SlideShare a Scribd company logo
1 of 46
Mesoscopic numerical methods for
              reactive flows: Lattice Boltzmann
                     method and beyond

                                      Candidate: Antonio F. Di Rienzo

                            Advisors: Prof. P. Asinari, Dr. E. Chiavazzo

       External Collaborators: Dr. J. Mantzaras, Dr. N. Prasianakis (Paul
                               Scherrer Institute)

                                Energy Department, Politecnico di Torino, Torino, Italy


                                   PhD Defence, Torino 20 April, 2012
Antonio F. Di Rienzo (Politecnico di Torino)          PhD Defence                   Torino, April 20th 2012   1 / 46
Outline of The Talk

   1    Summary of the talk

   2    Introduction

   3    The Lattice Boltzmann Method (LBM)

   4    Link-wise Artificial Compressibility Method (LW-ACM)

   5    Radiative Transfer Equation

   6    Combustion

   7    Conclusions & Acknowledgements



Antonio F. Di Rienzo (Politecnico di Torino)   PhD Defence   Torino, April 20th 2012   2 / 46
Summary of the talk

   Outline Compass

   1    Summary of the talk

   2    Introduction

   3    The Lattice Boltzmann Method (LBM)

   4    Link-wise Artificial Compressibility Method (LW-ACM)

   5    Radiative Transfer Equation

   6    Combustion

   7    Conclusions & Acknowledgements



Antonio F. Di Rienzo (Politecnico di Torino)                 PhD Defence   Torino, April 20th 2012   3 / 46
Summary of the talk

   Topics

           Link-wise Artificial Compressibility Method 1 : CFD by kinetic
           mock-up models.

           The Radiative transfer equation (RTE) is solved by means of
           lattice Boltzmann method (LBM) formalism 2 : intensity is updated
           according to lattice velocities.

           A consistent lattice Boltzmann model for reactive flows 3 is
           presented. It addresses the lack of accurate combustion models in
           LBM literature (in collaboration with the Paul Scherrer Institute,
           Switzerland).
       1
         Asinari P., Ohwada T., Chiavazzo E., Di Rienzo A.F., Link-wise Artificial Compressiniblity Method, Journ.
  Comp. Phys., 2012 (preprint), (Impact Factor: 2.345)
       2
         Di Rienzo A. F., Asinari P., Borchiellini R., Mishra S. C., Improved angular discretization and error analysis of
  the lattice Boltzmann method for solving radiative heat transfer in a participating medium, Int. Journ. Num. Meth.
  Heat Fluid Flow, 21 (5), 640-662, 2011, (Impact Factor: 0.53)
       3
         Di Rienzo A. F., Asinari P., Chiavazzo E., Prasianakis N., Mantzaras J., A Lattice Boltzmann model for reactive
  flows simulation, EuroPhys. Lett., 2012 (accepted), (Impact Factor: 2.752)
Antonio F. Di Rienzo (Politecnico di Torino)                 PhD Defence                              Torino, April 20th 2012   4 / 46
Introduction

   Outline Compass

   1    Summary of the talk

   2    Introduction

   3    The Lattice Boltzmann Method (LBM)

   4    Link-wise Artificial Compressibility Method (LW-ACM)

   5    Radiative Transfer Equation

   6    Combustion

   7    Conclusions & Acknowledgements



Antonio F. Di Rienzo (Politecnico di Torino)          PhD Defence   Torino, April 20th 2012   5 / 46
Introduction

   Simulating Reactive Flows

           Reactive flows can be found in several energy systems: internal
           combustion engines, industrial burners, gas turbine combustors...
           Numerical modeling of reactive flows is a challenging task per se:
           simultaneous processes must be taken into account (turbulent
           mixing, multi-phase fluid-dynamics, radiative heat transfer).
           Furthermore, in some applications, the numerical model must be
           able to deal with different length scales.
           The demand of energy systems with high energy efficiency and
           low environmental impact requires the development of new energy
           systems to be used in industrial and domestic applications.
           Porous media combustion (PMC) proved to be a feasible option.
           Porous media (PM) burners offer some advantages compared to
           classical burners. The internal process of heat recovery makes
           the combustion process more efficient, which comes with a lower
           emission of polluntants, such as NOx and CO.
Antonio F. Di Rienzo (Politecnico di Torino)          PhD Defence   Torino, April 20th 2012   6 / 46
Introduction

   Porous Media Burner




  Abdul Mujeebu at al., Application of porous media combustion technology - A review,
  Applied Energy, 2009
Antonio F. Di Rienzo (Politecnico di Torino)          PhD Defence   Torino, April 20th 2012   7 / 46
Introduction

   Fluid Flow in Porous Media




           If studies are carried out at the level of the pore-scale, detailed
           local informations of the flow can be obtained.
           In Lattice Boltzmann Method (LBM) the link between the
           pore-scale and the macro-scale is readily available.
           LBM is often referred to as mesoscopic approach.
Antonio F. Di Rienzo (Politecnico di Torino)          PhD Defence   Torino, April 20th 2012   8 / 46
The Lattice Boltzmann Method (LBM)

   Outline Compass

   1    Summary of the talk

   2    Introduction

   3    The Lattice Boltzmann Method (LBM)

   4    Link-wise Artificial Compressibility Method (LW-ACM)

   5    Radiative Transfer Equation

   6    Combustion

   7    Conclusions & Acknowledgements



Antonio F. Di Rienzo (Politecnico di Torino)              PhD Defence   Torino, April 20th 2012   9 / 46
The Lattice Boltzmann Method (LBM)

   What is Lattice Boltzmann Method?

           "The lattice Boltzmann method (LBM) is used for the numerical
           simulation of physical phenomena and serves as an alternative to
           classical solvers of partial differential equation (PDEs)"
           [www.lbmethod.org]. The main unknown is the discrete
           distribution function, from which all relevant macroscopic
           quantities (satisfying some target PDEs) can be derived.
           The operative formula consists of the (a) relaxation process and
           the (b) advection process:

                  fi (x + vi ∆t, t + ∆t) − fi (x, t) = ω [fieq (x, t) − fi (x, t)].
                          ˆ                                                                       (1)

           The updating of fi is link-wise, in the sense that only the
           informations along the directions identified by the lattice velocity vi
                                                                               ˆ
           are required.


Antonio F. Di Rienzo (Politecnico di Torino)              PhD Defence   Torino, April 20th 2012   10 / 46
The Lattice Boltzmann Method (LBM)

   Dealing with Complex Boundaries




Antonio F. Di Rienzo (Politecnico di Torino)              PhD Defence   Torino, April 20th 2012   11 / 46
The Lattice Boltzmann Method (LBM)

   State of Art




Antonio F. Di Rienzo (Politecnico di Torino)              PhD Defence   Torino, April 20th 2012   12 / 46
Link-wise Artificial Compressibility Method (LW-ACM)

   Outline Compass

   1    Summary of the talk

   2    Introduction

   3    The Lattice Boltzmann Method (LBM)

   4    Link-wise Artificial Compressibility Method (LW-ACM)

   5    Radiative Transfer Equation

   6    Combustion

   7    Conclusions & Acknowledgements



Antonio F. Di Rienzo (Politecnico di Torino)              PhD Defence   Torino, April 20th 2012   13 / 46
Link-wise Artificial Compressibility Method (LW-ACM)

   Limitation of Lattice Boltzmann Method
           LBM works with larger set of unknowns, including the higher-order
           moments beyond hydrodynamics:
                                                     ntimes         mtimes
                               Πxx···x yy···y (xx · · · x, yy · · · y) =                         ˆn ˆm
                                                                                                 vi,α vi,β fi .
                                                                                             i

           They are unessential as long as the continuum limit is the main
           concern and can lead to numerical instabilities 4 .
           Ohwada & Asinari 5,6 revived the Artificial Compressibility Method
           (ACM) as an high-order accurate numerical method for the
           incompressible Navier-Stokes equations (INSE).
           The asymptotic analysis of the LBM updating rule delivers the
           same governing equations of ACM.
      4
        Dellar P., Nonhydrodynamic modes and a priori construction of shallow water lattice Boltzmann equations,
  PRE, 2002, 65:036309
      5
        Ohwada T., Asinari P., Artificial compressibility method revisited: asymptotic numerical method for
  incompressible Navier-Stokes equations, JCP, 2010, 229:1698-1723
      6
        Ohwada T., Asinari P., Yabusaki D., Artificial compressibility method and lattice Boltzmann method: similarities
  and differences, Comp. Math. Appl., 2011, 61:3461-3474
Antonio F. Di Rienzo (Politecnico di Torino)                PhD Defence                             Torino, April 20th 2012   14 / 46
Link-wise Artificial Compressibility Method (LW-ACM)

   ACM vs LBM


           Unlike LBM, ACM deals with macroscopic variables only. It offers
           the opportunity of exploiting all existing Finite-Difference (FD)
           technologies.

