Sie sind auf Seite 1von 11

Problem 2.

Here we have the problem of heat transfer in one dimension. Taking


back the governing equation (3.1) and according to table 3.1, the steady state
equation is
d  d 
  kA   P     0 0 xL (2.1.1)
dx  dx 

The data given for this problem are:


a – the left end of the rod is maintained at temperature 0
b – the other (right) end is insulated
These data are translated as follow:
a → 0  0 (2.1.2)
d
b → Q( L)  kA  0 (2.1.3)
dx x L

Let the function  x   x   0 for 0  x  L . By introducing this


function in equation (2.1.1), when k , A ,  and P are constant, we obtain:
d 2
 kA 2  P  0 0 xL (2.1.4)
dx
With the boundary conditions
d
 0   0  0   and QL   kA  0 (2.1.5)
dx x L
The rod is assumed as a domain  = 0; L . We divide into a set of N
sub domains  e = x1e ; x2e  of equal lengths he 
L
. This is locally that we have this,
N
globally we have a set of point xi iN0

With the finite difference method, we do the approximation:


d 2  i 1  2 i   i 1
 (2.1.6)
dx 2 h2
where  i i 1 is a set of points which we want to be equal to the values of  at any
N 1

point x i :  i    xi  0i N
We obtain by substituting in (2.1.4):
k  2k  k
 2  i 1   2    i  2  i 1  0 (2.1.7)
h  h  h

Let take m 
2
. Then
k
 2

  i 1  2  mh   i   i 1  0 1 i  N (2.1.8)
And the boundary conditions are:
 N 1   N
 0   0 and  k  0   N 1   N (2.1.9)
h
Using equation (2.1.9), the equations of finite difference method are
  
2  mh   1   2   0
2


 
   1  2  mh   2   3  0

2

 .......... .......... ...... (2.1.10-a)


 

 
N  2  2  mh   N 1   N  0
2


 
  N 1  1  mh   N  0
2

Or in the matrix form
2  mh 2 1 0 0 0    1   0 
     
 1 2  mh  1    2   0 
2
0 0
 ... ... ... ... ...    ...    ...  (2.1.10-b)
 
 1 2  mh   1   N 1   0 
2
 0 0
 1 1  mh     N   0 
2
 0 0 0

With the finite element method, by taking linear elements we can have
the coefficients matrix for each element from equation (3.16) by taking ae  kA and
ce  P . Thus this matrix is:
 NkA PL NkA PL 
NkA  1  1 PL 2 1  L   
 
 
e
  3 N L 6N 
 NkA PL 
L  1 1  6 N 1 2  ( sym)
(2.1.11)
 
 L 3N 
The columns vectors are:

f e   f e L 1  0
2N 1 0
We can have the global system from equation (3.24).

To compare the finite element method and the finite difference method
with the exact solution, we will take for the steel rod:
- diameter: D = 0,02 m
- length: L = 0,05 m
- thermal conductivity k = 50 W/(m.°C)
- heat transfer coefficient β = 100 W/(m.°C)
- temperatures: T0 = 320 °C and T∞ = 20 °C
- number of linear element N=4

We deduce:
 0  0    320  20  300 C
  D 4 4  100
m      20
k k D 2
kD 50  0,02
4
2 2

mh    m L    20  0,05   0,0625


2

 N  4 
NkA Nk D 2
4  50 3,14  0,02 2
     1,256
L L 4 0,05 4
NkA Nk D 2 4  50 3,14  0,02 2
     1,256
L L 4 0,05 4
PL L 100  0,05
  D   3,14  0,02  0,013
6N 6N 6 4

Finite difference method


The matrix equation is:
2,0625 1 0 0  1  300 
 1
 2,0625 1 0     
 2   0 
     (2.1.12)
 0 1 2,0625  1   3   0 
 
 0 0 1 1,0625    4 
 
 0 

By solving, we obtain:
1   248 ,281 
  212 ,080 
 2  
   (en °C)

  
3 189 ,133 
 4  178 ,008 
  D 2 1   0 3,14  0,02 2 248 ,281  300
and Q0  k 1 0  k   50    64,959
h 4 L 4 0,05
4 4

