betaFeedback

WorksMathematics

Realizability-Based Kinetic-Fluid Domain Decomposition for Multitemperature Gas Mixtures on Unstructured Meshes

Abstract

We present a dynamic domain decomposition method for gas mixtures described by a multi-species BGK model in the regime where collisions between particles of the same species are dominant. In this regime the hydrodynamic limit is a system of Euler equations in which every species keeps its own velocity and temperature, the species being coupled through momentum and energy exchange terms. To decide locally whether this multi-velocity, multi-temperature Euler description is accurate, we extend the moment realizability criterion of Levermore, Morokoff and Nadiga: for each species, a realizability matrix is derived from the Chapman–Enskog expansion of the kinetic model, and its zeroth- and first-order approximations are compared. The indicator involves only macroscopic quantities and is evaluated species by species, so that in the same cell one species may be described by the Euler equations and another one by the kinetic equation. The kinetic solver is therefore used only where, and for the species for which, the indicator detects non-equilibrium effects, such as across shocks or near obstacles in rarefied flows. Both solvers are discretized on unstructured triangular meshes with second-order accuracy in space and time. The kinetic equation is integrated by an asymptotic-preserving IMEX Runge–Kutta scheme, whose limit for vanishing Knudsen number is a consistent discretization of the multi-velocity, multi-temperature Euler system. In the numerical tests the hybrid solution stays close to the kinetic one, while the computational time is reduced by a factor between about three and almost five.

Introduction

Rarefied gas mixtures appear in many applications, from high-altitude atmospheric flows and nuclear engineering to evaporation–condensation problems [41] and plasma physics, where electrons and ions are modelled as different interacting species [11]. In several of these problems the flow is in a transitional regime, as during the re-entry of space vehicles [27] or in micro-electro-mechanical systems (MEMS). At high altitude non-equilibrium effects become important and macroscopic models lose accuracy: the Navier–Stokes equations, for instance, do not give correct shock profiles in the hypersonic regime when rarefaction effects are significant, whereas a kinetic description remains applicable. Near strong shocks or in kinetic boundary layers fluid models are therefore not reliable, and for mixtures a multi-species kinetic description is needed.

The Boltzmann equation for gas mixtures has been studied in several works, see e.g. [12, 13]. In [49] the analysis was extended to granular mixtures, proving well-posedness in the spatially homogeneous case and studying the long-time behaviour. All these models share the same structure: a linear transport operator and one or more stiff, nonlinear collision operators, acting on distribution functions defined in a high-dimensional phase space [28].

The main difficulty in the numerical solution of kinetic equations is the dimension of the phase space, up to seven including time, which makes kinetic solvers much more expensive than fluid solvers. Moreover, kinetic models contain small scales, such as the mean free path. Asymptotic expansions in these small parameters lead to hydrodynamic equations, such as the Euler equations, which describe the average behaviour of the gas. Solving the kinetic equation everywhere in a near-continuum regime is very expensive, and wherever the hydrodynamic model is valid it is much cheaper to solve it instead.

2020 Mathematics Subject Classification. Primary: 76P05, 82C40; Secondary: 65M08, 65M55.

Key words and phrases. Gas mixtures, kinetic–fluid coupling, moment realizability, dynamic domain decomposition, asymptotic-preserving schemes, IMEX Runge–Kutta methods, unstructured meshes, finite volume methods.

Hybrid methods, or dynamic domain decomposition methods [28, 15, 31, 50, 17, 21, 20], are based on this observation: at every time step the spatial domain is divided into regions where different models are solved, according to the local degree of non-equilibrium. Compared with a purely hydrodynamic computation the solution is more accurate, and compared with a purely kinetic one it is cheaper. Other kinetic–fluid approaches for gas mixtures include the particle-based micro–macro method of [24]. In [25] a multi-species kinetic model is coupled with a multi-fluid model, and in [22] a multi-species Vlasov–BGK model is coupled with its single-velocity, single-temperature hydrodynamic limit through buffer regions; in both works the kinetic region is fixed in time. Here, instead, the partition is dynamic: it is updated at every time step with a realizability-based indicator, so that the kinetic and fluid regions of each species evolve in time with the flow.

