Date post: | 09-Dec-2015 |
Category: |
Documents |
Upload: | ervikram25 |
View: | 52 times |
Download: | 14 times |
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 1
A First Course in Finite Elements
Introduction
The finite element method has become a powerful tool for the numerical so-lution of a wide range of engineering problems. Applications range from deforma-tion and stress analysis of automotive, aircraft, building, and bridge structures to field analysis of heat flux, fluid flow, magnetic flux, seepage, and other flow prob-lems.
With the advances in computer technology and CAD systems, complex prob-lems can be modeled with relative ease. Several alternative configurations can be tried out on a computer before the first prototype is built. All of this suggests that we need to keep pace with these developments by understanding the basic theory, modeling techniques, and computational aspects of the finite element method.
In this method of analysis, a complex region defining a continuum is discre-tized into simple geometric shapes called finite elements. The material proper-ties and the governing relationships are considered over these elements and ex-pressed in terms of unknown values at element corners. An assembly process, duly considering the loading and constraints, results in a set of equations. Solu-tion of these equations gives us the approximate behavior of the continuum.
Historical Background
Basic ideas of the finite element method originated from advances in aircraft structural analysis. In 1941, Hrenikoff presented a solution of elasticity problems using the “frame work method.” Courant’s paper, which used piecewise polyno-mial interpolation over triangular subregions to model torsion problems, appeared in 1943. Turner et al. derived stiffness matrices for truss, beam, and other ele-ments and presented their findings in 1956. The term finite element was first coined and used by Clough in 1960.
In the early 1960s, engineers used the method for approximate solution of problems in stress analysis, fluid flow, heat transfer, and other areas. A book by
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 2
Argyris in 1955 on energy theorems and matrix methods laid a foundation for fur-ther developments in finite element studies. The first book on finite elements by Zienkiewicz and Chung was published in 1967. In the late 1960s and early 1970s, finite element analysis was applied to nonlinear problems and large de-formations. Oden’s book on nonlinear continua appeared in 1972.
Mathematical foundations were laid in the 1970s. New element development, convergence studies, and other related areas fall in this category.
Today, developments in mainframe computers and availability of powerful mi-crocomputers have brought this method within reach of students and engineers working in small industries.
Stress and Equilibrium
Consider a three-dimensional body of volume V having a surface S:
A point in the body is located by x, y, and z coordinates. On part of the
boundary, a distributed force per unit area T, also called traction is applied. Un-
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 3
der the force, the body deforms. The deformation of a point x (x, y, z) is given by the three components of its displacement vector:
[ ]Twvu ,,=u The distributed force per unit volume is given by force vector:
[ ]Tzyx FFF ,,=F
The traction T is given by its components at points along the surface:
[ ]Tzyx TTT ,,=T
A load Pi acting at a point i is given by its three components:
[ ] Tizyx PPP ,,=iP
The stresses acting on the element volume dV are:
When the volume dV shrinks to a point, the stresses may be represented by
placing its components in a (3 x 3) symmetric matrix. Stress can be represented by the six independent components:
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 4
[ ]Τ= yzxzxyzyx τττσσσσ
where σx, σy, and σz are called normal stress and τxy, τxz, and τyz are called shear stress. Consider the equilibrium of the element volume dV. Forces are developed by multiplying the stresses by the corresponding areas. Writing the equations of equilibrium, recognizing the dV = dx dy dz:
0=+∂
∂+
∂
∂+
∂∂
=∑ xxzxyx
x fzyx
F ττσ
0=+∂
∂+
∂∂
+∂
∂=∑ y
yzxxyy f
zyxF
τστ
0=+∂
∂+
∂
∂+
∂∂
=∑ zzyzxz
z fzyx
F σττ
Boundary Conditions
There are displacement boundary conditions and surface loading conditions.
u = c on Su
where c is a given displacement
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 5
Consider the equilibrium if an elemental tetrahedron ABCD:
where DA, DB, and DC are parallel to the x, y, and z axes, respectively, and the area ABC, denoted by dA, lies on the surface. If the unit vector normal to the surface dA is given as:
[ ]Tzyx nnn ,,=n
then the areas:
dAnx=BDC dAny=ADC dAnz=ADB
Consider equilibrium in each direction:
xzxzyxyxx Tnnn =++ ττσ
yzyzyyxxy Tnnn =++ τστ
yzzyxyxxz Tnnn =++ σττ
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 6
These conditions must be satisfied on the boundary, ST, where the tractions are applied. Point loads must be treated as loads distributed over small but finite areas.
Strain-Displacement Relations
The strains in vector form that corresponds to the stress are:
where εx, εy, and εz are normal strains and γxy, γxz, and γxz, are the engineering shears strains.
We can approximate the shear strains by considering a small deformation of the dx-dy face of the unit volume dV:
, , , , ,T
u v w u v u w u vx y z y x z x z y
ε ⎡ ⎤∂ ∂ ∂ ∂ ∂ ∂ ∂ ∂ ∂= + + +⎢ ⎥∂ ∂ ∂ ∂ ∂ ∂ ∂ ∂ ∂⎣ ⎦
[ ]Tyzxzxyzyx γγγεεεε ,,,,,=
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 7
Stress-Strain Relations
For linear elastic materials, the stress-strain relations come from the general-ized Hooke’s Law. For isotropic materials, the two material properties are Young’s modulus (or the modulus of elasticity) E and Poisson’s ratio ν. Consider-ing an elemental volume inside the body, Hooke’s Law gives:
EEEzyx
xσν
σνσε −−=
EEEzyx
yσν
σσνε −+−=
EEEzyx
zσσ
νσνε +−−=
where the shear modulus (or modulus of rigidity), G, is given by:
)1(2 ν+=
EG
From Hooke’s Law, strain and stress are related by:
( )zyxzyx Eσσσνεεε ++
−=++
)21(
By substituting the above relationships into Hooke’s Law we get an inverse rela-tionship:
εσ D=
Gxz
xzτγ =
Gxy
xyτ
γ =Gyz
yzτ
γ =
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 8
where D is a symmetric (6 x 6) material matrix:
⎥⎥⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢⎢⎢
⎣
⎡
−−
−−
−−
−+=
νν
νν
νν
νν
ν
νν
ν
νν
5.000000
05.00000
00
5.0000
000
1
000
1
000
1
)21)(1(ED
Special Cases
One-Dimension. In one-dimension, the normal stress s along x and the cor-responding normal strain e. The stress-strain relationship is simply:
εσ E=
Two-Dimension. A thin planar body subjected to in-plane loading on its edge surface is said to be plane stress. For example, consider the thin plates shown below.
Here the stresses σz, τxz, and τyz are assumed to be zero. Generally, members that are thin (those with a small z dimension compared to the in-plane x and y dimensions) and shoes loads only in the x-y plane can be consider under plane stress.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 9
For plane stress, Hooke’s Law reduces to:
EEyx
xσ
νσε −=
EEyx
yσσνε +−=
( )yxz Eσσνε +−=
xyxy Eτνγ )1(2 +
=
The inverse relationship σ = Dε reduces to:
⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧
⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢
⎣
⎡
−−=
⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧
xy
y
x
xy
y
x E
γεε
ν
νν
ντσσ
2)1(
00
01
0
1
1 2
If the body of uniform cross section is subjected to a transverse loading along
its length, a small thickness in the loaded area can be approximated by plane strain. For example:
Here the strains εz, γxz, and γyz are assumed to be zero.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 10
The inverse relationship σ = Dε reduces to:
( )( ) ⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡
−−
−
−+=
⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧
xy
y
x
xy
y
x E
γεε
νν
νν
ν
νντσσ
21
00
01
0
1
211
Temperature Effects
If there is a temperature change ∆T(x, y, z) with respect to the original state, then an additional deformation can be estimated. For isotropic material, the tem-perature rise ∆T results in a uniform strain; this depends on the coefficient of lin-ear expansion of the material.
The temperature strain dose not cause any stresses when the body is free to deform. The temperature strain is represented as an initial strain:
[ ]TTTT 0,0,0,,,0 ∆∆∆= αααε
The stress-strain relationship becomes:
)( 0εεσ −= D
In plane stress, we get:
[ ]TTT 0,,0 ∆∆= ααε
In plane strain, we get:
[ ]TTT 0,,)1(0 ∆∆+= αανε
For plane stress and plane strain σ, ε, and D are defined by the corresponding equations given above.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 11
Potential Energy and Equilibrium; the Rayleigh- Ritz Method
In mechanics of solids, our problem is to determine the displacement u of the body, satisfying the equilibrium equations.
Note that stresses are related to strains, which, in turn, are related to dis-
placements. This leads to requiring the solution of set of second-order partial dif-ferential equations. Solution of these equations is generally referred to as an ex-act solution. Such exact solutions are available for simple geometries and load-ing conditions. For problems of complex geometries and general boundary and loading conditions, obtaining exact solutions is an almost impossible task. Ap-proximate solution methods usually employ potential energy or variational methods, which place less stringent conditions on the functions.
Potential Energy, Π
The total potential energy Π of an elastic body is defined as the sum of the to-tal strain energy, U, and the work potential, WP:
Π = Strain Energy + Work Potential (U) (WP)
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 12
For linear elastic materials, the strain energy per unit volume is:
εσ T
21
The total strain energy U is given as:
∫=V
T dVεσ21U
The potential work WP is given as:
∑∫∫ −−−= iT
iS
T
V
T dSdV PuTufuWP
The total potential energy for a general elastic body is:
∑∫∫∫ −−−=Π iT
iS
T
V
T
V
T dSdVdV PuTufuεσ21
This is a conservative system, where the work potential is independent of the path taken. In other words, if the system is displaced from a given configuration and brought back to its original state, the forces do zero work regardless of the path.
Kinematically admissible displacements are those that satisfy the single-valued nature of displacements (compatibility) and the boundary conditions.
Principle of Minimum Potential Energy – For conservative sys-tems, of all the kinematically admissible displacement fields, those corresponding to equilibrium extermize the total potential energy. If the extremum condition is a minimum, the equilibrium state is stable.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 13
Example
Consider a discrete connected system. The figure below shows a system of springs.
The total potential energy of the system is:
( ) 33112
442
332
222
1121 qFqFkkkk −−+++=Π δδδδ
where δ1, δ 2, δ 3, and δ 4 are the extensions of the four springs.
Therefore, total potential energy of the system is:
[ ] 33112
342
2332
222
211 )()(21 qFqFqkqqkqkqqk −−+−++−=Π
where q1, q2, and q3 are the displacements of nodes 1, 2, and 3, respectively.
22 q=δ
233 qq −=δ 34 q−=δ
211 qq −=δ
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 14
For equilibrium of this three degree-of-freedom system, we need to minimize Π with respect to the displacements q1, q2, and q3.
0 1, 2, 3i
iq
∂Π= =
∂
Therefore, the three equations are:
0)( 12111
=−−=∂
Π∂ Fqqkq
0)()( 233222112
=−−+−−=∂
Π∂ qqkqkqqkq
0)( 3342333
=−+−=∂
Π∂ Fqkqqkq
These equations can be written in matrix form as:
⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧=
⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡
+−
−++
−−
3
1
3
2
1
43
3
3
321
1
1
1
00
0 F
F
qqq
kkk
kkkk
kk
k
Alternately, we could write the equations of equilibrium for each node separately.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 15
111 Fk =δ
0331122 =−− δδδ kkk
34433 Fkk =− δδ
Notice the equations for the displacements were obtained in a routine manner using the potential energy approach, without any reference to free body dia-grams. This feature makes the potential energy approach attractive for large and complex problems.
Rayleigh-Ritz Method
For continua, the total potential energy, Π, can be used for finding an ap-proximate solution. The Rayleigh-Ritz method involves the construction of an as-sumed displacement field [u, v, w]:
ltoizyxau ii 1),,( == ∑ φ
mtoljzyxav jj 1),,( +== ∑ φ
lmnntomkzyxaw kk >>+== ∑ 1),,(φ
The functions φi are usually taken as polynomials. Displacements u, v, and w must be kinematically admissible (that is u, v, and w must satisfy boundary conditions). Introducing stress-strain and strain-displacement relationships gives:
),...,,( 21 naaaΠ=Π where n is the number of independent unknowns. The extremum with respect to ai, (i = 1 to n) gives a set of n equations:
niai
,,2,10 ==∂
Π∂
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 16
Example
Consider the linear elastic one-dimensional rod with a body force shown below:
The potential energy of this system is:
10
2
221 udx
dxduEA
L
−⎟⎠⎞
⎜⎝⎛=Π ∫
where u1 = u(x=1). Consider the polynomial function:
2321 xaxaau ++=
The kinematically admissible function u must satisfy the boundary conditions u = 0 at both (x = 0) and (x = 2). Therefore:
0420 321 =+= aaa
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 17
Hence:
32 2aa −=
23 1 3( 2 ) ( 1)u a x x u x u a= − + = = = −
)1(23 −= xadxdu
The potential energy of this system using the function u is:
)(2)1(421
3
2
0
223 adxxaEA +−=Π ∫
22 23 3
0
2 (1 2 ) 2EAa x x dx a= − + +∫
23 3
4 23
a a= +
Applying the Rayleigh-Ritz method gives:
0238
33
=+=∂
Π∂ aa
Solving for a3 gives:
( )23 1 30.75 0.75 ( ) 0.75 2a u a u x x x= − = − = = −
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 18
The stress in the bar is given by:
)1(5.1 xdxduE −==σ
Notice that an exact solution is obtained if a piecewise polynomial interpolation is used in the construction of u.
Galerkin’s Method
Galerkin’s method uses the set of governing equations in the development of an integral form. It is usually presented as one of the weighted residual methods. For our discussion, let us consider a general representation of a governing equa-tion on a region V:
PLu = For the one-dimensional rod considered in the pervious example, the govern-
ing equation is:
0=⎟⎠⎞
⎜⎝⎛
dxduEA
dxd
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 19
If we consider L as the following operator:
()dxdEA
dxd
operating on u, the exact solution needs to satisfy L at every point x. If we seek an approximate solution, u , it introduces an error e(x), called the residual:
PuLxe −= ˆ)(
Approximate methods revolve around setting the residual relative to a weight-ing function Wi, to zero.
ntoidVPLuWv
i 10)( ==−∫
The choice of the weighting function, Wi, leads to various approximation
methods. In the Galerkin method, the weighting functions, Wi, are chosen from the basis functions used for constructing u . Let u be represented by:
∑=
=n
iiiGQu
1
ˆ
where Gi, i = 1 to n, are basis functions (usually polynomials of x, y, z). Here we choose the weighting function to a linear combination of the basis functions Gi. Consider an arbitrary function φ given by:
∑=
=n
iiiG
1
φφ
where the coefficients φi are arbitrary, except for requiring that φ satisfy homoge-neous (zero) boundary conditions where u is prescribed.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 20
Galerkin’s Method in Elasticity
Consider the equations of equilibrium we developed earlier. Galerkin’s method requires:
yyyzyxy
Vxx
xzxyx fzyx
fzyx
φτστ
φττσ⎟⎟⎠
⎞⎜⎜⎝
⎛+
∂∂
+∂
∂+
∂∂
+⎢⎣
⎡⎟⎟⎠
⎞⎜⎜⎝
⎛+
∂∂
+∂
∂+
∂∂
∫
0=⎥⎦
⎤⎟⎟⎠
⎞⎜⎜⎝
⎛+
∂∂
+∂
∂+
∂∂
+ dVfzyx zz
zyzxz φσττ
where
, ,T
x y zφ φ φ φ⎡ ⎤= ⎣ ⎦
is an arbitrary displacement consistent with the displacements, u. Consider inte-gration by parts using the following formula:
xV V S
dV dV n dSx xα θθ α αθ∂ ∂
= − +∂ ∂∫ ∫ ∫
where α and θ are functions of (x, y, z). For multi-dimensional problems the above equation is referred to as Green-Gauss theorem or the divergence theo-rem.
Galerkin’s Method – Chose basis functions Gi. Determine the coefficients φi such that
where φi are arbitrary except for requiring that φ satisfy homogeneous boundary conditions. The solution of the resulting equations for Gi then yields the approximate solution u .
∑∫=
==−n
iii
V
GdVPuL1
where0)ˆ( φφφ
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 21
xzxzyxyxx Tnnn =++ ττσ
( ) xzxzyxyxx PdSnnn =++ ττσ
Using the Green-Gauss theorem on the equations of equilibrium yields:
∫∫ +−V
T
V
T dVfdV φφεσ )(
( )x x y xy z xz xS
n n n dSσ τ τ φ+ + +∫
( )x xy y y z yz yS
n n n dSτ σ τ φ+ + +∫
( )x xz y yz z z zS
n n n dSτ τ σ φ+ + +∫
where ε(φ) is the strain field corresponding to the arbitrary displacement field φ.
T
yxzxzyzyx
xyxzyzzyx ⎥⎦
⎤⎢⎣
⎡∂∂
+∂∂
∂∂
+∂∂
∂∂
+∂∂
∂∂
∂∂
∂∂
=φφφφφφφφφφε ,,,,,)(
On the boundary we have: At a point loads: These are the natural boundary conditions in the problem. Therefore the Galerkin “weak form” or “variational form” for three-dimensional stress analysis is:
yzyzyyxxy Tnnn =++ τστ
yzzyxyxxz Tnnn =++ σττ
( ) yzyzyyxxy PdSnnn =++ τστ
( ) yzzyxyxxz PdSnnn =++ σττ
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 22
0)( =−−− ∑∫∫∫ PTf T
S
T
V
T
V
T dSdVdV φφφφεσ
where φ is an arbitrary displacement consistent with the boundary conditions. For problems of linear elasticity, the above equation is precisely the principle of vir-tual work. The function φ is the admissible virtual displacement. The principle of virtual work may be stated as follows:
Example
Let consider the pervious problem and solve it by Galerkin’s approach. The equilibrium equation is:
2at00at00 ===== xuxudxduEA
dxd
Multiplying the differential equation above by φ and integrating by parts gives:
1 22 2
0 10 0
0d du du d du duEA dx EA dx EA EAdx dx dx dx dx dx
φφ φ φ⎛ ⎞ ⎛ ⎞ ⎛ ⎞= − + + =⎜ ⎟ ⎜ ⎟ ⎜ ⎟⎝ ⎠ ⎝ ⎠ ⎝ ⎠∫ ∫
where φ is zero at (x = 0) and (x = 2) and EA(du/dx) is the tension in the rod, which make a jump of magnitude of 2 at (x = 1). Therefore:
Principle of Virtual Work – A body is in equilibrium if the internal work equals the external virtual work for every kinematically admissible displace-ment field.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 23
02 1
2
0
=+− ∫ φφ dxdxd
dxduEA
If we use the same polynomial function for u and φ and If u1 and φ1 are the values at (x = 1), we get:
( ) ( ) 12
12 22 φφ xxuxxu −=−=
Substituting these and E = A = 1 in the above integral:
02)22(2
0
211 =⎥
⎦
⎤⎢⎣
⎡+−− ∫ dxxuφ
0238
11 =⎟⎠⎞
⎜⎝⎛ +− uφ
This is to be satisfied for every φ1. We get:
75.01 =u
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 24
The stress in the bar is given by:
)1(5.1 xdxduE −==σ
Problems:
1. Obtain the D matrix given below using the generalized Hook’s law relation-ships.
⎥⎥⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢⎢⎢
⎣
⎡
−−
−−
−−
−+=
νν
νν
νν
νν
ν
νν
ν
νν
5.000000
05.00000
00
5.0000
000
1
000
1
000
1
)21)(1(ED
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 25
2. Determine the displacement of nodes of the spring system shown below using both equilibrium and principle of minimum potential energy method.
3. Use the Rayleigh-Ritz method to find the displacement of the midpoint of the rod shown below (assume ρg = 1 and u = a1 + a2x + a3x2):
4. Use Galerkin’s method to find the displacement of the midpoint of the rod in Problem 3.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 26
Role of Computers in Finite Element Methods
Until the early 1950s, matrix methods and the associated finite element method were not readily adaptable for solving complicated problems because of the large number of algebraic equations that resulted. Hence, even though the finite element method was being used to describe complicated structures, the re-sulting large number of equations associated with the finite element method of structural analysis made the method extremely difficult and impractical to use.
With the advent of the computer, the solution of thousands of equations in a matter of minutes became possible. The development of the computer resulted in computational program development. Numerous special-purpose and general-purpose programs have been written to handle various complicated structural (and non-structural) problems. To use the computer, the analyst, having defined the finite element model, inputs the information into the computer. This formation may include the position of the element nodal coordinates, the manner in which elements are connected together, the material properties of the elements, the applied loads, boundary conditions, or constraints, and the kind of analysis to be performed. The computer then uses this information to generate and solve the equations necessary to carry out the analysis.
General Steps of the Finite Element Method
The following section presents the general steps for applying the finite ele-ment method to obtain solutions of structural engineering problem. Typically, for the structural stress-analysis problem, the engineer seeks to determine dis-placements and stresses throughout the structure, which is in equilibrium and is subjected to applied loads. For many structures, it is difficult to determine the dis-tribution of deformation using conventional methods, and thus the finite element method is necessarily used.
There are two general approaches associated with the finite element method. One approach, called the force, or flexibility method, uses internal forces as the unknowns of the problem. To obtain the governing equations, first the equilib-rium equations are used. Then necessary additional equations are found by in-troducing compatibility equations. The result is a set of algebraic equations for determining the redundant or unknown forces.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 27
The second approach, called the displacement, or stiffness method, as-sumes the displacements of the nodes as the unknowns of the problem. The governing equations are expressed in terms of nodal displacements using the equations of equilibrium and an applicable law relating forces to displacements.
These two approaches result in different unknowns (forces or displacements) in the analysis and different matrices associated with their formulations (flexibil-ities or stiffnesses). It has been shown that, for computational purposes, the dis-placement (or stiffness) method is more desirable because its formulation is sim-pler for most structural analysis problems. Consequently, only the displacement method will be used throughout this text.
The finite element method involves modeling the structure using small inter-connected elements called finite elements. A displacement function is associ-ated with each finite element. Every interconnected element is linked, directly or indirectly, to every other element through common (or shared) interfaces, includ-ing nodes and/or boundary lines and/or surfaces. The total set of equations de-scribing the behavior of each node results in a series of algebraic equations best expressed in matrix notation.
Step 1 - Discretize and Select Element Types
Step 1 involves dividing the body into an equivalent system of finite elements with associated nodes and choosing the most appropriate element type. The total number of elements used and their variation in size and type within a given body are primarily matters of engineering judgment. The elements must be made small enough to give usable results and yet large enough to reduce computational ef-fort. Small elements (and possibly higher-order elements) are generally desirable where the results are changing rapidly, such as where changes in geometry oc-cur, whereas large elements can be used where results are relatively constant.
The primary line elements, consist of bar (or truss) and beam elements. They have a cross-sectional area but are usually represented by line segments. In general, the cross-sectional area within the element can vary, but it will be considered to be constant throughout this text.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 28
These elements are often used to model trusses and frame structures. The
simplest line element (called a linear element) has two nodes, one at each end, although higher-order elements having three nodes or more (called quadratic, cubic, etc. elements) also exist. The line elements are the simplest of elements to consider and will be used to illustrate many of the basic concepts of the finite element method.
The basic two-dimensional (or plane) elements are loaded by forces in their
own plane (plane stress or plane strain conditions). They are triangular or quadri-lateral elements.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 29
The simplest two-dimensional elements have corner nodes only (linear ele-
ments) with straight sides or boundaries although there are also higher-order elements, typically with mid-side nodes (called quadratic elements) and curved sides. The elements can have variable thicknesses throughout or be constant. They are often used to model a wide range of engineering problems.
The most common three-dimensional elements are tetrahedral and hexahe-dral (or brick) elements; they are used when it becomes necessary to perform a three-dimensional stress analysis. The basic three dimensional elements have corner nodes only and straight sides, whereas higher-order elements with mid-edge nodes (and possible mid-face nodes) have curved surfaces for their sides.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 30
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 31
The axisymmetric element is developed by rotating a triangle or quadrilateral about a fixed axis located in the plane of the element through 360°. This element can be used when the geometry and loading of the problem are axisymmetric
Step 2 - Select a Displacement Function
Step 2 involves choosing a displacement function within each element. The function is defined within the element using the nodal values of the element. Lin-ear, quadratic, and cubic polynomials are frequently used functions because they are simple to work with in finite element formulation. The functions are expressed in terms of the nodal unknowns (in the two-dimensional problem, in terms of an x and a y component). Hence, the finite element method is one in which a continu-ous quantity, such as the displacement throughout the body, is approximated by a discrete model composed of a set of piecewise-continuous functions defined within each finite domain or finite element.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 32
Step 3 - Define the Strain/Displacement and Stress/Strain Relationships
Strain/displacement and stress-strain relationships are necessary for deriving the equations for each finite element. For one-dimensional small strain deforma-tion, say, in the x direction, we have strain εx, related to displacement u by:
dxdu
x =ε
In addition, the stresses must be related to the strains through the stress-strain law (generally called the constitutive law). The ability to define the mate-rial behavior accurately is most important in obtaining acceptable results. The simplest of stress-strain laws, Hooke’s law, often used in stress analysis, is given by:
xx Eεσ =
Step 4 - Derive the Element Stiffness Matrix and Equations
Initially, the development of element stiffness matrices and element equations was based on the concept of stiffness influence coefficients, which presupposes a background in structural analysis. We now present alternative methods used in this text that do not require this special background.
Direct Equilibrium Method - According to this method, the stiffness matrix and element equations relating nodal forces to nodal displacements are obtained using force equilibrium conditions for a basic element, along with force-deformation relationships. This method is most easily adaptable to line or one-dimensional elements (spring, bar, and beam elements).
Work or Energy Methods - To develop the stiffness matrix and equations for two- and three-dimensional elements, it is much easier to apply a work or energy method. The principle of virtual work (using virtual displacements), the princi-ple of minimum potential energy, and Castigliano’s theorem are methods fre-quently used for the purpose of derivation of element equations. We will present the principle of minimum potential energy (probably the most well known of the three energy methods mentioned here).
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 33
Methods of Weighted Residuals - The methods of weighted residuals are useful for developing the element equations (particularly popular is Galerkin’s method). These methods yield the same results as the energy methods, wher-ever the energy methods are applicable. They are particularly useful when a functional such as potential energy is not readily available. The weighted resid-ual methods allow the finite element method to be applied directly to any differen-tial equation.
Step 5 - Assemble the Element Equations and Introduce Boundary Condi-
tions
The individual element equations generated in Step 4 can now be added to-gether using a method of superposition (called the direct stiffness method) whose basis is nodal force equilibrium (to obtain the global equations for the whole structure). Implicit in the direct stiffness method is the concept of continu-ity, or compatibility, which requires that the structure remain together and that no tears occur anywhere in the structure. The final assembled or global equation written in matrix form is:
[ ] dKF = where F is the vector of global nodal forces, [K] is the structure global or total stiffness matrix, and d is now the vector of known and unknown structure nodal degrees of freedom or generalized displacements.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 34
Step 6 - Solve for the Unknown Degrees of Freedom
(or Generalized Displacements)
Once the element equations are assembled and modified to account for the boundary conditions, a set of simultaneous algebraic equations that can be writ-ten in expanded matrix form as:
⎪⎪⎭
⎪⎪⎬
⎫
⎪⎪⎩
⎪⎪⎨
⎧
⋅⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢
⎣
⎡
⋅⋅⋅⋅⋅
⋅⋅
=
⎪⎪⎭
⎪⎪⎬
⎫
⎪⎪⎩
⎪⎪⎨
⎧
⋅
nnnnn
n
n
n d
dd
KKK
KKKKKK
F
FF
2
1
21
22221
11211
2
1
where n is the structure total number of unknown nodal degrees of freedom. These equations can be solved for the d’s by using an elimination method (such as Gauss’s method) or an iterative method (such as Gauss Seidel’s method).
Step 7 - Solve for the Element Strains and Stresses
For the structural stress-analysis problem, important secondary quantities of strain and stress (or moment and shear force) can be obtained in terms of the displacements determined in Step 6.
Step 8 - Interpret the Results
The final goal is to interpret and analyze the results for use in the de-sign/analysis process. Determination of locations in the structure where large de-formations and large stresses occur is generally important in making de-sign/analysis decisions. Post-processor computer programs help the user to in-terpret the results by displaying them in graphical.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 35
Applications of the Finite Element Method
The following applications will illustrate the variety, size, and complexity of problems that can be solved using the finite element method and the typical dis-cretization process and kinds of elements used.
The first example is of a control tower for a railroad. The tower is a three-dimensional frame comprising a series of beam-type elements. The 48 elements are labeled by the circled numbers, whereas the 28 nodes are indicated by the encircled numbers. Each node has three rotation and three displacement com-ponents associated with it. The rotations and displacements are called the de-grees of freedom.
The next example is the determination of displacements and stresses in an underground box culvert subjected to ground shock loading from a bomb explo-sion. The discretized model that included a total of 369 nodes, 40 one-dimensional bar or truss elements used to model the steel reinforcement in the box culvert, and 333 plane strain two-dimensional triangular and rectangular elements used to model the surrounding soil and concrete box culvert. With an assumption of symmetry, only half of the box culvert must be analyzed. This problem requires the solution of nearly 700 unknown nodal displacements.
Another two-dimensional problem is that of a hydraulic cylinder rod end. It was modeled by 120 nodes and 297 plane strain triangular elements. Symmetry was also applied to the whole rod end so that only half of the rod end had to be ana-lyzed, as shown. The purpose of this analysis was to locate areas of high stress concentration in the rod end.
The next example shows a chimney stack section that is four form heights high (or a total of 32 ft high). The engineer used 584 beams to model the vertical and horizontal stiffeners making up the formwork, whereas 252 flat-plate ele-ments were used to model the inner wooden form and the concrete shell.
The next example shows the finite element discretized model of a proposed steel die used in a plastic film-making process. Two hundred forty axisymmetric elements were used to model the three-dimensional die.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 36
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 37
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 38
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 39
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 40
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 41
The next example illustrates the use of a three-dimensional solid element to model a swing casting for a backhoe frame. The three-dimensional hexahedral elements are necessary to model the irregularly shaped three-dimensional cast-ing. Two-dimensional models certainly would not yield accurate engineering solu-tions to this problem.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 42
Advantages of the Finite Element Method
The finite element method has been applied to numerous problems, both structural and non-structural. This method has a number of advantages that have made it very popular.
1. Model irregularly shaped bodies quite easily 2. Handle general load conditions without difficulty 3. Model bodies composed of several different materials because the ele-
ment equations are evaluated individually 4. Handle unlimited numbers and kinds of boundary conditions 5. Vary the size of the elements to make it possible to use small elements
where necessary 6. Alter the finite element model relatively easily and cheaply 7. Include dynamic effects 8. Handle nonlinear behavior existing with large deformations and nonlin-
ear materials The finite element method of structural analysis enables the designer to detect
stress, vibration, and thermal problems during the design process and to evalu-ate design changes before the construction of a possible prototype.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 43
Introduction to the Stiffness (Displacement) Method
Introduction
This section introduces some of the basic concepts on which the direct stiff-ness method is based. The linear spring is simple and an instructive tool to illus-trate the basic concepts. The steps to develop a finite element model for a linear spring follow our general 8 step procedure.
1. Discretize and Select Element Types - Linear spring elements 2. Select a Displacement Function - Assume a variation of the dis-
placements over each element. 3. Define the Strain/Displacement and Stress/Strain Relationships -
We next illustrate how to assemble the total stiffness matrix for a struc-ture comprising an assemblage of spring elements by using elementary concepts of equilibrium and compatibility.
4. Derive the Element Stiffness Matrix and Equations - Define the stiff-ness matrix for an element and then consider the derivation of the stiff-ness matrix for a linear-elastic spring element.
5. Assemble the Element Equations to Obtain the Global or Total Equations and Introduce Boundary Conditions - We then show how the total stiffness matrix for the problem can be obtained by superimpos-ing the stiffness matrices of the individual elements in a direct manner. The term direct stiffness method evolved in reference to this method. After establishing the total structure stiffness matrix, impose boundary conditions (both homogeneous and nonhomogeneous).
6. Solve for the Unknown Degrees of Freedom (or Generalized Dis-placements) - Solve for the nodal displacements.
7. Solve for the Element Strains and Stresses - The reactions and inter-nal forces association with the bar element.
8. Interpret the Results
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 44
Step 1 - Select Element Type
Consider the linear spring shown below. The spring is of length L and is sub-jected to a nodal tensile force, T directed along the x axis.
Step 2 - Select a Displacement Function
A displacement function u is assumed.
xaau ˆˆ 21 += In general, the number of coefficients in the displacement function is equal to the total number of degrees of freedom associated with the element. We can write the displacement function in matrix forms as:
[ ]⎭⎬⎫
⎩⎨⎧
=2
1ˆ1ˆaa
xu
We can express u as a function of the nodal displacements d by evaluating u at each node and solving for a1 and a2.
11ˆ)0ˆ(ˆ adxu x ===
122ˆ)ˆ(ˆ aLadLxu x +===
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 45
Solving for a2:
Ldda xx 12
2
ˆˆ −=
Substituting a1 and a2 into u gives:
xxx dx
Lddu 1
12 ˆˆˆˆ
ˆ +⎟⎟⎠
⎞⎜⎜⎝
⎛ −=
In matrix form:
⎭⎬⎫
⎩⎨⎧⎥⎦
⎤⎢⎣
⎡⎟⎠⎞
⎜⎝⎛ −=
x
x
dd
Lx
Lxu
2
1
ˆˆˆˆ
1ˆ
Or in another form:
[ ]⎭⎬⎫
⎩⎨⎧
=x
x
ddNNu
2
121 ˆ
ˆˆ
where N1 and N2 are defined as:
The functions Ni are called interpolation functions because they describe how the assumed displacement function varies over the domain of the element. In this case the interpolation functions are linear.
Step 3 - Define the Strain/Displacement and Stress/Strain Relationships
Tensile forces produce a total elongation (deformation) δ of the spring. For lin-ear springs, the force T and the displacement δ are related by Hooke’s law:
δkT =
LxNˆ
11 −=LxNˆ
2 =
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 46
where deformation of the spring δ is given as:
)0(ˆ)(ˆ uLu −=δ
xx dd 12ˆˆ −=δ
Step 4 - Derive the Element Stiffness Matrix and Equations
We can now derive the spring element stiffness matrix as follows:
Rewrite the forces in terms of the nodal displacements:
( )xxx ddkfT 121ˆˆˆ −=−=
( )xxx ddkfT 122ˆˆˆ −==
We can write the last two force-displacement relationships in matrix form as:
⎭⎬⎫
⎩⎨⎧⎥⎦
⎤⎢⎣
⎡ −−
=⎭⎬⎫
⎩⎨⎧
x
x
x
x
dd
kk
kk
ff
2
1
2
1
ˆˆ
ˆˆ
This formulation is valid as long as the spring deforms along the x axis. The co-efficient matrix of the above equation is called the local stiffness matrix k :
⎥⎦
⎤⎢⎣
⎡ −−
=kk
kk
k
Tf x −=1 Tf x =2
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 47
Step 5 - Assemble the Element Equations
and Introduce Boundary Conditions
The global stiffness matrix and the global force vector are assembled using the nodal force equilibrium equations, and force/deformation and compatibility equations.
[ ] ∑=
==N
e
ekK1
)(ˆK
∑=
==N
e
efF1
)(ˆF
where k and f are the element stiffness and force matrices expressed in global coordinates.
Step 6 - Solve for the Nodal Displacements
Solve the displacements by imposing the boundary conditions and solving the following set of equations:
KdF =
Step 7 - Solve for the Element Forces
Once the displacements are found, the forces in each element may be calcu-lated from:
δkT =
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 48
Example 1 - Spring Problem
Consider the following two-spring system shown below:
where the element axis x coincides with the global axis x. For element 1:
1 11 1
3 31 1
x x
x x
f dk kf dk k
−⎧ ⎫ ⎧ ⎫⎡ ⎤=⎨ ⎬ ⎨ ⎬⎢ ⎥−⎣ ⎦⎩ ⎭ ⎩ ⎭
For element 2:
3 2 2 3
2 2 2 2
x x
x x
f k k df k k d
−⎧ ⎫ ⎡ ⎤ ⎧ ⎫=⎨ ⎬ ⎨ ⎬⎢ ⎥−⎩ ⎭ ⎣ ⎦ ⎩ ⎭
Both continuity and compatibility require that both elements remain connected at node 3.
(1) (2)3 3x xd d=
We can write the nodal equilibrium equation at each node as:
(1)1 1x xF f=
(2)2 2x xF f=
(1) (2)3 3 3x x xF f f= +
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 49
Therefore the force-displacement equations for this spring system are:
1 1 1 1 3x x xF k d k d= −
2 2 3 2 2x x xF k d k d= − +
3 1 1 1 3 2 3 2 2x x x x xF k d k d k d k d= − + + − In matrix form the above equations are:
⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡
+−−−−
=⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧
x
x
x
x
x
x
ddd
kkkkkkkk
FFF
3
2
1
2121
22
11
3
2
1
00
or
KdF = where F is the global nodal force vector, d is called the global nodal dis-placement vector, and K is called the global stiffness matrix.
Assembling the Total Stiffness Matrix by Superposition
Consider the spring system defined in the last example:
The elemental stiffness matrices may be written for each element. For element 1:
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 50
⎥⎦
⎤⎢⎣
⎡ −−
=1
1
1
1)1(
31
kk
kk
k
dd xx
For element 2:
⎥⎦
⎤⎢⎣
⎡ −−
=2
2
2
2)2(
23
kk
kk
k
dd xx
Write the stiffness matrix in global format for element 1 as follows:
⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧=
⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡
−
−
)1(3
)1(2
)1(1
)1(3
)1(2
)1(1
1
101000101
x
x
x
x
x
x
fff
ddd
k
For element 2:
⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧=
⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡
−−
)2(3
)2(2
)2(1
)2(3
)2(2
)2(1
2
110110
000
x
x
x
x
x
x
fff
ddd
k
Apply the force equilibrium equations at each node.
⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧=
⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧+
⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧
x
x
x
x
x
x
x
FFF
ff
f
f
3
2
1
)2(3
)2(2
)1(3
)1(1 00
The above equations give:
1 1 1 1
2 2 2 2
1 2 1 2 3 3
00
x x
x x
x x
k k d Fk k d F
k k k k d F
−⎡ ⎤ ⎧ ⎫ ⎧ ⎫⎪ ⎪ ⎪ ⎪⎢ ⎥− =⎨ ⎬ ⎨ ⎬⎢ ⎥⎪ ⎪ ⎪ ⎪⎢ ⎥− − +⎣ ⎦ ⎩ ⎭ ⎩ ⎭
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 51
To avoid the expansion of the each elemental stiffness matrix, we can use a more direct, shortcut form of the stiffness matrix. For element 1:
⎥⎦
⎤⎢⎣
⎡ −−
=1
1
1
1)1(
31
kk
kk
k
dd xx
For element 2:
⎥⎦
⎤⎢⎣
⎡ −−
=2
2
2
2)2(
23
kk
kk
k
dd xx
The global stiffness matrix may be constructed by directly adding terms associ-ated with the degrees of freedom in k(1) and k(2) into their corresponding locations in the K as follows:
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡
+−−
−−=
21
2
1
2
2
1
1
321
00
kkkk
kk
k
kddd xxx
K
Boundary Conditions
In order to solve the equations defined by the global stiffness matrix, we must apply some form of constraints or supports or the structure will be free to move as a rigid body.
Boundary conditions are of two general types: homogeneous boundary conditions (the most common) occur at locations that are completely prevented from movement; nonhomogeneous boundary conditions occur where finite
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 52
non-zero values of displacement are specified, such as the settlement of a sup-port.
Consider the equations we developed for the two-spring system. We will con-sider node 1 to be fixed d1x = 0. The equations describing the elongation of the spring system become:
⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧=
⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡
+−−−−
x
x
x
x
x
FFF
dd
kkkkkkkk
3
2
1
3
2
2121
22
11 00
0
Expanding the matrix equations gives:
xx dkF 311 −=
xxx dkdkF 22322 +−=
( ) xxx dkkdkF 321223 ++−= The second and third equation may be written in matrix form as:
⎭⎬⎫
⎩⎨⎧
=⎭⎬⎫
⎩⎨⎧⎥⎦
⎤⎢⎣
⎡+−
− x
x
x
x
FF
dd
kkk
kk
3
2
3
2
21
2
2
2
Once we have solved the above equations for the unknown nodal displacements, we can use the first equation in the original matrix to find the support reaction.
xx dkF 311 −= For homogeneous boundary conditions, we can delete the row and column cor-responding to the zero-displacement degrees-of-freedom.
Let’s again look at the equations we developed for the two-spring system. However, this time we will consider a nonhomogeneous boundary condition at node 1 d1x = δ. The equations describing the elongation of the spring system be-come:
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 53
⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧=
⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡
+−−−−
x
x
x
x
x
FFF
dd
kkkkkkkk
3
2
1
3
2
2121
22
11
00 δ
Expanding the matrix equations gives:
xx dkkF 3111 −= δ
xxx dkdkF 32222 −=
( ) xxx dkkdkkF 3212213 ++−−= δ By considering the second and third equations because they have known nodal forces we get:
xxx dkdkF 32222 −=
( ) xxx dkkdkkF 3212213 ++−=+ δ Writing the above equations in matrix form gives:
⎭⎬⎫
⎩⎨⎧
+=
⎭⎬⎫
⎩⎨⎧⎥⎦
⎤⎢⎣
⎡+−
− δ13
2
3
2
21
2
2
2
kFF
dd
kkk
kk
x
x
x
x
For nonhomogeneous boundary conditions, we must transfer the terms from the stiffness matrix to the right-hand-side force vector before solving for the unknown displacements.
Once we have solved the above equations for the unknown nodal displace-ments, we can use the first equation in the original matrix to find the support re-action.
