A Direct Arbitrary-Lagrangian-Eulerian ADER-WENO Finite Volume Scheme on Unstructured Tetrahedral Meshes for Conservative and Nonconservative Hyperbolic Systems in 3D Walter Boscheri and Michael Dumbser
[email protected] University of Trento (Italy) - Department of Civil, Environmental and Mechanical Engineering -
SHARK FV 2015 May, 18th − 22nd 2015 - Ofir
Outline 1
Introduction
2
Numerical Method Hyperbolic Balance Laws High Order WENO Reconstruction on Unstructured Meshes Local Space-Time DG Predictor on Moving Meshes Mesh Motion Solution Algorithm
3
Validation The Euler equations The Baer–Nunziato Model
4
Conclusions and Outlook
1. Introduction
Introduction: Lagrangian methods D ∂ () = () + v ∇() Dt ∂t Advantages availability of trajectory information; no convection terms → numerical stability; less numerical diffusion; material interfaces are precisely located and identified.
Walter Boscheri
High Order ALE ADER Finite Volume Schemes on Unstructured Meshes
2 / 19
1. Introduction
Introduction: Lagrangian methods D ∂ () = () + v ∇() Dt ∂t Advantages availability of trajectory information; no convection terms → numerical stability; less numerical diffusion; material interfaces are precisely located and identified.
Disadvantages high computational cost; mesh distortion. Walter Boscheri
High Order ALE ADER Finite Volume Schemes on Unstructured Meshes
2 / 19
2. Numerical Method
Hyperbolic Balance Laws
Outline 1 2
3
4
Introduction Numerical Method Hyperbolic Balance Laws High Order WENO Reconstruction on Unstructured Meshes Local Space-Time DG Predictor on Moving Meshes Mesh Motion Solution Algorithm Validation The Euler equations The Baer–Nunziato Model Conclusions and Outlook
Walter Boscheri
High Order ALE ADER Finite Volume Schemes on Unstructured Meshes
2 / 19
2. Numerical Method
Hyperbolic Balance Laws
Hyperbolic Balance Laws ∂Q + ∇ · F(Q) + B(Q) · ∇Q = S(Q), ∂t
x ∈ Ω ⊂ R3 , t ∈ R+ 0
(1)
where Q ∈ ΩQ ⊂ Rν
→
state vector
F = (f, g, h)
→
flux tensor (purely conservative part of (1))
B = (B1 , B2 , B3 )
→
non–conservative terms of (1)
S(Q)
→
vector of algebraic source terms
P(Q) = B(Q) · ∇Q
→
abbreviation
Walter Boscheri
High Order ALE ADER Finite Volume Schemes on Unstructured Meshes
3 / 19
2. Numerical Method
Hyperbolic Balance Laws
Hyperbolic Balance Laws Moving mesh TΩn =
NE [
Tin
i=1
k = 1, 2, 3, 4
∀Tin
x
=
(x, y , z)
ξ
=
(ξ, η, ζ)
Walter Boscheri
x
=
y
=
z
=
n n n n n n n X1,i + X2,i − X1,i ξ + X3,i − X1,i η + X4,i − X1,i ζ n n n n n n n Y1,i + Y2,i − Y1,i ξ + Y3,i − Y1,i η + Y4,i − Y1,i ζ n n n n n n n Z1,i + Z2,i − Z1,i ξ + Z3,i − Z1,i η + Z4,i − Z1,i ζ High Order ALE ADER Finite Volume Schemes on Unstructured Meshes
3 / 19
2. Numerical Method
Hyperbolic Balance Laws
Hyperbolic Balance Laws Moving mesh TΩn =
NE [
Tin
i=1
k = 1, 2, 3, 4
∀Tin
x
=
(x, y , z)
ξ
=
(ξ, η, ζ)
Data representation Qni =
Walter Boscheri
1 |Tin |
Z
Q(x, y , z, t n )dV
Tin
High Order ALE ADER Finite Volume Schemes on Unstructured Meshes
3 / 19
2. Numerical Method
High Order WENO Reconstruction on Unstructured Meshes
Outline 1 2
3
4
Introduction Numerical Method Hyperbolic Balance Laws High Order WENO Reconstruction on Unstructured Meshes Local Space-Time DG Predictor on Moving Meshes Mesh Motion Solution Algorithm Validation The Euler equations The Baer–Nunziato Model Conclusions and Outlook
Walter Boscheri
High Order ALE ADER Finite Volume Schemes on Unstructured Meshes
3 / 19
2. Numerical Method
High Order WENO Reconstruction on Unstructured Meshes
