Middle East Technical University

Department of Mechanical Engineering

ME 413 Introduction to Finite Element Analysis

Chapter 4
Trusses, Beams and Frames

These notes are prepared by

Dr. Cüneyt Sert
http://www.me.metu.edu.tr/people/cuneyt
csert@metu.edu.tr

These notes are prepared with the hope to be useful to those who want to learn and teach FEM. You are free to use
them. Please send feedbacks to the above email address.
4-1
 What Is This Chapter About?
• We’ll study FEM formulations of
• deformation of planar trusses
• bending of beams
• deformation of frames (as the superposition of planar truss and beam
formulations)
• These problems will be studied as 1D, but there will be multiple unknowns at a node.
• We’ll modify the 1D FEM code to solve these problems.

METU – Dept. of Mechanical Engineering – ME 413 Int. to Finite Element Analysis – Lecture Notes of Dr. Sert 4-2
 Deformation of a Bar
• A bar is a structural member that is loaded axially.

𝐸, 𝐴
𝐹
𝑥

• It is either in direct tension or compression.

• Axial deformation, 𝑢, is governed by the following DE

𝑑 𝑑𝑢
− 𝐸𝐴 =0
𝑑𝑥 𝑑𝑥

solution of which is linear for constant 𝐸 and 𝐴.

• Even a single linear element can solve this problem exactly.

METU – Dept. of Mechanical Engineering – ME 413 Int. to Finite Element Analysis – Lecture Notes of Dr. Sert 4-3
 Deformation of a Bar (cont’d)
• Elemental weak form of the problem is

𝑑𝑢 𝑑𝑤 𝑑𝑢 𝑑𝑢
𝐸𝐴 𝑑𝑥 = 𝑤𝐸𝐴 + −𝑤𝐸𝐴
Ω 𝑒 𝑑𝑥 𝑑𝑥 𝑑𝑥 𝑥𝑒
𝑑𝑥 𝑥1𝑒
2
𝑄2𝑒 𝑄1𝑒

𝑑𝑢
• SV of the problem is the axial force : 𝐸𝐴 𝑛
𝑑𝑥 𝑥

𝑑𝑆𝑖 1 𝑑𝑆𝑗 1 𝑒
• Elemental stiffness matrix is 𝐾𝑖𝑗𝑒 = 𝐸𝐴 𝑒 𝑒
𝐽 𝑑𝜉
Ω𝑒 𝑑𝜉 𝐽 𝑑𝜉 𝐽

𝐸𝐴 1 −1 𝑢1𝑒 𝑄1𝑒
• Elemental system is =
ℎ𝑒 −1 1 𝑢2𝑒 𝑄2𝑒

Elemental force
vector is zero

METU – Dept. of Mechanical Engineering – ME 413 Int. to Finite Element Analysis – Lecture Notes of Dr. Sert 4-4
 Planar Truss
• A truss consists of several bars connected with frictionless pin joints.
• Note that this is not the actual meaning of truss in civil engineering.
𝐹1

𝐹2
• Each member can only carry axial force, but no shear force or bending moment.
• All members of a planar truss lie on a 2D plane. Space truss is the 3D version.
• A truss can be loaded with multiple point forces at its joints.
• Typically there is at least one fixed joint.
• Some joints might have restricted motion.
• Deformations are small, i.e. general shape of the truss is similar before and after
loading.

METU – Dept. of Mechanical Engineering – ME 413 Int. to Finite Element Analysis – Lecture Notes of Dr. Sert 4-5
 Planar Truss – Local Coordinates
• Each member of a truss can be treated as an element of a FE mesh.
• The elemental system derived previously for a bar is valid for each member.
• But in order to be able to use it, different coordinate systems aligned with each
member should be used. These local coordinates are shown below with 𝑥 1 and 𝑥 2 .

For the 1st member For the 2nd member

𝑄21
1 e=2 2 𝑄22
2 𝑄12
𝑥2
e=1 Nodal deflections of
e=1 in 𝑥 1 direction
1 𝑥 1 𝐸𝐴 1 −1 𝑢12 𝑄12
= 2
Nodal forces of e=1 ℎ2 −1 1 𝑢22 𝑄2
in 𝑥 1 direction
𝑄11

Nodal deflections Nodal forces of

𝐸𝐴 1 −1 𝑢11 𝑄11 of e=2 in 𝑥 2 e=2 in 𝑥 2
=
ℎ1 −1 1 𝑢12 𝑄21 direction direction

METU – Dept. of Mechanical Engineering – ME 413 Int. to Finite Element Analysis – Lecture Notes of Dr. Sert 4-6
 Planar Truss – Transformation Matrix
• During the assembly of the elemental systems, PVs and SVs written for a common 𝑥𝑦
coordinate system should be used.
• For each element a transformation between local 𝑥 coordinate and the global 𝑥𝑦
coordinates is necessary.
• This is a purely geometrical transformation.

𝑄2𝑒 𝑒
𝑄2𝑦 𝑒
𝑢1𝑥 = 𝑢1𝑒 cos(𝜃 𝑒 )
𝑢2𝑒 𝑒 𝑒
𝑄2𝑥 𝑒
𝑥𝑒 2 𝑢2𝑦 𝑢1𝑦 = 𝑢1𝑒 sin(𝜃 𝑒 )
𝑒
e 𝑢2𝑥
Multiply the 1st eqn with cos(𝜃 𝑒 )
1 𝜃𝑒
𝑢1𝑒 𝑒
and the 2nd eqn with sin 𝜃 𝑒 and
𝑦 𝑢1𝑦 add them up.
𝑄1𝑒 𝑒
𝑒 𝑢1𝑥
𝑄1𝑦
𝑒
𝑄1𝑥 𝑒
𝑢1𝑥 𝑒
cos 𝜃 𝑒 + 𝑢1𝑦 sin 𝜃 𝑒 = 𝑢1𝑒 cos 2 𝜃 𝑒 + sin2 𝜃 𝑒
1
𝑥
𝑢1𝑒 = 𝑢1𝑥
𝑒 𝑒
cos 𝜃 𝑒 + 𝑢1𝑦 sin 𝜃 𝑒

METU – Dept. of Mechanical Engineering – ME 413 Int. to Finite Element Analysis – Lecture Notes of Dr. Sert 4-7
 Planar Truss – Transformation Matrix (cont’d)
• Similary for the 2nd node of element e : 𝑢2𝑒 = 𝑢2𝑥
𝑒 𝑒
cos 𝜃 𝑒 + 𝑢2𝑦 sin 𝜃 𝑒
• Together these two eqns become
𝑒
𝑢1𝑥
𝑒
𝑢1𝑒 cos(𝜃 𝑒 ) sin(𝜃 𝑒 ) 0 0 𝑢1𝑦
=
𝑢2𝑒 0 0 cos(𝜃 𝑒 ) sin(𝜃 𝑒 ) 𝑒
𝑢2𝑥
𝑒
𝑢2𝑦
Transformation matrix, [𝑇 𝑒 ]