1 1 1 3x xF k k dδ= −
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 54
Example 2 -Spring Problem
Consider the following three-spring system:
The elemental stiffness matrices for each element are:
⎥⎦
⎤⎢⎣
⎡ −−
=11
11
1000)1(
31
k ⎥⎦
⎤⎢⎣
⎡ −−
=11
11
2000)2(
43
k
⎥⎦
⎤⎢⎣
⎡ −−
=11
11
3000)3(
24
k
Using the concept of superposition (the direct stiffness method), the global stiff-ness matrix is:
⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢
⎣
⎡
−−
−
−
−−
=
5000200030000
20003000
01000
30000
30000
010000
1000
K
The global force-displacement equations are:
⎪⎪⎭
⎪⎪⎬
⎫
⎪⎪⎩
⎪⎪⎨
⎧
=
⎪⎪⎭
⎪⎪⎬
⎫
⎪⎪⎩
⎪⎪⎨
⎧
⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢
⎣
⎡
−−
−
−
−−
x
x
x
x
x
x
x
x
FFFF
dddd
4
3
2
1
4
3
2
1
5000200030000
20003000
01000
30000
30000
010000
1000
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 55
We have homogeneous boundary conditions at nodes 1 and 2 (d1x = 0 and d2x = 0). Deleting the first two rows and the first two columns and substituting for the known force at node 4 (F4x = 5000 lb) gives:
⎭⎬⎫
⎩⎨⎧
=⎭⎬⎫
⎩⎨⎧
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡−
− 50000
50002000
20003000
4
3
x
x
dd
Solving for d3x and d4x gives:
.1115.
1110
43 indind xx ==
To obtain the global forces, substitute the displacement in the force-displacement equations.
⎪⎪⎭
⎪⎪⎬
⎫
⎪⎪⎩
⎪⎪⎨
⎧
⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢
⎣
⎡
−−
−
−
−−
=
⎪⎪⎭
⎪⎪⎬
⎫
⎪⎪⎩
⎪⎪⎨
⎧
1115
1110
4
3
2
1
00
5000200030000
20003000
01000
30000
30000
010000
1000
x
x
x
x
FFFF
Solving for the forces gives:
lbFlbF xx 11000,45
11000,10
21 −=−=
lbFF xx 11000,550 43 ==
Next, use the local element equations to obtain the force in each spring. For element 1:
⎭⎬⎫
⎩⎨⎧⎥⎦
⎤⎢⎣
⎡ −−
=⎭⎬⎫
⎩⎨⎧
1110
3
1 010001000
10001000
ˆˆ
x
x
ff
The local forces are:
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 56
lbflbf xx 11000,10ˆ
11000,10ˆ
31 =−=
A free-body diagram of the spring element 1 is shown below.
For element 2:
⎭⎬⎫
⎩⎨⎧⎥⎦
⎤⎢⎣
⎡ −−
=⎭⎬⎫
⎩⎨⎧
1115
1110
4
3
20002000
20002000
ˆˆ
x
x
ff
The local forces are:
lbflbf xx 11000,10ˆ
11000,10ˆ
43 =−=
A free-body diagram of the spring element 2 is shown below.
For element 3:
⎭⎬⎫
⎩⎨⎧⎥⎦
⎤⎢⎣
⎡ −−
=⎭⎬⎫
⎩⎨⎧
030003000
30003000
ˆˆ 11
15
2
4
x
x
ff
The local forces are:
lbflbf xx 11000,45ˆ
11000,45ˆ
24 −==
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 57
A free-body diagram of the spring element 2 is shown below.
Example 3 -Spring Problem
Consider the following four-spring system:
The spring constant k = 200 kN/m and the displacement δ = 20 mm. Therefore, the elemental stiffness matrices are:
mkNkkkk /11
11
200)4()3()2()1(⎥⎦
⎤⎢⎣
⎡ −−
====
Using superposition (the direct stiffness method), the global stiffness matrix is:
⎥⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢⎢
⎣
⎡
−−
−−
−−
−−
=
200200
000
200400200
00
0200400200
0
00
200400200
000
200200
K
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 58
The global force-displacement equations are:
⎪⎪⎪
⎭
⎪⎪⎪
⎬
⎫
⎪⎪⎪
⎩
⎪⎪⎪
⎨
⎧
=
⎪⎪⎪
⎭
⎪⎪⎪
⎬
⎫
⎪⎪⎪
⎩
⎪⎪⎪
⎨
⎧
⎥⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢⎢
⎣
⎡
−−
−−
−−
−−
x
x
x
x
x
x
x
x
x
x
FFFFF
ddddd
5
4
3
2
1
5
4
3
2
1
200200
000
200400200
00
0200400200
0
00
200400200
000
200200
Applying the boundary conditions (d1x = 0 and d5x = 20 mm) and the known forces (F2x, F3x, and F4x equal to zero) give:
⎪⎪⎭
⎪⎪⎬
⎫
⎪⎪⎩
⎪⎪⎨
⎧
=
⎪⎪⎭
⎪⎪⎬
⎫
⎪⎪⎩
⎪⎪⎨
⎧
⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢
⎣
⎡
−−−
−−−
x
x
x
x
Fddd
5
4
3
2
000
02.0200200002004002000020040020000200400
Rearranging the first three equations gives:
⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧=
⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡−
−
−−
400
400200
0
200400200
0200400
4
3
2
x
x
x
ddd
Solving for d2x, d3x, and d4x gives:
mdmdmd xxx 015.001.0005.0 432 === Solving for the forces F1x and F5x gives:
kNF x 0.1)005.0(2001 −=−=
kNF x 0.1)02.0(200)015.0(2005 =+−=
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 59
Next, use the local element equations to obtain the force in each spring. For element 1:
⎭⎬⎫
⎩⎨⎧⎥⎦
⎤⎢⎣
⎡ −−
=⎭⎬⎫
⎩⎨⎧
005.00
200200
200200
ˆˆ
2
1
x
x
ff
The local forces are:
kNfkNf xx 0.1ˆ0.1ˆ21 =−=
For element 2:
⎭⎬⎫
⎩⎨⎧⎥⎦
⎤⎢⎣
⎡ −−
=⎭⎬⎫
⎩⎨⎧
01.0005.0
200200
200200
ˆˆ
3
2
x
x
ff
The local forces are:
kNfkNf xx 0.1ˆ0.1ˆ32 =−=
For element 3:
⎭⎬⎫
⎩⎨⎧⎥⎦
⎤⎢⎣
⎡ −−
=⎭⎬⎫
⎩⎨⎧
015.001.0
200200
200200
ˆˆ
4
3
x
x
ff
The local forces are:
kNfkNf xx 0.1ˆ0.1ˆ43 =−=
For element 4:
⎭⎬⎫
⎩⎨⎧⎥⎦
⎤⎢⎣
⎡ −−
=⎭⎬⎫
⎩⎨⎧
02.0015.0
200200
200200
ˆˆ
5
4
x
x
ff
The local forces are:
kNfkNf xx 0.1ˆ0.1ˆ54 =−=
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 60
Example 4 -Spring Problem
Consider the following spring system:
The boundary conditions are:
0431 === xxx ddd The compatibility condition at node 2 is:
xxxx dddd 2)3(
2)2(
2)1(
2 === Using the direct stiffness method: the elemental stiffness matrices for each ele-ment are:
⎥⎦
⎤⎢⎣
⎡ −−
=1
1
1
1)1(
21
kk
kk
k ⎥⎦
⎤⎢⎣
⎡ −−
=2
2
2
2)2(
32
kk
kk
k
⎥⎦
⎤⎢⎣
⎡ −−
=3
3
3
3)3(
42
kk
kk
k
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 61
Using the concept of superposition (the direct stiffness method), the global stiff-ness matrix is:
⎥⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢⎢
⎣
⎡
−−
−−
++−
−=
3
3
2
2
3
2
321
1
1
1
0
0
0
0
00
4321
k
kkk
kk
kkkk
kk
K
Applying the boundary conditions (d1x = d3x = d4x = 0) and the known forces (F2x
= P) gives:
( ) xdkkkP 2321 ++= Solving the equation gives:
3212 kkk
Pd x ++=
Solving for the forces gives:
xxxxxx dkFdkFdkF 234223211 −=−=−=
Potential Energy Approach to Derive Spring Element Equations
One of the alternative methods often used to derive the element equations and the stiffness matrix for an element is based on the principle of minimum potential energy. This method has the advantage of being more general than the methods involving nodal and element equilibrium equations, along with the stress/strain law for the element.
Thus, the principle of minimum potential energy is more adaptable for the de-termination of element equations for complicated elements (those with large numbers of degrees of freedom) such as the plane stress/strain element, the axi-
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 62
symmetric stress element, the plate bending element, and the three-dimensional solid stress element.
Total Potential Energy
The total potential energy is defined as the sum of the internal strain energy U and the potential energy of the external forces Ω:
Ω+=Upπ
Strain energy is the capacity of the internal forces (or stresses) to do work through deformations (strains) in the structure; Ω is the capacity of forces such as body forces, surface traction forces, and applied nodal forces to do work through the deformation of the structure.
Recall the force-displacement relationship for a linear spring:
kxF = The differential internal work (or strain energy) dU in the spring is the internal force multiplied by the change in displacement which the force moves through:
( )dxkxFdxdU ==
The total strain energy is:
( ) 2
0 21 kxdxkxdUU
x
L
=== ∫∫
The strain energy is the area under the force-displacement curve. The potential energy of the external forces is the work done by the external forces:
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 63
Fx−=Ω Therefore, the total potential energy is:
Fxkxp −= 2
21π
The concept of a stationary value of a function G is shown below:
The function G is expressed in terms of x. To find a value of x yielding a station-ary value of G(x), we use differential calculus to differentiate G with respect to x and set the expression equal to zero.
0=dxdG
We can replace G with the total potential energy πp and the coordinate x with a discrete value di. To minimize πp we first take the variation of πp (we will not cover the details of variational calculus):
nn
pppp d
dd
dd
dδ
πδ
πδ
πδπ
∂∂
++∂∂
+∂∂
= ...22
11
The principle states that equilibrium exist when the di define a structure state such that δπp = 0 for arbitrary admissible variations δdi from the equilibrium state. An admissible variation is one in which the displacement field still satisfies the boundary conditions and interelement continuity.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 64
To satisfy δπp = 0, all coefficients associated with δdi must be zero independently, therefore:
0,,2,10 =∂∂
==∂∂
dorni
dp
i
p ππ
Example 5 -Spring Problem
Consider the following linear-elastic spring system subjected to a force of 1,000 lb. Evaluate the potential energy for various displacement values and show that the minimum potential energy also corresponds to the equilibrium position of the spring.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 65
The total potential energy is defined as the sum of the internal strain energy U and the potential energy of the external forces Ω:
The variation of πp with respect to x is:
0=∂∂
= xx
pp δ
πδπ
Since δx is arbitrary and might not be zero, then:
0=∂∂
xpπ
Using our express for πp, we get:
( ) xlbxinlbFxkxp 1000)/(50021
21 22 −=−=π
.0.210005000 inxxx
p =−==∂∂π
If we had plotted the total potential energy function πp for various values of de-formation we would get:
Fx−=Ω2
21 kxU =
Ω+=Upπ
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 66
Let’s derive the spring element equations and stiffness matrix using the principal of minimum potential energy. Consider the linear spring subjected to nodal forces shown below:
The total potential energy πp is:
( ) xxxxxxp dfdfddk 2211
2
12ˆˆˆˆˆˆ
21
−−−=π
Expanding the above express gives:
( ) xxxxxxxxp dfdfddddk 22112
1212
2ˆˆˆˆˆˆˆ2ˆ
21
−−−−=π
Minimizing the total potential energy πp:
210 toidi
p ==∂∂π
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 67
( ) xxxp fddk
d 1121
ˆˆ2ˆ2210 −+−==
∂∂π
( ) xxxp fddk
d 2122
ˆˆ2ˆ2210 −−==
∂∂π
In matrix form the above equations are:
⎭⎬⎫
⎩⎨⎧⎥⎦
⎤⎢⎣
⎡ −−
=⎭⎬⎫
⎩⎨⎧
x
x
x
x
dd
kk
kk
ff
2
1
2
1
ˆˆ
ˆˆ
This result is identical to stiffness matrix we obtained using equilibrium.
Example 6 - Spring Problem
Obtain the total potential energy of the spring system shown below and find its minimum value.
The potential energy πp for element 1 is:
( ) xxxxxxp dfdfddk 33112
131)1( ˆ
21
−−−=π
The potential energy πp for element 2 is:
( ) xxxxxxp dfdfddk 44332
342)2(
21
−−−=π
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 68
The potential energy πp for element 3 is:
( ) xxxxxxp dfdfddk 44222
423)3(
21
−−−=π
The total potential energy πp for the spring system is:
∑=
=3
1
)(
e
epp ππ
Minimizing the total potential energy πp:
)1(11131
1
0 xxxx
p fdkdkd
−+−==∂∂π
)3(24323
2
0 xxxx
p fdkdkd
−−==∂∂π
)2(3
)1(332421131
3
0 xxxxxxx
p ffdkdkdkdkd
−−+−−==∂∂π
)3(4
)2(443233242
4
0 xxxxxxx
p ffdkdkdkdkd
−−+−−==∂∂π
In matrix form:
⎪⎪⎭
⎪⎪⎬
⎫
⎪⎪⎩
⎪⎪⎨
⎧
++
=
⎪⎪⎭
⎪⎪⎬
⎫
⎪⎪⎩
⎪⎪⎨
⎧
⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢
⎣
⎡
+−−
−+
−−
)3(
4
)2(
4
)2(
3
)1(
3
2
1
4
3
2
1
32
2
3
2
21
1
3
3
1
1 00
0
0
0
0
xx
xx
x
x
x
x
x
x
ffff
ff
dddd
kkkk
kkk
k
k
kk
k
Using the following force equilibrium equations:
xx Ff 1
)1(1 = xx Ff 2
)3(2 =
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 69
The global force-displacement equations are:
⎪⎪⎭
⎪⎪⎬
⎫
⎪⎪⎩
⎪⎪⎨
⎧
=
⎪⎪⎭
⎪⎪⎬
⎫
⎪⎪⎩
⎪⎪⎨
⎧
⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢
⎣
⎡
−−
−
−
−−
x
x
x
x
x
x
x
x
FFFF
dddd
4
3
2
1
4
3
2
1
5000200030000
20003000
01000
30000
30000
010000
1000
The above equations are identical to those we obtained through the direct stiff-ness method.
Problems:
5. Do problems 2.2, 2.9, and 2.17 on pages 59 - 62 in your textbook “A First Course in the Finite Element Method” by D. Logan.
xxx Fff 4)3(
4)2(
4 =+xxx Fff 3
)2(
3
)1(
3 =+
CIVL 7117 Finite Element Methods in Structural Mechanics Page 70
Development of Truss Equations
Introduction
Having set forth the foundation on which the direct stiffness method is based, we will now derive the stiffness matrix for a linear-elastic bar (or truss) element using the general steps outlined in Chapter 1. We will include the introduction of both a local coordinate system, chosen with the element in mind, and a global or reference coordinate system, chosen to be convenient (for numerical purposes) with respect to the overall structure.
We will also discuss the transformation of a vector from the local coordinate system to the global coordinate system, using the concept of transformation ma-trices to express the stiffness matrix of an arbitrarily oriented bar element in terms of the global system. We will solve example plane truss problems to illus-trate the procedure of establishing the total stiffness matrix and equations for so-lution of a structure.
Next we will describe how to handle inclined, or skewed, supports. We will then extend the stiffness method to include space trusses. We will develop the transformation matrix in three-dimensional space and analyze a space truss.
We will then use the principle of minimum potential energy and apply it to the bar element equations. Finally, we will introduce Galerkin’s residual method and then apply it to derive the bar element equations.
Derivation of the Stiffness Matrix for a Bar Element
Consider the derivation of the stiffness matrix for the linear-elastic, constant cross-sectional area (prismatic) bar element show below. This application is di-rectly applicable to the solution of pin-connected truss problems. Consider the idealized element as:
CIVL 7117 Finite Element Methods in Structural Mechanics Page 71
where T is the tensile force directed along the axis at nodes 1 and 2, x is the lo-cal coordinate system directed along the length of the bar, and x and y are the global coordinate system. The bar element has a constant cross-section A, an initial length L, and modulus of elasticity E. The nodal degrees of freedom are the local axial displacements xd1
ˆ and xd2ˆ at the ends of the bar.
The strain-displacement relationship is:
xdudEˆˆ
== εεσ
From equilibrium of forces, assuming no distributed loads acting on the bar, we get:
constant== TA xσ Combining the above equations gives:
constantˆˆ
== TxdudAE
CIVL 7117 Finite Element Methods in Structural Mechanics Page 72
taking the derivative of the above equation with respect to the local coordinate x gives:
0ˆˆ
ˆ=⎟
⎠⎞
⎜⎝⎛
xdudAE
xdd
where u is the axial displacement function in the x direction and A and E can be written as functions of x .
The following assumptions are considered in deriving the bar element stiffness matrix:
1. The bar cannot sustain shear force: 0ˆˆ21 == yy ff .
2. Any effect of transverse displacement is ignored. 3. Hooke’s law applies: stress is related to strain: xx Eεσ =
Step 1 - Select Element Type
We will consider the linear bar element shown below.
CIVL 7117 Finite Element Methods in Structural Mechanics Page 73
Step 2 - Select a Displacement Function
A linear displacement function u is assumed.
xaau ˆˆ 21 += The number of coefficients in the displacement function, ai, is equal to the total number of degrees of freedom associated with the element. Applying the bound-ary conditions and solving for the unknown coefficients gives:
xxx dx
Lddu 1
12 ˆˆˆˆ
ˆ +⎟⎟⎠
⎞⎜⎜⎝
⎛ −=
In matrix form:
⎭⎬⎫
⎩⎨⎧
⎥⎦
⎤⎢⎣
⎡⎟⎠⎞
⎜⎝⎛ −=
x
x
dd
Lx
Lxu
2
1
ˆˆˆˆ
1ˆ
Or in another form:
[ ]⎭⎬⎫
⎩⎨⎧
=x
x
ddNNu
2
121 ˆ
ˆˆ
where N1 and N2 are the interpolation functions gives as:
LxN
LxN
ˆˆ1 21 =−=
The linear displacement function u plotted over the length of the bar element is shown below.
CIVL 7117 Finite Element Methods in Structural Mechanics Page 74
Step 3 - Define the Strain/Displacement and Stress/Strain Relationships
The stress-displacement relationship is:
Ldd
xdud xx
x12
ˆˆˆˆ −
==ε
Step 4 - Derive the Element Stiffness Matrix and Equations
We can now derive the element stiffness matrix as follows:
xAT σ=
Substituting the stress-displacement relationship into the above equation gives:
⎟⎟⎠
⎞⎜⎜⎝
⎛ −=
LddAET xx 12ˆˆ
The nodal force sign convention, defined in element figure, is:
CIVL 7117 Finite Element Methods in Structural Mechanics Page 75
TfTf xx =−= 21ˆˆ
therefore,
⎟⎟⎠
⎞⎜⎜⎝
⎛ −=⎟⎟
⎠
⎞⎜⎜⎝
⎛ −=
LddAEf
LddAEf xx
xxx
x12
221
1
ˆˆˆˆˆˆ
Writing the above equations in matrix form gives:
⎭⎬⎫
⎩⎨⎧
⎥⎦
⎤⎢⎣
⎡ −−
=⎭⎬⎫
⎩⎨⎧
x
x
x
x
dd
LAE
ff
2
1
2
1
ˆˆ
11
11
ˆˆ
Notice that AE/L for a bar element is analogous to the spring constant k for a spring element.
Step 5 - Assemble the Element Equations and
Introduce Boundary Conditions
The global stiffness matrix and the global force vector are assembled using the nodal force equilibrium equations, and force/deformation and compatibility equations.
[ ] ∑∑==
====N
e
eN
e
e fFkK1
)(
1
)( ˆˆ FK
where k and f are the element stiffness and force matrices expressed in global coordinates.
Step 6 - Solve for the Nodal Displacements
Solve the displacements by imposing the boundary conditions and solving the following set of equations:
KdF =
CIVL 7117 Finite Element Methods in Structural Mechanics Page 76
Step 7 - Solve for the Element Forces
Once the displacements are found, the stress and strain in each element may be calculated from:
xxxx
x EL
ddxdud εσε =
−== 12
ˆˆˆˆ
Example 1 - Spring Problem
Consider the following three-bar system shown below. Assume for elements 1 and 2: A = 1 in2 and E = 30 (106) psi and for element 3: A = 2 in2 and E = 15 (106) psi.
Determine (a) the global stiffness matrix, (b) the displacement of nodes 2 and 3, and (c) the reactions at nodes 1 and 4. For elements 1 and 2:
( )( )in
lbin
lbkk ⎥⎦
⎤⎢⎣
⎡ −−
=⎥⎦
⎤⎢⎣
⎡ −−
×==
11
11
1011
11
3010301 6
6)2()1(
2elementfornumbersnode32
1elementfornumbersnode21
CIVL 7117 Finite Element Methods in Structural Mechanics Page 77
For element 3:
( )( )in
lbin
lbk ⎥⎦
⎤⎢⎣
⎡ −−
=⎥⎦
⎤⎢⎣
⎡ −−
×=
11
11
1011
11
3010152 6
6)3(
3elementfornumbersnode43
where, as before, the numbers above the matrices indicate the displacements associated with the matrix. Assembling the global stiffness matrix by the direct stiffness methods gives:
⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢
⎣
⎡
−−
−−
−−
=
1100
1210
0121
0011
106
4321
K
xxxx dddd
Relating global nodal forces related to global nodal displacements gives:
⎪⎪⎭
⎪⎪⎬
⎫
⎪⎪⎩
⎪⎪⎨
⎧
⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢
⎣
⎡
−−
−−
−−
=
⎪⎪⎭
⎪⎪⎬
⎫
⎪⎪⎩
⎪⎪⎨
⎧
x
x
x
x
x
x
x
x
dddd
FFFF
4
3
2
1
6
4
3
2
1
1100
1210
0121
0011
10
The boundary conditions are:
041 == xx dd Applying the boundary conditions and the known forces (F2x = 3000 lb.) gives:
⎭⎬⎫
⎩⎨⎧
⎥⎦
⎤⎢⎣
⎡ −−
=⎭⎬⎫
⎩⎨⎧
x
x
dd
3
26
21
12
100
3000
Solving for d2x and d3x gives:
CIVL 7117 Finite Element Methods in Structural Mechanics Page 78
.001.0.002.0 32 indind xx == The global nodal forces are calculated as:
lbs
FFFF
x
x
x
x
⎪⎪⎭
⎪⎪⎬
⎫
⎪⎪⎩
⎪⎪⎨
⎧
−
−
=
⎪⎪⎭
⎪⎪⎬
⎫
⎪⎪⎩
⎪⎪⎨
⎧
⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢
⎣
⎡
−−
−−
−−
=
⎪⎪⎭
⎪⎪⎬
⎫
⎪⎪⎩
⎪⎪⎨
⎧
10000
30002000
0001.0002.00
1100
1210
0121
0011
106
4
3
2
1
Selecting Approximation Functions for Displacements
Consider the following guidelines, as they relate to the one-dimensional bar element, when selecting a displacement function.
1. Common approximation functions are usually polynomials, where the function is expressed in terms of the shape functions.
2. The approximation function should be continuous within the bar element. 3. The approximating function should provide interelement continuity for all
degrees of freedom at each node for discrete line elements, and along common boundary lines and surfaces for two- and three-dimensional elements.
For the bar element, we must ensure that nodes common to two or more elements remain common to these elements upon deformation and thus prevent overlaps or voids between elements.
The linear function is then called a conforming (or compatible) function
for the bar element because it ensures both the satisfaction of continuity between adjacent elements and of continuity within the element.
CIVL 7117 Finite Element Methods in Structural Mechanics Page 79
4. The approximation function should allow for rigid-body displacement and for a state of constant strain within the element.
The interpolation function must allow for a rigid-body displacement, that
means the function must be capable of yielding a constant value. Consider the follow situation;
xx ddaau 2111ˆˆˆ ===
therefore,
( ) 1212211ˆˆˆ aNNdNdNu xx +=+=
since u = a1 then:
( ) 1211ˆ aNNau +== this means that:
121 =+ NN The displacement interpolation function must add to unity at every point within the element so the u will yield a constant value when a rigid-body displacement occurs.
Transformation of Vectors in Two Dimensions
In many problems it is convenient to introduce both local and global (or refer-ence) coordinates. Local coordinates are always chosen to conveniently repre-sent the individual element. Global coordinates are chosen to be convenient for the whole structure.
Given the nodal displacement of an element, represented by the vector d in the figure below, we want to relate the components of this vector in one coordi-nate system to components in another.
CIVL 7117 Finite Element Methods in Structural Mechanics Page 80
Let’s consider that d does not coincident with either the local or global axes. In
this case, we want to relate global displacement components to local ones. In so doing, we will develop a transformation matrix that will subsequently be used to develop the global stiffness matrix for a bar element. We define the angle θ to be positive when measured counterclockwise from x to x . We can express vector displacement d in both global and local coordinates by:
jdiddd yxyxˆˆˆˆ +=+= jid
where i and j are unit vectors in the x and y directions, and i and j are unit vec-tor in the x and y direction.
CIVL 7117 Finite Element Methods in Structural Mechanics Page 81
Consider the following diagram:
Using vector addition:
iba =+ Using the law of cosines, we get:
θθ cos||cos|||| == aia
Similarly:
θθ sin||sin|||| == bib The vector a is in the i direction and b is in the j direction, therefore:
( ) ( ) ( )( )jjbbiia ˆsinˆ||ˆcosˆ|| −=−=== θθa The vector i can be rewritten as:
ji ˆsinˆcos θθ −=i
CIVL 7117 Finite Element Methods in Structural Mechanics Page 82
In a similar manner, we could get an expression for j
jij ˆcosˆsin θθ += If we use the above express to describe the displacement vector we get:
( ) ( ) jdidjidjidjdid yxyxyxˆˆˆˆˆcosˆsinˆsinˆcos +=++−=+ θθθθ
If we combine like coefficients of i and j we get:
xyx ddd ˆsincos =+ θθ
yyx ddd ˆcossin =+− θθ
In matrix form the above equations are:
θθ
sincos
ˆˆ
==
⎭⎬⎫
⎩⎨⎧
⎥⎦
⎤⎢⎣
⎡−
=⎭⎬⎫
⎩⎨⎧
SC
dd
CSSC
dd
y
x
y
x
The above equation relates the global displacement d to the local displacements d . The matrix above is called the transformation matrix.
⎥⎦
⎤⎢⎣
⎡− C
SSC
The figure below shows xd expressed in terms of the global coordinates x and y.
CIVL 7117 Finite Element Methods in Structural Mechanics Page 83
The magnitude of xd is:
yxx SdCdd +=ˆ
Example 2 - Bar Element Problem
The global nodal displacement at node 2 is d2x = 0.1 in. and d2y = 0.2 in. for the bar element shown below. Determine the local x displacement.
Using the following expression we just derived, we get:
yxx SdCdd +=ˆ
CIVL 7117 Finite Element Methods in Structural Mechanics Page 84
.223.0)2.0(60sin)01.0(60cos2 ind oox =+=
Global Stiffness Matrix
We will now use the transformation relationship developed above to obtain the global stiffness matrix for a bar element. We known that for a bar element in local coordinates we have:
dkfdd
LAE
ff
x
x
x
x ˆˆˆˆˆ
11
11
ˆˆ
2
1
2
1 =⎭⎬⎫
⎩⎨⎧
⎥⎦
⎤⎢⎣
⎡ −−
=⎭⎬⎫
⎩⎨⎧
We want to relate the global element forces f to the global displacements d for a bar element with an arbitrary orientation.
kdf
dddd
k
ffff
y
x
y
x
y
x
y
x
=
⎪⎪⎭
⎪⎪⎬
⎫
⎪⎪⎩
⎪⎪⎨
⎧
=
⎪⎪⎭
⎪⎪⎬
⎫
⎪⎪⎩
⎪⎪⎨
⎧
2
2
1
1
2
2
1
1
Using the relationship between local and global components, we can develop the global stiffness matrix. We already know the transformation relationships:
xyx ddd ˆsincos =+ θθ
Combining both expression xd1
ˆ and xd2ˆ , in matrix form, we get:
dTd
dddd
SCSC
dd
y
x
y
x
x
x *
2
2
1
1
2
1 ˆ0000ˆ
ˆ=
⎪⎪⎭
⎪⎪⎬
⎫
⎪⎪⎩
⎪⎪⎨
⎧
⎥⎦
⎤⎢⎣
⎡=
⎭⎬⎫
⎩⎨⎧
where
⎥⎦
⎤⎢⎣
⎡=
SCSC
T00
00*
CIVL 7117 Finite Element Methods in Structural Mechanics Page 85
A similar expression for the force transformation can be developed.
fTf
ffff
SCSC
ff
y
x
y
x
x
x *
2
2
1
1
2
1 ˆ0000ˆ
ˆ=
⎪⎪⎭
⎪⎪⎬
⎫
⎪⎪⎩
⎪⎪⎨
⎧
⎥⎦
⎤⎢⎣
⎡=
⎭⎬⎫
⎩⎨⎧
Substituting the global force expression into element force equation gives:
dkfTdkf ˆˆˆˆˆ * =⇒= Substituting the transformation between local and global dispalcements gives:
dTkfT ** ˆ= The matrix T* is not a square matrix so we cannot invert it. Let’s expand our rela-tionship between local and global displacement.
Tdd
dddd
CS
SC
CS
SC
dddd
y
x
y
x
y
x
y
x
=
⎪⎪⎭
⎪⎪⎬
⎫
⎪⎪⎩
⎪⎪⎨
⎧
⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢
⎣
⎡
−
−=
⎪⎪⎭
⎪⎪⎬
⎫
⎪⎪⎩
⎪⎪⎨
⎧
ˆ00
00
00
00
ˆˆˆˆ
2
2
1
1
2
2
1
1
where T is:
⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢
⎣
⎡
−
−=
CS
SC
CS
SC
T00
00
00
00
We can write a similar expression for the relationship between local and global forces.
CIVL 7117 Finite Element Methods in Structural Mechanics Page 86
Tff
ffff
CS
SC
CS
SC
ffff
y
x
y
x
y
x
y
x
=
⎪⎪⎭
⎪⎪⎬
⎫
⎪⎪⎩
⎪⎪⎨
⎧
⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢
⎣
⎡
−
−=
⎪⎪⎭
⎪⎪⎬
⎫
⎪⎪⎩
⎪⎪⎨
⎧
ˆ00
00
00
00
ˆˆˆˆ
2
2
1
1
2
2
1
1
Therefore our original local coordinate force-displacement expression
dkfdd
LAE
ff
x
x
x
x ˆˆˆˆˆ
11
11
ˆˆ
2
1
2
1 =⎭⎬⎫
⎩⎨⎧
⎥⎦
⎤⎢⎣
⎡ −−
=⎭⎬⎫
⎩⎨⎧
May be expanded
⎪⎪⎭
⎪⎪⎬
⎫
⎪⎪⎩
⎪⎪⎨
⎧
⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢
⎣
⎡ −
−=
⎪⎪⎭
⎪⎪⎬
⎫
⎪⎪⎩
⎪⎪⎨
⎧
y
x
y
x
y
x
y
x
dddd
LAE
ffff
2
2
1
1
2
2
1
1
ˆˆˆˆ
0000
0101
0000
01
01
ˆˆˆˆ
The global force-displacement equations are:
TdkTfdkf ˆˆˆˆ =⇒= Multiply both side by T -1 we get:
TdkTf ˆ1−= where T-1 is the inverse of T. It can be shown that:
TTT =−1 The global force-displacement equations become:
TdkTf T ˆ=
CIVL 7117 Finite Element Methods in Structural Mechanics Page 87
Where the global stiffness matrix k is:
TkTk T ˆ= Expanding the above transformation gives:
⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢
⎣
⎡−−
−−
−−
−−
=
2
2
2
2
2
2
2
2
SCSSCS
CSCCSC
SCSSCS
CSC
CSC
LAEk
We can assemble the total stiffness matrix by using the above element stiffness matrix and the direct stiffness method.
[ ] ∑∑==
====N
e
eN
e
e fFkK1
)(
1
)( FK
dKF = Local forces can be computed as:
1 1 1
1 1 1
222
222
ˆ ˆ1 0 1 0 1 0 1 0 0 0ˆ ˆ0 0 0 0 0 0 0 0 0 0ˆ ˆ1 0 1 0 1 0 1 0 0 0
0 0 0 0 0 0 0 0 0 0ˆˆ
x x x
y y y
xxx
yyy
f d dC Sf d dS CAE AE
dC SL LdfdS Cdf
⎧ ⎫ ⎧ ⎫ ⎧ ⎫− −⎡ ⎤ ⎡ ⎤ ⎡ ⎤⎪ ⎪ ⎪ ⎪ ⎪ ⎪⎢ ⎥ ⎢ ⎥ ⎢ ⎥⎪ ⎪ −⎪ ⎪ ⎪ ⎪⎢ ⎥ ⎢ ⎥ ⎢ ⎥= =⎨ ⎬ ⎨ ⎬ ⎨ ⎬⎢ ⎥ ⎢ ⎥ ⎢ ⎥− −⎪ ⎪ ⎪ ⎪ ⎪ ⎪⎢ ⎥ ⎢ ⎥ ⎢ ⎥⎪ ⎪ ⎪ ⎪ ⎪ ⎪−⎣ ⎦ ⎣ ⎦ ⎣ ⎦ ⎩ ⎭⎩ ⎭⎩ ⎭
1 1 1 2 2
1
1 1 2 22
2
ˆ
ˆ 0ˆ
0ˆ
x x y x y
y
x y x yx
y
f Cd Sd Cd Sdf AE
Cd Sd Cd SdLf
f
⎧ ⎫ + − −⎡ ⎤⎪ ⎪ ⎢ ⎥⎪ ⎪ ⎢ ⎥=⎨ ⎬ ⎢ ⎥− − + +⎪ ⎪ ⎢ ⎥⎪ ⎪ ⎣ ⎦⎩ ⎭
CIVL 7117 Finite Element Methods in Structural Mechanics Page 88
Example 3 - Bar Element Problem
For the bar element shown below, evaluate the global stiffness matrix. As-sume the cross-sectional area is 2 in.2, the length is 60 in., and the E is 30 x 106 psi.
Therefore:
2130sin
2330cos30 ===== ooo SCθ
The global elemental stiffness matrix is:
( )in
lbk
⎥⎥⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢⎢⎢
⎣
⎡
−−
−−
−−
−−
×=
41
43
41
43
43
43
43
43
41
43
41
43
43
43
43
43
601030)2( 6
Simplifying the global elemental stiffness matrix is:
inlbk
⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢
⎣
⎡−−
−−
−−
−−
=
250.0433.0250.0433.0
433.0750.0433.0750.0
250.0433.0
250.0433.0
433.0750.0
433.0750.0
106
Computation of Stress for a Bar in the x-y Plane
For a bar element the local forces are related to the local displacements by:
CIVL 7117 Finite Element Methods in Structural Mechanics Page 89
⎭⎬⎫
⎩⎨⎧
⎥⎦
⎤⎢⎣
⎡ −−
=⎭⎬⎫
⎩⎨⎧
x
x
x
x
dd
LAE
ff
2
1
2
1
ˆˆ
11
11
ˆˆ
The force-displacement equation for xf2 is:
[ ]⎭⎬⎫
⎩⎨⎧
−=x
xx d
dL
AEf2
12 ˆ
ˆ11ˆ
Therefore the stress is:
[ ] 12
2
ˆˆ1 1
ˆxx
x
df EA L d
σ⎧ ⎫⎪ ⎪= = − ⎨ ⎬⎪ ⎪⎩ ⎭
The stress in terms of global displacement is:
[ ]1
11 1 2 2
2
2
0 01 1
0 0
x
yx y x y
x
y
ddC SE E Cd Sd Cd SddC SL Ld
σ
⎧ ⎫⎪ ⎪
⎡ ⎤ ⎪ ⎪ ⎡ ⎤= − = − − + +⎨ ⎬⎢ ⎥ ⎣ ⎦⎣ ⎦ ⎪ ⎪
⎪ ⎪⎩ ⎭
Example 4 - Bar Element Problem
For the bar element shown below, determine the axial stress. Assume the cross-sectional area is 4 x 10-4 m2, the length is 2 m, and the E is 210 GPa. The global displacements are known as d1x = 0.25 mm, d1y = 0, d2x = 0.5 mm, and d2y = 0.75 mm.
CIVL 7117 Finite Element Methods in Structural Mechanics Page 90
mKN⎥
⎦
⎤⎢⎣
⎡++−−
×= )75.0(
43)5.0(
21)0(
43)25.0(
21
210210 6
σ
MPamkN 32.81/1032.81 3 =×=σ
Solution of a Plane Truss
We will now illustrate the use of equations developed above along with the di-rect stiffness method to solve the following plane truss example problems. A plane truss is a structure composed of bar elements all lying in a common plane that are connected together by frictionless pins. The plane truss also must have loads acting only in the common plane.
Example 5 - Plane Truss Problem
The plane truss shown below is composed of three bars subjected to a down-ward force of 10 kips at node 1. Assume the cross-sectional area A = 2 in.2 and E is 30 x 106 psi for all elements. Determine the x and y displacement at node 1 and stresses in each element.
Element Node 1 Node 2 θ C S 1 1 2 90o 0 1 2 1 3 45o 0.707 0.707 3 1 4 0o 1 0
CIVL 7117 Finite Element Methods in Structural Mechanics Page 91
The global elemental stiffness matrix for element 1 is:
⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢
⎣
⎡−
−
×=
101
0
0000
1010
0000
120)1030)(2( 6
)1(
2211
k
yxyxdddd
The global elemental stiffness matrix for element 2 is:
⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢
⎣
⎡−−
−−
−−
−−
×=
1111
1111
11
11
11
11
2240)1030)(2( 6
)2(
3311
k
yxyxdddd
The global elemental stiffness matrix for element 3 is:
⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢
⎣
⎡ −
−×
=
0000
0101
0000
01
01
120)1030)(2( 6
)3(
4411
k
yxyxdddd
CIVL 7117 Finite Element Methods in Structural Mechanics Page 92
The total global stiffness matrix is:
inlbK
⎥⎥⎥⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢⎢⎢⎢
⎣
⎡
×=
−−−
−−
−
−−
−
−−−
00000000
01000001
00354.0354.000354.0354.0
00354.0354.000354.0354.0
0000101
0
00000000
00354.0354.01
0354.1354.0
01354.0354.0
00354.0354.1
5105
The total global force-displacement equations are:
⎪⎪⎪
⎭
⎪⎪⎪
⎬
⎫
⎪⎪⎪
⎩
⎪⎪⎪
⎨
⎧
⎪⎪⎪⎪
⎭
⎪⎪⎪⎪
⎬
⎫
⎪⎪⎪⎪
⎩
⎪⎪⎪⎪
⎨
⎧
=
−
0000001
1
4
4
3
3
2
2
000,100
ydxd
yFxFyFxFyFxF
K
The above equations reduce to:
⎭⎬⎫
⎩⎨⎧
⎥⎦
⎤⎢⎣
⎡×=
⎭⎬⎫
⎩⎨⎧
− y
x
dd
1
15
354.1354.0
354.0354.1
105000,10
0
Solving the equations gives:
.1059.1.10414.0 21
21 indind yx
−− ×−=×=
The stress in an element is:
[ ]yxyx SdCdSdCdLE
2211 ++−−=σ
CIVL 7117 Finite Element Methods in Structural Mechanics Page 93
Element Node 1 Node 2 θ C S
1 1 2 90o 0 1 2 1 3 45o 0.707 0.707 3 1 4 0o 1 0
The stress in element 1 is:
[ ] psid y 965,3120
10301
6)1( =−
×=σ
The stress in element 2 is:
[ ] psidd yx 471,1)707.0()707.0(120
103011
6)2( =+
×−=σ
The stress in element 3 is:
[ ] psid x 035,1120
10301
6)3( −=−
×=σ
Let’s check equilibrium at node 1: 0)2)(035,1()707.0)(2)(471,1( 22 =−−=∑ inpsiinpsiFx
0000,10)707.0)(2)(471,1()2)(965,3( 22 =−+=∑ inpsiinpsiFy
CIVL 7117 Finite Element Methods in Structural Mechanics Page 94
Example 6 - Plane Truss Problem
Develop the element stiffness matrices and system equations for the plane truss below. Assume the stiffness of each element is constant. Use the number-ing scheme indicated. Solve the equations for the displacements and compute the member forces. All elements have a constant value of AE/L
45
P1
P2
1
3
2
The node numbering is given in the diagram above (Note that this is the optimum numbering configuration). Develop the element information:
Member Node 1 Node 2 Elemental Stiffness
θ
1 1 2 k 0 2 2 3 k 3π/4 3 1 3 k π/2
Compute the elemental stiffness matrix for each element. The general form of the matrix is:
⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢
⎣
⎡−
−−−
−−
−−
=
2
2
2
2
2
2
2
2
SCSS
CS
CSCCSC
SCSSCS
CSC
CSC
LAEk
CIVL 7117 Finite Element Methods in Structural Mechanics Page 95
For element 1:
For element 2:
For element 3:
Assemble the global stiffness matrix by superimposing the elemental global ma-trices.
y
x
y
x
dddd
kk
yxyxdddd
2
2
1
1
)1(
0000
0101
0000
0101
2211
⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢
⎣
⎡ −
−=
y
x
y
x
dddd
kk
yxyxdddd
3
3
2
2
)2(
1111
1111
1111
1111
2
3322
⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢
⎣
⎡
−−
−
−
−
−
−−
=
y
x
y
x
dddd
kk
yxyxdddd
3
3
1
1
)3(
1010
0000
1010
0000
3311
⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢
⎣
⎡−
−
=
y
x
y
x
y
x
dddddd
kK
yxyxyx dddddd
3
3
2
2
1
1
311120
111100
111100
111302
200020
000202
2
332211
⎥⎥⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢⎢⎢
⎣
⎡
−−
−
−
−
−
−
−−
−
−
−=
CIVL 7117 Finite Element Methods in Structural Mechanics Page 96
The unconstrained (no boundary conditions satisfied) equations are
⎪⎪⎪⎪
⎭
⎪⎪⎪⎪
⎬
⎫
⎪⎪⎪⎪
⎩
⎪⎪⎪⎪
⎨
⎧
−=
⎪⎪⎪⎪
⎭
⎪⎪⎪⎪
⎬
⎫
⎪⎪⎪⎪
⎩
⎪⎪⎪⎪
⎨
⎧
⎥⎥⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢⎢⎢
⎣
⎡
−−
−
−
−
−
−
−−
−
−
−
y
x
y
x
y
x
y
x
y
x
FFP
PFF
dddddd
k
3
3
2
1
1
1
3
3
2
2
1
1
311120
111100
111100
111302
200020
000202
2
The displacement at nodes 1 and 3 are zero in both directions. Applying these conditions to the system equations gives
⎭⎬⎫
⎩⎨⎧−
=⎭⎬⎫
⎩⎨⎧
⎥⎦
⎤⎢⎣
⎡ −− 2
1
2
2
11
13
2 PP
ddk
y
x
Solving this set of equations is fairly easy. The solution is
kPPd
kPPd yx
212
212
3−=
−=
Using the force-displacement relationship the force in each member may be computed. Member (element) 1:
011221
1 =−=⎟⎠⎞
⎜⎝⎛ −−= yx fPP
kPPkf
1 22 1 2 2 0x y
P Pf k P P fk−⎛ ⎞= = − =⎜ ⎟
⎝ ⎠
Member (element) 2:
1 2 1 22 2
31 1 22 2x
P P P Pf k Pk k
⎡ ⎤− −⎛ ⎞ ⎛ ⎞⎛ ⎞ ⎛ ⎞= + − = −⎢ ⎥⎜ ⎟ ⎜ ⎟⎜ ⎟ ⎜ ⎟⎝ ⎠ ⎝ ⎠⎝ ⎠ ⎝ ⎠⎣ ⎦
CIVL 7117 Finite Element Methods in Structural Mechanics Page 97
1 2 1 23 2
31 1 22 2x
P P P Pf k Pk k
⎡ ⎤− −⎛ ⎞ ⎛ ⎞⎛ ⎞ ⎛ ⎞= − + =⎢ ⎥⎜ ⎟ ⎜ ⎟⎜ ⎟ ⎜ ⎟⎝ ⎠ ⎝ ⎠⎝ ⎠ ⎝ ⎠⎣ ⎦
00 32 == yy ff
Member (element) 3:
0000
33
11
==
==
yx
yx
ffff
The solution to this simple problem can be readily checked by using simple static equilibrium equations.
