Introduction to the Finite Element Method,

4th Edition
J. N. Reddy Chapter 5 Problem 5.11

Problem Introduction
Textbook Problem
Use the minimum number of Euler-Bernoulli beam nite elements to analyze the
beam shown below. In particular, give:
a. The assembled stiffness matrix and force vector;
b. The speci ed global displacements and forces, and the equilibrium
conditions,
c. The condensed matrix equations for the unknown generalized displacement
and unknown generalized forces separately.

Exploit symmetries, if any, in analyzing the problems. The positive convention

used for the generalized displacement and forces is the same as that shown in
Fig. 5.2.3 (https://www.accessengineeringlibrary.com/content/book/97812598
61901/toc-chapter/chapter5/section/section3#/ch05 g05).

 
 

Objective, Steps and Wrap-up

Problem Objective
This problem is similar to those in Chapter 4, where the nite element
procedure was applied to heat transfer, uid, and structural problems using
known stiffness and forcing vector equations. The difference here, with Euler-
Bernoulli beam elements, is there are two primary variables per node, thus
Hermite interpolation functions are required. Otherwise, the equations solve
very similarly.

GAMEPLAN
The problem statement asks for speci c items, but you will notice they are all
necessary parts of the six steps of the nite element method, as detailed in
Table 3.2.1 (https://www.accessengineeringlibrary.com/content/book/978125
9861901/toc-chapter/chapter3/section/section6#/c9781259861901ch03lev2s
ec04). As such, we will follow the typical approach:
1. Mesh the domain
2. Develop the elemental stiffness matrix and forcing vector
3. Assemble the global stiffness matrix and forcing vector
4. Impose boundary conditions to constrain the system
5. Solve the system of constrained nite element equations
6. Postprocess the results as necessary

 
 

Step 1
Mesh the domain

The problem statement requests that the fewest Euler-Bernoulli beam elements
be used. Recall that boundary conditions can only be applied at nodal locations.
Since there is a rigid constraint at x = 0, a roller constraint at x = 5 m, and a free
end at x = 10 m, this implies 3 Nodes are required.
A schematic representation of the resulting mesh is shown below, where
encircled numbers represent element numbering and plain numbers represent
node numbering:

Tip

Remember: each of the nodes has two degrees of freedom associated

with it (i.e. displacement in the transverse direction and nodal rotation).

 
 

Step 2
Develop the elemental stiffness matrix and forcing vector

The governing equations for a Euler-Bernoulli beam element is provided in

Equation (5.2.24) (https://www.accessengineeringlibrary.com/content/book/9
781259861901/toc-chapter/chapter5/section/section3#ch05eq5.2.24a). In
this example, the elastic modulus, second moment of area and distributed load
are all constant in x, thus, the simpli ed Equation (5.2.26) (https://www.access
engineeringlibrary.com/content/book/9781259861901/toc-chapter/chapter5/s
ection/section3#ch05eq5.2.26a) can be used. Noting that the foundation
stiffness, kef , for this problem is zero, the elemental equations are

e e e e
[K ]{Δ } = {q } + {Q }

e e
6 −3he −6 −3he v ⎫ ⎧ V1 ⎫
⎡ ⎤⎧ 1 ⎧ 6 ⎫
2 2 e e e
2Ee Ie −3he 2he 3he he θ q he −he M
1 0 1
⎨ ⎬ = ⎨ ⎬ + ⎨ ⎬
3 e e
h −6 3he 6 3he v 12 6 V
e 2 2
⎩ ⎭
⎣ 2 2 ⎦⎩ e
⎭ ⎩ e
⎭
−3he he 3he 2he θ he M
2 2

where
- vei represents a transverse displacement at Node i
- θei represents a nodal rotation at Node i
- Vie represents the shear load carried by the element at Node i
- Mie represents the bending moment carried by the element at Node i
- Ee is the elastic modulus of the beam
- Ie is the second moment of area of the beam
- he is the length of the element
- q0e represents the magnitude of the uniform distributed load acting on the
beam

 
 

 
⇒

Tip
⇒

Element 2:

⇒
2E1 I1

1
[K ] =
h

2E2 I2

2
[K ] =
h

2
3
1

[K ] = 1 × 10

3
2

[K ] = 1 × 10

⎢ ⎥⎪
The stiffness matrices and distributed loading vector for the two elements are
computed below:

Element 1:

⎡
−3h1

⎣−3h

−3h2

⎣−3h
6

−6

2(4 × 10 )

(5)

−6

2(4 × 10 )

(5)
3

3
6

3
−3h1

−3h2

⎣
2h

3h1

384

−960

−384

−960

2h

3h2

384

−960

−384

−960
2
1

2
2
2
1

−3(5)

2
2

−3(5)
6

−6

−3(5)

−6

−3(5)
−6

3h1

3h1

−960

3200

960

1600

−6

3h2

3h2

−960

3200

960

1600
−3h1

3h1

2h

−3(5)

2(5)

3(5)

2
(5)

−3h2

3h2

2h

−3(5)

2(5)

3(5)

2
(5)

IMPORTANT NOTE: The units are not consistent among the entries in the
h

2
2
1

2
1

−384

960

384

960

2
2

2
2

−384

stiffness or forcing vector. This is because the degrees of freedom include

a mixture of displacements and rotations. The units have been left off to
minimize clutter, but the student should do spot checks to ensure the units
are consistent when solving the nite element equations.
960

384

960
⎤

⎤
⎧v ⎫

⎦⎩

−6
⎨
1

3(5)

3(5)

⎧v ⎫

⎦⎩

−6
⎨
1

3(5)

3(5)
θ

θ
1
1

1⎭
2

2
1

2
2
1

1
2

−960

1600

960

3200

2
2

−960

1600

960

3200
⎬ =

−3(5)

(5)

3(5)

2(5)

⎬ =

⎭
⎤

−3(5)

(5)

3(5)

2(5)

⎦
2

2
2

2
q h1

q h2

,
1
0

12

2
0

12

⎦
,

,
⎧

⎩
−h1

h1

−h2

h2

2
6

⎭
⎫

⎬ + ⎨

⎫
⎩

{q } =

⎬ + ⎨

2
⎩

{q } =
M

M
V

V
2

2
⎧V

⎧V

2
1

1⎭

⎭
1

2
1

(400)(5)

3
{q } = 1 × 10 ⎨
12

1
⎧

−0.833

(400)(5)

3
{q } = 1 × 10 ⎨
12

1
⎧

−0.833

⎩
⎬

⎬
⎫

⎫
⎧

0.833

0.833
(5)
6

−(5)

−(5)

(5)
6

⎭
⎫

⎭
 

Step 3
Assemble the global stiffness matrix and forcing vector

Since there are 3 nodes in the system, and each node as 2 degree of freedom,
the global stiffness matrix will be 6 x 6 and the global forcing vector will be 6 x
1. Using the direct assembly approach (i.e. superposition of stiffnesses and
forces) as discussed in Section 3.3 (https://www.accessengineeringlibrary.co
m/content/book/9781259861901/toc-chapter/chapter3/section/section10#/c
h03para64)., the global system of equations [K]{Δ} = {q} + {Q} (see
Equation (5.2.23) (https://www.accessengineeringlibrary.com/content/book/9
781259861901/toc-chapter/chapter5/section/section3#ch05eq5.2.23)) is
assembled as follows.
1. Create a 6 x 6 matrix of zeros to initialize the global stiffness matrix and a 6 x
1 vector of zeros to initialize the global forcing vector.

0 0 0 0 0 0 v ⎧0⎫ ⎧0 ⎫
⎡ ⎤⎧ 1 ⎫

0 0 0 0 0 0 θ1 0 0

0 0 0 0 0 0 v2 0 0
⎨ ⎬ = ⎨ ⎬ + ⎨ ⎬
0 0 0 0 0 0 θ2 0 0

0 0 0 0 0 0 v3 0 0
⎣ ⎦ ⎩ ⎭ ⎩ ⎭ ⎩ ⎭
0 0 0 0 0 0 θ3 0 0

 
 

2. Superimpose the elemental stiffness matrix for the rst element [K 1 ] onto
the global stiffness matrix and the elemental distributed load vector and
forcing vector onto the associated global vectors.

Note that elemental degrees of freedom v11 , θ11 , v12 , and θ12 correspond to
global degrees of freedom v1 , θ1 , v2 , and θ2 , respectively. Therefore, [K 1 ]
shall go into rows and columns 1, 2, 3, and 4 (i.e. the rows and columns in
the global stiffness matrix associated with degrees of freedom v1 , θ1 , v2 ,
and θ2 ). This is similarly done for the forcing vectors.

