Lecture 2 1

Outline:
- The minimum of total potential energy
- Bar and beam analysis using total potential energy
- Direct variational method

-The Minimum of total potential energy

In solid mechanics some of the laws of physics take several alternative forms. For example the
equilibrium relations which require that vector sum of all applied forces on a body to be equal to zero can
be obtained using the variational principles such as the principle of minimum total potential energy. In
vectorial form, for cumbersome systems obtaining the governing equilibrium equations can be
cumbersome. In addition, the type of boundary conditions to be used in conjunction with the derived
equations may not be clear. Using variational approach the governing equilibrium equations can be
obtained based on purely geometrical considerations. Therefore, variational approach is suitable to apply
elasticity problems under kinematic postulates, such as beam problems.
In particular, the minimum of total potential energy principle states that in a conservative system, among
all kinematically possible configurations of the system the one that satisfies the equilibrium makes the
total potential energy a minimum. The total potential energy Π is the internal strain energy stored during
deformation U plus the external potential of the applied load Wext , i.e. Π= U + Wext .
Total potential energy is a function of the deformed configuration (i.e. displacement function). Thus, it is
a function of functions which is called a functional. A functional is a scalar quantity depending on some
function or several functions. The functional can be treated as a function of an infinite number of
independent variables. The subject matter of calculus of variation is searching of unknown functions
u , u ′... that give a stationary value (e.g., minimum) of a functional.
The minimum principle states that Π (u , u ′, u ′′...) ≤ Π (u , u ′, u ′′...) , where u is the true displacement field
and u is any kinematically admissible displacement field. Similar to the minimum of functions the
minimum of potential energy occurs where the derivative (called variation) of the potential energy
vanishes, i.e. δΠ =0 .
Assuming that the potential energy is a function of displacement u , its variation can be written as

 
δ  δu  0 , Assuming: δu  0  0
u u
For a system depends on n functions the variation of the total potential energy can be written as

  
 

 δ1  δu1 δu1  0 
 
 0
 u1 
  u1

  

 
 
      0
δ 2  δu2 δu2  0 

δ  δ1  δ 2    δ n  0   u2 Assuming : 
 u2

 
 


    
    

  
 
   0
δ n  u δun
 δun  0 
  u

 n  n


Note that based on calculus of variations, in the variation of total potential energy each derivative of the
function, i.e., u , u ′, u ′′... can be treated as separate functions, i.e., u1 u , u2 u  , u3 u  ….

Fundamentals of Finite Element Analysis of Structures Dr Emre Erkmen

Lecture 2 1
Total potential energy of a linear elastic spring
For an applied load of P on a spring with stiffness constant k, the corresponding displacement is u as
shown in the figure below. Note that external loads (or moments) in the positive displacement direction
(or rotation) produce negative potential. The total potential energy is the internal strain energy stored
during deformation U plus the external potential of the applied load which can be written as

Example 2 Problem with two variables

Determine the displacements a 1 and a2 for the mechanical system caused by the gradually applied static
force P = 5kN . The spring constants are k1 = 2kN / m and k2 = 3kN / m .
a1 a2
k1 k2
P

Figure: Mechanical system

The total potential energy Π in this problem can be written as

1 1
Π ( a1 , a2 )= k1a12 + k2 ( a2 − a1 ) − Pa2
2

2 2

Lecture 2 2
The necessary condition for the total potential energy to be minimum is that it should be a minimum with
respect to both variables a 1 and a 2 . Therefore, we need to calculate the partial derivatives as
∂Π
= 2a1 − 3 ( a2 − a1 ) = 0
∂a1

∂Π
= 3 ( a2 − a1 ) −=
5 0
∂a2

Solving for a 1 and a 2 gives

5 25
a1 = m a2 = m
2 6

The total potential energy value then becomes

2 2
 5 25  1  5  1  25 5  25
Π ,  = 2   + 3 −  − 5 = −10.41
2 6  2 2 2  6 2 6

a1 4.17 2.5
a2
-10.41

Figure: Total potential energy function and its minimum at a1 = 2.5 and a2 = 4.17

5  25 5 
Note that k1a=1 2= 5kN and k2 ( a2 − a1=) 3  − = 5kN , therefore the system is in equilibrium
2  6 2