𝑢𝑒 = [𝑇 𝑒 ] Δ𝑒
Δ𝑒 includes both 𝑢𝑥𝑒 ’s and 𝑢𝑦𝑒 ’s.
Each node has 2 unknowns and in
total one element has 4 unknowns.
• A similar eqn can be written for the SVs too
𝑄 𝑒 = [𝑇 𝑒 ] 𝑄𝑒
𝑒 𝑒 𝑒 𝑒
𝑄𝑒 = 𝑄1𝑥 𝑄1𝑦 𝑄2𝑥 𝑄2𝑦 𝑇

METU – Dept. of Mechanical Engineering – ME 413 Int. to Finite Element Analysis – Lecture Notes of Dr. Sert 4-8
 Planar Truss – [𝐾 𝑒 ] Transformation Matrix
• [𝑇 𝑒 ] can be used to transform the original 2x2 elemental system into a new 4x4
elemental system

𝐸𝐴 1 −1 𝑢1𝑒 𝑄1𝑒
• Original 2x2 elemental system using bars : =
ℎ𝑒 −1 1 𝑢2𝑒 𝑄2𝑒
or
𝐾 𝑒 𝑢𝑒 = {𝑄 𝑒 }

• Using 𝑢𝑒 = [𝑇 𝑒 ] Δ𝑒 and 𝑄 𝑒 = [𝑇 𝑒 ] 𝑄 𝑒

𝐾 𝑒 [𝑇 𝑒 ] Δ𝑒 = [𝑇 𝑒 ] 𝑄𝑒

• Premultiply this eqn by 𝑇 𝑒 𝑇

𝑇𝑒 𝑇
𝐾 𝑒 [𝑇 𝑒 ] Δ𝑒 = 𝑇 𝑒 𝑇 [𝑇 𝑒 ] 𝑄 𝑒

𝐾𝑒 1

METU – Dept. of Mechanical Engineering – ME 413 Int. to Finite Element Analysis – Lecture Notes of Dr. Sert 4-9
 Planar Truss – Transformed [𝐾 𝑒 ]
• Transformed 4x4 elemental system : [𝐾 𝑒 ] Δ𝑒 = 𝑄 𝑒

𝑇𝑒 𝑇 𝐾𝑒 𝑇𝑒

𝛼 0
𝛽 0 𝐸𝐴 1 −1 𝛼 𝛽 0 0
0 𝛼 ℎ𝑒 −1 1 0 0 𝛼 𝛽
0 𝛽

𝛼2 𝛼𝛽 −𝛼 2 −𝛼𝛽
𝐸𝐴 𝛽2 −𝛼𝛽 −𝛽2
𝐾𝑒 = 𝑒
ℎ 𝛼2 𝛼𝛽
sym 𝛽2

where 𝛼 = cos(𝜃 𝑒 ) and 𝛽 = sin(𝜃 𝑒 )

METU – Dept. of Mechanical Engineering – ME 413 Int. to Finite Element Analysis – Lecture Notes of Dr. Sert 4-10
 Planar Truss – 4x4 Elemental System
• Therefore in a planar truss solution elemental systems are 4x4.
• Each truss member is a linear element with 4 unknows

𝛼2 𝛼𝛽 −𝛼 2 −𝛼𝛽 Δ𝑒1 𝑄1𝑒

𝐸𝐴 𝛽2 −𝛼𝛽 −𝛽2 Δ𝑒2 𝑄2𝑒
=
ℎ𝑒 𝛼2 𝛼𝛽 Δ𝑒3 𝑄3𝑒
sym 𝛽2 Δ𝑒4 𝑄4𝑒

𝑄4𝑒
Δ𝑒4 𝑄3𝑒 Δ𝑒1 : Horizontal deflection of point 1
e Δ𝑒3 Δ𝑒2 : Vertical deflection of point 1
Δ𝑒3 : Horizontal deflection of point 2
𝑦 𝜃𝑒 Δ𝑒4 : Vertical deflection of point 2
Δ𝑒2
Δ𝑒1 𝜃 𝑒 is measured CCW 𝑄1𝑒 : Horizontal force at point 1
𝑄2𝑒
from the positive 𝑥 𝑄2𝑒 : Vertical force at point 1
𝑄1𝑒 𝑄3𝑒 : Horizontal force at point 2
axis.
𝑥 𝑄4𝑒 : Vertical force at point 2

METU – Dept. of Mechanical Engineering – ME 413 Int. to Finite Element Analysis – Lecture Notes of Dr. Sert 4-11
 Planar Truss – Local to Global Unknown Mapping
• Consider the following truss problem.

2𝑃

3 𝑃

e=3 e=2

1 e=1 2

• There are 𝑁𝐸 = 3 elements and 𝑁𝑁 = 3 nodes.

• At each node there are 𝑁𝑁𝑈 = 2 unknown deflections. Totally there are 𝑁𝑈 = 6
unknowns.
• 4 of these unknowns are known. Nodes 1 and 2 are fixed.
• We need to determine 2 deflections (horizontal and vertical) at node 3 and, if desired,
the reaction forces at nodes 1 and 2.

METU – Dept. of Mechanical Engineering – ME 413 Int. to Finite Element Analysis – Lecture Notes of Dr. Sert 4-12
 Planar Truss – Local to Global Unknown Mapping (cont’d)
• PVs and SVs of each element are
For e=1 For e=2 For e=3
Δ24 , 𝑄42 Δ34 , 𝑄43

Δ12 , 𝑄21 Δ14 , 𝑄41

2 Δ23 , 𝑄32 Δ33 , 𝑄33
2
Δ13 , 𝑄31 Δ32 , 𝑄23
1 2
Δ11 , 𝑄11 Δ22 , 𝑄22
Δ21 , 𝑄12 1 Δ31 , 𝑄13
1

Δ6 , 𝑄6
• PVs and SVs of the global system are
Δ5 , 𝑄5
3
• In general the global PVs (and SVs) of
Δ2 , 𝑄2
the 𝑖 𝑡ℎ
node are numbered as 2𝑖 − 1 Δ4 , 𝑄4
and 2𝑖. 1 2
Δ3 , 𝑄3
Δ1 , 𝑄1
METU – Dept. of Mechanical Engineering – ME 413 Int. to Finite Element Analysis – Lecture Notes of Dr. Sert 4-13
 Planar Truss – Local to Global Unknown Mapping (cont’d)
• Assembly process is about local-to- global unknown mapping for each element

Δ11
1
Δ
Unknowns of e=1 : Δ1 = 12
Δ3
Δ1
Δ14
Δ2
Δ21
Δ22 Δ3
2
Unknowns of e=2 : Δ = =Δ Global unknowns
Δ23 Δ4
Δ24
Δ5
Δ31 Δ6
3
Δ 2
Unknowns of e=3 : Δ3 =
Δ33
Δ34
METU – Dept. of Mechanical Engineering – ME 413 Int. to Finite Element Analysis – Lecture Notes of Dr. Sert 4-14
 Planar Truss – 𝐿𝑡𝑜𝐺 Matrix & Assembly Rule
• This graph can also be expressed as a local-to-global mapping matrix
1 2 3 4 e=1
𝐿𝑡𝑜𝐺 = 3 4 5 6 e=2
1 2 5 6 e=3

• 𝐿𝑡𝑜𝐺𝑖𝑗 gives the global unknown number of the 𝑖 𝑡ℎ element’s 𝑗𝑡ℎ local unknown.
• For example, 𝐿𝑡𝑜𝐺34 = 6 because the 3rd element’s 4th local unknown is the 6th global
unknown.

