Recent advances in kinetic theory for mixtures of polyatomic gases · M. Bisi, Parma Kinetic theory...
Transcript of Recent advances in kinetic theory for mixtures of polyatomic gases · M. Bisi, Parma Kinetic theory...
Recent advances in kinetic theoryfor mixtures of polyatomic gases
Marzia Bisi
Parma University, Italy
Conference “Problems on Kinetic Theory and PDE’s”
Novi Sad (Serbia), September 25–27, 2014
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 1 / 25
Summary
Kinetic Boltzmann model for (inert or reactive) polyatomic gasmixtures
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 2 / 25
Summary
Kinetic Boltzmann model for (inert or reactive) polyatomic gasmixtures
BGK relaxation model (joint work with M.J. Caceres)
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 2 / 25
Summary
Kinetic Boltzmann model for (inert or reactive) polyatomic gasmixtures
BGK relaxation model (joint work with M.J. Caceres)
Hydrodynamic limit leading to incompressible Navier–Stokesequations (based on a joint work with L. Desvillettes)
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 2 / 25
Summary
Kinetic Boltzmann model for (inert or reactive) polyatomic gasmixtures
BGK relaxation model (joint work with M.J. Caceres)
Hydrodynamic limit leading to incompressible Navier–Stokesequations (based on a joint work with L. Desvillettes)
Work in progress and open problems
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 2 / 25
Kinetic Boltzmann approach to polyatomic gases
Motivation: It is well known that gas mixtures involved in physicalapplications are usually composed also of polyatomic species (forinstance in simple dissociation and recombination processes, or inthe evolution of powders in the atmosphere)
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 3 / 25
Kinetic Boltzmann approach to polyatomic gases
Motivation: It is well known that gas mixtures involved in physicalapplications are usually composed also of polyatomic species (forinstance in simple dissociation and recombination processes, or inthe evolution of powders in the atmosphere)
∗ In kinetic approaches, each gas is endowed with a suitableinternal energy variable to mimic non–translational degrees offreedom
Groppi, Spiga, J. Math. Chem. (1999): discrete internal energy levels
Desvillettes, Monaco, Salvarani, Europ. J. Mech. B/Fluids (2005):continuous internal energy variable
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 3 / 25
Kinetic Boltzmann approach to polyatomic gases
Motivation: It is well known that gas mixtures involved in physicalapplications are usually composed also of polyatomic species (forinstance in simple dissociation and recombination processes, or inthe evolution of powders in the atmosphere)
∗ In kinetic approaches, each gas is endowed with a suitableinternal energy variable to mimic non–translational degrees offreedom
Groppi, Spiga, J. Math. Chem. (1999): discrete internal energy levels
Desvillettes, Monaco, Salvarani, Europ. J. Mech. B/Fluids (2005):continuous internal energy variable
Our physical frame:
Mixture of M polyatomic gases Gs , s = 1, . . . ,M
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 3 / 25
Kinetic Boltzmann approach to polyatomic gases
Motivation: It is well known that gas mixtures involved in physicalapplications are usually composed also of polyatomic species (forinstance in simple dissociation and recombination processes, or inthe evolution of powders in the atmosphere)
∗ In kinetic approaches, each gas is endowed with a suitableinternal energy variable to mimic non–translational degrees offreedom
Groppi, Spiga, J. Math. Chem. (1999): discrete internal energy levels
Desvillettes, Monaco, Salvarani, Europ. J. Mech. B/Fluids (2005):continuous internal energy variable
Our physical frame:
Mixture of M polyatomic gases Gs , s = 1, . . . ,M
Each gas Gs is considered as a mixture of Q monatomiccomponents A i , i = s, s + M, s + 2M, s + M(Q − 1),each one characterized by a different internal energy E i
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 3 / 25
In the frame of each species, energies are monotonicallyincreasing with their index:
∀ i, j ≡ s , i < j ⇒ E i < E j
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 4 / 25
In the frame of each species, energies are monotonicallyincreasing with their index:
∀ i, j ≡ s , i < j ⇒ E i < E j
Besides classical elastic scattering, particles may undergoalso inelastic transitions in which they change their internalenergy level
A i + A j⇋ Ah + Ak h ≡ i, k ≡ j
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 4 / 25
