• Keine Ergebnisse gefunden

Construction of semi-implicit methods for magnetohydrodynamics based on multiple scale asymptotics

N/A
N/A
Protected

Academic year: 2021

Aktie "Construction of semi-implicit methods for magnetohydrodynamics based on multiple scale asymptotics"

Copied!
33
0
0

Wird geladen.... (Jetzt Volltext ansehen)

Volltext

(1)

magnetohydrodynamics based on multiple scale asymptotics

Friedemann Kemm IAG Universit¨at Stuttgart kemm@iag.uni-stuttgart.de

Claus-Dieter Munz IAG Universit¨at Stuttgart munz@iag.uni-stuttgart.de

(2)

The non dimensional equations in primitive variables

ρt + uρx + ρux = 0 , ut + uux + 1

M2 px

ρ + 1 Av2

1

2ρ(B2)x = 0 , vt + uvx − 1

Av2 B1

ρ B2x = 0 , wt + uwx − 1

Av2 B1

ρ B3x = 0 , B2t + B2ux − B1vx + uB2x = 0 , B3t + B3ux − B1wx + uB3x = 0 ,

pt + γpux + upx = 0

M := v0

a0 , Av := v0 cA,0 .

(3)

The eigenvalues

u , u ± cs , u ± cA , u ± cf with

cA = 1

Avb1 , cf /s = 1

√2 s

1

M2a2 + 1

Av2b2 ±

r 1

M2a2 + 1

Av2b22

− 4 · 1

M2a2 1

Av2b21 , the abbreviations b1 = p

B12/ρ, b2 = B2/ρ and the acoustic speed a = γp/ρ.

(4)

The asymptotic Expansion

For small Mach numbers:

q(x, t; M) = q(0)(x, ξ, t) + M q(1)(x, ξ, t) + M2q(2)(x, ξ, t) + . . . with

q(i) : R × R × [0,∞) → Rm

(x, ξ, t) 7→ q(i)(x, ξ, t)

(i = 0,1,2, . . .)

and

ξ = M x .

(5)

For small Alfven numbers:

q(x, t; Av) = q(0)(x, ξ, t) + Av q(1)(x, ξ, t) + Av2q(2)(x, ξ, t) + . . . with

q(i) : R × R × [0,∞) → Rm

(x, ξ, t) 7→ q(i)(x, ξ, t)

(i = 0,1,2, . . .)

and

ξ = Av x . With the restriction

q(i)(x, ξ, t)

x → 0 for x → ∞ ∀i, t .

(6)

The velocity equation

Orders -2 and -1:

p(0)x = 0 , p(0)ξ + p(1)x = 0 .

(7)

The velocity equation

Orders -2 and -1:

p(0)x = 0 , p(0)ξ + p(1)x = 0 .

0 = 1

|I| Z

I

p(0)ξ (ξ, t) dx + 1

|I| Z

I

p(1)x (x, ξ, t) dx

= p(0)ξ + 1

|I| Z

I

p(1)x (x, ξ, t) dx

= p(0)ξ + 1

|I|

p(1)b

a , and therefore

p(0)ξ = 0 f¨ur |I| → ∞ .

(8)

The behavior and the meaning of the pressure variables

The meaning:

p(0)= p(0)(t) thermodynamic , p(1)= p(1)(ξ, t) acoustic ,

p(2)= p(2)(x, ξ, t) incompressible .

(9)

The behavior and the meaning of the pressure variables

The meaning:

p(0)= p(0)(t) thermodynamic , p(1)= p(1)(ξ, t) acoustic ,

p(2)= p(2)(x, ξ, t) incompressible .

The time behavior of p(0):

p(0)t + 1

|I|γp(0) ub

a = 0 .

(10)

The long wave length acoustic:

(ρu)t + p(1)ξ = 0 , p(1)t + γp(0)uξ = 0 .

(11)

The zero Mach number limit

The pressure is the only variable, whose higher order terms we must take into consideration. So we skip the upper indices of the others.

