-off radius are modeled in the subgrid till vaporization is complete.

24. School of Aerospace Engineering MITE Features of the Present Approach Present subgrid approach is more efficient than other LES schemes where a very fine multi- dimensional subgrid is needed to model the droplets. In conventional Lagrangian schemes, all the coupling between the droplet and the gas phase is via the supergrid. In the present approach, only the momentum of gas and liquid phase is coupled via the supergrid. Conventional Lagrangian schemes assume droplets vaporize instantaneously, below a cut-off radius. This can give erroneous results.

25. School of Aerospace Engineering MITE Mixing Layer Simulations with Droplets 3-D Shear layer, on which diturbances corresponding to first unstable mode are imposed. Seed Particles

26. School of Aerospace Engineering MITE Present Model Correctly Models Large and Small Particles St=Stokes No.

27. School of Aerospace Engineering MITE Simulation of a Mixing Layer, where the upper stream is laden with medium size particles (Stokes No. = 1). Experiment by Lazaros and Lasheras (1992)

28. School of Aerospace Engineering MITE Conventional LES Scheme Vs. Present 5 Micron Cut-Off Product mass Fraction

29. School of Aerospace Engineering MITE Conventional LES Scheme Vs. Present 5 Micron Cut-Off Temperature

30. School of Aerospace Engineering MITE Conventional LES Results are sensitive to Droplet Cut-Off Size 4 to 5 times expensive than present approach

31. School of Aerospace Engineering MITE Present Approach is less sensitive to Droplet Cut-Off Size

32. School of Aerospace Engineering MITE main Air Main Air Honeycomb Fuel Coflow Air Turbulence Generator Measurement Planes Optical Access Experimental Set Up for LES/LEM Validation

33. School of Aerospace Engineering MITE Comparisons with GA Tech Experiments Measured inflow velocities, droplet distribution and turbulence levels are input into the code

34. School of Aerospace Engineering MITE Comparisons with Ga Tech Experiments

35. School of Aerospace Engineering MITE Combustor Modeling- Conclusions Incompressible and Compressible Two-Phase Reacting Flow Solvers have been developed. Droplet convection, evaporation, turbulent mixing and reaction are all modeled from first principles. Present approach is less expensive than conventional LES, but more accurate. Flow solver has been validated with experiments.

36. School of Aerospace Engineering MITE Research Plans for Next Year Extend the new CFD based reduced order model to 3-D centrifugal configurations. Validate. Study stall and surge control of the Ga Tech centrifugal compressor configuration using CFD, and using the 3-D reduced order model. Perform further validations of the LES/LEM two-phase flow method with Georgia Tech data. Perform two-phase reacting flow simulations for a dump combustor configuration.