The assembly rule can now be defined as

• 𝐾𝑖𝑗𝑒 entry of an elemental system goes to 𝐾𝐼𝐽 entry of the global system.
• 𝐹𝑖𝑒 entry of an elemental system goes to 𝐹𝐼 entry of the global system.
• 𝑄𝑖𝑒 entry of an elemental system goes to 𝑄𝐼 entry of the global system.

where 𝐼 = 𝐿𝑡𝑜𝐺𝑒𝑖
𝐽 = 𝐿𝑡𝑜𝐺𝑒𝑗

METU – Dept. of Mechanical Engineering – ME 413 Int. to Finite Element Analysis – Lecture Notes of Dr. Sert 4-15
 Planar Truss – Assembly
• Using the assembly rule assemble system of the 3-member truss is

1 3 1 3 1 1 3 3
𝐾11 + 𝐾11 𝐾12 + 𝐾12 𝐾13 𝐾14 𝐾13 𝐾14 Δ1 𝑄11 + 𝑄13
1 3 1 3 1 1 3 3
𝐾21 + 𝐾21 𝐾22 + 𝐾22 𝐾23 𝐾24 𝐾23 𝐾24 Δ2 𝑄21 + 𝑄23
1 1 1 2 1 2 2 2
𝐾31 𝐾32 𝐾33 + 𝐾11 𝐾34 + 𝐾12 𝐾13 𝐾14 Δ3 𝑄31 + 𝑄12
=
1
𝐾41 1
𝐾42 1
𝐾43 2
+ 𝐾21 1
𝐾44 2
+ 𝐾22 2
𝐾23 2
𝐾24 Δ4 𝑄41 + 𝑄22
3
𝐾31 3
𝐾32 2
𝐾31 2
𝐾32 2
𝐾33 3
+ 𝐾33 2
𝐾34 3
+ 𝐾34 Δ5 𝑄32 + 𝑄33
3 3 Δ6
3
𝐾41 𝐾42 2
𝐾41 2
𝐾42 2
𝐾43 + 𝐾43 2
𝐾44 3
+ 𝐾44 𝑄42 + 𝑄43

1 2 3 4
which only depends on the 𝐿𝑡𝑜𝐺 matrix 𝐿𝑡𝑜𝐺 = 3 4 5 6
1 2 5 6

• 𝐿𝑡𝑜𝐺 matrix depends on

• how we number the nodes globally
• how we number the elements’ nodes locally, i.e. which node is the 1st and which
one is the 2nd ?

METU – Dept. of Mechanical Engineering – ME 413 Int. to Finite Element Analysis – Lecture Notes of Dr. Sert 4-16
 Planar Truss – Point Loads
• Consider the following truss problem with 2 point loads.
2𝑃
• How should we use the points loads?
• They are used in the boundary term vector. 3 𝑃
• 𝑄 of this problem is
e=3 e=2
𝑄1 𝑄11 + 𝑄13
1 e=1 2
𝑄2 𝑄21 + 𝑄23 𝑦
𝑄3 𝑄31 + 𝑄12
𝑄 = = 𝑥
𝑄4 𝑄41 + 𝑄22
𝑄5 𝑄32 + 𝑄33
𝑄6
𝑄42 + 𝑄43

• If there is no horizontal (or vertical) force at a node, the corresponding SV is set to zero.
• If there is a given point load at a node, the corresponding SV is set to the given value.
• At supports SV(s) are unknown and can be calculated during post-processing.
• Be careful with the direction (sign) of forces.

METU – Dept. of Mechanical Engineering – ME 413 Int. to Finite Element Analysis – Lecture Notes of Dr. Sert 4-17
 Planar Truss – Point Loads
• For our problem
2𝑃
• 𝑄1 and 𝑄2 are unknown reaction forces at node 1.
• 𝑄3 and 𝑄4 are unknown reaction forces at node 2. 3 𝑃
• 𝑄5 = 𝑃 (given horizontal point load at node 3).
e=3 e=2
• 𝑄6 = −2𝑃 (given vertical point load at node 3).
𝑄1 1 e=1 2
𝑄2 𝑦
𝑄3 𝑥
• Therefore the {𝑄} vector is
𝑄4
𝑃
−2𝑃
• 𝑄1 , 𝑄2 , 𝑄3 and 𝑄4 are not known, but the corresponding Δ1 , Δ2 , Δ3 and Δ4 are known.
• If there were no horizontal load at node 3, we should have set 𝑄5 to zero.
• If there were no vertical load at node 3, we should have set 𝑄6 to zero.
• If there were a roller support that restricts vertical motion but not horizontal motion,
we should have set 𝑄4 to zero.

METU – Dept. of Mechanical Engineering – ME 413 Int. to Finite Element Analysis – Lecture Notes of Dr. Sert 4-18
 Example 4.1
e.g. Example 4.1 : Solve the following truss problem.
• Find the deflection of the nodes.
• Determine the forces and stresses in each member.
• Determine the reaction forces at the supports.

𝐸 and 𝐴 values are the same for each member.

2𝑃

𝑃
3

e=3 e=2 𝐿
1 e=1 2

METU – Dept. of Mechanical Engineering – ME 413 Int. to Finite Element Analysis – Lecture Notes of Dr. Sert 4-19
 Example 4.1 (cont’d)
• Elemental system equation is given in slide 2-10.
• 𝜃 values are necessary and they depend on local node numbering of the nodes

2
3
2

e=3
1 e=2

1
1 e=1 2

1 2

• 𝜃 values of the elements are : 𝜃1 = 0 , 𝜃 2 = 𝜋/2 , 𝜃 3 = 𝜋/4

• Note : If the local node numbering of all the elements are reversed, 𝜃 values
change as 𝜃1 = 𝜋 , 𝜃 2 = 3𝜋/2 , 𝜃 3 = 5𝜋/4

METU – Dept. of Mechanical Engineering – ME 413 Int. to Finite Element Analysis – Lecture Notes of Dr. Sert 4-20
 Example 4.1 (cont’d)
• Elemental systems are

𝛼 = cos 𝜃1 = 1 1 0 −1 0
𝐸𝐴 0 0 0
For e=1 : 𝛽 = sin 𝜃1 = 0 → 𝐾1 =
ℎ1 = 𝐿
𝐿 sym. 1 0
0

𝛼 = cos 𝜃 2 = 0 0 0 0 0
𝐸𝐴 1 0 −1
For e=2 : 𝛽 = sin 𝜃 2 = 1 → 𝐾2 =
ℎ2 = 𝐿
𝐿 sym. 0 0
1

3 1 1 −1 −1
𝛼 = cos 𝜃 = 1/ 2 𝐸𝐴 1 −1 −1
For e=3 : 𝛽 = sin 𝜃 3 = 1/ 2 → 𝐾3 =
2 2𝐿 sym. 1 1
ℎ3 = 2𝐿 1