ρt + ρux + ρxu = 0 , ut + uux + 1

ρp(2)x + 1

2Av2ρ(B2)x = −1

ρp(1)ξ , vt + uvx − 1

Av2ρB1B2x = 0 , wt + uwx − 1

Av2ρB1B3x = 0 , B2t + B2ux − B1vx + uB2x = 0 , B3t + B3ux − B1wx + uB3x = 0 , p(0)t + γp(0)ux = 0 .

(12)

The behavior of the magnetic induction variables

The meaning:

Bi(0) = Bi(0)(t) , Bi(1) = Bi(1)(ξ, t) ,

Bi(2) = Bi(2)(x, ξ, t) . (i = 2,3)

(13)

The behavior of the magnetic induction variables

The meaning:

Bi(0) = Bi(0)(t) , Bi(1) = Bi(1)(ξ, t) ,

Bi(2) = Bi(2)(x, ξ, t) . (i = 2,3)

The time behavior of the leading order terms:

B2,t(0) + B2(0) 1

|I|

ub

a − B1 1

|I|

vb

a = 0 , B3,t(0) + B3(0) 1

|I|

ub

a − B1 1

|I|

wb

a = 0 .

(14)

Evolution equations for long wave magnetic phenomena:

(ρu)t + B2(0)B2,ξ(1) + B3(0)B3,ξ(1)

= 0 , (ρv)t − B1B2,ξ(1) = 0 , (ρw)t − B1B3,ξ(1) = 0 , B2,t(1) + B2(0)uξ − B1vξ = 0 , B3,t(1) + B3(0)uξ − B1vξ = 0 .

(15)

The zero Alfven number limit

ρt + ρux + ρxu = 0 , ut + uux + 1

M2 1

ρpx + 1

ρ(B2)(2)x = −1

ρ(B2)(1)ξ , vt + uvx − 1

ρB1B2,x(2) = 1

ρB1B2,ξ(1) , wt + uwx − 1

ρB1B3,x(2) = 1

ρB1B3,ξ(1) , B2,t(0) + B2(0)ux − B1vx = 0 ,

B3,t(0) + B3(0)ux − B1wx = 0 , pt + γpux + pxu = 0 .

(16)

The hyperbolic-elliptic splitting

ρt + uρx + ρux = 0 , ut + uux + 1

M2 px

ρ + 1

2Av2ρ(B2)x = 0 , vt + uvx − B1

Av2ρB2x = 0 , wt + uwx − B1

Av2ρB3x = 0 , B2t + B2ux − B1vx + uB2x = 0 , B3t + B3ux − B1wx + uB3x = 0 ,

pt + γpux + upx = 0

M := v0

a0 , Av := v0 cA,0 .

(17)

A semi-implicit method;

System with fast waves

For low Mach numbers: Discretization with backward differences in time:

ρn+1 − ρn

∆t + ρn+1un+1x = 0 , un+1 − un

∆t + 1 M2

1

ρn+1pn+1x = 0 , pn+1 − pn

∆t + γpn+1un+1x = 0 .

(18)

The scheme:

Step 1 Choose estimates ρ and p for density and pressure.

(19)

The scheme:

Step 1 Choose estimates ρ and p for density and pressure.

Step 2 Compute an estimate for the velocity via u = un − ∆t

M2ρpx .

(20)

Step 3 If the estimates are good enough → ready. If not then make correction Ansatz:

un+1 = u + δu , pn+1 = p + δp , ρn+1 = ρ + δρ write δu in terms of δp, ρ, M and ∆t:

δu = − ∆t

M2ρδpx and put this into the pressure equation

pn+1 − pn

∆t + γpun+1x = 0 .

(21)

This gives the “elliptic” condition

δp − ∆t2

M2 γp

1

ρδpx

x

= pn − p − ∆t γpux

for the pressure correction. Solve this equation numerically.

(22)

This gives the “elliptic” condition

δp − ∆t2

M2 γp

1

ρδpx

x

= pn − p − ∆t γpux

for the pressure correction. Solve this equation numerically.

Step 4 Compute corrected pressure, velocity and density. Use these values as the new estimates and return to Step 3.