Example 7 - Plane Truss Problem Consider the two bar truss shown below. Determine the displacement in the y di-rection of node 1 and the axial force in each element. Assume E = 210 GPa and A = 6 x 10 -4 m 2
CIVL 7117 Finite Element Methods in Structural Mechanics Page 98
The global elemental stiffness matrix for element 1 is:
8.054sin6.0
53cos )1()1( ==== θθ
⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢
⎣
⎡−−
−−
−−
−−
××=
−
64.048.064.048.0
48.036.048.036.0
64.048.0
64.048.0
48.036.0
48.036.0
5)106(10210 46
)1(k
Simplifying the above expression gives:
⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢
⎣
⎡−−
−−
−−
−−
=
64.048.064.048.0
48.036.048.036.0
64.048.0
64.048.0
48.036.0
48.036.0
200,25)1(
2211
k
yxyxdddd
The global elemental stiffness matrix for element 2 is:
1sin0cos )2()2( == θθ
⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢
⎣
⎡−−
−
××=
−
1010
0000
1010
0000
4)106)(10210( 46
)2(k
⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢
⎣
⎡−
−
=
25.1025.1
0
0000
25.1025.10
0000
200,25)2(
3311
k
yxyxdddd
CIVL 7117 Finite Element Methods in Structural Mechanics Page 99
The total global equations are:
⎪⎪⎪⎪
⎭
⎪⎪⎪⎪
⎬
⎫
⎪⎪⎪⎪
⎩
⎪⎪⎪⎪
⎨
⎧
⎥⎥⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢⎢⎢
⎣
⎡−−
−−−
−
−−
−−
=
⎪⎪⎪⎪
⎭
⎪⎪⎪⎪
⎬
⎫
⎪⎪⎪⎪
⎩
⎪⎪⎪⎪
⎨
⎧
y
x
y
x
y
x
y
x
y
x
y
x
dddddd
FFFFFF
3
3
2
2
1
1
3
3
2
2
1
1
25.1000
25.10
000000
00
64.048.064.048.0
00
48.036.048.036.0
25.10
64.048.089.148.0
00
48.036.048.036.0
200,25
The displacement boundary conditions are:
033221 ===== yxyxx ddddd δ
By applying the boundary conditions the force-displacement equations reduce to:
)89.148.0(200,25 1ydP += δ
Solving the equation gives:
δ25.0)101.2( 51 −×= − Pd y
By substituting P = 1000 kN and δ = -0.05 m in the above equation gives:
md y 0337.01 =
The local element forces for element 1 are:
⎪⎪⎭
⎪⎪⎬
⎫
⎪⎪⎩
⎪⎪⎨
⎧=
−=
⎥⎦
⎤⎢⎣
⎡⎥⎦
⎤⎢⎣
⎡ −−
=⎭⎬⎫
⎩⎨⎧
y
x
y
x
x
x
dd
dd
ff
2
2
1
1
2
1 0337.005.0
8.00
6.00
08.0
06.0
11
11
200,25
The element forces are:
CIVL 7117 Finite Element Methods in Structural Mechanics Page 100
kNfkNf xx 7.766.76 21 =−= The local element forces for element 2 are:
⎪⎪⎭
⎪⎪⎬
⎫
⎪⎪⎩
⎪⎪⎨
⎧=
−=
⎥⎦
⎤⎢⎣
⎡⎥⎦
⎤⎢⎣
⎡ −−
=⎭⎬⎫
⎩⎨⎧
y
x
y
x
x
x
dd
dd
ff
3
3
1
1
3
1 0337.005.0
10
00
01
00
11
11
500,31
The element forces are:
kNfkNf xx 10611061 31 −==
Transformation Matrix and Stiffness Matrix for a Bar in Three-Dimensional Space
Let’s derive the transformation matrix for the stiffness matrix for a bar element in three-dimensional space as shown below:
The coordinates at node 1 are x1, y1, and z1, and the coordinates of node 2 are x2, y2, and z2. Also, let θx, θy, and θz be the angles measured from the global x, y,
CIVL 7117 Finite Element Methods in Structural Mechanics Page 101
and z axes, respectively, to the local x axis. The three-dimensional vector repre-senting the bar element is gives as:
kdjdidddd zyxzyxˆˆˆˆˆˆ ++=++= kjid
where i, j, and k are unit vectors in the x, y, and z directions, and i , j , and k are unit vector in the x , y , and z direction. Taking the dot product of the above
equation with i gives:
xzyx dikdijdiid ˆ)ˆ()ˆ()ˆ( =•+•+•
By the definition of the dot product we get:
2 1 2 1 2 1ˆ ˆ ˆx y z
x x y y z zi i C j i C k i CL L L− − −
• = = • = = • = =
where
212
212
212 )()()( zzyyxxL −+−+−=
and
zzyyxx CCC θθθ coscoscos ===
where Cx, Cy, and Cz are projections of i on to i, j, and k, respectively.
zzyyxxx dCdCdCd ++=ˆ
CIVL 7117 Finite Element Methods in Structural Mechanics Page 102
The transformation between local and global displacements is:
dTd *ˆ =
⎪⎪⎪
⎭
⎪⎪⎪
⎬
⎫
⎪⎪⎪
⎩
⎪⎪⎪
⎨
⎧
⎥⎦
⎤⎢⎣
⎡=
⎭⎬⎫
⎩⎨⎧
z
y
x
z
y
x
zyx
zyx
x
x
dddddd
CCCCCC
dd
2
2
2
1
1
1
2
1000
000ˆˆ
where
⎥⎦
⎤⎢⎣
⎡=
zyx
zyx
CCCCCC
T000
000*
The transformation from the local to the global stiffness matrix is:
TkTk T ˆ=
⎥⎦
⎤⎢⎣
⎡⎥⎦
⎤⎢⎣
⎡ −−
⎥⎥⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢⎢⎢
⎣
⎡
=zyx
zyx
z
y
x
z
y
x
CCCCCC
LAE
CCC
CCC
k000
00011
110
00
000
CIVL 7117 Finite Element Methods in Structural Mechanics Page 103
Simplifying the above expression gives:
⎥⎥⎥⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢⎢⎢⎢
⎣
⎡
−−−
−−
−
−−−
−−−
−−
−
−−−
=
2
2
2
2
2
2
2
2
2
2
2
2
z
zy
zx
z
zy
zx
zy
y
yx
zy
y
yx
zx
yx
x
zx
yx
x
z
zy
zx
z
zy
zx
zy
y
yx
zy
y
yx
zx
yx
x
zx
yx
x
CCCCCCCCCC
CCCCCCC
CCC
CCCC
CCCCC
C
CCCCC
CCCCC
CCCCC
CCCCC
CCCC
CCCCC
C
LAEk
The global stiffness matrix can be written in a more convenient form as:
⎥⎦
⎤⎢⎣
⎡ −−
=λλ
λλ
LAEk
where
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡
=2
2
2
z
zy
zx
zy
y
yx
zx
yx
x
CCCCC
CCC
CC
CCCC
Cλ
Example 8 - Space Truss
Consider the space truss shown below. The modulus of elasticity, E, is 1.2 x 106 psi for all elements. Node 1 is constrained from movement in the y direction.
To simplify the stiffness matrices for the three elements, we will express each element in the following form:
⎥⎦
⎤⎢⎣
⎡ −−
=λλ
λλ
LAEk
CIVL 7117 Finite Element Methods in Structural Mechanics Page 104
Consider element 3:
)3(14
)3(14
)3(14
LzzC
LyyC
LxxC zyx
−=
−=
−=
where
214
214
214
)3( )()()( zzyyxxL −+−+−=
.5.86)48()72( 22)3( inL =−+−= then
550.05.86
480833.05.86
72−=
−==−=
−= zyx CCC
CIVL 7117 Finite Element Methods in Structural Mechanics Page 105
Therefore, the matrix λ is:
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡=
30.0046.0
000
46.0069.0
λ
The global element stiffness matrix is:
⎥⎦
⎤⎢⎣
⎡ −−
×=
λλ
λλ
5.86102.1)(187.0( )6
444111
k
zyxzyxdddddd
Consider element 1:
.5.80)36()72( 22)1( inL =+−= then
045.05.80
3689.05.80
72===−=
−= zyx CCC
Therefore, the matrix λ is:
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡ −−=
000
020.040.0
040.0
79.0λ
The global element stiffness matrix is:
⎥⎦
⎤⎢⎣
⎡ −−
×=
λλ
λλ
5.80102.1)(302.0( )6
222111
k
zyxzyxdddddd
CIVL 7117 Finite Element Methods in Structural Mechanics Page 106
Consider element 2:
.108)72()36()72( 222)2( inL =++−= then
667.01087233.0
10836667.0
10872
====−=−
= zyx CCC
Therefore, the matrix λ is:
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡ −−
−−=
45.045.045.0
45.011.022.0
45.022.0
45.0λ
The global element stiffness matrix is:
⎥⎦
⎤⎢⎣
⎡ −−
×=
λλ
λλ
108102.1)(729.0( )6
333111
k
zyxzyxdddddd
The boundary conditions are:
000 444333222 ========= zyxzyxzyx ddddddddd
01 =yd
Canceling the rows and the columns associated with the boundary conditions re-duces the global stiffness matrix to:
⎥⎦
⎤⎢⎣
⎡ −−
=44502450
24509000
11
K
zxdd
CIVL 7117 Finite Element Methods in Structural Mechanics Page 107
The global force-displacement equations are:
⎭⎬⎫
⎩⎨⎧
−=
⎭⎬⎫
⎩⎨⎧
⎥⎦
⎤⎢⎣
⎡ −− 1000
044502450
24509000
1
1
z
x
dd
Solving the equation gives:
.264.0.072.0 11 indind zx −=−= The local forces in an element are:
ˆ
ˆ
ix
iy
ix x y z x y z iz
x y z x y z jxjx
jy
jz
dd
f C C C C C C dAEC C C C C C dLf
dd
⎧ ⎫⎪ ⎪⎪ ⎪
⎧ ⎫ ⎪ ⎪− − −⎡ ⎤⎪ ⎪ ⎪ ⎪=⎨ ⎬ ⎨ ⎬⎢ ⎥− − −⎣ ⎦⎪ ⎪ ⎪ ⎪⎩ ⎭⎪ ⎪⎪ ⎪⎪ ⎪⎩ ⎭
The stress in an element is:
[ ]
⎪⎪⎪⎪
⎭
⎪⎪⎪⎪
⎬
⎫
⎪⎪⎪⎪
⎩
⎪⎪⎪⎪
⎨
⎧
−−−=
jz
jy
jx
iz
iy
ix
zyxzyx
dddddd
CCCCCCLEσ
CIVL 7117 Finite Element Methods in Structural Mechanics Page 108
The stress in element 1 is:
[ ]
⎪⎪⎪
⎭
⎪⎪⎪
⎬
⎫
⎪⎪⎪
⎩
⎪⎪⎪
⎨
⎧
−
−
−×
=
000264.00072.0
045.089.0045.089.05.80102.1 6
)1(σ
psi955)1( −=σ
The stress in element 2 is:
[ ]
⎪⎪⎪
⎭
⎪⎪⎪
⎬
⎫
⎪⎪⎪
⎩
⎪⎪⎪
⎨
⎧
−
−
−−−×
=
000264.00072.0
667.033.0667.0667.033.0667.0108
102.1 6)2(σ
psi423,1)2( =σ The stress in element 3 is:
[ ]
⎪⎪⎪
⎭
⎪⎪⎪
⎬
⎫
⎪⎪⎪
⎩
⎪⎪⎪
⎨
⎧
−
−
−−×
=
000264.00072.0
55.0083.055.0083.05.86102.1 6
)3(σ
psi843,2)3( =σ
CIVL 7117 Finite Element Methods in Structural Mechanics Page 109
Inclined, or Skewed, Supports
If a support is inclined, or skewed, at some angle α for the global x axis, as shown below, the boundary conditions on the displacements are not in the global x-y directions but in the x’-y’ directions.
We must transform the local boundary condition of d’3y = 0 (in local coordinates) into the global x-y system. Therefore, the relationship between of the compo-nents of the displacement in the local and the global coordinate systems at node 3 is:
⎭⎬⎫
⎩⎨⎧
⎥⎦
⎤⎢⎣
⎡−
=⎭⎬⎫
⎩⎨⎧
y
x
y
x
dd
dd
3
3
3
3
cossin
sincos
''
αα
αα
We can rewrite the above expression as:
[ ] ⎥⎦
⎤⎢⎣
⎡−
==αα
αα
cossin
sincos
][' 3333 tdtd
CIVL 7117 Finite Element Methods in Structural Mechanics Page 110
We can apply this sort of transformation to the entire displacement vector as:
'][][' 11 dTdordTd T== where the matrix [T1]T is:
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡=
][]0[]0[
]0[][]0[
]0[]0[][
][
3
1
tI
IT T
Both the identity matrix [I] and the matrix [t3] are 2 x 2 matrices.
The force vector can be transformed by using the same transformation.
fTf ][' 1= In global coordinates, the force-displacement equations are:
dKf ][= Applying the skewed support transformation to both sides of the equation gives:
dKTfT ]][[][ 11 = By using the relationship between the local and the global displacements, the force-displacement equations become:
1 1' [ ][ ][ ] 'Tf T K T d=
CIVL 7117 Finite Element Methods in Structural Mechanics Page 111
Therefore the global equations become:
⎪⎪⎪⎪
⎭
⎪⎪⎪⎪
⎬
⎫
⎪⎪⎪⎪
⎩
⎪⎪⎪⎪
⎨
⎧
=
⎪⎪⎪⎪
⎭
⎪⎪⎪⎪
⎬
⎫
⎪⎪⎪⎪
⎩
⎪⎪⎪⎪
⎨
⎧
y
x
y
x
y
x
T
y
x
y
x
y
x
dddddd
TKT
FFFFFF
3
3
2
2
1
1
11
3
3
2
2
1
1
''
]][][[
''
Example 9 - Space Truss
Consider the plane truss shown below. Assume E = 210 GPa, A = 6 x 10-4 m2 for element 1 and 2, and A = 6 2 x 10-4 m2 for element 3.
Determine the stiffness matrix for each element. The global elemental stiffness matrix for element 1 is:
1sin0cos )1()1( == θθ
CIVL 7117 Finite Element Methods in Structural Mechanics Page 112
1 1 2 2
6 4(1)
0 0 0 00 1 0 1(210 10 )(6 10 )0 0 0 010 1 0 1
x y x yd d d d
k−
⎡ ⎤⎢ ⎥−× × ⎢ ⎥=⎢ ⎥⎢ ⎥−⎣ ⎦
The global elemental stiffness matrix for element 2 is:
0sin1cos )2()2( == θθ
⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢
⎣
⎡ −
−××
=−
0000
0101
0000
01
01
1)106)(10210( 46
)2(
3322
k
yxyx dddd
The global elemental stiffness matrix for element 3 is:
22sin
22cos )3()3( == θθ
⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢
⎣
⎡−−
−−
−−
−−
××=
−
1111
1111
1111
11
11
22)1026)(10210( 46
)3(
3311
k
yxyxdddd
CIVL 7117 Finite Element Methods in Structural Mechanics Page 113
Using the direct stiffness method, the global stiffness matrix is:
⎥⎥⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢⎢⎢
⎣
⎡−−
−−−
−
−−−
−
−−
×=
5.05.0005.05.0
5.05.1015.05.0
001010
010100
5.05.0105.15.0
5.05.0005.05.0
10260,1 5
mNK
We must transform the global displacements into local coordinates. Therefore the transformation [T1] is:
1
3 2 22 2
2 22 2
1 0 0 0 0 00 1 0 0 0 0
[ ] [0] [0]0 0 1 0 0 0
[ ] [0] [ ] [0]0 0 0 1 0 0
[0] [0] [ ]0 0 0 00 0 0 0
IT I
t
−
⎡ ⎤⎢ ⎥⎢ ⎥⎡ ⎤⎢ ⎥⎢ ⎥= = ⎢ ⎥⎢ ⎥⎢ ⎥⎢ ⎥⎣ ⎦ ⎢ ⎥⎢ ⎥⎣ ⎦
The first step in the matrix transformation to find the product of [T1][K].
⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢
⎣
⎡
×=
−−
−
−−−
−
−−−
−0
707.0005.05.0
70.0414.1
015.05.0
001010
707.0707.0
0100
0707.0
105.15.0
0707.0
005.05.0
5101260]][[ mNKT1
CIVL 7117 Finite Element Methods in Structural Mechanics Page 114
The next step in the matrix transformation to find the product of [T1][K][T1]T.
⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢
⎣
⎡
×=
−−
−−−
−
−−−
−5.05.00
707.000
5.05.10
707.0707.0707.0
001010
707.0707.0
0100
0707.0
105.15.0
0707.0
005.05.0
5101260]][][[ mNT
11 TKT
The displacement boundary conditions are:
0'3211 ==== yyyx dddd
By applying the boundary conditions the global force-displacement equations are:
⎭⎬⎫
⎩⎨⎧
==
=⎭⎬⎫
⎩⎨⎧
⎥⎦
⎤⎢⎣
⎡ −−
×0'000,1
'5.1707.0
707.01
1012603
2
3
25
x
x
x
x
FkNF
dd
Solving the equation gives:
mmdmmd xx 61.5'91.11 32 == The global nodal forces are calculated as:
⎪⎪
⎭
⎪⎪
⎬
⎫
⎪⎪
⎩
⎪⎪
⎨
⎧
⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢
⎣
⎡
−=
⎪⎪⎪
⎭
⎪⎪⎪
⎬
⎫
⎪⎪⎪
⎩
⎪⎪⎪
⎨
⎧
−−
−−−
−
−−−061.5091.11
00
5.05.00
707.000
5.05.10
707.0707.0707.0
001010
707.0707.0
0100
0707.0
105.15.0
0707.0
005.05.0
1
3
3
2
2
1
1
''
B
FFFFFF
y
x
y
x
y
x
mmNB 2
1 10260,1 ×=
CIVL 7117 Finite Element Methods in Structural Mechanics Page 115
Therefore:
kNFkNF yx 500500 11 −=−=
kNFF yy 707'0 32 ==
Potential Energy Approach to Derive Bar Element Equations Let’s derive the equations for a bar element using the principle of minimum
potential energy. The total potential energy, πp, is defined as the sum of the inter-nal strain energy U and the potential energy of the external forces Ω:
Ω+= Upπ
The differential internal work (strain energy) dU in a one-dimensional bar element is:
xx dxzydU εσ ))()(( ∆∆∆=
If we let the volume of the element approach zero, then:
dVddU xx εσ= Summing the differential energy over the whole bar gives:
∫ ∫⎭⎬⎫
⎩⎨⎧=
Vxx dVdU
xε
εσ0
CIVL 7117 Finite Element Methods in Structural Mechanics Page 116
For a linear-elastic material (Hooke’s law) as shown below:
The internal strain energy statement becomes
∫=V
xx dVU εσ21
The potential energy of the external forces is:
∑∫∫=
−−−=ΩM
1iixixxb dSdV dfuTuX
SV
ˆˆˆˆˆˆ
where bX is the body force (force per unit volume), xT is the traction (force per
unit area), and ixf is the nodal concentrated force. All of these forces are consid-ered to act in the local x direction.
Apply the following steps when using the principle of minimum potential en-ergy to derive the finite element equations.
1. Formulate an expression for the total potential energy 2. Assume a displacement pattern 3. Obtain a set of simultaneous equations minimizing the total potential en-
ergy with respect to the displacement parameters.
xx Eεσ =
CIVL 7117 Finite Element Methods in Structural Mechanics Page 117
Consider the following bar element, as shown below:
We can approximate the axial displacement as:
[ ]⎭⎬⎫
⎩⎨⎧
=x
x
ddNNu
2
121 ˆ
ˆˆ
where N1 and N2 are the interpolation functions gives as:
LxN
LxN
ˆˆ1 21 =−=
Using the stress-strain relationships, the axial strain is:
⎭⎬⎫
⎩⎨⎧
⎥⎦⎤
⎢⎣⎡==
x
xx d
dxd
dNxd
dNxdud
2
121
ˆˆ
ˆˆˆˆ
ε
where N1 and N2 are the interpolation functions gives as:
∫∫
∫
−−
−−=
SV
xx1x1x
L
oxxp
uTuX
dfdfxd2A
dSdV xb ˆˆˆˆ
ˆˆˆˆˆ 22εσπ
CIVL 7117 Finite Element Methods in Structural Mechanics Page 118
][ˆˆ11
2
1 dBdd
LL x
x
xx =
⎭⎬⎫
⎩⎨⎧
⎥⎦⎤
⎢⎣⎡−= εε
where [B] are:
⎥⎦⎤
⎢⎣⎡−=
LLB 11
The axial stress-strain relationship is:
xx D εσ ][= where [D] = [E] for the one-dimensional stress-strain relationship and E is the modulus of elasticity. Therefore, stress can be related to nodal displacements as:
dBDxˆ]][[=σ
The total potential energy expressed in matrix form is:
∫∫∫ −−−=S
T
V
TTL
oxxp TuXuPdxd
2A dSdV xb
ˆˆˆˆˆˆεσπ
where P represented the concentrated nodal loads. If we substitute the rela-tionship between u and d into the energy equations we get:
[ ] [ ] [ ]
[ ] [ ] ∫∫
∫
−−
−=
S
TT
V
TT
TL
o
TTT
p
TNdXNd
PdxddBDBd2A
dSdV xbˆˆˆˆ
ˆˆˆˆπ
In the above expression for potential energy πp is a function of the d , that is: πp = πp )ˆ,ˆ( 21 xx dd . However, [B] and [D] and the nodal displacements d are not a function of x . Integration the energy expression with respect to x gives:
CIVL 7117 Finite Element Methods in Structural Mechanics Page 119
fddBDBdAL TTTT
pˆˆˆ][][][ˆ
2−=π
where
∫∫ ++=SV
dSXNdVXNPf bT
bT ][][ˆ
We can define the surface tractions and body-force matrices as:
∫=S
x dSTNf ˆ][ˆ Ts
∫=V
dVXNf bT
bˆ][ˆ
Minimization of πp with respect to each nodal displacement requires that:
0ˆ0ˆ21
=∂∂
=∂∂
x
p
x
p
ddππ
For convenience, let’s define the following
dBDBdU TTT ˆ][][][ˆ* = or
[ ]⎭⎬⎫
⎩⎨⎧
⎥⎦⎤
⎢⎣⎡−
⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧−=
x
xxx d
dLL
E
L
LddU2
121
*
ˆˆ11][1
1ˆˆ
Simplifying the above expression gives:
( )2221
212
* ˆˆˆ2ˆxxxx dddd
LEU +−=
CIVL 7117 Finite Element Methods in Structural Mechanics Page 120
The loading on a bar element is given as:
xxxx
Tfdfdfd 2211ˆˆˆˆˆˆ +=
Therefore, the minimum potential energy is:
( ) 0ˆˆ2ˆ22ˆ 121
1
=−−=∂∂
xxxx
p fddL
AEdπ
( ) 0ˆˆ2ˆ22ˆ 221
2
=−+−=∂∂
xxxx
p fddL
AEdπ
The above equations can be written in matrix form as:
( ) 0ˆˆ
ˆˆ
11
11
ˆ2
1
2
1 =⎟⎟⎠
⎞⎜⎜⎝
⎛−⎟⎟
⎠
⎞⎜⎜⎝
⎛⎥⎦
⎤⎢⎣
⎡ −−
=∂∂
x
x
x
xp
ff
dd
LAE
dπ
The stiffness matrix for a bar element is:
[ ] ⎥⎦
⎤⎢⎣
⎡ −−
=11
11ˆ
LAEk
This form of the stiffness matrix obtained from the principle of minimum potential energy is identical to the stiffness matrix derived from the equilibrium equations.
Example 10 - Bar Problem
Consider the bar shown below:
CIVL 7117 Finite Element Methods in Structural Mechanics Page 121
The energy equivalent nodal forces due to the distributed load are:
∫=S
x dSTNf ˆ][ˆ Ts
∫⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧ −=
⎭⎬⎫
⎩⎨⎧
=L
xdxC
Lx
Lx1
fff
0
ˆˆˆ
ˆ
ˆˆˆ2x
1x
⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧
=⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧ −=
⎭⎬⎫
⎩⎨⎧
3CL6
CL
3LxC
3LxC
2xC
ff
2
2L
3
32
0
ˆ
ˆˆ
ˆˆ
2x
1x
The total load is the area under the distributed load curve, or:
2))((
21 2CLCLLF ==
The equivalent nodal forces for a linearly varying load are:
loadtotaltheof31
31 ==Ff x
loadtotaltheof32
32
2 ==Ff x
CIVL 7117 Finite Element Methods in Structural Mechanics Page 122
Example 11 - Bar Problem
Consider the axially loaded bar shown below. Determine the axial displace-ment and axial stress. Let E = 30 x 10 6 psi, A = 2 in.2, and L = 60 in. Use (a) one and (b) two elements in the finite element solutions.
The one-element solution:
The distributed load can be converted into equivalent nodal forces using:
∫=S
x dSTNF ˆ][ T
The above expression can be expressed as:
∫ −⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧ −=
⎭⎬⎫
⎩⎨⎧
=L
2x
1x dx10x
Lx
Lx1
FF
F0
or
CIVL 7117 Finite Element Methods in Structural Mechanics Page 123
⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧
−
−=
⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧
−
+−=
⎭⎬⎫
⎩⎨⎧
310L
610L
310L
310L
210L
FF
2
2
2
22
2x
1x
or
⎭⎬⎫
⎩⎨⎧
−−
=⎭⎬⎫
⎩⎨⎧
lb12,000lb6,000
2x
1x
FF
The stiffness matrix for element 1 is:
⎥⎦
⎤⎢⎣
⎡ −−
=11
11
106)1(k
The element equations are:
⎭⎬⎫
⎩⎨⎧
−−
=⎭⎬⎫
⎩⎨⎧
⎥⎦
⎤⎢⎣
⎡ −− 000,12
000,601
111
102
16
x
x
Rd
Solving the equation gives:
.006.01 ind x −= The axial stress-strain relationship is:
][ xx D εσ =
dBEx ][ =σ
⎟⎠⎞
⎜⎝⎛ −
=⎭⎬⎫
⎩⎨⎧
⎥⎦⎤
⎢⎣⎡−=
LddE
dd
LLE xx
x
x 12
2
111
)(000,360
006.001030 6 Tpsi=⎟⎠⎞
⎜⎝⎛ +
×=
CIVL 7117 Finite Element Methods in Structural Mechanics Page 124
The two-element solution:
The distributed load can be converted into equivalent nodal forces. For element 1, the total force of the triangular-shaped distributed load is:
lbinlbin 500,4.)/300.)(30(21
−=
Based on equations developed for the equivalent nodal force of a triangular dis-tributed load, develop in the one-element problem, the nodal forces are:
⎭⎬⎫
⎩⎨⎧
−−
=⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧
−
−=
⎭⎬⎫
⎩⎨⎧
lblb
ff
x
x
000,3500,1
)500,4(32
)500,4(31
)1(2
)1(1
For element 2, the applied force is in two parts: a triangular-shaped distributed load and a uniform load. The uniform load is:
lbinlbin 000,9.)/300.)(30( −= The nodal forces for element 2 are:
⎭⎬⎫
⎩⎨⎧−−
=
⎪⎪⎭
⎪⎪⎬
⎫
⎪⎪⎩
⎪⎪⎨
⎧
⎥⎦⎤
⎢⎣⎡ +−
⎥⎦⎤
⎢⎣⎡ +−
=⎭⎬⎫
⎩⎨⎧
lblb
ff
x
x
500,7000,6
)500,4(32)000,9(
21
)500,4(31)000,9(
21
)2(3
)2(2
CIVL 7117 Finite Element Methods in Structural Mechanics Page 125
The final nodal force vector is:
⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧
−−−
=⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧
500,7000,9500,1
33
2
1
xx
x
x
RFFF
The element stiffness matrices are:
⎥⎦
⎤⎢⎣
⎡ −−
==11
112)2()1(
32
21
LAEkk
The assembled global stiffness matrix is:
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡−
−
−−×=
110
121
011
102 6K
The assembled global force-displacement equations are:
⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧
−−−
=⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡−
−
−−×
500,7000,9500,1
0110
121
011
102
3
2
1
6
x
x
x
Rdd
After the eliminating the row and column associated with d3x, we get:
⎭⎬⎫
⎩⎨⎧−−
=⎭⎬⎫
⎩⎨⎧
⎥⎦
⎤⎢⎣
⎡ −−
×000,9500,1
21
11
1022
16
x
x
dd
Solving the equation gives:
.00525.0.006.0 21 indind xx −=−=
CIVL 7117 Finite Element Methods in Structural Mechanics Page 126
The axial stress-strain relationship is:
)(75000525.0
006.0301
30111
2
1)1( TpsiEdd
LLE
x
xx =
⎭⎬⎫
⎩⎨⎧
−−
⎥⎦⎤
⎢⎣⎡−=
⎭⎬⎫
⎩⎨⎧
⎥⎦⎤
⎢⎣⎡−=σ
)(250,50
00525.0301
301)2( TpsiEx =
⎭⎬⎫
⎩⎨⎧−
⎥⎦⎤
⎢⎣⎡−=σ
Comparison of Finite Element Solution to Exact Solution
In order to be able to judge the accuracy of our finite element models, we will develop an exact solution for the bar element problem. The exact solution for the displacement may be obtained by:
∫=x
dxxPAE 0
)(1δ
where the force P is shown on the following free-body diagram.
Therefore:
1
32
3551 CAExdxx
AE
x
o+== ∫δ
Applying the boundary conditions:
AELCC
AExL
35
350)(
3
11
3
−=⇒+==δ
[ ] 25)10(21)( xxxxP ==
CIVL 7117 Finite Element Methods in Structural Mechanics Page 127
The exact solution for axial displacement is:
( )33
35)( LxAE
L −=δ
The exact solution for axial stress is:
Ax
AxPx
25)()( ==σ
A plot of the exact solution for displacement as compared to several different finite element solutions is shown below.
A plot of the exact solution for axial stress as compared to several different fi-
nite element solutions is shown below.
CIVL 7117 Finite Element Methods in Structural Mechanics Page 128
Galerkin’s Residual Method and Its Application to a One-Dimensional Bar
There are a number of weighted residual methods. However, the Galerkin’s method is more well-known and will be the only weighted residual method dis-cussed in this course.
In weighted residual methods, a trial or approximate function is chosen to ap-proximate the independent variable (in our case, displacement) in a problem de-fined by a differential equation. The trial function will not, in general, satisfy the governing differential equation. Therefore, the substitution of the trial function in the differential equation will create a residual over the entire domain of the prob-lem.
minimum=∫V
RdV
CIVL 7117 Finite Element Methods in Structural Mechanics Page 129
In the residual methods, we require that a weighted value of the residual be a minimum over the entire domain of the problem. The weighting function allows the weighted integral of the residuals to go to zero.
0=∫V
dVWR
Using Galerkin’s weighted residual method, we require the weighting functions to be the interpolation functions or the shape functions, Ni. Therefore:
nidVRNV
i ,,2,10 ==∫
Example 12 - Bar Element Formulation
Let’s derive the bar element formulation using Galerkin’s method. The govern-ing differential equation is:
0ˆˆ
ˆ=⎟
⎠⎞
⎜⎝⎛
xdudAE
xdd
Applying Galerkin’s method we get:
nixdNxdudAE
xddL
i ,,2,10ˆˆˆ
ˆ0
==⎟⎠⎞
⎜⎝⎛∫
We now apply integration by parts using the following general formula:
∫∫ −= vduuvudv
where u and v are variable in the general equation. If we assume the following:
xdxd
dNduNu ii ˆ
ˆ==
xdudAEvxd
xdudAE
xdddv
ˆˆˆ
ˆˆ
ˆ=⎟
⎠⎞
⎜⎝⎛=
CIVL 7117 Finite Element Methods in Structural Mechanics Page 130
then integration by parts gives:
0ˆˆˆ
ˆˆˆ
00
=−⎟⎠⎞
⎜⎝⎛ ∫
Li
L
i xdxd
dNxdudAE
xdudAEN
Recall that:
xx dxd
dNdxd
dNxdud
22
11 ˆ
ˆˆ
ˆˆˆ
+=
or
⎭⎬⎫
⎩⎨⎧
⎥⎦⎤
⎢⎣⎡−=
x
x
dd
LLxdud
2
1
ˆˆ11
ˆˆ
Our original weighted residual expression, with the approximation for u be-comes:
L
o
i
x
xL
i
xdudAEN
ddxd
LLxddNAE ⎟
⎠⎞
⎜⎝⎛=
⎭⎬⎫
⎩⎨⎧
⎥⎦⎤
⎢⎣⎡−∫ ˆ
ˆˆˆ
ˆ11ˆ
2
1
0
Substituting N1 for the weighting function Ni gives:
L
x
xL
xdudAEN
ddxd
LLxddNAE
01
2
1
0
1
ˆˆ
ˆˆ
ˆ11ˆ ⎟
⎠⎞
⎜⎝⎛=
⎭⎬⎫
⎩⎨⎧
⎥⎦⎤
⎢⎣⎡−∫
01
02
1
0 ˆˆˆ
ˆˆ
ˆˆ
ˆ111==
==⎭⎬⎫
⎩⎨⎧
⎥⎦⎤
⎢⎣⎡−⎥⎦
⎤⎢⎣⎡−∫
x
x
xx
xL
xdudAEf
xdudAE
ddxd
LLLAE
( ) xxx fddL
AE121ˆˆˆ =−
CIVL 7117 Finite Element Methods in Structural Mechanics Page 131
Substituting N2 for the weighting function Ni gives:
L
ox
xL
xdudAEN
ddxd
LLxddNAE ⎟
⎠⎞
⎜⎝⎛=
⎭⎬⎫
⎩⎨⎧
⎥⎦⎤
⎢⎣⎡−∫ ˆ
ˆˆˆ
ˆ11ˆ 2
2
1
0
2
Lx
x
Lxx
xL
xdudAEf
xdudAE
ddxd
LLLAE
==
==⎭⎬⎫
⎩⎨⎧
⎥⎦⎤
⎢⎣⎡−⎥⎦
⎤⎢⎣⎡
∫ ˆˆˆ
ˆˆ
ˆˆ
ˆ1112
2
1
0
( ) xxx fddL
AE212ˆˆˆ =−
In matrix form the equations may be written as:
⎟⎟⎠
⎞⎜⎜⎝
⎛=⎟⎟
⎠
⎞⎜⎜⎝
⎛⎥⎦
⎤⎢⎣
⎡ −−
x
x
x
x
ff
dd
LAE
2
1
2
1
ˆˆ
ˆˆ
11
11
Problems:
6. Verify the global stiffness matrix for a three-dimensional bar. Hint: First, expand T* to a 6 x 6 square matrix, then expand k to 6 x 6 square matrix by adding the appropriate rows and columns of zeros, and finally, perform the matrix triple product k = TT k T.
7. Do problems 3.3, 3.7, 3.12, 3.16, 3.18, 3.22, 3.36, 3.43, 3.47, 3.49, and 3.55 on pages 120 - 136 in your textbook “A First Course in the Finite Ele-ment Method” by D. Logan.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 132
Symmetry and Bandwidth
Introduction
In this chapter, we will introduce the concepts of symmetry to reduce the size of a problem and of banded-symmetric matrices and bandwidth. In many in-stances, we can use symmetry to facilitate the solution of a problem. Symmetry means correspondence in size, shape, and position of loads; material properties; and boundary conditions that are mirrored about a dividing line or plane. Use of symmetry allows us to consider a reduced problem instead of the actual problem. Thus, the order of the total stiffness matrix and total set of stiffness equations can be reduced.
Example 1 - Truss Symmetry Problem
Solve the plane truss problem shown below. The truss is composed of eight elements and five nodes.
A vertical load of 2P is applied at node 4. Nodes 1 and 5 are pin supports. Bar elements 1, 2, 7, and 8 have axial stiffnesses of 2 AE and bars 3, 4, 5, and 6 have axial stiffnesses of AE.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 133
In this problem, we will use a plane of symmetry. The vertical plane perpen-dicular to the plane truss passing through nodes 2, 4, and 3 is the plane of sym-metry because identical geometry, material, loading, and boundary conditions occur at the corresponding locations on opposite sides of this plane. For loads such as 2P, occurring in the plane of symmetry, one-half of the total load must be applied to the reduced structure. For elements occurring in the plane of symme-try, one-half of the cross-sectional area must be used in the reduced structure.
The truss information may be summarized in the following table.
Element θ C S C2 S2 CS 1 45° 0.707 0.707 0.5 0.5 0.5
2 315° 0.707 -0.707 0.5 0.5 -0.5
3 0° 1 0 1 0 0
4 270° 0 -1 0 1 0
5 90° 0 1 0 1 0
The global elemental stiffness matrix for element 1 is:
1 1 2 2
(1)
1 1 1 11 1 1 11 1 1 121 1 1 1
x y x yd d d d
AEkL
− −⎡ ⎤⎢ ⎥− −⎢ ⎥=⎢ ⎥− −⎢ ⎥− −⎣ ⎦
The global elemental stiffness matrix for element 2 is:
1 1 3 3
(2)
1 1 1 11 1 1 11 1 1 121 1 1 1
x y x yd d d d
AEkL
− −⎡ ⎤⎢ ⎥− −⎢ ⎥=⎢ ⎥− −⎢ ⎥− −⎣ ⎦
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 134
The global elemental stiffness matrix for element 3 is:
1 1 4 4
(3)
1 0 1 00 0 0 01 0 1 00 0 0 0
x y x yd d d d
AEkL
−⎡ ⎤⎢ ⎥−⎢ ⎥=⎢ ⎥⎢ ⎥⎣ ⎦
The global elemental stiffness matrix for element 4 is:
2 2 4 4
(4)
0 0 0 00 1 0 10 0 0 020 1 0 1
x y x yd d d d
AEkL
⎡ ⎤⎢ ⎥−⎢ ⎥=⎢ ⎥⎢ ⎥−⎣ ⎦
The global elemental stiffness matrix for element 5 is:
3 3 4 4
(5)
0 0 0 00 1 0 10 0 0 020 1 0 1
x y x yd d d d
AEkL
⎡ ⎤⎢ ⎥−⎢ ⎥=⎢ ⎥⎢ ⎥−⎣ ⎦
Since elements 4 and 5 lie in the plane of symmetry, one half of their original ar-eas have been used in developing the stiffness matrices. The displacement boundary conditions are:
1 1 2 3 4 0x y x x xd d d d d= = = = =
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 135
By applying the boundary conditions the force-displacement equations reduce to:
Solution by Partitioning the Matrix
We can solve the above equations by separating the matrices in submatrices (indicated by the dashed lines). Consider a general set of equations shown be-low:
or
0212111 =+ dKdK
FdKdK =+ 222121 Solving the first equation for d1 gives:
2121
111 dKKd −−= Substituting the above equation in the second matrix equation gives:
( ) FdKdKKK =+− −
2222121
1121 Simplifying this expression gives:
( ) FdKKKK =− −
212
1
112122 or
Fdkc =2 where
2 2
3 3
4 4
2 0 1 00 2 1 0
21 1 2
x x
x x
x x
F dAEF d
LF d P
−⎧ ⎫ ⎡ ⎤ ⎧ ⎫ ⎧ ⎫⎪ ⎪ ⎪ ⎪ ⎪ ⎪⎢ ⎥= − =⎨ ⎬ ⎨ ⎬ ⎨ ⎬⎢ ⎥⎪ ⎪ ⎪ ⎪ ⎪ ⎪⎢ ⎥− −⎩ ⎭ ⎣ ⎦ ⎩ ⎭ ⎩ ⎭
11 12 1
21 22 2
0K K dK K d F
⎡ ⎤ ⎧ ⎫ ⎧ ⎫=⎨ ⎬ ⎨ ⎬⎢ ⎥
⎩ ⎭⎣ ⎦ ⎩ ⎭
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 136
121
112122 KKKKkc−−=
Therefore, the displacements d2 are:
( ) Fkd c1
2−=
If we apply this solution technique to our example global stiffness equations we get:
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡
⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧
−
−⎥⎦
⎤⎢⎣
⎡⎥⎦⎤
⎢⎣⎡ −−−=
−
2121
10
01
21
21]1[
1
LAEkc
Simplifying:
⎥⎦⎤
⎢⎣⎡=
⎭⎬⎫
⎩⎨⎧
⎥⎦⎤
⎢⎣⎡−=
21
21]1[
LAE
LAEkc
Therefore:
( )AE
Lkc
21 =−
Therefore, the displacements d2 are:
AEPLdd y
242 −==
The remaining displacements can be found by substituting the result for d4y in the global force-displacement equations.
