Large Eddy Simulation of Piloted Turbulent Premixed Flame Veeraraghava Raju Hasti, Robert P Lucht and Jay P Gore Maurice J. Zucrow Laboratories School of Mechanical Engineering Purdue University West Lafayette, IN Gaurav Kumar, Shuaishuai Liu Convergent Science New Braunfels, TX September 27, 2017 Detroit, MI 1
Acknowledgement This work was supported by the U.S. Department of Energy (DOE), National Energy Technology Laboratory (NETL), University Turbine Systems Research (UTSR) Program with DOE award number DE- FE0011822. Authors acknowledge the experimental data and support from Dong Han, Jupyoung Kim, Aman Satija and Jo. 2
1. Motivation Overview 2. Objectives 3. Experimental methods 4. Computational methods 5. Results 6. Summary and conclusions 7. Future work 3
Motivation To understand the effect of CO 2 addition on turbulent premixed flame structure Turbulent premixed flames are widely utilized in combustion applications Accurate prediction of turbulent flame characteristics play a crucial role in the development of low emission combustors EGR for Emission Reduction in GTCC Power Generation 4
Objectives Experimental characterization using CARS, OH PLIF and PIV CO 2 dilution Inlet Turbulence Intensity Pressure Reynolds Number Perform LES simulations to validate the Turbulent Combustion Models Understand CO 2 Dilution on Premixed Flame Structure 5
Piloted Axisymmetric Reactor Assisted Turbulent Burner (PARAT) D=18mm No co-flow High Pressure < 20 bar High Temperature < 800 K Pilot Stabilized Burner to Study Premixed Turbulent Flames 6
Coherent Anti-Stokes Raman Scattering (CARS) Experimental Methods Particle Image Velocimetry (PIV) Heated Al 2 O 3 ( 0.5 µm ) Sampling frequency: 10 Hz Number of image pairs: 1000 Sampling frequency: 10 Hz Pulse Energy: 20 mj (532 nm), 20 mj (560 nm), and 30 mj (607 nm) Spatial resolution: 1.5mm x 150 μm Number of shots: 1000 Planar Laser Induced Fluorescence (PLIF) Detecting species: OH radical Sampling frequency: 9 khz Number of images: 1000 7
CFD Simulations Cold flow simulations to study mixing and jet development RANS RNG k-ε Model Reacting Simulations Premixing tube excluded Domain : 36D x 64D x 36 D, D= 18 mm SAGE detailed chemistry solver with DRM19 mechanism Turbulence LES with 1 equation Dynamic structure model Average velocity at the nozzle exit 8.7 m/s Base grid : 10 mm, 8 mm, 6 mm AMR : Velocity 4 levels with sub-grid criterion 0.1 Temperature 4 levels with sub-grid criterion 2.5 Model Validation 8
CFD Domain and BCs Non Reacting Premixing section Flame Region (Open to Atmosphere) Flow Direction Air Inlet Velocity = 4 m/s Outlet Fuel injection All 4 sides are Open boundaries Bluff body Air Inlet Turbulent plate 1 Turbulent plate 2 Understand the Flow Inside Premixing Tube 9
y/d Cold flow Results with Steady RANS RNG k-ԑ model Turbulent Plate near Nozzle Causes Non-Uniform Velocity Profile at the Exit With intensity ~ 14-15% 10
Cold flow Results with RANS based RNG k-ԑ model Good Agreement with Experimental Data 11
CFD Domain and BCs - Reacting Outflow BC Nozzle Exit: Specified mean velocity profile. Turbulent Fluctuations: Random Fourier Approach Davidson et.al, IJHF, 27, 2006 Pilot H 2 CH 4 +Air (Premixed) Pilot H 2 Burner surface Premixing Tube Excluded to Reduce Computational Time 12
Instantaneous Temperature [K] Flame 1 No CO 2 Temperature distribution for Flame 1 AMR Region: 5 < y/d < 6 Mean Temperature [K] Mesh Size : 10M with AMR 13 Mesh Optimization With AMR
Velocity Profile for Flame 1 (No CO 2 ) Mean Axial Velocity [m/s] At y/d =0.2 y/d =0.2 At y/d =0.2 z=0 plane Axisymmetric Turbulent Round Jet 14
Centerline Centerline Mean Temperature Profile for Flame 1 Centerline Profile, x/d=0 Mean Temperature [K] MFL URL Potential Core and Flame Shape Predicted Very Well 15
Centerline Centerline RMS Temperature Profile for Flame 1 RMS Temperature [K] Centerline Profile, x/d=0 Vibrational CARS is not accurate at low temperature CARS centerline data also include more data in the radial direction Efforts are in progress to reduce CARS Probe volume 16
Mesh on z=0 Plane at t=2.68 s Centerline Profile, x/d=0 CARS Spatial Resolution : 150 µm x 1.5 mm CFD : Minimum cell size after AMR : 0.5 mm x 0.5 mm Temperature fluctuations are larger in the radial direction than along the center line after y/d =6 CFD includes more data in the axial direction but less in the radial direction 17
Radial Temperature Profile for Flame 1 At y/d =5 y/d =5 y/d =1.94 At y/d =1.94 Good match with Experimental Data 18
Conclusions and Future Work Cold flow RANS simulations to understand the mixing inside the premixing tube Reacting flow simulations with LES and SAGE detailed chemistry solver to model the piloted turbulent premixed flame Good agreement for mean and rms values of velocity and temperature profile Paper accepted for AIAA SciTech 2018 Reasonable Agreement for Flame 1; Efforts will continue for Flame 2 and 3 19
Thank You 20