           ACM relies on standard meshing techniques for dealing with
           complex boundaries. Unstructured body-fitted meshes are used in
           order to adapt the computational grid to the real object. This
           requires advanced algorithms, which imply an additional
           computational overhead.

           A possible improvements would be to have the LBM updating rule
           of the distribution function depend only on the local equilibrium,
           but retain the link-wise formulation.



Antonio F. Di Rienzo (Politecnico di Torino)              PhD Defence   Torino, April 20th 2012   15 / 46
Link-wise Artificial Compressibility Method (LW-ACM)

   Link-wise Artificial Compressibility Method (LW-ACM)
           In the attempt of making ACM more similar to LBM, the following
           formula is proposed 7 :
                                               (e)
               fi (x, t + ∆t) = fi                   (x − vi ∆t, t)
                                                          ˆ
                                               ω−1                (e,o)                (e,o)
                                                                                                                          (2)
                                     +2                      fi           (x, t) − fi          (x − vi ∆t, t) ,
                                                                                                    ˆ
                                                ω
           Similarly to LBM, the updating rule of Eq.(2) is link-wise.
             (e)             (e,o)
           fi and fi     are local functions of ρ = i fi and ρu = i vi fi ,
                                                                     ˆ
           but they do not depend on the higher-order moments.
           Eq. (2) recovers the incompressible Navier-Stokes equations with
           a kinematic viscosity defined as:
                                                              1       1   1
                                                       ν=               −          .
                                                              3       ω 2
      7
        Asinari P., Ohwada T., Chiavazzo E., Di Rienzo A.F., Link-wise Artificial Compressibility Method, Journ. Comp.
  Phys., 2012 (preprint)
Antonio F. Di Rienzo (Politecnico di Torino)               PhD Defence                          Torino, April 20th 2012   16 / 46
Link-wise Artificial Compressibility Method (LW-ACM)

   Simple Boundaries




                                                                            L2 [¯ ]
                                                                                u           L2 [¯ ]
                                                                                                u
           ∆x                 M a ∝ ∆t/∆x                 ν∝      Re−1      Test 1          Test 2
        1 × 10−1                3 × 10−2                  3 × 10−2       1.74 × 10−3     4.59 × 10−4
        5 × 10−2               1.5 × 10−2                 3 × 10−2       4.49 × 10−4     1.21 × 10−4
       2.5 × 10−2              7.5 × 10−2                 3 × 10−2       1.20 × 10−4     3.11 × 10−5


Antonio F. Di Rienzo (Politecnico di Torino)              PhD Defence             Torino, April 20th 2012   17 / 46
Link-wise Artificial Compressibility Method (LW-ACM)

   Complex Boundaries




                        ∆x                     M a ∝ ∆t/∆x          ν ∝ Re−1      L2 [¯ ]
                                                                                      u
                     5 × 10−2                    3 × 10−1           4 × 10−2   1.84 × 10−3
                    1.5 × 10−2                  1.5 × 10−1          4 × 10−2   3.83 × 10−4
                   1.25 × 10−2                  7.5 × 10−1          4 × 10−2   1.11 × 10−4

Antonio F. Di Rienzo (Politecnico di Torino)              PhD Defence            Torino, April 20th 2012   18 / 46
Link-wise Artificial Compressibility Method (LW-ACM)

   2D Lid Driven Cavity: Streamlines




             Comparison between LW-ACM and BGK at Re = 5000.


Antonio F. Di Rienzo (Politecnico di Torino)              PhD Defence   Torino, April 20th 2012   19 / 46
Link-wise Artificial Compressibility Method (LW-ACM)

   2D Lid Driven Cavity: Pressure Contours




             Comparison between LW-ACM and BGK at Re = 5000.


Antonio F. Di Rienzo (Politecnico di Torino)              PhD Defence   Torino, April 20th 2012   20 / 46
Link-wise Artificial Compressibility Method (LW-ACM)

   LW-ACM vs other INSE solvers

   2D lid driven cavity flow at Re = 5000: comparison between the LW-
   ACM and alternative solvers for INSE from literature.
    Scheme                    Grid                   (xp , yp )              (xlr , ylr )      Energy         Enstrophy
    Present                 128 × 128          (0.51652, 0.53754)       (0.81081, 0.079079)   0.039845          29.247
  ACM, 2010                 128 × 128          (0.52052, 0.53954)       (0.82883, 0.071071)   0.027430          41.249
Opt. ACM, 2010              128 × 128          (0.51652, 0.53854)       (0.80981, 0.072072)   0.038371          37.704
  BGK-LBM                   128 × 128               unstable                 unstable         unstable         unstable
  MRT-LBM                   128 × 128          (0.51652, 0.53554)       (0.80881, 0.075075)   0.043600          37.404
Bruneau, 2006               128 × 128          (0.51562, 0.53906)       (0.80469, 0.070313)   0.043566          30.861
    Present                 256 × 256          (0.51552, 0.53554)       (0.80581, 0.074074)   0.044391          34.821
  ACM, 2010                 256 × 256          (0.51652, 0.53654)       (0.80881, 0.072072)   0.040896          43.198
Opt. ACM, 2010              256 × 256          (0.51451, 0.53654)       (0.80380, 0.072072)   0.048114          42.290
  BGK-LBM                   250 × 250          (0.51752, 0.54054)       (0.80781, 0.074074)   0.041614          40.455
  MRT-LBM                   256 × 256          (0.51552, 0.53554)       (0.80681, 0.074074)   0.045222          40.833
Bruneau, 2006               256 × 256          (0.51562, 0.53516)       (0.80859, 0.074219)   0.046204          34.368
Bruneau, 2006              2048 × 2048         (0.51465, 0.53516)       (0.80566, 0.073242)   0.047290          40.261



Antonio F. Di Rienzo (Politecnico di Torino)              PhD Defence                    Torino, April 20th 2012   21 / 46
Radiative Transfer Equation

   Outline Compass

   1    Summary of the talk

   2    Introduction

   3    The Lattice Boltzmann Method (LBM)

   4    Link-wise Artificial Compressibility Method (LW-ACM)

   5    Radiative Transfer Equation

   6    Combustion

   7    Conclusions & Acknowledgements



Antonio F. Di Rienzo (Politecnico di Torino)                PhD Defence   Torino, April 20th 2012   22 / 46
Radiative Transfer Equation

   Radiative Lattice Boltzmann Model

           Combustion generally incorporates also radiation process, which
           plays a significant role if high temperature and a participating
           medium are involved. Treatment of thermal radiation is of key
           importance to develop mathematical model of combustion system.

           Due to its kinetic nature, the Radiative Transfer Equation (RTE)
           can be formulated according to the LBM formalism: the intensity is
           already a particle (photon) distribution function.

           Recently a LBM model 8 has been proposed to solve RTE in a
           participating medium. However, further improvements are
           required: the polar angle is not discretized, spoiling the accuracy
           of the method with respect to standard Finite Volume Method
           (FVM).

      8
        Asinari P., Mishra S.C., Borchiellini R., A lattice Boltzmann formulation to the analysis of radiative heat transfer
  problems in a participating medium, NHT B, 2010, 57:1-21
Antonio F. Di Rienzo (Politecnico di Torino)                  PhD Defence                              Torino, April 20th 2012   23 / 46
Radiative Transfer Equation

   Radiative Transfer Equation
           For an absorbing, emitting and scattering participating medium
           and under the assumption of isotropic scattering, radiative LBM
           reads:
           Iij (x + v∆t, t + ∆t) = Iij (x, t) + Vi,j (κa + σs ) [Ib (x, t) − Iij (x, t)].
                                                                                     (3)
           The link-wise formulation makes unnecessary to march from each
           single corner (e.g. Finite Volumes).
           The data structures are the same as the fluid flow.




Antonio F. Di Rienzo (Politecnico di Torino)                PhD Defence   Torino, April 20th 2012   24 / 46
Radiative Transfer Equation

   Azimuthal Angle Discretization




           The azimuthal angle δ is discretized by introducing a finite number
           of discrete velocities vi = (vx,i , vy,i ) lying on the lattice.
           Even if not discretizing the polar angle saves a lot of computational
           time, this approximation spoils the accuracy of the method.
Antonio F. Di Rienzo (Politecnico di Torino)                PhD Defence   Torino, April 20th 2012   25 / 46
Radiative Transfer Equation

   Azimuthal Angle Discretization




                                                                          Dimensionless       heat
                                                                          flux as function for
                                                                          different extinction co-
                                                                          efficients β = κa + σs .
                                                                          The polar angle is not
                                                                          discretized (Asinari P.
                                                                          et al., NHT B, 2010).