⎥⎦⎤
⎢⎣⎡−
⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧
−
−⎥⎦
⎤⎢⎣
⎡−=
⎭⎬⎫
⎩⎨⎧
AEPL
dd
y
y 2
2121
10
01
3
2
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 137
Expanding the above equations gives the values for the displacements.
⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧
−
−=
⎭⎬⎫
⎩⎨⎧
AEPLAEPL
dd
y
y
3
2
Banded-Symmetric Matrices and Bandwidth
The coefficient matrix (stiffness matrix) for the linear equations that occur in structural analysis is always symmetric and banded. Because a meaningful analysis generally requires the use of a large number of variables, the implemen-tation of compressed storage of the stiffness matrix is desirable both from the viewpoint of fitting into memory (immediate access portion of the computer) and computational efficiency.
Another method, based on the concept of the skyline of the stiffness matrix, is often used to improve the efficiency in solving the equations. The skyline is an envelope that begins with the first nonzero coefficient in each column of the stiff-ness matrix (see the following figure). In skyline, only the coefficients between the main diagonal and the skyline are stored. In general, this procedure takes even less storage space in the computer and is more efficient in terms of equa-tion solving than the conventional banded format.
A matrix is banded if the nonzero terms of the matrix are gathered about the main diagonal. To illustrate this concept, consider the plane truss shown on be-low. We can see that element 2 connects nodes 1 and 4. Therefore, the 2 x 2 submatrices at positions 1-1, 1-4, 4-1, and 4-4 will have nonzero coefficients. The total stiffness matrix of the plane truss, shown in the figure below, denotes non-zero coefficients with X’s. The nonzero terms are within the some band. Using a banded storage format, only the main diagonal and the nonzero upper codiago-nals need be stored.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 138
We now define the semibandwidth nb as nb = nd(m + 1), where nd is the num-ber of degrees of freedom per node and m is the maximum difference in node numbers determined by calculating the difference in node numbers for each ele-ment of a finite element model. In the example for the plane truss shown above, m = 4 - 1 = 3 and nd = 2, so that nb = 2(3 + 1) = 8.
Execution time (primarily, equation-solving time) is a function of the number of equations to be solved. Without using banded storage of global stiffness matrix K, the execution time is proportional to (1/3)n3, where n is the number of equa-tions to be solved. Using banded storage of K, the execution time is proportional to n(nb)2. The ratio of time of execution without banded storage to that using banded storage is then (1/3)(n/nb)2. For the plane truss example, this ratio is (1/3)(24/8)2 = 3. Therefore, it takes about three times as long to execute the solu-tion of the example truss if banded storage is not used.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 139
Several automatic node renumbering schemes have been computerized. This option is available in most general-purpose computer programs. Alternatively, the wavefront or frontal method is becoming popular for optimizing equation solution time. In the wavefront method, elements, instead of nodes, are automatically renumbered.
In the wavefront method the assembly of the equations alternates with their solution by Gauss elimination. The sequence in which the equations are proc-essed is determined by element numbering rather than by node numbering. The first equations eliminated are those associated with element 1 only. Next the con-tributions to stiffness coefficients from the adjacent element, element 2, are eliminated. If any additional degrees of freedom are contributed by elements 1 and 2 only these equations are eliminated (condensed) from the system of equa-tions. As one or more additional elements make their contributions to the system of equations and additional degrees of freedom are contributed only by these elements, those degrees of freedom are eliminated from the solution. This repeti-tive alternation between assembly and solution was initially seen as a wavefront that sweeps over the structure in a pattern determined by the element number-ing.
The wavefront method, although somewhat more difficult to understand and to program than the banded-symmetric method, is computationally more efficient. A banded solver stores and processes any blocks of zeros created in assembling the stiffness matrix. These blocks of zero coefficients are not stored or processed using the wavefront method. Many large-scale computer programs are now using the wavefront method to solve the system of equations.
Problems
8. Do problem B.9 on page 651 in your textbook “A First Course in the Finite Element Method” by D. Logan.
9. Develop a proposal for a truss analysis problem to be handed in with your take-home mid-term exam.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 140
10. Solve the following truss problems. You may use SAP2000 to do truss analysis.
a) For the plane truss shown below, determine the nodal displacements and element stresses. Nodes 1 and 2 are pin joints. Let E = 107 psi and the A = 2.0 in2 for all elements.
1
3
5
7
2
4
6
85 k
10 k
10 k
10 ft
10 ft
10 ft
10 ft
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 141
b) For the 25-bar truss shown below, determine the displacements and ele-mental stresses. Nodes 7, 8, 9, and 10 are pin connections. Let E = 107 psi and the A = 2.0 in2 for the first story and A = 1.0 in2 for the top story. Table 1 lists the coordinates for each node. Table 2 lists the values and directions of the two loads cases applied to the 25-bar space truss.
Table 1. Coordinates for the 25-Bar Truss
Node x (in) y (in) z (in)
1 -37.5 0.0 200.0 2 37.5 0.0 200.0 3 -37.5 37.5 100.0 4 37.5 37.5 100.0 5 37.5 -37.5 100.0 6 -37.5 -37.5 100.0 7 -100.0 100.0 0.0 8 100.0 100.0 0.0 9 100.0 -100.0 0.0
10 -100.0 -100.0 0.0 Note: 1 in = 2.54 cm
Table 2. Multiple Loading Conditions for the 25-Bar Truss
Case Node Fx (kip) Fy (kip) Fz (kip)
1 1.0 10.0 -5.0 2 0.0 10.0 -5.0 3 0.5 0.0 0.0
1
6 0.5 0.0 0.0 1 0.0 20.0 -5.0 2 2 0.0 -20.0 -5.0
Note: 1 kip = 4.45 kN
1 2
3 4
5 6
7 9
8
10
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 142
c) For the 72-bar truss shown below, determine the displacements and ele-mental stresses. Nodes 1, 2, 3, and 4 are pin connections. Let E = 107 psi and the A = 1.0 in2 for the first two stories and A = 0.5 in2 for the top two stories. Table 3 lists the values and directions of the two loads cases ap-plied to the 72-bar space truss.
Table 3. Multiple Loading Conditions for the 72-Bar Truss
Case Node Fx (kip) Fy (kip) Fz (kip) 17 0.0 0.0 -5.0 18 0.0 0.0 -5.0 19 0.0 0.0 -5.0
1
20 0.0 0.0 -5.0 2 17 5.0 5.0 -5.0
Note: 1 kip = 4.45 kN.
120 in
60 in
60 in
60 in
60 in
1
5
9
13
17
2
6
10
14
18
1
2
3
4
5 6
9
7
8 10
11
12
13
14 15
16 17
18
(a)
(b)
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 143
Development of Beam Equations
Introduction
In this section, we will develop the stiffness matrix for a beam element, the most common of all structural elements. The beam element is considered to be straight and to have constant cross-sectional area. We will derive the beam ele-ment stiffness matrix by using the principles of simple beam theory.
The degrees of freedom associated with a node of a beam element are a transverse displacement and a rotation. We will discuss procedures for handling distributed loading and concentrated nodal loading. We will include the nodal shear forces and bending moments and the resulting shear force and bending moment diagrams as part of the total solution.
To further acquaint you with the potential energy approach for developing stiffness matrices and equations, we will again develop the beam bending ele-ment equations using this approach. Finally, the Galerkin residual method is ap-plied to derive the beam element equations.
Beam Stiffness
A beam is a long, slender structural member generally subjected to transverse loading that produces significant bending effects as opposed to twisting or axial effects. The bending deformation is measured as a transverse displacement and a rotation. The degrees of freedom at each node of a beam element will be a transverse displacement and a rotation (as opposed to only an axial displace-ment for the bar element).
Consider the beam element shown below. The beam is of length L with axial local coordinate x and transverse local coordinate y . The local transverse nodal
displacements are given by iyd and the rotations by iφ . The local nodal forces
are given by iyf and the bending moments by im . We initially neglect all axial ef-
fects.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 144
At all nodes, the following sign conventions are used:
1. Moments are positive in the counterclockwise direction. 2. Rotations are positive in the counterclockwise direction. 3. Forces are positive in the positive y direction.
4. Displacements are positive in the positive y direction.
The differential equation governing simple linear-elastic beam behavior can be
derived as follows. Consider the beam shown below.
Write the equations of equilibrium for the differential element:
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 145
0ˆ2ˆˆ)ˆ(ˆ)(0 2 ≈⎟⎠⎞
⎜⎝⎛+−−+==∑ − xdxdxdxwxVdMdMMM sideright
dxxwdVVVFy )ˆ()(0 −+−==∑ From force and moment equilibrium of a differential beam element, we get:
xddMVdMxVdM sideright ˆ
or0ˆ0 ==+−⇒=∑ −
⎟⎠⎞
⎜⎝⎛−=⇒−==−−⇒=∑ xddM
xddw
xddVwdVxwdFy ˆˆˆ
or0ˆ0
The curvature k of the beam is related to the moment by:
EIMk ==
ρ1
where ρ is the radius of the deflected curve, v is the transverse displacement function in the y direction, E is the modulus of elasticity, and I is the principle moment of inertia about y direction, as shown below.
The curvature for small slopes xdvd ˆ/ˆ=θ is given as:
2
2
ˆˆ
xdvdk =
Therefore:
2
2
2
2
ˆˆ
ˆˆ
xdvdEIM
EIM
xdvd
=⇒=
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 146
Substituting the moment expression into the moment-load equations gives:
( )xwxdvdEI
xdd ˆ
ˆˆ
ˆ 2
2
2
2
−=⎟⎠
⎞⎜⎝
⎛
For constant values of EI, the above equation reduces to:
( )xwxdvdEI ˆ
ˆˆ4
4
−=⎟⎠
⎞⎜⎝
⎛
Step 1 - Select Element Type
We will consider the linear-elastic beam element shown below.
Step 2 - Select a Displacement Function
Assume the transverse displacement function v is:
432
23
1 ˆˆˆ axaxaxav +++= The number of coefficients in the displacement function, ai, is equal to the total number of degrees of freedom associated with the element (displacement and rotation at each node). The boundary conditions are:
yy dLxvdxv 21ˆ)ˆ(ˆˆ)0ˆ(ˆ ====
21ˆ
ˆ)ˆ(ˆˆ
ˆ)0ˆ(ˆ φφ =
==
=xd
Lxvdxd
xvd
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 147
Applying the boundary conditions and solving for the unknown coefficients gives:
41ˆ)0(ˆ adv y ==
432
23
12ˆ)(ˆ aLaLaLadLv y +++==
31ˆ)0(ˆ a
xdvd
== φ
322
12 23ˆ)(ˆ aLaLadx
Lvd++== φ
Solving these equations for a1, a2, a3, and a4 gives:
( ) ( ) ( ) ( ) yyyyy dxxL
ddL
xL
ddL
v 112
212123
212213ˆˆˆˆˆˆ21ˆˆ3ˆˆˆ1ˆˆ2ˆ ++⎥⎦
⎤⎢⎣⎡ +−−−+⎥⎦
⎤⎢⎣⎡ −+−= φφφφφ
In matrix form the above equations are:
dNv ˆ][ˆ = where
[ ]4321
2
2
1
1
][ˆ NNNNNd
d
dy
y
=
⎪⎪⎭
⎪⎪⎬
⎫
⎪⎪⎩
⎪⎪⎨
⎧
=
φ
φ
and
( ) ( )322332
32331 ˆˆ2ˆ1ˆ3ˆ21 LxLxLx
LNLLxx
LN +−=+−=
( ) ( )22334
2333 ˆˆ1ˆ3ˆ21 LxLx
LNLxx
LN −=+−=
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 148
N1, N2, N3, and N4 are called the shape functions for a beam element.
N1
-0.200
0.000
0.200
0.400
0.600
0.800
1.000
0.00 1.00
N2
-0.200
0.000
0.200
0.400
0.600
0.800
1.000
0.00 1.00
N3
-0.200
0.000
0.200
0.400
0.600
0.800
1.000
0.00 1.00
N4
-0.200
0.000
0.200
0.400
0.600
0.800
1.000
0.00 1.00
Step 3 - Define the Strain/Displacement and Stress/Strain Relationships
The stress-displacement relationship is:
( )xdudyxx ˆˆˆ,ˆ =ε
where u is the axial displacement function. We can relate the axial displacement to the transverse displacement by considering the beam element shown below:
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 149
xdvdyuˆˆˆˆ −=
One of the basic assumptions in simple beam theory is that planes remain planar after deformation, therefore:
( ) ⎟⎠
⎞⎜⎝
⎛−= 2
2
ˆˆˆˆ,ˆ
xdvdyyxxε
Moments and shears are related to the transverse displacement as:
( ) ( ) ⎟⎠
⎞⎜⎝
⎛=⎟
⎠
⎞⎜⎝
⎛= 3
3
2
2
ˆˆˆ
ˆˆˆˆ
xdvdEIxV
xdvdEIxm
Step 4 - Derive the Element Stiffness Matrix and Equations
Using beam theory sign convention for shear force and bending moment we obtain the following equations:
( )221133
3
1ˆ6ˆ12ˆ6ˆ12
ˆ)0(ˆˆˆ φφ LdLd
LEI
xdvdEIVf yyy +−+===
( )221133
3
2ˆ6ˆ12ˆ6ˆ12
ˆ)(ˆˆˆ φφ LdLd
LEI
xdLvdEIVf yyy −+−−==−=
( )22
212
132
2
1ˆ2ˆ6ˆ4ˆ6
ˆ)0(ˆˆˆ φφ LdLLdL
LEI
xdvdEImm yy +−+=−=−=
( )22
212
132
2
2ˆ4ˆ6ˆ2ˆ6
ˆ)(ˆˆˆ φφ LdLLdL
LEI
xdLvdEImm yy +−+===
xdvdˆˆ
≈θxdds ˆ≈
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 150
In matrix form the above equations are:
⎪⎪⎭
⎪⎪⎬
⎫
⎪⎪⎩
⎪⎪⎨
⎧
⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢
⎣
⎡
−−
−−
−−=
⎪⎪⎭
⎪⎪⎬
⎫
⎪⎪⎩
⎪⎪⎨
⎧
2
2
1
1
2
2
2
2
3
2
2
1
1
ˆˆˆˆ
46
26
612612
2646
612612
ˆ
ˆˆ
ˆ
φ
φ
y
y
y
y
d
d
LL
LL
L
L
LLLL
L
LLEI
mfmf
where the stiffness matrix is:
⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢
⎣
⎡
−−
−−
−−=
2
2
2
2
3
46
26
612612
26
46
612612
LL
LL
L
L
LL
LL
L
LLEIk
Step 5 - Assemble the Element Equations and Introduce Boundary Conditions
Consider a beam modeled by two beam elements, shown below:
Assume the EI to be constant throughout the beam. A force of 1,000 lb and mo-ment of 1,000 lb-ft are applied to the mid-point of the beam. The beam element stiffness matrices are:
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 151
⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢
⎣
⎡
−−
−−
−−=
2
2
2
2
3)1(
46
26
612612
26
46
612612
2211
LL
LL
L
L
LL
LL
L
LLEIk
ydyd φφ
⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢
⎣
⎡
−−
−−
−−=
2
2
2
2
3)2(
46
26
612612
26
46
612612
3322
LL
LL
L
L
LL
LL
L
LLEIk
ydyd φφ
In this example, the local coordinates coincide with the global coordinates of the whole beam (therefore there is no transformation required for this problem). The total stiffness matrix can be assembled as:
The boundary conditions are:
1 1 3 0y yd dφ= = =
By applying the boundary conditions the beam equations reduce to:
⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡=
⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧⋅
−
3
2
2
22
223
4262806024
0000,1
000,1
φφ
yd
LLLLLL
LEIftlb
lb
⎪⎪⎪
⎭
⎪⎪⎪
⎬
⎫
⎪⎪⎪
⎩
⎪⎪⎪
⎨
⎧
⎥⎥⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢⎢⎢
⎣
⎡
−−−−
−++−−+−+−−
−−
=
⎪⎪⎪
⎭
⎪⎪⎪
⎬
⎫
⎪⎪⎪
⎩
⎪⎪⎪
⎨
⎧
3
3
2
2
1
1
22
2222
22
3
3
3
2
2
1
1
46260061261200
2644662661266121261200264600612612
φ
φ
φ
y
y
y
y
y
y
d
d
d
LLLLLL
LLLLLLLLLLLL
LLLLLL
LEI
MFMFMF
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 152
Solving the above equations gives:
⎪⎪⎪⎪⎪
⎭
⎪⎪⎪⎪⎪
⎬
⎫
⎪⎪⎪⎪⎪
⎩
⎪⎪⎪⎪⎪
⎨
⎧
−−
−−
−−
=⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧
radEI
LL
radEI
LL
lbEI
LL
d y
1252125
46252125
12
23753875
3
2
2
φφ
Example 1 - Beam Problem
Consider the beam shown below. Assume that EI is constant and the length is 2L.
The beam element stiffness matrices are:
⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢
⎣
⎡
−−
−−
−−=
2
2
2
2
3)1(
46
26
612612
26
46
612612
2211
LL
LL
L
L
LL
LL
L
LLEIk
ydyd φφ
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 153
⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢
⎣
⎡
−−
−−
−−=
2
2
2
2
3)2(
3
46
26
612612
26
46
612612
322
LL
LL
L
L
LL
LL
L
LLEIk
ydyd φφ
The local coordinates coincide with the global coordinates of the whole beam (therefore there is no transformation required for this problem). The total stiffness matrix can be assembled as:
The governing beam equations are:
⎪⎪⎪
⎭
⎪⎪⎪
⎬
⎫
⎪⎪⎪
⎩
⎪⎪⎪
⎨
⎧
⎥⎥⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢⎢⎢
⎣
⎡
−−−−
−−−−
−−
=
⎪⎪⎪
⎭
⎪⎪⎪
⎬
⎫
⎪⎪⎪
⎩
⎪⎪⎪
⎨
⎧
3
3
2
2
1
1
3
3
3
2
2
1
1
2462260061261200
2262802266120246120022624600612612
φ
φ
φ
y
y
y
y
y
y
d
d
d
LLLLLL
LLLLLLL
LLLLLL
LEI
MFMFMF
The boundary conditions are:
0332 === φyy dd
⎥⎥⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢⎢⎢
⎣
⎡
−−−−
−−−−
−−
=
2462260061261200
2262802266120246120022624600612612
3
LLLLLL
LLLLLLL
LLLLLL
LEIK
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 154
By applying the boundary conditions the beam equations reduce to:
⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡=
⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧−
2
1
1
2
2
2
23
826
246
6612
00
φφ
yd
LLL
LLL
LL
LEI
P
Solving the above equations gives:
⎪⎪⎪
⎭
⎪⎪⎪
⎬
⎫
⎪⎪⎪
⎩
⎪⎪⎪
⎨
⎧
=⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧−
1
34
37
2
2
1
1
L
EIPL
d y
φφ
The positive signs for the rotations indicate that both are in the counterclockwise direction. The negative sign on the displacement indicates a deformation in the
y− direction.
⎪⎪⎪
⎭
⎪⎪⎪
⎬
⎫
⎪⎪⎪
⎩
⎪⎪⎪
⎨
⎧
⎥⎥⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢⎢⎢
⎣
⎡
−−−−
−−−−
−−
=
⎪⎪⎪
⎭
⎪⎪⎪
⎬
⎫
⎪⎪⎪
⎩
⎪⎪⎪
⎨
⎧ −
00103
2462260061261200
2262802266120246120022624600612612
4
37
3
3
2
2
1
1L
y
y
y
LLLLLL
LLLLLLL
LLLLLL
LP
MFMFMF
The local nodal forces for element 1:
⎪⎪⎭
⎪⎪⎬
⎫
⎪⎪⎩
⎪⎪⎨
⎧
−
−
=
⎪⎪⎭
⎪⎪⎬
⎫
⎪⎪⎩
⎪⎪⎨
⎧−
⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢
⎣
⎡
−−−−
−−
=
⎪⎪⎭
⎪⎪⎬
⎫
⎪⎪⎩
⎪⎪⎨
⎧
PLP
P
LLLLLL
LLLLLL
LP
mfmf L
y
y
0
001
4626612612
2646612612
4ˆ
ˆˆ
ˆ 37
22
22
2
2
1
1
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 155
The local nodal forces for element 2:
⎪⎪⎭
⎪⎪⎬
⎫
⎪⎪⎩
⎪⎪⎨
⎧
−=
⎪⎪⎭
⎪⎪⎬
⎫
⎪⎪⎩
⎪⎪⎨
⎧
⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢
⎣
⎡
−−−−
−−
=
⎪⎪⎭
⎪⎪⎬
⎫
⎪⎪⎩
⎪⎪⎨
⎧
PLP
PLP
LLLLLL
LLLLLL
LP
mfmf
y
y
5.05.1
5.1
0010
4626612612
2646612612
4ˆ
ˆˆ
ˆ
22
22
3
3
2
2
The free-body diagrams for the each element are shown below.
Combining the elements gives the forces and moments for the original beam.
Therefore, the shear force and bending moment diagrams are:
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 156
Example 2 - Beam Problem
Consider the beam shown below. Assume E = 30 x 106 psi and I = 500 in.4 are constant throughout the beam. Use four elements of equal length to model the beam.
The beam element stiffness matrices are:
⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢
⎣
⎡
−−
−−
−−=
++
2
2
2
2
3)(
46
26
612612
26
46
612612
1)1(
LL
LL
L
L
LL
LL
L
LLEIk i
iyidiiyd φφ
Using the direct stiffness method, the four beam element stiffness matrices are superimposed to produce the global stiffness matrix. As shown on the next page. The boundary conditions for this problem are:
1 1 3 5 5 0y y yd d dφ φ= = = = =
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 157
Ele
men
t 1
Ele
men
t 2
Ele
men
t 3
Ele
men
t 4
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 158
After applying the boundary conditions the global beam equations reduce to:
⎪⎪⎪
⎭
⎪⎪⎪
⎬
⎫
⎪⎪⎪
⎩
⎪⎪⎪
⎨
⎧
−
−
=
⎪⎪⎪
⎭
⎪⎪⎪
⎬
⎫
⎪⎪⎪
⎩
⎪⎪⎪
⎨
⎧
⎥⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢⎢
⎣
⎡
−−
0000,1000000,10
80200024600
2682600280006024
4
4
3
2
2
22
222
22
3
lb
lb
d
d
LLL
LLLLLLLL
LEI
y
y
φ
φφ
Substituting L = 120 in., E = 30 x 106 psi, and I = 500 in.4 into the above equa-tions and solving for the unknowns gives:
0048.0 43242 ===−== φφφindd yy The global forces and moments can be determined as:
ftkipsMMMM
ftkipsM
kipsFkipsFkipsFkipsF
kipsF
y
y
y
y
y
·25000
·25
51010105
5
4
3
2
1
5
4
3
2
1
−=====
=====
The local nodal forces for element 1:
⎪⎪⎭
⎪⎪⎬
⎫
⎪⎪⎩
⎪⎪⎨
⎧
−=
⎪⎪⎭
⎪⎪⎬
⎫
⎪⎪⎩
⎪⎪⎨
⎧
−⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢
⎣
⎡
−−
−−
−−=
⎪⎪⎭
⎪⎪⎬
⎫
⎪⎪⎩
⎪⎪⎨
⎧
ftkkips
ftkkips
LL
LL
L
L
LL
LL
L
LLEI
mfmf
y
y
·255
·255
0048.000
46
26
612612
26
46
612612
ˆ
ˆˆ
ˆ
2
2
2
2
3
2
2
1
1
The local nodal forces for element 2:
⎪⎪⎭
⎪⎪⎬
⎫
⎪⎪⎩
⎪⎪⎨
⎧
−
−−
=
⎪⎪⎭
⎪⎪⎬
⎫
⎪⎪⎩
⎪⎪⎨
⎧−
⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢
⎣
⎡
−−
−−
−−=
⎪⎪⎭
⎪⎪⎬
⎫
⎪⎪⎩
⎪⎪⎨
⎧
ftkkips
ftkkips
LL
LL
L
L
LL
LL
L
LLEI
mfmf
y
y
·255
·255
000048.0
46
26
612612
26
46
612612
ˆ
ˆˆ
ˆ
2
2
2
2
3
3
3
2
2
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 159
The local nodal forces for element 3:
⎪⎪⎭
⎪⎪⎬
⎫
⎪⎪⎩
⎪⎪⎨
⎧
−=
⎪⎪⎭
⎪⎪⎬
⎫
⎪⎪⎩
⎪⎪⎨
⎧
−⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢
⎣
⎡
−−
−−
−−=
⎪⎪⎭
⎪⎪⎬
⎫
⎪⎪⎩
⎪⎪⎨
⎧
ftkkips
ftkkips
LLLL
L
L
LLLL
L
LLEI
mfmf
y
y
·255
·255
0048.000
4626
612612
2646
612612
ˆ
ˆˆ
ˆ
2
2
2
2
3
4
4
3
3
The local nodal forces for element 4:
⎪⎪⎭
⎪⎪⎬
⎫
⎪⎪⎩
⎪⎪⎨
⎧
−
−−
=
⎪⎪⎭
⎪⎪⎬
⎫
⎪⎪⎩
⎪⎪⎨
⎧−
⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢
⎣
⎡
−−
−−
−−=
⎪⎪⎭
⎪⎪⎬
⎫
⎪⎪⎩
⎪⎪⎨
⎧
ftkkips
ftkkips
LLLL
L
L
LLLL
L
LLEI
mfmf
y
y
·255
·255
000048.0
4626
612612
2646
612612
ˆ
ˆˆ
ˆ
2
2
2
2
3
5
5
4
4
Note: Due to symmetry about the vertical plane at node 3, we could have worked just half the beam, as shown below.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 160
Example 3 - Beam Problem
Consider the beam shown below. Assume E = 210 GPa and I = 2 x 10-4 m4 are constant throughout the beam and the spring constant k = 200 kN/m. Use two beam elements of equal length and one spring element to model the struc-ture.
The beam element stiffness matrices are:
⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢
⎣
⎡
−−
−−
−−=
2
2
2
2
3)1(
46
26
612612
26
46
612612
2211
LL
LL
L
L
LL
LL
L
LLEIk
ydyd φφ
⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢
⎣
⎡
−−
−−
−−=
2
2
2
2
3)2(
3
46
26
612612
26
46
612612
322
LL
LL
L
L
LL
LL
L
LLEIk
ydyd φφ
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 161
The spring element stiffness matrix is:
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡
−
−=⇒⎥
⎦
⎤⎢⎣
⎡ −−
=kk
kkk
kk
kk
k
dddd yyyy
0000
0)3()3(
43343 φ
Using the direct stiffness method and superposition gives the global beam equa-tions.
The boundary conditions for this problem are:
1 1 2 4 0y y yd d dφ= = = =
After applying the boundary conditions the global beam equations reduce to:
⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧−=
⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡
−−+−
−=
⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧
0
0
4626'126
268
3
3
2
22
22
3
3
3
2
PdLLLLkL
LLL
LEI
MFM
yy
φ
φ
Solving the above equations gives:
EIkLk
kEIPL
kEIPL
kEIPL
d y
3
2
3
2
3
3
2
'
'71219
'71217
'71213
=
⎪⎪⎪
⎭
⎪⎪⎪
⎬
⎫
⎪⎪⎪
⎩
⎪⎪⎪
⎨
⎧
⎟⎠⎞
⎜⎝⎛
+−
⎟⎠⎞
⎜⎝⎛
+−
⎟⎠⎞
⎜⎝⎛
+−
=⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧
φ
φ
EIkLk
d
d
d
d
kkLLLL
kLkLLLLLLLL
LLLLLL
LEI
FMFMFMF
y
y
y
y
y
y
y
y
3
4
3
3
2
2
1
1
22
222
22
3
4
3
3
2
2
1
1
'
'0'00000462600
'6'1261200026802606120246120002646000612612
=
⎪⎪⎪⎪
⎭
⎪⎪⎪⎪
⎬
⎫
⎪⎪⎪⎪
⎩
⎪⎪⎪⎪
⎨
⎧
⎥⎥⎥⎥⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢⎢⎢⎢⎢
⎣
⎡
−−
−−+−−−−−−
−−
=
⎪⎪⎪⎪
⎭
⎪⎪⎪⎪
⎬
⎫
⎪⎪⎪⎪
⎩
⎪⎪⎪⎪
⎨
⎧
φ
φ
φ
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 162
Substituting L = 3 m, E = 210 GPa, I = 2 x 10-4 m4, and k = 200 kN/m in the above equations gives:
radradmd y
00747.000249.00174.0
3
2
3
−=−=−=
φφ
Substituting the solution back into the global equations gives:
⎪⎪⎪⎪
⎭
⎪⎪⎪⎪
⎬
⎫
⎪⎪⎪⎪
⎩
⎪⎪⎪⎪
⎨
⎧
−
⋅−−
=
⎪⎪⎪⎪
⎭
⎪⎪⎪⎪
⎬
⎫
⎪⎪⎪⎪
⎩
⎪⎪⎪⎪
⎨
⎧
kN
kN
kNmkN
kN
FMFMFMF
y
y
y
y
5.30
5004.116
7.699.69
4
3
3
2
2
1
1
A free-body diagram, including forces and moments acting on the beam is shown below.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 163
Distributed Loadings
Beam members can support distributed loading as well as concentrated nodal loading. Therefore, we must be able to account for distributed loading. Consider the fixed-fixed beam subjected to a uniformly distributed loading w shown the fig-ure below. The reactions, determined from structural analysis theory, are called fixed-end reactions. In general, fixed-end reactions are those reactions at the ends of an element if the ends of the element are assumed to be fixed (dis-placements and rotations are zero). Therefore, guided by the results from struc-tural analysis for the case of a uniformly distributed load, we replace the load by concentrated nodal forces and moments tending to have the same effect on the beam as the actual distributed load.
The figure below illustrates the idea of equivalent nodal loads for a general beam. We can replace the effects of a uniform load by a set of nodal forces and moments.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 164
Work Equivalence Method
This method is based on the concept that the work done by the distributed load is equal to the work done by the discrete nodal loads. The work done by the distributed load is:
( ) ( )∫=0
ˆˆˆˆ xdxvxwW ddistribute
where ( )xv ˆˆ is the transverse displacement. The work done by the discrete nodal forces is:
yyyynodes dfdfmmW 22112211ˆˆˆˆˆˆˆˆ +++= φφ
The nodal forces can be determined by setting Wdistributed = Wnodes for arbitrary displacements and rotations.
nodesddistribute WW =
Example 4 - Load Replacement
Consider the beam, shown below, and determine the equivalent nodal forces for the given distributed load.
Using the work equivalence method or:
nodesddistribute WW = we get:
( ) ( ) yyyy dfdfmmxdxvxwL
221122110
ˆˆˆˆˆˆˆˆˆˆˆˆ +++=∫ φφ
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 165
Evaluating the left-hand-side of the above expression using ( )xw ˆ = -w and ( )xv ˆˆ equal to:
( ) ( ) ( ) ( ) ( ) yyyyy dxxL
ddL
xL
ddL
xv 112
212123
212213ˆˆˆˆˆˆ21ˆˆ3ˆˆˆ1ˆˆ2ˆˆ ++⎥⎦
⎤⎢⎣⎡ +−−−+⎥⎦
⎤⎢⎣⎡ ++−= φφφφφ
gives:
( ) ( ) ( ) ( ) ( ) yyyyy dwLwLwLddLwwLddLwxdxvwL
11
2
21
2
1221
2
210
ˆˆ2
ˆˆ23
ˆˆˆˆ4
ˆˆ2
ˆˆˆ −−++−−+−−=∫ φφφφφ
Using a set of arbitrary nodal displacements, such as:
10 1221 ==== φφyy dd The resulting nodal equivalent force or moment is:
12232
4)1(ˆ
222
2
1
wLwLwLwLm −=⎟⎠
⎞⎜⎝
⎛+−−=
Using another set of arbitrary nodal displacements, such as:
10 2121 ==== φφyy dd The resulting nodal equivalent force or moment is:
1234)1(ˆ
222
2
wLwLwLm =⎟⎠
⎞⎜⎝
⎛−−=
Setting the nodal rotations equal zero except for the yd1
ˆ and yd2ˆ gives:
22)1(1
LwLwLwLWf y −=−+−=
22)1(2
LwLwLWf y −=−=
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 166
General Formulation
We can account for the distributed loads or concentrated loads acting on a beam elements by considering the following formulation for a general structure:
0FKdF −= where F0 are the equivalent nodal forces, expressed in terms of the global-coordinate components. These force would yield the same displacements as the original distributed load. If we assume that the global nodal forces are not pre-sent (F = 0) then:
KdF =0 We now solve for the displacements, d, given the nodal forces F0. Next, substi-tute the displacements and the equivalent nodal forces F0 back into the original expression and solve for the global nodal forces.
Example 5 - Load Replacement
Consider the beam shown below, determine the equivalent nodal forces for the given distributed load.
The work equivalent nodal forces are shown above. Using the beam stiffness equations, with the boundary conditions applied, we can solve for the displace-ments
⎭⎬⎫
⎩⎨⎧⎥⎦
⎤⎢⎣
⎡ −−
=⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧−
2
2232 ˆ
ˆ
46
612
12
2φ
ydLL
LLEI
wL
wL
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 167
Therefore:
⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧
−
−=
⎭⎬⎫
⎩⎨⎧
EIwL
EIwL
d y
6
8ˆˆ
3
4
2
2
φ
In this case, the method of equivalent nodal forces gives the exact solution for the displacements and rotations. To obtain the global nodal forces, we will first define the product of Kd to be Fe, where Fe is called the effective global nodal forces. Therefore:
⎪⎪⎭
⎪⎪⎬
⎫
⎪⎪⎩
⎪⎪⎨
⎧
−−
⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢
⎣
⎡
−−−−
−−
=
⎪⎪⎭
⎪⎪⎬
⎫
⎪⎪⎩
⎪⎪⎨
⎧
EIwL
EIwL
e
ye
e
ye
LLLLLL
LLLLLL
LEI
MFMF
6
8
22
22
3
1
2
1
1
3
4
00
4626612612
2646612612
Simplifying the above expression gives:
⎪⎪⎪⎪
⎭
⎪⎪⎪⎪
⎬
⎫
⎪⎪⎪⎪
⎩
⎪⎪⎪⎪
⎨
⎧
−=
⎪⎪⎭
⎪⎪⎬
⎫
⎪⎪⎩
⎪⎪⎨
⎧
12
2
125
2
2
2
2
2
1
1
wL
wL
wL
wL
MFMF
e
ye
e
ye
Using the above expression and the fix-end moments in:
0FKdF −= gives the correct global nodal forces as:
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 168
⎪⎪⎪⎪
⎭
⎪⎪⎪⎪
⎬
⎫
⎪⎪⎪⎪
⎩
⎪⎪⎪⎪
⎨
⎧
=
⎪⎪⎪⎪
⎭
⎪⎪⎪⎪
⎬
⎫
⎪⎪⎪⎪
⎩
⎪⎪⎪⎪
⎨
⎧
−
−
−
−
⎪⎪⎪⎪
⎭
⎪⎪⎪⎪
⎬
⎫
⎪⎪⎪⎪
⎩
⎪⎪⎪⎪
⎨
⎧
−=
⎪⎪⎭
⎪⎪⎬
⎫
⎪⎪⎩
⎪⎪⎨
⎧
0
02
12
2
12
2
12
2
125
22
2
2
2
2
2
2
1
1 wL
wL
wL
wL
wL
wL
wL
wL
wL
wL
MFMF
y
y
Example 6 - Cantilever Beam
Consider the beam, shown below, determine the vertical displacement and ro-tation at the free-end and the nodal forces, including reactions. Assume EI is constant throughout the beam.
We will use one element and replace the concentrated load with the appropriate nodal forces. The beam stiffness equations become:
⎭⎬⎫
⎩⎨⎧⎥⎦
⎤⎢⎣
⎡ −−
=⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧−
2
223 ˆ
ˆ
46
612
8
2φ
ydLL
LLEI
PL
P
Therefore:
⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧
−
−=
⎭⎬⎫
⎩⎨⎧
EIPL
EIPL
d y
8
485
ˆˆ
2
3
2
2
φ
To obtain the global nodal forces, we begin by evaluating the effective nodal
forces.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 169
⎪⎪⎭
⎪⎪⎬
⎫
⎪⎪⎩
⎪⎪⎨
⎧
−−
⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢
⎣
⎡
−−−−
−−
=
⎪⎪⎭
⎪⎪⎬
⎫
⎪⎪⎩
⎪⎪⎨
⎧
EIPL
EIPL
e
ye
e
ye
LLLLLL
LLLLLL
LEI
MFMF
8
485
22
22
3
1
2
1
1
2
3
00
4626612612
2646612612
Simplifying the above expression gives:
⎪⎪⎪⎪
⎭
⎪⎪⎪⎪
⎬
⎫
⎪⎪⎪⎪
⎩
⎪⎪⎪⎪
⎨
⎧
−=
⎪⎪⎭
⎪⎪⎬
⎫
⎪⎪⎩
⎪⎪⎨
⎧
8
2
83
2
1
2
1
1
PL
P
PL
P
MFMF
e
ye
e
ye
Using the above expression in the following equation, gives:
0FKdF −= The correct global nodal forces as:
⎪⎪⎪⎪
⎭
⎪⎪⎪⎪
⎬
⎫
⎪⎪⎪⎪
⎩
⎪⎪⎪⎪
⎨
⎧
=
⎪⎪⎪⎪
⎭
⎪⎪⎪⎪
⎬
⎫
⎪⎪⎪⎪
⎩
⎪⎪⎪⎪
⎨
⎧
−
−
−
−
⎪⎪⎪⎪
⎭
⎪⎪⎪⎪
⎬
⎫
⎪⎪⎪⎪
⎩
⎪⎪⎪⎪
⎨
⎧
−=
⎪⎪⎭
⎪⎪⎬
⎫
⎪⎪⎩
⎪⎪⎨
⎧
0
02
8
2
8
2
8
2
83
2
2
2
1
1 PL
P
PL
P
PL
P
PL
P
PL
P
MFMF
y
y
In general, for any structure in which an equivalent nodal force replacement is made, the actual nodal forces acting on the structure are determined by first evaluating the effective nodal forces Fe for the structure and then subtracting off the equivalent nodal forces F0 for the structure. Similarly, for any element of a structure in which equivalent nodal force replacement is made, the actual local nodal forces acting on the element are determined by first evaluating the effec-
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 170
tive local nodal forces )(ˆ ef for the element and then subtracting off the equivalent local nodal forces 0f associated only with the element.
Beam Element with Nodal Hinge
Consider the beam, shown below, with an internal hinge. An internal hinge causes a discontinuity in the slope of the deflection curve at the hinge and the bending moment is zero at the hinge.
For a beam with a hinge on the right end, the moment 2m is zero and we can partition the matrix to eliminate the degree of freedom associated with 2φ .
We can condense out the degree of freedom by using the partitioning method discussed earlier. Recall, the form of kc
][]][[][ 211
221211 KKKKkc−−=
[ ]LLLL
LLL
LEIL
LLL
LLEIkc 626
41
626
12612
646
12612
22
23
23
−⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧
−−
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡−−
−−=
⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢
⎣
⎡
−−
−−
−−=
2
2
2
2
3
46
26
612612
26
46
612612
ˆ
LL
LL
L
L
LL
LL
L
LLEIk
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 171
Therefore, the condensed stiffness matrix is:
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡
−−−−
=11
113 2
3
LLLL
L
LEIkc
The element force-displacement equations are:
⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡
−−−−
=⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧
y
y
y
y
d
d
LLLL
L
LEI
fmf
2
1
12
3
2
1
1
ˆˆˆ
11
113
ˆˆ
ˆ
φ
Expanding the element force-displacement equations and maintaining 2m = 0 gives:
⎪⎪⎭
⎪⎪⎬
⎫
⎪⎪⎩
⎪⎪⎨
⎧
⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢
⎣
⎡
−−−−
=
⎪⎪⎭
⎪⎪⎬
⎫
⎪⎪⎩
⎪⎪⎨
⎧
2
2
1
12
3
2
2
1
1
ˆˆˆˆ
00000110011
3
ˆ
ˆˆ
ˆ
φ
φ
y
y
y
y
d
d
LLLL
L
LEI
mfmf
The element force-displacement equations maintaining 1m = 0 gives:
⎪⎪⎭
⎪⎪⎬
⎫
⎪⎪⎩
⎪⎪⎨
⎧
⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢
⎣
⎡
−−−
−
=
⎪⎪⎭
⎪⎪⎬
⎫
⎪⎪⎩
⎪⎪⎨
⎧
2
2
1
1
2
3
2
2
1
1
ˆˆˆˆ
0101
0000101
3
ˆ
ˆˆ
ˆ
φ
φ
y
y
y
y
d
d
LLLL
L
LEI
mfmf
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 172
Example 7 - Beam With Hinge
In the following beam, shown below, determine the vertical displacement and rotation at node 2 and the element forces for the uniform beam with an internal hinge at node 2. Assume EI is constant throughout the beam.