5kN 5kN 5kN 5kN 5kN

Figure: System in equilibrium under load P=5kN

- Total potential energy formulation for bar analysis

Recall-Mechanics of Solids

x ∆x N
N N N N
∆x
L u
∆u
N
σ σ A
∆u du ( x) N
Strain:
= ε ( x) lim
= (Strain-deflection relation) σ= (Equilibrium equation)
∆x → 0 ∆x dx A
du ( x) N
σ = Eε (Hooke’s law) =
dx EA

Lecture 2 3
For an axially loaded bar,
q N0 q NL
N N+dN dN d 2u
EA +q =0 EA +q =0 L
dx dx 2
x,u
dx
2
1 1  du 
the strain energy can be written as U  E   2 dV  EA   dx .
2 V 2  dx 
L

Considering the possible boundary conditions, the potential energy functional becomes,
L L
1
Π (=
u, u′) EA( u ′ ) dx − ∫ qudx=
2

20∫ 0
+ N 0 u x 0=
− NL u x L

For the total potential energy to be minimum the first variation of the total potential energy functional
should vanish, i.e.
L L
δΠ = ∫ EAu ′δu ′dx − ∫ qδudx + N 0 δu x=0 − N L δu x= L = 0
0 0
L L
L
By using integration by parts, i.e. ∫ u ′ δu ′dx =
− ∫ u ′′ δu dx + u ′ δu 0 the above equation becomes
0 0
L L
δΠ = − ∫ EAu ′′δudx − ∫ qδudx + (EAu ′)δu 0 + N 0 δu x=0 − N L δu x= L = 0
L

0 0

in which δu is any arbitrary function (called test function or the weighting function). The differential
equilibrium equation in general can be written as
d 2u
EA +q =0
dx 2
For a case with no specified tip loads, the possible boundary conditions are
du
EA = N0 or δu ( 0 ) = 0 ( u ( 0 ) is specified)
dx x =0

du
EA = NL or δu ( L ) = 0 ( u ( L ) is specified)
dx x=L

- Total potential energy formulation for Euler-Bernoulli beam analysis

q
q
v M M+dM d2M d 4v M0 ML
EI +q =0 EI 4 + q =0
x dx 2 dx V0 VL
V L
dx V+dV
d 2v
For Euler-Bernoulli beam analysis considering that ε xx =− y , σ x = Eε x , σ y= ε y= τ xy= γ xy= 0 and
dx 2
2
1 1  d 2v 
∫A 2 V∫
ε
2 ∫L  dx 2 
2 2
y dA = I , The strain energy can be written as U = E d V = EI   dx

L L 2
1  d 2v  dv dv
Π(v)
= ∫ EI  2 =
2 0  dx 
dx + ∫ qvdx − V0 v x 0=
0 =
+ M0 + VL v x L − M L
dx x 0 = dx x L

Lecture 2 4
The first variation of the total potential energy functional should vanish, i.e.

L L
d 2v  d 2v  dv dv
=δΠ ∫ EI 2 δ  2= dx + ∫ qδvdx − V0 δv x 0=
+ M0 δ + VL δv x L − M L δ
0
dx  dx  0 = dx x 0 = dx x L

By using integration by parts

L
=x L=x L
 d 4v   d 3v   d 2 v  dv dv dv
δΠ= ∫  EI 4 + q  δvdx −  EI 3  δv +  EI= 2 
δ − V0 δv x 0=
+ M0 δ + VL δv x L − M L δ = 0
0
dx  =  dx  x 0=  dx  dx x 0= dx x 0= dx x L

The first term gives the differential equilibrium equation as

d 4v
EI = −q
dx 4
The second and third terms produce the possible boundary conditions of the problem
d 3v
EI = V0 or δv( 0 ) = 0 ( v( 0 ) is specified)
dx 3 x =0

d 3v
EI = VL or δv( L ) = 0 ( v( L ) is specified)
dx 3 x=L

d 2v dv
EI = M0 or δ = 0 ( v′( 0 ) is specified)
dx 2 x =0
dx x =0

d 2v dv
EI 2 = ML or δ = 0 ( v′( L ) is specified)
dx x=L
dx x=L

- Total potential energy formulation for Timoshenko beam analysis

−θ + v′≠ 0 how will be the equilibrium equations,
If the normality assumption is relaxed, i.e. γ xy =