Finite element method


So the coefficients matrix for each element is:
 2  0,013  1,256  0,013   1,282  1,243 
 e   1,256  
 1,256  0,013 1,256  2  0,013   1,243 1,282 
The matrix form of equations of the system is
 1,282  1,243 0 0 0  300  Q0 
 1,243 2,564  1,243
 0 0   1   0 
   
 0  1,243 2,564  1,243 0    2    0  (2.1.7-a)
     
 0 0  1,243 2,564  1,243   0
 3   
 0 0 0  1,243 1,282    4   0 
In the linear form these equations are:
 1,282  300  1,243 1  Q0
 1,243  300  2,564  1,243 0
 1 2

  1,243 1  2,564 2  1,243 3 0 (2.1.7-b)


  1,243 2  2,564 3  1,243 4  0

  1,243 3  1,828 4  0
The primary variables are determined by solving the last four equations
of the system. After transformation, these four equations in linear form are:
 2,5641  1,243 2  372 ,9
 1,243  2,564  1,243 3 0
 1 2
 (2.1.8-b)
  1, 243  2  2,564  3  1, 243  4  0
  1,243 3  1,282 4 0
Or in the matrix form
 2;564  1,243 0 0  1  372 ,9
 1;243 2,564  1,243     0 
 0    2    
 (2.1.8-a)
 0  1,243 2,564  1,243   3   0 
 
 0 0  1,243 1,282   4   0 

By solving with a computer, we obtain:


 1   251,750 
  219 ,298 
 2  
   (en °C)
 3  200 ,607 
 4  194 ,504 

From the first equation of (2.1.7-b), we have :


Q0  1,282  300  1,2431  1,282  300  1,243  251,750  71,675

The following table shows all our results:


Finite Difference Finite Element
True Solution
Method Method
θ1 (°C) 248,281 251,750 251,708

θ2 (°C) 212,080 219,298 219,229

θ3 (°C) 189,133 200,607 200,524

θ4 (°C) 178,008 194,504 194,416

Q0 (W) 64,959 71,675 71,742


Problem 2.2

First of all, let determine the


area of the section at any position x.
x   0,50,5  2a 
According to THALES theorem,
x a x
  a
2 0,5 4
So

1 x
 x  
4

Finite element models

2 linear elements 4 linear elements

1
0  20  0  20   5 kN
4
For the other charges Pi or Fi, we have:
25hi  i h 
i  25Vi  25 x x dx  x 1  x dx 
xi
25 x i

 x1  1  i 
2 2
(en kN)
4  2
i i
1 4 1

L
Where hi  x2  x1  .
i i

N
Thus,
L 2
- 2 linear elements: N = 2  hi   1
2 2
25  1  75
1   0 1   kN
4  2 8
25  1  125
2   1  1    kN
4  2 8

L 2
- 4 linear elements: N = 4  hi    0,5
4 4
25 1  0,5  125
1    0 1  kN
4 2  2  32
25 1  0,5  175
2     0,5  1   kN
4 2  2  32
25 1  0,5  225
3    1  1   kN
4 2  2  32
25 1  0,5  275
4    1,5  1   kN
4 2  2  32

Let now establish formulae for linear elements:


 is the function which takes the value 1 at x  x1i and the value 0 at x  x 2i
i
1

x 2i  x x 2i  x d 1i 1
   i
i
 and  
i

x 2  x1i
1 1
hi dx hi
 i2 is the function which takes the value 0 at x  x1i and the value 1 at x  x 2i
x  x1i x  x1i d i2 1
   i
i
 and  2 
i

x2  x1i
2
hi dx hi

For an element the matrix is determined as follow:

 i     i   i dx   2 11   1 1 xx 1  x dx


x2i i
2

i
i
1x 4hi   1

  i h   1  1
   
i
 x1  1  i   
4hi  2   1 1 
For the 2 linear elements,
 1   1  1 3  1  1
 
1   0  1     
4 2   1 1  8  1 1 
 1 5  1  1
    1  1  1   11
2

1  8  1 1 
4 2 
And the general stiffness matrix is
3 3 0  3 3 0 
 
 2    3 3  5  5   3 8  5
8 8
 0 5 5   0  5 5 

The equations system is


 3  3 0  U 1   40   0 
    1   
 3 8  5  U 2    75    0 
8  8 125   R 
 0  5 5   U 3     
The boundary condition is U3 = 0, so we solve
 3  3 U 1  40  U 1  1,298 
           (en mm)
  3 8   2  
U 75 U 2  0,821
125 
And  R   (5U 2  5U 3 )  R  30 kN
8 8