The stiffness matrix for element 1 (with hinge) is:
⎪⎪⎭
⎪⎪⎬
⎫
⎪⎪⎩
⎪⎪⎨
⎧
⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢
⎣
⎡
−−−−
=
2
2
1
12
3)1(
ˆˆˆˆ
00000110011
3
2211
φ
φ
φφ
y
y
d
d
aaaa
a
aEIk
ydy
d
The stiffness matrix for element 2 (without hinge) is:
⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢
⎣
⎡
−−
−−
−−=
2
2
2
2
3)2(
46
26
612612
26
46
612612
3322
bb
bb
b
b
bb
bb
b
bbEIk
ydy
d φφ
The boundary conditions for this problem are:
03131 ==== φφyy dd
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 173
After applying the boundary conditions the global beam equations reduce to:
3 3 22
22
3 12 6
6 4 0φ
⎡ ⎤+⎢ ⎥ −⎧ ⎫ ⎧ ⎫=⎢ ⎥ ⎨ ⎬ ⎨ ⎬⎩ ⎭⎩ ⎭⎢ ⎥
⎢ ⎥⎣ ⎦
yd Pa b bEI
b b
Solving the above equations gives:
( )
( )
3 3
3 32
3 32
3 3
3
2φ
⎧ ⎫−⎪ ⎪+⎧ ⎫ ⎪ ⎪=⎨ ⎬ ⎨ ⎬
⎩ ⎭ ⎪ ⎪⎪ ⎪+⎩ ⎭
y
a b Pb a EIda b P
b a EI
The element force-displacement equations for element 1 are:
( ) ⎪⎪⎪
⎭
⎪⎪⎪
⎬
⎫
⎪⎪⎪
⎩
⎪⎪⎪
⎨
⎧
+−
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡
−−−−
=⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧
EIabPbaa
aaaa
aEI
fmf
y
y
33
33
23
2
1
1
3
0
0
11
113
ˆˆ
ˆ
Therefore:
3
3 3
1 3
1 3 3
32
3 3
ˆ
ˆˆ
⎧ ⎫⎪ ⎪+⎧ ⎫ ⎪ ⎪
⎪ ⎪ ⎪ ⎪⎪ ⎪ =⎨ ⎬ ⎨ ⎬+⎪ ⎪ ⎪ ⎪⎪ ⎪ ⎪ ⎪⎩ ⎭
−⎪ ⎪+⎩ ⎭
y
y
b Pb afab Pm
b af b P
b a
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 174
The element force-displacement equations for element 2 are:
( )
( )
3 3
3 32
2 2 3 22
3 3 33
2 23
ˆ 312 6 12 6ˆ 6 4 6 2ˆ 12 6 12 6 2
6 2 6 4ˆ 00
⎧ ⎫−⎪ ⎪+⎧ ⎫ −⎡ ⎤ ⎪ ⎪
⎪ ⎪ ⎢ ⎥ ⎪ ⎪−⎪ ⎪ ⎪ ⎪⎢ ⎥= −⎨ ⎬ ⎨ ⎬⎢ ⎥− − − +⎪ ⎪ ⎪ ⎪⎢ ⎥⎪ ⎪ ⎪ ⎪−⎣ ⎦⎩ ⎭ ⎪ ⎪⎪ ⎪⎩ ⎭
y
y
a b Pb a EIf b b
m b b b bEI a b Pb b b b a EIf
b b b bm
Therefore:
3
3 3
2
23
33 3
33
3 3
ˆ0ˆ
ˆ
ˆ
⎧ ⎫−⎪ ⎪+⎪ ⎪⎧ ⎫ ⎪ ⎪⎪ ⎪ ⎪ ⎪⎪ ⎪ ⎪ ⎪=⎨ ⎬ ⎨ ⎬
⎪ ⎪ ⎪ ⎪⎪ ⎪ ⎪ ⎪+⎩ ⎭ ⎪ ⎪
⎪ ⎪⎪ ⎪+⎩ ⎭
y
y
a Pb a
fm
a Pfb amba P
b a
Potential Energy Approach to Derive Beam Element Equations
Let’s derive the equations for a beam element using the principle of minimum potential energy. The procedure for applying the principle of minimum potential energy is similar to that used for the bar element. The total potential energy, πp, is defined as the sum of the internal strain energy U and the potential energy of the external forces Ω:
Ω+=Upπ The differential internal work (strain energy) dU in a one-dimensional beam ele-ment is:
∫=V
xx dVU εσ21
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 175
For a single beam element, shown below, subjected to both distributed and con-centrated nodal forces, the potential energy due to forces (or the work done these forces) is:
∑∑∫==
−−−=Ω2
1
2
1
ˆˆˆˆˆˆi
iii
iyiyS
y mdPdSvT φ
If the beam element has a constant cross-sectional area A, then the differential volume of the beam is given as:
xddAdV ˆ= and the differential element where the surface loading acts is given as:
xdbdS ˆ= where b is the width of the beam element. Therefore the total potential energy is:
( )∑∫∫ ∫=
−−−=2
1ˆˆ21 ˆˆˆˆˆˆˆˆ
iiiiyiy
xy
x Axxp mdPxdbvTxddA φεσπ
The strain-displacement relationship is:
2
2
ˆˆˆ
xdvdyx −=ε
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 176
We can express the strain in terms of nodal displacements and rotations as:
dL
LxLL
LxL
LxLL
Lxyxˆ2661246612
3
2
33
2
3 ⎥⎦
⎤⎢⎣
⎡ −+−−−−=ε
or: dByx
ˆ][ˆ−=ε where [B] is:
⎥⎦
⎤⎢⎣
⎡ −+−−−= 3
2
33
2
3
2661246612][L
LxLL
LxL
LxLL
LxB
The stress-strain relationship in one-dimension is:
xx E εσ ][= where E is the modulus of elasticity. Therefore:
dBEyxˆ]][[ˆ−=σ
The total potential energy can be written in matrix form as:
PdxdvTbxddAT
x
Ty
x Vx
Txp
ˆˆˆˆˆˆˆˆ
21 −−= ∫∫ ∫ εσπ
If we define, yTbw ˆ= as a line load (load per unit length) in the y direction and the substitute the definitions of σx and εx the total potential energy can be written in matrix form as:
PdxdNdwxdAddBBdyE TLTT
x A
TT
pˆˆˆ][ˆˆˆ][][ˆ
2 0ˆ
2 −−= ∫∫ ∫π
Use the following definition for moment of inertia:
∫=A
dAyI 2
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 177
Then the total potential energy expression becomes:
PdxdNdwxddBBdEI TLTT
x
TT
pˆˆˆ][ˆˆˆ][][ˆ
2 0ˆ
−−= ∫∫π
Differentiating the total potential energy with respect to the displacement and ro-tations ( 2121
ˆ,ˆ,ˆ,ˆ ffdd yy ) and equating each term to zero gives:
0ˆˆ][ˆˆ][][0ˆ
=−− ∫∫ PxdNwdxdBBEIL
T
x
T
The nodal forces vector is:
0ˆˆ][ˆ0
=−− ∫ PxdNwfL
T
The elemental stiffness matrix is:
[ ] ∫=x
T xdBBEIkˆ
ˆ][][ˆ
Integrating the above matrix expression gives:
[ ]⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢
⎣
⎡
−−
−−
−−=
2
2
2
2
3
46
26
612612
26
46
612612
ˆ
LL
LL
L
L
LL
LL
L
LLEIk
Galerkin’s Method to Derive Beam Element Equations
The governing differential equation for a one-dimensional beam is:
4
4
ˆ0
ˆd vEI wdx
⎛ ⎞+ =⎜ ⎟
⎝ ⎠
We can define the residual R as:
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 178
4
40
ˆ ˆ 0 1, 2, 3, and 4ˆ
L
id vEI w N dx idx
⎛ ⎞⎛ ⎞+ = =⎜ ⎟⎜ ⎟
⎝ ⎠⎝ ⎠∫
If we apply integration by parts twice to the first term we get:
( ) ( )( ) ( ) ( )( )[ ]Lxxxixxxi
L
xxixx
L
ixxxx vNvNEIxdNvEIxdNvEI0ˆˆˆ,ˆˆˆ
0ˆˆ,ˆˆ
0ˆˆˆˆ ˆˆˆˆˆˆ −+= ∫∫
where the subscript x indicates a derivative with respect to x . Since dNv ˆ][ˆ = , then the second derivative of v with respect to x is:
dL
LLxL
LxL
LLxL
Lxv xxˆ2ˆ66ˆ124ˆ66ˆ12ˆ
3
2
33
2
3ˆˆ ⎥⎦
⎤⎢⎣
⎡ −+−−−=
or dBv xxˆ][ˆˆ =
Therefore the integration by parts becomes:
( ) ( ) ˆ ˆ ˆ, ,00 0
ˆ ˆˆˆ ˆ[ ] 0 1, 2, 3, and 4L L L
i ii xx i xN EI B dx d N w dx NV N m d i⎡ ⎤+ + − = =⎣ ⎦∫ ∫ The above expression is really four equations (one for each Ni) and can be writ-ten in matrix form as:
ˆ 00 0
ˆ ˆˆˆ ˆ[ ] [ ] [ ] [ ] [ ]L L LT T T T
xB EI B dx d N w dx N m N V= − + −∫ ∫
The interpolation function in the last two terms can be evaluated:
( ) ( ) ]1000[ˆ][]0010[0ˆ][ ˆˆ ==== LxNxN xx
( ) ( ) ]0100[ˆ][]0001[0ˆ][ ==== LxNxN
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 179
Therefore, the last two terms of the matrix form of the Galerkin formulation be-come (see the figure below):
)0(ˆ2)0(ˆ1 miVi ⇒=⇒=
)(ˆ4)(ˆ3 LmiLVi ⇒=⇒=
When beam elements are assembled, as shown below, two shear forces and two moments form adjoining elements contribute to the concentrated force and the concentrated moment at the node common to both elements.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 180
Problems:
11. Verify the four beam element equations are contained in the following ma-trix expression.
0ˆˆ][ˆˆ][][0
=−− ∫∫ PxdNwdxdBBEIL
T
A
T
12. Do problems 4.6, 4.11, 4.19, 4.23, 4.29 and 4.34 on pages 181 - 187 in your textbook “A First Course in the Finite Element Method” by D. Logan.
13. Work problem 4.21 (on page 185 in your textbook “A First Course in the Fi-nite Element Method” by D. Logan) using the SAP2000 analysis system (you may used any available structural analysis software system). Start with one element and increase the number of beam elements until you get good agreement with the exact solution. Plot you results for each discreti-zation to the exact solution. How many elements are enough to accurately model this problem?
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 181
Plane Frame and Grid Equations
Introduction
Many structures, such as buildings and bridges, are composed of frames and/or grids. This chapter develops the equations and methods for solution of plane frames and grids. First, we will develop the stiffness matrix for a beam element arbitrarily oriented in a plane. We will then include the axial nodal displacement degree of freedom in the local beam element stiffness matrix. Then we will combine these results to develop the stiffness matrix, including axial deformation effects, for an arbitrarily oriented beam element. We will also consider frames with inclined or skewed supports.
Two-Dimensional Arbitrarily Oriented Beam Element
We can derive the stiffness matrix for an arbitrarily oriented beam element, shown in the figure below, in a manner similar to that used for the bar element. The local axes x and y are located along the beam element and transverse to the beam element, respectively, and the global axes x and y are located to be convenient for the total structure.
The transformation from local displacements to global displacements is given in matrix form as:
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 182
θθ
sincos
ˆˆ
==
⎭⎬⎫
⎩⎨⎧⎥⎦
⎤⎢⎣
⎡−
=⎪⎭
⎪⎬⎫
⎪⎩
⎪⎨⎧
SC
dd
CSSC
dd
y
x
y
x
Using the second equation for the beam element, we can relate local nodal degrees of freedom to global degree of freedom:
1
1 1
1 1
22
22
2
ˆ 0 0 0 0ˆ 0 0 1 0 0 0 ˆ
ˆ 0 0 0 00 0 0 0 0 1ˆ
φ φ
φφ
⎧ ⎫⎪ ⎪⎧ ⎫ −⎡ ⎤ ⎪ ⎪⎪ ⎪ ⎢ ⎥ ⎪ ⎪⎪ ⎪ ⎢ ⎥= = − +⎨ ⎬ ⎨ ⎬⎢ ⎥−⎪ ⎪ ⎪ ⎪⎢ ⎥⎪ ⎪ ⎪ ⎪⎣ ⎦⎩ ⎭ ⎪ ⎪⎩ ⎭
X
y y
y x yXy
y
dd dS C
d Sd CddS Cdd
For a beam we will define the following as the transformation matrix:
⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢
⎣
⎡
−
−
=
10000000000001000000
CS
CS
T
Notice that the rotations are not affected by the orientation of the beam. Substituting the above transformation into the general form of the stiffness matrix
TkTk T ˆ= gives:
Let’s know consider the effects of an axial force in the general beam transformation.
2 2
2 2
2 2
3 2 2
2 2
2 2
12 12 6 12 12 612 12 6 12 12 66 6 4 6 6 212 12 6 12 12 6
12 12 6 12 12 66 6 2 6 6 4
⎡ ⎤− − − −⎢ ⎥− −⎢ ⎥⎢ ⎥− −
= ⎢ ⎥− −⎢ ⎥
⎢ ⎥− − − −⎢ ⎥− −⎢ ⎥⎣ ⎦
S SC LS S SC LSSC C LC SC C LCLS LC L LS LC LEIk
L S SC LS S SC LSSC C LC SC C LCLS LC L LS LC L
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 183
Recall the simple axial deformation, define in the spring element:
⎭⎬⎫
⎩⎨⎧⎥⎦
⎤⎢⎣
⎡ −−
=⎭⎬⎫
⎩⎨⎧
x
x
x
x
dd
LAE
ff
2
1
2
1
ˆˆ
11
11
ˆˆ
Combining the axial effects with the shear force and bending moment effects, in local coordinates, gives:
where
321 LEIC
LAEC ==
11 1 1
11 2 2 2 22 2
1 2 2 2 2 1
1 1 22
2 2 2 222 2 2
2 2 2 22 2
ˆˆ 0 0 0 0ˆˆ 0 12 6 0 12 6ˆˆ 0 6 4 0 6 2
ˆ ˆ0 0 0 00 12 6 0 12 6 ˆˆ0 6 2 0 6 4ˆ ˆ
φ
φ
⎧ ⎫⎧ ⎫ −⎡ ⎤ ⎪ ⎪⎪ ⎪ ⎢ ⎥ ⎪ ⎪−⎪ ⎪ ⎢ ⎥ ⎪⎪ ⎪ ⎢ ⎥−⎪ ⎪ ⎪=⎨ ⎬ ⎨ ⎬⎢ ⎥−⎪ ⎪ ⎪⎢ ⎥⎪ ⎪ ⎪⎢ ⎥− − −⎪ ⎪ ⎪⎢ ⎥
−⎣ ⎦⎪ ⎪ ⎪⎩ ⎭ ⎩
xx
xy
xx
yy
df C Cdf C LC C LC
m LC C L LC C LC C df
C LC C LC dfLC C L LC C Lm
⎪⎪
⎪⎪⎪⎪⎭
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 184
Therefore:
⎥⎥⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢⎢⎢
⎣
⎡
−−−−
−−−
−
=
222
222
2222
11
222
222
2222
11
460260612061200000
26046061206120
0000
ˆ
LCLCLCLCLCCLCC
CCLCLCLCLC
LCCLCCCC
k
The above stiffness matrix include the effects of axial force in the x direction, shear force in the y , and bending moment about the z axis. The local degrees of freedom may be related to the global degrees of freedom by:
⎪⎪⎪
⎭
⎪⎪⎪
⎬
⎫
⎪⎪⎪
⎩
⎪⎪⎪
⎨
⎧
⎥⎥⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢⎢⎢
⎣
⎡
−
−
=
⎪⎪⎪⎪
⎭
⎪⎪⎪⎪
⎬
⎫
⎪⎪⎪⎪
⎩
⎪⎪⎪⎪
⎨
⎧
2
2
2
1
1
1
2
2
2
1
1
1
1000000000000000010000000000
ˆˆˆˆˆˆ
φ
φ
φ
φ
y
x
x
x
y
x
x
x
dd
dd
CSSC
CSSC
dd
dd
where the transformation matrix, including axial effects is:
⎥⎥⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢⎢⎢
⎣
⎡
−
−
=
1000000000000000010000000000
CSSC
CSSC
T
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 185
Substituting the above transformation into the general form of the stiffness matrix TkTk T ˆ= gives:
The analysis of a rigid plane frame can be undertaken by applying stiffness matrix. A rigid plane frame is defined here as a series of beam elements rigidly connected to each other; that is, the original angles made between elements at their joints remain unchanged after the deformation. Furthermore, moments are transmitted from one element to another at the joints. Hence, moment continuity exists at the rigid joints. In addition, the element centroids, as well as the applied loads, lie in a common plane. We observe that the element stiffnesses of a frame are functions of E, A, L, I, and the angle of orientation θ of the element with respect to the global-coordinate axes.
EkL
=
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 186
Rigid Plane Frame Example
Consider the frame shown in the figure below.
The frame is fixed at nodes 1 and 4 and subjected to a positive horizontal force of 10,000 lb applied at node 2 and to a positive moment of 5,000 lb-in. applied at node 3. Let E = 30 x 106 psi and A = 10 in.2 for all elements, and let I = 200 in.4 for elements 1 and 3, and I = 100 in.4 for element 2. Element 1: The angle between x and x is 90°
10 == SC where
( )32
22 0.10120
)200(66167.0120
)200(1212 inLIin
LI
====
3
6
/000,250120
1030 inlbLE
=×
=
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 187
Therefore, for element 1:
(1)
1 1 1 2 2 2
0.167 0 10 0.167 0 100 10 0 0 10 010 0 800 10 0 400
250,0000.167 0 10 0.167 0 10
0 10 0 0 10 010 0 400 10 0 800
φ φ
− − −⎡ ⎤⎢ ⎥−⎢ ⎥⎢ ⎥−
= ⎢ ⎥−⎢ ⎥⎢ ⎥−⎢ ⎥
−⎣ ⎦
d d d dx y x y
lbk in
Element 2: The angle between x and x is 0°
01 == SC
( )32
22 0.5120
)100(660835.0120
)100(1212 inLIin
LI
====
Therefore, for element 2:
inlbk
yd
xd
yd
xd
⎥⎥⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢⎢⎢
⎣
⎡
−−−
−−
−
=
400502005050835.0050835.00
00100010200504005050835.0050835.0000100010
000,250)2(
333222φφ
Element 3: The angle between x and x is 270°
10 −== SC
( )32
22 0.10120
)200(66167.0120
)200(1212 inLIin
LI
====
3
6
/000,250120
1030 inlbLE
=×
=
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 188
Therefore, for element 3:
inlbk
yd
xd
yd
xd
⎥⎥⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢⎢⎢
⎣
⎡
−−
−−−−
−−
=
80001040001001000100100167.0100167.0
40001080001001000100
100167.0100167.0
000,250)3(
444333φφ
The boundary conditions for this problem are:
1 1 1 4 4 4 0x y x yd d d dφ φ= = = = = =
After applying the boundary conditions the global beam equations reduce to:
⎪⎪⎪
⎭
⎪⎪⎪
⎬
⎫
⎪⎪⎪
⎩
⎪⎪⎪
⎨
⎧
⎥⎥⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢⎢⎢
⎣
⎡
−−−−
−−
−−
×=
⎪⎪⎪
⎭
⎪⎪⎪
⎬
⎫
⎪⎪⎪
⎩
⎪⎪⎪
⎨
⎧
3
3
3
2
2
2
5
12005102005050835.10050835.00
100167.10001020050120051050835.0050835.1000010100167.10
105.2
000,50000000,10
φ
φ
y
x
y
x
dd
dd
Solving the above equations gives:
2
2
2
3
3
3
0.2110.00148
0.001530.209
0.001480.00149
φ
φ
⎧ ⎫ ⎧ ⎫⎪ ⎪ ⎪ ⎪⎪ ⎪ ⎪ ⎪⎪ ⎪ ⎪ ⎪−
=⎨ ⎬ ⎨ ⎬⎪ ⎪ ⎪ ⎪⎪ ⎪ ⎪ ⎪−⎪ ⎪ ⎪ ⎪
−⎩ ⎭⎩ ⎭
x
y
x
y
d ind in
radd ind in
rad
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 189
Element 1: The element force-displacement equations can be obtained using dTkf ˆˆ = . Therefore, dT is:
⎪⎪⎪
⎭
⎪⎪⎪
⎬
⎫
⎪⎪⎪
⎩
⎪⎪⎪
⎨
⎧
−−
=
⎪⎪⎪
⎭
⎪⎪⎪
⎬
⎫
⎪⎪⎪
⎩
⎪⎪⎪
⎨
⎧
−===
===
⎥⎥⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢⎢⎢
⎣
⎡
−
−
=
radinin
radind
ind
dd
dT
y
x
y
x
00153.0211.0
00148.0000
00153.000148.0
211.0000
100000001000010000000100000001000010
2
2
2
1
1
1
φ
φ
Recall the elemental stiffness matrix is:
⎥⎥⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢⎢⎢
⎣
⎡
−−−−
−−−
−
=
222
222
2222
11
222
222
2222
11
460260612061200000
26046061206120
0000
ˆ
LCLCLCLCLCCLCC
CCLCLCLCLC
LCCLCCCC
k
Therefore, the local force-displacement equations are:
⎪⎪⎪
⎭
⎪⎪⎪
⎬
⎫
⎪⎪⎪
⎩
⎪⎪⎪
⎨
⎧
−−
⎥⎥⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢⎢⎢
⎣
⎡
−−−−
−−
−−
×==
radinin
dTkf
00153.0211.0
00148.0000
80010040010010167.0010167.00
10010001040010080010010167.0010167.0000100010
105.2ˆˆ 5)1(
Simplifying the above equations gives:
⎪⎪⎪
⎭
⎪⎪⎪
⎬
⎫
⎪⎪⎪
⎩
⎪⎪⎪
⎨
⎧
⋅−
⋅
−
=
⎪⎪⎪⎪
⎭
⎪⎪⎪⎪
⎬
⎫
⎪⎪⎪⎪
⎩
⎪⎪⎪⎪
⎨
⎧
inklblbinklblb
mffmff
y
x
y
x
223990,4700,3
376990,4700,3
ˆ
ˆˆˆ
ˆˆ
2
2
2
1
1
1
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 190
Element 2: The element force-displacement equations are:
⎪⎪⎪
⎭
⎪⎪⎪
⎬
⎫
⎪⎪⎪
⎩
⎪⎪⎪
⎨
⎧
−−
−
−
=
⎪⎪⎪
⎭
⎪⎪⎪
⎬
⎫
⎪⎪⎪
⎩
⎪⎪⎪
⎨
⎧
−=−==
−===
⎥⎥⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢⎢⎢
⎣
⎡
=
radinin
radinin
radind
indradind
ind
dT
y
x
y
x
00149.000148.0
209.000153.0
00148.0211.0
00149.000148.0209.0
00153.000148.0
211.0
100000010000001000000100000010000001
3
3
3
2
2
2
φ
φ
Therefore, the local force-displacement equations are:
⎪⎪⎪
⎭
⎪⎪⎪
⎬
⎫
⎪⎪⎪
⎩
⎪⎪⎪
⎨
⎧
−−
−
⎥⎥⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢⎢⎢
⎣
⎡
−−−−
−−
−−
×==
radinin
radinin
dTkf
00149.000148.0
209.000153.0
00148.0211.0
400502005050833.0050833.00
00100010200504005050833.0050833.0000100010
105.2ˆˆ 5)2(
Simplifying the above equations gives:
⎪⎪⎪
⎭
⎪⎪⎪
⎬
⎫
⎪⎪⎪
⎩
⎪⎪⎪
⎨
⎧
⋅−
−⋅−
−
=
⎪⎪⎪⎪
⎭
⎪⎪⎪⎪
⎬
⎫
⎪⎪⎪⎪
⎩
⎪⎪⎪⎪
⎨
⎧
inklblbinklblb
mffmff
y
x
y
x
221700,3010,5
223700,3010,5
ˆ
ˆˆˆ
ˆˆ
3
3
3
2
2
2
Element 3: The element force-displacement equations are:
⎪⎪⎪
⎭
⎪⎪⎪
⎬
⎫
⎪⎪⎪
⎩
⎪⎪⎪
⎨
⎧
−=
⎪⎪⎪
⎭
⎪⎪⎪
⎬
⎫
⎪⎪⎪
⎩
⎪⎪⎪
⎨
⎧
===
−=−==
⎥⎥⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢⎢⎢
⎣
⎡
−
−
=
000
00149.0209.0
00148.0
000
00149.000148.0209.0
100000001000010000000100000001000010
4
4
4
3
3
3
radin
in
dd
radind
ind
dT
y
x
y
x
φ
φ
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 191
Therefore, the local force-displacement equations are:
⎪⎪⎪
⎭
⎪⎪⎪
⎬
⎫
⎪⎪⎪
⎩
⎪⎪⎪
⎨
⎧
−
⎥⎥⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢⎢⎢
⎣
⎡
−−−−
−−
−−
×==
000
00149.0209.0
00148.0
80010040010010167.0010167.00
10010001040010080010010167.0010167.0000100010
105.2ˆˆ 5)3( radin
in
dTkf
Simplifying the above equations gives:
⎪⎪⎪
⎭
⎪⎪⎪
⎬
⎫
⎪⎪⎪
⎩
⎪⎪⎪
⎨
⎧
⋅−−
⋅=
⎪⎪⎪⎪
⎭
⎪⎪⎪⎪
⎬
⎫
⎪⎪⎪⎪
⎩
⎪⎪⎪⎪
⎨
⎧
inklblbinklblb
mffmff
y
x
y
x
375010,5700,3
226010,5700,3
ˆ
ˆˆˆ
ˆˆ
4
4
4
3
3
3
Rigid Plane Frame Example
Consider the frame shown in the figure below.
The frame is fixed at nodes 1 and 3 and subjected to a positive distributed load of 1,000 lb/ft applied along element 2. Let E = 30 x 106 psi and A = 100 in.2 for all elements, and let I = 1,000 in.4 for all elements.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 192
First we need to replace the distributed load with a set of equivalent nodal forces and moments acting at nodes 2 and 3. For a beam with both end fixed, subjected to a uniform distributed load, w, the nodal forces and moments are:
2 3(1,000 / )40 20
2 2y ywL lb ft ftf f k= = − = − = −
2 2
2 3(1,000 / )(40 ) 133,333 1,600
12 12wL lb ft ftm m lb ft k in= − = − = − = − ⋅ = ⋅
If we consider only the parts of the stiffness matrix associated with the three degrees of freedom at node 2, we get: Element 1: The angle between x and x is 45º
707.0707.0 == SC where
( )6
3 222
3
30 10 12 12(1,000)58.93 / 0.0463509 12 30 2
6 6(1,000) 11.7855112 30 2
E Ik in inL L
I inL
×= = = =
×
= =×
Therefore, for element 1:
2 2 2
(1)
50.02 49.98 8.3358.93 49.98 50.02 8.33
8.33 8.33 4000
x yd d
kk in
φ
⎡ ⎤⎢ ⎥= −⎢ ⎥⎢ ⎥−⎣ ⎦
Simplifying the above equation:
2 2 2
(1)
2,948 2,945 4912,945 2,948 491491 491 235,700
x yd d
kk in
φ
⎡ ⎤⎢ ⎥= −⎢ ⎥⎢ ⎥−⎣ ⎦
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 193
Element 2: The angle between x and x is 0°
01 == SC where
( )
63 2
22
3
30 10 12 12(1,000)62.5 / 0.0521480 12 40
6 6(1,000) 12.512 40
E Ik in inL L
I inL
×= = = =
×
= =×
Therefore, for element 2: 2 2 2
(2)
100 0 062.50 0 0.052 12.5
0 12.5 4,000
x yd d
kk in
φ
⎡ ⎤⎢ ⎥= ⎢ ⎥⎢ ⎥⎣ ⎦
Simplifying the above equation:
2 2 2
(2)
6,250 0 00 3.25 781.250 781.25 250,000
x yd d
kk in
φ
⎡ ⎤⎢ ⎥= ⎢ ⎥⎢ ⎥⎣ ⎦
The global beam equations reduce to:
⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡=
⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧
⋅−−
2
2
2
700,485290491290951,2945,2491945,2198,9
600,1200
φy
x
dd
inkk
Solving the above equations gives:
⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧
−−=
⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧
radin
indd
y
x
0033.00097.0
0033.0
2
2
2
φ
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 194
Element 1: The element force-displacement equations can be obtained using dTkf ˆˆ = . Therefore, dT is:
⎪⎪⎪
⎭
⎪⎪⎪
⎬
⎫
⎪⎪⎪
⎩
⎪⎪⎪
⎨
⎧
−−−
=
⎪⎪⎪
⎭
⎪⎪⎪
⎬
⎫
⎪⎪⎪
⎩
⎪⎪⎪
⎨
⎧
−−
⎥⎥⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢⎢⎢
⎣
⎡
−
−
=
radinin
radin
indT
0033.00092.0
00452.0000
0033.00097.0
0033.0000
1000000707.0707.00000707.0707.00000001000000707.0707.00000707.0707.0
Recall the elemental stiffness matrix is a function of values C1, C2, and L
( ) ink
ink
LEIC
LAEC 2273.0
23012
)000,1(1030893,5230121030)100(
3
6
32
6
1 =×
×===
××
==
Therefore, the local force-displacement equations are:
(1)
5,893 0 10 5,893 0 0 00 2.730 694.8 0 2.730 694.8 0
10 694.8 117,900 0 694.8 117,000 0ˆ ˆ5,893 0 0 5,983 0 0 0.00452
0 2.730 694.8 0 2.730 694.8 0.00920 694.8 117,000 0 694.8 235,800 0.0033
−⎡ ⎤ ⎧⎢ ⎥ ⎪−⎢ ⎥ ⎪⎢ ⎥ ⎪−
= = ⎨⎢ ⎥− −⎢ ⎥⎢ ⎥− − − −⎢ ⎥
− −⎣ ⎦
f kTdin
inrad
⎫⎪⎪⎪⎬
⎪ ⎪⎪ ⎪⎪ ⎪⎩ ⎭
Simplifying the above equations gives:
1
1
1
2
2
2
ˆ 26.64ˆ 2.268ˆ 389.1ˆ 26.64
2.268ˆ778.2ˆ
⎧ ⎫ ⎧ ⎫⎪ ⎪ ⎪ ⎪−⎪ ⎪ ⎪ ⎪⎪ ⎪ ⎪ ⎪− ⋅⎪ ⎪ =⎨ ⎬ ⎨ ⎬−⎪ ⎪ ⎪ ⎪⎪ ⎪ ⎪ ⎪⎪ ⎪ ⎪ ⎪
− ⋅⎩ ⎭⎪ ⎪⎩ ⎭
x
y
x
y
f kf km k in
kfkf
k inm
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 195
Element 2: The element force-displacement equations are:
⎪⎪⎪
⎭
⎪⎪⎪
⎬
⎫
⎪⎪⎪
⎩
⎪⎪⎪
⎨
⎧
−−
=
⎪⎪⎪
⎭
⎪⎪⎪
⎬
⎫
⎪⎪⎪
⎩
⎪⎪⎪
⎨
⎧
−−
⎥⎥⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢⎢⎢
⎣
⎡
=
000
0033.00097.0
0033.0
000
0033.00097.0
0033.0
100000010000001000000100000010000001
radin
in
radin
in
dT
Recall the elemental stiffness matrix is a function of values C1, C2, and L
( ) ink
ink
LEIC
LAEC 2713.0
4012)000,1(1030250,6
40121030)100(
3
6
32
6
1 =×
×===
××
==
Therefore, the local force-displacement equations are:
(2)
6,250 0 0 6,250 0 0 0.00330 3.25 781.1 0 3.25 781.1 0.00970 781.1 250,000 0 781.1 125,000 0.0033ˆ ˆ
6,250 0 0 6,250 0 0 00 3.25 781.1 0 3.25 781.1 00 781.1 125,000 0 781.1 250,00 0
− −⎡ ⎤ ⎧ ⎫⎢ ⎥ ⎪ ⎪− −⎢ ⎥ ⎪ ⎪⎢ ⎥ ⎪− −
= = ⎨ ⎬⎢ ⎥− ⎪⎢ ⎥⎪⎢ ⎥− − −⎪⎢ ⎥
−⎣ ⎦ ⎩
inin
radf kTd
⎪
⎪⎪⎪⎭
Simplifying the above equations gives:
⎪⎪⎪
⎭
⎪⎪⎪
⎬
⎫
⎪⎪⎪
⎩
⎪⎪⎪
⎨
⎧
⋅−
−⋅−
−
=
inkkkinkkk
dk
50.41258.263.20
57.83258.263.20
ˆˆ
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 196
To obtain the actual element local forces, we must subtract the equivalent nodal forces.
2
2
2
3
3
3
ˆ 20.63 0 20.63ˆ 2.58 20 17.42ˆ 832.57 1600 767.4ˆ 20.63 0 20.63
2.58 20 22.58ˆ412.50 1600 2,013ˆ
⎧ ⎫ ⎧ ⎫ ⎧ ⎫⎪ ⎪ ⎪ ⎪ ⎪ ⎪− −⎪ ⎪ ⎪ ⎪ ⎪ ⎪⎪ ⎪ ⎪ ⎪ ⎪ ⎪− ⋅ − ⋅ ⋅⎪ ⎪ = − =⎨ ⎬ ⎨ ⎬ ⎨ ⎬− −⎪ ⎪ ⎪ ⎪ ⎪ ⎪⎪ ⎪ ⎪ ⎪ ⎪ ⎪−⎪ ⎪ ⎪ ⎪ ⎪ ⎪
− ⋅ ⋅ − ⋅⎩ ⎭ ⎩ ⎭⎪ ⎪⎩ ⎭
x
y
x
y
f k kf k k km k in k in k in
k kfk k kfk in k in k inm
⎧ ⎫⎪ ⎪⎪ ⎪⎪ ⎪⎨ ⎬⎪ ⎪⎪ ⎪⎪ ⎪⎩ ⎭
Rigid Plane Frame Example
Consider the frame shown in the figure below. In this example will illustrate the equivalent joint force replacement method for a frame subjected to a load acting on an element instead of at one of the joints of the structure. Since no distributed loads are present, the point of application of the concentrated load could be treated as an extra joint in the analysis.
This approach has the disadvantage of increasing the total number of joints, as well as the size of the total structure stiffness matrix K. For small structures solved by computer, this does not pose a problem. However, for very large structures, this might reduce the maximum size of the structure that could be analyzed.
The frame is fixed at nodes 1, 2, and 3 and subjected to a concentrated load of 15 k applied at mid-length of element 1. Let E = 30 x 106 psi, A = 8 in2, and let I = 800 in4 for all elements.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 197
Solution Procedure 1. Express the applied load in the element 1 local coordinate system (here
x is directed from node 1 to node 4).
2. Next, determine the equivalent joint forces at each end of element 1,
using the table in Appendix D (see figure below).
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 198
3. Then transform the equivalent joint forces from the local coordinate
system forces into the global coordinate system forces, using the equation fTf T ˆ= . These global joint forces are shown below.
4. Then we analyze the structure, using the equivalent joint forces (plus
actual joint forces, if any) in the usual manner.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 199
5. The final internal forces developed at the ends of each element may be obtained by subtracting Step 2 joint forces from Step 4 joint forces.
Element 1: The angle between x and x is 63.43°
895.0447.0 == SC where
( )2 3
2212 12(800) 6 6(800)0.0334 8.95
44.7 1244.7 12= = = =
××
I Iin inL L
6
330 10 55.9 /44.7 12
E k inL
×= =
×
Therefore, for element 1:
(1)
4 4 4
90.0 178 448178 359 244448 244 179,000
kin
d dx y
k
φ
⎡ ⎤⎢ ⎥= −⎢ ⎥⎢ ⎥−⎣ ⎦
Element 2: The angle between x and x is 116.57°
895.0447.0 =−= SC where
( )2 3
2212 12(800) 6 6(800)0.0334 8.95
44.7 1244.7 12= = = =
××
I Iin inL L
6
330 10 55.9 /44.7 12
E k inL
×= =
×
Therefore, for element 2:
(2)
4 4 4
90.0 178 448178 359 244448 244 179,000
kin
d dx y
k
φ
−⎡ ⎤⎢ ⎥= −⎢ ⎥⎢ ⎥⎣ ⎦
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 200
Element 3: The angle between x and x is 0° (The author of your textbook directed the element from node 4 to 3. In general, as we have discussed in class, we usually number the element numerically or from 3 to 4. In this case the angle between x and x is 180°)
6330 101 0 50 /
50 12EC S k inL
×= = = =
×
( )2 3
2212 12(800) 6 6(800)0.0267 8.0
50 1250 12= = = =
××
I Iin inL L
Therefore, for element 3:
(2)
4 4 4
400 0 00 1.334 4000 400 160,000
kin
d dx y
k
φ
⎡ ⎤⎢ ⎥= ⎢ ⎥⎢ ⎥⎣ ⎦
The global beam equations reduce to:
4
4
4
7.5 582 0 8960 0 719 400
900 896 400 518,000 φ
−⎧ ⎫ ⎡ ⎤ ⎧ ⎫⎪ ⎪ ⎪ ⎪⎢ ⎥=⎨ ⎬ ⎨ ⎬⎢ ⎥⎪ ⎪ ⎪ ⎪⎢ ⎥− ⋅⎩ ⎭ ⎣ ⎦ ⎩ ⎭
x
y
k dd
k in
Solving the above equations gives:
4
4
4
0.01030.0009560.00172φ
−⎧ ⎫ ⎧ ⎫⎪ ⎪ ⎪ ⎪=⎨ ⎬ ⎨ ⎬⎪ ⎪ ⎪ ⎪−⎩ ⎭ ⎩ ⎭
x
y
d ind in
rad
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 201
Element 1: The element force-displacement equations can be obtained using ˆ ˆf kTd= . Therefore, dT is:
895.0447.0
1000000000000000010000000000
==
⎥⎥⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢⎢⎢
⎣
⎡
−
−
= SC
CSSC
CSSC
T
0.447 0.895 0 0 0 0 0 00.895 0.447 0 0 0 0 0 00 0 1 0 0 0 0 00 0 0 0.447 0.895 0 0.0103 0.003740 0 0 0.895 0.447 0 0.000956 0.009630 0 0 0 0 1 0.00172 0.00172
⎡ ⎤ ⎧ ⎫ ⎧ ⎫⎢ ⎥ ⎪ ⎪ ⎪ ⎪−⎢ ⎥ ⎪ ⎪ ⎪ ⎪⎢ ⎥ ⎪ ⎪ ⎪ ⎪
= =⎨ ⎬ ⎨ ⎬⎢ ⎥ − −⎪ ⎪ ⎪ ⎪⎢ ⎥⎪ ⎪ ⎪ ⎪⎢ ⎥−⎪ ⎪ ⎪ ⎪⎢ ⎥− −⎣ ⎦ ⎩ ⎭ ⎩ ⎭
Tdin inin in
rad rad
Recall the elemental stiffness matrix is:
⎥⎥⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢⎢⎢
⎣
⎡
−−−−
−−−
−
=
222
222
2222
11
222
222
2222
11
460260612061200000
26046061206120
0000
ˆ
LCLCLCLCLCCLCC
CCLCLCLCLC
LCCLCCCC
k
( ) ink
ink
LEIC
LAEC 155.0
72.4412)800(10302.447
72.44121030)8(
3
6
32
6
1 =××
===×
×==
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 202
Therefore, the local force-displacement equations are:
⎪⎪⎪
⎭
⎪⎪⎪
⎬
⎫
⎪⎪⎪
⎩
⎪⎪⎪
⎨
⎧
−
−
⎥⎥⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢⎢⎢
⎣
⎡
−−−−
−−−
−
==
radinin
dTkf
00172.000963.000374.0
000
000,1795.5000490,895.50005.500868.105.500868.10
0044700447490,895.5000000,1795.5000
5.500868.105.500868.100044700447
ˆˆ)1(
Simplifying the above equations gives:
(1)
1.670.88
158ˆ ˆ ˆ1.670.88
311
⎧ ⎫⎪ ⎪−⎪ ⎪⎪ ⎪− ⋅
= = ⎨ ⎬−⎪ ⎪⎪ ⎪⎪ ⎪− ⋅⎩ ⎭
kk
k inf kd
kk
k in
To obtain the actual element local forces, we must subtract the equivalent nodal forces.