In collision-dominated regions the collisions drive the gas quickly towards local equilibrium, and the collision operator is stiff. Explicit schemes then require time steps of the order of the relaxation time. Asymptotic-preserving (AP) schemes [23, 38, 16, 43, 39, 42, 44, 35, 26, 30, 29, 41, 37, 7, 8] remove this restriction: they are stable for time steps independent of the Knudsen number and reduce, in the limit, to a consistent discretization of the hydrodynamic equations [40]. We use this property for the kinetic solver.

In this work we consider mixtures in which collisions between particles of the same species are much more frequent than collisions between different species. This happens, for instance, in mixtures of light and heavy gases, where the energy exchange between species is slow [33, 4]. Each species then relaxes to its own Maxwellian, and the hydrodynamic limit is a system of Euler equations with one velocity and one temperature per species [4]. We propose a hybrid method that couples this multi-velocity, multi-temperature Euler system with the multi-species BGK model and switches between the two automatically. To our knowledge, this is the first realizability-based dynamic spatial domain decomposition method for a kinetic mixture model whose hydrodynamic limit has a velocity and a temperature for each species. Since the fluid model keeps these quantities separate, the choice between fluid and kinetic description can be made species by species.

The switching criterion is based on a moment realizability matrix that we derive for this regime. The computations are carried out for the BGK model of [5, 4]. The collision model enters the indicator only through the Chapman–Enskog correction terms, so the same construction can be used with the full Boltzmann operator for mixtures once these corrections are known. Since the indicator is based on the Chapman–Enskog expansion, it can also be extended to more than two levels, as done for a single gas in [17, 20]. In [17] the hierarchy consists of the Euler equations, the ellipsoidal statistical BGK (ES-BGK) model [1] and the full Boltzmann equation on Cartesian meshes. In [20], on unstructured meshes, the intermediate level is given by the Navier–Stokes equations, which correspond to the ES-BGK model at first order in the Chapman–Enskog expansion: on unstructured meshes the cost is dominated by transport, and the ES-BGK model does not give a clear computational advantage.

The coupling is non-intrusive: the two solvers only exchange moments and Maxwellians, and they only need to use the same Runge–Kutta stages. The kinetic solver is used only in the cells, and for the species, selected by the indicator. We implement the method on unstructured triangular meshes with a second-order finite volume reconstruction, so that problems with non-trivial geometries can be treated.

The opposite situation, in which all collisions occur on the same time scale and the mixture relaxes to a single velocity and a single temperature, leads to a different hydrodynamic limit and requires a different indicator. It is considered in the companion paper [19].

The paper is organized as follows. § 2 presents the kinetic and macroscopic models. In § 3 we derive the moment realizability matrices and the regime indicators. § 4 describes the numerical schemes, the AP property and the coupling between kinetic and fluid solvers. Numerical tests are presented in § 5, and the computational cost is discussed in § 6. § 7 contains some conclusions.

Multi-temperature, multi-velocity BGK model for gas mixtures

We recall the kinetic model considered in this work, following [4, 5]. We consider a mixture of NN monatomic gases whose particles undergo elastic collisions. Species ii is described by its distribution function fi(t,x,v)f_i(t,x,v), where x∈R3x \in\mathbb{R}^3 is the position, v∈R3v \in\mathbb{R}^3 the velocity and t≥0t \ge0 the time. The evolution is given by the Boltzmann equations

∂fi∂t+v⋅∇xfi=Qi[f],Qi[f]=∑k=1NQi,k[fi,fk],fi(0,x,v)=fi,0(x,v),(1)\begin{aligned} \frac{\partial f_i}{\partial t}+v\cdot\nabla_x f_i&=Q_i[f],\qquad Q_i[f]=\sum_{k=1}^{N}Q_{i,k}[f_i,f_k],\\ f_i(0,x,v)&=f_{i,0}(x,v), \tag*{(1)} \end{aligned}

where QiQ_i is the collision operator of species ii and the binary operators read

