Graphics Reference
In-Depth Information
Kass and Miller [ 10 ] find that a first-order implicit method (Eqs. 8.20 and 8.21 ) is sufficient for this
application. Using these equations to solve for h ( n ) and substituting Equation 8.19 for the second deriv-
ative ( h ( n )) results in Equation 8.22 .
hðnÞhðn
1
Þ
_ ðnÞ¼
(8.20)
Dt
_ ðnÞ _ ðn 1 Þ
Dt
hðnÞ¼
(8.21)
h i ðnÞ¼
2 h i ðn
Þh i ðn
Þ
1
2
2 d i 1 þ d i
2 Dx
gðDtÞ
ðh i ðnÞh i 1 ðnÞÞ
2
(8.22)
2 d i 1 þ d i
2
þgðDtÞ
ðh i þ 1 ðnÞh i ðnÞÞ
Dx
2
Assuming d is constant during the iteration, the next value of h can be calculated from previous
values with the symmetric tridiagonal linear system represented by Equations 8.23-8.25 .
Ah i ðnÞ¼
2 h i ðn
1
Þþh i ðn
2
Þ
(8.23)
2
4
3
5
e 0
f 0 000 0 0
f 0
e 0
f i 00 0 0
0
f i
e 2
...
0
0
0
A ¼
00
... ... ...
0
0
(8.24)
000
... e n 3
f n 3
0
0000 f n 3
e n 2
f n 2
00000 f n 2
e n 1
!
d 0 þ d 1
2
2
e 0 ¼
1
þ gðDtÞ
2
ðDxÞ
!
ð
d i 1 þ
2 d 1 þ d i þ 1
2
e i ¼
1
þ gðDtÞ
0
< i < n
1
Þ
2
ðDxÞ
2
!
d n 2 þ d n 1
2
2
e n 1 ¼
1
þ gðDtÞ
2
ðDxÞ
!
!
d i þ d i þ 1
2
2
f i ¼ gðDtÞ
(8.25)
2
ðDxÞ
To simulate a viscous fluid, Equation 8.23 can be modified to incorporate a parameter that controls
the viscosity, thus producing Equation 8.26 . The parameter t ranges between 0 and 1. When
0,
Equation 8.26 reduces to Equation 8.3 .
Ah i ðnÞ¼h i ðn 1 Þþð 1 tÞðh i ðn 1 Þh i ðn 2 ÞÞ
(8.26)
 
Search WWH ::




Custom Search