v’ q
M0 ML
V0 VL
L

∂u dθ
For Timoshenko beam analysis considering the corresponding strains ε xx = =− y , σ x =Eε x ,
∂x dx
τ xy =Gγ xy , −θ + v′ ,
γ xy = σ y= ε y= 0 , ∫ y dA= I and ∫ dA= A ,
2
the potential energy functional
A A
L
1
Π ∫ ( ε xσ x + ε yσ y + γ xyτ xy ) dV +∫ q y vdx=
= M 0 θ x 0=
− V0 v x 0 += MLθ x
+ VL v x L −= becomes
2V 0
L

L 2 L 2 L
1  dθ  1  dv 
Π (θ , v )
= ∫ EI   dx + GA∫  − θ  dx + ∫ qvdx= M 0 θ x 0=
− V0 v x 0 += MLθ x
+ VL v x L −=
2 0  dx  2 0
dx  0
L

Lecture 2 5
The first variation of the total potential energy functional should vanish, i.e.
L L
dΠ dθ  dθ   dv 
0 ⇒ ∫ EI
= δ dx − GA∫  − θ δθ=
+ M 0 δθ x 0=
− M L δθ x 0
=
dθ 0
dx  dx  0  dx 
L

L L
dΠ  dv  dv
0 ⇒ GA∫  − θ δ dx + ∫ qδvdx=
= − V0 δv x 0=
+ VL δv x 0
=
dv 0
dx  dx 0
L

By using integration by parts

L
 d 2θ  dv 
L
  d 2 v dθ  
δΠ = 0 ⇒ − ∫  EI 2 + GA  − θ   δθ dx − ∫ GA  2 −  − q  δvdx
0 
dx  dx  0   dx dx  
x=L x=L
 dθ    dv  
+  EI δθ  + GA  − θ  δv  = − V0 δv x 0=
+ VL δv x 0
=
 dx  x = 0   dx
L
  x =0

The first two terms provide two differential equilibrium equations

d 2θ  dv 
EI 2
+ GA  − θ  =
0
dx  dx 

 d 2 v dθ 
GA  2 − −q =0
 dx dx 

The third and fourth terms produce the possible boundary conditions of the problem
dθ
EI = M0 or δθ = 0 ( θ ( 0 ) is specified)
dx x =0
x =0

dθ
EI = ML or δθ = 0 ( θ ( L ) is specified)
dx x=L
x=L

 dv 
GA  − θ  = −V0 or δv( 0 ) = 0 ( v( 0 ) v is specified)
 dx  x =0
 dv 
GA  − θ  −VL
= or δv( L ) = 0 ( v( L ) v is specified)
 dx  x=L

Direct variational method-Approximate solution via Ritz method

The method developed by Ritz applies the principle of minimum potential energy. In the Ritz method we
approximate the dependent unknown of a given problem by a linear combination of the form
N
u = ∑ Ci fi
i =1

where fi are some individual functions and Ci are unknown constants to be determined from the
minimum of the total potential energy. Introducing the expression for u, we obtain the total potential
energy as a function of the unknown coefficients Ci , i.e.
Π = Π ( C1 , C2 ,...)

Lecture 2 6
Therefore, the extremum problem of calculus of variations is transformed into the problem of differential
N
∂Π
calculus. δΠ =∑ δCi
i =1 ∂Ci

The minimum of the potential energy is found as

∂Π ∂Π ∂Π
=0, = 0, … = 0, …
∂C1 ∂C2 ∂Ci

The individual approximation functions fi must satisfy the following conditions,

• The kinematic boundary conditions

• They should be continuous

• They should be linearly independent, in other words none of them can be expressed by a linear
combination of the others

• They should not produce zero energy.

In summary the following steps are followed in Ritz approximate solution technique
Step 1 Assume an approximate solution;
Step 2 Write the total potential energy function;
Step 3 Perform minimization;
Step 4 Solve simultaneous equations.