For the 4 linear elements,


 1   1  1 5  1  1
 
1   0 1     
2 4   1 1  8  1 1 
1 1   1  1 7  1  1
 
 2   1    
22 4   1 1  8  1 1 
 1   1  1 9  1  1
 
 3  1  1     
2 4   1 1  8  1 1 
3 1   1  1 11  1  1
 
 4   1    
22 4   1 1  8  1 1 
And the general stiffness matrix is:
5 5 0 0 0   5 5 0 0 0 
 5 5  7  7 0 0   5 12  7 0 0 
 
 4     0  7 7  9  9 0     0  7 16  9 0 
8  8 
 0 0  9 9  11  11 0 0  9 20  11
 0 0 0  11 11   0 0 0  11 11 
The general equation is
 5 5 0 0 0  V1  160   0 
 5 12  7 0
 0  V2  125   0 
   1    
 0  7 16  9 0   V3   175    0 
8    32 
 0 0  9 20  11  V4  225   0 
   
 0 0 0  11 11  V5  275   R 
With the boundary condition U5 = 0; so we solve
 5 5 0 0  V1  160  V1  1,662 
 5 12  7 0  V  125  V  1,376 
  1    2  
    
2
     (en mm)
 0  7 16  9 V3  4 175  V3  1,012 
 
0 0  9 20   V4  
225 
 
V4   
0,556 

275 
And  R   (11U 4  11U 5 )  R  30,000 kN
32 8

Stresses are determined for each element by:


 1 1  U 1i  
 i   i   U        i    U 2  U 1 
i i i i

 hi hi  U 2  hi
For the two linear elements,
 1  28  0,821  1,298   13,356 kN/mm2
 2  28  0  0,821  22,988 kN/mm2

For the four linear elements


 1  2  28  1,376  1,662   16,016 kN/mm2
 2  2  28  1,012  1,376   20,384 kN/mm2
 3  2  28  0,556  1,012   25,536 kN/mm2
 4  2  28  0  0,556   31,136 kN/mm2

We compare the two interpolations in the following table:


Two linear elements Four linear elements Exact solution
Displ.(mm) Stress(Mpa) Displ.(mm) Stress(Mpa) Displ.(mm) Stress(Mpa)
Top
1,298 13,356 1,662 16,016 20
node
Mid
0,821 18,172 1,012 22,96 28,75
node
Bottom
0 22,988 0 31,136 0 40
node
reaction 30 kN 30,000 kN 30 kN
Problem 2.3
1-)

For this truss, n = 76, so


P = n = 76 kN
Q = (100 – n) = 24 kN
1824 mm 2 for horizontal and vertical members
  n 100  n   1824   
 2188 ,8 mm 2 for diagonal members

This part is done on excel file. We have tried to develop a little program
which will automatically determine the general stiffness matrix of a truss. The
important informations are given on sheet 2. When you wrote data on this sheet
automatically, a stiffness matrix is created on the sheet “brouillon”. If you the matrix
of your structure,
- go on sheet 1
- select a number of cells equal to the number of values of the matrix (for
example 10 lines x 10 columns if the matrix is 10x10)
- write the expression “=matrice$” where $ is the number of nodes of the
structure (for example “=matrice15”)
- press CTRL+SHIFT+ENTER
The matrix appears but you cannot modify it; when you want to modify
the matrix to obtain the condensed one, copy the first one and do a “special paste”
(this is proper to EXCEL), choose to paste values of selection.
By doing our structure which gives a matrix 78x78 we realized that
EXCEL cannot gives the inverse matrix of a squared matrix of an order greater than
73 and since we realized it, we didn’t find any software which can do it that is why
our structure is not solved and why our work is not finished, we apologize for this.

Das könnte Ihnen auch gefallen