Qi,k[fi,fk]=∫R3∫S2gi,k(∣y∣,y^⋅Ω)[fi(v′)fk(w′)−fi(v)fk(w)] dΩ dw.(2)Q_{i,k}[f_i,f_k]=\int_{\mathbb{R}^3}\int_{\mathbb{S}^2}g_{i,k}(|y|,\hat{y}\cdot\Omega)\left[f_i(v')f_k(w')-f_i(v)f_k(w)\right]\,d\Omega\,dw. \tag*{(2)}

Here y=v−wy=v-w is the relative velocity, v′v' and w′w' are the pre-collisional velocities, and the collision kernel gi,kg_{i,k} depends on ∣y∣|y| and on μ=y^⋅Ω\mu=\hat{y}\cdot\Omega, with y^=y/∣y∣\hat{y}=y/|y|. The number density, mean velocity and temperature of species ii, whose particles have mass mim_i, are

ni=∫R3fi dv,ui=1ni∫R3vfi dv,Ti=mi3ni∫R3∣v−ui∣2fi dv.(3)n_i=\int_{\mathbb{R}^3}f_i\,dv,\qquad u_i=\frac{1}{n_i}\int_{\mathbb{R}^3}v f_i\,dv,\qquad T_i=\frac{m_i}{3n_i}\int_{\mathbb{R}^3}|v-u_i|^2f_i\,dv. \tag*{(3)}

The collision equilibria of (1) are local Maxwellians with a common velocity uu and a common temperature TT,

fieq=ni(mi2πT)3/2exp⁡(−mi2T∣v−u∣2),(4)f_i^{\mathrm{eq}}=n_i\left(\frac{m_i}{2\pi T}\right)^{3/2}\exp\left(-\frac{m_i}{2T}|v-u|^2\right), \tag*{(4)}

where

n=∑i=1Nni,ρ=∑i=1Nmini,u=1ρ∑i=1Nminiui,n=\sum_{i=1}^{N}n_i,\qquad\rho=\sum_{i=1}^{N}m_i n_i,\qquad u=\frac{1}{\rho}\sum_{i=1}^{N}m_i n_i u_i,
T=13n∑i=1Nmi∫R3∣v−u∣2fi dv=1n∑i=1NniTi+13n∑i=1Nmini∣ui−u∣2.(5)T=\frac{1}{3n}\sum_{i=1}^{N}m_i\int_{\mathbb{R}^3}|v-u|^2f_i\,dv =\frac{1}{n}\sum_{i=1}^{N}n_iT_i+\frac{1}{3n}\sum_{i=1}^{N}m_i n_i|u_i-u|^2. \tag*{(5)}

In [4] the authors consider the BGK model of [5], in which the collision operators are replaced by relaxation operators,

Qi[f]=∑k=1Nνik(niMik−fi),(6)Q_i[f]=\sum_{k=1}^{N}\nu_{ik}\left(n_iM_{ik}-f_i\right), \tag*{(6)}

where νik\nu_{ik} are collision frequencies, independent of vv, and MikM_{ik} are Maxwellians with mass mim_i, velocity uiku_{ik} and temperature TikT_{ik},

Mik=(mi2πTik)3/2exp⁡(−mi2Tik∣v−uik∣2).(7)M_{ik}=\left(\frac{m_i}{2\pi T_{ik}}\right)^{3/2}\exp\left(-\frac{m_i}{2T_{ik}}|v-u_{ik}|^2\right). \tag*{(7)}

The parameters uiku_{ik}, TikT_{ik} are fixed by requiring that the exchange rates of momentum and energy coincide with those of the Boltzmann operator for Maxwellian interactions; this guarantees the conservation of total momentum and total energy [4]. One obtains

uik=(1−aik)ui+aikuk,Tik=(1−bik)Ti+bikTk+γik∣ui−uk∣2.(8)u_{ik}=(1-a_{ik})u_i+a_{ik}u_k,\qquad T_{ik}=(1-b_{ik})T_i+b_{ik}T_k+\gamma_{ik}|u_i-u_k|^2. \tag*{(8)}

with

aik=λiknkmkνik(mi+mk),bik=2aikmimi+mk,γik=miaik3(2mkmi+mk−aik),(9)a_{ik}=\frac{\lambda_{ik}n_km_k}{\nu_{ik}(m_i+m_k)},\qquad b_{ik}=\frac{2a_{ik}m_i}{m_i+m_k},\qquad\gamma_{ik}=\frac{m_ia_{ik}}{3}\left(\frac{2m_k}{m_i+m_k}-a_{ik}\right), \tag*{(9)}