METU – Dept. of Mechanical Engineering – ME 413 Int. to Finite Element Analysis – Lecture Notes of Dr. Sert 4-21
 Example 4.1 (cont’d)
• Assembled system is given in slide 4-16. With numbers it becomes

1 + 0.3536 0 + 0.3536 −1 0 −0.3536 −0.3536 Δ1 𝑄1

0 + 0.3536 0 0 −0.3536 −0.3536 Δ2 𝑄2
𝐸𝐴 1+0 0+0 0 0 Δ3 𝑄3
=
𝐿 0+1 0 −1 Δ4 𝑄4
symmetric 0 + 0.3536 0 + 0.3536 Δ5 𝑃
1 + 0.3536 Δ6 −2𝑃

• We know that
Δ1 = Δ2 = Δ3 = Δ4 = 0

• Reduction can be applied to get the following 2x2 system

0.3536 0.3536 ∆5 𝑃
=
0.3536 1.3536 ∆6 −2𝑃
• Solving this we get
𝑃𝐿 𝑃𝐿
∆5 = 5.828 , ∆6 = −3
𝐸𝐴 𝐸𝐴
METU – Dept. of Mechanical Engineering – ME 413 Int. to Finite Element Analysis – Lecture Notes of Dr. Sert 4-22
 Example 4.1 (cont’d)
• To calculate axial forces in each member we can go back to local coordinates aligned
with the elements. 𝑒
𝑄2

• From slide 4-9 1

𝑥𝑒

𝑄1𝑒 𝐸𝐴 1 𝑒
−1 𝑢1
= 𝑒 𝑄1𝑒
𝑄2𝑒 ℎ −1 1 𝑢2𝑒

• Using the transformation matrix definition from Slide 4-8

Δ𝑒1
𝑄1𝑒 𝐸𝐴 1 −1 Δ𝑒2
= 𝑒 𝑇𝑒
𝑄2𝑒 ℎ −1 1 Δ𝑒3
Δ𝑒4

METU – Dept. of Mechanical Engineering – ME 413 Int. to Finite Element Analysis – Lecture Notes of Dr. Sert 4-23
 Example 4.1 (cont’d)
• Axial forces in each member are

0
𝑄11 𝐸𝐴 1 −1 1 0 0 0 0 0
• For e=1 : = =
𝑄21 𝐿 −1 1 0 0 1 0 0 0
0

0
0
𝑄12 𝐸𝐴 1 −1 0 1 0 0 𝑃𝐿 3𝑃
• For e=2 : = 5.828 =
𝑄22 𝐿 −1 1 0 0 0 1 𝐸𝐴 −3𝑃
𝑃𝐿
−3
𝐸𝐴

0
0
𝑄13 𝐸𝐴 1 −1 1 1 1 0 0 𝑃𝐿 −1.414𝑃
• For e=3 : = 5.828 =
𝑄23 2𝐿 −1 1 2 0 0 1 1 𝐸𝐴 1.414𝑃
𝑃𝐿
−3
𝐸𝐴

METU – Dept. of Mechanical Engineering – ME 413 Int. to Finite Element Analysis – Lecture Notes of Dr. Sert 4-24
 Example 4.1 (cont’d)
• As seen for each member
𝑄1𝑒 = −𝑄2𝑒

i.e. net axial force on each member is zero.

• 1st element carries no axial force, as expected, because its both ends are fixed.
• 2nd element is in compression because 𝑄12 > 0 (or 𝑄22 < 0).
• 3rd element is in tension because 𝑄13 < 0 (or 𝑄23 > 0).

𝑄22 < 0 𝑄23 > 0

2 2
e=3
e=2
𝑥2
1
1 𝑥3
𝑄13 < 0
𝑄12 > 0

METU – Dept. of Mechanical Engineering – ME 413 Int. to Finite Element Analysis – Lecture Notes of Dr. Sert 4-25
 Example 4.1 (cont’d)
• Axial stresses in each element can be calculated.

𝑄21
For e=1 : 𝜎1 = =0
𝐴
𝑄22 𝑃
For e=2 : 𝜎2 = = −3 (Negative stress indicates compression)
𝐴 𝐴
𝑄23 𝑃
For e=3 : 𝜎3 = = 1.414 (Positive stress indicates tension)
𝐴 𝐴

• Finally forces at the supports can be calculated using the 6x6 system of Slide 4-22.
𝐸𝐴
At node 1 : 𝑄1 = 1.3536∆1 + 0.3536∆2 − ∆3 − 0.3536∆5 − 0.3536∆6 = −𝑃
𝐿
𝐸𝐴
𝑄2 = 0.3536∆1 + 0.3536∆2 − 0.3536∆5 − 0.3536∆6 = −𝑃
𝐿

𝐸𝐴
At node 2 : 𝑄3 = −∆1 + ∆3 = 0 𝑃 3𝑃 Forces at the
𝐿
𝐸𝐴 supports are the
𝑄4 = ∆4 − ∆6 = 3𝑃 opposite of the
𝐿 𝑃
calculated ones
METU – Dept. of Mechanical Engineering – ME 413 Int. to Finite Element Analysis – Lecture Notes of Dr. Sert 4-26
 Planar Truss – Constrained Motion
• Sometimes horizontal and vertical deflections of a node are not independent.
• This happens at a roller support inclined at an angle to the global 𝑥𝑦 system.

𝑖 𝛼

• At node 𝑖 horizontal and vertical deflections are related to each other.

∆2𝑖−1 sin(𝛼) − ∆2𝑖 cos(𝛼) = 0

Horizontal deflection Vertical deflection

of node 𝑖 of node 𝑖

• Details of how to handle these cases can be found in FEM textbooks.

METU – Dept. of Mechanical Engineering – ME 413 Int. to Finite Element Analysis – Lecture Notes of Dr. Sert 4-27
 Beam Bending
• Beams are long, slender structural members, generally subjected to transverse loading
that produces significant bending effects.
• Axial deformation or twisting is not considerable for beams.

𝑦 𝐹
𝑞(𝑥)

𝑀
𝑥

• 𝑞(𝑥) is the distributed transverse loading.

• 𝐹 is a point transverse load and 𝑀 is a point bending moment.
• Transverse deflection 𝑣(𝑥) is in the 𝑦 direction.

METU – Dept. of Mechanical Engineering – ME 413 Int. to Finite Element Analysis – Lecture Notes of Dr. Sert 4-28
 Euler-Bernoulli Beam Theory
• The assumption behind the Euler-Bernoulli beam theory is that plane cross sections
perpendicular to the longitudinal axis of the beam before bending, remain
perpendicular to the longitudional axis after bending.

Before bending : 𝑥

After bending :

𝑑2 𝑑2 𝑣
• Governing DE is 2
𝐸𝐼 2 = 𝑞 𝑥 , 0<𝑥<𝐿
𝑑𝑥 𝑑𝑥

• 𝑣(𝑥) : Unknown transverse deflection

• 𝑞(𝑥) : Known distributed transverse load
• 𝐸𝐼 : Known flexural rigidity of the beam, i.e. product of modulus of elasticity and the
second moment of inertia.