High Order of Accuracy in Space Some spatial basis functions ψl (ξ, η, ζ) are used to write the reconstructed polynomials whs for each reconstruction stencil s: whs (x, y , z, t n ) =
M X
n,s n,s ˆ l,i ˆ l,i ψl (ξ, η, ζ)w := ψl (ξ, η, ζ)w
l=1
Integral conservation
1 |Tjn |
Z
ne [
Sis
=
n Tm(j)
M
=
(M + 1)(M + 2)(M + 3)/6
ne
=
3M
j=1 n,s ˆ l,i ψl (ξ, η, ζ)w dV = Qnj ,
Tjn
Walter Boscheri
∀Tjn ∈ Sis
High Order ALE ADER Finite Volume Schemes on Unstructured Meshes
4 / 19
2. Numerical Method
High Order WENO Reconstruction on Unstructured Meshes
High Order of Accuracy in Space Some spatial basis functions ψl (ξ, η, ζ) are used to write the reconstructed polynomials whs for each reconstruction stencil s: whs (x, y , z, t n ) =
M X
n,s n,s ˆ l,i ˆ l,i ψl (ξ, η, ζ)w := ψl (ξ, η, ζ)w
l=1
Integral conservation
1 |Tjn |
Z
ne [
Sis
=
n Tm(j)
M
=
(M + 1)(M + 2)(M + 3)/6
ne
=
3M
j=1 n,s ˆ l,i ψl (ξ, η, ζ)w dV = Qnj ,
∀Tjn ∈ Sis
Tjn
WENO polynomials wh (x, y , z, t n ) =
M X
n ˆ l,i ψl (ξ, η, ζ)w ,
with
n ˆ l,i w =
X
nonlinear WENO weights
oscillation indicators
ω ˜s =
λs (σs + )r
Walter Boscheri
,
ω ˜s ωs = P ˜q qω
n,s ˆ l,i ωs w
s
l=1
σs =
X
Z
α+β+γ≤M T e
∂ α+β+γ ψl (ξ, η, ζ) ∂ξα ∂η β ∂ζ γ
·
∂ α+β+γ ψm (ξ, η, ζ) ∂ξα ∂η β ∂ζ γ
High Order ALE ADER Finite Volume Schemes on Unstructured Meshes
dξdηdζ · 4 / 19
2. Numerical Method
Local Space-Time DG Predictor on Moving Meshes
Outline 1 2
3
4
Introduction Numerical Method Hyperbolic Balance Laws High Order WENO Reconstruction on Unstructured Meshes Local Space-Time DG Predictor on Moving Meshes Mesh Motion Solution Algorithm Validation The Euler equations The Baer–Nunziato Model Conclusions and Outlook
Walter Boscheri
High Order ALE ADER Finite Volume Schemes on Unstructured Meshes
4 / 19
2. Numerical Method
Local Space-Time DG Predictor on Moving Meshes
High Order of Accuracy in Time ˜ x = (x, y , z, t) x = (x, y , z) physical coordinates
ξ˜ = (ξ, η, ζ, τ ) ξ = (ξ, η, ζ) reference coordinates
˜ = θl (ξ, η, ζ, τ ) θl = θl (ξ) space–time basis functions
Isoparametric approach qh = qh (ξ, η, ζ, τ ) = θl (ξ, η, ζ, τ )b ql,i
bl,i Fh = (fh , gh , hh ) = Fh (ξ, η, ζ, τ ) = θl (ξ, η, ζ, τ )F
b l,i ˜ = θl (ξ) ˜ P Ph = B(qh ) · ∇qh = Ph (ξ)
bl,i Sh = Sh (ξ, η, ζ, τ ) = θl (ξ, η, ζ, τ )S ⇓
x(ξ, η, ζ, τ ) = θl (ξ, η, ζ, τ )b xl,i , z(ξ, η, ζ, τ ) = θl (ξ, η, ζ, τ )b zl,i ,
y (ξ, η, ζ, τ ) = θl (ξ, η, ζ, τ )b yl,i t(ξ, η, ζ, τ ) = θl (ξ, η, ζ, τ )tbl
Jacobian
xξ yξ ∂˜ x Jst = = ∂ ξ˜ zξ 0 Walter Boscheri
xη yη zη 0
xζ yζ zζ 0
xτ yτ zτ ∆t
Jst−1
ξx ∂ ξ˜ ηx = = ζx ∂˜ x 0
ξy ηy ζy 0
ξz ηz ζz 0
High Order ALE ADER Finite Volume Schemes on Unstructured Meshes
ξt ηt ζt
1 ∆t
5 / 19
2. Numerical Method
Local Space-Time DG Predictor on Moving Meshes
High Order of Accuracy in Time The PDE in the reference coordinate system
∇ξ =
∂ ∂ξ ∂ ∂η ∂ ∂ζ
,
∇=
∂ ∂x ∂ ∂y ∂ ∂z
ξx = ξy ξz
ηx ηy ηz
ζx ζy ζz
∂ ∂ξ ∂ ∂η ∂ ∂ζ
=
∂ξ ∂x
T ∇ξ
⇓ " # T T ∂Q ∂Q ∂ξ ∂ξ ∂ξ + ∆t · + ∇ξ · F + B(Q) · ∇ξ Q = ∆tS(Q) ∂τ ∂ξ ∂t ∂x ∂x Let define the following operators: [f , g ]τ =
Z1 Z
Z f (ξ, η, ζ, τ )g (ξ, η, ζ, τ )dξdηdζ, Te
Walter Boscheri
hf , g i =
f (ξ, η, ζ, τ )g (ξ, η, ζ, τ )dξdηdζdτ. 0 Te
High Order ALE ADER Finite Volume Schemes on Unstructured Meshes
5 / 19
2. Numerical Method
Local Space-Time DG Predictor on Moving Meshes
High Order of Accuracy in Time ∂Q = ∆tH ∂τ "
∂Q ∂ξ H = ∆tS − ∆t · + ∂ξ ∂t
∂ξ ∂x
T
∇ξ · F + B(Q) ·
∂ξ ∂x
#
T
∇ξ Q ,
b l,i . ˜ H Hh = θl (ξ)
Multiplication with θk (ξ) and integration over Te × [0, 1] yields the weak formulation of the PDE, i.e.