⎧V 1 ⎫
⎡ 384 −960 −384 −960 0 0 ⎤⎧ v1 ⎫ ⎧ 1 ⎫
1

1
−960 3200 960 1600 0 0 θ1 −0.833 M
1

1
−384 960 384 960 0 0 v2 1 V
3 3 2
1 × 10 ⎨ ⎬ = 1 × 10 ⎨ ⎬+ ⎨ ⎬
−960 1600 960 3200 0 0 θ2 0.833 1
M
2

0 0 0 0 0 0 v3 0 0
⎩ ⎭ ⎩ ⎭ ⎩ ⎭
⎣ 0 0 0 0 0 0⎦ θ3 0 0

Superimpose the elemental stiffness matrix for the second element [K 2 ] onto
the global stiffness matrix and the elemental distributed load vector and forcing
vector onto the associated global vectors.

Note that elemental degrees of freedom v21 , θ21 , v22 , and θ22 correspond to
global degrees of freedom v2 , θ2 , v3 , and θ3 , respectively. Therefore, [K 2 ]
shall go into rows and columns 3, 4, 5, and 6 (i.e. the rows and columns in
the global stiffness matrix associated with degrees of freedom v2 , θ2 , v3 ,
and θ3 ). This is similarly done for the forcing vectors.
1
⎧ V ⎫
⎡ 384 −960 −384 −960 0 0 ⎤ ⎧v1 ⎫ ⎧ 1 ⎫ 1

1
−960 3200 960 1600 0 0 θ1 −0.833 M
1

1 2
−384 960 384 + 384 960 − 960 −384 −960 v2 1 + 1 V + V
3 3 2 1
1 × 10 ⎨ ⎬ = 1 × 10 ⎨ ⎬ + ⎨ ⎬
1 2
−960 1600 960 − 960 3200 + 3200 960 1600 θ2 0.833 − 0.833 M + M
2 1

0 0 −384 960 384 960 v3 1 2

V
2
⎣ ⎩ ⎭ ⎩ ⎭
0 0 −960 1600 960 3200 ⎦ θ3 0.833 ⎩ 2 ⎭
M
2

 
 

 
Step 4
3
1 × 10

⎢ ⎥⎪
Thus, the unconstrained system of equations is

⎣
384

−960

−384

−960

0
−960

3200

960

1600

0
−384

960

768

−384

−960
−960

Apply boundary conditions and forces

1600

6400

960

1600
0

−384

960

384

960
0

−960

1600

960

3200 ⎦

and Node 2 is constrained in the transverse direction.

This means that v1 ,
= 0 θ1 = 0
⎤

, and v2
⎧v1 ⎫

⎩
θ1

v2

θ2

v3

θ3
⎭

As detailed in Section 3.4.5 (https://www.accessengineeringlibrary.com/conten

t/book/9781259861901/toc-chapter/chapter3/section/section17#/c97812598
61901ch03lev2sec17), every global degree of freedom will either have the
primary variable known (i.e. the global nodal displacement or rotation) or the
secondary variable known (i.e. the external point load or point moment
applied).
From the problem statement, we see that Node 1 is rigidly xed to the wall,

= 0

Since the displacements are known, that means the external reaction
forces and moments at those nodes must be unknown
The other degrees of freedom are not constrained, thus:
The nodal degrees of freedom θ2 , v3 , and θ3 are unknown
This means the external point loads and point moments for these degrees
.

of freedom are known. Since no point loads or moments are applied, these
terms are zero.
3
⎬ = 1 × 10 ⎨
⎧

⎩
1

−0.833

0.833
⎫

⎭
⎧

⎬ + ⎨

⎩
M
1
V
2

+ M
1

2
2
⎬
V
1

+ V

V
2

M
1

1
1

2
2
2
1

1
⎫

⎭
  

Let's consider the global forcing vector {F  }for a moment. From Equation
(5.2.25) (https://www.accessengineeringlibrary.com/content/book/978125986
1901/toc-chapter/chapter5/section/section3#ch05eq5.2.25) we have

