The shifted box scheme for scalar transport problems Bertil - - PowerPoint PPT Presentation

the shifted box scheme for scalar transport problems
SMART_READER_LITE
LIVE PREVIEW

The shifted box scheme for scalar transport problems Bertil - - PowerPoint PPT Presentation

The shifted box scheme for scalar transport problems Bertil Gustafsson Uppsala University and Stanford University Joint work with: Yaser Khalighi, Stanford .


slide-1
SLIDE 1

The shifted box scheme for scalar transport problems

Bertil Gustafsson Uppsala University and Stanford University Joint work with: Yaser Khalighi, Stanford

slide-2
SLIDE 2

.

slide-3
SLIDE 3

Scalar transport equation, incompressible flow

✂ ✄✆☎
✞ ✂ ✄✆✟
✠☛✡ ☞ ✌ ☎ ✞ ✂ ✟ ✠ ✡ ☞

Velocity components

☎ ✡ ☎ ✄✆✍ ✌ ✎ ✌✑✏ ✝ ✌ ✟ ✡ ✟ ✄✆✍ ✌ ✎ ✌✑✏ ✝

Energy conservation:

✒ ✒ ✏ ✔✓ ✒ ✍ ✒ ✎ ✡ ☞

Alternative formulation:

✂ ☎
✂ ✟
  • ✠☛✡
slide-4
SLIDE 4

Requirements on numerical method:

  • 1. Very fast algorithm
  • 2. No stability limit on the time step
  • 3. Energy conservation (no loss of energy)
  • 4. No unphysical oscillations

Standard centered finite difference or finite volume method, trapezoidal rule in time: (2), (3)

Implicit upwind scheme: (2), (4)

Method of characteristics: (1), (2), (3), (4)

Method of characteristics on a grid (interpolation): (1), (2), (4) The box scheme?

slide-5
SLIDE 5

Model problem 1-D

✂ ✖ ✗ ✄ ✄ ☎
✞ ✂ ☎
✝ ✡ ☞ ✌ ☞ ✘ ✍ ✘ ✖ ✌ ☞ ✘ ✏ ✌
☞ ✌✑✏ ✝ ✡ ✙ ✄ ✏ ✝ ✌
✍ ✌ ☞ ✝ ✡ ✚ ✄✆✍ ✝✜✛ ☎ ✡ ☎ ✄✆✍ ✝✣✢ ☞ ✒ ✒ ✏ ✤ ✤
✤ ✓ ✡ ✥ ☎ ✄ ✖ ✝ ✤
✖ ✌ ✏ ✝ ✤ ✓ ✌
slide-6
SLIDE 6

The box scheme

x t

slide-7
SLIDE 7
✡ ✖ ✗ ✄
★ ✩ ✂
✝ ✌ ✪
✡ ✖ ✫ ✄
★ ✩ ✥
✝✜✛ ✦ ✔✬ ★ ✩ ✧ ✂ ✭ ✮ ✄ ✪ ✄ ☎ ✬ ★ ✩ ✧ ✔✬ ★ ✩ ✧ ✝ ✂ ✄ ✦ ☎ ✬ ★ ✩ ✧ ✝ ✪ ✔✬ ★ ✩ ✧ ✝ ✡ ✦
✧ ✥ ✭ ✮ ✄ ✪ ✄ ☎ ✬ ✧ ✯✬ ✧ ✝ ✂ ✄ ✦ ☎ ✬ ✧ ✝ ✪
✧ ✝ ✔✬ ★ ✩ ✰ ✡ ✙ ✬ ★ ✩
✧✱✡ ✚ ✧
slide-8
SLIDE 8

Box scheme vs standard Crank-Nicholson: Parasitic solution?

✒ ☎ ✒ ✍ ✡ ☞

Crank-Nicholson

☎ ✧ ★ ✩ ✥ ☎ ✧✳✲ ✩ ✡ ☞ ✴ ☎ ✧ ✡ ✵ ✩ ✂ ✵ ✓ ✄ ✥ ✖ ✝ ✧

Box scheme

☎ ✧ ★ ✩ ✥ ☎ ✧ ✡ ☞ ✴ ☎ ✧ ✡ ✵ ✩ ✡

const

slide-9
SLIDE 9

Accuracy? One step:

☎ ✄ ✍ ✌ ✫ ✝ ✡ ✶✸✷✹ ✺ ✞ ★ ✻ ✼

. Approximation of

✽ ✫

?