Walter Boscheri
∂θl θk , ∂τ
b l,i bl,i = hθk , θl i ∆t H q
High Order ALE ADER Finite Volume Schemes on Unstructured Meshes
5 / 19
2. Numerical Method
Local Space-Time DG Predictor on Moving Meshes
High Order of Accuracy in Time The nonlinear algebraic equation system can be written in a more compact matrix form as n b l,i bl,i = F0 w ˆ l,i + MH K1 q
1
K1 = [θk (ξ, 1), θl (ξ, 1)] −
Walter Boscheri
∂θk , θl ∂τ
,
F0 = [θk (ξ, 0), ψl (ξ)] ,
M = hθk , θl i .
High Order ALE ADER Finite Volume Schemes on Unstructured Meshes
5 / 19
2. Numerical Method
Local Space-Time DG Predictor on Moving Meshes
High Order of Accuracy in Time The nonlinear algebraic equation system can be written in a more compact matrix form as n b l,i bl,i = F0 w ˆ l,i + MH K1 q
1
K1 = [θk (ξ, 1), θl (ξ, 1)] −
∂θk , θl ∂τ
,
F0 = [θk (ξ, 0), ψl (ξ)] ,
M = hθk , θl i .
local PDE evolution n br brl,i+1 = K1−1 F0 w ˆ l,i q + MH l,i
Walter Boscheri
High Order ALE ADER Finite Volume Schemes on Unstructured Meshes
5 / 19
2. Numerical Method
Local Space-Time DG Predictor on Moving Meshes
High Order of Accuracy in Time The nonlinear algebraic equation system can be written in a more compact matrix form as n b l,i bl,i = F0 w ˆ l,i + MH K1 q
1
K1 = [θk (ξ, 1), θl (ξ, 1)] −
∂θk , θl ∂τ
local PDE evolution n br brl,i+1 = K1−1 F0 w ˆ l,i q + MH l,i
,
F0 = [θk (ξ, 0), ψl (ξ)] ,
M = hθk , θl i .
geometry evolution dx = V (Q, x, t) dt
b l,i Vh = θl (ξ, τ )V b l,i = V(ˆ V ql,i , ˆxl,i , tˆl )
0 b l,i K1 b xl,i = [θk (ξ, 0), x(ξ, t n )] + ∆tM V Walter Boscheri
High Order ALE ADER Finite Volume Schemes on Unstructured Meshes
5 / 19
2. Numerical Method
Mesh Motion
Outline 1 2
3
4
Introduction Numerical Method Hyperbolic Balance Laws High Order WENO Reconstruction on Unstructured Meshes Local Space-Time DG Predictor on Moving Meshes Mesh Motion Solution Algorithm Validation The Euler equations The Baer–Nunziato Model Conclusions and Outlook
Walter Boscheri
High Order ALE ADER Finite Volume Schemes on Unstructured Meshes
5 / 19
2. Numerical Method
Mesh Motion
Nodal solvers 1) The node solver of Cheng and Shu N S cs (JCP 227: 1567-1596, 2007) 1 Z 1 X n n n b l,j Vk = Vk,j with Vk,j = θl (ξe,m(k) , ηe,m(k) , ζe,m(k) , τ )dτ V µk j∈V k
0
X
µk =
1 = Nk
Tjn ∈Vk
or µk =
X
µk,j
µk,j = ρnj |Tjn |
Tjn ∈Vk
Walter Boscheri
High Order ALE ADER Finite Volume Schemes on Unstructured Meshes
6 / 19
2. Numerical Method
Mesh Motion
Nodal solvers 2) The node solver of Maire N S m n
(JCP 228: 2391-2425, 2009)
Vk = Mk−1
X
(Lk,j Pj nj + Mp,j Vj )
j∈Vk
with Mk,j =
X f ∈j
Walter Boscheri
j Zk,f Ljk,f njk,f ⊗ njk,f ,
Mk =
X
Mk,j ,
j Zk,f = ρj cj
j∈Vk
High Order ALE ADER Finite Volume Schemes on Unstructured Meshes
6 / 19
2. Numerical Method
Mesh Motion
Geometry Evolution Lagrangian coordinates n
xn+1,Lag = xnk + Vk · ∆t k Rezoned coordinates
Fk (xk ) =
X j∈Vk
K(Jj ),
x j − xkj k+3 j Jj = yk+3 − ykj j zk+3 − zkj xrez k
→
j xk+1 − xkj j yk+1 − ykj j zk+1 − zkj
min {Fk (xk )}
New coordinates (relaxation algorithm 1 ) n+1,Lag n+1,Lag rez + ω x − x , xn+1 = x k k k k k 1
j xk+2 − xkj j yk+2 − ykj j zk+2 − zkj
ωk = F xnk , xn+1,Lag k
Galera et al., JCP 229: 5755-5787, 2010
Walter Boscheri
High Order ALE ADER Finite Volume Schemes on Unstructured Meshes
7 / 19
2. Numerical Method
Solution Algorithm
