Arts

2.
Variational Formulation and Numerical Scheme

2.1. Variational Formulation
We adopt the conventional Sobolev space Wk,p(Ω)Wk,p(Ω) of functions
defined on ΩΩ for k≥0k≥0 and 1≤p≤∞1≤p≤∞, and
denote Lp(Ω)=W0,p(Ω)Lp(Ω)=W0,p(Ω) and Hk(Ω)=Wk,2(Ω)Hk(Ω)=Wk,2(Ω)
. Then we take the notations W1,p0(Ω)W01,p(Ω) representing the space of
functions in W1,p(Ω)W1,p(Ω) with zero traces on the boundary ∂Ω∂Ω, and
naturally H10(Ω):=W1,20(Ω)H01(Ω):=W01,2(Ω). The corresponding vector
spaces are given by
Lp(Ω)=[Lp(Ω)]d,W1,p0(Ω)=[W1,p0(Ω)]d,H1(Ω)={v∈H1(Ω):v×n|
∂Ω=0},Wk,p(Ω)=[Wk,p(Ω)]d,H10(Ω)=W1,20(Ω),Lp(Ω)=[Lp(Ω)]d,Wk,p(Ω)=[W
k,p(Ω)]d,W01,p(Ω)=[W01,p(Ω)]d,H01(Ω)=W01,2(Ω),H1(Ω)={v∈H1(Ω):v×n|
∂Ω=0},
where d is the dimension of space. As usual (⋅,⋅)(⋅,⋅) denotes the inner
product of L2(Ω)L2(Ω).
Consequently, we obtain that the exact solutions of (1.1)-(1.3) satisfy
μ(∂tH,w)+σ−1(∇×H,∇×w)−μ(u×H,∇×w)=0,μ(∂tH,w)+σ−1(∇×H,∇×w)
−μ(u×H,∇×w)=0,(2.1)
(∂tu,v)+b(u,u,v)+ν(∇u,∇v)−(p,∇⋅v)+μ(H×(∇×H),v)=0,(∂tu,v)+b(u,u,v)
+ν(∇u,∇v)−(p,∇⋅v)+μ(H×(∇×H),v)=0,(2.2)
(∇⋅u,q)=0,(∇⋅u,q)=0,(2.3)
for any test functions (w,v,q)∈(H1(Ω),H10(Ω),L20(Ω))
(w,v,q)∈(H1(Ω),H01(Ω),L02(Ω)). The trilinear operator b(⋅,⋅,⋅)b(⋅,⋅,⋅) is
defined as
b(u,v,w):=(u⋅∇v,w)+12((∇⋅u)v,w) =12[(u⋅∇v,w)−(u⋅∇w,v)],
∀u,v,w∈H10(Ω),b(u,v,w):=(u⋅∇v,w)+12((∇⋅u)v,w) =12[(u⋅∇v,w)−
(u⋅∇w,v)], ∀u,v,w∈H01(Ω),(2.4)
and then it is natural that
b(u,v,v)=0, ∀u,v∈H10(Ω).b(u,v,v)=0, ∀u,v∈H01(Ω).(2.5)
In order to treat the nonlinear term explicitly and ensure the energy stability,
we introduce the function of “zero energy contribution”, which is defined as
Qt=−μ(u×H,∇×H)+μ(H×(∇×H),u)+b(u,u,u)≡0, and
Q0=1.Qt=−μ(u×H,∇×H)+μ(H×(∇×H),u)+b(u,u,u)≡0, and Q0=1.(2.6)
and then rewrite the Equations (2.1)-(2.3) as follows
μ(∂tH,w)+σ−1(∇×H,∇×w)−μQ(u×H,∇×w)=0,μ(∂tH,w)+σ−1(∇×H,∇×w)
−μQ(u×H,∇×w)=0,(2.7)
(∂tu,v)+Qb(u,u,v)+ν(∇u,∇v)−(p,∇⋅v)+μQ(H×(∇×H),v)=0,(∂tu,v)
+Qb(u,u,v)+ν(∇u,∇v)−(p,∇⋅v)+μQ(H×(∇×H),v)=0,(2.8)
(∇⋅u,q)=0,(∇⋅u,q)=0,(2.9)
for any test functions (w,v,q)∈(H1(Ω),H10(Ω),L20(Ω))
(w,v,q)∈(H1(Ω),H01(Ω),L02(Ω)). The variational Formulation (2.7)-(2.9) will
be studied in this paper later.
2.2. Numerical Discretization
In this subsection, we propose a fully discrete decoupled finite element
method for solving the system (1.1)-(1.3). Let Ihℑh denote a quasi-uniform