{F  } = {q} + {Q}
  





  
global nodal loads external distributed loads generalized element f orces

Also, as described in Section 3.2 (https://www.accessengineeringlibrary.com/c

ontent/book/9781259861901/toc-chapter/chapter3/section/section6#/c9781
259861901ch03lev2sec04), we have the equilibrium conditions at each node.
As shown in Equation (5.2.30) (https://www.accessengineeringlibrary.com/con
tent/book/9781259861901/toc-chapter/chapter5/section/section3#ch05eq5.
2.30), we see that this results in the generalized element forces being
equivalent to any external point forces or moments applied directly to a node.
Thus, we can write the global nodal load vector as:

⎧ V1 ⎫ ⎧ 1 ⎫ ⎧ unknown reaction f rom wall constraint ⎫

M1 −0.833 unknown moment f rom wall constraint

V2 2 unknown reaction f rom roller constraint
3
⎨ ⎬ = 1 × 10 ⎨ ⎬ + ⎨ ⎬
M2 0 0

V3 1 0
⎩ ⎭ ⎭ ⎩ ⎭ ⎩
M3 0.833 0






        
global nodal loads external distributed loads generalzed element f orces = nodal point loads and moments

 
 

We see that the rst three terms in the loading vector are unknown, due to the
reactions from the boundary conditions, thus we can write the global,
constrained system of equations as

384 −960 −384 −960 0 0 ⎧0 ⎫ ⎧ V1 ⎫

⎡ ⎤

−960 3200 960 1600 0 0 0 M1

−384 960 768 0 −384 −960 0 V2

3
1 × 10 ⎨ ⎬ = ⎨ ⎬
−960 1600 0 6400 960 1600 θ2 0

0 0 −384 960 384 960 v3 1000

⎩ ⎭ ⎩ ⎭
⎣ ⎦
0 0 −960 1600 960 3200 θ3 833

Once again, notice that when the primary variable (i.e. displacement/rotation
value) is known, the secondary variable (i.e. the external load/moment) is
unknown, and vice-versa.

Step 5
Solving the constrained system of equations

Our system of nite element equations [K]{Δ} = {F  } has knowns on both

sides of the equation (i.e. in both {Δ} and {F  }).
From a linear algebra point of view, this is troublesome. We would like all the
unknowns to be on one side of the equation to make the system of equations
"easy" to solve.
This can be accomplished using the approach discuss in Section 3.4.5 (https://
www.accessengineeringlibrary.com/content/book/9781259861901/toc-chapte
r/chapter3/section/section17#/c9781259861901ch03lev2sec17) termed the
condensed equations for the unknown primary variables and the condensed
equations for the unknown secondary variables.

 
 

As this method has been explained in previous walk-throughs (e.g. P3.3), the
intermediate steps will be skipped for conciseness; only the partitioned
equations will be presented.

Sub step 1
The partitioned equations for our system is:

11 12 1 1
[K ] [K ] {U } {F }
[ ]{ } = { }
21 22 2 2
[K ] [K ] {U } {F }

such that

384 −960 −384 −960 0 0

⎡ ⎤ ⎡ ⎤
[K
11
] = 1 × 10
3
−960 3200 960 , [K 12
] = 1 × 10
3
1600 0 0
⎣ ⎦ ⎣ ⎦
−384 960 768 0 384 −960

−960 1600 0 6400 960 1600

⎡ ⎤ ⎡ ⎤
[K
21
] = 1 × 10
3
0 0 −384 , [K 22
] = 1 × 10
3
960 384 960
⎣ ⎦ ⎣ ⎦
0 0 −960 1600 960 3200

384 −960 −384 −960 0 0 ⎧0 ⎫ ⎧ V1 ⎫

⎡ ⎤

−960 3200 960 1600 0 0 0 M1

−384 960 768 0 384 −960 0 V2

3
1 × 10 ⎨ ⎬ = ⎨ ⎬
−960 1600 0 6400 960 1600 θ2 0

0 0 −384 960 384 960 v3 1000

⎩ ⎭ ⎩ ⎭
⎣ ⎦
0 0 −960 1600 960 3200 θ3 833

⎧ 0⎫ ⎧θ2 ⎫ ⎧ V1 ⎫ ⎧ 0 ⎫

{U
1
} = ⎨ 0⎬ {U , 2
} = ⎨v3 ⎬ {F , 1
} = ⎨M1 ⎬ , and {F 2
} = ⎨ 1000⎬
⎩ ⎭ ⎩ ⎭ ⎩ ⎭ ⎩ ⎭
0 θ3 V2 833

 
 

Using the second row of the partitioned system of equations, we nd:

−1
2 22 2 21 1
{U } = [K ] ({F } − [K ]{U })