METU – Dept. of Mechanical Engineering – ME 413 Int. to Finite Element Analysis – Lecture Notes of Dr. Sert 4-29
 Euler-Bernoulli Beam Theory
𝑑2 𝑑2 𝑣
2
𝐸𝐼 2 = 𝑞 𝑥 , 0<𝑥<𝐿
𝑑𝑥 𝑑𝑥
• This is a 4th order DE.
• IBP should be applied two times to get the weak form.
• First IBP gives
𝑑𝑤 𝑑 𝑑2 𝑣 𝑑 𝑑2 𝑣 𝑑 𝑑2 𝑣
− 𝐸𝐼 2 𝑑𝑥 = 𝑤𝑞 𝑑𝑥 + −𝑤 𝐸𝐼 2 + 𝑤 𝐸𝐼 2
Ω𝑒 𝑑𝑥 𝑑𝑥 𝑑𝑥 Ω𝑒 𝑑𝑥 𝑑𝑥 𝑥𝑒
𝑑𝑥 𝑑𝑥 𝑥1𝑒
2

• Second IBP gives

𝑑2 𝑤 𝑑2 𝑣
2
𝐸𝐼 2 𝑑𝑥 = 𝑤𝑞 𝑑𝑥 +
Ω 𝑒 𝑑𝑥 𝑑𝑥 Ω𝑒

𝑑 𝑑2 𝑣 𝑑 𝑑2 𝑣 𝑑𝑤 𝑑 2 𝑣 𝑑𝑤 𝑑 2 𝑣
+ −𝑤 𝐸𝐼 2 + 𝑤 𝐸𝐼 2 + 𝐸𝐼 2 + − 𝐸𝐼 2
𝑑𝑥 𝑑𝑥 𝑥𝑒
𝑑𝑥 𝑑𝑥 𝑥𝑒
𝑑𝑥 𝑑𝑥 𝑥𝑒
𝑑𝑥 𝑑𝑥 𝑥1𝑒
2 1 2

METU – Dept. of Mechanical Engineering – ME 413 Int. to Finite Element Analysis – Lecture Notes of Dr. Sert 4-30
 Euler-Bernoulli Beam Theory (cont’d)
• There are two PVs
Transverse deflection : 𝑣
𝑑𝑣
Slope :
𝑑𝑥

• There are two SVs

𝑑 𝑑2 𝑣
Shear force : 𝐸𝐼 2
𝑑𝑥 𝑑𝑥
𝑑2 𝑣
Bending moment : 𝐸𝐼 2
𝑑𝑥

• Sign conventions are

• Deflection in +𝑦 direction (upward) is positive.
• CCW rotation of the beam corresponds to positive slope.
• Shear force in +𝑦 direction (upward) is positive.
• CCW moment (in +𝑧 direction) is positive.

METU – Dept. of Mechanical Engineering – ME 413 Int. to Finite Element Analysis – Lecture Notes of Dr. Sert 4-31
 Two Node Beam Element
• 2 node beam element has 4 PVs and 4 SVs

Δ𝑒1 Δ𝑒1 Transverse deflection at node 1

Δ𝑒3 𝑒
Δ 2 Slope at node 1
Δ𝑒2 Δ𝑒4 Δ𝑒 =
Δ𝑒3 Transverse deflection at node 2
e
1 2 Δ𝑒4 Slope at node 2

𝑄1𝑒 𝑄3𝑒 𝑄1𝑒 Shear force at node 1

𝑄2𝑒 Bending moment at node 1
𝑄2𝑒 𝑄4𝑒
𝑒
𝑄 =
e 𝑄3𝑒 Shear force at node 2
1 2 𝑄4𝑒 Bending moment at node 2

METU – Dept. of Mechanical Engineering – ME 413 Int. to Finite Element Analysis – Lecture Notes of Dr. Sert 4-32
 Hermite Type Shape Functions
• Weak form of the problem contains 2nd derivative of the transverse deflection.
• Not only the transverse deflection, but also its first derivative, i.e. slope should be
continuous, because slope is a PV too.
• Over each element FE solution is
4

𝑣𝑒 = 𝑆𝑗 Δ𝑗𝑒
𝑗=1

• A 𝐶 0 continuous solution for Δ𝑒 is NOT enough. It should be at least 𝐶 1 continuous.

• Lagrange type shape functions used previously are not suitable.

• Hermite type shape functions should be used.

METU – Dept. of Mechanical Engineering – ME 413 Int. to Finite Element Analysis – Lecture Notes of Dr. Sert 4-33
 Hermite Type Shape Functions (cont’d)
• Over each beam element there are 4 unknowns.
• Continuity of two variables (𝑣 and 𝑑𝑣/𝑑𝑥) at two ends of an element results in 4
conditions to be satisfied.
• To satisfy these 4 conditions at least a cubic polynomial is necessary for 𝑣 𝑒 .

𝑣 𝑒 = 𝐴 + 𝐵𝜉 + 𝐶𝜉 2 + 𝐷𝜉 3

• Four continuity restrictions are

At 𝜉 = −1 ∶ 𝑣 𝑒 = Δ𝑒1
𝑑𝑣 𝑒 𝑑𝑣 𝑒 1
At 𝜉 = −1 ∶ = Δ𝑒2 → = Δ𝑒
2
𝑑𝑥 𝑑𝜉 𝐽𝑒
At 𝜉 = +1 ∶ 𝑣 𝑒 = Δ𝑒3
𝑑𝑣 𝑒 𝑑𝑣 𝑒 1
At 𝜉 = +1 ∶ = Δ𝑒4 → = Δ𝑒
4
𝑑𝑥 𝑑𝜉 𝐽𝑒

METU – Dept. of Mechanical Engineering – ME 413 Int. to Finite Element Analysis – Lecture Notes of Dr. Sert 4-34
 Hermite Type Shape Functions (cont’d)
• Using 𝐽𝑒 these 4 conditions become
Δ𝑒1 = 𝐴 − 𝐵 + 𝐶 − 𝐷
2
Δ𝑒2 = 𝐵 − 2𝐶 + 3𝐷
ℎ𝑒
Δ𝑒3 = 𝐴 + 𝐵 + 𝐶 + 𝐷
2
Δ𝑒4 = 𝐵 + 2𝐶 + 3𝐷 𝑒
ℎ

• Solve for 𝐴, 𝐵, 𝐶 and 𝐷 in terms of Δ𝑒1 , Δ𝑒2 , Δ𝑒3 and Δ𝑒4 .

• Substitute them into the following equation
4

𝑆𝑗 Δ𝑗𝑒 = 𝐴 + 𝐵𝜉 + 𝐶𝜉 2 + 𝐷𝜉 3
𝑗=1

𝑣𝑒 𝑣𝑒

• And identify the 4 Hermite type cubic shape functions.