In the frame of each species, energies are monotonicallyincreasing with their index:
∀ i, j ≡ s , i < j ⇒ E i < E j
Besides classical elastic scattering, particles may undergoalso inelastic transitions in which they change their internalenergy level
A i + A j⇋ Ah + Ak h ≡ i, k ≡ j
Boltzmann equations for distribution functions of single components
∂f i
∂t+ v · ∇xf i =
∑
(j, h, k )∈Di
"K ijhk
i [f ](v,w, n′)dwdn′ 1≤ i≤QM
K ijhki [f ] = Θ
(
g2 −2∆Ehk
ij
µij
)
gσhkij (g, n · n′)
[
fh(
vhkij
)
fk(
whkij
)
− f i(v)f j(w)]
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 4 / 25
In the frame of each species, energies are monotonicallyincreasing with their index:
∀ i, j ≡ s , i < j ⇒ E i < E j
Besides classical elastic scattering, particles may undergoalso inelastic transitions in which they change their internalenergy level
A i + A j⇋ Ah + Ak h ≡ i, k ≡ j
Boltzmann equations for distribution functions of single components
∂f i
∂t+ v · ∇xf i =
∑
(j, h, k )∈Di
"K ijhk
i [f ](v,w, n′)dwdn′ 1≤ i≤QM
K ijhki [f ] = Θ
(
g2 −2∆Ehk
ij
µij
)
gσhkij (g, n · n′)
[
fh(
vhkij
)
fk(
whkij
)
− f i(v)f j(w)]
Here the set Di includes all possible collisions,vhk
ij , whkij are post–collision velocities, g = v −w = g n,
Θ is the Heaviside function providing a threshold for all endothermicinteractions in which ∆Ehk
ij = Eh + Ek − E i − E j > 0
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 4 / 25
Collision invariants:number density of each gas Ns =
∑
i≡s
ni , s = 1, . . . ,M,
global velocity u, global energy32
nKT +QM∑
i=1
E ini
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 5 / 25
Collision invariants:number density of each gas Ns =
∑
i≡s
ni , s = 1, . . . ,M,
global velocity u, global energy32
nKT +QM∑
i=1
E ini
Collision equilibria of the Boltzmann equations:
f iM(v) = ni
(
ms
2πKT
)3/2
exp
[
− ms
2KT|v − u|2
]
∀i ≡ s, ∀s = 1, . . . ,M,
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 5 / 25
Collision invariants:number density of each gas Ns =
∑
i≡s
ni , s = 1, . . . ,M,
global velocity u, global energy32
nKT +QM∑
i=1
E ini
Collision equilibria of the Boltzmann equations:
f iM(v) = ni
(
ms
2πKT
)3/2
exp
[
− ms
2KT|v − u|2
]
∀i ≡ s, ∀s = 1, . . . ,M,
with equilibrium number densities related by the constraints
ni = Ns ψ(E i ,T) i ≡ swhere
ψ(E i ,T) =exp
(
−E i−Es
KT
)
∑
i≡s exp(
−E i−Es
KT
) =exp
(
−E i−Es
KT
)
Zs(T)
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 5 / 25
Collision invariants:number density of each gas Ns =
∑
i≡s
ni , s = 1, . . . ,M,
global velocity u, global energy32
nKT +QM∑
i=1
E ini
Collision equilibria of the Boltzmann equations:
f iM(v) = ni
(
ms
2πKT
)3/2
exp
[
− ms
2KT|v − u|2
]
∀i ≡ s, ∀s = 1, . . . ,M,
with equilibrium number densities related by the constraints
ni = Ns ψ(E i ,T) i ≡ swhere
ψ(E i ,T) =exp
(
−E i−Es
KT
)
∑
i≡s exp(
−E i−Es
KT
) =exp
(
−E i−Es
KT
)
Zs(T)
Remark: ψ(E i ,T) represents the fraction of particles Gs (s ≡ i) belonging to the
component A i in any equilibrium configuration; for any i, j with i ≡ j and i < j, we
have ψ(E i ,T) > ψ(E j , T)
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 5 / 25
Generalization to chemically reactive frames (4 species)
Besides elastic scattering and inelastic transitions (with transfer ofinternal energy), particles may undergo the binary and reversiblechemical reaction G1 + G2 ⇋ G3 + G4, hence, for singlecomponents,
A i + A j⇋ Ah + Ak i ≡ 1, j ≡ 2, h ≡ 3, k ≡ 4
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 6 / 25
Generalization to chemically reactive frames (4 species)
Besides elastic scattering and inelastic transitions (with transfer ofinternal energy), particles may undergo the binary and reversiblechemical reaction G1 + G2 ⇋ G3 + G4, hence, for singlecomponents,
A i + A j⇋ Ah + Ak i ≡ 1, j ≡ 2, h ≡ 3, k ≡ 4
Basic properties:
For reactive encounters
K ijhki [f ] = Θ
(
g2 −2∆Ehk
ij
µij
)
gσhkij (g, n · n′)
[(
µij
µhk
)3fh
(
vhkij
)
fk(
whkij
)
− f i(v)f j(w)]
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 6 / 25
Generalization to chemically reactive frames (4 species)
Besides elastic scattering and inelastic transitions (with transfer ofinternal energy), particles may undergo the binary and reversiblechemical reaction G1 + G2 ⇋ G3 + G4, hence, for singlecomponents,
A i + A j⇋ Ah + Ak i ≡ 1, j ≡ 2, h ≡ 3, k ≡ 4
Basic properties:
For reactive encounters
K ijhki [f ] = Θ
(
g2 −2∆Ehk
ij
µij
)
gσhkij (g, n · n′)
[(
µij
µhk
)3fh
(
vhkij
)
fk(
whkij
)
− f i(v)f j(w)]
Collision invariants: N1 + N3, N1 + N4, N2 + N4,global momentum, and total energy
