Scribd
Upload
33 views

Diffusion

Artificial diffusion and flux splitting are equivalent techniques used in computational fluid dynamics to add numerical dissipation that stabilizes simulations. Artificial diffusion adds diffusion terms to discretized governing equations, smoothing scalar and vector fields. Flux splitting decomposes the Jacobian matrix into positive and negative components, equivalent to adding second-order diffusion. Both techniques add numerical dissipation that is analogous to physical diffusion and helps prevent numerical instabilities.

Uploaded by

Tarak Nath Nandi
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
33 views

Diffusion

Artificial diffusion and flux splitting are equivalent techniques used in computational fluid dynamics to add numerical dissipation that stabilizes simulations. Artificial diffusion adds diffusion terms to discretized governing equations, smoothing scalar and vector fields. Flux splitting decomposes the Jacobian matrix into positive and negative components, equivalent to adding second-order diffusion. Both techniques add numerical dissipation that is analogous to physical diffusion and helps prevent numerical instabilities.

Uploaded by

Tarak Nath Nandi
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 3

1

1.1

Flux Splitting, Diusion and Articial Dissipation


What is Diusion?

Diusion is the tendency for certain quantity of a medium to spread out evenly. It drives the quantity of any medium from high density locations towards low density locations. It drives in the direction of the opposite of the gradient. In other words, it smoothes scalar or vector elds.

1.2

What is Articial Diusion

Articial diusion is introduced by adding additional terms to the ux of the discretized equations. In our example, the governing convective (non-dissipated) equations from Euler in one dimension is w f (w) + =0 t x which can be written in the ux form wn+1 = wn t hi+ 1 hi 1 2 2 (1.2) (1.1)

The uxes hi+ 1 contains the convection part and the diusion part, where in 2 the 1st order scalar diusion setting, the articial dissipation is usually dened as below hi+ 1 2 =
1 fi+ 1 di+ 2 2

convection

fi+1 + fi 2

(2)

(wi+1 wi )
dif f usion

(1.3)

1.3

Equivalence Between Articial Diusion and Flux Splitting


w w +A =0 t x

The non-conservative form of the equation is (1.4) (1.5) where the Jacobian is f (w) x A = N 1 N A= 1 (1.6) (1.7)

Flux split is done by splitting the Jacobians. A+ = N 1 + N A = N 1 N this is equivalent to below A = A2 I 2 (1.10) (1.8) (1.9)

Using second order diusion without active pressure sensor switches,


1 hj+ 2 =

1 (fj+1 + fj ) 2

(2)

(wj+1 + wj )

(1.11)

therefore, hj+ 1 2 wj 1 fj + (2) I 2 wj 1 = Aj + 2 (2) I 2 = A+ j =

(1.12)

and hj+ 1 2 wj+1 = = = similarly, hj 1 2 wj and hj 1 2 wj1 = A+ j1 (1.15) = A j (1.14) 1 fj+1 (2) I 2 wj+1 1 Aj+1 2 (2) I 2 A j+1

(1.13)

Overall, for the ADI type methods, where the tridiagonal Jacobian matrices looks like below, D C 0 0 0 0 w2 A D C 0 w3 0 0 ... 0 A D C w4 0 0 (1.16) = R.H.S. 0 0 A D C w5 0 ... . . . .. .. .. . . . . . . . . . 2

the diusion of the dierence of the uxes at j + 1 and j written in rst order expansion of Taylor series,
A D C
1 hj+ 1 hj 2 2

hj+ 1 hj 1 2 2 wj1

wj1 + = = =

hj+ 1 hj 1 2 2 wj

wj +

wj+1

wj+1

A wj+1 A wj + A+ wj A+ wj1 j+1 j j j1


+ Dx A + Dx A+ w

A+ wj1 + A+ A wj + A wj+1 j j j+1 j1


A D C

(1.17)

this results in A + A+ 2 2 A+ 2 0 0 . . .

A 3 A3 + A+ 3 A+ 3 0 . . .

0 A 4 A + A+ 4 4 A+ 4

0 0 A 5 A + A+ 5 5 .. .

0 0 0 A 6 .. .

0 0 0 0 .. .

w2 w3 w4 w5 . . .

... = R.H.S. ...

(1.18) Therefore, ux splitting is equivalent to added diusion.

You might also like