                                                                          The error between
                                                                          LBM and FVM is 2%.




Antonio F. Di Rienzo (Politecnico di Torino)                PhD Defence         Torino, April 20th 2012   26 / 46
Radiative Transfer Equation

   Polar Angle Discretization
           An additional velocity vz,j is introduced along the z-axis (off the
           lattice) in order to discretize the polar angle 9 :
                                                 π
                                    vz,j = tan     − γj vi                     (4)
                                                 2
           The projection on the lattice of the total velocity Vij must overlap
           vi .




       9
         Di Rienzo A.F., Asinari P., Borchiellini R., Mishra S.C., Improved angular discretization and error analysis of
  the lattice Boltzmann method for solving radiative heat transfer in a participating medium, Int. Jour. Num. Meth. Heat
  Fluid Flow, 2011, 21:640-662
Antonio F. Di Rienzo (Politecnico di Torino)                PhD Defence                            Torino, April 20th 2012   27 / 46
Radiative Transfer Equation

   2D Square Enclosure


                                                                          Dimensionless heat flux for
                                                                          β = 2 and β = 5. Radiative
                                                                          equilibrium condition is as-
                                                                          sumed, i.e. · qrad = 0.

                                                                          The error between LBM
                                                                          and FVM is reduced to 0.5%.




                               β = 2.0                                           β = 5.0
             Nx       Nδ         Nγ               L2 [Ψ]            Nx      Nδ     Nγ          L2 [Ψ]
             40       8           4            7.31 × 10−2          100      8      4       5.54 × 10−2
             80       16          8            3.16 × 10−2          200     16      8       2.01 × 10−2
             160      32         16            2.00 × 10−2          400     32     16       1.30 × 10−2
Antonio F. Di Rienzo (Politecnico di Torino)                PhD Defence                Torino, April 20th 2012   28 / 46
Combustion

   Outline Compass

   1    Summary of the talk

   2    Introduction

   3    The Lattice Boltzmann Method (LBM)

   4    Link-wise Artificial Compressibility Method (LW-ACM)

   5    Radiative Transfer Equation

   6    Combustion

   7    Conclusions & Acknowledgements



Antonio F. Di Rienzo (Politecnico di Torino)         PhD Defence   Torino, April 20th 2012   29 / 46
Combustion

   Simulating Reactive Flows: What’s Required?

           Despite the last decade has witnessed a significant improvement
           in describing various problems (flow in porous media, thermal
           flows, etc.), application of LBM to combustion is still limited.
           "In principle, once lattice Boltzmann models can properly account
           for large temperature variation, extension to reactive flows
           essentially involves adding appropriate source terms..."[C. E.
           Frouzakis, Fluid Mechanics and Its Applications, 2011]
           So far, no reactive LBM models has been proposed that
           satisfactorily fulfill this requirement.
           The LBM model is requested to accurately recover the Navier-
           Stokes-Fourier equations, coupled to a transport equation for each
           chemical species, and thus to behave macroscopically like a
           compressible solver.

Antonio F. Di Rienzo (Politecnico di Torino)         PhD Defence   Torino, April 20th 2012   30 / 46
Combustion

   Governing Equations for Reactive Flows
           The system of macroscopic governing equations reads:
                                                ∂t ρ +      · (ρu) = 0,                                      (5a)

                                       ∂t (ρu) +     · (ρu ⊗ u + pI) =            · Π,                       (5b)
                                                                             N
                                               dp
     ∂t (ρhs ) +             · (ρuhs − κ T ) =    +Π:                  u−         h0 ωk Wk −
                                                                                   k˙                    · qrad ,
                                               dt
                                                                            k=1
                                                                                          Qh
                                                                                                             (5c)
                             ∂t (ρYk ) +       · (ρuYk ) =         · (ρDk Yk ) + ωk Wk .
                                                                                 ˙                            (6)
           The corresponding kinetic equations reads:
                           2∆t                    2∆tτ
             gt+∆t = gt +         (f eq − gt ) +         [Ψt + Φt + Qh ],                                     (7)
                          ∆t + 2τ t              ∆t + 2τ
                                                      (∗)     eq(∗)
                                     ξt+∆t = ξt + ωk        ξt        − ξ t + ωk W k .
                                                                              ˙                               (8)
Antonio F. Di Rienzo (Politecnico di Torino)         PhD Defence                   Torino, April 20th 2012    31 / 46
Combustion

   Navier-Stokes-Fourier in the Compressible Limit

                                        2∆t                    2∆tτ
               gt+∆t = gt +                    (fteq − ft ) +         [Ψt + Φt + Qh ]                                          (9)
                                       ∆t + 2τ                ∆t + 2τ

           The correction terms Ψ and Φ are added to the discrete kinetic
           equation in order to remove the deviations in the momentum and
           energy equations.
           By means of these correction terms, it is possible to accurately
           recover Navier-Stokes-Fourier equations in the compressible
           limit.10,11
           This partially bridges over the lacks of the existing models: also
           species transport equation must reckon with compressibility
           effects.
       10
          Prasianakis N. I., Karlin I. V., Lattice Boltmann method for simulation of thermal flows on standard lattice, PRE,
  2007, 76:016702
       11
          Prasianakis N. I., Karlin I. V., Lattice Boltzmann method for simulation of compressible flows on standard
  lattices, PRE, 2008, 78:016704
Antonio F. Di Rienzo (Politecnico di Torino)                  PhD Defence                            Torino, April 20th 2012   32 / 46
Combustion

   Species Equation: Basic Model

           Standard LBM12 emulates the species transport equation by
           means of the following equation:
                                                        eq
           ξi,k (x + vi ∆t, t + ∆t) = ξi,k (x, t)+ωk ξi,k (x, t) , −ξi,k (x, t) +ωk Wk ,
                     ˆ                                                           ˙
                                                                                 (10)
           where
                                  eq
                                 ξi,k = wi ρYk [1 + 3 (ˆ i · u)] ,
                                                       v                         (11)
           and
                                                              1     1    1
                                                   Dk =                −            .                                   (12)
                                                              3     ωk   2
           Eq. (10) recovers the following equation in the continuum limit:

             ∂t (ρYk )+            ·(ρuYk ) =          ·(ρDk Yk )+                · (Dk Yk ρ)+ ωk Wk . (13)
                                                                                               ˙
     12
        Yamamoto K., He X., Doolen G.D., Simulation of combustion field with lattice Boltzmann method, JSP, 2002,
  107:367-383
Antonio F. Di Rienzo (Politecnico di Torino)             PhD Defence                          Torino, April 20th 2012    33 / 46
Combustion

   Species Equation in the Compressible Limit

           Standard LBM solves the species transport equation with the
           following deviation term:

                                                       · (Dk Yk ρ).                                        (14)

           Deviation in the species equation is activated in case of significant
           compressibility effects (i.e. large ρ).
           Due to the simpler nature of the equation, the deviation can be
           removed without adding any corrections term. A strategy fully
           relying upon the LBM formulation can be followed, instead.
           The equilibrium distribution function is modified as follows:

                  eq(∗)            2                         eq(∗)      1
                ξ0,k      = ρYk 1 − ϕ ,                     ξ1,...,4,k = ρYk [ϕ + 3 (ˆ i · u)],
                                                                                     v                     (15)
                                   3                                    6

           with ϕ = ρmin /ρ.
Antonio F. Di Rienzo (Politecnico di Torino)         PhD Defence                 Torino, April 20th 2012    34 / 46
Combustion

   Species Equation in the Compressible Limit

           The relaxation frequency is redefined as follows:

                                                  (∗)                       1
                                                ωk =                                         .                               (16)
                                                             1   1           1    1
                                                               +                −
                                                             2 ϕ             ωk   2

           Through the suggested procedure13 , the advection-
           diffusion-reaction equation in the compressible limit is recovered.
           The particular cases of weak-compressible and incompressible
           flows are readily provided, by setting ϕ = 1.

           Since compressibility effects are accounted for both in Navier-
           Stokes-Fourier and species equations, the proposed model is
           actually suitable for simulating reactive flows.
     13
        Di Rienzo A.F., Asinari P., Chiavazzo E., Prasianakis N.I., Mantzaras J., A lattice Boltzmann model for reactive
  flows simulation, EPL, 2012 (accepted)
Antonio F. Di Rienzo (Politecnico di Torino)                PhD Defence                            Torino, April 20th 2012    35 / 46
Combustion

   Reactor of Compact Gas-Turbines




  Karagiannidis et al.,Experimental and numerical investigation of a propane-fueled, cat-
  alytic mesoscale combustor, Catalysis Today, 2010
Antonio F. Di Rienzo (Politecnico di Torino)         PhD Defence   Torino, April 20th 2012   36 / 46
Combustion

   Combustion in a Narrow Channel
           We consider combustion of premixed stoichiometric H2 -air
           reactive mixture. Radiation is neglected.




