Вы находитесь на странице: 1из 10

TECHNICAL UNIVERSITY OF CIVIL ENGINEERING - BUCHAREST

DEPARTMENT OF HYDRAULIC STRUCTURES

THE FINITE ELEMENT METHOD


CLASS AND HOMEWORKS

EXERCISE No. 1

1.1 For the gravity dam with the cross section shown in the following figure, determine the
displacements, the state of stress at the bottom of downstream slope and the reaction forces at
the contact with the rock layer (along the embedment).
NNR
H

B/3
B

Problems data:

a). Geometry of the dam - height H = N+15,0 m


- width of embedment B = N+10,0 m

b). Material properties

- elasticity module E = 200.000 Dan/cm2 for N = odd number


E = 250.000 Dan/ cm2 for N = even number
- Poisson's coefficient = 0,16

c). Assumptions

- water level at the crest of the dam;


- plane strain state;
- only water pressure effect is taken into account.

The problem will be solved using the F.E.M., by decomposing (meshing) the structure into 4
identical, triangular, linear finite elements.

1
1.2 Theoretical considerations

a). The triangular finite element

According to the problem, the domain should be meshed in triangular elements.

vl
y ul
l

0 x
D vn
vm n un
m um

General shape:
- defined by 3 nodes;
- coordinate system arbitrary chosen;
- element thickness t, area A
Degrees of freedom: 6 displacement components on both directions
Nodal displacement vector
T T
{ } e= l m n = ul vl um vm un vn

Displacement field on element domain, x, y D, expressed be polynomial interpolation (1st


degree) is:

u(x,y) = a1 + a2 x + a3 y
(1.1)
v(x,y) = a4 + a5 x + a6 y

In matrix form:
a1
a
2
u( x , y ) 1 x y 0 0 0 a3

e
{ (x,y)} = =
v( x , y ) 0 0 0 1 x y a4
a5

a6

and condensed:

{ (x,y)}e = [ f(x,y) ] { a } (1.2)


2
To express the displacement of a random point in the domain, as a function of nodal
displacements, we will follow the next stages:

- express the nodal displacement vector by using equation (1.2) for particular points l, m and
n:
l ( xl , yl ) f ( xl , yl )

{}e = m = ( xm , ym ) = f ( xm , ym ) { a } = [ ] { a } (1.3)
( x ,y ) f ( x ,y )
n n n n n

- by calculating the inverted matrix results:

{a} = []-1{}e (1.4)

1
where []-1 = [] (1.5)
2A

xmyn xn ym 0 xn yl xl yn 0 xl ym xm yl 0
y y 0 yn yl 0 yl ym 0
m n
[]= ( xm xn ) 0 ( xn xl ) 0 ( xl xm ) 0
0 xmyn xn ym 0 xn yl xl yn 0 xl ym xmyl

0 ym yn 0 yn yl 0 yl ym

0 ( xm xn ) 0 ( xn xl ) 0 ( xl xm )

{ (x,y)}e = [ f(x,y) ] []-1{}e = [] {}e (1.6)


Matrix expresses the interpolation (approximation) of displacements on the element
domain according to nodal displacements.

Nl ( x, y ) 0 Nm( x, y ) 0 Nn( x, y ) 0
[] = (1.7)
0 Nl ( x, y ) 0 Nm( x, y ) 0 Nn( x, y )

1
where Nl(x,y) = (l + ymn x + xmn y)
2A
l = xm yn - xn ym (1.8)
ymn = ym - yn
xmn = - (xm - xn)
A = area of domain D - triangle (l, m, n),
and so on, by circular permutations.

Stain field on element - expressed by the theory of elasticity - plane stress

3
u

x x
v
{} = y = (1.9)
y
xy u v
+
y x
using equation (1.1)

u
x = = ( a1 +a 2 x + a3 y ) = a2
x x
y = a6 (1.10)
xy = a3 + a5
In matrix expression:

a2 0 1 0 0 0 0
T
{} = a6 = 0 0 0 0 0 1 a1 a 2 a3 a4 a5 a6 =
a +a 0 0 1 0 1 0
3 5

= [c] {a} = [c] []-1{}e = [B] {}e (1.11)

[B] = matrix of derivatives of interpolating functions.

