Lab Report 9

Exercise 1

Consider the shown system, where the potential energy representation can be given by:

,
Where we assume the deflection has the polynomial form:

, which satisfies our kinematic boundary conditions at x=0.

Thus, v and v are defined as:
N

dv
x i 1
v = = ( i+1 ) ci
( )
dx i=1
L L
'

v' '=

()

d2 v
x
= i ( i+1 ) c i
2
L
d x i=1

i1

1
L

( ) ()

Plugging in our terms into our potential energy equation gives:

1
EI
2

i ( i+1 ) c i
i =1

i1

2 2

( ) ( ))
x
L

1
L

dx P
0

1
2

( i+ 1 ) c i
i=1

x i 1
( ) dx
L L

() )

( v ( x ) ) =
0

Using the principle of stationary potential energy, we take our partial derivative with respect to

cj , and, setting to zero, for force equilibrium yields:

EI v ' '

''

'

v
v
dx P v '
dx=0
cj
cj
0
L

d
( v ( x ) ) =
d cj
0

Where

v ''
x
= j ( j+1 )
cj
L

j1

1
L

() ()

Substituting 2 and 3 into 1

2 and

v' (
x j 1
= j+1 )
( )
cj
L L

()

i1

( ) ( ))

x
EI i ( i+ 1 ) c i
L
i=1

1
L

x
( j ) ( j +1 )
L

j1

1 2
dx P
L
0

() ()

( ) ( ))

( i+1 ) ci Lx
i=1

1
1 x
( j+1)( )( ) dx=0
L
L L

d
( v ( x ) )=
dc j
0
Manipulating, we obtain:
N

d
1
( v ( x ) ) =EI ( i+ 1 ) iC i ( j+1 ) j
dc j
L
i=1

4 L

) (
0

x
L

i1

x
L

) ()

j1

1
dxP (i+1 ) Ci ( j+1 ) Ci
L
i =1

We note that
L

(
0

x
L

i1

x
L

) ()

j1

dx=

L
i+ j1

and

(
0

x
L

x
L
dx=
L
i+ j+1

)( )

Substituting 5 and 6 into 4

4

1
1
L
L
N ( i +1 ) C i ( j +1 )
L
L
( i+1 )( ij ) Ci ( j + 1)
P
=0
i + j1
i + j +1
i=1
N
d
( v ( x ) ) =EI
dc j
i=1

()

Dividing by

EI
L3

on both sides

( i+1 ) ij Ci ( j+1 ) P L2 N ( i+1 ) C i ( j+ 1 )

i+ j1 EI i+ j+1 =0
i=1
i=1
Reducing to eigen form:
By factoring Ci , we note that

(i+1 ) ( j+1 ) ij
(i +1 ) ( j+1 )
P L2
K ij =
Gij =
=
i+ j1
i+ j+1
EI
Which proves C.37 and C.38 respectively.

()

5&6

2 L

) ( xL ) ( Lx ) dx=0
0

Exercise 2

K and G are defined:

% Form K
K(i,j) = (i+1)*(j+1)*i*j/(i+j-1); %% COMPLETE this line
% Form G
G(i,j) = (i+1)*(j+1)/(i+j+1); %% COMPLETE this line

The critical load and buckling mode are stored:

approx = D(1); , where D(1) corresponds to the minimum buckling load

c =V(:,1);, where the columns in V correspond to the buckling loads

Exercise 3
1) What is the error with just one term in the approximation?

We obtain an error of 21.58%

2) What is the error with 3 terms in the approximation?

We obtain an error of 0.014%.

3) Make a plot of error versus the number of parameters in the approximation for n = 1,
2, . . . , 10. Use semilog axes (log error versus number of terms).

Exercise 4

We consider the load case for the following system.