0.5 1 1.5 2 2.5 3 3.5 0.5 1 1.5 2 2.5 3 3.5 4

Box

✾ ✡ ☞ ✛ ✿

, Exact and Box

✾ ✡ ✖ ✛ ☞

, Box

✾ ✡ ✖ ✛ ✗

, C-N(4,2)

✾ ✡ ✖ ✛ ✗

, C-N(2,2)

✾ ✡ ✖ ✛ ✗
slide-10
SLIDE 10

Approximation at

✽ ✫ ✡ ❀ ❁ ✗

(4 points per wavelength) Phase speed for

☞ ✛ ❂ ✘ ✾ ✘ ✖ ✛ ❂

0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3

Box, C-N(4,2), C-N(2,2)

slide-11
SLIDE 11

Constant coefficient

, periodic solutions

✂ ☎
✡ ☞

Fourier transform

❃❄❅ ❆ ✗ ✄ ❇
★ ✩ ✥ ❇ ✯✬ ✝ ✂ ✾ ☎❈ ❅ ❉❊ ❆ ✗ ✄ ❇ ✯✬ ★ ✩ ✂ ❇ ✯✬ ✝ ✡ ☞ ✾ ✡ ✭ ✫ ✌ ❆ ✡ ✽ ✫ ❇ ❋✬ ★ ✩ ✡ ❃❄ ❅ ✄ ❆ ❁ ✗ ✝ ✥ ✾ ☎❈ ❅ ❉❊ ✄ ❆ ❁ ✗ ✝ ❃❄ ❅ ✄ ❆ ❁ ✗ ✝ ✂ ✾ ☎❈ ❅ ❉❊ ✄ ❆ ❁ ✗ ✝ ❇ ❋✬
★ ✩ ✤ ✡ ✤ ❇ ❋✬ ✤

Unconditional stability if

☎ ❍ ✡ ☞

. Note: problem with

☎ ✡ ☞ ✌ ❆ ✡ ❀

!

slide-12
SLIDE 12

Initial-boundary value problem:

☞ ✌✑✏ ✝ ✡ ✙ ✄ ✏ ✝ ✌
  • ✄✆✍
✌ ☞ ✝ ✡ ✚ ✄ ✍ ✝

Solve from left to right

t x

slide-13
SLIDE 13 ☎ ✡

const

✢ ☞

Energy conservation:

✤ ✤ ✦ ■✬ ★ ✩ ✤ ✤ ✓ ✻ ✂ ✤ ✤ ☎ ✭ ✗ ✪
★ ✩ ✤ ✤ ✓ ✻ ✡ ✤ ✤ ✦
✤ ✤ ✓ ✻ ✂ ✤ ✤ ☎ ✭ ✗ ✪
✤ ✤ ✓ ✻ ✥ ☎ ✭ ✗ ✄ ✤
❏ ✤ ✓ ✂ ✤ ■✬ ★ ✩ ❏ ✤ ✓ ✝✜✛

(1) Stability estimate

✤ ✤ ❋✬ ✤ ✤ ✻ ✘ ❑ ✓ ❁ ❑ ✩ ✤ ✤ ✚ ✤ ✤ ✻ ❑ ✩ ✡ ▲ ❉ ❊ ✄ ✾ ✓ ✌ ✖ ✝ ✌ ❑ ✓ ✡ ▲ ▼ ◆ ✄ ✾ ✓ ✌ ✖ ✝ ✛ ✾ ✡ ☎ ✭ ❁ ✫
slide-14
SLIDE 14 ☎ ✄ ✍ ✝✣❖ ☞ ✌ ✍ P ✘ ✍ ✘ ✍ ◗

?