References

  1. [1]Andries, P., Le Tallec, P., Perlat, J.-P., and Perthame, B. The Gaussian-BGK model of Boltzmann equation with small Prandtl number. Eur. J. Mech. B Fluids 19, 6 (2000), 813–830.DOI
  2. [2]Ascher, U. M., Ruuth, S. J., and Spiteri, R. J. Implicit-explicit Runge-Kutta methods for time-dependent partial differential equations. Applied Numerical Mathematics 25, 2 (1997), 151–167. Special Issue on Time Integration.DOI
  3. [3]Bisi, M., Boscheri, W., Dimarco, G., Groppi, M., and Martalò, G. A new mixed Boltzmann-BGK model for mixtures solved with an IMEX finite volume scheme on unstructured meshes. Applied Mathematics and Computation 433 (2022), 127416.DOI
  4. [4]Bisi, M., Groppi, M., and Martalò, G. Macroscopic equations for inert gas mixtures in different hydrodynamic regimes. Journal of Physics A: Mathematical and Theoretical 54, 8 (2021), 085201.DOI
  5. [5]Bobylev, A., Bisi, M., Groppi, M., Spiga, G., and Potapenko, I. A general consistent BGK model for gas mixtures. Kinetic & Related Models 11 (2018), 1377–1393.DOI
  6. [6]Boscarino, S., Cho, S. Y., and Russo, G. A local velocity grid conservative semi-Lagrangian schemes for BGK model. Journal of Computational Physics 460 (2022), 111178.DOI
  7. [7]Boscarino, S., Cho, S.-Y., Russo, G., and Yun, S.-B. High Order Conservative Semi-Lagrangian Scheme for the BGK Model of the Boltzmann Equation. Commun. Comput. Phys. 29 (2021), 1–56.arxiv.org/abs/1905.03660
  8. [8]Boscarino, S., Pareschi, L., and Russo, G. Implicit-Explicit Methods for Evolutionary Partial Differential Equations, vol. 24 of Mathematical Modeling and Computation. SIAM, 2024.DOI
  9. [9]Boscheri, W., and Dimarco, G. High order central WENO-Implicit-Explicit Runge Kutta schemes for the BGK model on general polygonal meshes. Journal of Computational Physics 422 (2020), 109766.
  10. [10]Boscheri, W., and Dimarco, G. High order finite volume schemes with IMEX time stepping for the Boltzmann model on unstructured meshes. Computer Methods in Applied Mechanics and Engineering 387 (2021), 114180.
  11. [11]Bott, A. F. A., Cowley, S. C., and Schekochihin, A. A. Kinetic stability of Chapman–Enskog plasmas. J. Plasma Phys. 90, 2 (2024), 975900207.arxiv.org/abs/2310.17754
  12. [12]Briant, M. Stability of global equilibrium for the multi-species Boltzmann equation in L∞ settings. Discrete Contin. Dyn. Syst. 36, 12 (2016), 6669–6688.DOI
  13. [13]Briant, M., and Daus, E. S. The Boltzmann Equation for a Multi-species Mixture Close to Global Equilibrium. Arch. Ration. Mech. Anal. 222, 3 (2016), 1367–1443.DOI
  14. [14]Buffard, T., and Clain, S. Monoslope and multislope MUSCL methods for unstructured meshes. Journal of Computational Physics 229, 10 (2010), 3745–3776.DOI
  15. [15]Burt, J. M., and Boyd, I. D. A hybrid particle approach for continuum and rarefied flow simulation. J. Comput. Phys. 228, 2 (2009), 460–475.DOI
  16. [16]Caflisch, R. E., Jin, S., and Russo, G. Uniformly Accurate Schemes for Hyperbolic Systems with Relaxation. SIAM J. Numer. Anal. 34, 1 (1997), 246–281.DOI
  17. [17]Caparello, D., Pareschi, L., and Rey, T. Hierarchical dynamic domain decomposition for the multiscale Boltzmann equation. arXiv e-prints (2025), arXiv:2505.03360.arxiv.org/abs/2505.03360