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 6 / 25
Generalization to chemically reactive frames (4 species)
Besides elastic scattering and inelastic transitions (with transfer ofinternal energy), particles may undergo the binary and reversiblechemical reaction G1 + G2 ⇋ G3 + G4, hence, for singlecomponents,
A i + A j⇋ Ah + Ak i ≡ 1, j ≡ 2, h ≡ 3, k ≡ 4
Basic properties:
For reactive encounters
K ijhki [f ] = Θ
(
g2 −2∆Ehk
ij
µij
)
gσhkij (g, n · n′)
[(
µij
µhk
)3fh
(
vhkij
)
fk(
whkij
)
− f i(v)f j(w)]
Collision invariants: N1 + N3, N1 + N4, N2 + N4,global momentum, and total energy
Collision equilibria: Maxwellian distributions f i = f iM(ni ,u,T)
with ni = Ns ψ(E i ,T) plus the mass action law of chemistry
N1N2
N3N4=
(
µ12
µ34
)3/2 Z1(T)Z2(T)
Z3(T)Z4(T)e
∆E3412
KT ∆E3412 = E3 + E4 − E1 − E2 > 0
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 6 / 25
BGK approximation for polyatomic INERT mixtures
∂f i
∂t+ v · ∇xf i = νi(Mi − f i) i = 1, . . . ,QM
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 7 / 25
BGK approximation for polyatomic INERT mixtures
∂f i
∂t+ v · ∇xf i = νi(Mi − f i) i = 1, . . . ,QM
where attractorsMi =Mi(ni , u, T) take the form
Mi(v) = ni(
mi
2πKT
)3/2
exp
[
− mi
2KT|v − u|2
]
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 7 / 25
BGK approximation for polyatomic INERT mixtures
∂f i
∂t+ v · ∇xf i = νi(Mi − f i) i = 1, . . . ,QM
where attractorsMi =Mi(ni , u, T) take the form
Mi(v) = ni(
mi
2πKT
)3/2
exp
[
− mi
2KT|v − u|2
]
with densities ni bound together as
ni = Ns ψ(E i , T)
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 7 / 25
BGK approximation for polyatomic INERT mixtures
∂f i
∂t+ v · ∇xf i = νi(Mi − f i) i = 1, . . . ,QM
where attractorsMi =Mi(ni , u, T) take the form
Mi(v) = ni(
mi
2πKT
)3/2
exp
[
− mi
2KT|v − u|2
]
with densities ni bound together as
ni = Ns ψ(E i , T)
Key points:Only one relaxation operator for each component[Andries, Aoki, Perthame, J. Stat. Phys. (2002)]
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 7 / 25
BGK approximation for polyatomic INERT mixtures
∂f i
∂t+ v · ∇xf i = νi(Mi − f i) i = 1, . . . ,QM
where attractorsMi =Mi(ni , u, T) take the form
Mi(v) = ni(
mi
2πKT
)3/2
exp
[
− mi
2KT|v − u|2
]
with densities ni bound together as
ni = Ns ψ(E i , T)
Key points:Only one relaxation operator for each component[Andries, Aoki, Perthame, J. Stat. Phys. (2002)]AttractorsMi fulfill the equilibrium conditions[Bisi, Groppi, Spiga, Proceedings RGD26 (2009)]
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 7 / 25
BGK approximation for polyatomic INERT mixtures
∂f i
∂t+ v · ∇xf i = νi(Mi − f i) i = 1, . . . ,QM
where attractorsMi =Mi(ni , u, T) take the form
Mi(v) = ni(
mi
2πKT
)3/2
exp
[
− mi
2KT|v − u|2
]
with densities ni bound together as
ni = Ns ψ(E i , T)
Key points:Only one relaxation operator for each component[Andries, Aoki, Perthame, J. Stat. Phys. (2002)]AttractorsMi fulfill the equilibrium conditions[Bisi, Groppi, Spiga, Proceedings RGD26 (2009)]Ns , u, T are M + 4 independent free parameters to beproperly determined as functions of the actual macroscopicfields ni, ui , T i
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 7 / 25
Strategy: we impose that the BGK model preserves the correctcollision invariants
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 8 / 25
Strategy: we impose that the BGK model preserves the correctcollision invariants
∑
i≡s
νi∫
(Mi − f i)dv = 0 s = 1, . . . ,M (a)
M∑
s=1
∑
i≡s
νi∫
msv(Mi − f i)dv = 0 (b)
M∑
s=1
∑
i≡s
νi∫ (
12
ms |v|2 + E i)
(Mi − f i)dv = 0 (c)
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 8 / 25
Strategy: we impose that the BGK model preserves the correctcollision invariants
∑
i≡s
νi∫
(Mi − f i)dv = 0 s = 1, . . . ,M (a)
M∑
s=1
∑
i≡s
νi∫
msv(Mi − f i)dv = 0 (b)
M∑
s=1
∑
i≡s
νi∫ (
12
ms |v|2 + E i)
(Mi − f i)dv = 0 (c)
• For any s = 1, . . . ,M, condition (a) provides∑
i≡s
νini =∑
i≡s
νini
from which, bearing in mind the constraint ni = Ns ψ(E i , T) we get
Ns =
∑
i≡s
νini
/
∑
i≡s
νiψ(E i , T)
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 8 / 25
Strategy: we impose that the BGK model preserves the correctcollision invariants
∑
i≡s
νi∫
(Mi − f i)dv = 0 s = 1, . . . ,M (a)
M∑
s=1
∑
i≡s
νi∫
msv(Mi − f i)dv = 0 (b)
M∑
s=1
∑
i≡s
νi∫ (
12
ms |v|2 + E i)
(Mi − f i)dv = 0 (c)
• For any s = 1, . . . ,M, condition (a) provides∑
i≡s
νini =∑
i≡s
νini
from which, bearing in mind the constraint ni = Ns ψ(E i , T) we get
Ns =
∑
i≡s
νini
/
∑
i≡s
νiψ(E i , T)
• Momentum conservation (b) yields u =
M∑
s=1
ms∑
i≡s
νiniui
/
M∑