(23)

For very small Mach numbers put the multiple pressure variables into the scheme.

The “elliptic” condition then gets the form

δp(2) − ∆t2

M2 γp

1

ρδp(2)x

x

= pn − p − ∆t γpux . The corrected pressure is then computed by

pn+1 = p + M2δp(2) .

(24)

For low Alfven numbers: Analogous precedure. Now the “elliptic” condition is δB2 − ∆t2

Av2

B2(B2 + δB2) + B12

ρ δB2,x

x

− ∆t2 Av2

B2(B3 + δB3)

ρ δB3,x

x

− ∆t2

Av2B2 δB2B2,x + δB3B3,x

= B2n − B2 − ∆t B2ux − B1vx , δB3 − ∆t2

Av2

B3(B3 + δB3) + B12

ρ δB3,x

x

− ∆t2 Av2

B3(B2 + δB2)

ρ δB2,x

x

− ∆t2

Av2B3 δB3B3,x + δB2B2,x

= B3n − B3 − ∆t B3ux − B1vx .

(25)

If we use multiple induction variables it is

δB2(2) − ∆t2

B2B2 + B12

ρ δB2,x(2)

x

− ∆t2

B2B3

ρ δB3,x(2)

x

− ∆t2B2 δB2(2)B2,x + δB3(2)B3,x

= B2n − B2 − ∆t B2ux − B1vx , δB3(2) − ∆t2

B3B3 + B12

ρ δB3,x(2)

x

− ∆t2

B3B2

ρ δB2,x(2)

x

− ∆t2B3 δB3(2)B3,x + δB2(2)B2,x

= B3n − B3 − ∆t B3ux − B1vx .

(26)

In the case of multiple induction variables the corrected magnetic field is computed by

B2n+1 = B2 + Av2δB2(2) , B3n+1 = B3 + Av2δB3(2) .

(27)

Conclusions

• Non dimensional equations

(28)

Conclusions

• Non dimensional equations

• Multiple scale asymptotics (one space dimension)

(29)

Conclusions

• Non dimensional equations

• Multiple scale asymptotics (one space dimension)

• Elliptic-hyperbolic splitting

(30)

Conclusions

• Non dimensional equations

• Multiple scale asymptotics (one space dimension)

• Elliptic-hyperbolic splitting

• Semi-implicit method:

(31)

Conclusions

• Non dimensional equations

• Multiple scale asymptotics (one space dimension)

• Elliptic-hyperbolic splitting

• Semi-implicit method:

? slow waves explicit

(32)

Conclusions

• Non dimensional equations

• Multiple scale asymptotics (one space dimension)

• Elliptic-hyperbolic splitting

• Semi-implicit method:

? slow waves explicit

? fast waves implicit

(33)

Conclusions

• Non dimensional equations

• Multiple scale asymptotics (one space dimension)

• Elliptic-hyperbolic splitting

• Semi-implicit method:

? slow waves explicit

? fast waves implicit

• correction of pressure or magnetic induction via elliptic equation

Referenzen

ÄHNLICHE DOKUMENTE

However, mainly in astrophysics [2 – 4], space physics [5, 6], and plasma physics [7,8], there arises very often the problem of the motion of a charged particle when the intensity

Therefore five different alloys with carbon contents in a range from 0.001% up to 0.8% are sampled in the earliest stage of melting, the refining period, in the VIDP

Methodologically, we make a three-fold contribution: (i) we suggest a transformation from textual-based disease similarity values to confidence values that are learned automatically

(d) ambient, diffuse/specular, shadow (e) ambient, diffuse/specular, shadow, textures (f) standard local illumination model Figure 5: Illumination components for the

Our controller design approach is based on the first main result in this paper, Theorem 4, below, which characterizes ISpS of an event-based closed loop system by means of

While plateau airway pressure alone is an unreliable estimate of lung overdistension inspiratory transpulmonary pressure (PL) is an important parameter to reflect it in patients

Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.. The original article can be found online

Together with a suitable asymptotic condition at the cusp i∞ given in terms of multiple zeta values [31, 32], the differential equation yields a canonical decompo- sition of