★ ✩ ✧ ★ ✩ ✥
✧ ★ ✩ ✂ ❘✬ ★ ✩ ✧ ✥ ✯✬ ✧ ✂ ✾ ☎ ✄ ❘✬ ★ ✩ ✧ ★ ✩ ✥ ❘✬ ★ ✩ ✧ ✂ ✯✬ ✧ ✥ ✯✬ ✧ ✝ ✡ ☞
★ ✩ ✧ ★ ✩ ✡ ✥ ✖ ✥ ✾ ☎ ✖ ✂ ✾ ☎ ❘✬ ★ ✩ ✧ ✂ ❙ ✄
✧ ✝ ✌ ❚ ✡ ❚ ✰ ✄ ✫ ✝ ✌ ❚ ✰ ✂ ✖ ✌ ✛ ✛ ✛ ✌ ❚ ✩ ✄ ✫ ✝ ✾ ✡ ✭ ❁ ✫

Unstable!

slide-15
SLIDE 15

The shifted box scheme Add constant

to the velocity:

❯ ✁ ✂ ✖ ✗ ✄ ✄ ❑ ✂ ☎ ✝ ❯ ✝ ❱ ✂ ✖ ✗ ✄ ❑ ✂ ☎ ✝ ❯ ❱ ✡ ☞ ✛

True solution obtained by a shift

✍ ✌✑✏ ✬ ★ ✩ ✝ ✡ ❯ ✄✆✍ ✂ ❑ ✭ ✌ ✏ ✬ ★ ✩ ✝✜✛

Choose

❑ ✭ ✡ ✫ ✌

(or more generally

❑ ✭ ✡ ❲ ✫ ✌ ❲

integer)

slide-16
SLIDE 16

The shifted box scheme

x t

slide-17
SLIDE 17

0.2 0.4 0.6 0.8 1 −0.2 0.2 0.4 0.6 0.8 1 1.2

(a) Velocity

❳ ❨❬❩ ❭

0.2 0.4 0.6 0.8 1 0.2 0.4 0.6 0.8 1

(b) Initial function

❪ ❨ ❩ ❭
slide-18
SLIDE 18

Characteristics

x 0.4 0.6

slide-19
SLIDE 19

Original box scheme unstable

0.2 0.4 0.6 0.8 1 −0.5 0.5 1 1.5 2 2.5

(c) Original box scheme

0.2 0.4 0.6 0.8 1 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6

(d) Shifted box scheme

  • ✄✆✍
✌ ✏ ✝

after 18 steps,

❫ ✡ ✿ ☞ ☞ ✌ ✾ ✡ ✖
slide-20
SLIDE 20

0.2 0.4 0.6 0.8 1 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6

(e) t=2

0.2 0.4 0.6 0.8 1 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6

(f) t=3

✍ ✌✑✏ ✝ ✌ ❑ ✡ ✖ ✌ ✾ ✡ ✖ ✌ ❫ ✡ ✿ ☞ ☞
slide-21
SLIDE 21

0.2 0.4 0.6 0.8 1 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6

(g) t=2

0.2 0.4 0.6 0.8 1 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6

(h) t=3

  • ✄✆✍
✌ ✏ ✝ ✌ ❑ ✡ ☞ ✛ ❂ ✌ ✾ ✡ ✗ ✌ ❫ ✡ ✿ ☞ ☞
slide-22
SLIDE 22

0.5 1 1.5 2 2.5 3 0.002 0.004 0.006 0.008 0.01

✤ ✤
✏ ✝ ✤ ✤ ✻ ✌ ☞ ✘ ✏ ✘❵❴ ✌ ❫ ✡ ✿ ☞ ☞ ✌ ✾ ✡ ✗
slide-23
SLIDE 23

10

−4

10

−3

10

−2

10

−1

10

−6

10

−5

10

−4

10

−3

Grid Spacing Error

Periodic BC Inlet BC

Second order accuracy

slide-24
SLIDE 24

2 space dimensions

y x u > 0 v > 0

slide-25
SLIDE 25
✂ ✄ ☎
✞ ✂ ✄ ✟
✠☛✡ ☞ ✌ ☞ ✘ ✍ ✌ ✎ ❖ ✖ ✌ ☞ ✘ ✏ ✌
☞ ✌ ✎ ✌ ✏ ✝ ✡ ✙ ✩ ✄ ✎ ✌✑✏ ✝ ✌ ☞ ✘ ✎ ✘ ✖ ✌
✍ ✌ ☞ ✌ ✏ ✝ ✡ ✙ ✓ ✄✆✍ ✌ ✏ ✝ ✌ ☞ ✘ ✍ ✘ ✖ ✌
  • ✄✆✍
✌ ✎ ✌ ☞ ✝ ✡ ✚ ✄ ✍ ✌ ✎ ✝ ✛
slide-26
SLIDE 26