s=1
ms∑
i≡s
νini
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 8 / 25
• Energy conservation (c) provides
M∑
s=1
∑
i≡s
νi[
12
ms ni |u|2 +32
niKT + E ini − 12
msni |ui |2 − 32
niKT i − E ini]
= 0
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 9 / 25
• Energy conservation (c) provides
M∑
s=1
∑
i≡s
νi[
12
ms ni |u|2 +32
niKT + E ini − 12
msni |ui |2 − 32
niKT i − E ini]
= 0
that, recalling the explicit expressions for ni, u, may be written as atranscendental equation for T :
F(T) = Λ where F(T) =M∑
s=1
∑
j≡s
νjnj
32
KT +
∑
i≡s νiE iψ(E i , T)
∑
j≡s νjψ(E j , T)
and Λ is a known explicit function of the actual macroscopic fieldsthat turns out to fulfill Λ ≥ ∑M
s=1
(
∑
i≡s νini
)
Es
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 9 / 25
Lemma: For any Λ, the equation F(T) = Λ has a unique positivesolution
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 10 / 25
Lemma: For any Λ, the equation F(T) = Λ has a unique positivesolution
Steps of the proof:by direct computations, we check that F(T) is a monotonicallyincreasing function of its argument;
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 10 / 25
Lemma: For any Λ, the equation F(T) = Λ has a unique positivesolution
Steps of the proof:by direct computations, we check that F(T) is a monotonicallyincreasing function of its argument;recalling that, for each gas Gs , energy levels are such thatEs < Es+M < Es+2M < · · · < Es+QM, we get
∑
i≡s νiE iψ(E i , T)
∑
j≡s νjψ(E j , T)
=νsEs +
∑
i≡si,sνiE i exp
(
− E i−Es
KT
)
νs +∑
i≡si,sνi exp
(
− E i−Es
KT
) ≥ mini≡s
E i = Es
therefore
limT→0
F(T) =M∑
s=1
∑
i≡s
νini
Es , limT→+∞
F(T) = +∞
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 10 / 25
Lemma: For any Λ, the equation F(T) = Λ has a unique positivesolution
Steps of the proof:by direct computations, we check that F(T) is a monotonicallyincreasing function of its argument;recalling that, for each gas Gs , energy levels are such thatEs < Es+M < Es+2M < · · · < Es+QM, we get
∑
i≡s νiE iψ(E i , T)
∑
j≡s νjψ(E j , T)
=νsEs +
∑
i≡si,sνiE i exp
(
− E i−Es
KT
)
νs +∑
i≡si,sνi exp
(
− E i−Es
KT
) ≥ mini≡s
E i = Es
therefore
limT→0
F(T) =M∑
s=1
∑
i≡s
νini
Es , limT→+∞
F(T) = +∞
⇒ Since for T > 0 the function F(T) monotonically increasesfrom the minimum admissible value for Λ to +∞, existence anduniqueness of solution to F(T) = Λ is guaranteed
•M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 10 / 25
Remarks and basic properties
We have thus proved that the proposed BGK model is welldefined, in the sense that all auxiliary parameters are uniquelydetermined in terms of the actual fields
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 11 / 25
Remarks and basic properties
We have thus proved that the proposed BGK model is welldefined, in the sense that all auxiliary parameters are uniquelydetermined in terms of the actual fields
Collision equilibria:
f i(v) =Mi(v) ∀v ∈ R3 ⇒ ni = ni , ui = u, T i = T
hence also the actual number densities fulfill the constraintsni = Ns ψ(E i ,T) (i ≡ s)
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 11 / 25
Remarks and basic properties
We have thus proved that the proposed BGK model is welldefined, in the sense that all auxiliary parameters are uniquelydetermined in terms of the actual fields
Collision equilibria:
f i(v) =Mi(v) ∀v ∈ R3 ⇒ ni = ni , ui = u, T i = T
hence also the actual number densities fulfill the constraintsni = Ns ψ(E i ,T) (i ≡ s)
It may be proved that the usual H–functional
H =M∑
s=1
∑
i≡s
∫
f i log f i dv
is a Lyapunov functional even for the BGK kinetic model
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 11 / 25
BGK approximation for polyatomic REACTING mixtures(4 species)
∂f i
∂t+ v · ∇xf i = νi(Mi − f i) i = 1, . . . , 4Q
with parameters of the Maxwellian attractorsMi =Mi(ni , u, T)bound together as
ni = Ns ψ(E i , T) ,N1N2
N3N4=
(
µ12
µ34
)3/2 Z1(T)Z2(T)
Z3(T)Z4(T)e
∆E3412
KT (a)
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 12 / 25
BGK approximation for polyatomic REACTING mixtures(4 species)
∂f i
∂t+ v · ∇xf i = νi(Mi − f i) i = 1, . . . , 4Q
with parameters of the Maxwellian attractorsMi =Mi(ni , u, T)bound together as
ni = Ns ψ(E i , T) ,N1N2
N3N4=
(
µ12
µ34
)3/2 Z1(T)Z2(T)
Z3(T)Z4(T)e
∆E3412
KT (a)
⇒ 7 independent free parameters (u, T , three among Ns) to bedetermined imposing the preservation of correct collision invariants
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 12 / 25
BGK approximation for polyatomic REACTING mixtures(4 species)
∂f i
∂t+ v · ∇xf i = νi(Mi − f i) i = 1, . . . , 4Q