1
1
1
4
4
4
ˆ 1.67 3.36 5.03ˆ 0.88 6.71 7.59ˆ 158 900 1,058ˆ 1.67 3.36 1.68
0.88 6.71 5.83ˆ311 900 589ˆ
⎧ ⎫ −⎧ ⎫ ⎧ ⎫ ⎧⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪− −⎪ ⎪ ⎪ ⎪ ⎪ ⎪⎪ ⎪ ⎪ ⎪ ⎪ ⎪− ⋅ ⋅ − ⋅⎪ ⎪ = − =⎨ ⎬ ⎨ ⎬ ⎨ ⎬ ⎨− −⎪ ⎪ ⎪ ⎪ ⎪ ⎪⎪ ⎪ ⎪ ⎪ ⎪ ⎪ −⎪ ⎪ ⎪ ⎪ ⎪ ⎪
− ⋅ − ⋅ ⋅⎩ ⎭ ⎩ ⎭⎪ ⎪⎩ ⎭
x
y
x
y
f k k kf k k km k in k in k in
k k kfk k kf
k in k in k inm
⎫⎪
⎪ ⎪⎪ ⎪
⎬⎪ ⎪⎪ ⎪⎪ ⎪⎩ ⎭
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 203
Element 2: The element force-displacement equations can be obtained using dTkf ˆˆ = . Therefore, dT is:
895.0447.0
1000000000000000010000000000
=−=
⎥⎥⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢⎢⎢
⎣
⎡
−
−
= SC
CSSC
CSSC
T
⎪⎪⎪
⎭
⎪⎪⎪
⎬
⎫
⎪⎪⎪
⎩
⎪⎪⎪
⎨
⎧
−
=
⎪⎪⎪
⎭
⎪⎪⎪
⎬
⎫
⎪⎪⎪
⎩
⎪⎪⎪
⎨
⎧
−
−
⎥⎥⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢⎢⎢
⎣
⎡
−−−
−−−
=
radinin
radinin
dT
00172.000879.000546.0
000
00172.0000956.0
0103.0000
1000000447.0895.00000895.0447.00000001000000447.0895.00000895.0447.0
Therefore, the local force-displacement equations are:
( ) ink
ink
LEIC
LAEC 155.0
72.4412)800(10302.447
72.44121030)8(
3
6
32
6
1 =××
===×
×==
⎪⎪⎪
⎭
⎪⎪⎪
⎬
⎫
⎪⎪⎪
⎩
⎪⎪⎪
⎨
⎧
−⎥⎥⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢⎢⎢
⎣
⎡
−−−−
−−−
−
==
radinin
dTkf
00172.000879.000546.0
000
000,1795.5000490,895.50005.500868.105.500868.10
0044700447490,895.5000000,1795.5000
5.500868.105.500868.100044700447
ˆˆ)2(
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 204
Simplifying the above equations gives:
(2)
2.440.877
158ˆ ˆ ˆ2.440.877
312
⎧ ⎫⎪ ⎪−⎪ ⎪⎪ ⎪− ⋅
= = ⎨ ⎬⎪ ⎪⎪ ⎪⎪ ⎪− ⋅⎩ ⎭
kk
k inf kd
kk
k in
Since there are no applied loads on element 2, there are no equivalent nodal forces to account for. Therefore, the above equations are the final local nodal forces
Element 3: The element force-displacement equations can be obtained using ˆ ˆf kTd= . Therefore, dT is:
⎪⎪⎪
⎭
⎪⎪⎪
⎬
⎫
⎪⎪⎪
⎩
⎪⎪⎪
⎨
⎧
−
−
=
⎪⎪⎪
⎭
⎪⎪⎪
⎬
⎫
⎪⎪⎪
⎩
⎪⎪⎪
⎨
⎧
−
−
⎥⎥⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢⎢⎢
⎣
⎡
=
000
00172.0000956.0
0103.0
000
00172.0000956.0
0103.0
100000010000001000000100000010000001
radinin
radinin
dT
Therefore, the local force-displacement equations are:
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 205
( ) ink
ink
LEIC
LAEC 111.0
5012)800(1030400
50121030)8(
3
6
32
6
1 =×
×===
××
==
⎪⎪⎪
⎭
⎪⎪⎪
⎬
⎫
⎪⎪⎪
⎩
⎪⎪⎪
⎨
⎧
−
−
⎥⎥⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢⎢⎢
⎣
⎡
−−−−
−−−
−
==
000
00172.0000956.0
0103.0
000,1604000000,804000400335.10400335.100040000400000,804000000,1604000
400335.10400335.100040000400
ˆˆ)3(
radinin
dTkf
Simplifying the above equations gives:
(3)
4.120.687
275ˆ ˆ ˆ4.120.687
137
−⎧ ⎫⎪ ⎪−⎪ ⎪⎪ ⎪− ⋅
= = ⎨ ⎬⎪ ⎪⎪ ⎪⎪ ⎪− ⋅⎩ ⎭
kk
k inf kd
kk
k in
Since there are no applied loads on element 3, there are no equivalent nodal forces to account for. Therefore, the above equations are the final local nodal forces. The free-body diagrams are shown below.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 206
Rigid Plane Frame Example
The frame shown on the right is fixed at nodes 2 and 3 and subjected to a concentrated load of 500 kN applied at node 1. For the bar, A = 1 x 10-3 m2, for the beam, A = 2 x 10-3 m2, I = 5 x 10-5 m4, and L = 3 m. Let E = 210 GPa for both elements.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 207
Beam Element 1: The angle between x and x is 0°
01 == SC where
345
252
5
2 103
)105(661067.6)3(
)105(1212 mLIm
LI −
−−
−
=×
=×=×
=
36
6
/10703
10210 mkNLE
×=×
=
Therefore, for element 1:
mkNk
yx dd
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡×=
20.010.0010.0067.00002
1070 3)1(
111 φ
Bar Element 2: The angle between x and x is 45°
707.0707.0 == SC
where
( )m
kNm
mkNmk
yx dd
⎥⎦
⎤⎢⎣
⎡×=
−
5.05.05.05.0
24.4/1021010 2623
)2(
11
mkNk
yx dd
⎥⎦
⎤⎢⎣
⎡×=
354.0354.0354.0354.0
1070 3)2(
11
Assembling the elemental stiffness matrices we obtain the global stiffness matrix
mkNK⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡×=
20.010.0010.0421.0354.00354.0354.2
1070 3
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 208
The global equations are:
⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡×=
⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧−
1
1
13
20.010.0010.0421.0354.00354.0354.2
10700
5000
φy
x
dd
mkNkN
Solving the above equations gives:
⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧−=
⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧
radmm
dd
y
x
0113.00225.0
00388.0
1
1
1
φ
Bar Element: The bar element force-displacement equations can be obtained using ˆ ˆf kTd= .
Therefore, the forces in the bar element are:
( ) kNSdCdL
AEf yxx 670ˆ111 −=+=
( ) kNSdCdL
AEf yxx 670ˆ113 =+−=
Beam Element: The beam element force-displacement equations can be obtained using ˆ ˆ ˆf kd= . Since the local axis coincides with the global coordinate system, and the displacements at node 2 are zero. Therefore, the local force-displacement equations are:
⎪⎪⎭
⎪⎪⎬
⎫
⎪⎪⎩
⎪⎪⎨
⎧
⎥⎦
⎤⎢⎣
⎡⎥⎦
⎤⎢⎣
⎡−
−=
⎭⎬⎫
⎩⎨⎧
y
x
y
x
x
x
dddd
SCSC
LAE
ff
3
3
1
1
3
1
0000
1111
ˆˆ
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 209
32
1
222
222
2222
11
222
222
2222
11
460260612061200000
26046061206120
0000
ˆ
LEIC
LAEC
LCLCLCLCLCCLCC
CCLCLCLCLC
LCCLCCCC
k
=
=
⎥⎥⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢⎢⎢
⎣
⎡
−−−−
−−−
−
=
⎪⎪⎪
⎭
⎪⎪⎪
⎬
⎫
⎪⎪⎪
⎩
⎪⎪⎪
⎨
⎧
⋅−
⎥⎥⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢⎢⎢
⎣
⎡
−−−−
−−−
−
×==
000
0113.00225.0
00388.0
20.010.0010.010.0010.0067.0010.0067.00
00200210.010.0020.010.0010.0067.0010.0067.00002002
1070ˆˆˆ 3)1(
mkNmm
dkf
Substituting numerical values into the above equations gives:
⎪⎪⎪
⎭
⎪⎪⎪
⎬
⎫
⎪⎪⎪
⎩
⎪⎪⎪
⎨
⎧
⋅−
−
−
=
⎪⎪⎪⎪
⎭
⎪⎪⎪⎪
⎬
⎫
⎪⎪⎪⎪
⎩
⎪⎪⎪⎪
⎨
⎧
mkNkNkN
kNkN
mffmff
y
x
y
x
3.785.26
4730.05.26
473
ˆ
ˆˆˆ
ˆˆ
2
2
2
1
1
1
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 210
Rigid Plane Frame Example
Consider the frame shown in the figure below.
The frame is fixed at nodes 1 and 3 and subjected to a moment of 20 kN-m applied at node 2. Assume A = 2 x 10-2 m2, I = 2 x 10-4 m4, and E = 210 GPa for all elements.
Beam Element 1: The angle between x and x is 90°
10 == SC where
344
242
4
2 1034
)102(66105.1)4(
)102(1212 mLIm
LI −
−−
−
×=×
=×=×
=
37
6
/1025.54
10210 mkNLE
×=×
=
Therefore, the stiffness matrix for element 1, considering only the parts associated with node 2, is:
mkNk
yx dd
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡×=
08.0003.002003.00015.0
1025.5 5)1(
222 φ
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 211
Beam Element 2: The angle between x and x is 0°
01 == SC where
344
252
4
2 104.25
)102(66106.9)5(
)102(1212 mLIm
LI −
−−
−
×=×
=×=×
=
37
6
/102.45
10210 mkNLE
×=×
=
Therefore, the stiffness matrix for element 2, considering only the parts associated with node 2, is:
mkNk
yx dd
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡×=
08.0024.00024.00096.00002
102.4 5)2(
222 φ
Assembling the elemental stiffness matrices we obtain the global stiffness matrix:
mkNK⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡=
0756.00101.00158.00101.00500.100158.008480.0
106
The global equations are:
⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡=
⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧
⋅ 2
2
26
0756.00101.00158.00101.00500.100158.008480.0
1020
00
φy
x
dd
mkN
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 212
Solving the above equations gives:
62
62
42
4.95 102.56 102.66 10φ
−
−
−
⎧ ⎫− ×⎧ ⎫⎪ ⎪⎪ ⎪ = − ×⎨ ⎬ ⎨ ⎬
⎪ ⎪ ⎪ ⎪×⎩ ⎭ ⎩ ⎭
x
y
d md m
rad
Element 1: The beam element force-displacement equations can be obtained using ˆ ˆf kTd= .
6 6
6 6
4 4
0 1 0 0 0 0 0 01 0 0 0 0 0 0 0
0 0 1 0 0 0 0 00 0 0 0 1 0 4.95 10 2.56 100 0 0 1 0 0 2.56 10 4.95 100 0 0 0 0 1 2.66 10 2.66 10
− −
− −
− −
⎡ ⎤ ⎧ ⎫ ⎧ ⎫⎢ ⎥ ⎪ ⎪ ⎪ ⎪−⎢ ⎥ ⎪ ⎪ ⎪ ⎪⎢ ⎥ ⎪ ⎪ ⎪ ⎪
= =⎨ ⎬ ⎨ ⎬⎢ ⎥ − × − ×⎪ ⎪ ⎪ ⎪⎢ ⎥⎪ ⎪ ⎪ ⎪⎢ ⎥− − × ×⎪ ⎪ ⎪ ⎪⎢ ⎥
× ×⎣ ⎦ ⎩ ⎭ ⎩ ⎭
Tdm mm m
rad rad
Therefore, the local force-displacement equations are:
⎥⎥⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢⎢⎢
⎣
⎡
−−−−
−−−
−
=
222
222
2222
11
222
222
2222
11
460260612061200000
26046061206120
0000
ˆ
LCLCLCLCLCCLCC
CCLCLCLCLC
LCCLCCCC
k
( )
2 66
1
6 4
2 33
(2 10 )210 10 1.05 104
210 10 (2 10 ) 656.254
kNm
kNm
AECL
EICL
−
−
× ×= = = ×
× ×= = =
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 213
⎪⎪⎪
⎭
⎪⎪⎪
⎬
⎫
⎪⎪⎪
⎩
⎪⎪⎪
⎨
⎧
×××−
⎥⎥⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢⎢⎢
⎣
⎡
−−−−
−−−
−
×==
−
−
−
radmm
dTkf
4
6
63
)1(
1066.21095.41056.2
000
83043035.1035.10
002000020043083035.1035.100020000200
1025.5ˆˆ
Solving for the forces and moments gives:
⎪⎪⎪
⎭
⎪⎪⎪
⎬
⎫
⎪⎪⎪
⎩
⎪⎪⎪
⎨
⎧
⋅−
−⋅
=
⎪⎪⎪⎪
⎭
⎪⎪⎪⎪
⎬
⎫
⎪⎪⎪⎪
⎩
⎪⎪⎪⎪
⎨
⎧
mkNkNkNmkN
kNkN
mffmff
y
x
y
x
17.112.4
69.259.5
2.469.2
ˆ
ˆˆˆ
ˆˆ
2
2
2
1
1
1
Element 2: The beam element force-displacement equations can be obtained using ˆ ˆf kTd= .
⎪⎪⎪
⎭
⎪⎪⎪
⎬
⎫
⎪⎪⎪
⎩
⎪⎪⎪
⎨
⎧
××−×−
=
⎪⎪⎪
⎭
⎪⎪⎪
⎬
⎫
⎪⎪⎪
⎩
⎪⎪⎪
⎨
⎧
××−×−
⎥⎥⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢⎢⎢
⎣
⎡
=
−
−
−
−
−
−
radmm
radmm
dT
4
6
6
4
6
6
1066.21056.21095.4
000
1066.21056.21095.4
000
100000010000001000000100000010000001
Therefore, the local force-displacement equations are:
( )
2 66
1
6 4
2 33
(2 10 )210 10 0.84 105
210 10 (2 10 ) 3365
kNm
kNm
AECL
EICL
−
−
× ×= = = ×
× ×= = =
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 214
⎪⎪⎪
⎭
⎪⎪⎪
⎬
⎫
⎪⎪⎪
⎩
⎪⎪⎪
⎨
⎧
××−×−
⎥⎥⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢⎢⎢
⎣
⎡
−−−−
−−−
−
×==−
−
−
0001066.2
1056.21095.4
840.20440.2040.296.0040.296.00
0020000200440.20840.2040.296.0040.296.000020000200
102.4ˆˆ4
6
6
3)2(
radmm
dTkf
Solving for the forces and moments gives:
2
2
2
3
3
3
ˆ 4.16ˆ 2.69ˆ 8.92ˆ 4.16
2.69ˆ4.47ˆ
⎧ ⎫ ⎧ ⎫⎪ ⎪ ⎪ ⎪−⎪ ⎪ ⎪ ⎪⎪ ⎪ ⎪ ⎪⋅⎪ ⎪ =⎨ ⎬ ⎨ ⎬−⎪ ⎪ ⎪ ⎪⎪ ⎪ ⎪ ⎪⎪ ⎪ ⎪ ⎪
⋅⎩ ⎭⎪ ⎪⎩ ⎭
x
y
x
y
f kNf kNm kN m
kNfkNfkN mm
Inclined or Skewed Supports
If a support is inclined, or skewed, at some angle α for the global x axis, as shown below, the boundary conditions on the displacements are not in the global x-y directions but in the x’-y’ directions.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 215
We must transform the local boundary condition of d’3y = 0 (in local coordinates) into the global x-y system. Therefore, the relationship between of the components of the displacement in the local and the global coordinate systems at node 3 is:
⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡−=
⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧
3
3
3
3
3
3
1000cossin0sincos
'''
φαααα
φy
x
y
x
dd
dd
We can rewrite the above expression as:
[ ]⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡−==
1000cossin0sincos
][' 3333 αααα
tdtd
We can apply this sort of transformation to the entire displacement vector as:
'][or][' dTddTd Tii ==
where the matrix [Ti] is:
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡=
][]0[]0[
]0[][]0[
]0[]0[][
][
3tI
ITi
Both the identity matrix [I] and the matrix [t3] are 3 x 3 matrices.
The force vector can be transformed by using the same transformation.
fTf i ][' = In global coordinates, the force-displacement equations are:
dKf ][=
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 216
Applying the skewed support transformation to both sides of the force-displacement equation gives:
dKTfT ii ]][[][ = By using the relationship between the local and the global displacements, the force-displacement equations become:
']][][['']][][[][ dTKTfdTKTfT Tii
Tiii =⇒=
Therefore the global equations become:
⎪⎪⎪⎪⎪⎪
⎭
⎪⎪⎪⎪⎪⎪
⎬
⎫
⎪⎪⎪⎪⎪⎪
⎩
⎪⎪⎪⎪⎪⎪
⎨
⎧
=
⎪⎪⎪⎪⎪⎪
⎭
⎪⎪⎪⎪⎪⎪
⎬
⎫
⎪⎪⎪⎪⎪⎪
⎩
⎪⎪⎪⎪⎪⎪
⎨
⎧
1
3
3
2
2
2
1
1
1
3
3
3
2
2
2
1
1
1
''
]][][[
''
φ
φ
φ
y
x
y
x
y
x
Tii
y
x
y
x
y
x
dd
dd
dd
TKT
MFFMFFMFF
Grid Equations
A grid is a structure on which the loads are applied perpendicular to the plane of the structure, as opposed to a plane frame where loads are applied in the plane of the structure. Both torsional and bending moment continuity are maintained at each node in a grid element. Examples of a grid structure are floors and bridge deck systems. A typical grid structure is shown in the figure below.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 217
A representation of the grid element is shown below:
The degrees of freedom for a grid element are: a vertical displacement iyd (normal to the grid), a torsional rotation ixφ about the x axis, and a bending rotation izφ about the z axis. The nodal forces are: a transverse force iyf a torsional moment ixm about the x axis, and a bending moment izm about the z axis.
Let’s derive the torsional rotation components of the element stiffness matrix. Consider the sign convention for nodal torque and angle of twist shown the figure below.
A linear displacement function φ is assumed.
xaa ˆ21 +=φ Applying the boundary conditions and solving for the unknown coefficients gives:
xxx x
L 112 ˆˆˆˆ
φφφφ +⎟⎟⎠
⎞⎜⎜⎝
⎛ −=
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 218
Or in matrix form:
[ ]⎭⎬⎫
⎩⎨⎧
==x
xNN2
121 ˆ
ˆˆφφ
φ
where N1 and N2 are the interpolation functions gives as:
LxN
LxN =−= 21
ˆ1
To obtain the relationship between the shear strain γ and the angle of twist φ consider the torsional deformation of the bar as shown below.
If we assume that all radial lines, such as OA, remain straight during twisting or torsional deformation, then the arc length AB is:
φγ ˆˆmax RdxdAB == Therefore;
xdRd
ˆˆ
max
φγ =
At any radial position, r, we have, from similar triangles OAB and OCD:
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 219
( )xxLr
xddr 12
ˆˆˆˆ
φφφγ −==
The relationship between shear stress and shear strain is:
τ γ=G where G is the shear modulus of the material. From elementary mechanics of materials, we get:
ˆ τ=x
JmR
where J is the polar moment of inertia for a circular cross section or the torsional constant for non-circular cross sections. Rewriting the above equation we get:
( )2 1ˆ ˆˆ φ φ= −x x x
GJmL
The nodal torque sign convention gives:
xxxx mmmm ˆˆˆˆ 21 =−= Therefore;
( ) ( )1 1 2 2 2 1ˆ ˆ ˆ ˆˆ ˆφ φ φ φ= − = −x x x x x x
GJ GJm mL L
In matrix form the above equations are:
⎭⎬⎫
⎩⎨⎧⎥⎦
⎤⎢⎣
⎡−
−=
⎭⎬⎫
⎩⎨⎧
x
x
x
x
LGJ
mm
2
1
2
1
ˆˆ
1111
ˆˆ
φφ
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 220
Combining the torsional effects with shear and bending effects, we obtain the local stiffness matrix equations for a grid element.
⎪⎪⎪⎪
⎭
⎪⎪⎪⎪
⎬
⎫
⎪⎪⎪⎪
⎩
⎪⎪⎪⎪
⎨
⎧
⎥⎥⎥⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢⎢⎢⎢
⎣
⎡
−−
−−−−
−−
=
⎪⎪⎪⎪
⎭
⎪⎪⎪⎪
⎬
⎫
⎪⎪⎪⎪
⎩
⎪⎪⎪⎪
⎨
⎧
z
x
y
z
x
y
LEI
LEI
LEI
LEI
LGJ
LGJ
LEI
LEI
LEI
LEI
LEI
LEI
LEI
LEI
LGJ
LGJ
LEI
LEI
LEI
LEI
z
x
y
z
x
y
d
d
mmf
mmf
2
2
2
1
1
1
4626
612612
2646
612612
2
2
2
1
1
1
ˆˆˆˆˆˆ
000000
0000
000000
ˆˆ
ˆˆˆ
ˆ
22
2323
22
2323
φφ
φφ
The transformation matrix relating local to global degrees of freedom for a grid is:
⎥⎥⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢⎢⎢
⎣
⎡
−
−=
CSSC
CSSC
TG
00000000
00100000000000000001
where θ is now positive taken counterclockwise from x to x in the x-z plane: therefore;
cos sinj i j ix x z zC S
L Lθ θ
− −= = = =
The global stiffness matrix for a grid element arbitrary oriented in the x-z plane is given by:
GGT
GG TkTk ˆ=
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 221
Grid Example Consider the frame shown in the figure below.
The frame is fixed at nodes 2, 3, and 4, and is subjected to a load of 100 kips applied at node 1. Assume I = 400 in.4, J = 110 in.4, G = 12 x 10 3 ksi, and E = 30 x 10 3 ksi for all elements.
To facilitate a timely solution, the boundary conditions at nodes 2, 3, and 4 are applied to the local stiffness matrices at the beginning of the solution.
000
444
333
222
=========
zxy
zxy
zxy
ddd
φφφφφφ
Beam Element 1:
447.036.221020sin894.0
36.22200cos )1(
12)1(
12 =−
=−
==−=−
=−
==L
zzSL
xxC θθ
where
3 3
3 3 2 212 12(30 10 )(400) 6 6(30 10 )(400)7.45 1,000
(22.36 12) (22.36 12)× ×
= = = =× ×
EI EIk kinL L
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 222
3 34 4(30 10 )(400) (12 10 )(110)179,000 4,920(22.36 12) (22.36 12)
× ×= = ⋅ = = ⋅
× ×EI GJk in k inL L
The global stiffness matrix for element 1, considering only the parts associated with node 1, and the following relationship:
GGT
GG TkTk ˆ=
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡
−−−=
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡
−−−=
894.0447.00447.0894.00001
894.0447.00447.0894.00001
TGG TT
inkk
zxyd
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡=
000,1790000,10920,40
000,1045.7ˆ )1(
111 φφ
Therefore, the global stiffness matrix is
inkk
zxyd
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡
−−
−−=
000,144600,69894600,69700,39447
89444745.7)1(
111 φφ
Beam Element 2:
447.036.22100sin894.0
36.22200cos )2(
13)2(
13 −=−
=−
==−=−
=−
==L
zzSL
xxC θθ
where
3 3
3 3 2 212 12(30 10 )(400) 6 6(30 10 )(400)7.45 1,000
(22.36 12) (22.36 12)× ×
= = = =× ×
EI EIk kinL L
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 223
3 34 4(30 10 )(400) (12 10 )(110)179,000 4,920
(22.36 12) (22.36 12)× ×
= = ⋅ = = ⋅× ×
EI GJk in k inL L
The global stiffness matrix for element 2, considering only the parts associated with node 1, and the following relationship:
GGT
GG TkTk ˆ=
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡
−−−
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡
−−−=
894.0447.00447.0894.00001
000,1790000,10920,40
000,1045.7
894.0447.00447.0894.00001
)2(k
Therefore, the global stiffness matrix is
inkk
zxyd
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡
−−−−
=000,144600,69894600,69700,39447
89444745.7)2(
111 φφ
Beam Element 3:
110
100sin010
2020cos )3(14
)3(14 −=
−=
−===
−=
−==
LzzS
LxxC θθ
where
kLEIink
LEI 000,5
)1210()400)(1030(66/3.83
)1210()400)(1030(1212
2
3
23
3
3=
××
==×
×=
inkL
GJinkLEI
⋅=×
×=⋅=
××
= 000,11)1210(
)110)(1012(000,400)1210(
)400)(1030(44 33
The global stiffness matrix for element 3, considering only the parts associated with node 1, and the following relationship:
GGT
GG TkTk ˆ=
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 224
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡−
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡
−=
010100
001
000,4000000,50000,110000,503.83
010100001
)3(k
Therefore, the global stiffness matrix is
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡=
000,11000000,400000,50000,53.83
)3(
111
k
zxyd φφ
Superimposing the three elemental stiffness matrices gives:
1 1 1
98.2 5,000 1,7905,000 479,000 01,790 0 299,000
φ φ
−⎡ ⎤⎢ ⎥= ⎢ ⎥⎢ ⎥−⎣ ⎦
y x zd
K
The global equations are:
1 1
1 1
1 1
100 98.2 5,000 1,7900 5,000 479,000 00 1,790 0 299,000
φφ
= − −⎧ ⎫ ⎡ ⎤ ⎧ ⎫⎪ ⎪ ⎪ ⎪⎢ ⎥= =⎨ ⎬ ⎨ ⎬⎢ ⎥⎪ ⎪ ⎪ ⎪⎢ ⎥= −⎩ ⎭ ⎣ ⎦ ⎩ ⎭
y y
x x
z z
F k dMM
Solving the above equations gives:
⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧
−
−=
⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧
radrad
ind
z
x
y
0169.00295.0
83.2
1
1
1
φφ
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 225
Element 1: The grid element force-displacement equations can be obtained using dTkf GG
ˆˆ = .
⎪⎪⎪
⎭
⎪⎪⎪
⎬
⎫
⎪⎪⎪
⎩
⎪⎪⎪
⎨
⎧−−
=
⎪⎪⎪
⎭
⎪⎪⎪
⎬
⎫
⎪⎪⎪
⎩
⎪⎪⎪
⎨
⎧
−
−
⎥⎥⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢⎢⎢
⎣
⎡
−−−
−−−
=
000
00192.00339.0
83.2
000
0169.00295.0
83.2
894.0447.00000447.0894.00000001000000894.0447.00000447.0894.00000001
radradin
radradin
dTG
Therefore, the local force-displacement equations are:
⎪⎪⎪
⎭
⎪⎪⎪
⎬
⎫
⎪⎪⎪
⎩
⎪⎪⎪
⎨
⎧−−
⎥⎥⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢⎢⎢
⎣
⎡
−−
−−−−
−−
==
000
00192.00339.0
83.2
000,1790000,1500,890000,10920,400920,40000,1045.7000,1045.7500,890000,1000,1790000,10920,400920,40
000,1045.7000,1045.7
ˆˆ)1(
radradin
dTkf
Solving for the forces and moments gives:
1
1
1
2
2
2
ˆ 19.2ˆ 167ˆ 2,480ˆ 19.2
167ˆ2,260ˆ
⎧ ⎫ −⎧ ⎫⎪ ⎪ ⎪ ⎪− ⋅⎪ ⎪ ⎪ ⎪⎪ ⎪ ⎪ ⎪− ⋅⎪ ⎪ =⎨ ⎬ ⎨ ⎬⎪ ⎪ ⎪ ⎪⎪ ⎪ ⎪ ⎪⋅⎪ ⎪ ⎪ ⎪
− ⋅⎩ ⎭⎪ ⎪⎩ ⎭
y
x
z
y
x
z
f km k inm k in
kfk inmk inm
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 226
Element 2: The grid element force-displacement equations can be obtained using ˆ ˆ
G Gf k T d= .
⎪⎪⎪
⎭
⎪⎪⎪
⎬
⎫
⎪⎪⎪
⎩
⎪⎪⎪
⎨
⎧−−
=
⎪⎪⎪
⎭
⎪⎪⎪
⎬
⎫
⎪⎪⎪
⎩
⎪⎪⎪
⎨
⎧
−
−
⎥⎥⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢⎢⎢
⎣
⎡
−−−
−−−
=
000
0283.00188.0
83.2
000
0169.00295.0
83.2
894.0447.00000447.0894.00000001000000894.0447.00000447.0894.00000001
radradin
radradin
dTG
Therefore, the local force-displacement equations are:
⎪⎪⎪
⎭
⎪⎪⎪
⎬
⎫
⎪⎪⎪
⎩
⎪⎪⎪
⎨
⎧−−
⎥⎥⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢⎢⎢
⎣
⎡
−−
−−−−
−−
==
000
0283.00188.0
83.2
000,1790000,1500,890000,10920,400920,40000,1045.7000,1045.7500,890000,1000,1790000,10920,400920,40
000,1045.7000,1045.7
ˆˆ)2(
radradin
dTkf
Solving for the forces and moments gives:
1
1
1
3
3
3
ˆ 7.23ˆ 92.5ˆ 2,240ˆ 7.23
92.5ˆ295ˆ
⎧ ⎫ ⎧ ⎫⎪ ⎪ ⎪ ⎪− ⋅⎪ ⎪ ⎪ ⎪⎪ ⎪ ⎪ ⎪− ⋅⎪ ⎪ =⎨ ⎬ ⎨ ⎬
−⎪ ⎪ ⎪ ⎪⎪ ⎪ ⎪ ⎪⋅⎪ ⎪ ⎪ ⎪
− ⋅⎩ ⎭⎪ ⎪⎩ ⎭
y
x
z
y
x
z
f km k inm k in
kfk inm
k inm
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 227
Element 3: The grid element force-displacement equations can be obtained using ˆ ˆ
G Gf k T d= .
⎪⎪⎪
⎭
⎪⎪⎪
⎬
⎫
⎪⎪⎪
⎩
⎪⎪⎪
⎨
⎧ −
=
⎪⎪⎪
⎭
⎪⎪⎪
⎬
⎫
⎪⎪⎪
⎩
⎪⎪⎪
⎨
⎧
−
−
⎥⎥⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢⎢⎢
⎣
⎡
−
−
=
000
0295.00169.0
83.2
000
0169.00295.0
83.2
010000100000
001000000010000100000001
radradin
radradin
dTG
Therefore, the local force-displacement equations are:
⎪⎪⎪
⎭
⎪⎪⎪
⎬
⎫
⎪⎪⎪
⎩
⎪⎪⎪
⎨
⎧ −
⎥⎥⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢⎢⎢
⎣
⎡
−−
−−−−
−−
==
000
0295.00169.0
83.2
000,4000000,5000,2000000,50000,1100000,110000,503.83000,503.83000,2000000,5000,4000000,5
0000,1100000,110000,503.83000,503.83
ˆˆ)3(
radradin
dTkf
Solving for the forces and moments gives:
1
1
1
4
4
4
ˆ 88.1ˆ 186ˆ 2,340ˆ 88.1
186ˆ8,240ˆ
⎧ ⎫ −⎧ ⎫⎪ ⎪ ⎪ ⎪⋅⎪ ⎪ ⎪ ⎪⎪ ⎪ ⎪ ⎪− ⋅⎪ ⎪ =⎨ ⎬ ⎨ ⎬⎪ ⎪ ⎪ ⎪⎪ ⎪ ⎪ ⎪− ⋅⎪ ⎪ ⎪ ⎪
− ⋅⎩ ⎭⎪ ⎪⎩ ⎭
y
x
z
y
x
z
f km k inm k in
kfk inmk inm
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 228
To check the equilibrium of node 1 the local forces and moments for each element need to be transformed to global coordinates. Recall, that:
1ˆˆ −==⇒= TTfTfTff TT Since we are only checking the forces and moments at node 1, we need only the upper-left-hand portion of the transformation matrix TG. Therefore; for Element 1:
1
1
1
1 0 0 19.2 19.20 0.894 0.447 167 1,2600 0.447 0.894 2,480 2,150
− −⎧ ⎫ ⎡ ⎤ ⎧ ⎫ ⎧ ⎫⎪ ⎪ ⎪ ⎪ ⎪ ⎪⎢ ⎥= − − − ⋅ = ⋅⎨ ⎬ ⎨ ⎬ ⎨ ⎬⎢ ⎥⎪ ⎪ ⎪ ⎪ ⎪ ⎪⎢ ⎥− − ⋅ ⋅⎩ ⎭ ⎣ ⎦ ⎩ ⎭ ⎩ ⎭
y
x
z
f k km k in k inm k in k in
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 229
Therefore; for Element 2:
1
1
1
1 0 0 7.23 7.230 0.894 0.447 92.5 1,0800 0.447 0.894 2,240 1,960
−⎧ ⎫ ⎡ ⎤ ⎧ ⎫ ⎧ ⎫⎪ ⎪ ⎪ ⎪ ⎪ ⎪⎢ ⎥= − − ⋅ = ⋅⎨ ⎬ ⎨ ⎬ ⎨ ⎬⎢ ⎥⎪ ⎪ ⎪ ⎪ ⎪ ⎪⎢ ⎥− − − ⋅ − ⋅⎩ ⎭ ⎣ ⎦ ⎩ ⎭ ⎩ ⎭
y
x
z
f k km k in k inm k in k in
Therefore; for Element 3:
1
1
1
1 0 0 88.1 88.10 0 1 2,340 2,3400 1 0 186 186
− −⎧ ⎫ ⎡ ⎤ ⎧ ⎫ ⎧ ⎫⎪ ⎪ ⎪ ⎪ ⎪ ⎪⎢ ⎥= − ⋅ = − ⋅⎨ ⎬ ⎨ ⎬ ⎨ ⎬⎢ ⎥⎪ ⎪ ⎪ ⎪ ⎪ ⎪⎢ ⎥− − ⋅ − ⋅⎩ ⎭ ⎣ ⎦ ⎩ ⎭ ⎩ ⎭
y
x
z
f k km k in k inm k in k in
The forces and moments that are applied to node 1 by each element are equal in magnitude and opposite direction. Therefore the sum of the forces and moments acting on node 1 are:
The forces and moments accurately satisfy equilibrium considering the amount of truncation error inherent in results of the calculations presented in this example.
kF y 07.01.882.1923.71001 =++−−=∑
inkM x ⋅=+−−=∑ 0.0340,2080,1260,11
inkM z ⋅−=++−=∑ 0.4186060,1150,21
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 230
Grid Example Consider the frame shown in the figure below.
The frame is fixed at nodes 1 and 3, and is subjected to a load of 22 kN applied at node 2. Assume I = 16.6 x 10-5 m4, J = 4.6 x 10-5 m4, G = 84 GPa, and E = 210 GPa for all elements.
To facilitate a timely solution, the boundary conditions at nodes 1 and 3 are applied to the local stiffness matrices at the beginning of the solution.
00
333
111
======
zxy
zxy
dd
φφφφ
Beam Element 1: the local x axis coincides with the global x axis
030sin1
33cos )1(
12)1(
12 ==−
====−
==L
zzSL
xxC θθ
where
mkNLEI /1055.1
)3()106.16)(10210(1212 4
3
56
3 ×=××
=−
kNLEI 4
2
56
2 1032.2)3(
)106.16)(10210(66×=
××=
−
mkNLEI ·1065.4
3)106.16)(10210(44 4
56
×=××
=−
mkNL
GJ ·10128.03
)106.4)(1084( 456
×=××
=−
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 231
The global stiffness matrix for element 1, considering only the parts associated with node 2, may be obtained from the following relationship:
GGT
GG TkTk ˆ=
mkNk⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡
−
−
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡=
100010001
65.4032.20128.0032.2055.1
100010001
104)1(
Therefore, the global stiffness matrix is
mkNk
zxyd
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡
−
−=
65.4032.20128.0032.2055.1
104)1(
222 φφ
Beam Element 2: the local x axis is located from node 2 to node 3
3 2 3 2(2) (1)
0 3cos 0 sin 13 3
x x z zC SL L
θ θ− − −= = = = = = = = −
The global stiffness matrix for element 2, considering only the parts associated with node 2, may be obtained using:
GGT
GG TkTk ˆ=
mkNk⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡−
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡
−
−
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡
−=
010100
001
65.4032.20128.0032.2055.1
010100001
104)2(
Therefore, the global stiffness matrix is
mkNk
zxyd
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡=
128.000065.432.2032.255.1
104)2(
222 φφ
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 232
Superimposing the two elemental stiffness matrices gives:
mkNK
zxyd
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡
−
−=
78.4032.2078.432.2
32.232.210.310 4
222 φφ
The global equations are:
⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡
−
−=
⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧
==
−=
x
x
y
z
x
y d
MM
kNF
2
2
24
2
2
2
78.4032.2078.432.232.232.210.3
1000
22
φφ
Solving the above equations gives:
2
2
2
0.002590.001260.00126
φφ
−⎧ ⎫ ⎧ ⎫⎪ ⎪ ⎪ ⎪=⎨ ⎬ ⎨ ⎬⎪ ⎪ ⎪ ⎪−⎩ ⎭ ⎩ ⎭
y
x
z
d mradrad
Element 1: The grid element force-displacement equations can be obtained using ˆ ˆ
G Gf k T d= .
⎪⎪⎪
⎭
⎪⎪⎪
⎬
⎫
⎪⎪⎪
⎩
⎪⎪⎪
⎨
⎧
−
−=
⎪⎪⎪
⎭
⎪⎪⎪
⎬
⎫
⎪⎪⎪
⎩
⎪⎪⎪
⎨
⎧
−
−
⎥⎥⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢⎢⎢
⎣
⎡
=
radrad
m
radrad
mdTG
00126.000126.0
00259.0000
00126.000126.0
00259.0000
100000010000001000000100000010000001
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 233
Therefore, the local force-displacement equations are:
⎪⎪⎪
⎭
⎪⎪⎪
⎬
⎫
⎪⎪⎪
⎩
⎪⎪⎪
⎨
⎧
−
−
⎥⎥⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢⎢⎢
⎣
⎡
−−
−−−−
−−
==
radrad
mdTkf
00126.000126.0
00259.0000
65.4032.233.2032.20128.000128.0032.2055.132.2055.1
33.2032.265.4032.20128.000128.0032.2055.132.2055.1
10ˆˆ 4)1(
Solving for the forces and moments gives:
1
1
1
2
2
2
ˆ 11.0ˆ 1.50ˆ 31.0ˆ 11.0
1.50ˆ1.50ˆ
⎧ ⎫ ⎧ ⎫⎪ ⎪ ⎪ ⎪− ⋅⎪ ⎪ ⎪ ⎪⎪ ⎪ ⎪ ⎪⋅⎪ ⎪ =⎨ ⎬ ⎨ ⎬
−⎪ ⎪ ⎪ ⎪⎪ ⎪ ⎪ ⎪⋅⎪ ⎪ ⎪ ⎪
⋅⎩ ⎭⎪ ⎪⎩ ⎭
y
x
z
y
x
z
f kNm kN mm kN m
kNfkN mmkN mm
Element 2: The grid element force-displacement equations can be obtained using ˆ ˆ
G Gf k T d= .
⎪⎪⎪
⎭
⎪⎪⎪
⎬
⎫
⎪⎪⎪
⎩
⎪⎪⎪
⎨
⎧−
=
⎪⎪⎪
⎭
⎪⎪⎪
⎬
⎫
⎪⎪⎪
⎩
⎪⎪⎪
⎨
⎧
−
−
⎥⎥⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢⎢⎢
⎣
⎡
−
−
=
000
00126.000126.0
00259.0
000
00126.000126.0
00259.0
010000100000
001000000010000100000001
radrad
m
radrad
m
dTG
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 234
Therefore, the local force-displacement equations are:
⎪⎪⎪
⎭
⎪⎪⎪
⎬
⎫
⎪⎪⎪
⎩
⎪⎪⎪
⎨
⎧−
⎥⎥⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢⎢⎢
⎣
⎡
−−
−−−−
−−
==
000
00126.000126.0
00259.0
65.4032.233.2032.20128.000128.0032.2055.132.2055.1
33.2032.265.4032.20128.000128.0032.2055.132.2055.1
10ˆˆ 4)2(
radrad
m
dTkf
Solving for the forces and moments gives:
2
2
2
3
3
3
ˆ 11.0ˆ 1.50ˆ 1.50ˆ 11.0
1.50ˆ31.0ˆ
⎧ ⎫ −⎧ ⎫⎪ ⎪ ⎪ ⎪⋅⎪ ⎪ ⎪ ⎪⎪ ⎪ ⎪ ⎪− ⋅⎪ ⎪ =⎨ ⎬ ⎨ ⎬⎪ ⎪ ⎪ ⎪⎪ ⎪ ⎪ ⎪− ⋅⎪ ⎪ ⎪ ⎪
− ⋅⎩ ⎭⎪ ⎪⎩ ⎭
y
x
z
y
x
z
f kNm kN mm kN m
kNfkN mm
kN mm
The resulting free-body diagrams:
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 235
Beam Element Arbitrarily Oriented in Space In this section,we will develop a beam element that is arbitrarily oriented in
three-dimensions. This element can be used to analyze three-dimensional frames. Let consider bending about axes, as shown below.
The y axis is the principle axis for which the moment of inertia is minimum, Iy. The right-hand rule is used to establish the z axis and the maximum moment of inertia, Iz. Bending in the zx ˆˆ − plane: The bending in the zx − plane is defined by ym . The stiffness matrix for bending the in the x-z plane is:
⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢
⎣
⎡
−−
−−
−−=
3
2
3
2
2
2
3
2
3
2
2
2
4
4626
612612
2646
612612
ˆ
LLLL
LL
LL
LLLL
LL
LL
LEI
k yY
where Iy is the moment of inertia about the y axis (the weak axis). Bending in the yx ˆˆ − plane: The bending in the yx − plane is defined by ˆ zm . The stiffness matrix for bending the in the yx − plane is:
⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢
⎣
⎡
−−
−−
−−=
3
2
3
2
2
2
3
2
3
2
2
2
4
4626
612612
2646
612612
ˆ
LLLL
LL
LL
LLLL
LL
LL
LEIk z
z
where Iz is the moment of inertia about the z axis (the strong axis).
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 236
Direct superposition of the bending stiffness matrices with the effects of axial forces and torsional rotation give:
The global stiffness matrix may be obtained using:
TkTk T ˆ= where
⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢
⎣
⎡
=
33
33
33
33
x
x
x
x
T
λλ
λλ
where
zyxzyxzyxzyx dddddd 222222111111ˆˆˆˆˆˆˆˆˆˆˆˆ φφφφφφ
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 237
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡=
zzzyzx
yzyyyx
xzxyxx
x
CCCCCCCCC
ˆˆˆ
ˆˆˆ
ˆˆˆ
33λ
where the direction cosines,
jiC ˆ , are defined as shown below
The direction cosines of the x axis are:
kjix xzxyxx ˆˆˆ coscoscosˆ θθθ ++= where
nL
zzmL
yylL
xxxzxyxx =
−==
−==
−= 12
ˆ12
ˆ12
ˆ coscoscos θθθ
The y axis is selected to be perpendicular to the x and the z axes is such a way that the cross product of global z with x results in the y axis as shown in the figure below.
yxz ˆˆ =×
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 238
3 3 0x
l m n
m lD Dln mn DD D
λ
⎡ ⎤⎢ ⎥⎢ ⎥⎢ ⎥= −⎢ ⎥⎢ ⎥⎢ ⎥−⎢ ⎥⎣ ⎦
jDli
Dm
nml
kji
Dyxz +−===× 1001ˆˆ
where 22 mlD +=
The z axis is determined by the condition that yxz ˆˆˆ ×=
kDjD
mniDnl
lmnmlkji
Dyxz +−−=
−=×=
0
1ˆˆˆ
Therefore, the transformation matrix becomes:
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡=
zzzyzx
yzyyyx
xzxyxx
x
CCCCCCCCC
ˆˆˆ
ˆˆˆ
ˆˆˆ
33λ
There are two exceptions that arise when using the above expressions for mapping the local coordinates to the global system: (1) when the positive x coincides with z; and (2) when the positive x is in the opposite direction as z. For the first case, it is assumed that y is y.