Outline 1 2
3
4
Introduction Numerical Method Hyperbolic Balance Laws High Order WENO Reconstruction on Unstructured Meshes Local Space-Time DG Predictor on Moving Meshes Mesh Motion Solution Algorithm Validation The Euler equations The Baer–Nunziato Model Conclusions and Outlook
Walter Boscheri
High Order ALE ADER Finite Volume Schemes on Unstructured Meshes
7 / 19
2. Numerical Method
Solution Algorithm
Finite Volume Scheme The starting PDE system can be rewritten as ˜ + B(Q) ˜ ˜ ·F ˜ = S(Q) ∇ · ∇Q with ˜ = ∇
∂ ∂ ∂ ∂ , , , ∂x ∂y ∂z ∂t
T ,
˜ = (f, g, h, Q) , F
˜ = (B1 , B2 , B3 ). B
Integration over the space–time control volume Cin = Ti (t) × t n ; t n+1 yields Z Cin
Walter Boscheri
˜ dxdt + ˜ ·F ∇
Z Cin
˜ ˜ dxdt = B(Q) · ∇Q
Z S(Q) dxdt Cin
High Order ALE ADER Finite Volume Schemes on Unstructured Meshes
8 / 19
2. Numerical Method
Solution Algorithm
Finite Volume Scheme Application of Gauss’ theorem leads to Z Z Z ˜+D ˜ · n ˜ ˜ dxdt = S(Q) dxdt F ˜ dS + B(Q) · ∇Q ∂Cin
Cin
Cin \∂Cin
where n ˜ = (˜ nx , n˜y , n˜z , n˜t ),
1 ˜ ·n D ˜= 2
Z1
∂Ψ ˜ Ψ(Q− , Q+ , s) · n B ˜ ds. ∂s
0
Walter Boscheri
High Order ALE ADER Finite Volume Schemes on Unstructured Meshes
8 / 19
2. Numerical Method
Solution Algorithm
Finite Volume Scheme Application of Gauss’ theorem leads to Z Z Z ˜+D ˜ · n ˜ ˜ dxdt = S(Q) dxdt F ˜ dS + B(Q) · ∇Q ∂Cin
Cin
Cin \∂Cin
where n ˜ = (˜ nx , n˜y , n˜z , n˜t ),
1 ˜ ·n D ˜= 2
Z1
∂Ψ ˜ Ψ(Q− , Q+ , s) · n B ˜ ds. ∂s
0
Linear parametrization ∂Cijn = ˜ x (χ1 , χ2 , τ ) =
6 X
˜ nij,k βk (χ1 , χ2 , τ )X
k=1
0≤χ ˜ = (χ1 , χ2 , τ ) ≤ 1 |∂Cijn | Walter Boscheri
∂˜ x = , ∂χ ˜
n ˜ij =
High Order ALE ADER Finite Volume Schemes on Unstructured Meshes
∂˜ x /|∂Cijn | ∂χ ˜ 8 / 19
2. Numerical Method
Solution Algorithm
Finite Volume Scheme ALE–type One–Step Finite Volume Scheme |Tin+1 | Qn+1 i
=
|Tin | Qni
−
1 1 X Z Z Tj ∈Ni
Walter Boscheri
0
0
˜ ij dχdτ + |∂Cijn |G
Z (Sh − Ph ) dxdt
Cin \∂Cin
High Order ALE ADER Finite Volume Schemes on Unstructured Meshes
8 / 19
2. Numerical Method
Solution Algorithm
Finite Volume Scheme ALE–type One–Step Finite Volume Scheme |Tin+1 | Qn+1 i
=
|Tin | Qni
1 1 X Z Z
−
Tj ∈Ni
0
˜ ij dχdτ + |∂Cijn |G
Z (Sh − Ph ) dxdt
Cin \∂Cin
0
Rusanov numerical flux 1 Z 1 − ˜ ij = 1 F(q ˜ + ˜ − ˜ ˜ij + B(Ψ) G ·n ˜ ds − |λmax |I q+ h ) + F(qh ) · n h − qh 2 2 0
˜ n˜ = A ˜ ·n A ˜=
Walter Boscheri
q
n˜x2 + n˜y2 + n˜z2
∂F + B · n − (V · n) I , ∂Q
(˜ nx , n˜y , n˜z )T n= p n˜x2 + n˜y2 + n˜z2
High Order ALE ADER Finite Volume Schemes on Unstructured Meshes
8 / 19
2. Numerical Method
Solution Algorithm
Finite Volume Scheme ALE–type One–Step Finite Volume Scheme |Tin+1 | Qn+1 i
=
|Tin | Qni
−
1 1 X Z Z Tj ∈Ni
0
˜ ij dχdτ + |∂Cijn |G
Z (Sh − Ph ) dxdt
Cin \∂Cin
0
Osher numerical flux 1 Z 1 ˜ + − ˜ ij = 1 F(q ˜ + ˜ − ˜ ˜ij + G B(Ψ) ·n ˜ − A n ˜ (Ψ) ds qh − qh h ) + F(qh ) · n 2 2 0
Ψ = Ψ(Q− , Q+ , s) = Q− + s(Q+ − Q− ), |A| = R|Λ|R−1 , Walter Boscheri
0≤s≤1
|Λ| = diag (|λ1 |, |λ2 |, ..., |λν |)
High Order ALE ADER Finite Volume Schemes on Unstructured Meshes
8 / 19
3. Validation
The Euler equations
Outline 1 2
3
4
Introduction Numerical Method Hyperbolic Balance Laws High Order WENO Reconstruction on Unstructured Meshes Local Space-Time DG Predictor on Moving Meshes Mesh Motion Solution Algorithm Validation The Euler equations The Baer–Nunziato Model Conclusions and Outlook