Example–Simply supported beam under uniform load

Given v = ex 4 + dx3 + cx 2 + bx + a determine the deflection function v of the simply supported beam
caused by the uniformly distributed load q. The cross-sectional bending rigidity is EI.
q

+ v(x) v’
+

x
L
The essential boundary conditions are those that are related to the displacement function v . Boundary
conditions that are related to the moment are not essential.

For a given trial function, v = ex 4 + dx3 + cx 2 + bx + a

v( x = 0) = 0 → a = 0 and v( x =L) =0 → b =−eL3 − dL2 − cL

Note that we do not have to satisfy M ( x= 0)= 0 and M (=x L=

) 0.
Now we can use the trial function
v = ex 4 + dx3 + cx 2 − ( eL3 + dL2 + cL ) x
and seek the minimum of the total potential energy.
L L
1
( c, d , e ) ∫ EI (12ex 2 + 6dx + 2c ) dx + ∫ q ex 4 + dx3 + cx 2 − ( eL3 + dL2 + cL ) x dx
2
Π=
0
2 0

Lecture 2 7
 ∂Π 1 1 3
= 2 EI ( 4eL3 + 3dL2 + 2cL ) + qL3 − =qL 0
∂c 3 2

∂Π 1 1 4
= 6 EI ( 3eL4 + 2dL3 + cL2 ) + qL4 − =qL 0
∂d 4 2

∂Π  12 3 2  1 1 5
= 12 EI  eL5 + dL4 + cL3  + qL5 − = qL 0
∂e  5 2 3  5 2
From the solution of the above set of equations
q qL qL3
e= − , d= , c=0 −eL3 − dL2 − cL ⇒ b =−
b=
24 EI 12 EI 24

The deflection function is

1  q qL 3 qL3 
v( x) =  − x 4 + x − x
EI  24 12 24 
Note that the conditions M ( x= 0)= 0 and M (=x L=
) 0 are satisfied as a result of the solution
d 2v q qL
M=
EI 2 = − x2 + x
dx 2 2

Example–Approximate solution for Euler-Bernoulli beam bending

Given v = dx3 + cx 2 + bx + a determine the deflection function v of the simply supported beam caused by
the uniformly distributed load q. The cross-sectional bending rigidity is EI.

The essential boundary conditions should be satisfied for the given trial function, v = dx3 + cx 2 + bx + a

v( x = 0) = 0 → a = 0 and v( x =L) =0 → b =−dL2 − cL

Now we can use the trial function

v = dx 3 + cx 2 − ( dL2 + cL ) x

and seek the minimum of the total potential energy.

L L
1
( c, d ) ∫ EI ( 6dx + 2c ) dx + ∫ q  dx3 + cx 2 − ( dL2 + cL ) x dx
2
Π=
0
2 0

∂Π 1 1 3
= 2 EI ( 3dL2 + 2cL ) + qL3 − qL
= 0
∂c 3 2

∂Π 1 1 4
= 6 EI ( 2dL3 + cL2 ) + qL4 − =qL 0
∂d 4 2

qL2 qL3
d = 0, c = −dL2 − cL ⇒ b =−
b=
24 EI 24 EI

Lecture 2 8
The deflection function is
qL2 2 qL3
= v( x) x − x
24 EI 24 EI

Note that the conditions M ( x= 0)= 0 and M (=x L=

) 0 is also not satisfied as a result of the solution
d 2 v qL2
=M EI
=
dx 2 12

For q=EI=L=1 the deflection and moment functions are plotted and compared with the exact ones

0 0.14
0 0.2 0.4 0.6 0.8 1 0.12
-0.002
0.1
-0.004
Deflection v

Moment
0.08
-0.006 Error in
approximation 0.06
-0.008
0.04
-0.01 Exact
0.02 Approximate
-0.012 0
Exact
-0.014 Approximate 0 0.2 0.4 0.6 0.8 1
x x

Approximate
solution
e a,b,c,d

Exact solution

Geometric interpretation of the approximate solution

Above figure indicates that for different selections of the trial function, the total potential energy provides
different minimums. Unless there are sufficient parameters the solution is approximate and the
approximate solution can be conceived as the solution sought in a plane rather than in a wider space. The
minimum of the total potential energy functional does not necessarily match that within the cut plane.