  18. [18]Caparello, D., Pareschi, L., and Rey, T. High-Order Asymptotic-Preserving IMEX schemes for an ES-BGK model for Gas Mixtures. arXiv preprint arXiv:2511.22304 (2025).arxiv.org/abs/2511.22304
  19. [19]Caparello, D., Pareschi, L., and Rey, T. A Realizability-Based Kinetic-Fluid Domain Decomposition Method for Single-Temperature Gas Mixtures. Companion manuscript, 2026.
  20. [20]Caparello, D., Pareschi, L., and Rey, T. An IMEX multi-level domain-decomposition method for the Boltzmann equation on unstructured meshes. arXiv preprint arXiv:2610.21740 (2026).arxiv.org/abs/2610.21740
  21. [21]Caparello, D., and Tenna, T. A Coupled IMEX Domain Decomposition Method for High-Order Time Integration of the ES-BGK Model of the Boltzmann Equation. arXiv preprint arXiv:2512.03586 (2025).arxiv.org/abs/2512.03586
  22. [22]Chuna, T., Sagert, I., Muriullo, M. S., and Haack, J. R. Multi-species kinetic-fluid coupling for high-energy density simulations. Journal of Computational Physics 505 (2024), 112908.DOI
  23. [23]Coron, F., and Perthame, B. Numerical passage from kinetic to fluid equations. SIAM J. Numer. Anal. 28 (1991), 26–42.DOI
  24. [24]Crestetto, A., Klingenberg, C., and Pirner, M. Kinetic/Fluid Micro-Macro Numerical Scheme for a Two Component Gas Mixture. Multiscale Model. Simul. 18, 3 (2020), 970–998.arxiv.org/abs/1806.11485
  25. [25]Datta, I. A. M., and Shumlak, U. Computationally efficient high-fidelity plasma simulations by coupling multi-species kinetic and multi-fluid models on decomposed domains. Journal of Computational Physics 483 (2023), 112073.DOI
  26. [26]Diaz, M. A., Chen, M.-H., and Yang, J.-Y. High-Order Conservative Asymptotic-Preserving Schemes for Modeling Rarefied Gas Dynamical Flows with Boltzmann-BGK Equation. Commun. Comput. Phys. 18, 4 (2015), 1012–1049.DOI
  27. [27]Dillmann, R., and Corliss, J. Overview of the Mars Sample Return Earth Entry Vehicle. In Proceedings of the 6th International Planetary Probe Workshop (Atlanta, GA, 2008). NASA Technical Reports Server, Document ID 20080023907.
  28. [28]Dimarco, G., and Pareschi, L. Numerical methods for kinetic equations. Acta Numer. 23 (2014), 369–520.DOI
  29. [29]Filbet, F., and Jin, S. A class of asymptotic-preserving schemes for kinetic equations and related problems with stiff sources. J. Comput. Phys. 229, 20 (2010), 7625–7648.DOI
  30. [30]Filbet, F., and Jin, S. An Asymptotic Preserving Scheme for the ES-BGK Model of the Boltzmann Equation. J. Sci. Comput. 46 (2011), 204–224.DOI
  31. [31]Filbet, F., and Rey, T. A Hierarchy of Hybrid Numerical Method for Multi-Scale Kinetic Equations. SIAM J. Sci. Comput. 37, 3 (2015), A1218–A1247.arxiv.org/abs/1402.6304
  32. [32]Filbet, F., and Xiong, T. A hybrid discontinuous Galerkin scheme for multi-scale kinetic equations. J. Comput. Phys. 372 (2018), 841–863.DOI
  33. [33]Galkin, V., and Makashev, N. Kinetic derivation of the gas-dynamic equations for multicomponent mixtures of light and heavy particles. Fluid dynamics 29, 1 (1994), 140–155.DOI
  34. [34]Gamba, I., and Tharkabhushanam, S. Spectral - Lagrangian methods for Collisional Models of Non - Equilibrium Statistical States. Journal of Computational Physics 228 (2009), 2012–2036.arxiv.org/abs/0710.5308
