LES Lecture

download LES Lecture

of 28

Transcript of LES Lecture

  • 8/10/2019 LES Lecture

    1/28

    School of somethingFACULTY OF OTHER

    CFD CentreEnergy Technology Innovation Initiative

    Large Eddy Simulation:

    an introduction

    University of Leeds, 4-5-2012

  • 8/10/2019 LES Lecture

    2/28

    Turbulent Flows

    Eddies (structures) with wide range of length and timescales.

    3-D

    Highly unsteady

    A. Pranzitelli (2012). Large Eddy Simulation: an introduction

  • 8/10/2019 LES Lecture

    3/28

    CFD approaches

    DNS

    LES

    RANS C o

    m p u t a t i o n a l c o s t

    A. Pranzitelli (2012). Large Eddy Simulation: an introduction

  • 8/10/2019 LES Lecture

    4/28

    RANS, URANS, LES and DNS

    URANS

    LES

    RANS

    DNS

    Direct calculation of the statisticalaverage of the solution

    Direct calculation of only certain low-frequency modes in time and theaverage field

    Direct calculation of only the

    low-frequency modes in space

    Simulation of all the scales

    A. Pranzitelli (2012). Large Eddy Simulation: an introduction

  • 8/10/2019 LES Lecture

    5/28

    The rationale for the LES

    Momentum, mass, energy and other passive scalars aretransported mostly by large eddies.

    Large eddies are anisotropic, subject to history effects andproblem-dependent (flow configuration, boundaryconditions, etc.)

    Small eddies are less dependent on the geometry, tend tobe more isotropic, and are consequently more universal.

    The chance of finding a universal turbulence model is much

    higher for small eddies.

    A. Pranzitelli (2012). Large Eddy Simulation: an introduction

  • 8/10/2019 LES Lecture

    6/28

    Why LES?

    Bluff body aerodynamics Manoeuvrability of vehicles

    Safety and comfort (wake of building and ships)

    Unsteady loading on structures

    Aerodynamically generated noise Landing gear

    Jet engine, propeller

    Multi-component airfoils (flaps)

    Fluid-structure interaction buffeting

    Combustion instability A. Pranzitelli (2012). Large Eddy Simulation: an introduction

  • 8/10/2019 LES Lecture

    7/28

    Kolmogorovs theory

    The energy cascade

    Large eddies are unstable and break up, transferring energy to thesmaller eddies, energy continues to be passed to smaller scaled eddies

    This process continues until Re( l ) u(l)l / is sufficiently small that the

    eddy motion is stable, and molecular viscosity is effective in dissipating thekinetic energy

    At these small scales, the kinetic energy of turbulence is converted intoheat

    No dissipation in the energy cascade

    determined by the first process, proportional to u03

    /l 0 independent of

    (at high Re)

    A. Pranzitelli (2012). Large Eddy Simulation: an introduction

  • 8/10/2019 LES Lecture

    8/28

    Kolmogorovs theory

    Kolmogorovs hypotheses: At sufficiently high Reynolds numbers, the small-scale turbulent motions (l l EI) andthe small scale isotropic eddies ( l

  • 8/10/2019 LES Lecture

    9/28

    Kolmogorovs theory

    Inertial subrangeDissipation range

    Energycontaining

    range

    Universal equilibrium range

    l DI l EI l 0 L

    Kolmogorovlength scale

    Integrallength scale

    41

    3

    1Re

    A. Pranzitelli (2012). Large Eddy Simulation: an introduction

  • 8/10/2019 LES Lecture

    10/28

    Filtering

    Transport equations are filtered such that only larger eddiesneed be resolved.

    N-Sequation

    Filtered N-S

    equation

    Random variable filtered usinga space-filter function G

    Need to be modelledSub-grid scale models

    A. Pranzitelli (2012). Large Eddy Simulation: an introduction

  • 8/10/2019 LES Lecture

    11/28

    Filtering

    A. Pranzitelli (2012). Large Eddy Simulation: an introduction

  • 8/10/2019 LES Lecture

    12/28

    Sub-grid Scale models

    Smagorinsky

    In Fluent: Smagorinsky-Lilly

    Dynamic models (e.g. Germano et al ., 1991): variable

    In Fluent: Dynamic Smagorinsky-Lilly model, with clipped at 0 and0.23 by default

    Shortcomings: constant. No value universally applicable to different types of flow Difficulty with transitional (laminar) flows Damping function needed in near-wall region

    A. Pranzitelli (2012). Large Eddy Simulation: an introduction

  • 8/10/2019 LES Lecture

    13/28

    Sub-grid Scale models

    Wall-Adapting Local Eddy-Viscosity (WALE) Model

    The WALE SGS model adapts to local near-wall flow structure

    In Fluent

    A. Pranzitelli (2012). Large Eddy Simulation: an introduction

  • 8/10/2019 LES Lecture

    14/28

    Quality of your LES

    The resolved kinetic energy depends on the filter width

    Most of commercial software useimplicit filtering

    The filter width isdetermined by the grid

    resolution

    The cell size determines the amount of energy that isbeing resolved

    80% of turbulent kinetic energy should be resolved for anaccurate LES

    A. Pranzitelli (2012). Large Eddy Simulation: an introduction

  • 8/10/2019 LES Lecture

    15/28

    Quality of your LES

    The grid plays a key-role in the LES

    Need for knowing the turbulence scales of our own case

    Need for knowing the distribution of the turbulent kineticenergy on these scales

    Simulation of enough energy to catch the importantfeatures of our flow field

    A. Pranzitelli (2012). Large Eddy Simulation: an introduction

  • 8/10/2019 LES Lecture

    16/28

    What is usually done

    Use of the same grid used for RANS Incorrect, grid too coarse

    Not related to the vortex distribution

    Grid independence study, as for the RANS simulations Expensive

    No local refinement

    Not possible in case of implicit filtering

    A. Pranzitelli (2012). Large Eddy Simulation: an introduction

  • 8/10/2019 LES Lecture

    17/28

    Quality of your LES

    To assess the quality of your LES:

    A priori study

    A posteriori study

    Determination of the mesh resolution

    Accuracy is difficult to measure in DNS and LES because of the nature itself of the turbulent flows

    A. Pranzitelli (2012). Large Eddy Simulation: an introduction

  • 8/10/2019 LES Lecture

    18/28

    Cell-size estimation

    A first approach, from the integral length scale estimate from correlation

    from a RANS simulation (e.g. with the standard k- ) on a coarse gridand l 0 =(k 1.5 )/

    cell size l 0 /6, 80% turb. kinetic energy resolved (l 0 /6= l EI demarcation line between Energy and Inertial ranges)

    Issues: k is grid dependent / solution dependent (?)

    rough estimation of the turbulent kinetic energy to be resolved

    A. Pranzitelli (2012). Large Eddy Simulation: an introduction

  • 8/10/2019 LES Lecture

    19/28

    Example:E.ON Combustion Test Facility

    Located in Ratcliffe-on-Soar, UK

    1 MW th

    Single, wall-fired, low NO xburner

    Variety of fuels (coal, gas,biomass, )

    Time-temperature scaledto 500 MW facility

    A. Pranzitelli (2012). Large Eddy Simulation: an introduction

  • 8/10/2019 LES Lecture

    20/28

    CFD simulations: approach

    commercial code: ANSYS FLUENT V12.1

    Preparation of thegeometry

    Mesh generation

    Steady RANS simulation

    LES

    Hardware:

    RANS: 1 quad-core intel Q9550, 2.83GHzprocessor, 8GB RAM LES: Leeds ARC1 HPC cluster, 50 intel NehalemX5560, 2.8GHz CPU cores, 2GB RAM/core

    CPU time:

    RANS: 25 days LES: 15 days/second of simulation

    A. Pranzitelli (2012). Large Eddy Simulation: an introduction

  • 8/10/2019 LES Lecture

    21/28

    Example:E.ON Combustion Test Facility (CTF)

    Mesh:

    3M cells Unstructured, but regular cell

    distribution Mostly hexahedral cells,small number of polyhedralcells

    High quality, low cell aspectratio and squish, notetrahedral cells

    5.1

    0k

    L 12

    0 Ll cell Cell sizing: A. Pranzitelli (2012). Large Eddy Simulation: an introduction

  • 8/10/2019 LES Lecture

    22/28

    E.ON CTF:results

    RANS LES (instantaneous values)

    Predicted Temperature distribution (K)

    Air-coal combustionRANS vs LES

    A. Pranzitelli (2012). Large Eddy Simulation: an introduction

  • 8/10/2019 LES Lecture

    23/28

    E.ON CTF:results

    vertical mid plane,distance from the burner centre line

    Gas temperature

    A. Pranzitelli (2012). Large Eddy Simulation: an introduction

  • 8/10/2019 LES Lecture

    24/28

    Oxy-coal combustionRANS vs LES

    RANS LES (instantaneousvalues)

    Temperature distribution (K)

    A. Pranzitelli (2012). Large Eddy Simulation: an introduction

  • 8/10/2019 LES Lecture

    25/28

    Oxy-coal combustionflame

    RANS LES (inst.) LES (mean) A. Pranzitelli (2012). Large Eddy Simulation: an introduction

  • 8/10/2019 LES Lecture

    26/28

    Oxy-coal combustionLES

    A. Pranzitelli (2012). Large Eddy Simulation: an introduction

  • 8/10/2019 LES Lecture

    27/28

    References

    Sagaut, P. Large Eddy Simulation for Incompressible Flows. Springer, 2006.

    Ferziger J.H., Peri, M. Computational methods for Fluid Dynamics. Springer, 2002

    ANSYS Fluent Users Manual

    Pope, Stephen B. Turbulent Flows. Cambridge University Press, 2000

    Andr Bakker, lecture on Applied Computational Fluid Dynamics, Fluent Inc. 2002

    Best practice advice for Large Eddy Simulation, ANSYS Celik, I., Klein, M., Janica, J. Assessment Measures for Engineering LES Applications.

    Journal of Fluids Engineering, Vol. 131 (2009)

    M. Gharebaghi, R.M.A. Irons, L. Ma, M. Pourkashanian, A. Pranzitelli. Large eddysimulation of oxy-coal combustion in an industrial combustion test facility. Int. J.Greenhouse Gas Control (2011), doi:10.1016/j.ijggc.2011.05.030

    D. Caridi, D. Couling, M. Gharebaghi, S.R. Gubba, R.M.A. Irons, L. Ma, M. Pourkashanian, A. Pranzitelli and A. Williams. Comparison of RANS and LES Turbulence Models for Predicting Air-coal and Oxy-coal Combustion Behaviours. 2nd Oxyfuel CombustionConference, 12th 16th September 2011, Yeppoon ,QLD, Australia.http://www.ieaghg.org/index.php?/20110526255/occ2-programme-ts-5.html

    A. Pranzitelli (2012). Large Eddy Simulation: an introduction

    http://www.ieaghg.org/index.php?/20110526255/occ2-programme-ts-5.htmlhttp://www.ieaghg.org/index.php?/20110526255/occ2-programme-ts-5.html
  • 8/10/2019 LES Lecture

    28/28

    Any questions?

    Contact details: [email protected] rd floor, Energy Building

    A. Pranzitelli (2012). Large Eddy Simulation: an introduction

    mailto:[email protected]:[email protected]