METU – Dept. of Mechanical Engineering – ME 413 Int. to Finite Element Analysis – Lecture Notes of Dr. Sert 4-35
 Hermite Type Shape Functions (cont’d)
• Hermite type cubic shape functions are
1 3
𝑆1 = 𝜉 − 3𝜉 + 2
4
ℎ𝑒 3
𝑆2 = 𝜉 − 𝜉2 − 𝜉 + 1
8
1
𝑆3 = −𝜉 3 + 3𝜉 + 2
4
ℎ𝑒 3
𝑆4 = 𝜉 + 𝜉2 − 𝜉 − 1
8

𝑆1 𝑆3
𝑆2 /ℎ𝑒
0.15
0.8
0.1

0.6 0.05

0
0.4 -0.05

-0.1
0.2
-0.15
𝑆4 /ℎ𝑒
0 -0.2
-1 -0.5 0 0.5 1 -1 -0.5 0 0.5 1

METU – Dept. of Mechanical Engineering – ME 413 Int. to Finite Element Analysis – Lecture Notes of Dr. Sert 4-36
 Elemental System for a Beam
• From Slide 4-30, elemental stiffness matrix and force vector are
2 2
𝑒 𝑑12
𝑆𝑖 1 𝑑 2 𝑆𝑗 1
𝐾𝑖𝑗 = 𝐸𝐼 2 𝑒 𝐽𝑒 𝑑𝜉
−1 𝑑𝜉 𝐽 𝑑𝜉 2 𝐽𝑒
1
𝐹𝑖𝑒 = 𝑞 𝑆𝑖 𝐽𝑒 𝑑𝜉 𝑞 is the part of the
−1
distributed transverse
load, simplified as
• Evaluating these using Hermite type shape functions uniform over element e

6 3ℎ𝑒 −6 3ℎ𝑒 Δ𝑒1 6 𝑄1𝑒

𝑒
2𝐸𝐼 2 ℎ𝑒 2 −3ℎ𝑒 ℎ𝑒 2 Δ𝑒2 𝑞ℎ ℎ𝑒 𝑄2𝑒
= +
(ℎ𝑒 )3 6 −3ℎ𝑒 Δ𝑒3 12 6 𝑄3𝑒
sym 2 ℎ𝑒 2 Δ𝑒4 −ℎ𝑒 𝑄4𝑒

𝐾𝑒 𝐹𝑒

METU – Dept. of Mechanical Engineering – ME 413 Int. to Finite Element Analysis – Lecture Notes of Dr. Sert 4-37
 Example 4-2
Example 4.2 : For the following clamped beam with 𝐸𝐼 = 4 × 106 Nm, use two
e.g.
equal length elements to determine
• the transverse deflection of the tip
• the reaction force at the middle support.

𝑞 = 400 N/m

5m 5m

METU – Dept. of Mechanical Engineering – ME 413 Int. to Finite Element Analysis – Lecture Notes of Dr. Sert 4-38
 Example 4-2 (cont’d)
• 𝐸𝐼, ℎ𝑒 and 𝑞 are the same for both elements.
• 𝐾 𝑒 and 𝐹 𝑒 will be the same for both elements.
• There are four unknowns for each element.

Δ11 Δ13 Δ21 Δ23

Δ12 Δ14 Δ22 Δ24
1 e=1 2 1 e=2 2

• Overall there are 3 nodes and 6 unknowns

Δ1 Δ3 Δ5
Δ2 Δ4 Δ6
e=1 e=2
1 2 3

METU – Dept. of Mechanical Engineering – ME 413 Int. to Finite Element Analysis – Lecture Notes of Dr. Sert 4-39
 Example 4-2 (cont’d)
• Local-to-global mapping of the unknowns are as follows

Δ11 Δ1
1
Δ
Unknowns of e=1 : Δ1 = 12 Δ2
Δ3
Δ14 Δ3
=Δ Global unknowns
Δ21 Δ4

2 Δ22 Δ5
Unknowns of e=2 : Δ =
Δ23
Δ24 Δ6

1 2 3 4
𝐿𝑡𝑜𝐺 =
3 4 5 6

METU – Dept. of Mechanical Engineering – ME 413 Int. to Finite Element Analysis – Lecture Notes of Dr. Sert 4-40
 Example 4-2 (cont’d)
• Assembled system is

6 3ℎ𝑒 −6 3ℎ𝑒 0 0 Δ1
3ℎ𝑒 2(ℎ𝑒 )2 −3ℎ𝑒 (ℎ𝑒 )2 0 0 Δ2
2𝐸𝐼 −6 −3ℎ𝑒 6+6 −3ℎ𝑒 + 3ℎ𝑒 −6 3ℎ𝑒 Δ3
(ℎ𝑒 )3 3ℎ𝑒 (ℎ𝑒 )2 −3ℎ𝑒 + 3ℎ𝑒 2(ℎ𝑒 )2 +2(ℎ𝑒 )2 −3ℎ𝑒 (ℎ𝑒 )2 Δ4
0 0 −6 −3ℎ𝑒 6 −3ℎ𝑒 Δ5
0 0 3ℎ𝑒 (ℎ𝑒 )2 −3ℎ𝑒 2(ℎ𝑒 )2 Δ6

𝑄11
6
ℎ𝑒 𝑄21
𝑞ℎ𝑒 6+6 𝑄31 + 𝑄12
= +
12 −ℎ𝑒 + ℎ𝑒 𝑄41 + 𝑄22
6 𝑄32
−ℎ𝑒
𝑄42

METU – Dept. of Mechanical Engineering – ME 413 Int. to Finite Element Analysis – Lecture Notes of Dr. Sert 4-41
 Example 4-2 (cont’d)
• Boundary conditions need to be applied.

• Known deflections and slopes are EBCs.

• At the clamped end, transverse deflection and slope are zero.

Δ1 = 0, Δ2 = 0
• At the middle support transverse deflection is zero.
Δ3 = 0

• Known shear forces and moments are NBCs.

• Middle support can not carry any bending moment

𝑄4 = 𝑄41 + 𝑄22 = 0
• Free end can not cary and shear force or bending moment

𝑄5 = 0 , 𝑄6 = 0
METU – Dept. of Mechanical Engineering – ME 413 Int. to Finite Element Analysis – Lecture Notes of Dr. Sert 4-42
 Example 4-2 (cont’d)
• Only 3 PVs (Δ4 , Δ5 , Δ6 ) are actually unknown.
• Reduction can be applied to the original 6x6 system.

𝑒 2 2 Δ4
2𝐸𝐼 4ℎ −3ℎ𝑒 ℎ𝑒 𝑞ℎ𝑒 0 0
3 −3ℎ 𝑒 6 −3ℎ𝑒 Δ5 = 6 + 0
ℎ𝑒 12
ℎ𝑒
2
−3ℎ𝑒 2ℎ𝑒
2 Δ6 −ℎ𝑒 0

𝐹 is not changed
• Using 𝐸𝐼 = 4 × 106 Nm , ℎ𝑒 = 5 m , 𝑞 = −400 N/m because ∆1 = ∆2 = ∆3 = 0

𝑞 is in – 𝑦 direction

• Unknown PVs are calculated as

∆4 = −0.00130 rad Slope at the middle support. Beam rotation is CW.