Antonio F. Di Rienzo (Politecnico di Torino)         PhD Defence   Torino, April 20th 2012   37 / 46
Combustion

   Reactive LBM at Work: Uin = 0.85 m/s


                                                                   Density, x-velocity and
                                                                   temperature along the
                                                                   horizontal symmetry plane
                                                                   (top).
                                                                   Species mass fractions
                                                                   (bottom) along the channel
                                                                   walls predicted by stan-
                                                                   dard LBM (dashed-line)
                                                                   are compared to those
                                                                   recovered by the pro-
                                                                   posed model (solid-lines).
                                                                   Symbols are the FLUENT
                                                                   solution:    triangles H2 ,
                                                                   squares O2 , circles H2 O.



Antonio F. Di Rienzo (Politecnico di Torino)         PhD Defence          Torino, April 20th 2012   38 / 46
Combustion

   Reactive LBM at Work: Uin = 0.48 m/s


                                                                   Density, x-velocity, tem-
                                                                   perature (top) and species
                                                                   mass fractions (bottom)
                                                                   along the horizontal sym-
                                                                   metry plane.
                                                                   The maximun temperature
                                                                   ratio is 5.5, twice greater
                                                                   than previous models.

                                                                   Maximum relative errors
                                                                   are: 2% for the density,
                                                                   4.2% for the velocity, 5.7%
                                                                   for the temperature, 3% for
                                                                   the species.



Antonio F. Di Rienzo (Politecnico di Torino)         PhD Defence          Torino, April 20th 2012   39 / 46
Conclusions & Acknowledgements

   Outline Compass

   1    Summary of the talk

   2    Introduction

   3    The Lattice Boltzmann Method (LBM)

   4    Link-wise Artificial Compressibility Method (LW-ACM)

   5    Radiative Transfer Equation

   6    Combustion

   7    Conclusions & Acknowledgements



Antonio F. Di Rienzo (Politecnico di Torino)            PhD Defence   Torino, April 20th 2012   40 / 46
Conclusions & Acknowledgements

   Conclusions

      1    The Link-wise Artificial Compressiblity Method (LW-ACM) has
           been presented. The updating rule of the distribution function
           depends only on the hydrodynamic variables. This fully alleviates
           the problems of additional higher-order moments (e.g. at
           boundaries) in pseudo-kinetic schemes.
      2    With the LW-ACM, the ability of LBM to easily deal with complex
           boundaries is preserved (no body-fitting is required) and all
           existing techniques for classical computational fluid dynamic
           (CFD) readily available.
      3    The LW-ACM represents an alternative to both classical CFD and
           LBM, as long as incompressible and weak-compressible flows are
           investigated. It is a first step towards the development of models
           for reactive flows. Compressibility effects are not taken into
           account and its application to combustion simulation will be
           considered in a near future.
Antonio F. Di Rienzo (Politecnico di Torino)            PhD Defence   Torino, April 20th 2012   41 / 46
Conclusions & Acknowledgements

   Conclusions


      4    A general lattice Boltzmann scheme for simulating reactive flows
           at the low Mach number limit has been developed, so as to
           compensate for the limitations of the models available in literature.
      5    Accounting for compressibility effects in the Navier-Stokes-Fourier
           and species equations, significant density (temperature) variations
           can be handled. This extension allows to apply lattice Boltzmann
           method (LBM) to a wide range of phenomena, which were not
           properly addressed so far.
      6    The accuracy of the previous radiative LBM model has been
           improved by discretizing also the polar angle. The possibility to
           use the same data structures makes the coupling with the reactive
           flow solver straightforward.



Antonio F. Di Rienzo (Politecnico di Torino)            PhD Defence   Torino, April 20th 2012   42 / 46
Conclusions & Acknowledgements




  Journal Publications
      1    Di Rienzo A. F., Asinari P., Chiavazzo E., Prasianakis N. I., Mantzaras J., A
           lattice Boltzmann model for reactive flows simulations, EuroPhysics Letters,
           2012 (accepted), (Impact Factor: 2.752)
      2    Asinari P., Ohwada T., Chiavazzo E., Di Rienzo A. F., Link-wise Artificial
           Compressibility Method, Journal of Computational Physics, 2012 (preprint),
           (Impact Factor: 2.345)
      3    Di Rienzo A. F., Asinari P., Borchiellini R., Mishra S. C., Improved angular
           discretization and error analysis of the lattice Boltzmann method for solving
           radiative heat transfer in a participating medium, Int. Jour. Num. Meth. for Heat
           and Fluid Flow, 2011, 21:640-662, (Impact Factor: 0.53)


  Proceedings
      1    Di Rienzo A. F., Asinari P., Chiavazzo E., Link-wise equilibrium-based lattice
           Boltzmann method, Proceedings of the XXIX UIT Heat Transfer Conference,
           2011



Antonio F. Di Rienzo (Politecnico di Torino)            PhD Defence    Torino, April 20th 2012   43 / 46
Conclusions & Acknowledgements


  Conference Presentations
      1    Di Rienzo A. F., Asinari P., Chiavazzo E., Prasianakis N. I., Mantzaras J.,
           Coupling lattice Boltzmann model with reduced chemical kinetics for combustion
           simulations, 8th International Conference for Mesoscopic Methods in
           Engineering and Science (ICMMES), July 4-8 2011, Lyon, France
      2    Di Rienzo A. F., Asinari P., Chiavazzo E., Link-wise equilibrium based lattice
           Boltzmann method, 29th UIT Heat Transfer Conference, June 20-22, 2011, Turin,
           Italy
      3    Di Rienzo A. F., Chiavazzo E., Asinari P., Radiative lattice Boltzmann method
           applied to combustion simulation and reduced chemical kinetics, 19th
           International Conference on the Discrete Simulation of Fluid Dynamics (DSFD),
           July 5-9, Rome, Italy
      4    Di Rienzo A. F., Izquierdo S., Asinari P., Mishra S. C., Borchiellini R., The lattice
           Boltzmann method in solving radiative heat transfer in a participating medium,
           First International Conference on Computational Methods for Thermal Problems
           (ThermaComp), September, 8-10, 2009, Naples, Italy
  Award
      1    Best presentation at the Energetics PhD Day, Politecnico di Torino, December
           20, 2011

Antonio F. Di Rienzo (Politecnico di Torino)            PhD Defence      Torino, April 20th 2012   44 / 46
Conclusions & Acknowledgements

   Acknowledgements




                                               Dr. John Mantzaras, Dr. Nikolaos Prasianakis




Antonio F. Di Rienzo (Politecnico di Torino)            PhD Defence         Torino, April 20th 2012   45 / 46
Conclusions & Acknowledgements




        Thank you for your attention!




Antonio F. Di Rienzo (Politecnico di Torino)            PhD Defence   Torino, April 20th 2012   46 / 46

More Related Content

Similar to Mesoscopic numerical methods for reactive flows

The stuff that proteins are made of
The stuff that proteins are made ofThe stuff that proteins are made of
The stuff that proteins are made ofkhinsen
 
THE LATTICE BOLTZMANN METHOD IN SOLVING RADIATIVE HEAT TRANSFER IN A PARTICIP...
THE LATTICE BOLTZMANN METHOD IN SOLVING RADIATIVE HEAT TRANSFER IN A PARTICIP...THE LATTICE BOLTZMANN METHOD IN SOLVING RADIATIVE HEAT TRANSFER IN A PARTICIP...
THE LATTICE BOLTZMANN METHOD IN SOLVING RADIATIVE HEAT TRANSFER IN A PARTICIP...FabioDiRienzo
 
The Algorithms of Life - Scientific Computing for Systems Biology
The Algorithms of Life - Scientific Computing for Systems BiologyThe Algorithms of Life - Scientific Computing for Systems Biology
The Algorithms of Life - Scientific Computing for Systems Biologyinside-BigData.com
 
Modeling ultrasonic attenuation coefficient and comparative study with the pr...
Modeling ultrasonic attenuation coefficient and comparative study with the pr...Modeling ultrasonic attenuation coefficient and comparative study with the pr...
Modeling ultrasonic attenuation coefficient and comparative study with the pr...IOSR Journals
 