Thus, we can solve for the unknown variables:

⎧θ2 ⎫ ⎧−0.00130 rad⎫
2
{U } = ⎨v3 ⎬ = ⎨ 0.01432 m ⎬
⎩ ⎭ ⎩ ⎭
θ3 −0.00339 rad

Using the rst row of the partitioned system of equations, we nd:

11 1 12 2 1
[K ]{U } + [K ]{U } = {F }

Thus, we can solve for the unknown variables:

⎧ V1 ⎫ ⎧ 1250 N ⎫
1
{F } = ⎨M1 ⎬ = ⎨ −2083 Nm ⎬
⎩ ⎭ ⎩ ⎭
V2 −2250 N

 
  

Step 6
Post-processing the results to compute the unknown variables.

There is no post-processing requested in this problem, but let's use this space
to do a sanity check on the result.
Recall the force equilibrium we computed in Step 4

⎧ V1 ⎫ ⎧ 1 ⎫ ⎧ unknown reaction f rom wall constraint ⎫

M1 −0.833 unknown moment f rom wall constraint

V2 2 unknown reaction f rom roller constraint
3
⎨ ⎬ = 1 × 10 ⎨ ⎬ + ⎨ ⎬
M2 0 0

V3 1 0
⎩ ⎭ ⎭ ⎩ ⎭ ⎩
M3 0.833 0
  





      
global nodal loads external distributed loads generalzed element f orces = nodal point loads and moments

We have also just nished computing the force vector, {F  } , where in this
example it was

⎧ 1250 N ⎫

−2083 Nm

−2250 N
{F  } = ⎨ ⎬
0

1000 N
⎩ ⎭
833 Nm

Thus, we can compute for the unknown reactions:

⎧ unknown reaction f rom wall constraint ⎫ ⎧ 1250 N ⎫ ⎧ 1000 N ⎫ ⎧ 250 N ⎫

unknown moment f rom wall constraint −2083 Nm −833 Nm −1250 Nm

unknown reaction f rom roller constraint −2250 N 2000 N −4250 N

⎨ ⎬ = ⎨ ⎬ − ⎨ ⎬ = ⎨ ⎬
0 0 0 0

0 1000 N 1000 N 0
⎩ ⎭ ⎩ ⎭ ⎩ ⎭ ⎩ ⎭
0 833 Nm 833 Nm 0

 
 

From this we can draw the Free Body Diagram of the beam:

If we sum the forces and moments, we see the beam is in static equilibrium, as
expected:

Sum of the forces in the vertical direction (up is positive):

∑ F = 250 N − 4250 N + (400 N/m)(10 m) = 0

Sum of the moments about the left end (clockwise is positive):

∑ M = 1250 Nm − (4250 N)(5 m) + (400 N/m)(10 m)(5 m) = 0

Wrap-up

Reminders for Moving Forward

Working with beam elements is just like working with the previous elements
you've seen (e.g. bar, heat transfer, uid ow), but now each node has two
degrees of freedom. The result is larger (and more complicated) nite element
matrices and vectors, but if you employ the 6-step method, you should be able
to complete all of the problems in this section.

 
 

Related Problems
Problems 5.7 (https://www.accessengineeringlibrary.com/content/book/97812
59861901/toc-chapter/chapter5/section/section27#ch05pb7) – 5.23 (https://
www.accessengineeringlibrary.com/content/book/9781259861901/toc-chapte
r/chapter5/section/section27#ch05 g23p) all deal with structural problems
involving beam elements

References
Section 5.2 (https://www.accessengineeringlibrary.com/content/book/978125
9861901/toc-chapter/chapter5/section/section3): overview of the Euler-
Bernoulli beam elements
Section 3.2.1 (https://www.accessengineeringlibrary.com/content/book/97812
59861901/toc-chapter/chapter3/section/section6#/c9781259861901ch03lev2
sec04): overview of the 6-step nite element method
Section 3.4.5 (https://www.accessengineeringlibrary.com/content/book/97812
59861901/toc-chapter/chapter3/section/section17#/c9781259861901ch03lev
2sec17): discussion of boundary conditions and solving for unknowns
Walk-through including in-depth instructions on solving constrained system of
equations: P3.3 (https://www.accessengineeringlibrary.com/content/book/978
1259861901/toc-chapter/chapter3/section/section38#/ch03pb3)