Turbulent mixing for a jet in crossflow and plans for turbulent combustion simulations - PowerPoint PPT Presentation

Loading...

PPT – Turbulent mixing for a jet in crossflow and plans for turbulent combustion simulations PowerPoint presentation | free to download - id: 7ea7cf-ODI1O



Loading


The Adobe Flash plugin is needed to view this content

Get the plugin now

View by Category
About This Presentation
Title:

Turbulent mixing for a jet in crossflow and plans for turbulent combustion simulations

Description:

Turbulent mixing for a jet in crossflow and plans for turbulent combustion simulations James Glimm – PowerPoint PPT presentation

Number of Views:38
Avg rating:3.0/5.0
Slides: 27
Provided by: YanY1
Category:

less

Write a Comment
User Comments (0)
Transcript and Presenter's Notes

Title: Turbulent mixing for a jet in crossflow and plans for turbulent combustion simulations


1
Turbulent mixing for a jet in crossflow and
plansfor turbulent combustion simulations
  • James Glimm

2
The Team/Collaborators
  • Stony Brook University
  • James Glimm
  • Xiaolin Li
  • Xiangmin Jiao
  • Yan Yu
  • Ryan Kaufman
  • Ying Xu
  • Vinay Mahadeo
  • Hao Zhang
  • Hyunkyung Lim
  • Drew University
  • Srabasti Dutta
  • Los Alamos National Laboratory
  • David H. Sharp
  • John Grove
  • Bradley Plohr
  • Wurigen Bo
  • Baolian Cheng

3
Outline of Presentation
  • Problem specification and dimensional analysis
  • Experimental configuration
  • HyShot II configuration
  • Plans for combustion simulations
  • Fine scale simulations for VV purposes
  • HyShot II simulation plans
  • Preliminary simulation results for mixing
  • Comments on the nature of LES convergence

4
Main Objective
  • Compare to the Stanford code development effort.
    Chemistry to be computed without a model (beyond
    dynamic turbulence model). Hereby we can offer a
    UQ assessment of the accuracy of the Stanford
    code.
  • If the comparison is satisfactory and the two
    codes agree, the UQ analysis of the Stanford code
    (in this aspect) will be complete.
  • If the comparison is unsatisfactory, we will
    attempt to determine which of the differing
    results are to be believed.

5
Problem Specification andDimensional Analysis
  • Experimental configuration
  • Problem dimensions 8.6 x 2 x 2 cm
  • Parameters for crossflow (air)
  • Crossflow Ma 2.4 flow velocity 1800 m/s
  • Crossflow pressure 0.4 Bar
  • Crossflow Temperature 1548K
  • L (air) distance of nozzle downstream 0.067 m
  • Viscosity (air) 5.36e-4 m2/s
  • Re (air) 2.25e5
  • Kolmogorov scale (air) L Re-3/4 6.5 microns
  • Parameters for H2
  • H2 flow M 1 H2 velocity 1205 m/s
  • H2 pressure 20.2 Bar
  • H2 Temperature 300 K
  • Viscosity of H2 0.16e-4 m2/s
  • L (H2) nozzle diameter 2 mm
  • Re (H2) 1.5e5
  • Kolmogorov scale (H2) LRe-3/4 11 microns
  • Flame width (OH, from experiment) 200 microns

6
Problem Specification andDimensional Analysis
  • HyShot II Scramjet configuration
  • Combustion chamber dimensions 29.5 x 0.98 x 7.5
    cm
  • Reduced by symmetry to 29.5 x 0.98 x 0.9375 cm
  • Volume is 0.79 as a fraction of the experimental
    combustion chamber (after symmetry reduction)
  • Crossflow (air) parameters
  • Crossflow Ma 2.4 flow velocity 1720 m/s
  • Crossflow pressure 130 KPa
  • Crossflow Temperature 1300 K
  • Viscosity of air 0.000182 m/s
  • L (air) 5 cm (from inflow plane to injector)
  • Re (air) 4.7e5
  • Kolmogorov scale (air) LRe-3/4 2.8 microns
  • H2 parameters (at injector exit)
  • H2 flow M 1 velocity 1200 m/s
  • H2 pressure 4.6 bar
  • H2 Temperature 300 K
  • Viscosity of H2 2.22e-5 m2/s
  • L (H2) nozzle diameter 2 mm
  • Re (H2) 1.1 e5

7
Problem Specification andDimensional Analysis
  • Simulation Parameters Experimental Configuration
  • Fine grid approximately 60 micron grid
  • Mesh 1500 x 350 x 350 183 M cells
  • If necessary, we can simulate only a fraction of
    the experimental domain
  • If necessary, a few levels of AMR can be used
  • Current simulations 120 microns, about 10 M
    cells
  • HyShot II configuration
  • Resolution problem is similar
  • 3/4 volume after symmetry reduction compared to
    experiment
  • Full (symmetry reduced) domain needed to model
    unstart
  • Resolved chemistry might be feasible
  • Wall heating an important issue

8
Flow and Chemistry Regime
  • Turbulence scale ltlt chemistry scale
  • Broken reaction zone
  • Autoignition flow regime
  • Tc ltlt T
  • Makes flame stable against extinction from
    turbulent fluctuations within flame structure
  • Unusual regime for turbulent combustion
  • Broken reaction zone autoignition distributed
    flame regime
  • Query to Stanford team literature on this flow
    regime?
  • Knudsen and Pitsch Comb and Flame 2009
  • Modification to FlameMaster for this regime?
  • Opportunity to develop validated combustion
    models for this regime, for use in other
    applications
  • Some applications of DOE interest