0 0 10 0 01 0 0
λ⎡ ⎤⎢ ⎥= ⎢ ⎥⎢ ⎥−⎣ ⎦
In case two, it is assumed that y is y.
0 0 10 0 01 0 0
λ−⎡ ⎤
⎢ ⎥= ⎢ ⎥⎢ ⎥⎣ ⎦
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 239
If the effects of axial force, both shear forces, twisting moment, and both bending moments are considered, the stiffness matrix for a frame element is:
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 240
In this case the symbol φ are:
2 2
12 12y zy z
s s
EI EIGA L GA L
φ φ= =
where As is the effective beam cross-section in shear. Recall the shear modulus of elasticity or the modulus of rigidity, G, is related to the modulus of elasticity and the Poisson’s ratio, ν as:
( )2 1
EGν
=+
If φy and φz are set to zero, the stiffness matrix reduces to that shown previously on page 235. This is the form of the stiffness matrix used by SAP2000 for its frame element.
zyxzyxzyxzyx dddddd 222222111111ˆˆˆˆˆˆˆˆˆˆˆˆ φφφφφφ
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 241
Example Frame Application
A bus subjected to a static roof-crush analysis. In this model 599 frame elements and 357 nodes are used.
Concept of Substructure Analysis
Sometimes structures are too large to be analyzed as a single system or treated as a whole; that is, the final stiffness matrix and equations for solution exceed the memory capacity of the computer. A procedure to overcome this problem is to separate the whole structure into smaller units called substructures. For example, the space frame of an airplane, as shown below, may require thousands of nodes and elements to completely model and describe the response of the whole structure. If we separate the aircraft into substructures,
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 242
such as parts of the fuselage or body, wing sections, etc., as shown below, then we can solve the problem more readily and on computers with limited memory.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 243
Problems
14. Do problems 5.3, 5.8, 5.13, 5.28, 5.41, and 5.43 on pages 240 - 263 in your textbook “A First Course in the Finite Element Method” by D. Logan.
15. Do problems 5.23, 5.25, 5.35, 5.39, and 5.55 on pages 240 - 263 in your textbook “A First Course in the Finite Element Method” by D. Logan. You may use the SAP2000 to do frame analysis.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 243
Development of the Plane Stress and Plane Strain Stiffness Equations
Introduction
In Chapters 2 through 5, we considered only line elements. Line elements are connected only at common nodes, forming framed or articulated structures such as trusses, frames, and grids. Line elements have geometric properties such as cross-sectional area and moment of inertia associated with their cross sections. However, only one local coordinate along the length of the element is required to describe a position along the element (hence, they are called line elements). Nodal compatibility is then enforced during the formulation of the nodal equilib-rium equations for a line element.
This chapter considers the two-dimensional finite element. Two-dimensional (planar) elements are thin-plate elements such that two coordinates define a po-sition on the element surface.
The elements are connected at common nodes and/or along common edges
to form continuous structures. Nodal compatibility is then enforced during the formulation of the nodal equilibrium equations for two-dimensional elements. If proper displacement functions are chosen, compatibility along common edges is also obtained.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 244
The two-dimensional element is extremely important for: (1) Plane stress analysis, which includes problems such as plates with
holes, fillets, or other changes in geometry that are loaded in their plane resulting in local stress concentrations; and
(2) Plane strain analysis, which includes problems such as a long under-ground box culvert subjected to a uniform load acting constantly over its length or a long cylindrical control rod subjected to a load that remains constant over the rod length (or depth).
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 245
Plane Stress Problems
Plane Strain Problems We begin this chapter with the development of the stiffness matrix for a basic
two-dimensional or plane finite element, called the constant-strain triangular element. The constant-strain triangle (CST) stiffness matrix derivation is the simplest among the available two-dimensional elements.
We will derive the CST stiffness matrix by using the principle of minimum po-tential energy because the energy formulation is the most feasible for the devel-opment of the equations for both two- and three-dimensional finite elements.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 246
General Steps in the Formulation of the Plane Triangular Element Equations
We will now follow the steps described in Chapter 1 to formulate the governing equations for a plane stress/plane strain triangular element. First, we will de-scribe the concepts of plane stress and plane strain. Then we will provide a brief description of the steps and basic equations pertaining to a plane triangular ele-ment.
Plane Stress Plane stress is defined to be a state of stress in which the normal stress and
the shear stresses directed perpendicular to the plane are assumed to be zero. That is, the normal stress σz and the shear stresses τxz and τyz are assumed to be zero. Generally, members that are thin (those with a small z dimension com-pared to the in-plane x and y dimensions) and whose loads act only in the x-y plane can be considered to be under plane stress.
Plane Strain Plane strain is defined to be a state of strain in which the strain normal to the
x-y plane εz and the shear strains γxz and γyz are assumed to be zero. The as-sumptions of plane strain are realistic for long bodies (say, in the z direction) with constant cross-sectional area subjected to loads that act only in the x and/or y di-rections and do not vary in the z direction.
Two-Dimensional State of Stress and Strain The concept of two-dimensional state of stress and strain and the stress/strain
relationships for plane stress and plane strain are necessary to understand fully the development and applicability of the stiffness matrix for the plane stress/plane strain triangular element.
A two-dimensional state of stress is shown in the figure below. The infinitesi-mal element with sides dx and dy has normal stresses σz and σy acting in the x and y directions (here on the vertical and horizontal faces), respectively. The shear stress τxy acts on the x edge (vertical face) in the y direction. The shear stress τyx acts on the y edge (horizontal face) in the x direction.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 247
Since τxy equals τyx, three independent stress exist:
Tx y xyσ σ σ τ⎡ ⎤= ⎣ ⎦
Recall, the relationships for principal stresses in two-dimensions are:
σ σ σ σσ τ σ
+ −⎛ ⎞= + + =⎜ ⎟
⎝ ⎠
22
1 max2 2x y x y
xy
min2
2
2 22στ
σσσσσ =+⎟⎟
⎠
⎞⎜⎜⎝
⎛ −−
+= xy
yxyx
Also, θp is the principal angle which defines the normal whose direction is per-pendicular to the plane on which the maximum or minimum principle stress acts.
2tan2 xy
px y
τθ
σ σ=
−
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 248
The general two-dimensional state of strain at a point is show below.
The general definitions of normal and shear strains are:
xv
yu
xv
xu
xyyx ∂∂
+∂∂
=∂∂
=∂∂
= γεε
The strain may be written in matrix form as:
Tx y xyε ε ε γ⎡ ⎤= ⎣ ⎦
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 249
For plane stress, the stresses σz, τxz, and τyz are assumed to be zero. The stress-strain relationship is:
( ) ⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡
−−
=⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧
xy
y
x
xy
y
x E
γεε
νν
ν
ντσσ
15.0000101
1 2
where
( )⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡
−−
=ν
νν
ν15.0000101
1][
2
ED
is called the stress-strain matrix (or the constitutive matrix), E is the modulus of elasticity, and ν is Poisson’s ratio.
For plane strain, the strains εz, γxz, and γyz are assumed to be zero. The stress-strain relationship is:
( )( ) ⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡
−−
−
−+=
⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧
xy
y
x
xy
y
x E
γεε
ννν
νν
νντσσ
5.0000101
211
where
( )( )⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡
−−
−
−+=
ννν
νν
νν5.0000101
211][ ED
The partial differential equations for plane stress are:
⎟⎟⎠
⎞⎜⎜⎝
⎛∂∂
∂−
∂∂+
=∂∂
+∂∂
yxv
yu
yu
xu 2
2
2
2
2
2
2
21 ν
⎟⎟⎠
⎞⎜⎜⎝
⎛∂∂
∂−
∂∂+
=∂∂
+∂∂
yxu
yv
yv
xv 2
2
2
2
2
2
2
21 ν
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 250
Steps in the Formulation of the Element Stiffness Equations
Consider the problem of a thin plate subjected to a tensile load as shown in the figure below:
Step 1 - Discretize and Select Element Types Discretize the thin plate into a set of triangular elements. Each element is de-
fine by nodes i, j, and m.
We use triangular elements because boundaries of irregularly shaped bodies can be closely approximated, and because the expressions related to the triangular element are comparatively simple. This discretization is called a coarse-mesh generation if few large elements are used. Each node has two degrees of free-dom: displacements in the x and y directions. We will let ui and vi represent the node i displacement components in the x and y directions, respectively. The nodal displacements for an element with nodes i, j, and m are:
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 251
⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧=
m
j
i
ddd
d
where the nodes are ordered counterclockwise around the element, and
⎭⎬⎫
⎩⎨⎧
=i
ii v
ud
Therefore:
⎪⎪⎪
⎭
⎪⎪⎪
⎬
⎫
⎪⎪⎪
⎩
⎪⎪⎪
⎨
⎧
=
m
m
j
j
i
i
vuvuvu
d
Step 2 - Select Displacement Functions The general displacement function is:
⎭⎬⎫
⎩⎨⎧
=Ψ),(),(
yxvyxu
i
The functions u(x, y) and v(x, y) must be compatible with the element type.
Step 3 - Define the Strain-Displacement and Stress-Strain Relationships
The general definitions of normal and shear strains are:
xv
yu
xv
xu
xyyx ∂∂
+∂∂
=∂∂
=∂∂
= γεε
For plane stress, the stresses σz, τxz, and τyz are assumed to be zero. The stress-strain relationship is:
( ) ⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡
−−
=⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧
xy
y
x
xy
y
x E
γεε
νν
ν
ντσσ
15.0000101
1 2
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 252
For plane strain, the strains εz, γxz, and γyz are assumed to be zero. The stress-strain relationship is:
( )( ) ⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡
−−
−
−+=
⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧
xy
y
x
xy
y
x E
γεε
ννν
νν
νντσσ
5.0000101
211
Step 4 - Derive the Element Stiffness Matrix and Equations Using the principle of minimum potential energy, we can derive the element
stiffness matrix.
dkf ][= This approach is better than the direct methods used for one-dimensional ele-ments.
Step 5 - Assemble the Element Equations and Introduce Boundary Conditions
The final assembled or global equation written in matrix form is:
dKF ][= where F is the equivalent global nodal loads obtained by lumping distributed edge loads and element body forces at the nodes and [K] is the global structure stiffness matrix.
Step 6 - Solve for the Nodal Displacements Once the element equations are assembled and modified to account for the
boundary conditions, a set of simultaneous algebraic equations that can be writ-ten in expanded matrix form as:
Step 7 - Solve for the Element Forces (Stresses) For the structural stress-analysis problem, important secondary quantities of
strain and stress (or moment and shear force) can be obtained in terms of the displacements determined in Step 6.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 253
Derivation of the Constant-Strain Triangular Element Stiffness Matrix and Equations
Consider the problem of a thin plate subjected to a tensile load as shown in the figure below:
Step 1 - Discretize and Select Element Types The basic triangular element, shown below, is define by nodes i, j, and m la-
beled in a counterclockwise manner.
The nodal displacement matrix is:
⎪⎪⎪
⎭
⎪⎪⎪
⎬
⎫
⎪⎪⎪
⎩
⎪⎪⎪
⎨
⎧
=
m
m
j
j
i
i
vuvuvu
d
Step 2 - Select Displacement Functions We will select a linear displacement function for each triangular element, de-
fined as:
⎭⎬⎫
⎩⎨⎧
++++
=⎭⎬⎫
⎩⎨⎧
=Ψyaxaayaxaa
yxvyxu
i654
321
),(),(
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 254
A linear function ensures that the displacements along each edge of the element and the nodes shared by adjacent elements are equal.
⎪⎪⎪
⎭
⎪⎪⎪
⎬
⎫
⎪⎪⎪
⎩
⎪⎪⎪
⎨
⎧
⎥⎦
⎤⎢⎣
⎡=
⎭⎬⎫
⎩⎨⎧
++++
=Ψ
6
5
4
3
2
1
654
321
10000001
aaaaaa
yxyx
yaxaayaxaa
i
To obtain the values for the a’s substitute the coordinated of the nodal points into the above equations:
= + + = + +1 2 3 4 5 6i i i i i iu a a x a y v a a x a y jjjjjj yaxaavyaxaau 654321 ++=++=
mmmmmm yaxaavyaxaau 654321 ++=++= Solving for the a’s and writing the results in matrix forms gives:
[ ] uxaaaa
yxyxyx
uuu
mm
jj
ii
m
j
i1
3
2
1
111
−=⇒⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡=
⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧
(xj, yj)
(xi, yi) (xm, ym)
uiuj
um
u
y
x
Linear representation of u(x, y)
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 255
The inverse of the [x] matrix is:
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡=−
mji
mji
mji
Ax
γγγβββααα
21][ 1
where
mm
jj
ii
yxyxyx
A111
2 =
is the determinant of [x].
( ) ( ) ( )jimimjmji yyxyyxyyxA −+−+−=2 where A is the area of the triangle and
jmimjimjmji xxyyxyyx −=−=−= γβα
mijimjmimij xxyyxyyx −=−=−= γβα
ijmjimjijim xxyyxyyx −=−=−= γβα The values of a may be written matrix form as:
⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡=
⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧
m
j
i
mji
mji
mji
uuu
Aaaa
γγγβββααα
21
3
2
1
and
⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡=
⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧
m
j
i
mji
mji
mji
vvv
Aaaa
γγγβββααα
21
6
5
4
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 256
We will now derive the displacement function in terms of the coordinates x and y.
[ ]⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧=
3
2
1
1aaa
yxu
Substituting the values for a into the above equation gives:
[ ]1 12
i j m i
i j m j
i j m m
uu x y u
Au
α α αβ β βγ γ γ
⎡ ⎤ ⎧ ⎫⎪ ⎪⎢ ⎥= ⎨ ⎬⎢ ⎥⎪ ⎪⎢ ⎥ ⎩ ⎭⎣ ⎦
Expanding the above equations
[ ]1 12
i i j j m m
i i j j m m
i i j j m m
u u uu x y u u u
Au u u
α α αβ β βγ γ γ
⎡ ⎤+ +⎢ ⎥= + +⎢ ⎥⎢ ⎥+ +⎣ ⎦
Multiplying the matrices in the above equations gives
( ) ( ) ( ) 1( , )2 i i i i j j j j m m m mu x y x y u x y u x y u
Aα β γ α β γ α β γ= + + + + + + + +
A similar expression can be obtained for the y displacement
( ) ( ) ( ) 1( , )2 i i i i j j j j m m m mv x y x y v x y v x y v
Aα β γ α β γ α β γ= + + + + + + + +
The displacements can be written in a more convenience form as:
mmjjiimmjjii vNvNvNyxvuNuNuNyxu ++=++= ),(),( where
( ) ( ) ( )yxA
NyxA
NyxA
N mmmmjjjjiiii γβαγβαγβα ++=++=++=21
21
21
The elemental displacements can be summarized as:
⎭⎬⎫
⎩⎨⎧
++++
=⎭⎬⎫
⎩⎨⎧
=Ψmmjjii
mmjjiii vNvNvN
uNuNuNyxvyxu
),(),(
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 257
In another form the above equations are:
0 0 0 [ ]
0 0 0
i
i
i j m j
i j m j
m
m
uv
N N N uN d
N N N vuv
ψ
⎧ ⎫⎪ ⎪⎪ ⎪⎪ ⎪⎡ ⎤
Ψ = =⎨ ⎬⎢ ⎥⎣ ⎦ ⎪ ⎪
⎪ ⎪⎪ ⎪⎩ ⎭
where
[ ] ⎥⎦
⎤⎢⎣
⎡=
mji
mji
NNNNNN
N000
000
The linear triangular shape functions are illustrated below
Step 3 - Define the Strain-Displacement and Stress-Strain Relationships Elemental Strains: The strains over a two-dimensional element are:
⎪⎪⎪
⎭
⎪⎪⎪
⎬
⎫
⎪⎪⎪
⎩
⎪⎪⎪
⎨
⎧
∂∂
+∂∂∂∂∂∂
=⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧=
xv
yu
yvxu
xy
y
x
γεε
ε
j
i m
1
Ni
y
x
j
i m 1
Nj
y
x
j
i m
1
Nm
y
x
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 258
Substituting our approximation for the displacement gives:
( )mmjjiix uNuNuNx
uxu
++∂∂
==∂∂
,
mxmjxjixix uNuNuNu ,,,, ++= where the comma indicates differentiation with respect to that variable. The de-rivatives of the interpolation functions are:
( )A
NA
NA
yxxA
N mxm
jxj
iiiixi 2222
1,,,
βββγβα ===++∂∂
=
Therefore:
( )mmjjii uuuAx
u βββ ++=∂∂
21
In a similar manner, the remaining strain terms are approximated as:
( )12 i i j j m m
v v v vy A
γ γ γ∂= + +
∂
( )mmmmjjjjiiii vuvuvuAx
vyu γβγβγβ +++++=
∂∂
+∂∂
21
We can write the strains in matrix form as:
or
[ ]⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧=
m
j
i
mji
ddd
BBBε
0 0 01 0 0 0
2
i
ix i j m
jy i j m
jxy i i j j m m
m
m
uuvxuvvy Auu v
y x v
ε β β βε ε γ γ γ
γ γ β γ β γ β
⎧ ⎫⎧ ⎫∂ ⎪ ⎪⎪ ⎪∂ ⎪ ⎪⎪ ⎪⎧ ⎫ ⎡ ⎤
⎪ ⎪⎪ ⎪∂⎪ ⎪ ⎢ ⎥== = =⎨ ⎬ ⎨ ⎬ ⎨ ⎬⎢ ⎥∂⎪ ⎪ ⎪ ⎪ ⎪ ⎪⎢ ⎥⎩ ⎭ ⎣ ⎦⎪ ⎪ ⎪ ⎪∂ ∂+⎪ ⎪ ⎪ ⎪∂ ∂⎩ ⎭ ⎩ ⎭
0 0 01 0 0 0
2
i
ix i j m
jy i j m
jxy i i j j m m
m
m
uuvxuvvy Auu v
y x v
ε β β βε ε γ γ γ
γ γ β γ β γ β
⎧ ⎫⎧ ⎫∂ ⎪ ⎪⎪ ⎪∂ ⎪ ⎪⎪ ⎪⎧ ⎫ ⎡ ⎤
⎪ ⎪⎪ ⎪∂⎪ ⎪ ⎢ ⎥== = =⎨ ⎬ ⎨ ⎬ ⎨ ⎬⎢ ⎥∂⎪ ⎪ ⎪ ⎪ ⎪ ⎪⎢ ⎥⎩ ⎭ ⎣ ⎦⎪ ⎪ ⎪ ⎪∂ ∂+⎪ ⎪ ⎪ ⎪∂ ∂⎩ ⎭ ⎩ ⎭
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 259
where
[ ] [ ] [ ]⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡=
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡=
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡=
mm
m
m
m
jj
j
j
j
ii
i
i
i AB
AB
AB
βγγ
β
βγγ
β
βγγ
β0
0
210
0
210
0
21
These equations can be written in matrix form as:
][ dB=ε Stress-Strain Relationship: The in-plane stress-strain relationship is:
⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧=
⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧
xy
y
x
xy
y
x
Dγεε
τσσ
][
where [D] for plane stress is:
( )⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡
−−
=ν
νν
ν15.0000101
1][
2
ED
and [D] for plane strain is:
( )( )⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡
−−
−
−+=
ννν
νν
νν5.0000101
211][ ED
In-plane stress can be related to displacements by:
]][[ dBD=σ
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 260
Step 4 - Derive the Element Stiffness Matrix and Equations The total potential energy is defined as the sum of the internal strain energy U
and the potential energy of the external forces Ω:
spbp U Ω+Ω+Ω+=π where the strain energy is:
∫=V
T dVU 21 σε
or
∫=V
T dVDU ][21 εε
The potential energy of the body force term is:
∫ Ψ−=ΩV
Tb dVX
where Ψ is the general displacement function, and X is the body weight per unit volume.
The potential energy of the concentrated forces is:
Pd Tp −=Ω
where P are the concentrated forces, and d are the nodal displacements.
The potential energy of the distributed loads is:
∫ Ψ−=ΩS
Ts dST
where Ψ is the general displacement function, and T are the surface tractions. Then the total potential energy expression becomes:
1 [ ] [ ][ ] [ ] [ ] 2
T T T TT T Tp
V V S
d B D B d dV d N X dV d P d N T dSπ = − − −∫ ∫ ∫
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 261
The nodal displacements d are independent of the general x-y coordinates, therefore
1 [ ] [ ][ ] [ ] [ ] 2
T T T TT T Tp
V V S
d B D B dV d d N X dV d P d N T dSπ = − − −∫ ∫ ∫
We can define the last three terms as:
∫∫ ++=S
T
V
T dSTNPdVXNf ][][
Therefore:
1 [ ] [ ][ ]2
T TTp
V
d B D B dV d d fπ = −∫
Minimization of πp with respect to each nodal displacement requires that:
[ ] [ ][ ] 0p T
V
B D B dV d fdπ∂
= − =∂ ∫
The above relationship requires:
[ ] [ ][ ]T
V
B D B dV d f=∫
The stiffness matrix can be defined as:
[ ] [ ] [ ][ ]T
V
k B D B dV= ∫
For an element of constant thickness, t, the above integral becomes:
∫=A
T dydxBDBtk ]][[][][
The integrand in the above equation is not a function of x or y (global coordi-nates); therefore, the integration reduces to:
∫=A
T dydxBDBtk ]][[][][
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 262
or ]][[][][ BDBtAk T=
where A is the area of the triangular element. Expanding the stiffness relationship gives:
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡=
][][][][][][][][][
][
mmmjmi
jmjjji
imijii
kkkkkkkkk
k
where each [kii] is a 2 x 2 matrix define as:
tABDBktABDBktABDBk mT
iimjT
iijiT
iii ]][[][][]][[][][]][[][][ === Recall:
[ ] [ ] [ ]⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡=
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡=
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡=
mm
m
m
m
jj
j
j
j
ii
i
i
i AB
AB
AB
βγγ
β
βγγ
β
βγγ
β0
0
210
0
210
0
21
Step 5 - Assemble the Element Equations to Obtain the Global Equations and Introduce the Boundary Conditions
The global stiffness matrix can be found by the direct stiffness method.
∑=
=N
e
ekK1
)( ][][
The global equivalent nodal load vector is obtained by lumping body forces
and distributed loads at the appropriate nodes as well as including any concen-trated loads.
∑=
=N
e
efF1
)(
The resulting global equations are:
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 263
][ dKF = where d is the total structural displacement vector.
In the above formulation of the element stiffness matrix, the matrix has been derived for a general orientation in global coordinates. Therefore, no transforma-tion form local to global coordinates is necessary. However, for completeness, we will now describe the method to use if the local axes for the constant-strain triangular element are not parallel to the global axes for the whole structure.
To relate the local to global displacements, force, and stiffness matrices we
will use:
TkTkTffTdd T ˆˆˆ === The transformation matrix T for the triangular element is:
where C = cos θ and S = sin θ, and θ is shown in the figure above.
Step 6 - Solve for the Nodal Displacements
0 0 0 00 0 0 0
0 0 0 00 0 0 00 0 0 00 0 0 0
C SS C
C ST
S CC SS C
⎡ ⎤⎢ ⎥−⎢ ⎥⎢ ⎥
= ⎢ ⎥−⎢ ⎥⎢ ⎥⎢ ⎥
−⎣ ⎦
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 264
Step 7 - Solve for Element Forces and Stress
Having solved for the nodal displacements, we can obtain strains and stresses in x and y directions in the elements by using:
[ ] [ ][ ] B d D B dε σ= =
Plane Stress Example
Consider the structure shown in the figure below.
Assume plane stress conditions. All coordinates are shown on the figure. Let E = 30 x 106 psi, ν = 0.25, and t = 1 in. Assume the element nodal displacements have been determined to be u1 = 0.0, v1 = 0.0025 in., u2 = 0.0012 in., v2 = 0.0, u3 = 0.0, and v3 = 0.0025 in. Determine the element stiffness matrix and the element stresses. First, we calculate the element β’s and γ’s as:
220110 −=−=−=−=−=−= jmimji xxyy γβ
0002)1(0 =−=−==−−=−= mijimj xxyy γβ
202101 =−=−=−=−−=−= ijmjim xxyy γβ Therefore, the [B] matrix is:
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 265
[ ]0 0 0 1 0 2 0 1 0
1 10 0 0 0 2 0 0 0 22 2(2)
2 1 0 2 2 1
i j m
i j m
i i j j m m
BA
β β βγ γ γ
γ β γ β γ β
⎡ ⎤ − −⎡ ⎤⎢ ⎥ ⎢ ⎥= = −⎢ ⎥ ⎢ ⎥⎢ ⎥ ⎢ ⎥− − −⎣ ⎦⎣ ⎦
For plane stress conditions, the [D] matrix is:
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡
−×
=375.0000125.0025.01
)25.0(11030][ 2
6
D
Substitute the above expressions for [D] and [B] into the general equations for the stiffness matrix:
]][[][][ BDBtAk T=
6
1 0 20 2 1
1 0.25 0 1 0 2 0 1 02 0 1(2)30 10 10.25 1 0 0 2 0 0 0 2
4(0.9375) 2(2)2 0 20 0 0.375 2 1 0 2 2 1
1 0 20 2 1
k
− −⎡ ⎤⎢ ⎥− −⎢ ⎥ − −⎡ ⎤ ⎡ ⎤⎢ ⎥× ⎢ ⎥ ⎢ ⎥= −⎢ ⎥ ⎢ ⎥ ⎢ ⎥⎢ ⎥ ⎢ ⎥ ⎢ ⎥− − −⎣ ⎦ ⎣ ⎦⎢ ⎥−⎢ ⎥
−⎣ ⎦ Performing the matrix triple product gives:
inlbk
⎥⎥⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢⎢⎢
⎣
⎡
−−−−−−−−−−
−−−−−−−
−−−
×=
375.425.175.01625.325.025.15.25.1225.05.075.05.15.1075.05.1
120412625.325.075.01375.425.125.05.05.1225.15.2
104 6
The in-plane stress can be related to displacements by:
]][[ dBD=σ
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 266
6
0.00.0025
1 0.25 0 1 0 2 0 1 00.001230 10 10.25 1 0 0 2 0 0 0 2
0.9375 2(2) 0.00 0 0.375 2 1 0 2 2 1
0.00.0025
x
y
xy
inin
in
σστ
⎧ ⎫⎪ ⎪⎪ ⎪⎧ ⎫ − −⎡ ⎤ ⎡ ⎤⎪ ⎪×⎪ ⎪ ⎢ ⎥ ⎢ ⎥= −⎨ ⎬ ⎨ ⎬⎢ ⎥ ⎢ ⎥
⎪ ⎪ ⎪ ⎪⎢ ⎥ ⎢ ⎥− − −⎣ ⎦ ⎣ ⎦⎩ ⎭ ⎪ ⎪⎪ ⎪⎩ ⎭
The stresses are:
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡
−=
⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧
psipsipsi
xy
y
x
000,15800,4200,19
τσσ
Recall, the relationships for principal stresses and principal angle in two-dimensions are:
max2
2
1 22στ
σσσσσ =+⎟⎟
⎠
⎞⎜⎜⎝
⎛ −+
+= xy
yxyx
min2
2
2 22στ
σσσσσ =+⎟⎟
⎠
⎞⎜⎜⎝
⎛ −−
+= xy
yxyx
⎥⎦
⎤⎢⎣
⎡
−= −
yx
xyp σσ
τθ
2tan
21 1
Therefore:
( ) psi639,28000,152
800,4200,192
800,4200,19 22
1 =−+⎟⎠⎞
⎜⎝⎛ −
++
=σ
( ) psi639,4000,152
800,4200,192
800,4200,19 22
1 −=−+⎟⎠⎞
⎜⎝⎛ −
−+
=σ
op 3.32
800,4200,19)000,15(2tan
21 1 −=⎥⎦
⎤⎢⎣⎡
−−
= −θ
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 267
Treatment of Body and Surface Forces
The general force vector is defined as:
∫∫ ++=S
T
V
T dSTNPdVXNf ][][
Body Force
Let’s consider the first term of the above equation.
[ ] = ∫ Tb
V
f N X dV
where
⎧ ⎫= ⎨ ⎬⎩ ⎭
b
b
XX
Y
where Xb and Yb are the weight densities in the x and y directions, respectively. The force may reflect the effects of gravity, angular velocities, or dynamic inertial forces.
The integration of the fb is simplified if the origin of the coordinate system is chosen at the centroid of the element, as shown in the figure below. With the ori-gin placed at the centroid, we can use the definition of a centroid.
For a given thickness, t, the body force term becomes: [ ] [ ] T T
bV A
f N X dV t N X dA= =∫ ∫
0=∫A
x dA
0=∫A
y dA
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 268
Recall the interpolation functions for a place stress/strain triangle:
( ) ( ) ( )yxA
NyxA
NyxA
N mmmmjjjjiiii γβαγβαγβα ++=++=++=21
21
21
Therefore the terms in the integrand are:
0β γ= =∫ ∫i iA A
x dA y dA
and 23i j mAα α α= = =
The body force at node i is given as:
3⎧ ⎫
= ⎨ ⎬⎩ ⎭
bbi
b
XtAfY
The general body force vector is:
3
bix b
biy b
bjx bb
bjy b
bmx b
bmy b
f Xf Yf XtAff Yf Xf Y
⎧ ⎫ ⎧ ⎫⎪ ⎪ ⎪ ⎪⎪ ⎪ ⎪ ⎪⎪ ⎪ ⎪ ⎪⎪ ⎪= =⎨ ⎬ ⎨ ⎬⎪ ⎪ ⎪ ⎪⎪ ⎪ ⎪ ⎪⎪ ⎪ ⎪ ⎪⎪ ⎪ ⎩ ⎭⎩ ⎭
Surface Force
The third term in the general force vector is defined as:
[ ] = ∫ Ts
S
f N T dS
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 269
Let’s consider the example of a uniform stress p acting between nodes 1 and 3 on the edge of element 1 as shown in figure below.
The surface traction becomes:
0
⎧ ⎫ ⎧ ⎫= =⎨ ⎬ ⎨ ⎬
⎩ ⎭⎩ ⎭
x
y
p pT
p
and [N]T is:
1
1
2
2
3
3
00
0[ ]
00
0
⎡ ⎤⎢ ⎥⎢ ⎥⎢ ⎥
= ⎢ ⎥⎢ ⎥⎢ ⎥⎢ ⎥⎣ ⎦
T
NN
NN
NN
N
evaluated at x=a
Therefore, the traction force vector is:
1
1
2
20 0
3
3
00
00 0
00
⎡ ⎤⎢ ⎥⎢ ⎥⎢ ⎥ ⎧ ⎫
= ⎢ ⎥ ⎨ ⎬⎩ ⎭⎢ ⎥
⎢ ⎥⎢ ⎥⎣ ⎦
∫ ∫t L
s
NN
N pf dy dz
NN
N
evaluated at x=a
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 270
After some simplification, the traction force vector is:
1
2
0
3
0
0
0
⎡ ⎤⎢ ⎥⎢ ⎥⎢ ⎥
= ⎢ ⎥⎢ ⎥⎢ ⎥⎢ ⎥⎣ ⎦
∫L
s
N p
N pf t dy
N p
The interpolation function for i = 1 is:
( )12
α β γ= + +i i i iN x yA
For convenience, let’s choose the coordinate system shown in the figure below.
Recall:
α = −i j m j mx y y x
with i = 1, j = 2, and m = 3, we get
1 2 3 2 3α = −x y y x If we substitute the coordinates of the triangle show above in the above equation we get:
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 271
1 0α = Similarly, we can find:
1 10β γ= = a Therefore, the interpolation function, N1 is:
1 2=
ayNA
The remaining interpolation function, N2 and N2 are:
2 3( )2 2− −
= =L a x Lx ayN N
A A
Substituting the interpolation function in the traction force vector expression gives:
1
1
2
2
3
3
100
2 010
⎧ ⎫ ⎡ ⎤⎪ ⎪ ⎢ ⎥⎪ ⎪ ⎢ ⎥⎪ ⎪ ⎢ ⎥⎪ ⎪= =⎨ ⎬ ⎢ ⎥⎪ ⎪ ⎢ ⎥⎪ ⎪ ⎢ ⎥⎪ ⎪ ⎢ ⎥⎪ ⎪ ⎣ ⎦⎩ ⎭
s x
s y
s xs
s y
s x
s y
fff pLtffff
Explicit Expression for the Constant-Strain Triangle Stiffness Matrix
Usually the stiffness matrix is computed internally by computer programs, but since we are not computers, we need to explicitly evaluate the stiffness matrix. For a constant-strain triangular element, considering the plane strain case, recall that:
[ ] [ ] [ ][ ]= Tk tA B D B where [D] for plane strain is:
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 272
( )( )⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡
−−
−
−+=
ννν
νν
νν5.0000101
211][ ED
Substituting the appropriate definition into the above triple product gives:
00
1 00
[ ] 1 004 (1 )(1 2 )
0 0 0.50
0
β γγ β
ν νβ γ
ν νγ βν ν
νβ γ
γ β
⎡ ⎤⎢ ⎥⎢ ⎥ −⎡ ⎤⎢ ⎥ ⎢ ⎥= −⎢ ⎥ ⎢ ⎥+ − ⎢ ⎥ ⎢ ⎥−⎣ ⎦⎢ ⎥⎢ ⎥⎣ ⎦
i i
i i
j j
j j
m m
m m
tEkA
0 0 0
0 0 0β β β
γ γ γγ β γ β γ β
⎡ ⎤⎢ ⎥×⎢ ⎥⎢ ⎥⎣ ⎦
i j m
i j m
i i j j m m
Therefore the global stiffness matrix is:
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 273
The stiffness matrix is a function of the global coordinates x and y, the material properties, and the thickness and area of the element.
Finite Element Solution of a Plane Stress Problem
Consider the thin plate subjected to the surface traction shown in the figure below.
Assume plane stress conditions. Let E = 30 x 106 psi, ν = 0.30, and t = 1 in. De-termine the nodal displacements and the element stresses.
Discretization Let’s discretize the plate into two elements as shown below:
This level of discretization will probably not yield practical results for displace-ment and stresses: however; it is useful example for a longhand solution.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 274
The tensile traction forces can be converted into nodal forces as follows:
1
1
2
2
3
3
1 1 5,0000 0 00 0 01,000 (1 )10
2 20 0 01 1 5,0000 0 0
s x
s y
s xs
s y
s x
s y
f lbff pLt psi in infff lbf
⎧ ⎫ ⎡ ⎤ ⎡ ⎤ ⎡ ⎤⎪ ⎪ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥⎪ ⎪ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥⎪ ⎪ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥⎪ ⎪= = = =⎨ ⎬ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥⎪ ⎪ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥⎪ ⎪ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥⎪ ⎪ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥⎪ ⎪ ⎣ ⎦ ⎣ ⎦ ⎣ ⎦⎩ ⎭
The governing global matrix equations are:
[ ] =F K d Expanding the above matrices gives:
1 11
1 11
2 22
2 22
3 3 3
3 3 3
4 4 4
4 4 4
0000
[ ] [ ]5,000
05,000
0
⎧ ⎫ ⎧ ⎫ ⎧ ⎫⎧ ⎫⎪ ⎪ ⎪ ⎪ ⎪ ⎪⎪ ⎪⎪ ⎪ ⎪ ⎪ ⎪ ⎪⎪ ⎪⎪ ⎪ ⎪ ⎪ ⎪ ⎪⎪ ⎪⎪ ⎪ ⎪ ⎪ ⎪ ⎪⎪ ⎪⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪= = =⎨ ⎬ ⎨ ⎬ ⎨ ⎬ ⎨ ⎬⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪⎩ ⎭⎩ ⎭ ⎩ ⎭ ⎩ ⎭
x xx
y yy
x xx
y yy
x x x
y y y
x x x
y y y
F dRF dRF dRF dR
K KF d dlbF d dF d dlbF d d
⎪⎪⎪⎪⎪
where [K] is an 8 x 8 matrix before deleting the rows and columns accounting for the boundary supports.