with parameters of the Maxwellian attractorsMi =Mi(ni , u, T)bound together as
ni = Ns ψ(E i , T) ,N1N2
N3N4=
(
µ12
µ34
)3/2 Z1(T)Z2(T)
Z3(T)Z4(T)e
∆E3412
KT (a)
⇒ 7 independent free parameters (u, T , three among Ns) to bedetermined imposing the preservation of correct collision invariants
Main difference with respect to the inert mixture:the constraint coming from energy conservation and mass actionlaw (a) are two transcendental equations for the unknowns (N1, T)
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 12 / 25
Incompressible hydrodynamic limit of the Boltzmannequations (four reacting species)
ε ∂t fiε + v · ∇xf i
ε =1ε
4Q∑
j=1
Q ij(f iε, f
jε) + ε Ji i = 1, . . . , 4Q
where Q ij(f iε, f
jε) is the classical elastic operator and Ji takes into
account all non–conservative collisions (inelastic transitions andchemical reactions)
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 13 / 25
Incompressible hydrodynamic limit of the Boltzmannequations (four reacting species)
ε ∂t fiε + v · ∇xf i
ε =1ε
4Q∑
j=1
Q ij(f iε, f
jε) + ε Ji i = 1, . . . , 4Q
where Q ij(f iε, f
jε) is the classical elastic operator and Ji takes into
account all non–conservative collisions (inelastic transitions andchemical reactions)
As in [Bardos, Golse, Levermore, J. Stat. Phys. (1991)] and in[Bisi, Desvillettes, ESAIM - Math. Model. Numer. Anal. (2014)], welook for solutions in the form
f iε = ρi Mi
(1,0,1)(1 + ε giε)
(perturbations of collision equilibria)
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 13 / 25
Here Mi(1,0,1)
are absolute normalized Maxwellians
Mi(v) =
(
mi
2π
)3/2
e−mi2 v2
and ρi > 0 are constants (without loss of generality ρ =∑4Q
i=1 = 1)such that ρi Mi
(1,0,1)are equilibria even of the non–conservative
operators Ji
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 14 / 25
Here Mi(1,0,1)
are absolute normalized Maxwellians
Mi(v) =
(
mi
2π
)3/2
e−mi2 v2
and ρi > 0 are constants (without loss of generality ρ =∑4Q
i=1 = 1)such that ρi Mi
(1,0,1)are equilibria even of the non–conservative
operators Ji
In other words, if we denote
Ns =∑
i≡s
ρi , Zs =∑
i≡s
e−(Ei−Es), s = 1, . . . , 4 ,
for any i ≡ s the constant ρi has to be related to Ns and Zs as
ρi =Ns
Zs e−(Ei−Es)
and global number densities have to fulfill the mass action law
N1N2
Z1Z2=
(
µ12
µ34
)3/2
e∆E3412
N3N4
Z3Z4
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 14 / 25
Incompressible Navier–Stokes equations
By inserting the ansatz f iε = ρi Mi
(1,0,1)(1 + εgi
ε) into the scaledBoltzmann equations we get
4Q∑
j=1
ρiρj[
Q ij(giεM
i ,Mj) + Q ij(Mi , gjεM
j)]
= O(ε)
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 15 / 25
Incompressible Navier–Stokes equations
By inserting the ansatz f iε = ρi Mi
(1,0,1)(1 + εgi
ε) into the scaledBoltzmann equations we get
4Q∑
j=1
ρiρj[
Q ij(giεM
i ,Mj) + Q ij(Mi , gjεM
j)]
= O(ε)
hencegiε(v) = αi + mi v · u +
(
12
mi v2 − 32
)
T + O(ε)
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 15 / 25
Incompressible Navier–Stokes equations
By inserting the ansatz f iε = ρi Mi
(1,0,1)(1 + εgi
ε) into the scaledBoltzmann equations we get
4Q∑
j=1
ρiρj[
Q ij(giεM
i ,Mj) + Q ij(Mi , gjεM
j)]
= O(ε)
hencegiε(v) = αi + mi v · u +
(
12
mi v2 − 32
)
T + O(ε)
The parameters αi, u, T (depending on t and x) are perturbationsof number densities, velocity and temperature∫
f iε(v) dv = ρi(1 + ε αi) + O(ε2)
∫
v f iε(v) dv = ε ρi u + O(ε2)
mi∫
v2f iε(v) dv = 3 ρi + ε 3 ρi(αi + T) + O(ε2)
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 15 / 25
Incompressible Navier–Stokes equations
By inserting the ansatz f iε = ρi Mi
(1,0,1)(1 + εgi
ε) into the scaledBoltzmann equations we get
4Q∑
j=1
ρiρj[
Q ij(giεM
i ,Mj) + Q ij(Mi , gjεM
j)]
= O(ε)
hencegiε(v) = αi + mi v · u +
(
12
mi v2 − 32
)
T + O(ε)
The parameters αi, u, T (depending on t and x) are perturbationsof number densities, velocity and temperature∫
f iε(v) dv = ρi(1 + ε αi) + O(ε2)
∫
v f iε(v) dv = ε ρi u + O(ε2)
mi∫
v2f iε(v) dv = 3 ρi + ε 3 ρi(αi + T) + O(ε2)
We look for evolution equations for αi (i = 1, . . . , 4Q), u, T
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 15 / 25
We consider interactions of Maxwell molecules type: gσij(g, χ) = ϑij(χ)
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 16 / 25
We consider interactions of Maxwell molecules type: gσij(g, χ) = ϑij(χ)and define
κij = 2π∫ π
0ϑij(χ)(1−cos χ) sin χ dχ νij = 2π
∫ π
0ϑij(χ)(1−cos2 χ) sin χ dχ
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 16 / 25
We consider interactions of Maxwell molecules type: gσij(g, χ) = ϑij(χ)and define
κij = 2π∫ π