  35. [35]Gosse, L., and Toscani, G. An asymptotic-preserving well-balanced scheme for the hyperbolic heat equations. Cr. Math. 334, 4 (2002), 337–342.DOI
  36. [36]Horn, R. A., and Johnson, C. R. Matrix Analysis, 2 ed. Cambridge University Press, 2012.
  37. [37]Hu, J., and Wang, L. An asymptotic-preserving scheme for the semiconductor Boltzmann equation toward the energy-transport limit. J. Comput. Phys. 281 (2015), 806–824.arxiv.org/abs/1309.1330
  38. [38]Jin, S. Runge-Kutta Methods for Hyperbolic Conservation Laws with Stiff Relaxation Terms. J. Comput. Phys. 122, 1 (1995), 51–67.DOI
  39. [39]Jin, S. Efficient Asymptotic-Preserving (AP) Schemes For Some Multiscale Kinetic Equations. SIAM J. Sci. Comput. 21, 2 (1999), 441–454.DOI
  40. [40]Jin, S. Asymptotic preserving (AP) schemes for multiscale kinetic and hyperbolic equations: a review. Riv. Mat. Univ. Parma (N.S.) 3, 2 (2012), 177–216.
  41. [41]Jin, S., and Li, Q. A BGK-penalization-based asymptotic-preserving scheme for the multispecies Boltzmann equation. Numer. Methods Partial Differential Equations 29, 3 (2013), 1056–1080.DOI
  42. [42]Jin, S., and Pareschi, L. Asymptotic-Preserving (AP) Schemes for Multiscale Kinetic Equations: a Unified Approach. In Hyperbolic Problems: Theory, Numerics, Applications (Basel, 2001), H. Freistühler and G. Warnecke, Eds., Birkhäuser Basel, pp. 573–582.DOI
  43. [43]Jin, S., Pareschi, L., and Toscani, G. Diffusive Relaxation Schemes For Multiscale Discrete-Velocity Kinetic Equations. SIAM J. Numer. Anal. 35 (1998), 2405–2439.DOI
  44. [44]Klar, A. An Asymptotic-Induced Scheme for Nonstationary Transport Equations in the Diffusive Limit. SIAM J. Numer. Anal. 35, 3 (1998), 1073–1094.DOI
  45. [45]Levermore, C. D., Morokoff, W. J., and Nadiga, B. T. Moment realizability and the validity of the Navier-Stokes equations for rarefied gas dynamics. Phys. Fluids 10, 12 (1998), 3214–3226.DOI
  46. [46]Li, W., Song, P., and Wang, Y. A Hybrid Moment Method for Multi-scale Kinetic Equations Based on Maximum Entropy Principle. J. Sci. Comput. 89, 2 (2021), 35.arxiv.org/abs/2004.05893
  47. [47]Mieussens, L. Discrete Velocity Model and Implicit Scheme for the BGK Equation of Rarefied Gas Dynamics. Math. Models Methods Appl. Sci. 10, 8 (2000), 1121–1149.DOI
  48. [48]Mitrinovic, D. S. Analytic Inequalities, vol. 1. Springer, 1970.
  49. [49]Rey, T., and Tenna, T. The Boltzmann Equation for a Multi-Species Inelastic Mixture. J. Stat. Phys. 192, 11 (2025), 151.arxiv.org/abs/2410.17027
  50. [50]Tiwari, S. Coupling of the Boltzmann and Euler equations with automatic domain decomposition. J. Comput. Phys. 144, 2 (1998), 710–726.DOI
  51. [51]Tiwari, S. Application of Moment Realizability Criteria for the Coupling of the Boltzmann and Euler Equations. Transport Theor. Stat. 29, 7 (2000), 759–783.DOI
  52. [52]Tyrtyshnikov, E. E. A Brief Introduction to Numerical Analysis. Birkhäuser, Boston, MA, 1997. Lecture 4, pp. 31–37.DOI
  53. [53]Xiong, T., and Qiu, J.-M. A hierarchical uniformly high order DG-IMEX scheme for the 1D BGK equation. J. Comput. Phys. 336 (2017), 164–191.DOI

Paper details

vlnc:W_mau6bq2f
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Mathematics

a generative AI tool

View statement
Looking for related papers…
Contents