Walter Boscheri
High Order ALE ADER Finite Volume Schemes on Unstructured Meshes
8 / 19
3. Validation
The Euler equations
The Euler equations
Qt + fx + gy + hz = S(Q) with
ρ Q = ρvj , ρE
ρvi F = (f, g, h) = ρvi vj + pδij . vi (ρE + p)
ρ → fluid density v = (vx , vy , vz ) → fluid’s velocity vector S → vector of source term (S = 0) p → fluid pressure
Walter Boscheri
1 p = (γ−1) ρE − ρ(vx2 + vy2 + vz2 ) 2 ↓ equation of state
High Order ALE ADER Finite Volume Schemes on Unstructured Meshes
9 / 19
3. Validation
The Euler equations
Convergence studies: Shu Vortex (Hu and Shu, JCP 150: 97-127 , 1999 )
IC: (ρ, vx , vy , vz , p) = (1 + δρ, 1 + δvx , 1 + δvy , 0, 1 + δp) Computational domain Ω(0) = [0; 10] × [0; 10] × [0; 5] Convective velocity vc = (1, 1, 0) Perturbations
δvx δvy δvz
=
δS
=
δT Walter Boscheri
=
Parameters −(y − 5) 1−r2 2 (x − 5) e 2π 0 0 −
(γ − 1)
r2
=
(x − 5)2 + (y − 5)2 radius
=
5 vortex strength
γ
=
1.4 ratio of specific heat
2
2 e 1−r High Order ALE ADER Finite Volume Schemes on Unstructured Meshes
10 / 19
3. Validation
The Euler equations
Convergence studies: Shu Vortex (Hu and Shu, JCP 150: 97-127 , 1999 ) h(Ω(tf )) 3.43E-01 2.85E-01 2.09E-01 1.47E-01 2.89E-01 2.17E-01 1.52E-01 1.13E-01 2.89E-01 2.17E-01 1.52E-01 1.13E-01 Walter Boscheri
L 2 O1 1.081E-01 9.159E-02 6.875E-02 4.899E-02 O3 1.718E-02 7.641E-03 2.601E-03 1.049E-03 O5 2.272E-03 6.605E-04 1.234E-04 2.932E-05
O(L2 )
h(Ω, tf )
0.9 0.9 1.0
2.89E-01 2.16E-01 1.52E-01 1.13E-01
2.8 3.1 3.1
2.89E-01 2.17E-01 1.52E-01 1.13E-01
4.3 4.8 4.9
2.89E-01 2.17E-01 1.52E-01 1.13E-01
L 2 O2 2.214E-02 1.202E-02 5.865E-03 3.254E-03 O4 4.116E-03 1.369E-03 3.273E-04 9.802E-05 O6 1.015E-03 2.312E-04 3.090E-05 6.576E-06
O(L2 ) 2.1 2.0 2.0 3.8 4.1 4.1 5.1 5.7 5.2
High Order ALE ADER Finite Volume Schemes on Unstructured Meshes
10 / 19
3. Validation
The Euler equations
Three–dimensional Explosion Problem Inner state (Qi ) Outer state (Qo )
ρ 1.0 0.125
vx 0.0 0.0
vy 0.0 0.0
vz 0.0 0.0
p 1.0 0.1
Parameters r2
=
x 2 + y 2 + z 2 radius
R
=
0.5 internal radius
γ
=
1.4 ratio of specific heat
Analytical solution Qt + F(Q)r = S(Q)
ρ Q = ρu , ρE Walter Boscheri
ρu F = ρu + p , u(ρE + p) 2
α S=− r
ρu 2 ρu u(ρE + p)
High Order ALE ADER Finite Volume Schemes on Unstructured Meshes
α=2
11 / 19
3. Validation
The Euler equations
Three–dimensional Explosion Problem
t = 0.00
t = 0.08
4th order accurate explosion problem with NE = 7225720. Walter Boscheri
High Order ALE ADER Finite Volume Schemes on Unstructured Meshes
11 / 19
3. Validation
The Euler equations
Three–dimensional Explosion Problem
t = 0.16
t = 0.25
4th order accurate explosion problem with NE = 7225720. Walter Boscheri
High Order ALE ADER Finite Volume Schemes on Unstructured Meshes
11 / 19
3. Validation
The Euler equations
Three–dimensional Explosion Problem Density profile Reference solution ALE WENO (O4)
1.3 1.2 1.1 1 0.9 0.8
rho
0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 -0.1
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
x
4th order accurate explosion problem with NE = 7225720. Walter Boscheri
High Order ALE ADER Finite Volume Schemes on Unstructured Meshes
11 / 19
3. Validation
The Euler equations