0ϑij(χ)(1−cos χ) sin χ dχ νij = 2π
∫ π
0ϑij(χ)(1−cos2 χ) sin χ dχ
We formally get that parameters αi (i = 1, . . . , 4Q), u, T satisfythe following Navier–Stokes system for polyatomic mixtures
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 16 / 25
We consider interactions of Maxwell molecules type: gσij(g, χ) = ϑij(χ)and define
κij = 2π∫ π
0ϑij(χ)(1−cos χ) sin χ dχ νij = 2π
∫ π
0ϑij(χ)(1−cos2 χ) sin χ dχ
We formally get that parameters αi (i = 1, . . . , 4Q), u, T satisfythe following Navier–Stokes system for polyatomic mixtures
Incompressibility condition:
∇x · u = 0
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 16 / 25
We consider interactions of Maxwell molecules type: gσij(g, χ) = ϑij(χ)and define
κij = 2π∫ π
0ϑij(χ)(1−cos χ) sin χ dχ νij = 2π
∫ π
0ϑij(χ)(1−cos2 χ) sin χ dχ
We formally get that parameters αi (i = 1, . . . , 4Q), u, T satisfythe following Navier–Stokes system for polyatomic mixtures
Incompressibility condition:
∇x · u = 0
Boussinesq identity:
∇x
4Q∑
i=1
(
ρi αi)
+ T
= 0
[Such constraints follow from conservation of total number densityand of global momentum, respectively]
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 16 / 25
Convection-diffusion equations for the densities of thecomponents (main difference with respect to the singlespecies frame):
∂t
[
∑
j,i
ρjµij κij(αi − αj)
]
+ u · ∇x
[
∑
j,i
ρjµijκij(αi − αj)
]
= ∆x
[
∑
j,i
ρj(αi − αj)
]
+
(
∑
j,i
ρj
ρiµij κij
) ∫
Ji(1) dv −
∑
j,i
µij κij∫
Jj(1)
dv i = 1, . . . , 4Q − 1 ,
where µij = mi mj
mi+mj is the reduced mass and Ji(1)
is the O(ε) part of
the operator Ji (contributions will be made explicit for Maxwellmolecules)
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 17 / 25
Convection-diffusion equations for the densities of thecomponents (main difference with respect to the singlespecies frame):
∂t
[
∑
j,i
ρjµij κij(αi − αj)
]
+ u · ∇x
[
∑
j,i
ρjµijκij(αi − αj)
]
= ∆x
[
∑
j,i
ρj(αi − αj)
]
+
(
∑
j,i
ρj
ρiµij κij
) ∫
Ji(1) dv −
∑
j,i
µij κij∫
Jj(1)
dv i = 1, . . . , 4Q − 1 ,
where µij = mi mj
mi+mj is the reduced mass and Ji(1)
is the O(ε) part of
the operator Ji (contributions will be made explicit for Maxwellmolecules)
More precisely, if D iin and D i
ch denote the sets of all inelastictransitions and chemical reactions involving particles A i , we have
Ji(1)
=∑
(j,h,k )∈D iin∪D i
ch
{
ρhρk[
Jiijhk+(gh
εMh ,Mk ) + Jiijhk+(Mh , gk
εMk )]
− ρiρj[
Jiijhk−(g
iεM
i ,Mj) + Jiijhk−(M
i , gjεM
j)]}
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 17 / 25
Convection-diffusion equation for the momentum
∂tu + u · ∇xu + ∇xp = d1 ∆xu
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 18 / 25
Convection-diffusion equation for the momentum
∂tu + u · ∇xu + ∇xp = d1 ∆xu
Convection-diffusion equation for the temperature
∂tT + u · ∇xT = d2 ∆xT +4Q∑
i=1
∫ (
15
miv2 − 1
)
Ji(1) dv
where diffusion coefficients d1, d2 are the unique solutions ofsuitable linear systems and depend on masses and on averagedcollision frequencies κij , νij
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 18 / 25
Convection-diffusion equation for the momentum
∂tu + u · ∇xu + ∇xp = d1 ∆xu
Convection-diffusion equation for the temperature
∂tT + u · ∇xT = d2 ∆xT +4Q∑
i=1
∫ (
15
miv2 − 1
)
Ji(1) dv
where diffusion coefficients d1, d2 are the unique solutions ofsuitable linear systems and depend on masses and on averagedcollision frequencies κij , νij
Remarks:The final system is not strongly coupled, evolution equationfor u could be solved separately
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 18 / 25
Convection-diffusion equation for the momentum
∂tu + u · ∇xu + ∇xp = d1 ∆xu
Convection-diffusion equation for the temperature
∂tT + u · ∇xT = d2 ∆xT +4Q∑
i=1
∫ (
15
miv2 − 1
)
Ji(1) dv
where diffusion coefficients d1, d2 are the unique solutions ofsuitable linear systems and depend on masses and on averagedcollision frequencies κij , νij
Remarks:The final system is not strongly coupled, evolution equationfor u could be solved separately
Velocities or temperatures specific to each species wouldappear only if we considered higher orders in expansions ofdistributions, or if we took as dominant operator (of order 1/ε)only Q ii(f i
ε, fiε)
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 18 / 25
In the one-species case concentration α is completely knownfrom the equation for T , while for a mixture 4Q − 1 additionalindependent evolution equations for αi are needed
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 19 / 25