Replacing from equation (1.7)


N l N m N n
0 0 0
x x x
N l N m N n
{} = 0 0 0 {}e (1.12)
y y y
N N l N m N m
l N n N n
y x y x y x

N l 1 N l 1
with , = y mn and = xmn , and so on.
x 2 A y 2 A

4
N l N m N n
0 0 0
x x x
N l N m N n
[B] = 0 0 0 =
y y y
N N l N m N m
l N n N n
y x y x y x

ymn 0 ynl 0 ylm 0


1
= 0 xmn 0 xnl 0 xlm (1.13)
2A
xmn ymn xnl ynl xlm ylm

Stress field - expressed by using Hooke's law

{ (x,y)} = [E] { (x,y)}


or
x x

y = [E] y (1.14)

xy xy
Replacing in (1.9)

{} = [E] [B] {}e (1.15)


The elasticity matrix for plane strain is:


1 0
E
[E] = 1 0
( 1 + )( 1 2 ) 1 2
0 0
2
The product [E][B] :

e11( ym yn ) e12( xn xm ) e11( yn yl ) e12( xl xn ) e11( yl ym ) e12( xm xl )


1
[E][B]= e21( ym yn ) e22( xn xm ) e21( yn yl ) e22( xl xn ) e21( yl ym ) e22( xm xl )
2A
e33( xn xm ) e33( ym yn ) e33( xl xn ) e33( yn yl ) e33( xm xl ) e33( yl ym )

(1.16)
with:

5
E( 1 ) E( 1 )
e11 = e22 = , e12 = e21 = ,
( 1 + )( 1 2 ) ( 1 + )( 1 2 )
E
e33 = (1.17')
2( 1 + )

Displacements of structure subjected to a system of forces are proportional with


displacements.

{F}e = [K]e{}e (1.18)

[K] = stiffness matrix of the structure, [K]e = t [B]T[E][B] AdA = [B]T[E][B]tA (1.19)

For the triangular element,


a11 a12 a13 a14 a15 a16
a a 22 a 23 a 24 a 25 a 26
12
t a13 a 23 a 33 a 34 a 35 a 36
[K]e = =
4 A a14 a 24 a 34 a 44 a 45 a 46

a15 a 25 a 35 a 45 a 55 a 56

a16 a 26 a 36 a 46 a 56 a 66

k11 k12 k13 k14 k15 k16


k k 22 k 23 k 24 k 25 k 26
12
k13 k 23 k 33 k 34 k 35 k 36
= (1.20)
k14 k 24 k 34 k 44 k 45 k 46
k15 k 25 k 35 k 45 k 55 k 56

k16 k 26 k 36 k 46 k 56 k 66

with term aij given in annex 1.

6
1.3 Meshing in finite elements

First stage for every analysis is the domain meshing. Elements are coupled in nodes; thus
displacements of common nodes and mutual reactions are equal.

y 4

4 5

2 l m

1 3

x
0
1 2 3

1.4 Boundary conditions.

For this problem, the boundary conditions are expressed by displacement contraints applied
for nodes 1, 2 and 3 - all degrees of freedom u1 , v1 , u2 , v2, u3 , v3 are suppressed. The
applied forces are calculated by assesing the volume of the triangular prism diagram of water
preasure, as shown in the next figure.

F1

H /4 A1

A2 F2
H H /2

7
1.5 The stiffness matrix of the structure.