Example- Euler-Bernoulli beam

For the given trial function v = dx3 + cx 2 + bx + a by using the minimum of total potential energy
principle for the Euler-Bernoulli beam, find the tip deflection of the cantilever under the applied point
load of P. The cross-sectional bending rigidity is EI , the span of the beam is L
P

EI + v(x)
x
L

Lecture 2 9
For a given trial function, v = dx3 + cx 2 + bx + a

dv( x = 0)
v( x = 0) = 0 → a = 0 and =0→b=0
dx
Now we can use the trial function
=v dx 3 + cx 2

and seek the minimum of the total potential energy.

L
1
Π ( c=
,d) ∫ 2 EI ( 6dx + 2c ) dx + P ( dL + cL )
2 3 2

∂Π
= 2 EI ( 3dL2 + 2cL ) + =
PL2 0
∂c

∂Π
= 6 EI ( 2dL3 + cL2 ) +=
PL3 0
∂d
From the solution of the above set of equations
P PL − Px 2
d= , c= − and the deflection function can be written =
as v( x) ( 3L − x ) ,
6 EI 2 EI 6 EI
− PL3
The tip deflection can be calculated as v(=x L=)
3EI
M=-PL(1-x/L)

d 2v  x x
The moment function can be written as M ( x ) =
EI 2 = − PL 1 − 
dx  L

Example- Euler-Bernoulli beam-revisited

For the given trial function v = cx 2 + bx + a by using the minimum of total potential energy principle for
the Euler-Bernoulli beam, find the tip deflection of the cantilever under the applied point load of P. The
cross-sectional bending rigidity is EI , the span of the beam is L
P

EI + v(x)
x
L

For a given trial function, v = cx 2 + bx + a

dv( x = 0)
v( x = 0) = 0 → a = 0 and =0→b=0
dx
Now we can use the trial function
v = cx 2
and seek the minimum of the total potential energy.
L
1
( c, d ) ∫ 2 EI ( 2c ) dx + P ( cL )
2 2
Π=
0

∂Π
= 2 EI ( 2cL ) +=
PL2 0
∂c

Lecture 2 10
From the solution of the above set of equations
PL − PLx 2
c= − and the deflection function can be written as v( x) =
4 EI 4 EI
− PL3
The tip deflection can be calculated as v(=x L=
)
4 EI
M=-PL(1-x/L)
M=-PL/2

d 2 v − PL x
The moment function can be written =
as M EI
=
dx 2 2

Example –Timoshenko beam

For the given trial functions v = dx3 + cx 2 + bx + a and θ= d ( 3 x 2 + 6 EI GA ) + 2cx + b , and by using them
in the minimum of total potential energy principle for the Timoshenko beam, find the deflection function
of the cantilever under the applied point load of P. The cross-sectional bending rigidity is EI , shear
rigidity is GA and the span of the beam is L .
P

+ v(x) θ
+

EI,GA
x
L

For a given trial functions, v = dx3 + cx 2 + bx + a and θ= d ( 3 x 2 + 6 EI GA ) + 2cx + b

EI
v( x = 0) = 0 → a = 0 and θ ( x =0) → b =−6d
GA
Now we can use the trial function
 EI 
v= d  x3 − 6 x  + cx 2 and θ 3dx 2 + 2cx
=
 GA 

L 2 L 2
1 dθ 1 dv
in Π
= (θ , v ) ∫ EI   dx + GA∫  − θ  dx + Pv ( L )
2 0  dx  2 0
dx 

and seek the minimum of the total potential energy, i.e.,

∂Π
= EI ( 4cL + 6dL2 ) + PL=
2
0
∂c

∂Π   EI    EI 
= EI 6cL2 + d 12 L3 + 36 L   + P  L3 − 6  0
L=
∂d   GA    GA 
From the solution of the above set of equations

P PL
d= , c= −
6 EI 2 EI

and the deflection function can be written as

Lecture 2 11
 Px 2 Px
v( x) =− ( 3L − x ) −
6 EI GA

PL3 PL PL3 ( 4 + α ) 12EI

The tip deflection can be calculated as v( L) =
− − =
− in which α =
3EI GA EI 12 GAL2

P 2 PL
as θ
Rotation function can be written= x − x
2 EI EI

PL2
The tip rotation can be calculated as θ ( L ) = −
2 EI

Lecture 2 12