partition of ΩΩ into tetrahedrons KjKj in R3ℝ3 (or triangles in R2ℝ2 ), j=1,2,
…,Mj=1,2,…,M, with mesh size h=max1≤j≤M{diam Kj}h=max1≤j≤M{diam
Kj}. To approximate uu and p in the system (1.1)-(1.3), we introduce the
Taylor-Hood finite element space Xh×MhXh×Mh, defined by
Xh={lh∈H10(Ω):lh|Kj∈Pr(Kj)},Mh={qh∈L2(Ω):qh|Kj∈Pr−1(Kj),∫Ω
qhdx=0},Xh={lh∈H01(Ω):lh|Kj∈Pr(Kj)},Mh={qh∈L2(Ω):qh|Kj∈Pr−1(Kj),∫Ω
qhdx=0},
for any integer r≥2r≥2, where Pr(Kj)Pr(Kj) is the space of polynomials with
degree r on KjKj for all Kj∈IhKj∈ℑh and Pr(Kj):=[Pr(Kj)]dPr(Kj):=[Pr(Kj)]d.
To approximate the magnetic field HH, we introduce the finite element
space ShSh defined by
Sh={wh∈H1(Ω):wh|Kj∈Pr(Kj)}.Sh={wh∈H1(Ω):wh|Kj∈Pr(Kj)}.
Let {tn=nτ}Nn=0{tn=nτ}n=0N be a uniform partition of the time interval [0,T]
[0,T], and τ=T/Nτ=T/N denotes the temporal step size.
Furthermore, vnvn represents the value of v(x,tn)v(x,tn), and we adopt the
abbreviation
vñ+1:=2vn−vn−1.vñ+1:=2vn−vn−1.
Additionally, we adopt the following notations:
Dtvn+1h=32vn+1h−2vnh+12vn−1hτ,Dtvhn+1=32vhn+1−2vhn+12vhn−1τ,
where v could be an arbitrary function except
that uˆn+1hu^hn+1, Rhun+1ˆRhun+1^ and eˆn+1ueûn+1, for which we
denote by
Dtuˆn+1h=3uˆn+1h−4unh+un−1h2τ,DtRhun+1ˆ=3Rhun+1ˆ−4Rhun+Rhun−12τ,Dteˆn+1u=
3eˆn+1u−4enu+en−1u2τ.Dtu^hn+1=3u^hn+1−4uhn+uhn−12τ,DtRhun+1^=3Rhun+
1^−4Rhun+Rhun−12τ,Dteûn+1=3eûn+1−4eun+eun−12τ.
Thus, we propose a fully decoupled discrete scheme with FEM and BDF for the
incompressible MHD Equations (1.1)-(1.3) that,
find Qn+1hQhn+1 and (Hn+1h,un+1h,uˆn+1h,pn+1h)∈(Sh,Xh,Xh,Mh)
(Hhn+1,uhn+1,u^hn+1,phn+1)∈(Sh,Xh,Xh,Mh) such that
μ(DtHn+1h,wh)+σ−1(∇×Hn+1h,∇×wh)+σ−1(∇⋅Hn+1h,∇⋅wh)
−μQn+1h(uñ+1h×Hñ+1h,∇×wh)=0,μ(DtHhn+1,wh)+σ−1(∇×Hhn+1,∇×wh)
+σ−1(∇⋅Hhn+1,∇⋅wh) −μQhn+1(u˜hn+1×H˜hn+1,∇×wh)=0,(2.10)
(Dtuˆn+1h,vh)+Qn+1hb(uñ+1h,uñ+1h,vh)+ν(∇uˆn+1h,∇vh)−(pnh,∇⋅vh)
+μQn+1h(Hñ+1h×(∇×Hñ+1h),vh)=0,(Dtu^hn+1,vh)
+Qhn+1b(u˜hn+1,u˜hn+1,vh)+ν(∇u^hn+1,∇vh)−(phn,∇⋅vh)
+μQhn+1(H˜hn+1×(∇×H˜hn+1),vh)=0,(2.11)
(un+1h−uˆn+1hτ,sh)−23(pn+1h−pnh,∇⋅sh)=0,(uhn+1−u^hn+1τ,sh)
−23(phn+1−phn,∇⋅sh)=0,(2.12)

Arts

Uploaded by

Copyright:

Available Formats

Arts

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Arts

Uploaded by

Copyright:

Available Formats

2.

Variational Formulation and Numerical Scheme

that uˆn+1hu^hn+1, Rhun+1ˆRhun+1^ and eˆn+1ue^un+1, for which we

You might also like