Three–dimensional Explosion Problem Velocity and pressure profile Reference solution ALE WENO (O4)
1.2
1.1
1.1
1
1
0.9
0.9
0.8
0.8
0.7
0.7
0.6
0.6
0.5
0.5
0.4
0.4
0.3
0.3
0.2
0.2
0.1
0.1
0 -0.1
Reference solution ALE WENO (O4)
1.3
1.2
p
u
1.3
0 0
0.1
0.2
0.3
0.4
0.5
x
0.6
0.7
0.8
0.9
1
-0.1
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
x
4th order accurate explosion problem with NE = 7225720. Walter Boscheri
High Order ALE ADER Finite Volume Schemes on Unstructured Meshes
11 / 19
3. Validation
The Kidder problem IC:
The Euler equations
(Kidder, Nucl. Fusion 1: 3-14 , 1976 )
ρ0 = ρ(r , 0) =
2 − r2 re,0 2 − r2 re,0 i,0
γ−1 ρi,0 +
!
2 r 2 − ri,0 2 − r2 re,0 e,0
1 γ−1
ργ−1 e,0
p0 =1 ργ0 p0 (r ) = s0 ρ0 (r )γ s0 =
Z Y X
Parameters 1
re (0) = 1.0 external radius
0.8
ri (0) = 0.9 internal radius
0.6
z
γ = 5/3 ratio of specific heat ρi,0 = 1.0 initial internal density
0.4
1
0.2 0.8 0.6
y
0
ρe,0 = 2.0 initial external density
0.4 0
0.2
0.2 0.4
0.6
x
Walter Boscheri
0.8
0 1
High Order ALE ADER Finite Volume Schemes on Unstructured Meshes
12 / 19
3. Validation
The Euler equations
The Kidder problem
(Kidder, Nucl. Fusion 1: 3-14 , 1976 )
Analytical Solution
R(r , t) ρ0 ρ (R(r , t), t) = h(t) h(t) R(r , t) d h(t) ur (R(r , t), t) = dt h(t) 2γ R(r , t) p (R(r , t), t) = h(t)− p0 γ−1 h(t) 2 − γ−1
with r h(t) =
t2 1− 2, τ
s τ =
2
2
γ − 1 (re,0 − ri,0 ) 2 2 2 ce,0 − ci,0
and the internal and external sound speeds ci and ce defined as r r pi pe ci = γ , ce = γ ρi ρe Walter Boscheri
High Order ALE ADER Finite Volume Schemes on Unstructured Meshes
12 / 19
3. Validation
The Kidder problem
The Euler equations
(Kidder, Nucl. Fusion 1: 3-14 , 1976 )
Rinternal: exact solution Rinternal: ALE WENO (O4) Rexternal: exact solution Rexternal: ALE WENO (O4)
1.1
1
Radius
0.9
0.8
0.7
0.6
0.5
0.4 -0.05
0
0.05
0.1
0.15
0.2
0.25
time
Internal radius External radius Walter Boscheri
rex 0.450000 0.500000
rnum 0.449765 0.499728
|err | 2.35E-04 2.72E-04
High Order ALE ADER Finite Volume Schemes on Unstructured Meshes
12 / 19
3. Validation
The Euler equations
The Saltzman problem
(Dukovicz et al., JCP 99: 115-134 , 1992 ) Z Y X
Computational domain Ω(0) = [0; 1] × [0; 0.1] × [0; 0.1]
0
Piston velocity vp = (1, 0, 0)
0.2
0.4
x
0.6
0.1
0.08
z
0.06
0.8
0.04
0.02
0 0.1 0.08 0.06 1 0.04
y
0.02 0
Z Y
Mesh size 100 × 10 × 10 cubes
X
0
IC: ρ0 = 1, v0 = 0, e0 = 10−4
0.2
0.4
x
0.6
0.1
0.08
z
0.06
0.8
0.04
0.02
0 0.1 0.08 0.06 1 0.04
y
0.02 0
Analytical solution ρ u p
Left state 1.0 1.0 6.67 · 10−5
Walter Boscheri
Right state 1.0 -1.0 6.67 · 10−5
final time: shifting: final shock location: exact density:
tf = 0.6 d = up · tf x = 0.8 ρe = 4.0
High Order ALE ADER Finite Volume Schemes on Unstructured Meshes
13 / 19
3. Validation
The Saltzman problem
Walter Boscheri
The Euler equations
(Dukovicz et al., JCP 99: 115-134 , 1992 )
High Order ALE ADER Finite Volume Schemes on Unstructured Meshes
13 / 19
3. Validation
The Euler equations
The Saltzman problem
(Dukovicz et al., JCP 99: 115-134 , 1992 )
Exact solution ALE WENO (O3)
4.5
Exact solution ALE WENO (O3)
1.4 1.2
4
1
3.5
0.8 3
u
rho
0.6 2.5
0.4 2 0.2 1.5
0
1 0.5 0.6
-0.2
0.65
0.7
0.75
0.8
x
0.85
0.9
0.95
1
-0.4 0.6
0.65
0.7
0.75
0.8
0.85
0.9
0.95
1
x
3rd order numerical results of density and velocity for the Saltzman problem.