9
Flow, Simulation and Chemistry Scales
Experimental Regime
  • Turbulence scale ltlt grid scale ltlt chemistry scale
  • Turbulence scale 10 microns
  • ltlt grid scale 60 microns
  • ltlt chemistry scale 200 microns
  • Resolved chemistry, but not resolved turbulence
  • Need for dynamic SGS models for turbulence
  • Transport in chemistry simulations must depend on
    turbulent laminar fluid transport, not on
    laminar transport alone

10
Chemistry Simulation Plans
  • Resolved Chemistry vs. Flamelets
  • Flamelets
  • assumes diffusion flame,
  • Resolved chemistry
  • makes no assumption of flame structure
  • thus resolved chemistry is more suitable for an
    autoignition flame
  • FlameMaster has been or will be extended to
    support autoignition flame structure?
  • Flamelets
  • use FlameMaster,
  • Resolved chemistry
  • uses FlameMaster subroutine for chemical source
    terms
  • Flamelets
  • assumes a quasi equilibrium solution, thus
    suppresses certain transients.
  • (This can be either/both a strength or a
    weakness.)
  • speed and/or memory advantages
  • Flamelets feasible for coarser grids
  • Resolved chemistry
  • allows UQ assessment of flamelet model in
    Scramjet context.
  • Has value for Scramjet UQ analysis even if too
    slow to be feasible for most simulations

11
Simulation Plans Experimental Regime
  • Mixed fluid physics
  • Accurate multifluid viscosity, diffusion
    parameters
  • Diffusion velocity
  • Numerical issues
  • Finer resolution grids
  • No need to track fronts
  • AMR needed?
  • Turbulent inflow needed (nozzle/cross flow)?
  • VV for pure mixing
  • Add chemistry
  • VV for resolved chemistry
  • Comparison to flamelet simulations
  • VV for flamelets

12
Simulation PlansHyShot II Regime
  • Work with autoignition version of FlameMaster
  • Add this capability if necessary
  • Compare to laboratory experimental regime and
    resolved chemistry simulations (VV)
  • Simulate in representative flow regimes defined
    by the large scale MC reduced order model, both
    for failure conditions (unstart) and for
    successful conditions.
  • Provide improved combustion modeling to the MC
    low order model, for the next iteration of an MC
    full system search.
  • Investigate gates which serve to couple system
    components into full system
  • For combustion chamber, primarily fuel nozzle,
    inlet flow and exit nozzle
  • Essential step for relating UQ of components to
    UQ of full system

13
Preliminary Simulation ResultsMixing Only
3D simulation. 67 H2 mass concentration isosurfac
e plot compared to experimental OH-PLIF image
(courtesy of Mirko Gamba). The grid is 120
microns, 2 times coarser than the Intended fine
grid mesh size.
14
Preliminary Simulation ResultsMixing Only
Black dots are the flame front extracted from the
experimental OH-PLIF image.
15
Preliminary Simulation ResultsMixing Only
Velocity divergence plotted at the midline plane.
Bow shock, boundary layer separation, barrel
shock and Mach disk are visible from the plot.
16
Preliminary Simulation ResultsMixing Only
H2 mass fraction contour plotted at the midline
plane
17
Preliminary Simulation ResultsMixing Only
Stream-wise velocity contour plotted at the
midline plane
18
Preliminary Simulation ResultsMixing Only
H2 mass fraction contour plotted at x/d2.4
19
Preliminary Simulation ResultsMixing Only
Stream-wise velocity contour plotted x/d2.4
20
Preliminary Simulation ResultsMixing Only
Mixture fraction plot courtesy of Catherine
Gorle 0 represents Hydrogen,1 represents Air
Mass fraction plot of our simulation 1 represents
Hydrogen, 0 represents Air
21
Preliminary Simulation ResultsMixing Only
Comparison between Smagorinsky model (left) and
dynamic model (right) Mass fraction
plot, using 240 micron grid
22
Preliminary Simulation ResultsMixing Only
Comparison between 240 micron grid and 120 micron
grid With dynamic model, mass fraction plot
23
The Nature of LES Convergence
  • A mathematical theorem (G-Q Chen and JG)
  • If one assumes K41 as a property of LES solutions
    of NS (this is incompressible, but with possible
    passive scalars) then after to passing to
    subsequences
  • Velocity field converges in some Lq space
  • Passive scalars (are proved to) converge only as
    Young measures, ie measure valued distributions

24
Young Measures and PDFs
  • Assume this applies to compressible NS as well
  • Implication is that the LES regime defines a
    convergent PDF, not a convergent point valued
    function, even as a weak solution.
  • If point values are extracted, these are the
    means of the pdfs
  • For application to nonlinear processes
    (chemistry), mean values of pds are not good
    enough, and fluctuations are needed.
  • Thus pdf (Young measure) convergence is an
    important conceptual issue for LES applied to
    combustion

25
Young Measure Convergence
  • How would one define or notice the effects of pdf
    convergence?
  • We take a unit size filter (implicit filter). But
    the values at a point are not meaningful. Rather
    the ensemble of solution values in an extended
    neighborhood of the unit filter define an
    ensemble, which is the pdf for the point or for
    all points in the extended neighborhood
  • At the larger filter level size, the limit looks
    like an ensemble, a pdf, not a point value.

26
Queries for Stanford
  • What is the status/need for autoignition in
    FlameMaster?
  • In the broken flame regime, with turbulence
    inside the flame,
  • what is used for the binary diffusion
    coefficients that drive the effective diffusivity
    of species k into the mixture? Laminar, from
    kinetic theory, or turbulent, from an SGS model?
  • Or are the SGS diffusion terms just a Fickean add
    on to the multicomponent diffusion?
  • In this case they should be dominant for most
    grids, and so the multicomponent theory of
    diffusion might not be needed?
  • References for the broken flame-autoignition
    regime?
About PowerShow.com