A Technique for Partially Solving a Family of Diffusion Problems
A Technique for Partially Solving a Family of Diffusion ProblemsA Technique for Partially Solving a Family of Diffusion Problems
A Technique for Partially Solving a Family of Diffusion Problemsijtsrd
 
MSc_Thesis_Michael_Aldam
MSc_Thesis_Michael_AldamMSc_Thesis_Michael_Aldam
MSc_Thesis_Michael_AldamMichael Aldam
 
Damage detection in cfrp plates by means of numerical modeling of lamb waves ...
Damage detection in cfrp plates by means of numerical modeling of lamb waves ...Damage detection in cfrp plates by means of numerical modeling of lamb waves ...
Damage detection in cfrp plates by means of numerical modeling of lamb waves ...eSAT Journals
 
optics-03-00017.pdf
optics-03-00017.pdfoptics-03-00017.pdf
optics-03-00017.pdfRocoNapan
 
3 di metrology-slideshare
3 di metrology-slideshare3 di metrology-slideshare
3 di metrology-slideshareKeshab Paudel
 
2012 tus lecture 6
2012 tus lecture 62012 tus lecture 6
2012 tus lecture 6AllenHermann
 

Similar to Mesoscopic numerical methods for reactive flows (20)

351 b p.3
351 b p.3351 b p.3
351 b p.3
 
351 b p.3
351 b p.3351 b p.3
351 b p.3
 
12098
1209812098
12098
 
The stuff that proteins are made of
The stuff that proteins are made ofThe stuff that proteins are made of
The stuff that proteins are made of
 
THE LATTICE BOLTZMANN METHOD IN SOLVING RADIATIVE HEAT TRANSFER IN A PARTICIP...
THE LATTICE BOLTZMANN METHOD IN SOLVING RADIATIVE HEAT TRANSFER IN A PARTICIP...THE LATTICE BOLTZMANN METHOD IN SOLVING RADIATIVE HEAT TRANSFER IN A PARTICIP...
THE LATTICE BOLTZMANN METHOD IN SOLVING RADIATIVE HEAT TRANSFER IN A PARTICIP...
 
peerj-cs-17
peerj-cs-17peerj-cs-17
peerj-cs-17
 
The Algorithms of Life - Scientific Computing for Systems Biology
The Algorithms of Life - Scientific Computing for Systems BiologyThe Algorithms of Life - Scientific Computing for Systems Biology
The Algorithms of Life - Scientific Computing for Systems Biology
 
Modeling ultrasonic attenuation coefficient and comparative study with the pr...
Modeling ultrasonic attenuation coefficient and comparative study with the pr...Modeling ultrasonic attenuation coefficient and comparative study with the pr...
Modeling ultrasonic attenuation coefficient and comparative study with the pr...
 
nte.pdf
nte.pdfnte.pdf
nte.pdf
 
LOY-DISSERTATION-2013
LOY-DISSERTATION-2013LOY-DISSERTATION-2013
LOY-DISSERTATION-2013
 
LSANCHEZ_THESIS_MC
LSANCHEZ_THESIS_MCLSANCHEZ_THESIS_MC
LSANCHEZ_THESIS_MC
 
A Technique for Partially Solving a Family of Diffusion Problems
A Technique for Partially Solving a Family of Diffusion ProblemsA Technique for Partially Solving a Family of Diffusion Problems
A Technique for Partially Solving a Family of Diffusion Problems
 
MSc_Thesis_Michael_Aldam
MSc_Thesis_Michael_AldamMSc_Thesis_Michael_Aldam
MSc_Thesis_Michael_Aldam
 
Damage detection in cfrp plates by means of numerical modeling of lamb waves ...
Damage detection in cfrp plates by means of numerical modeling of lamb waves ...Damage detection in cfrp plates by means of numerical modeling of lamb waves ...
Damage detection in cfrp plates by means of numerical modeling of lamb waves ...
 
optics-03-00017.pdf
optics-03-00017.pdfoptics-03-00017.pdf
optics-03-00017.pdf
 
MEMS Sensor.pptx
MEMS Sensor.pptxMEMS Sensor.pptx
MEMS Sensor.pptx
 
Solid Catalyzed Reactions
Solid Catalyzed ReactionsSolid Catalyzed Reactions
Solid Catalyzed Reactions
 
3 di metrology-slideshare
3 di metrology-slideshare3 di metrology-slideshare
3 di metrology-slideshare
 
