LES Investigation of Fuel Effects on Lean Blow off (LBO) for a Realistic Two-Phase Flow Combustor J.W. LABAHN, P. C. M A, L. E SCLAPE, M. I HME S TANFORD U NIVERSITY 2017 SIAM International Conference on Numerical Combustion Orlando, FL April 4 th 2017 Financial support through FAA National Jet Fuel Combustion Program is gratefully acknowledged
Background and Motivation Modern gas turbines operate at lean conditions to increase efficiency and educe emissions Operating at lean conditions make the gas turbines susceptible to blowout as the flame stabilization mechanisms are weakened Numerical investigations are needed to study the combustion behavior close to lean blowout (LBO) for different fuels to determine next generation aviation fuels Investigations of fuel sensitivity of LBO and the ability of CFD to predict LBO trends for different fuels need to be determined
Background and Motivation Processes impacting LBO: Spray formation, droplet breakup Multicomponent droplet evaporation Chemistry
Background and Motivation Previous numerical work for a referee-rig combustor: Produced stable flames with accurate axial and radial droplet velocities for a range of fuels Sauter mean diameter (SMD) was severely under-predicted at all axial locations, L. Esclapez, et al., Combust. Flame
Background and Motivation Objectives: To investigate and quantify the impact of the secondary breakup (SBU) model on the droplet SMD Determine if the modeling of the spray can be simplified Can the spray boundary conditions be defined in such a way that the SBU model is not required
Geometry and Boundary Conditions Referee-rig combustor configuration 1 Operating conditions: T air : 394.0 K, T liq : 322.0 K Pressure: 2 atm Air inflow: 3% P -> 391 g/s 1. Pressurized plenum 2. Injection system 3. Combustion chamber 4. Exhaust plenum LES simulation on full combustion chamber with FPV model to examine fuel effects on lean blow-out Lagrangian particle method for the liquid phase with secondary breakup model S. Stouffer et al., UDRI
Spray Boundary Conditions Numerical Methodology Low-Mach solver VIDA Mesh: patched version of Multi-block grid from UTRC (20M cells) 2 nd /4 th order in space on regular/hex mesh Predicter/corrector scheme in time Poisson solver for pressure 3 rd order RK scheme for dispersed phase Closures WALE subgrid model FPV with presumed PDF for chemistry and turbulence-chemistry interactions Spray Models Reaction chemistry and transport properties Detailed chemistry (Hai Wang) for Cat{A2,C1,C5} Reported liquid transport properties from T. Edwards, Fuel Handbook Combustion model FPV with presumed PDF for turbulence/chemistry interaction Consideration of latent heat of vaporization in spray-combustion Droplet model Deterministic Lagrangian model Stochastic model for secondary droplet breakup
Secondary break up model Stochastic droplet breakup model: Breakup of parent drops into secondary governed by the Fokker Planck (FP) differential equation!" #, %!% = ( )!" #, %!# + 1 2 ( )-!- " #, %!# - with # =./ 0 Steady state distribution given by " 1 #, % + 1 = 1 2 1 + 203 # # 1 ) 2 ) - Closure needed for ) and ) - which are given by ) - = 0.1.DE F2 GH F2 1, ) = 0.1
Stanford: sensitivity to secondary break-up 0 Droplet frequency [-] -1-2 -3-4 0 5 15 20 25 30 Weber number [-] Although the instantaneous number of droplets above critical We is low, most of the droplet with Dp > 60µm experiences Web > 12 following the injection 35
Secondary break-up position Critical Weber Number 5 Nb Count [-] 4 3 2 0 2 4 6 8 Weber Number 12 14 Secondary break-up mainly in the radial swirler flow/irz shear layer Droplet Weber number larger than the critical value (ongoing break-up)
Spray Boundary Conditions Fuel Stable A2 New Spray injection location 0.25 mm downstream of spray nozzle Spray injection point Ring injection with ring radius of r=0.9 mm Spray injection speed 8.6 m/s 8 m/s Drop size distribution Rosin-Rammler, q = 2.25 SMD: 54 microns SMD: 45 microns Full spray angle 90 degrees 80 degrees Secondary breakup model Yes No
Effect of variation in SMD Near blow-out simulation (phi=0.096); no secondary breakup: SMD=45um
Effect of variation in SMD Near blow-out simulation (phi=0.096); no secondary break-up
Impact of Secondary Break-up Importance of initial particle distribution defined at injector (solid line-injector, dashed line particle distribution at -5mm 0.07 0.06 0.05 0.04 0.03 Impact of secondary breakup 35 SMD/No SBU black line 45 SMD/No SBU red line 48 SMD/SBU blue line 0.02 0.01 0 0 20 40 60 80 0 120 140 160 d [um] Large difference between SBU and no SBU SBU particle distribution at -5 mm defined completely by SBU model For No SBU model distribution at -5mm defined completely by boundary condition
Comparison to PDPA measurements 5 mm
Comparison to PDPA measurements mm
Summary and Conclusions LES simulation with and without secondary droplet breakup are performed Secondary droplet breakup and the initial droplet distribution are shown to have large impact on lean blowout (LBO) predictions Current simulations are sensitive to spray boundary conditions and SBU model employed An adhoc method to determine spray boundary conditions to replace SBU is possible but requires a large amount of tuning -> not computationally feasible Additional work required to determine cause of underprediction of SMD in previous simulations Computationally simple and inexpensive method required for spray to allow for comparison between different LES-combustion codes
Thank You!