The stiffness matrix of the structure results as a 12 12 matrix, by adding all terms of the
elementary stiffness matrices (algebraic sum taking into account the nodal degrees of freedom
connections).

element 1 element 2

element 3 element 4

By solving the linear equation system [K] {} = {F}, we obtain the displacements of nodes
4, 5 and 6. The force vector has the next components:

{F}T = R1x R1y R2x R2 y R3x R3 y F2 0 0 0 F1 0 .

where Rix, Riy are the reaction forces in constrained nodes and F1, F2 the applied loads on the
structure. The constraint displacements of nodes 1, 2 and 3 is equivalent to elimination of
lines and columns 1- 6 in the equation system. Consequently, the linear equation system has
the dimensions 6 6. After solving the system (the unknowns are the 6 displacements of
nodes 4, 5 and 6), the deformed shape could be represented, as shown in the next figure.

1.6 State of stress assessment

For the bottom of the downstream slope, assessment of the Stress State means calculation of
the stress matrix for element 3. For this element we should apply the above mentioned

8
expression, with the displacement of nodes 2,3 and 5 (calculated before) replaced in the
general form:

{}(3) = [E][B]{}(3)
or
x

y = [E] [B] u 2 v 2 u3 v3 u5 v5 T
(1.21)

xy
1.7 Reaction forces assessment

To calculate the reaction forces at the constrained nodes 1, 2 and 3, we shall replace the
calculated displacements in the first 6 lines of the entire system (12 12).

[K ]ij==11,6,12 0 0 0 0 0 0 u4 v4 u5 v5 u6 v6 T=
T
= R1x R1y R2x R2 y R3x R3y

For results checking:


5
Hi = 0 ,
1
3
Vi = 0
1

v6
u6

H v
(m) (mm)
u5
v4
u4 v5

u(mm)
B(m)

Deformed shape of the structure

9
ANNEX 1

Expressions of aij terms of the stiffness matrix for the linear triangular element

a11 = e11( y 2 y3 )2 + e33 ( x3 x2 )2


a12 = e21( x3 x2 )( y 2 y3 ) + e33 ( x3 x2 )( y 2 y3 )
a13 = e11( y 2 y3 )( y3 y1 ) + e33 ( x1 x3 )( x3 x2 )
a14 = e21( x1 x3 )( y 2 y3 ) + e33 ( x3 x2 )( y3 y1 )
a15 = e11( y1 y 2 )( y 2 y3 ) + e33 ( x2 x1 )( x3 x2 )
a16 = e21( x2 x1 )( y 2 y3 ) + e33 ( x3 x2 )( y1 y 2 )
a22 = e22 ( x3 x2 )2 + e33 ( y 2 y3 )2
a23 = e12 ( x3 x2 )( y3 y1 ) + e33 ( x1 x3 )( y 2 y3 )
a24 = e22 ( x1 x3 )( x3 x2 ) + e33 ( y 2 y3 )( y3 y1 )
a25 = e12 ( x3 x2 )( y1 y 2 ) + e33 ( x2 x1 )( y 2 y3 )
a26 = e22 ( x2 x1 )( x3 x2 ) + e33 ( y1 y 2 )( y 2 y3 )
a33 = e11( y3 y1 )2 + e33 ( x1 x3 )2
a34 = e12 ( x1 x3 )( y3 y1 ) + e33 ( x1 x3 )( y3 y1 )
a35 = e11( y1 y 2 )( y3 y1 ) + e33 ( x1 x3 )( x2 x1 )
a36 = e12 ( x2 x1 )( y3 y1 ) + e33 ( x1 x3 )( y1 y 2 )
a44 = e22 ( x1 x3 )2 + e33 ( y3 y1 )2
a45 = e12 ( x1 x3 )( y1 y 2 ) + e33 ( x2 x1 )( y3 y1 )
a46 = e22 ( x1 x3 )( x2 x1 ) + e33 ( y1 y 2 )( y3 y1 )
a55 = e11( y1 y 2 )2 + e33 ( x2 x1 )2
a56 = e21( x2 x1 )( y1 y 2 ) + e33 ( x2 x1 )( y1 y 2 )
a66 = e22 ( x2 x1 )2 + e33 ( y1 y 2 )2

10

Вам также может понравиться