In the one-species case concentration α is completely knownfrom the equation for T , while for a mixture 4Q − 1 additionalindependent evolution equations for αi are needed
For a mixture of only two monatomic species, the additionalequation is simply provided by the difference of the two kineticequations
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 19 / 25
In the one-species case concentration α is completely knownfrom the equation for T , while for a mixture 4Q − 1 additionalindependent evolution equations for αi are needed
For a mixture of only two monatomic species, the additionalequation is simply provided by the difference of the two kineticequations
Computation of diffusion coefficients d1 and d2 and the proofthat they are strictly positive is not a trivial extension of theone-species case
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 19 / 25
In the one-species case concentration α is completely knownfrom the equation for T , while for a mixture 4Q − 1 additionalindependent evolution equations for αi are needed
For a mixture of only two monatomic species, the additionalequation is simply provided by the difference of the two kineticequations
Computation of diffusion coefficients d1 and d2 and the proofthat they are strictly positive is not a trivial extension of theone-species case
Contributions due to the polyatomic nature of gases (and tochemical reactions) affect equations for number densities andglobal temperature
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 19 / 25
Some steps of the derivation
Equations for concentrations (i = 1, . . . , 4Q)
ε ∂t
∫
(giεM
i)(v) dv + ∇x ·∫
v (giεM
i)(v) dv = ε1
ρi
∫
Ji(1) dv
We need a closure for the streaming terms to O(ε) accuracy
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 20 / 25
Some steps of the derivation
Equations for concentrations (i = 1, . . . , 4Q)
ε ∂t
∫
(giεM
i)(v) dv + ∇x ·∫
v (giεM
i)(v) dv = ε1
ρi
∫
Ji(1) dv
We need a closure for the streaming terms to O(ε) accuracy
We resort to momentum equations of each component
ε2 ρi∂t
∫
v(giεM
i)(v) dv + ε ρi∇x ·∫
v ⊗ v(giεM
i)(v) dv =
=∑
j,i
{
ρiρj∫
v[
Q ij(giεM
i ,Mj) + Q ij(Mi , gjεM
j)]
dv
+ ε ρiρj∫
v Q ij(giεM
i , gjεM
j)dv}
+ ε2∫
v Ji(1) dv
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 20 / 25
Some steps of the derivation
Equations for concentrations (i = 1, . . . , 4Q)
ε ∂t
∫
(giεM
i)(v) dv + ∇x ·∫
v (giεM
i)(v) dv = ε1
ρi
∫
Ji(1) dv
We need a closure for the streaming terms to O(ε) accuracy
We resort to momentum equations of each component
ε2 ρi∂t
∫
v(giεM
i)(v) dv + ε ρi∇x ·∫
v ⊗ v(giεM
i)(v) dv =
=∑
j,i
{
ρiρj∫
v[
Q ij(giεM
i ,Mj) + Q ij(Mi , gjεM
j)]
dv
+ ε ρiρj∫
v Q ij(giεM
i , gjεM
j)dv}
+ ε2∫
v Ji(1) dv
We find an explicit relation between the terms in red
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 20 / 25
∑
j,i
ρiρj∫
v[
Q ij(giεM
i ,Mj) + Q ij(Mi , gjεM
j)]
dv
= −
∑
j,i
ρjµij κij
ρi
mi
∫
v (giεM
i)(v) dv +ρi
mi
∑
j,i
ρjµij κij∫
v (gjεM
j)(v) dv
hence we can “insert” the i–th momentum equation into a suitablelinear combination of number densities equations, and we evaluatethen each term recalling thatgiε(v) = αi + mi v · u +
(
12 mi v2 − 3
2
)
T + O(ε)
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 21 / 25
∑
j,i
ρiρj∫
v[
Q ij(giεM
i ,Mj) + Q ij(Mi , gjεM
j)]
dv
= −
∑
j,i
ρjµij κij
ρi
mi
∫
v (giεM
i)(v) dv +ρi
mi
∑
j,i
ρjµij κij∫
v (gjεM
j)(v) dv
hence we can “insert” the i–th momentum equation into a suitablelinear combination of number densities equations, and we evaluatethen each term recalling thatgiε(v) = αi + mi v · u +
(
12 mi v2 − 3
2
)
T + O(ε)
∗ Analogously for closure of momentum and temperature equations:
integrals of streaming terms ∇x ·
4Q∑
i=1
ρi∫
Bin(v)(gi
εMi)(v) dv
with Bi1(v) = mi
(
v ⊗ v − 13
v2I)
, Bi2(v) =
(
12
miv2 − 52
)
v
are proportional to4Q∑
i,j=1
ρiρj θin
∫
Bin(v)
[
Q ij(giεM
i ,Mj)+Q ij(Mi , gjεM
j)]
dv (n = 1, 2)
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 21 / 25
Inelastic collision contributions∫
Ji(1)dv =
∑
(j,h,k )∈D iin∪D i
ch
K iijhk
where K iijhk represents the net production of particles of species A i
due to the interaction A i + A j ⇄ Ah + Ak
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 22 / 25
Inelastic collision contributions∫
Ji(1)dv =
∑
(j,h,k )∈D iin∪D i
ch
K iijhk
where K iijhk represents the net production of particles of species A i
due to the interaction A i + A j ⇄ Ah + Ak
Obvious symmetry property: K iijhk = K j
ijhk = −Khijhk = −Kk
ijhk
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 22 / 25
Inelastic collision contributions∫
Ji(1)dv =
∑
(j,h,k )∈D iin∪D i
ch
K iijhk