2012 tus lecture 6
2012 tus lecture 62012 tus lecture 6
2012 tus lecture 6
 
Turin Bnct Moss
Turin Bnct MossTurin Bnct Moss
Turin Bnct Moss
 

Mesoscopic numerical methods for reactive flows

  • 1. Mesoscopic numerical methods for reactive flows: Lattice Boltzmann method and beyond Candidate: Antonio F. Di Rienzo Advisors: Prof. P. Asinari, Dr. E. Chiavazzo External Collaborators: Dr. J. Mantzaras, Dr. N. Prasianakis (Paul Scherrer Institute) Energy Department, Politecnico di Torino, Torino, Italy PhD Defence, Torino 20 April, 2012 Antonio F. Di Rienzo (Politecnico di Torino) PhD Defence Torino, April 20th 2012 1 / 46
  • 2. Outline of The Talk 1 Summary of the talk 2 Introduction 3 The Lattice Boltzmann Method (LBM) 4 Link-wise Artificial Compressibility Method (LW-ACM) 5 Radiative Transfer Equation 6 Combustion 7 Conclusions & Acknowledgements Antonio F. Di Rienzo (Politecnico di Torino) PhD Defence Torino, April 20th 2012 2 / 46
  • 3. Summary of the talk Outline Compass 1 Summary of the talk 2 Introduction 3 The Lattice Boltzmann Method (LBM) 4 Link-wise Artificial Compressibility Method (LW-ACM) 5 Radiative Transfer Equation 6 Combustion 7 Conclusions & Acknowledgements Antonio F. Di Rienzo (Politecnico di Torino) PhD Defence Torino, April 20th 2012 3 / 46
  • 4. Summary of the talk Topics Link-wise Artificial Compressibility Method 1 : CFD by kinetic mock-up models. The Radiative transfer equation (RTE) is solved by means of lattice Boltzmann method (LBM) formalism 2 : intensity is updated according to lattice velocities. A consistent lattice Boltzmann model for reactive flows 3 is presented. It addresses the lack of accurate combustion models in LBM literature (in collaboration with the Paul Scherrer Institute, Switzerland). 1 Asinari P., Ohwada T., Chiavazzo E., Di Rienzo A.F., Link-wise Artificial Compressiniblity Method, Journ. Comp. Phys., 2012 (preprint), (Impact Factor: 2.345) 2 Di Rienzo A. F., Asinari P., Borchiellini R., Mishra S. C., Improved angular discretization and error analysis of the lattice Boltzmann method for solving radiative heat transfer in a participating medium, Int. Journ. Num. Meth. Heat Fluid Flow, 21 (5), 640-662, 2011, (Impact Factor: 0.53) 3 Di Rienzo A. F., Asinari P., Chiavazzo E., Prasianakis N., Mantzaras J., A Lattice Boltzmann model for reactive flows simulation, EuroPhys. Lett., 2012 (accepted), (Impact Factor: 2.752) Antonio F. Di Rienzo (Politecnico di Torino) PhD Defence Torino, April 20th 2012 4 / 46
  • 5. Introduction Outline Compass 1 Summary of the talk 2 Introduction 3 The Lattice Boltzmann Method (LBM) 4 Link-wise Artificial Compressibility Method (LW-ACM) 5 Radiative Transfer Equation 6 Combustion 7 Conclusions & Acknowledgements Antonio F. Di Rienzo (Politecnico di Torino) PhD Defence Torino, April 20th 2012 5 / 46
  • 6. Introduction Simulating Reactive Flows Reactive flows can be found in several energy systems: internal combustion engines, industrial burners, gas turbine combustors... Numerical modeling of reactive flows is a challenging task per se: simultaneous processes must be taken into account (turbulent mixing, multi-phase fluid-dynamics, radiative heat transfer). Furthermore, in some applications, the numerical model must be able to deal with different length scales. The demand of energy systems with high energy efficiency and low environmental impact requires the development of new energy systems to be used in industrial and domestic applications. Porous media combustion (PMC) proved to be a feasible option. Porous media (PM) burners offer some advantages compared to classical burners. The internal process of heat recovery makes the combustion process more efficient, which comes with a lower emission of polluntants, such as NOx and CO. Antonio F. Di Rienzo (Politecnico di Torino) PhD Defence Torino, April 20th 2012 6 / 46
  • 7. Introduction Porous Media Burner Abdul Mujeebu at al., Application of porous media combustion technology - A review, Applied Energy, 2009 Antonio F. Di Rienzo (Politecnico di Torino) PhD Defence Torino, April 20th 2012 7 / 46
  • 8. Introduction Fluid Flow in Porous Media If studies are carried out at the level of the pore-scale, detailed local informations of the flow can be obtained. In Lattice Boltzmann Method (LBM) the link between the pore-scale and the macro-scale is readily available. LBM is often referred to as mesoscopic approach. Antonio F. Di Rienzo (Politecnico di Torino) PhD Defence Torino, April 20th 2012 8 / 46
  • 9. The Lattice Boltzmann Method (LBM) Outline Compass 1 Summary of the talk 2 Introduction 3 The Lattice Boltzmann Method (LBM) 4 Link-wise Artificial Compressibility Method (LW-ACM) 5 Radiative Transfer Equation 6 Combustion 7 Conclusions & Acknowledgements Antonio F. Di Rienzo (Politecnico di Torino) PhD Defence Torino, April 20th 2012 9 / 46
  • 10. The Lattice Boltzmann Method (LBM) What is Lattice Boltzmann Method? "The lattice Boltzmann method (LBM) is used for the numerical simulation of physical phenomena and serves as an alternative to classical solvers of partial differential equation (PDEs)" [www.lbmethod.org]. The main unknown is the discrete distribution function, from which all relevant macroscopic quantities (satisfying some target PDEs) can be derived. The operative formula consists of the (a) relaxation process and the (b) advection process: fi (x + vi ∆t, t + ∆t) − fi (x, t) = ω [fieq (x, t) − fi (x, t)]. ˆ (1) The updating of fi is link-wise, in the sense that only the informations along the directions identified by the lattice velocity vi ˆ are required. Antonio F. Di Rienzo (Politecnico di Torino) PhD Defence Torino, April 20th 2012 10 / 46
  • 11. The Lattice Boltzmann Method (LBM) Dealing with Complex Boundaries Antonio F. Di Rienzo (Politecnico di Torino) PhD Defence Torino, April 20th 2012 11 / 46
  • 12. The Lattice Boltzmann Method (LBM) State of Art Antonio F. Di Rienzo (Politecnico di Torino) PhD Defence Torino, April 20th 2012 12 / 46
  • 13. Link-wise Artificial Compressibility Method (LW-ACM) Outline Compass 1 Summary of the talk 2 Introduction 3 The Lattice Boltzmann Method (LBM) 4 Link-wise Artificial Compressibility Method (LW-ACM) 5 Radiative Transfer Equation 6 Combustion 7 Conclusions & Acknowledgements Antonio F. Di Rienzo (Politecnico di Torino) PhD Defence Torino, April 20th 2012 13 / 46
  • 14. Link-wise Artificial Compressibility Method (LW-ACM) Limitation of Lattice Boltzmann Method LBM works with larger set of unknowns, including the higher-order moments beyond hydrodynamics: ntimes mtimes Πxx···x yy···y (xx · · · x, yy · · · y) = ˆn ˆm vi,α vi,β fi . i They are unessential as long as the continuum limit is the main concern and can lead to numerical instabilities 4 . Ohwada & Asinari 5,6 revived the Artificial Compressibility Method (ACM) as an high-order accurate numerical method for the incompressible Navier-Stokes equations (INSE). The asymptotic analysis of the LBM updating rule delivers the same governing equations of ACM. 4 Dellar P., Nonhydrodynamic modes and a priori construction of shallow water lattice Boltzmann equations, PRE, 2002, 65:036309 5 Ohwada T., Asinari P., Artificial compressibility method revisited: asymptotic numerical method for incompressible Navier-Stokes equations, JCP, 2010, 229:1698-1723 6 Ohwada T., Asinari P., Yabusaki D., Artificial compressibility method and lattice Boltzmann method: similarities and differences, Comp. Math. Appl., 2011, 61:3461-3474 Antonio F. Di Rienzo (Politecnico di Torino) PhD Defence Torino, April 20th 2012 14 / 46
  • 15. Link-wise Artificial Compressibility Method (LW-ACM) ACM vs LBM Unlike LBM, ACM deals with macroscopic variables only. It offers the opportunity of exploiting all existing Finite-Difference (FD) technologies. ACM relies on standard meshing techniques for dealing with complex boundaries. Unstructured body-fitted meshes are used in order to adapt the computational grid to the real object. This requires advanced algorithms, which imply an additional computational overhead. A possible improvements would be to have the LBM updating rule of the distribution function depend only on the local equilibrium, but retain the link-wise formulation. Antonio F. Di Rienzo (Politecnico di Torino) PhD Defence Torino, April 20th 2012 15 / 46
  • 16. Link-wise Artificial Compressibility Method (LW-ACM) Link-wise Artificial Compressibility Method (LW-ACM) In the attempt of making ACM more similar to LBM, the following formula is proposed 7 : (e) fi (x, t + ∆t) = fi (x − vi ∆t, t) ˆ ω−1 (e,o) (e,o) (2) +2 fi (x, t) − fi (x − vi ∆t, t) , ˆ ω Similarly to LBM, the updating rule of Eq.(2) is link-wise. (e) (e,o) fi and fi are local functions of ρ = i fi and ρu = i vi fi , ˆ but they do not depend on the higher-order moments. Eq. (2) recovers the incompressible Navier-Stokes equations with a kinematic viscosity defined as: 1 1 1 ν= − . 3 ω 2 7 Asinari P., Ohwada T., Chiavazzo E., Di Rienzo A.F., Link-wise Artificial Compressibility Method, Journ. Comp. Phys., 2012 (preprint) Antonio F. Di Rienzo (Politecnico di Torino) PhD Defence Torino, April 20th 2012 16 / 46