∆5 = −0.01432 m Transverse deflection at the tip. It is downward.
∆6 = −0.00339 rad Slope at the tip. Beam rotation is CW.

METU – Dept. of Mechanical Engineering – ME 413 Int. to Finite Element Analysis – Lecture Notes of Dr. Sert 4-43
 Example 4-2 (cont’d)
• To find the unknown SVs we can use the calculated PVs in the original 6x6 system.
• Unknown SVs can be calculated as

𝑄1 = −250 N Force applied by the wall at the clamped end.

𝑄2 = −1250 Nm Moment applied by the wall at the clamped end.
𝑄3 = 4250 N Force acting by the middle support.

400 N/m

1250 Nm
250 N
4250 N

METU – Dept. of Mechanical Engineering – ME 413 Int. to Finite Element Analysis – Lecture Notes of Dr. Sert 4-44
 Example 4-3
e.g. Example 4.3 : Solve the same problem, but this time remove the distributed load
and put a point load at the tip

4000 N

5m 5m

• One detail you need to pay attention is that this time

𝑄5 = −4000

METU – Dept. of Mechanical Engineering – ME 413 Int. to Finite Element Analysis – Lecture Notes of Dr. Sert 4-45
 Planar Frames
• Frames look like trusses, but the connections are rigid, i.e. welded or riveted.
• Each member can carry axial force, shear force and bending moment.

𝐹1 𝐹2 𝐹1 𝐹2

• The above bicycle frame has 7 members.

• Each member can be modeled as a single element or multiple elements.
• It is possible to think of a frame element as the superposition of truss and beam
elements.

METU – Dept. of Mechanical Engineering – ME 413 Int. to Finite Element Analysis – Lecture Notes of Dr. Sert 4-46
 Arbitrarily Oriented Beam Element
• Frame elements are based on arbitrarily oriented beam elements.
• Similar to a truss element, it is possible to study the beam element using either the
local 𝑥 𝑒 , 𝑦 𝑒 coordinates or the global 𝑥, 𝑦 coordinates.

𝑦𝑒 Δ𝑒3
Δ𝑒5 Δ𝑒6
Δ𝑒4 𝑥𝑒 Δ𝑒4
Δ𝑒1 e 2
Δ𝑒2 Δ𝑒3
e 2 Δ𝑒1
Δ𝑒2 𝑦
𝜃𝑒 1
1 𝑥

Unknowns in local coordinates Unknowns in global coordinates

Node 1 : Δ𝑒1 , Δ𝑒2 Node 1 : Δ𝑒1 , Δ𝑒2 , Δ𝑒3
Node 2 : Δ𝑒3 , Δ𝑒3 Node 2 : Δ𝑒4 , Δ𝑒5 , Δ𝑒6

METU – Dept. of Mechanical Engineering – ME 413 Int. to Finite Element Analysis – Lecture Notes of Dr. Sert 4-47
 Transformation Matrix of a Beam Element
• Relation between local and global unknowns are

Δ𝑒1 = − sin 𝜃 𝑒 Δ𝑒1 + cos(𝜃 𝑒 ) Δ𝑒2

Δ𝑒2 = Δ𝑒3
Δ𝑒3 = − sin 𝜃 𝑒 Δ𝑒4 + cos(𝜃 𝑒 ) Δ𝑒5
Δ𝑒4 = Δ𝑒6

• These relations can be expressed using the following transformation matrix.

Δ𝑒1
Δ𝑒1 −𝛽 𝛼 0 0 0 0 Δ2
𝑒
𝑒
Δ𝑒2 0 0 1 0 0 0 Δ3
=
Δ𝑒3 0 0 0 −𝛽 𝛼 0 Δ𝑒4
Δ𝑒4 0 0 0 0 0 1 Δ𝑒5
Δ𝑒6 Transformation
matrix of the beam
element
where 𝛼 = cos(𝜃 𝑒 ) and 𝛽 = sin(𝜃 𝑒 )
METU – Dept. of Mechanical Engineering – ME 413 Int. to Finite Element Analysis – Lecture Notes of Dr. Sert 4-48
 Frame Element
• We now have transformation matrices for arbitrarily oriented beam and truss
elements.
• Frame elements carry axial force, shear force and bending moment.
• They can be obtained by the superposition of beam and truss elements.
• Frame element has 3 unknowns at each node.

Frame element in Frame element in

local coordinates global coordinates
Δ𝑒5 𝑦 Δ𝑒5
𝑦𝑒
Δ𝑒4 𝑥𝑒
Δ𝑒2 Δ𝑒6 Δ𝑒6 Δ𝑒4
Δ𝑒2
2 2
Δ𝑒1 e e
𝜃𝑒
Δ𝑒3 Δ𝑒3 𝑥
1 1 Δ𝑒1

METU – Dept. of Mechanical Engineering – ME 413 Int. to Finite Element Analysis – Lecture Notes of Dr. Sert 4-49
 Frame Element (cont’d)
• Elemental system of the frame element in local unknowns is obtained by the proper
combination of those of truss and beam elements
𝑞 is the constant
𝑒 𝑒 𝑒
𝐾 Δ = 𝐹 distributed
transverse load
These are coming from 𝐸𝐴 −𝐸𝐴
0 0 0 0
𝐾 𝑒 of the truss ℎ𝑒 ℎ𝑒 0
element (Slide 4-9) 12𝐸𝐼 6𝐸𝐼 −12𝐸𝐼 6𝐸𝐼 𝑞ℎ𝑒
0 Δ𝑒1
ℎ𝑒 3 ℎ𝑒 2 ℎ𝑒 3 ℎ𝑒 2 2
4𝐸𝐼 −6𝐸𝐼 2𝐸𝐼 Δ𝑒2 𝑞 ℎ𝑒 2
0
ℎ𝑒 ℎ𝑒 2 ℎ𝑒 Δ𝑒3 12
The rest is coming from =
𝐸𝐴 Δ𝑒4 0
𝐾 𝑒 of the beam 0 0 𝑞ℎ𝑒
ℎ𝑒 Δ𝑒5
element (Slide 4-37) 12𝐸𝐼 −6𝐸𝐼 2
Δ𝑒6
Symmetric ℎ𝑒 3 ℎ𝑒 2 −𝑞 ℎ𝑒 2
4𝐸𝐼 12
ℎ𝑒
{𝐹 𝑒 }
𝐾𝑒

METU – Dept. of Mechanical Engineering – ME 413 Int. to Finite Element Analysis – Lecture Notes of Dr. Sert 4-50
 Frame Element (cont’d)
• Similarly transformation matrix of the frame element is obtained by the proper
combination of those of truss and beam elements
These 2 eqns are coming
Δ𝑒1 𝛼 𝛽 0 0 0 0 Δ𝑒1 from the transformation
Δ𝑒2 −𝛽 𝛼 0 0 0 0 Δ𝑒2 matrix of the truss element
(Slide 4-8)
Δ𝑒3 0 0 1 0 0 0 Δ𝑒3
=
Δ𝑒4 0 0 0 𝛼 𝛽 0 Δ𝑒4 Remaining 4 eqns are
Δ𝑒5 0 0 0 −𝛽 𝛼 0 Δ𝑒5 coming from the
Δ𝑒6 0 0 0 0 0 1 Δ𝑒6 transformation matrix of the
beam element (Slide 4-48)
𝑇𝑒

• Elemental system in global unknowns is obtained as

𝐾 𝑒 Δ𝑒 = {𝐹 𝑒 }

𝐾𝑒 = 𝑇𝑒 𝑇
𝐾𝑒 𝑇𝑒 {𝐹 𝑒 } = 𝑇 𝑒 𝑇 {𝐹 𝑒 }

METU – Dept. of Mechanical Engineering – ME 413 Int. to Finite Element Analysis – Lecture Notes of Dr. Sert 4-51
 Example 4-3
e.g. Example 4.3 : Using three elements, determine the deflections and rotations at
the joints of the following frame. Draw bending moment and shear force diagrams
for all elements. Calculate the reactions at the supports.

