Information Technology Reference
In-Depth Information
TABLE 10.2
Procedure for Computer Implementation of the Central Difference Method
Initial Step
1.
Select
t
2.
Calculate the constants related to
t
c 1 =
1
/
t 2 , c 2 =
1
/
t , c 3 =
2 c 1 , and c 4 =
c 2 /
2
Initialize V 0 , V 0 , and V 0
3.
4.
Calculate V 1 from Eq. (10.101)
5.
Calculate the effective stiffness matrix of Eq. (10.100)
K
=
c 1 M
+
c 4 C
Decompose (triangularize)
K
6.
=
LDL T
At Each Step n:
1.
Calcula te the effective load from Eq. (10.100)
P n + 1 =
P n (
K
c 3 M
)
V n (
c 1 M
c 4 C
)
V n 1
Solve KV n + 1 =
P n + 1 using K
LDL T
2.
=
(matrix decomposition)
K can be decomposed into LDT T . L is a lower triangular matrix (elements only on the diagonal and below) with
the diagonal elements equal to 1, and D is a diagonal matrix.
solution, the spectral radius, i.e., the maximum eigenvalue of matrix A , has to be smaller
than 1. This can be readily proved. It follows from Eq. (10.102) that
X n =
AX n 1
(10.104)
Substitute Eq. (10.104) into Eq. (10.102)
A 2 X n 1
X n + 1
=
(10.105)
Thus, in general,
A n + 1 X 0
X n + 1
=
(10.106)
From the theory of matrices,
Z λ Z 1
A
=
(10.107)
where λ are the eigenvalues of A , which are assumed to be distinct, and each column of Z
is a corresponding eigenvector, whose elements are finite.
Also, from the theory of matrix analysis,
A n + 1
Z λ n + 1 Z 1
=
(10.108)
Since the elements of Z are bounded, the elements of A n + 1
will be bounded provided
| λ i |≤
0 , X n + 1 cannot be very large or approach infinity,
and the integration is stable. The condition of
1
.
0 ,i
=
1 , 2 ,
...
,n, i.e., if
| λ i |≤
1
.
| λ i |≤
1
.
0 leads to the critical time step for
stability
2
ω
t
t crit =
(10.109)
Search WWH ::




Custom Search