Walter Boscheri
High Order ALE ADER Finite Volume Schemes on Unstructured Meshes
13 / 19
3. Validation
The Euler equations
The Sedov Problem Computational domain Ω(0) = [0; 1.2] × [0; 1.2] × [0; 1.2] Mesh size 40 × 40 × 40 cubes IC: ρ0 = 1, v0 = 0, γ = 1.4, Etot = 0.851072 p0 =
Etot (γ − 1)ρ0 8·V or −6 10
in cor , elsewhere.
Analytical solution ⇒ Kamm et al., Technical Report LA-UR-07-2849, LANL, 2007 Walter Boscheri
High Order ALE ADER Finite Volume Schemes on Unstructured Meshes
14 / 19
3. Validation
The Euler equations
The Sedov Problem
Exact solution ALE WENO (O3)
6
5
4
rho
3
2
1
0
-1
0
0.2
0.4
0.6
0.8
1
1.2
r
3rd order accurate numerical results for the Sedov problem.
Walter Boscheri
High Order ALE ADER Finite Volume Schemes on Unstructured Meshes
14 / 19
3. Validation
The Euler equations
The Sedov Problem
3rd order accurate numerical results for the Sedov problem.
Walter Boscheri
High Order ALE ADER Finite Volume Schemes on Unstructured Meshes
14 / 19
3. Validation
The Baer–Nunziato Model
Outline 1 2
3
4
Introduction Numerical Method Hyperbolic Balance Laws High Order WENO Reconstruction on Unstructured Meshes Local Space-Time DG Predictor on Moving Meshes Mesh Motion Solution Algorithm Validation The Euler equations The Baer–Nunziato Model Conclusions and Outlook
Walter Boscheri
High Order ALE ADER Finite Volume Schemes on Unstructured Meshes
14 / 19
3. Validation
The Baer–Nunziato Model
The Baer-Nunziato Model of Compressible Multi-Phase Flows ∂ ∂t
∂ (φ ρ u ) + ∇ · (φ ρ u u ) + ∇φ p = p ∇φ − λ (u − u ) , 1 1 1 1 1 1 1 1 1 1 I 1 2 ∂t ∂ (φ ρ E ) + ∇ · ((φ ρ E + φ p ) u ) = −p ∂ φ − λ u · (u − u ) , 1 1 1 1 1 1 1 1 t 1 1 I I 1 2 ∂t ∂ (φ ρ ) + ∇ · (φ ρ u ) = 0, 2 2 2 2 2 ∂t ∂ (φ ρ u ) + ∇ · (φ ρ u u ) + ∇φ p = p ∇φ − λ (u − u ) , 2 2 2 2 2 2 2 2 2 2 I 2 1 ∂t ∂ (φ ρ E ) + ∇ · ((φ ρ E + φ p ) u ) = p ∂ φ − λ u · (u − u ) , 2 2 2 2 2 2 2 2 t 1 2 I I 2 1 ∂t ∂ φ + u ∇φ = ν(p − p ). 1 1 2 I ∂t 1 (φ1 ρ1 ) + ∇ · (φ1 ρ1 u1 ) = 0,
Walter Boscheri
High Order ALE ADER Finite Volume Schemes on Unstructured Meshes
15 / 19
3. Validation
The Baer–Nunziato Model
The Baer-Nunziato Model of Compressible Multi-Phase Flows Definitions k φk uk subscripts 1 subscripts 2 subscripts I ek Ek = ek + 21 uk 2
→ → → → → → → →
phase volume fraction of phase k velocity vector of phase k solid phase gas phase interface internal energy specific total energy of phase k
Assumptions φ1 + φ2 = 1 uI = u1 pI = p2
volume fractions must sum to the unity interface velocity is the solid phase velocity interface pressure is the gas phase pressure a
a
according to the original formulation of the model (M.R. Baer and J.W. Nunziato, J. of Multiphase Flows 1986) Walter Boscheri
High Order ALE ADER Finite Volume Schemes on Unstructured Meshes
15 / 19
3. Validation
The Baer–Nunziato Model
Riemann problems Initial condition for the left state (L) and the right state (R) for the Riemann problems solved with the Baer-Nunziato model. Values for γk , πk and the final time tf are also given. ρs
RP1: L R RP2 : L R RP3 : L R RP4 : L R Walter Boscheri
us ps ρg ug pg φs tf γs = 1.4, πs = 0, γg = 1.4, πg = 0, λ = µ = 0 1.0 0.0 1.0 0.5 0.0 1.0 0.4 0.10 2.0 0.0 2.0 1.5 0.0 2.0 0.8 γs = 3.0, πs = 100, γg = 1.4, πg = 0, λ = µ = 0 800.0 0.0 500.0 1.5 0.0 2.0 0.4 0.10 1000.0 0.0 600.0 1.0 0.0 1.0 0.3 γs = 1.4, πs = 0, γg = 1.67, πg = 0, λ = 103 , µ = 102 1.0 0.0 1.0 1.0 0.0 1.0 0.99 0.2 0.125 0.0 0.1 0.125 0.0 0.1 0.01 γs = 2.0, πs = 0, γg = 1.4, πg = 0, λ = 103 , µ = 102 1.0 0.0 2.0 1.0 0.0 2.0 0.99 0.2 0.125 0.0 0.1 0.125 0.0 0.1 0.01 High Order ALE ADER Finite Volume Schemes on Unstructured Meshes
16 / 19
3. Validation
The Baer–Nunziato Model
Riemann problems RP1 2.2
Reference solution ALE WENO (O3)
2
1.8
rhos
1.6
1.4
1.2
1
0.8 -0.5
-0.4
-0.3
-0.2
-0.1
0
0.1
0.2
0.3
0.4
0.5
x
3rd order accurate density distribution for the Riemann problem RP1.