  • 17. Link-wise Artificial Compressibility Method (LW-ACM) Simple Boundaries L2 [¯ ] u L2 [¯ ] u ∆x M a ∝ ∆t/∆x ν∝ Re−1 Test 1 Test 2 1 × 10−1 3 × 10−2 3 × 10−2 1.74 × 10−3 4.59 × 10−4 5 × 10−2 1.5 × 10−2 3 × 10−2 4.49 × 10−4 1.21 × 10−4 2.5 × 10−2 7.5 × 10−2 3 × 10−2 1.20 × 10−4 3.11 × 10−5 Antonio F. Di Rienzo (Politecnico di Torino) PhD Defence Torino, April 20th 2012 17 / 46
  • 18. Link-wise Artificial Compressibility Method (LW-ACM) Complex Boundaries ∆x M a ∝ ∆t/∆x ν ∝ Re−1 L2 [¯ ] u 5 × 10−2 3 × 10−1 4 × 10−2 1.84 × 10−3 1.5 × 10−2 1.5 × 10−1 4 × 10−2 3.83 × 10−4 1.25 × 10−2 7.5 × 10−1 4 × 10−2 1.11 × 10−4 Antonio F. Di Rienzo (Politecnico di Torino) PhD Defence Torino, April 20th 2012 18 / 46
  • 19. Link-wise Artificial Compressibility Method (LW-ACM) 2D Lid Driven Cavity: Streamlines Comparison between LW-ACM and BGK at Re = 5000. Antonio F. Di Rienzo (Politecnico di Torino) PhD Defence Torino, April 20th 2012 19 / 46
  • 20. Link-wise Artificial Compressibility Method (LW-ACM) 2D Lid Driven Cavity: Pressure Contours Comparison between LW-ACM and BGK at Re = 5000. Antonio F. Di Rienzo (Politecnico di Torino) PhD Defence Torino, April 20th 2012 20 / 46
  • 21. Link-wise Artificial Compressibility Method (LW-ACM) LW-ACM vs other INSE solvers 2D lid driven cavity flow at Re = 5000: comparison between the LW- ACM and alternative solvers for INSE from literature. Scheme Grid (xp , yp ) (xlr , ylr ) Energy Enstrophy Present 128 × 128 (0.51652, 0.53754) (0.81081, 0.079079) 0.039845 29.247 ACM, 2010 128 × 128 (0.52052, 0.53954) (0.82883, 0.071071) 0.027430 41.249 Opt. ACM, 2010 128 × 128 (0.51652, 0.53854) (0.80981, 0.072072) 0.038371 37.704 BGK-LBM 128 × 128 unstable unstable unstable unstable MRT-LBM 128 × 128 (0.51652, 0.53554) (0.80881, 0.075075) 0.043600 37.404 Bruneau, 2006 128 × 128 (0.51562, 0.53906) (0.80469, 0.070313) 0.043566 30.861 Present 256 × 256 (0.51552, 0.53554) (0.80581, 0.074074) 0.044391 34.821 ACM, 2010 256 × 256 (0.51652, 0.53654) (0.80881, 0.072072) 0.040896 43.198 Opt. ACM, 2010 256 × 256 (0.51451, 0.53654) (0.80380, 0.072072) 0.048114 42.290 BGK-LBM 250 × 250 (0.51752, 0.54054) (0.80781, 0.074074) 0.041614 40.455 MRT-LBM 256 × 256 (0.51552, 0.53554) (0.80681, 0.074074) 0.045222 40.833 Bruneau, 2006 256 × 256 (0.51562, 0.53516) (0.80859, 0.074219) 0.046204 34.368 Bruneau, 2006 2048 × 2048 (0.51465, 0.53516) (0.80566, 0.073242) 0.047290 40.261 Antonio F. Di Rienzo (Politecnico di Torino) PhD Defence Torino, April 20th 2012 21 / 46
  • 22. Radiative Transfer Equation Outline Compass 1 Summary of the talk 2 Introduction 3 The Lattice Boltzmann Method (LBM) 4 Link-wise Artificial Compressibility Method (LW-ACM) 5 Radiative Transfer Equation 6 Combustion 7 Conclusions & Acknowledgements Antonio F. Di Rienzo (Politecnico di Torino) PhD Defence Torino, April 20th 2012 22 / 46
  • 23. Radiative Transfer Equation Radiative Lattice Boltzmann Model Combustion generally incorporates also radiation process, which plays a significant role if high temperature and a participating medium are involved. Treatment of thermal radiation is of key importance to develop mathematical model of combustion system. Due to its kinetic nature, the Radiative Transfer Equation (RTE) can be formulated according to the LBM formalism: the intensity is already a particle (photon) distribution function. Recently a LBM model 8 has been proposed to solve RTE in a participating medium. However, further improvements are required: the polar angle is not discretized, spoiling the accuracy of the method with respect to standard Finite Volume Method (FVM). 8 Asinari P., Mishra S.C., Borchiellini R., A lattice Boltzmann formulation to the analysis of radiative heat transfer problems in a participating medium, NHT B, 2010, 57:1-21 Antonio F. Di Rienzo (Politecnico di Torino) PhD Defence Torino, April 20th 2012 23 / 46
  • 24. Radiative Transfer Equation Radiative Transfer Equation For an absorbing, emitting and scattering participating medium and under the assumption of isotropic scattering, radiative LBM reads: Iij (x + v∆t, t + ∆t) = Iij (x, t) + Vi,j (κa + σs ) [Ib (x, t) − Iij (x, t)]. (3) The link-wise formulation makes unnecessary to march from each single corner (e.g. Finite Volumes). The data structures are the same as the fluid flow. Antonio F. Di Rienzo (Politecnico di Torino) PhD Defence Torino, April 20th 2012 24 / 46
  • 25. Radiative Transfer Equation Azimuthal Angle Discretization The azimuthal angle δ is discretized by introducing a finite number of discrete velocities vi = (vx,i , vy,i ) lying on the lattice. Even if not discretizing the polar angle saves a lot of computational time, this approximation spoils the accuracy of the method. Antonio F. Di Rienzo (Politecnico di Torino) PhD Defence Torino, April 20th 2012 25 / 46
  • 26. Radiative Transfer Equation Azimuthal Angle Discretization Dimensionless heat flux as function for different extinction co- efficients β = κa + σs . The polar angle is not discretized (Asinari P. et al., NHT B, 2010). The error between LBM and FVM is 2%. Antonio F. Di Rienzo (Politecnico di Torino) PhD Defence Torino, April 20th 2012 26 / 46
  • 27. Radiative Transfer Equation Polar Angle Discretization An additional velocity vz,j is introduced along the z-axis (off the lattice) in order to discretize the polar angle 9 : π vz,j = tan − γj vi (4) 2 The projection on the lattice of the total velocity Vij must overlap vi . 9 Di Rienzo A.F., Asinari P., Borchiellini R., Mishra S.C., Improved angular discretization and error analysis of the lattice Boltzmann method for solving radiative heat transfer in a participating medium, Int. Jour. Num. Meth. Heat Fluid Flow, 2011, 21:640-662 Antonio F. Di Rienzo (Politecnico di Torino) PhD Defence Torino, April 20th 2012 27 / 46
  • 28. Radiative Transfer Equation 2D Square Enclosure Dimensionless heat flux for β = 2 and β = 5. Radiative equilibrium condition is as- sumed, i.e. · qrad = 0. The error between LBM and FVM is reduced to 0.5%. β = 2.0 β = 5.0 Nx Nδ Nγ L2 [Ψ] Nx Nδ Nγ L2 [Ψ] 40 8 4 7.31 × 10−2 100 8 4 5.54 × 10−2 80 16 8 3.16 × 10−2 200 16 8 2.01 × 10−2 160 32 16 2.00 × 10−2 400 32 16 1.30 × 10−2 Antonio F. Di Rienzo (Politecnico di Torino) PhD Defence Torino, April 20th 2012 28 / 46
  • 29. Combustion Outline Compass 1 Summary of the talk 2 Introduction 3 The Lattice Boltzmann Method (LBM) 4 Link-wise Artificial Compressibility Method (LW-ACM) 5 Radiative Transfer Equation 6 Combustion 7 Conclusions & Acknowledgements Antonio F. Di Rienzo (Politecnico di Torino) PhD Defence Torino, April 20th 2012 29 / 46
  • 30. Combustion Simulating Reactive Flows: What’s Required? Despite the last decade has witnessed a significant improvement in describing various problems (flow in porous media, thermal flows, etc.), application of LBM to combustion is still limited. "In principle, once lattice Boltzmann models can properly account for large temperature variation, extension to reactive flows essentially involves adding appropriate source terms..."[C. E. Frouzakis, Fluid Mechanics and Its Applications, 2011] So far, no reactive LBM models has been proposed that satisfactorily fulfill this requirement. The LBM model is requested to accurately recover the Navier- Stokes-Fourier equations, coupled to a transport equation for each chemical species, and thus to behave macroscopically like a compressible solver. Antonio F. Di Rienzo (Politecnico di Torino) PhD Defence Torino, April 20th 2012 30 / 46
  • 31. Combustion Governing Equations for Reactive Flows The system of macroscopic governing equations reads: ∂t ρ + · (ρu) = 0, (5a) ∂t (ρu) + · (ρu ⊗ u + pI) = · Π, (5b) N dp ∂t (ρhs ) + · (ρuhs − κ T ) = +Π: u− h0 ωk Wk − k˙ · qrad , dt k=1 Qh (5c) ∂t (ρYk ) + · (ρuYk ) = · (ρDk Yk ) + ωk Wk . ˙ (6) The corresponding kinetic equations reads: 2∆t 2∆tτ gt+∆t = gt + (f eq − gt ) + [Ψt + Φt + Qh ], (7) ∆t + 2τ t ∆t + 2τ (∗) eq(∗) ξt+∆t = ξt + ωk ξt − ξ t + ωk W k . ˙ (8) Antonio F. Di Rienzo (Politecnico di Torino) PhD Defence Torino, April 20th 2012 31 / 46