where K iijhk represents the net production of particles of species A i
due to the interaction A i + A j ⇄ Ah + Ak
Obvious symmetry property: K iijhk = K j
ijhk = −Khijhk = −Kk
ijhk
We adopt a Maxwell molecule assumption for any option (i, j, h, k)corresponding to an endothermic direct interaction (i.e. ∆Ehk
ij > 0):
νhkij =
∫
gσhkij (g, χ) dn′
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 22 / 25
Inelastic collision contributions∫
Ji(1)dv =
∑
(j,h,k )∈D iin∪D i
ch
K iijhk
where K iijhk represents the net production of particles of species A i
due to the interaction A i + A j ⇄ Ah + Ak
Obvious symmetry property: K iijhk = K j
ijhk = −Khijhk = −Kk
ijhk
We adopt a Maxwell molecule assumption for any option (i, j, h, k)corresponding to an endothermic direct interaction (i.e. ∆Ehk
ij > 0):
νhkij =
∫
gσhkij (g, χ) dn′
The relation for the cross section of the reverse exothermicinteraction σij
hk follows from the microreversibility condition
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 22 / 25
Inelastic collision contributions∫
Ji(1)dv =
∑
(j,h,k )∈D iin∪D i
ch
K iijhk
where K iijhk represents the net production of particles of species A i
due to the interaction A i + A j ⇄ Ah + Ak
Obvious symmetry property: K iijhk = K j
ijhk = −Khijhk = −Kk
ijhk
We adopt a Maxwell molecule assumption for any option (i, j, h, k)corresponding to an endothermic direct interaction (i.e. ∆Ehk
ij > 0):
νhkij =
∫
gσhkij (g, χ) dn′
The relation for the cross section of the reverse exothermicinteraction σij
hk follows from the microreversibility condition
⇒∫
Ji(1)dv =
∑
(j,h,k )∈D iEn
K iijhk −
∑
(j,h,k )∈D iEx
Khhkij
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 22 / 25
Recalling that giε(v) = αi + mi v · u +
(
12 mi v2 − 3
2
)
T + O(ε) and
the assumptions on the leading order number densities ρi we get∫
Ji(1)dv =
2√π
∑
(j,h,k )∈D iin∪D i
ch
Λhkij
[
αh + αk − αi − αj − T∆Ehkij
]
Γ
(
32,∣
∣
∣
∣
∆Ehkij
∣
∣
∣
∣
)
where
Λhkij =
νhkij ρ
iρj if (j, h, k) ∈ D iEn
νijhkρ
hρk if (j, h, k) ∈ D iEx
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 23 / 25
Recalling that giε(v) = αi + mi v · u +
(
12 mi v2 − 3
2
)
T + O(ε) and
the assumptions on the leading order number densities ρi we get∫
Ji(1)dv =
2√π
∑
(j,h,k )∈D iin∪D i
ch
Λhkij
[
αh + αk − αi − αj − T∆Ehkij
]
Γ
(
32,∣
∣
∣
∣
∆Ehkij
∣
∣
∣
∣
)
where
Λhkij =
νhkij ρ
iρj if (j, h, k) ∈ D iEn
νijhkρ
hρk if (j, h, k) ∈ D iEx
The content of the square brackets is the linearization (i.e.,the O(ε) terms) of the mass action law for global distributionfunctions (f i
ε, fjε, fh
ε , fkε )
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 23 / 25
Recalling that giε(v) = αi + mi v · u +
(
12 mi v2 − 3
2
)
T + O(ε) and
the assumptions on the leading order number densities ρi we get∫
Ji(1)dv =
2√π
∑
(j,h,k )∈D iin∪D i
ch
Λhkij
[
αh + αk − αi − αj − T∆Ehkij
]
Γ
(
32,∣
∣
∣
∣
∆Ehkij
∣
∣
∣
∣
)
where
Λhkij =
νhkij ρ
iρj if (j, h, k) ∈ D iEn
νijhkρ
hρk if (j, h, k) ∈ D iEx
The content of the square brackets is the linearization (i.e.,the O(ε) terms) of the mass action law for global distributionfunctions (f i
ε, fjε, fh
ε , fkε )
Suitable combinations of∫
Ji(1)dv complete the derivation of
incompressible Navier–Stokes equations for number densitiesand global temperature
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 23 / 25
Work in progress and open problems
Work in progress (joint with S. Brull): formal compressiblehydrodynamic limit (at Navier–Stokes level) for polyatomic gasmixtures, owing to the Chapman–Enskog method
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 24 / 25
Work in progress and open problems
Work in progress (joint with S. Brull): formal compressiblehydrodynamic limit (at Navier–Stokes level) for polyatomic gasmixtures, owing to the Chapman–Enskog method
Open problem: Fredholm alternative for the linearizedBoltzmann operator for polyatomic gases (with discrete orcontinuous internal energy)
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 24 / 25
Work in progress and open problems
Work in progress (joint with S. Brull): formal compressiblehydrodynamic limit (at Navier–Stokes level) for polyatomic gasmixtures, owing to the Chapman–Enskog method
Open problem: Fredholm alternative for the linearizedBoltzmann operator for polyatomic gases (with discrete orcontinuous internal energy)
Open problem: more appropriate descriptions for chains ofchemical reactions occurring in physical applications
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 24 / 25
Thank you for your attention
M. Bisi, Parma Kinetic theory for polyatomic gas mixtures 25 / 25