Walter Boscheri
High Order ALE ADER Finite Volume Schemes on Unstructured Meshes
16 / 19
3. Validation
The Baer–Nunziato Model
Riemann problems RP2 Reference solution ALE WENO (O3)
1050
1000
rhos
950
900
850
800
750 -0.5
-0.4
-0.3
-0.2
-0.1
0
0.1
0.2
0.3
0.4
0.5
x
3rd order accurate density distribution for the Riemann problem RP2.
Walter Boscheri
High Order ALE ADER Finite Volume Schemes on Unstructured Meshes
16 / 19
3. Validation
The Baer–Nunziato Model
Riemann problems RP3 1.1
Exact solution ALE WENO (O3)
1 0.9 0.8
rho
0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 -0.5
-0.4
-0.3
-0.2
-0.1
0
0.1
0.2
0.3
0.4
0.5
x
3rd order accurate density distribution for the Riemann problem RP3.
Walter Boscheri
High Order ALE ADER Finite Volume Schemes on Unstructured Meshes
16 / 19
3. Validation
The Baer–Nunziato Model
Riemann problems RP4 Exact solution ALE WENO (O3)
1.2 1.1 1 0.9 0.8
rho
0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 -0.5
-0.4
-0.3
-0.2
-0.1
0
0.1
0.2
0.3
0.4
0.5
x
3rd order accurate density distribution for the Riemann problem RP4.
Walter Boscheri
High Order ALE ADER Finite Volume Schemes on Unstructured Meshes
16 / 19
3. Validation
The Baer–Nunziato Model
Explosion problems EP1 2.2
Reference solution ALE WENO (O3)
2.1 2 1.9 1.8 1.7
rhos
1.6 1.5 1.4 1.3 1.2 1.1 1 0.9 0.8
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
x
3rd order accurate density distribution for the explosion problem EP1. Walter Boscheri
High Order ALE ADER Finite Volume Schemes on Unstructured Meshes
17 / 19
3. Validation
The Baer–Nunziato Model
Explosion problems EP2 1050
Reference solution ALE WENO (O3)
1000
rhos
950
900
850
800
750
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
x
3rd order accurate density distribution for the explosion problem EP2. Walter Boscheri
High Order ALE ADER Finite Volume Schemes on Unstructured Meshes
17 / 19
4. Conclusions and Outlook
Conclusions and Outlook
Conclusions we showed high order ALE WENO FV schemes on tetrahedral meshes; two different node solvers have been used; application to nonconservative hyperbolic balance laws with stiff source terms.
Walter Boscheri
High Order ALE ADER Finite Volume Schemes on Unstructured Meshes
18 / 19
4. Conclusions and Outlook
Conclusions and Outlook
Conclusions we showed high order ALE WENO FV schemes on tetrahedral meshes; two different node solvers have been used; application to nonconservative hyperbolic balance laws with stiff source terms. Outlook extension to moving curved elements; extension to high order DG direct ALE schemes; application to free surface flows.
Walter Boscheri
High Order ALE ADER Finite Volume Schemes on Unstructured Meshes
18 / 19
Thank you!
[email protected] [1] W. Boscheri and M. Dumbser Arbitrary–Lagrangian–Eulerian One–Step WENO Finite Volume Schemes on Unstructured Triangular Meshes. Communications in Computational Physics, 14: 1174-1206, 2013. [2] M. Dumbser and W. Boscheri High-order unstructured Lagrangian one–step WENO finite volume schemes for non–conservative hyperbolic systems: Applications to compressible multi–phase flows. Computers and Fluids, 86: 405-432, 2013. [3] W. Boscheri and M. Dumbser A Direct Arbitrary-Lagrangian-Eulerian ADER-WENO Finite Volume Scheme on Unstructured Tetrahedral Meshes for Conservative and Nonconservative Hyperbolic Systems in 3D. Journal of Computational Physics, 275: 484-523, 2014.