The box scheme

✦ ✞ ✦ ✠ ✄ ❘✬ ★ ✩ ✷ ✧ ✥ ✯✬ ✷ ✧ ✝ ✂ ✭ ✗❜❛ ✦ ✠ ✪ ✞ ❝ ✄ ☎
✬ ★ ✩ ✷ ✧ ✂ ✄✆☎
✬ ✷ ✧❵❞ ✂ ✦ ✞ ✪ ✠ ❝ ✄✆✟
✬ ★ ✩ ✷ ✧ ✂ ✄ ✟
✬ ✷ ✧ ❞❡ ✡ ☞ ✌

x y t

slide-27
SLIDE 27 ☎ ✢ ☞ ✌ ✟ ✢ ☞

Unconditional stability:

✤ ✤ ✦ ✞ ✦ ✠ ✯✬ ★ ✩ ✤ ✤ ✓ ✻ ✂ ✤ ✤ ✭ ✗ ✄✆☎ ✦ ✠ ✪ ✞ ✂ ✟ ✦ ✞ ✪ ✠ ✝
★ ✩ ✤ ✤ ✓ ✻ ✡ ✤ ✤ ✦ ✞ ✦ ✠ ✯✬ ✤ ✤ ✓ ✻ ✂ ✤ ✤ ✭ ✗ ✄✆☎ ✦ ✠ ✪ ✞ ✂ ✟ ✦ ✞ ✪ ✠ ✝
✤ ✤ ✓ ✻ ✥ ✾ ✞ ✖ ❢ ❏ ✲ ✩ ✧❤❣ ✰ ❝ ✄
✐ ✧ ✂
✐ ❥ ✧ ★ ✩ ✝ ✓ ✂ ✄ ❘✬ ★ ✩ ✐ ✧ ✂
★ ✩ ✐ ❥ ✧ ★ ✩ ✝ ✓ ❞ ✫ ✠ ✥ ✾ ✠ ✖ ❢ ✐ ✲ ✩ ✷ ❣ ✰ ❝ ✄ ✯✬ ✷ ❏ ✂
✷ ★ ✩ ❥ ❏ ✝ ✓ ✂ ✄
★ ✩ ✷ ❏ ✂
★ ✩ ✷ ★ ✩ ❥ ❏ ✝ ✓ ❞ ✫ ✞ ✛
slide-28
SLIDE 28

Velocity field

☎ ✡ ✖ ✛ ☞ ✂ ☞ ✛ ✖ ❃ ❄ ❅ ✄ ✗ ❀ ✍ ✝ ❃ ❄ ❅ ✄ ✗ ❀ ✎ ✝ ✌ ✟ ✡ ☞ ✛ ✮ ✂ ☞ ✛ ✖ ❅ ❉❊ ✄ ✗ ❀ ✍ ✝ ❅ ❉ ❊ ✄ ✗ ❀ ✎ ✝ ✛

Divergence free in the box-scheme sense:

✦ ✠ ✪ ✞ ☎ ✷ ✧ ✂ ✦ ✞ ✪ ✠ ✟ ✷ ✧ ✡ ☞

Initial data

✍ ✌ ✎ ✌ ☞ ✝ ✡ ✶ ✲ ❦ ✰ ✰ ✺ ✺ ✞ ✲ ✰ ❧ ✓ ✼♥♠ ★ ✺ ✠ ✲ ✰ ❧ ✓ ✼♥♠ ✼ ✛
slide-29
SLIDE 29

(i)

♦ ❨ ❩ ♣ q ♣sr ❭

(j)

♦ ❨ ❩ ♣ q ♣ r t ✉ ❭ ✈ ✡ ❫ ✡ ✗ ☞ ☞ ✌ ✾ ✡ ✖
slide-30
SLIDE 30

(k)

♦ ❨❬❩ ♣ q ♣ r t ✇ ① ❭

0.2 0.4 0.6 0.8 1 0.01 0.02 0.03 0.04 0.05

(l)

② ② ♦ ❨❬③ ❭ ② ② ④ ♣ r ⑤ ③ ⑤⑦⑥
slide-31
SLIDE 31

and/or

negative in part of the domain. The shifted box scheme

❯ ✁ ✂ ✄ ✄ ❑ ✂ ☎ ✝ ❯ ✝ ❱ ✂ ✄ ✄ ✒ ✂ ✟ ✝ ❯ ✝ ⑧ ✡ ☞ ✌ ❑ ✂ ☎ ✢ ☞ ✌ ✒ ✂ ✟ ✢ ☞

Velocity field:

☎ ✡ ☞ ✛ ☞ ❢ ✂ ☞ ✛ ☞ ✿ ❃ ❄ ❅ ✄ ✗ ❀ ✍ ✝ ❃❄ ❅ ✄ ☞ ✛ ✮ ✎ ✝ ✌ ✟ ✡ ☞ ✛ ❂ ✂ ☞ ✛ ✮ ❅ ❉❊ ✄ ✗ ❀ ✍ ✝ ❅ ❉❊ ✄ ☞ ✛ ✮ ✎ ✝ ✌

Initial data

✍ ✌ ✎ ✌ ☞ ✝ ✡ ✶ ✲ ⑨ ✰ ✰ ✺ ✺ ✞ ✲ ✰ ❧ ⑩ ✼❶♠ ★ ✺ ✠ ✲ ✰ ❧ ✓ ✼❶♠ ✼ ✌
slide-32
SLIDE 32

(m)

③ ❷ r

(n)

③ ❷ ⑥ t ❸ ✤ ✤
✏ ✝ ✤ ✤ ✌ ✈ ✡ ❫ ✡ ✗ ☞ ☞ ✌ ❑ ✡ ✖ ✌ ✒ ✡ ☞ ✌ ✾ ✡ ☞ ✛ ❂
slide-33
SLIDE 33

Robust scheme. Random data distributed on

❝ ☞ ✌ ✖ ❞

for initial and boundary functions

20 40 60 80 100 0.2 0.4 0.6 0.8 1

(o)

❹ ❷ ⑥

20 40 60 80 100 0.2 0.4 0.6 0.8 1

(p)

❹ ❷ ① r ✤ ✤ ❯ ✄ ✏ ✝ ✤ ✤ ✻ ✌ ☞ ✘ ✏ ✘ ✖ ☞ ☞
slide-34
SLIDE 34

Irregular grid Box scheme well defined over one interval

❝ ✍ ✧ ✌ ✍ ✧ ★ ✩ ❞

Step size

✫ ✧ ✡

random numbers

50 100 150 200 0.002 0.004 0.006 0.008 0.01 0.012 0.014

slide-35
SLIDE 35

Solution

  • ✄✆✍

with random step size

✫ ✧

0.2 0.4 0.6 0.8 1 −0.2 0.2 0.4 0.6 0.8 1

slide-36
SLIDE 36

The Taylor-Green problem T=5

x c

1 2 3 0.2 0.4 0.6 0.8 1 1.2 1.4

Analytical Shifted Box-Scheme Dissipative

slide-37
SLIDE 37

T=10

x c

1 2 3 0.2 0.4 0.6 0.8 1 1.2 1.4

Analytical Shifted Box-Scheme Dissipative

slide-38
SLIDE 38

T=30

x c

1 2 3 0.2 0.4 0.6 0.8 1 1.2 1.4

Analytical Shifted Box-Scheme Dissipative

slide-39
SLIDE 39

T=50

x c

1 2 3 0.2 0.4 0.6 0.8 1 1.2 1.4

Analytical Shifted Box-Scheme Dissipative

slide-40
SLIDE 40

High resolution T=10

x c

1 2 3 0.2 0.4 0.6 0.8 1 1.2 1.4

Analytical Shifted Box-Scheme 256 Shifted Box-Scheme 1024

slide-41
SLIDE 41

High resolution, T=30

x c

1 2 3 0.2 0.4 0.6 0.8 1 1.2 1.4

Analytical Shifted Box-Scheme 256 Shifted Box-Scheme 1024

slide-42
SLIDE 42

High resolution, T=50

x c

1 2 3 0.2 0.4 0.6 0.8 1 1.2 1.4

Analytical Shifted Box-Scheme 256 Shifted Box-Scheme 1024

slide-43
SLIDE 43

Conclusions: * New shifted box scheme * Work

❺ ❫
  • n grid with

points for any number of space dimensions * Unconditionally stable, “energy” conserving * No parasitic solution * No “extra” boundary conditions * Stable and accurate also for irregular grids