For all members 𝐸 = 200 GPa , 𝐼 = 2.7 × 10−6 m4 , 𝐴 = 4.4 × 10−4 m2

𝑞 = 7.5 kN/m

15 kN

2.5 m

3.5 m

METU – Dept. of Mechanical Engineering – ME 413 Int. to Finite Element Analysis – Lecture Notes of Dr. Sert 4-52
 Example 4-3 (cont’d)
• Element and global/local node numbering are shown below.

1 e=3 2

2 2 3 2
e=3
e=1 e=1 e=2 e=2

1 1 4 1

• Orientation of the elements are

𝜃1 = 𝜋/2 , 𝜃 2 = 𝜋/2 , 𝜃 3 = 0
• Element lengths are
ℎ1 = 2.5 , ℎ2 = 2.5 , ℎ3 = 3.5
• 𝐸, 𝐼 and 𝐴 are the same for all elements.

METU – Dept. of Mechanical Engineering – ME 413 Int. to Finite Element Analysis – Lecture Notes of Dr. Sert 4-53
 Example 4-3 (cont’d)
• Using Slide 4-50 calculate [𝐾 𝑒 ] and {𝐹 𝑒 } for each element

• For e=1: 𝐾 1 = 106

𝐹1 = 0 0 0 0 0 0 𝑇 (all zero because 𝑞 = 0 for e=1)

• For e=2: 𝐾 2 = 𝐾1 (because 𝐸, 𝐴, 𝐼 and ℎ are the same for both elements)

𝐹 2 = 𝐹1 (because 𝑞 = 0 for e=2, too)

METU – Dept. of Mechanical Engineering – ME 413 Int. to Finite Element Analysis – Lecture Notes of Dr. Sert 4-54
 Example 4-3 (cont’d)

• For e=3: 𝐾 3 = 106

𝐹 3 = 104

METU – Dept. of Mechanical Engineering – ME 413 Int. to Finite Element Analysis – Lecture Notes of Dr. Sert 4-55
 Example 4-3 (cont’d)
• Now the transformation matrices for each element need to be calculated using Slide
4-51.
0 1 0 0 0 0
−1 0 0 0 0 0
0 0 1 0 0 0
• For e=1: 𝑇1 =
0 0 0 0 1 0
0 0 0 −1 0 0
0 0 0 0 0 1

• For e=2: 𝑇 2 = 𝑇1 (because 𝜃 2 = 𝜃1 )

1 0 0 0 0 0
0 1 0 0 0 0
0 0 1 0 0 0
• For e=3: 𝑇3 = (This is the unity matrix because 𝑥 3 = 𝑥)
0 0 0 1 0 0
0 0 0 0 1 0
0 0 0 0 0 1

METU – Dept. of Mechanical Engineering – ME 413 Int. to Finite Element Analysis – Lecture Notes of Dr. Sert 4-56
 Example 4-3 (cont’d)
• [𝐾 𝑒 ] and {𝐹 𝑒 } of each element can be calculated using

𝐾𝑒 = 𝑇𝑒 𝑇 𝐾𝑒 𝑇𝑒 and {𝐹 𝑒 } = 𝑇 𝑒 𝑇 {𝐹 𝑒 }

• For e=1: 𝐾 1 = 104

𝐹1 = 0 0 0 0 0 0

• For e=2: 𝐾 2 = 𝐾1 , 𝐹 2 = 𝐹1

• For e=3: 𝐾3 = 𝐾3 , {𝐹 3 } = {𝐹 3 }

METU – Dept. of Mechanical Engineering – ME 413 Int. to Finite Element Analysis – Lecture Notes of Dr. Sert 4-57
 Example 4-3 (cont’d)
• Now we have three 6x6 elemental systems.
• Let’s assemble them into the 12x12 global system.
• LtoG mapping is as follows
1 2 3 4 5 6
𝐿𝑡𝑜𝐺 = 10 11 12 7 8 9
4 5 6 7 8 9

1 e=3 2
Δ31 , Δ32 , Δ33 Δ34 , Δ35 , Δ36

Δ14 , Δ15 , Δ16 Δ4 , Δ5 , Δ6 Δ7 , Δ8 , Δ9 Δ24 , Δ25 , Δ26

2 2 e=3 3 2

e=1 e=1 e=2 e=2

1 1 4 1
Δ1 , Δ2 , Δ3 Δ10 , Δ11 , Δ12 Δ21 , Δ22 , Δ23
Δ11 , Δ12 , Δ13

METU – Dept. of Mechanical Engineering – ME 413 Int. to Finite Element Analysis – Lecture Notes of Dr. Sert 4-58
 Example 4-3 (cont’d)
• Assembled global system is 12x12.
• At nodes 1 and 4 all three unknowns are known and they are zero.
∆1 = ∆2 = ∆3 = 0 , ∆10 = ∆11 = ∆12 =0

Δ4
Δ5
6 Δ6 𝑄4 is the given
10
Δ7 horizontal
Δ8 point load at
node 2
Δ9

15000
𝑄5 and 𝑄6 are zero because there is no 0
vertical point load or point moment at node 2 0
= 104 +
0
𝑄7 , 𝑄8 and 𝑄9 are zero because there is no
0
horizontal or vertical point load or point 0
moment at node 3.
METU – Dept. of Mechanical Engineering – ME 413 Int. to Finite Element Analysis – Lecture Notes of Dr. Sert 4-59
 Example 4-3 (cont’d)
• Solving for the unknown deflecions we get

Δ4
Δ5 Horizontal deflection, vertical
Δ6 deflection and rotation of node 2
=
Δ7 Horizontal deflection, vertical
Δ8 deflection and rotation of node 3
Δ9

• Both nodes 2 and 3 move in +𝑥 and −𝑦 directions. Also they rotate CW.

• Now the forces and moments at the supports (𝑄1 , 𝑄2 , 𝑄3 , 𝑄10 , 𝑄11 , 𝑄12 ) can be calculated.

• Also axial stress, shear stress and bending stress over each element can be determined.

• Question : Will the solution improve by using more elements?

METU – Dept. of Mechanical Engineering – ME 413 Int. to Finite Element Analysis – Lecture Notes of Dr. Sert 4-60