The given potential energy for the system is:

,
Where x0 is defined to be the point of application of the transverse load P 0
We further define :
N

v ( x )= C i sin (
i=1

ix
)
L

Applying force equilibrium by taking the partials, and setting it to stationary potential energy,

i
Ci
L

( )
j
jx
sin (
sin ( ix
)
[
]
L
L
L )
2

i=1

0
d
( v ( x ) )=EI
dcj

We note the following identities:

L

sin
0

( ) ( )

cos
0

{
{

L
ix
jx
sin
dx= 2 ,i= j
L
L
0,i j
L
ix
jx
cos
dx= 2 ,i= j
L
L
0,i j

( ) ( )

Plugging this into our force equilibrium statement,

j x 0
d
EIL j 4
PL j 2
( v (x ))=
C j
C jP0 sin
=0
dc j
2 L
2 L
L

( )

( )

With further algebraic manipulation, we obtain:

( )

[ ( ) ( )]
4

EIL j
PL j

2 L
2 L

C j=P0 sin (

j x 0
)
L

We note again, that we have rearranged our expression to the form:

K
( ijGij ) Cj=Fj

We note that unlike the eigenform, this can be solved directly by simple algebraic manipulation:

C j=

Fj
,
K ij PG ij

Where

F j=P0 sin (

j x 0
) ,
L

K ij =

interchanged with I,
Thus,

i x 0
)
L
Ci =
EIL i 4
L i
P
2 L
2 L
P 0 sin (

( )

( )

EIL j
L j
Gij =
2 L
2 L

( )

( )

, we also note that the j index in Cj can be freely

Exercise 5 & 6

1) Ci is obtained:
for i = 1:N
kij=EI*0.5*L*(i*pi/L)^4;
Gij=0.5*L*(i*pi/L)^2;
Fj=P0*sin(i*pi*x0/L);
c(i) =Fj/(kij-P*Gij); %% COMPLETE this line
end

2)

uapprox

is defined by plugging our obtained c values back into our original assumed
deflection expression v(x)

for k = 1:length(x)
for i = 1:N
uapprox(k) =uapprox(k) + c(i)*sin(i*pi*x(k)/L); %% COMPLETE this line
end
end

1)

2)

As revealed from the plots, v(0.5L) increases with increasing P, as expected.

Code:
%% Exercise 6-2
Pspan
= linspace(0,0.95*Peuler,200); % Declare axial compression laod as a vector
expression spanning subset [0,0.95Peuler]
% Define number of terms in expansion (leave as 5 for this exercise)
N=5;
vLo2=zeros(1,200); %preallocate v(L/2) for speed
for n=1:200
L
= 10;
% Length of beam
x
= linspace(0,L,200);
% Domain of beam
EI
= 100;
% Bending stiffness
x0
=0.6*L;
% Define point of application fo transverse load,
x0
P
=Pspan(n);
% Define P as element of Pspan
Peuler = pi^2*EI/L^2;
% Euler critical load
P0
= 3;
% Transverse load
% Compute coefficients in approximation
for i = 1:N
kij=EI*0.5*L*(i*pi/L)^4;
Gij=0.5*L*(i*pi/L)^2;
Fj=P0*sin(i*pi*x0/L);
c(i) =Fj/(kij-P*Gij); %% COMPLETE this line
end

% Evaluate approximate solution for plotting

uapprox = zeros(1,length(x));
for k = 1:length(x)
for i = 1:N
uapprox(k) =uapprox(k) + c(i)*sin(i*pi*x(k)/L); %% COMPLETE this line
end
end
vLo2(n)=uapprox(L/2); %Declare vLo2
end
plot(vLo2,Pspan); % plot
ylabel('P');
xlabel('v(0.5L)');
title('Plot of v(0.5L) against P from P=0:0.95Peuler')

