Скачиваний:
14
Добавлен:
14.06.2022
Размер:
1.42 Mб
Скачать

Weighted residual method (метод взвешенных невязок)

Initial equation:

2U (x, y)

 

2U (x, y)

0

or

2U (x, y) 0

x2

 

y2

 

 

 

(x, y)

- arbitrary function (произвольная функция)

Evidently: 2U (x, y) (x, y) 0

2U (x, y) (x, y)dS 0 S - is the problem domain area

S

21

Galerkin method (метод Бубнова-Галеркина)

2U (x, y) (x, y)dS 0

This relation should be valid for all

 

possible functions (x, y)

S

Let us replace the unknown function by its approximation:

2U~(x, y) (x, y)dS R 0

S

R is called ‘weighted residual’

(x, y) is called ‘weighting function’

22

Galerkin method

Main ideas of the method are:

 

 

to use the weighted residual method;

 

to use approximation functions for weighting:

j (x, y) j (x, y)

to set residuals to zero;

Rj 0

 

to apply integration-by-parts procedure to the integral (week formulation)

integration-by-parts = интегрирование по частям week formulation = ослабленная формулировка

23

Week formulation

 

Initial equation is:

2U~(x, y) (x, y)dS R

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

S

 

 

 

 

 

 

 

 

 

 

2 ~

(x, y) j (x, y)dS 0

 

 

 

 

 

For Galerkin formulation:

U

 

 

 

 

 

 

 

 

S

 

 

 

 

 

 

 

 

 

 

~

 

3

 

 

 

 

 

Remember:

 

U (x, y) Ui i (x, y)

 

 

 

 

 

 

 

 

 

 

i 1

 

 

 

 

 

 

3

2

 

 

 

 

 

 

Ui (x, y) (x, y)dS 0

i

 

j

 

 

So:

 

 

 

 

 

 

 

 

 

 

i 1 S

 

 

 

 

 

x

 

y

2-nd order derivative from the linear polynomial is = 0:

24

Week formulation

3

 

Ui 2 i (x,

i 1

S

Let us transform the integral

For this purpose we shall use a relation from vector algebra

y) j (x, y)dS 0

2 i (x, y) j (x, y)dS

S

F G F G F G

F i (x, y)

2

G j (x, y)

i (x, y) j (x, y) i (x, y) j (x, y) i (x, y) j (x, y)

 

So we can

2 i (x, y) j (x, y) i (x, y) j (x, y) i (x, y) j (x, y)

express:

 

25

Week formulation

Integrating the last relation over the problem domain:

 

 

 

 

 

2 i (x, y) j (x, y)dS i (x, y) j (x, y) dS

i (x, y) j (x, y)dS

S

S

 

 

 

S

The Gauss theorem:

 

FdS F d F d

 

 

 

 

n

 

 

S

 

 

 

 

j (x, y)

 

 

2 i (x, y) j (x, y)dS

i (x, y)

d i (x, y) j (x, y)dS

S

 

 

 

 

S

 

 

is the border of the problem domain

 

For the elements inside the problem domain the function:

i (x, y) 0

So

2 i (x, y) j (x, y)dS i (x, y) j (x, y)dS

 

 

S

S

 

26

Week formulation

So the equation takes form of:

N

 

 

Ui i (x, y) j (x,

i 1

S

 

Finally we have got a system of equations:

N

aij Ui 0 i 1

With the coefficients

y)dS 0

aij i (x, y) j (x, y)dS

S

If the boundary potentials are known in advance, several equations in the system will have non-zero right hand sides

27

1-st type boundary conditions

1-st type boundary conditions:

U F1

Un U

If the potential at the boundary is equal to zero:

Un 0

The 1-st type boundary conditions keep a symmetry of the main problem matrix

28

The potential and field intensity

After a system of equations is solved we can express a potential as:

~

 

N

 

 

 

 

 

U

(r) Ui i (x, y)

 

 

 

 

 

i 1

 

 

 

 

 

 

 

 

 

 

 

Inside a triangle the same expression:

 

~

3

 

U

(r) Ui i (x, y)

 

 

 

 

 

 

 

i 1

 

 

~

 

 

 

 

 

 

 

 

N

 

The field intensity:

E(r) U Ui

i (x, y)

 

 

 

 

 

i 1

 

 

 

 

 

 

 

 

 

 

 

~

3

 

 

 

Inside a triangle:

E(r) Ui i (x, y)

 

 

 

i 1

 

 

 

 

29

2-nd type boundary conditions

2-nd type boundary conditions:

U

 

F2

 

n

 

 

 

 

 

 

For the elements near the boundary:

 

 

2 i (x, y) j (x, y)dS

i (x, y) j

S

 

 

Now this term can not be neglected

Equation for the potential:

N

i

 

 

i

 

 

 

 

 

j

(x, y)dS

 

U

 

(x, y)

i 1

 

S

 

 

 

(x, y) d i (x, y) j (x, y)dS

S

 

 

a

(x, y)

(x, y)dS

 

 

 

ij

i

 

j

 

 

 

 

 

S

 

 

 

(x, y)

 

 

 

 

 

 

i n

 

j (x, y)d

0

 

 

 

 

 

 

 

 

 

30