  • 32. Combustion Navier-Stokes-Fourier in the Compressible Limit 2∆t 2∆tτ gt+∆t = gt + (fteq − ft ) + [Ψt + Φt + Qh ] (9) ∆t + 2τ ∆t + 2τ The correction terms Ψ and Φ are added to the discrete kinetic equation in order to remove the deviations in the momentum and energy equations. By means of these correction terms, it is possible to accurately recover Navier-Stokes-Fourier equations in the compressible limit.10,11 This partially bridges over the lacks of the existing models: also species transport equation must reckon with compressibility effects. 10 Prasianakis N. I., Karlin I. V., Lattice Boltmann method for simulation of thermal flows on standard lattice, PRE, 2007, 76:016702 11 Prasianakis N. I., Karlin I. V., Lattice Boltzmann method for simulation of compressible flows on standard lattices, PRE, 2008, 78:016704 Antonio F. Di Rienzo (Politecnico di Torino) PhD Defence Torino, April 20th 2012 32 / 46
  • 33. Combustion Species Equation: Basic Model Standard LBM12 emulates the species transport equation by means of the following equation: eq ξi,k (x + vi ∆t, t + ∆t) = ξi,k (x, t)+ωk ξi,k (x, t) , −ξi,k (x, t) +ωk Wk , ˆ ˙ (10) where eq ξi,k = wi ρYk [1 + 3 (ˆ i · u)] , v (11) and 1 1 1 Dk = − . (12) 3 ωk 2 Eq. (10) recovers the following equation in the continuum limit: ∂t (ρYk )+ ·(ρuYk ) = ·(ρDk Yk )+ · (Dk Yk ρ)+ ωk Wk . (13) ˙ 12 Yamamoto K., He X., Doolen G.D., Simulation of combustion field with lattice Boltzmann method, JSP, 2002, 107:367-383 Antonio F. Di Rienzo (Politecnico di Torino) PhD Defence Torino, April 20th 2012 33 / 46
  • 34. Combustion Species Equation in the Compressible Limit Standard LBM solves the species transport equation with the following deviation term: · (Dk Yk ρ). (14) Deviation in the species equation is activated in case of significant compressibility effects (i.e. large ρ). Due to the simpler nature of the equation, the deviation can be removed without adding any corrections term. A strategy fully relying upon the LBM formulation can be followed, instead. The equilibrium distribution function is modified as follows: eq(∗) 2 eq(∗) 1 ξ0,k = ρYk 1 − ϕ , ξ1,...,4,k = ρYk [ϕ + 3 (ˆ i · u)], v (15) 3 6 with ϕ = ρmin /ρ. Antonio F. Di Rienzo (Politecnico di Torino) PhD Defence Torino, April 20th 2012 34 / 46
  • 35. Combustion Species Equation in the Compressible Limit The relaxation frequency is redefined as follows: (∗) 1 ωk = . (16) 1 1 1 1 + − 2 ϕ ωk 2 Through the suggested procedure13 , the advection- diffusion-reaction equation in the compressible limit is recovered. The particular cases of weak-compressible and incompressible flows are readily provided, by setting ϕ = 1. Since compressibility effects are accounted for both in Navier- Stokes-Fourier and species equations, the proposed model is actually suitable for simulating reactive flows. 13 Di Rienzo A.F., Asinari P., Chiavazzo E., Prasianakis N.I., Mantzaras J., A lattice Boltzmann model for reactive flows simulation, EPL, 2012 (accepted) Antonio F. Di Rienzo (Politecnico di Torino) PhD Defence Torino, April 20th 2012 35 / 46
  • 36. Combustion Reactor of Compact Gas-Turbines Karagiannidis et al.,Experimental and numerical investigation of a propane-fueled, cat- alytic mesoscale combustor, Catalysis Today, 2010 Antonio F. Di Rienzo (Politecnico di Torino) PhD Defence Torino, April 20th 2012 36 / 46
  • 37. Combustion Combustion in a Narrow Channel We consider combustion of premixed stoichiometric H2 -air reactive mixture. Radiation is neglected. Antonio F. Di Rienzo (Politecnico di Torino) PhD Defence Torino, April 20th 2012 37 / 46
  • 38. Combustion Reactive LBM at Work: Uin = 0.85 m/s Density, x-velocity and temperature along the horizontal symmetry plane (top). Species mass fractions (bottom) along the channel walls predicted by stan- dard LBM (dashed-line) are compared to those recovered by the pro- posed model (solid-lines). Symbols are the FLUENT solution: triangles H2 , squares O2 , circles H2 O. Antonio F. Di Rienzo (Politecnico di Torino) PhD Defence Torino, April 20th 2012 38 / 46
  • 39. Combustion Reactive LBM at Work: Uin = 0.48 m/s Density, x-velocity, tem- perature (top) and species mass fractions (bottom) along the horizontal sym- metry plane. The maximun temperature ratio is 5.5, twice greater than previous models. Maximum relative errors are: 2% for the density, 4.2% for the velocity, 5.7% for the temperature, 3% for the species. Antonio F. Di Rienzo (Politecnico di Torino) PhD Defence Torino, April 20th 2012 39 / 46
  • 40. Conclusions & Acknowledgements Outline Compass 1 Summary of the talk 2 Introduction 3 The Lattice Boltzmann Method (LBM) 4 Link-wise Artificial Compressibility Method (LW-ACM) 5 Radiative Transfer Equation 6 Combustion 7 Conclusions & Acknowledgements Antonio F. Di Rienzo (Politecnico di Torino) PhD Defence Torino, April 20th 2012 40 / 46
  • 41. Conclusions & Acknowledgements Conclusions 1 The Link-wise Artificial Compressiblity Method (LW-ACM) has been presented. The updating rule of the distribution function depends only on the hydrodynamic variables. This fully alleviates the problems of additional higher-order moments (e.g. at boundaries) in pseudo-kinetic schemes. 2 With the LW-ACM, the ability of LBM to easily deal with complex boundaries is preserved (no body-fitting is required) and all existing techniques for classical computational fluid dynamic (CFD) readily available. 3 The LW-ACM represents an alternative to both classical CFD and LBM, as long as incompressible and weak-compressible flows are investigated. It is a first step towards the development of models for reactive flows. Compressibility effects are not taken into account and its application to combustion simulation will be considered in a near future. Antonio F. Di Rienzo (Politecnico di Torino) PhD Defence Torino, April 20th 2012 41 / 46
  • 42. Conclusions & Acknowledgements Conclusions 4 A general lattice Boltzmann scheme for simulating reactive flows at the low Mach number limit has been developed, so as to compensate for the limitations of the models available in literature. 5 Accounting for compressibility effects in the Navier-Stokes-Fourier and species equations, significant density (temperature) variations can be handled. This extension allows to apply lattice Boltzmann method (LBM) to a wide range of phenomena, which were not properly addressed so far. 6 The accuracy of the previous radiative LBM model has been improved by discretizing also the polar angle. The possibility to use the same data structures makes the coupling with the reactive flow solver straightforward. Antonio F. Di Rienzo (Politecnico di Torino) PhD Defence Torino, April 20th 2012 42 / 46
  • 43. Conclusions & Acknowledgements Journal Publications 1 Di Rienzo A. F., Asinari P., Chiavazzo E., Prasianakis N. I., Mantzaras J., A lattice Boltzmann model for reactive flows simulations, EuroPhysics Letters, 2012 (accepted), (Impact Factor: 2.752) 2 Asinari P., Ohwada T., Chiavazzo E., Di Rienzo A. F., Link-wise Artificial Compressibility Method, Journal of Computational Physics, 2012 (preprint), (Impact Factor: 2.345) 3 Di Rienzo A. F., Asinari P., Borchiellini R., Mishra S. C., Improved angular discretization and error analysis of the lattice Boltzmann method for solving radiative heat transfer in a participating medium, Int. Jour. Num. Meth. for Heat and Fluid Flow, 2011, 21:640-662, (Impact Factor: 0.53) Proceedings 1 Di Rienzo A. F., Asinari P., Chiavazzo E., Link-wise equilibrium-based lattice Boltzmann method, Proceedings of the XXIX UIT Heat Transfer Conference, 2011 Antonio F. Di Rienzo (Politecnico di Torino) PhD Defence Torino, April 20th 2012 43 / 46
  • 44. Conclusions & Acknowledgements Conference Presentations 1 Di Rienzo A. F., Asinari P., Chiavazzo E., Prasianakis N. I., Mantzaras J., Coupling lattice Boltzmann model with reduced chemical kinetics for combustion simulations, 8th International Conference for Mesoscopic Methods in Engineering and Science (ICMMES), July 4-8 2011, Lyon, France 2 Di Rienzo A. F., Asinari P., Chiavazzo E., Link-wise equilibrium based lattice Boltzmann method, 29th UIT Heat Transfer Conference, June 20-22, 2011, Turin, Italy 3 Di Rienzo A. F., Chiavazzo E., Asinari P., Radiative lattice Boltzmann method applied to combustion simulation and reduced chemical kinetics, 19th International Conference on the Discrete Simulation of Fluid Dynamics (DSFD), July 5-9, Rome, Italy 4 Di Rienzo A. F., Izquierdo S., Asinari P., Mishra S. C., Borchiellini R., The lattice Boltzmann method in solving radiative heat transfer in a participating medium, First International Conference on Computational Methods for Thermal Problems (ThermaComp), September, 8-10, 2009, Naples, Italy Award 1 Best presentation at the Energetics PhD Day, Politecnico di Torino, December 20, 2011 Antonio F. Di Rienzo (Politecnico di Torino) PhD Defence Torino, April 20th 2012 44 / 46
  • 45. Conclusions & Acknowledgements Acknowledgements Dr. John Mantzaras, Dr. Nikolaos Prasianakis Antonio F. Di Rienzo (Politecnico di Torino) PhD Defence Torino, April 20th 2012 45 / 46
  • 46. Conclusions & Acknowledgements Thank you for your attention! Antonio F. Di Rienzo (Politecnico di Torino) PhD Defence Torino, April 20th 2012 46 / 46