P0=1

From the plot, we observe that, with increasing P0, there is a greater centerspan
deflection, v(0.5L) for a given value of P.
Consider the code below:
%% Exercise 6-2
L
= 10;
% Length of beam
x
= linspace(0,L,200);
% Domain of beam
EI
= 100;
% Bending stiffness
x0
=0.6*L;
% Define point of application fo transverse load,
x0
Peuler = pi^2*EI/L^2;
% Euler critical load
Pspan
= linspace(0,0.95*Peuler,200); % Declare axial compression laod as a vector
expression spanning subset [0,0.95Peuler]
P0span = [1,3,5];
% Vector of the 3 desired transverse loads
vL02P0 = zeros(1,200,3);
% Declare vector for different vlo2 values at
different P0 values.Preallocate for speed
% Define number of terms in expansion (leave as 5 for this exercise)
N=5;
vLo2=zeros(1,200); %preallocate v(L/2) for speed
for m=1:3
for n=1:200
P0
=P0span(m);
%Define P0 as element of P0span

=Pspan(n);

% Define P as element of Pspan

% Compute coefficients in approximation

for i = 1:N
kij=EI*0.5*L*(i*pi/L)^4;
Gij=0.5*L*(i*pi/L)^2;
Fj=P0*sin(i*pi*x0/L);
c(i) =Fj/(kij-P*Gij); %% COMPLETE this line
end

% Evaluate approximate solution for plotting

uapprox = zeros(1,length(x));
for k = 1:length(x)
for i = 1:N
uapprox(k) =uapprox(k) + c(i)*sin(i*pi*x(k)/L); %% COMPLETE this line
end
end
vLo2(n)=uapprox(L/2); %Declare vLo2
end
vL02P0(:,:,m)=vLo2;
plot(vLo2,Pspan); % plot
hold on
end
ylabel('P');
xlabel('v(0.5L)');
title('Plot of v(0.5L) against P from P=0:0.95Peuler')
legend('P0=1','P0=3','P0=5')

We note that VL02P0 is a 3 dimensional matrix array to store VL02 values for P0=1,3,5.
By comparing our VL02P0 arrays, which corresponds to VL02, or centerspan deflection values for
P0=1,3,5 respectively and performing element by element division,
We run the command:
alldifferent=any(diff(sort(vL02P0(:,:,1)./vL02P0(:,:,2))==0))
alldifferent=any(diff(sort(vL02P0(:,:,2)./vL02P0(:,:,3))==0))
alldifferent=any(diff(sort(vL02P0(:,:,1)./vL02P0(:,:,3))==0))
All commands return
alldifferent=0;
We note that our command tells us that all elements in the resulting matrices from element by
element division are equal, ie- VL02P0(i,j,k)/VL02P0(i,j,k+1) result in a constant K, or,

( L2 , P )
L
V ( ,P )
2
V

01

=some constant K.

02

This indicates that VL02 values have a linear relationship with VL02 values of different P0 values.

We note that
vL02P0(:,:,1)./vL02P0(:,:,2) gives a matrix whereby each element is 1/3.
vL02P0(:,:,2)./vL02P0(:,:,3) gives a matrix whereby each element is 0.6.
vL02P0(:,:,1)./vL02P0(:,:,3) gives a matrix whereby each element is 0.2.
We note that vL02P0(:,:,1)corresponds to VL02 values for P0=1,
vL02P0(:,:,3)corresponds to VL02 values for P0=3
vL02P0(:,:,5)corresponds to VL02 values for P0=5
Therefore,
vL02P0(:,:,1)./vL02P0(:,:,2)=1/3=P01/P02
vL02P0(:,:,2)./vL02P0(:,:,3)=3/5=P02/P03
vL02P0(:,:,1)./vL02P0(:,:,3)=1/5=P01/P03
We observe that this corresponds to our mathematical expressions whereby,
N

ix
v ( x )= c i sin (
) , where
L
i=1

i x 0
Fj
L
c i=
=
K ij PGij
K ij P Gij
P 0 sin

( )

Therefore,
N

v ( x )=P0
i=1

sin

( iLx ) sin ( ix )
0

K ij P G ij

We note then that v(x) mathematically scales linearly with P0, as demonstrated in our
comparisons. Thus, our plots match our expectations, and express the linear relationship
between v(0.5L) and P0, and, v(x) and P0 in general.