Assemblage of the Stiffness Matrix The global stiffness matrix is assembled by superposition of the individual ele-ment stiffness matrices. The element stiffness matrix is:
[ ] [ ] [ ][ ]= Tk tA B D B
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 275
For element 1: the coordinates are xi = 0, yi = 0, xj = 20, yj = 10, xm = 0, and ym = 10. The area of the triangle is:
The matrix [B] is:
0 0 01[ ] 0 0 0
2
β β βγ γ γ
γ β γ β γ β
⎡ ⎤⎢ ⎥= ⎢ ⎥⎢ ⎥⎣ ⎦
i j m
i j m
i i j j m m
BA
We need to calculate the element β’s and γ’s as:
10 10 0 0 20 20β γ= − = − = = − = − = −i j m i m jy y x x
1 10 0 10 0 0 0β γ= − = − = = − = − =j m j i my y x x
0 10 10 20 0 20β γ= − = − = − = − = − =m i j m i jy y x x
Therefore, the [B] matrix is:
0 0 10 0 10 01 1[ ] 0 20 0 0 0 20
20020 0 0 10 20 10
−⎡ ⎤⎢ ⎥= −⎢ ⎥⎢ ⎥− −⎣ ⎦
B in
For plane stress conditions, the [D] matrix is:
61 0.3 0
30 10[ ] 0.3 1 00.91
0 0 0.35
⎡ ⎤× ⎢ ⎥= ⎢ ⎥
⎢ ⎥⎣ ⎦
D psi
2=
bhA
2(20)(10) 100 .2
= =A in
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 276
Therefore:
6
0 0 200 20 0
1 0.3 010 0 030(10 )[ ] [ ] 0.3 1 0
200(0.91) 0 0 100 0 0.35
10 0 200 20 10
−⎡ ⎤⎢ ⎥−⎢ ⎥ ⎡ ⎤⎢ ⎥ ⎢ ⎥= ⎢ ⎥ ⎢ ⎥⎢ ⎥ ⎢ ⎥⎣ ⎦⎢ ⎥−⎢ ⎥
−⎣ ⎦
TB D
Simplifying the above expression gives:
6
0 0 76 20 0
10 3 030(10 )[ ] [ ]200(0.91) 0 0 3.5
10 3 76 20 3.5
TB D
−⎡ ⎤⎢ ⎥− −⎢ ⎥⎢ ⎥
= ⎢ ⎥⎢ ⎥⎢ ⎥− −⎢ ⎥
−⎣ ⎦
The element stiffness matrix is:
[ ] [ ] [ ][ ]= Tk tA B D B therefore:
6
0 0 76 20 0
10 3 0(0.15)(10 )[ ] [ ][ ] 1(100)0.91 0 0 3.5
10 3 76 20 3.5
TtA B D B
−⎡ ⎤⎢ ⎥− −⎢ ⎥⎢ ⎥
= ⎢ ⎥⎢ ⎥⎢ ⎥− −⎢ ⎥
−⎣ ⎦
0 0 10 0 10 0
1 0 20 0 0 0 20200
20 0 0 10 20 10
−⎡ ⎤⎢ ⎥× −⎢ ⎥⎢ ⎥− −⎣ ⎦
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 277
Simplifying the above expression gives:
1 1 3 3 2 2
140 0 0 70 140 700 400 60 0 60 4000 60 100 0 100 6075,000[ ]
0.91 70 0 0 35 70 35140 60 100 70 240 130
70 400 60 35 130 435
u v u v u v
k
− −⎡ ⎤⎢ ⎥− − −⎢ ⎥⎢ ⎥− −
= ⎢ ⎥− −⎢ ⎥⎢ ⎥− − −⎢ ⎥
− − −⎣ ⎦
For element 2: the coordinates are xi = 0, yi = 0, xj = 20, yj = 0, xm = 20, and ym = 10. The area of the triangle is:
We need to calculate the element β’s and γ’s as:
0 10 10 20 20 0β γ= − = − = − = − = − =i j m i m jy y x x
1 10 0 10 0 20 20β γ= − = − = = − = − = −j m j i my y x x
0 0 0 20 0 20β γ= − = − = = − = − =m i j m i jy y x x
Therefore, the [B] matrix is:
10 0 10 0 0 01 1[ ] 0 0 0 20 0 20
2000 10 20 10 20 0
−⎡ ⎤⎢ ⎥= −⎢ ⎥⎢ ⎥− −⎣ ⎦
B in
2(20)(10) 100 .2
= =A in
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 278
For plane stress conditions, the [D] matrix is:
61 0.3 0
30 10[ ] 0.3 1 00.91
0 0 0.35
⎡ ⎤× ⎢ ⎥= ⎢ ⎥
⎢ ⎥⎣ ⎦
D psi
Therefore:
6
10 0 00 0 10
1 0.3 010 0 2030(10 )[ ] [ ] 0.3 1 0
200(0.91) 0 20 100 0 0.35
0 0 200 20 0
−⎡ ⎤⎢ ⎥−⎢ ⎥ ⎡ ⎤⎢ ⎥− ⎢ ⎥= ⎢ ⎥ ⎢ ⎥−⎢ ⎥ ⎢ ⎥⎣ ⎦⎢ ⎥⎢ ⎥⎣ ⎦
TB D
Simplifying the above expression gives:
6
10 3 00 0 3.5
10 3 730(10 )[ ] [ ]200(0.91) 0 20 3.5
6 0 76 20 0
− −⎡ ⎤⎢ ⎥−⎢ ⎥⎢ ⎥−
= ⎢ ⎥−⎢ ⎥⎢ ⎥−⎢ ⎥⎣ ⎦
TB D
The element stiffness matrix is:
[ ] [ ] [ ][ ]= Tk tA B D B therefore:
6
10 3 00 0 3.5
10 3 7(0.15)(10 )[ ] [ ][ ] 1(100)0.91 0 20 3.5
6 0 76 20 0
− −⎡ ⎤⎢ ⎥−⎢ ⎥⎢ ⎥−
= ⎢ ⎥−⎢ ⎥⎢ ⎥−⎢ ⎥⎣ ⎦
TtA B D B
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 279
10 0 10 0 0 0
1 0 0 0 20 0 20200
0 10 20 10 20 0
−⎡ ⎤⎢ ⎥× −⎢ ⎥⎢ ⎥− −⎣ ⎦
Simplifying the above expression gives:
1 1 4 4 3 3
100 0 100 60 0 600 35 70 35 70 0100 70 240 130 140 6075,000[ ]
0.91 60 35 130 435 70 4000 70 140 70 140 060 0 60 400 0 400
u v u v u v
k
− −⎡ ⎤⎢ ⎥− −⎢ ⎥⎢ ⎥− − −
= ⎢ ⎥− − −⎢ ⎥⎢ ⎥− −⎢ ⎥− −⎣ ⎦
Element 1:
Element 2:
1 1 2 2 3 3 4 4
28 0 28 14 0 14 0 0
0 80 12 80 12 0 0 0
28 12 48 26 20 14 0 0
14 80 26 87 12 7 0 0375,000[ ]
0.91 0 12 20 12 20 0 0 0
14 0 14 7 0 7 0 0
0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0
− −
− −
− − −
− − −=
− −
− −
⎡ ⎤⎢ ⎥⎢ ⎥⎢ ⎥⎢ ⎥⎢ ⎥⎢ ⎥⎢ ⎥⎢ ⎥⎢ ⎥⎢ ⎥⎣ ⎦
u v u v u v u v
k
1 1 2 2 3 3 4 4
20 0 0 0 0 12 20 12
0 7 0 0 14 0 14 7
0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0375,000[ ]
0.91 0 14 0 0 28 0 28 14
12 0 0 0 0 80 12 80
20 14 0 0 28 12 48 26
12 7 0 0 14 80 26 87
− −
− −
=− −
− −
− − −
− − −
⎡ ⎤⎢ ⎥⎢ ⎥⎢ ⎥⎢ ⎥⎢ ⎥⎢ ⎥⎢ ⎥⎢ ⎥⎢ ⎥⎢ ⎥⎣ ⎦
u v u v u v u v
k
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 280
Using the superposition, the total global stiffness matrix is:
1 1 2 2 3 3 4 4
48 0 28 14 0 26 20 12
0 87 12 80 26 0 14 7
28 12 48 26 20 14 0 0
14 80 26 87 12 7 0 0375,000[ ]
0.91 0 26 20 12 48 0 28 14
26 0 14 7 0 87 12 80
20 14 0 0 28 12 48 26
12 7 0 0 14 80 26 87
− − −
− − −
− − −
− −=
− − −
− − −
− − −
− − −
⎡ ⎤⎢ ⎥⎢ ⎥⎢ ⎥⎢ ⎥⎢ ⎥⎢ ⎥⎢ ⎥⎢ ⎥⎢ ⎥⎢ ⎥⎣ ⎦
u v u v u v u v
k
The governing global matrix equations are:
[ ] =F K d
1
1
2
2
48 0 28 14 0 26 20 12
0 87 12 80 26 0 14 7
28 12 48 26 20 14 0 0
14 80 26 87 12 7 0 0
0 26 20 12 48 0 28 14
26 0 14 7 0 87 12 80
20 14 0 0 28 12 48 26
12 7 0 0 14 80 26 87
375,0000.915,000
0500
0
− − −
− − −
− − −
− −
− − −
− − −
− − −
− − −
⎧ ⎫ ⎡⎪ ⎪ ⎢⎪ ⎪ ⎢⎪ ⎪ ⎢⎪ ⎪⎪ ⎪ =⎨ ⎬⎪ ⎪⎪ ⎪⎪ ⎪⎪ ⎪⎪ ⎪ ⎣⎩ ⎭
x
y
x
y
RRRR
lb
lb
1
1
2
2
3
3
4
4
⎧ ⎫⎤⎪ ⎪⎥⎪ ⎪⎥⎪ ⎪⎥⎪ ⎪⎢ ⎥⎪ ⎪⎢ ⎥ ⎨ ⎬⎢ ⎥ ⎪ ⎪⎢ ⎥ ⎪ ⎪⎢ ⎥ ⎪ ⎪⎢ ⎥ ⎪ ⎪⎢ ⎥ ⎪ ⎪⎦ ⎩ ⎭
x
y
x
y
x
y
x
y
dddddddd
Applying the boundary conditions:
1 1 2 2 0= = = =x y x yd d d d
The governing equations are:
3
3
4
4
5,000 48 0 28 140 0 87 12 80375,000
0.915,000 28 12 48 260 14 80 26 87
x
y
x
y
dlbddlbd
⎧ ⎫−⎧ ⎫ ⎡ ⎤⎪ ⎪⎪ ⎪ ⎢ ⎥−⎪ ⎪ ⎪ ⎪⎢ ⎥=⎨ ⎬ ⎨ ⎬⎢ ⎥− −⎪ ⎪ ⎪ ⎪⎢ ⎥⎪ ⎪ ⎪ ⎪− −⎩ ⎭ ⎣ ⎦ ⎩ ⎭
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 281
Solving the equations gives:
( )3
3 6
4
4
609.64.2
10663.7104.1
−
⎧ ⎫ ⎧ ⎫⎪ ⎪ ⎪ ⎪⎪ ⎪ ⎪ ⎪=⎨ ⎬ ⎨ ⎬⎪ ⎪ ⎪ ⎪⎪ ⎪ ⎪ ⎪⎩ ⎭⎩ ⎭
x
y
x
y
dd
indd
The exact solution for the displacement at the free end of the one-dimensional bar subjected to a tensile force is:
66
(10,000)20 670 1010(30 10 )
δ −= = = ××
PL inAE
The in-plane stress can be related to displacements by:
[ ][ ] =s D B d
( )2
1 0 0 0 0 1 0 0 0 0
2 (1 )0 0 0.5 1
ν β β βσ ν γ γ γ
νν γ β γ β γ β
⎧ ⎫⎪ ⎪⎪ ⎪⎡ ⎤⎡ ⎤⎪ ⎪⎪ ⎪⎢ ⎥⎢ ⎥= ⎨ ⎬⎢ ⎥⎢ ⎥− ⎪ ⎪⎢ ⎥⎢ ⎥−⎣ ⎦ ⎣ ⎦ ⎪ ⎪⎪ ⎪⎪ ⎪⎩ ⎭
ix
iyi j m
jxi j m
jyi i j j m m
mx
my
dddEdAdd
Element 1:
( )
1
11 3 2
31 3 22
31 1 3 3 2 2
2
2
1 0 0 0 0 1 0 0 0 0
2 (1 )0 0 0.5 1
ν β β βσ ν γ γ γ
νν γ β γ β γ β
⎧ ⎫⎪ ⎪⎪ ⎪⎡ ⎤ ⎡ ⎤ ⎪ ⎪⎪ ⎪⎢ ⎥ ⎢ ⎥= ⎨ ⎬⎢ ⎥ ⎢ ⎥− ⎪ ⎪⎢ ⎥ ⎢ ⎥− ⎣ ⎦⎣ ⎦ ⎪ ⎪⎪ ⎪⎪ ⎪⎩ ⎭
x
y
x
y
x
y
dddEdAdd
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 282
6 6
0.00.0
1 0.3 0 0 0 10 0 10 0609.630(10 )(10 ) 0.3 1 0 0 20 0 0 0 20
0.96(200) 4.20 0 0.35 20 0 0 10 20 10
0.00.0
σστ
−
⎧ ⎫⎪ ⎪⎪ ⎪⎧ ⎫ −⎡ ⎤ ⎡ ⎤⎪ ⎪⎪ ⎪ ⎢ ⎥ ⎢ ⎥= −⎨ ⎬ ⎨ ⎬⎢ ⎥ ⎢ ⎥
⎪ ⎪ ⎪ ⎪⎢ ⎥ ⎢ ⎥− −⎣ ⎦ ⎣ ⎦⎩ ⎭ ⎪ ⎪⎪ ⎪⎩ ⎭
x
y
xy
The stresses are:
1,0053012.4
σστ
⎧ ⎫ ⎡ ⎤⎪ ⎪ ⎢ ⎥=⎨ ⎬ ⎢ ⎥⎪ ⎪ ⎢ ⎥⎣ ⎦⎩ ⎭
x
y
xy
psipsipsi
Element 2:
( )
1
11 4 3
41 4 32
41 1 4 4 3 3
3
3
1 0 0 0 0 1 0 0 0 0
2 (1 )0 0 0.5 1
ν β β βσ ν γ γ γ
νν γ β γ β γ β
⎧ ⎫⎪ ⎪⎪ ⎪⎡ ⎤ ⎡ ⎤ ⎪ ⎪⎪ ⎪⎢ ⎥ ⎢ ⎥= ⎨ ⎬⎢ ⎥ ⎢ ⎥− ⎪ ⎪⎢ ⎥ ⎢ ⎥− ⎣ ⎦⎣ ⎦ ⎪ ⎪⎪ ⎪⎪ ⎪⎩ ⎭
x
y
x
y
x
y
dddEdAdd
6 6
0.00.0
1 0.3 0 10 0 10 0 0 0663.730(10 )(10 ) 0.3 1 0 0 0 0 20 0 20
0.96(200) 104.10 0 0.35 0 10 20 10 20 0
609.64.2
σστ
−
⎧ ⎫⎪ ⎪⎪ ⎪⎧ ⎫ ⎡ ⎤ ⎡ ⎤⎪ ⎪⎪ ⎪ ⎢ ⎥ ⎢ ⎥= −⎨ ⎬ ⎨ ⎬⎢ ⎥ ⎢ ⎥
⎪ ⎪ ⎪ ⎪⎢ ⎥ ⎢ ⎥−⎣ ⎦ ⎣ ⎦⎩ ⎭ ⎪ ⎪⎪ ⎪⎩ ⎭
x
y
xy
The stresses are:
9951.22.4
σστ
⎧ ⎫ ⎡ ⎤⎪ ⎪ ⎢ ⎥= −⎨ ⎬ ⎢ ⎥⎪ ⎪ ⎢ ⎥−⎣ ⎦⎩ ⎭
x
y
xy
psipsipsi
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 283
The principal stresses and principal angle are:
22
1995 1.2 995 1.2 ( 2.4) 995
2 2− +⎛ ⎞= + + − =⎜ ⎟
⎝ ⎠s psi
22
2995 1.2 995 1.2 ( 2.4) 1.1
2 2σ − +⎛ ⎞= − + − = −⎜ ⎟
⎝ ⎠psi
11 2( 2.4) 02 995 1.2
θ − −⎡ ⎤= ≈⎢ ⎥+⎣ ⎦o
p tan
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 284
WinFElt FEM Program To demonstrate a simple FEM solution for CST we will use the FElt (Finite ELe-menT) program. FElt is a free system for introductory level finite element analy-sis. It is primarily intended as a teaching tool for introductory type courses in finite elements - probably in the mechanical/structural/civil fields. In a command line environment, FElt uses an intuitive, straightforward input syntax to describe prob-lems. It also includes a graphical user interface for workstations that allows the user to set-up, solve and post-process the problem in a single CAD-like environ-ment. The Windows interface, WinFElt, consists of an editor and an encapsula-tor for the command line applications. It has some graphical post-processing ca-pabilities. The current version of FElt knows how to solve linear static and dynamic struc-tural and thermal analysis problems; it can also do modal and spectral analysis for dynamic problems. FElt's element library currently contains fourteen ele-ments. FElt uses an intuitive, ASCII based syntax for problem definition. This powerful syntax allows you to substitute analytic functions in place of numeric values (sin(60) instead of 0.866025) and even more importantly allows for time-dependent forcing and boundary conditions to be specified as analytic functions of time or in the more traditional fashion as a series of discrete time, magnitude pairs. This feature makes it quite easy to specify a wide range of functions. We can generate a color plot of stress contours for the above wrench problem with one simple menu selection after we have solved the problem in velvet. Dis-placement contours, and two- and three-dimensional wire frame plots of the dis-placed shape are also available post-processing options.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 285
WinFElt Manual Version 1.1.3 Problem description
The problem description section is used to define the problem title and the number of nodes and ele-ments in the problem. The problem description section is the only section which you cannot repeat within a given input file. The format for a problem-description is given below. problem description [ title = string ] [ nodes = integer ] [ elements = integer ] [ analysis = static | transient | modal | static-thermal | transient-thermal | spectral ]
nodes = integer elements = integer
These numbers will be used for error checking so the specifications given here must match the actual number of nodes and elements given in the definition sections. Note that the definitions for nodes and elements do not have to be given in numerical order, as long as nodes 1 ... m and elements 1 ... n (where m is the number of nodes and n is the number of elements) all get defined in one of the element and node definition sections in the file.
analysis = problem type
Defines the type of problem that you wish to solve. Currently it can either be static, transient, static-substitution, modal, static-thermal, transient-thermal, or spectral. If you do not specify anything, static analysis will be assumed.
Analysis parameters
The analysis parameters section is required only if you are doing some type of transient, modal, or spectral analysis (e.g., analysis=transient, analysis=spectral in the problem description section). For modal analysis it is simply used to set the type of element mass matrices that will be formed, but for tran-sient and spectral analyses it contains information that further defines the problem and the parameters for the numerical integration in time. analysis parameters [ alpha = expression ] [ beta = expression ] [ gamma = expression ] [ step = expression ] [ stop = expression ] [ Rm = expression ] [ Rk = expression ] [ nodes = [ node-list] ] [ dofs = [ dof-list] ] [ mass-mode = limped | consistent ] The alpha, beta, and gamma parameters are used in numerical integration schemes (transient and tran-sient-thermal analysis). start, stop, and step define the range of time or frequency interest for transient or spectral analyses. In transient analyses, start is meaningless and duration and dt can be used as ali-ases for stop and step, respectively. Rk and Rm are global Rayleigh (stiffness and mass) damping pro-portionality constants. The node-list is a comma or white space separated list of node numbers that are of interest in the analysis. Similarly, the dof-list is a list of the degrees of freedom (Tx, Ty , Tz, Rx, Ry, and Rz) that are of interest.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 286
An object-definition section defines objects of a specified type. Objects include nodes, elements, materi-als, constraints, forces, and distributed loads. Each of these types of objects is discussed below. Multiple object-definition sections are allowed and the sections may occur in any order.
Nodes
Nodes are points in Cartesian space to which elements are attached. A node must have a constraint and may have an optional force. A node is identified by a natural number. The syntax is as follows: nodes node-definitions where a node-definition takes the following form: node-number [ x = expression ] [ y = expression ] [ z = expression ] [ constraint = constraint-name ] [ force = force-name ] [ mass = expression ] The node-number starts the definition. Each node must have a unique number. If a Cartesian coordinate is not given then the coordinate of the previous node is used. Similarly, if no constraint is given then the constraint applied to the previous node is used. As above, the assignments can appear in any order and any number of times. As indicated above, some objects are identified by their name and some by their number. Elements and nodes have numbers while materials, forces, loads, and constraints have names.
Elements
Elements are linear, planar, or solid objects which are attached to nodes. Each element must have a ma-terial and may have optional loads. Furthermore, each element has a type, or definition. Like nodes, ele-ments are identified by a unique natural number. Elements of specific type are defined with the following syntax: element-type elements element-definition where an element-type is one of the following: spring truss beam beam3d CSTPlaneStrain CSTPlaneStress iso2d-PlaneStrain iso2d-PlaneStress quad-PlaneStrain quad-PlaneStress timoshenko htk brick ctg rod and an element-definition has the following form: element-number nodes = [ node-list ] [ material = material-name ] [ load = load-name-list ] The element-number starts the definition. Each element must have a unique number. If no material is given then the material applied to the previous element is used. The load-name-Iist is a list of up to three loads to apply to the element. The node-list is a comma or white space separated list of node numbers.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 287
Each type of element requires a certain number of nodes and in some cases a special "null node" which is numbered zero may be used to indicate a gap or filler in the list.
Materials
Elements are made of a type of material. Each material has a name and certain physical properties not all of which may be used by anyone element. The syntax for defining materials is as follows: material properties material-definitions where material-definition has the following form: material-name [ E = expression ] # Young's modulus [ Ix = expression ] # moment of inertia about x-x axis [ Iy = expression ] # moment of inertia about y-y axis [ Iz = expression ] # moment of inertia about z-z axis [ A = expression ] # cross-sectional area [ J = expression] # polar moment of inertia [ G = expression ] # bulk (shear) modulus [ t = expression ] # thickness [ rho = expression ] # density [ nu = expression ] # Poisson's ratio [ kappa = expression ] # shear force correction [ Rk = expression ] # Rayleigh damping coefficient (K) [ Rm = expression ] # Rayleigh damping coefficient (M) [ Kx = expression ] # thermal conductivity in the x-direction [ Ky = expression ] # thermal conductivity in the y-direction [ Ky = expression ] # thermal conductivity in the z-direction [ c = expression ] # heat capacitance The material-name starts the definition. If an attribute of a material is not specified then that attribute is zero. The assignments may occur in any order.
Constraints
Constraints are applied to nodes to indicate about which axes a node can move. The syntax for defining a constraint is as follows: constraints constraint -definitions where constraint-definition has the following form: constraint-name [ tx = c I u I expression ] # boundary translation along x axis [ ty = c I u I expression ] # boundary translation along y axis [ tz = c I u I expression ] # boundary translation along z axis [ rx = c I u I expression I h ] # boundary rotation about x axis [ ry = c I u I expression I h ] # boundary rotation about y axis [ rz = c I u I expression I h ] # boundary rotation about z axis [ itx = expression ] # initial displacement along x axis [ ity = expression ] # initial displacement along y axis [ itz = expression ] # initial displacement along z axis [ irx = expression ] # initial rotation about x axis [ iry = expression ] # initial rotation about y axis [ irz = expression ] # initial rotation about z axis
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 288
[ vx = expression ] # initial velocity along x axis [ vy = expression ] # initial velocity along y axis [ vz = expression ] # initial velocity along z axis [ ax = expression ] # initial acceleration. along x axis [ ay = expression ] # initial acceleration. along y axis [ az = expression ] # initial acceleration. along z axis The constraint-name starts the definition. A value of c for a boundary condition indicates that the axis is constrained; a value of u indicates that the axis is unconstrained. An expression indicates a displacement (non-zero) boundary condition and may contain the t variable for time varying boundary conditions in tra-nsient analysis problems. The initial displacement, velocity and acceleration specifications are only used in transient problems. A value of h for a rotational boundary condition indicates a hinge. By default, all axes are unconstrained.
Forces
Forces, or point loads, may be applied to nodes. The syntax for a force definition is as follows: forces force-definitions where a force-definition has the following form: force-name [ Fx = expression ] # force along x axis [ Fy = expression ] # force along y axis [ Fz = expression ] # force along z axis [ Mx = expression ] # moment about x axis [ My = expression ] # moment about y axis [ Mz = expression ] # moment about z axis [ Sfx = expression ] # frequency-domain spectra of force along x axis [ Sfy = expression ] # frequency-domain spectra of force along y axis [ Sfz = expression ] # frequency-domain spectra of force along z axis [ Smx = expression ] # frequency-domain spectra of moment about x axis [ Smy = expression ] # frequency-domain spectra of moment about y axis [ Smz = expression ] # frequency-domain spectra of moment about z axis The force-name starts the definition. If the force or moment is not specified then it is assumed to be zero. The expressions for forces may be time-varying. Time-varying expressions include the single variable t to represent the current time in the solution of a dynamic problem or consist of a list of discrete (time, value) pairs. Frequency varying expressions for spectra can also use w to represent the independent variable (radial frequency).
Loads
Distributed loads, or loads for short, are applied to elements. The syntax for a defining a distributed load is as follows: distributed loads load-definitions where a load-definition has the following form: load-name [ direction = dir ] # direction [ values = pair-list ] # local nodes and magnitudes The load-name starts the definition. The dir is one of LocalX, LocalY, LocalZ (local coordinate system), GlobalX, GlobalY, GlobalZ (global coordinate system), parallel, or perpendicular. The pair-list is a se-
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 289
quence of pairs. A pair is a node number and an expression enclosed in parentheses. The node number refers to the position within the element rather than referring to an actual node.
Expressions
An expression can be either constant or time-varying. As discussed above, time-varying expressions con-tain the variable t or consist of a list of discrete (time, value) pairs. If a time-varying expression is given where a constant expression is expected, the expression is evaluated at time zero. An expression has one of the following forms, where all operators have the precedences and associativities given to them in the C programming language. expression ? expression: expression # in-line conditional expression II expression # logical or expression && expression # logical and expression I expression # integer inclusive or expression ^expression # integer exclusive or expression & expression # integer and expression == expression # equality expression != expression # inequality expression < expression # less than expression > expression # greater than expression <= expression # less than or equal expression >= expression # greater than or equal expression << expression # integer shift left expression >> expression # integer shift right expression + expression # addition expression -expression # subtraction expression * expression # multiplication expression / expression # division expression % expression # integer remainder -expression # arithmetic negation ! expression # logical negation ~ expression # integer bitwise negation ( expression ) # enforce precedence sin ( expression ) # sine cos ( expression ) # cosine tan ( expression ) # tangent pow ( expression, expression ) # power (exponentiation) exp ( expression ) # exponential log ( expression ) # natural logarithm log10 ( expression ) # base-IO logarithm sqrt ( expression ) # square root hypot ( expression, expression ) # Euclidean distance floor ( expression ) # floor ceil ( expression ) # ceiling Cmod ( expression, expression ) # floating point remainder Cabs ( expression ) # absolute value number # literal value t # current time Finally, a discretely valued expression has the following syntax, where the optional + indicates that the list represents one cycle of an infinite waveform. ( expression ',' expression ) ...[ + ]
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 290
Below are a few examples of loading expressions
5 10 15 20 t(s)
500
F (lbs)
F (0,0.0) (4.99,500.0)= +
F 100 * fmod( ,5.0)t=
1 2 t(s)
1500
F (lbs)
F (0,0.0) (0.5,1500.0)(1,0)= +
F (fmod( ,1.0) 0.5 ? 3000 * fmod( ,0.5) :3000 * (0.5 - fmod( ,0.5)))
t tt
= <
2 4 6 8 t(s)
2000
F (lbs)
F (fmod( ,4.0) 2.0 ? 2000 : 0)t= <
F (0,2000) (1.99,2000) (2,0) (3.99,0)= +
12 t(s)
1500
F (lbs)
F (0,0.0) (3,650) (9,650) (12,0) (13,0)=
F 3 ? 650/3 * :( 9 ? 650 : ( 12 ? 650/3 * (12 ) : 0))
t tt t t
= << <= −
10 9 3
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 291
WinFElt Example 1: Use WinFElt to determine the nodal displacements and the element stresses.
Assume plane stress conditions. Let E = 30 x 106 psi, ν = 0.30, and t = 1 in. Con-sider the following discretization of two plane stress CST.
WinFElt input file: problem description title="CST Sample Problem (Logan 6.2, p.291)" nodes=4 elements=2 nodes 1 x=0 y=0.0 constraint=pin 2 x=0 y=10.0 constraint=pin 3 x=20.0 y=10.0 constraint=free force=point 4 x=20.0 y=0.0 constraint=free force=point CSTPlaneStress elements 1 nodes=[1,3,2] material=steel 2 nodes=[1,4,3] material=steel material properties steel e=30e06 nu=0.30 t=1.0 forces point fx=5000 constraints pin tx=c ty=c free tx=u ty=u end
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 292
WinFElt output file: ** CST Sample Problem (Logan 6.2, p.291) ** Nodal Displacements ----------------------------------------------------------------------------- Node # DOF 1 DOF 2 DOF 3 DOF 4 DOF 5 DOF 6 ----------------------------------------------------------------------------- 1 0 0 0 0 0 0 2 0 0 0 0 0 0 3 0.00060958 4.1633e-06 0 0 0 0 4 0.0006637 0.00010408 0 0 0 0 Element Stresses ------------------------------------------------------------------------------- 1: 1004.8 301.44 2.4019 1004.8 301.43 0.19566 2: 995.2 -1.201 -2.4019 995.2 -1.2068 -0.13812 Reaction Forces ----------------------------------- Node # DOF Reaction Force ----------------------------------- 1 Tx -5000 1 Ty -3002.4 2 Tx -5000 2 Ty 3002.4
The results are similar to those presented in the textbook on page 298. In addi-tion, WinFElt has some graphical visualization capabilities. By selecting the ap-propriate options on the Controls menu, a color contour of the displacements can be plotted.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 293
If more elements are utilized it may be possible to visualize the smoothed stress contours.
Note that the stress contours are not correct and most likely due to the fact the CST elements do not provide continuous inter-element stress values. WinFElt Example 2: Use WinFElt to determine the nodal displacements and the element stresses for the thin plate shown below.
Assume plane stress conditions. Let E = 210 GPa, ν = 0.30, and t = 5 mm. Con-sider the following discretization of four plane stress CST.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 294
WinFElt input file: problem description title="CST Sample Problem (Logan 6.11, p.305)" nodes=5 elements=4 nodes 1 x= 0.0 y= 0.0 constraint=free 2 x=400.0 y= 0.0 constraint=free 3 x=400.0 y=400.0 constraint=pin 4 x= 0.0 y=400.0 constraint=pin 5 x=200.0 y=200.0 constraint=free force=point CSTPlaneStress elements 1 nodes=[1,2,5] material=steel 2 nodes=[2,3,5] 3 nodes=[3,4,5] 4 nodes=[1,5,4] material=steel material properties steel e=210 nu=0.30 t=5.0 forces point fy=-30 constraints pin tx=c ty=c free tx=u ty=u end
WinFElt output file: ** CST Sample Problem (Logan 6.11, p.305) ** Nodal Displacements ----------------------------------------------------------------------------- Node # DOF 1 DOF 2 DOF 3 DOF 4 DOF 5 DOF 6 ----------------------------------------------------------------------------- 1 -0.0016515 -0.012505 0 0 0 0 2 0.0016515 -0.012505 0 0 0 0 3 0 0 0 0 0 0 4 0 0 0 0 0 0 5 4.9049e-19 -0.016279 0 0 0 0 Element Stresses ------------------------------------------------------------------------------- 1: 0.00059891 -0.003784 -1.0875e-19 0.00059891 -0.003784 -1.4217e-15 2: 0.0031171 0.0075 0.003716 0.0096226 0.00099449 -29.735 3: 0.0056352 0.018784 -1.9808e-19 0.018784 0.0056352 0 4: 0.0031171 0.0075 -0.003716 0.0096226 0.00099449 29.735 Reaction Forces ----------------------------------- Node # DOF Reaction Force ----------------------------------- 3 Tx 6.2341 3 Ty 15 4 Tx -6.2341 4 Ty 15
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 295
WinFElt Example 3: Use WinFElt to determine the nodal displacements and ro-tations, element forces, and the reactions for the rigid frame shown below (as-sume E = 3 x 106 psi, A = 15 in2, and I = 250 in4).
WinFElt input file: problem description title="Simple Frame Problem (Logan 5.8 p.243)" nodes=4 elements=3 nodes 1 x= 0.0 y= 0.0 constraint=fixed 2 x= 0.0 y=240.0 constraint=free 3 x=240.0 y=240.0 constraint=free 4 x=240.0 y= 0.0 constraint=fixed beam elements 1 nodes=[1,2] material=steel load=uni 2 nodes=[2,3] 3 nodes=[3,4] material=steel material properties steel E=30e06 A=15 Ix=250 distributed loads uni direction=perpendicular values=(1,-20.833) (2,-20.833) constraints fixed tx=c ty=c rz=c free tx=u ty=u rz=u end
WinFElt output file: ** Simple Frame Problem (Logan 5.8 p.243) ** Nodal Displacements ----------------------------------------------------------------------------- Node # DOF 1 DOF 2 DOF 3 DOF 4 DOF 5 DOF 6 ----------------------------------------------------------------------------- 1 0 0 0 0 0 0 2 0.24746 0.00038057 0 0 0 -0.00019435 3 0.2469 -0.00038057 0 0 0 -0.00072536 4 0 0 0 0 0 0 Element Stresses ------------------------------------------------------------------------------- 1: -713.57 3959.2 2.8118e+05 713.57 1040.7 69034 2: 1040.7 -713.57 -69034 -1040.7 713.57 -1.0222e+05 3: 713.57 1040.7 1.0222e+05 -713.57 -1040.7 1.4756e+05
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 296
Reaction Forces ----------------------------------- Node # DOF Reaction Force ----------------------------------- 1 Tx -3959.2 1 Ty -713.57 1 Rz 2.8118e+05 4 Tx -1040.7 4 Ty 713.57 4 Rz 1.4756e+05
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 297
Problems
16. Do problems 6.5, 6.6, 6.9, 6.10, and 6.13 on pages 301 - 306 in your text-book “A First Course in the Finite Element Method” by D. Logan.
17. Rework the plane stress problem given on page 291 in your textbook “A First Course in the Finite Element Method” by D. Logan using WinFElt to do analysis. Start with the simple two element model. Continuously refine your discretization by a factor of two each time until your FEM solution is in agreement with the exact solution for both displacements and stress. How many elements did you need?
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 298
Practical Considerations in Modeling; Interpreting Results
Introduction
In this section we will discuss some modeling considerations and guidelines, including mesh size, natural subdivisions, and the use of symmetry and associ-ated boundary conditions,
We will also introduce the concept of static condensation, which enables us to apply the basis of the CST stiffness matrix to a quadrilateral element.
Finite Element Modeling
Finite element modeling is partly an art guided by visualizing physical interac-tions taking place within a body, In modeling the user is confronted with the diffi-cult task of understanding physical behavior taking place and understanding the physical behavior of various elements available for use. Matching the appropriate finite element to the physical behavior being modeling is one of many decisions that must be made by the modeler. Understanding the boundary conditions can be one of the most difficult tasks a modeler must face in construction a useable finite element model.
Aspect Ratio and Element Shape The aspect ratio is define as the ratio of the longest dimension to the shortest
dimension of a quadrilateral element, In general, as the aspect ratio increases, the inaccuracy of the finite element solution increases,
Consider the five different finite element model shown in the figure below. A plot of the resulting error in the displacement at point A of the beam verse aspect ratio is given. In addition, the numerical answers are given in the following table.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 299
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 300
Case Aspect Ratio
Number of Nodes
Number of Elements
Point A Point B % Error
1 1.1 84 60 -1.093 -0.346 5.2 2 1.5 85 64 -1.078 -0.339 6.4 3 3.6 77 60 -1.014 -0.238 11.9 4 6.0 81 64 -0.886 -0.280 23.0 5 24.0 85 64 -0.500 -0.158 56.0
Exact Solution -1.152 -0.360
In general, elements that yield the best results are compact and regular in
shape will: (1) aspect ratios near one; and (2) corner angles of quadrilaterals near 90°.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 301
Use of Symmetry The use of symmetry will often expedite the modeling of a problem. Symme-
try allows us to consider a reduced problem instead of the actual problem. This will allow us to use a finer discretization of element with less computational cost.
Natural Subdivisions at Discontinuities There are a variety of natural subdivisions for finite element discretizations.
For example, natural locations of nodes occur at concentrated loads or disconti-nuities in loading, other types of boundary conditions, and abrupt changes in ge-ometry of materials.
Sizing of Elements and Mesh Refinement A discretization depends on the geometry of the structure, the loading, and the
boundary conditions. For example, areas of high, rapidity changing stresses re-quire a finer mesh then regions where the stress is constant.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 302
Here the use of symmetry is applied to a soil mass subjected to a foundation
loading (66 nodes and 50 elements). Note that at the place of symmetry the dis-placements in the direction perpendicular to the plane must be zero. This is mod-eled by rollers at nodes 2 - 6.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 303
The figure above illustrates the use of triangular elements for transitions from smaller quadrilaterals to larger quadrilaterals. The transitions are required since CST elements do not have immediate nodes along their edges. If an element had an intermediate node, the resulting equations would be inconsistent with the en-ergy formulation for the CST equations.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 304
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 305
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 306
Infinite Medium A typical example of infinite medium is a soil foundation problem. The guide-
line for the finite element model is that enough material must be included such that the displacements at nodes and stresses within the elements become negli-gibly small at locations far from the foundation load. The level of discretization can be determined by a trail-and-error procedure in which the horizontal and ver-tical distances from the load are varied and the resulting effects on the displace-ments and stresses are observed. For a homogeneous soil mass, experience has shown the influence of a footing becomes insignificant if the horizontal dis-tance of the model is taken as approximately four and six times the width of the footing and the vertical distance is taken as approximately four to ten times the width of the footing.
Checking the Model The discretized finite element model should be checked carefully before re-
sults are computed. Ideally, a model should be checked by an analyst not in-volved in the preparation of the model, who is then more likely to be objective. Preprocessors with their detailed graphical display capabilities now make it com-paratively easy to find errors, particularly with a misplaced node or missing ele-ment or a misplaced load or boundary condition. Preprocessors include the abil-ity to color, shrink, rotate, and section a model mesh.
Checking the Results and Typical Postprocessor Results An analyst should probability spend as much time processing, checking, and
analyzing results as spent in data preparation.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 307
Plate with a hole in the center.
Deformed shape of plate superimposed over an unreformed shape
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 308
The wrench in this example is modeled by 307 constraint strain triangular ele-ments (plane stress assumption). Below is a plot of the deformed shape of the wrench over the original mesh.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 309
On the left is a plot of the stress in the wrench and the plot on the right is a
plot of the defected shape. This analysis was preformed using WinFElt finite element structural analysis system.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 310
Equilibrium and Compatibility of Finite Element Results
An approximate solution for a stress analysis problem using the finite element method based on assumed displacement fields does not generally satisfy all the requirements for equilibrium and compatibility that an exact theory-of-elasticity solution satisfies. However, remember that relatively few exact solutions exist. Hence, the finite element method is a very practical one for obtaining reasonable, but approximate, numerical solutions.
We now describe some of the approximations generally inherent with finite element solutions.
1. Equilibrium of nodal forces and moments is satisfied. This is true be-cause the global equation F = Kd is a nodal equilibrium equation whose solution for d is such that the sums of all forces and moments applied to each node are zero. Equilibrium of the whole structure is also satisfied because the structure reactions are included in the global forces, and hence, in the nodal equilibrium equations.
2. Equilibrium within an element is not always satisfied. However, for the constant-strain bar and the constant-strain triangle, element equilibrium is satisfied. Also the cubic displacement function is shown to satisfy the basic beam equilibrium differential equation, and hence, to satisfy ele-ment force and moment equilibrium.
3. Equilibrium is not usually satisfied between elements. A differential ele-ment including parts of two adjacent finite elements is usually not in equilibrium (see the figure below). For line elements, such as used for truss and frame analysis, interelement equilibrium is satisfied. However, for two- and three-dimensional elements, interelement equilibrium is not usually satisfied. Also, the coarseness of the mesh causes this lack of interelement equilibrium to be even more pronounced.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 311
4. Compatibility is satisfied within an element as long as the element dis-
placement field is continuous; hence, individual elements do not tear apart.
5. In the formulation of the element equations, compatibility is invoked at the nodes. Hence, elements remain connected at their common nodes. Similarly, the structure remains connected to its support nodes because boundary conditions are invoked at these nodes.
6. Compatibility may or may not be satisfied along interelement bounda-ries. For line elements such as bars and beams, interelement bounda-ries are merely nodes. The constant-strain triangle remain straight sided when deformed and therefore, interelement compatibility exists for these elements. Incompatible elements, those that allow gaps or overlaps be-tween elements, can be acceptable and even desirable.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 312
Convergence of Solution
When the mesh size is reduced - that is the number of elements is increased - we are ensured of monotonic convergence of the solution when compatible and complete displacement functions are sued.
Case Number of Elements Number of Nodes Aspect Ratio
Point A
1 12 21 2 -0.740 2 24 39 1 -0.980 3 32 45 3 -0.875 4 64 85 1.5 -1.078 5 80 105 1.2 -1.100
Exact Solution -1.152
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 313
Interpretation of Stresses
In the stiffness or displacement formulation of the finite element method, used in this course, the primary quantities determined are the interelement nodal dis-placements of the assemblage. Secondary quantities, such as stress and strain, are computed based on these nodal displacements. In the case of the bar and constant-strain triangles, stresses are constant over the element. For these ele-ments, it is common practice to assign the stress to the centroid of the element with acceptable results.
An alternative procedure sometimes is to use an average (possibly weighted) value of the stresses evaluated at each node of the element. This averaging method is often based interpolating the element nodal values using the element shape functions. The averaging method is called smoothing. While the results from smoothing may be pleasing to the eye, they may not indicate potential prob-lems with the model and the results. You should always view the unsmoothed contour plots as well. Highly discontinuous contours between elements in a re-gion of an unsmoothed plot indicate modeling problems and typically require ad-ditional refinement of the element mesh in the suspect region.
Static Condensation
Let’s consider the concept of static condensation and used it to develop the stiffness matrix of a quadrilateral element. Consider a general quadrilateral ele-ment as shown below.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 314
An imaginary node 5 is temporary introduced at the intersection of the diagonals of the quadrilateral to create four triangles. We can superimpose the stiffness matrices of the four triangles to create the stiffness matrix of the quadrilateral element, where the internal imaginary node 5 degrees of freedom are said to be condensed out so that they never enter into the final equations.
Let’s start by partitioning the equilibrium equations:
where di is the vector of displacements corresponding to the imaginary internal node, Fi is the vector of loads at the internal node, and de and Fe are the actual displacements and loads, respectively. Rewriting the above equations we gives:
+ =11 12e i eK d K d F
21 22e i iK d K d F+ = Solving for di gives:
1 122 21 22i e id K K d K F− −= − +
Substituting the above equation, we obtain the condensed equilibrium equation:
=c c ck d F where
−= − 111 12 22 21ck K K K K
−= − 112 22c e iF F K K F
where kc and Fc are called the condensed stiffness matrix and the condensed load vector, respectively. An advantage of the four-CST quadrilaterals is that the solution becomes less dependent on the skew of the subdivision mesh. The skew means a directional stiffness bias that is built into a model through certain discretization patterns.
11 12
21 22
e e
i i
K K d FK K d F⎡ ⎤ ⎧ ⎫ ⎧ ⎫
=⎨ ⎬ ⎨ ⎬⎢ ⎥⎣ ⎦ ⎩ ⎭ ⎩ ⎭
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 315
The stiffness matrix of a typical triangular element, call it element 1, labeled with nodes 1, 2, and 5 is given as:
(1) (1) (1)11 12 15
(1) (1) (1) (1)21 22 25(1) (1) (1)51 52 55
k k kk k k k
k k k
⎡ ⎤⎢ ⎥⎡ ⎤ = ⎢ ⎥⎣ ⎦⎢ ⎥⎣ ⎦
where kij
(1) is a 2 x 2 matrix. The assembled stiffness matrix for the quadrilateral is:
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 316
Example Problem
Consider the quadrilateral with internal node 5 and dimensions as shown be-low. Apply the static condensation technique.
Using the CST stiffness matrix for plain strain, we get:
(1) (3)
1 2 5
3 4 5
1.5 1.0 0.1 0.2 1.6 1.21.0 3.0 0.2 2.6 0.8 5.60.1 0.2 1.5 1.0 1.6 1.2
[ ] [ ]4.16 0.2 2.6 1.0 3.0 0.8 5.6
1.6 0.8 1.6 0.8 3.2 0.01.2 5.6 1.2 5.6 0.0 11.2
Ek k
− −⎡ ⎤⎢ ⎥− − −⎢ ⎥⎢ ⎥− − −
= = ⎢ ⎥− −⎢ ⎥⎢ ⎥− − −⎢ ⎥− − −⎣ ⎦
(2) (4)
4 1 5
2 3 5
1.5 1.0 0.1 0.2 1.6 1.21.0 3.0 0.2 2.6 0.8 5.60.1 0.2 1.5 1.0 1.6 1.2
[ ] [ ]4.16 0.2 2.6 1.0 3.0 0.8 5.6
1.6 0.8 1.6 0.8 3.2 0.01.2 5.6 1.2 5.6 0.0 11.2
Ek k
− −⎡ ⎤⎢ ⎥− − −⎢ ⎥⎢ ⎥− − −
= = ⎢ ⎥− −⎢ ⎥⎢ ⎥− − −⎢ ⎥− − −⎣ ⎦
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 317
The resulting assembled matrix before static condensation is:
After partitioning, the condensed stiffness matrix is:
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 318
Flowchart for the Solution of Place Stress/Strain Problems The following flowchart is typical for a finite element process used for the
analysis of plane stress and plane strain problems.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 319
Problems
18. Do problems 7.1, 7.3, and 7.4 on pages 332 - 343 in your textbook “A First Course in the Finite Element Method” by D. Logan.
19. Work problems 7.7, 7.10, and 7.15 on pages 332 - 343 in your textbook “A First Course in the Finite Element Method” by D. Logan using WinFElt.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 320
Development of the Linear-Strain Triangle Equations
Introduction
In this section we will develop a higher-order triangular element, called the linear-strain triangle (LST). This element has many advantages over the con-stant-strain (CST). The LST element has six nodes and twelve displacement de-grees of freedom. The displacement function for the triangle is quadratic.
Derivation of the Linear-Strain Triangular Elemental Stiffness Matrix and Equations
The procedure to derive the LST element stiffness matrix and element equa-tions is identical to that used for the CST element.
Step 1 - Discretize and Select Element Types Consider the triangular element shown in the figure below:
Each node has two degrees of freedom: displacements in the x and y directions. We will let ui and vi represent the node i displacement components in the x and y directions, respectively.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 321
The nodal displacements for an LST element are:
⎧ ⎫⎪ ⎪⎪ ⎪⎪ ⎪⎪ ⎪
⎧ ⎫ ⎪ ⎪⎪ ⎪ ⎪ ⎪⎪ ⎪ ⎪ ⎪⎪ ⎪ ⎪ ⎪= =⎨ ⎬ ⎨ ⎬⎪ ⎪ ⎪ ⎪⎪ ⎪ ⎪ ⎪⎪ ⎪ ⎪ ⎪⎩ ⎭ ⎪ ⎪
⎪ ⎪⎪ ⎪⎪ ⎪⎪ ⎪⎩ ⎭
1
1
2
1 2
2 3
3 3
4 4
5 4
6 5
5
6
6
uvu
d vd ud v
dd ud vd u
vuv
where the nodes are ordered counterclockwise around the element, and
Step 2 - Select Displacement Functions Consider a straight-sided triangular element shown below
The variation of the displacements over the element may be expressed as
= + + + + +2 21 2 3 4 5 6( , )u x y a a x a y a x a xy a y
= + + + + +2 27 8 9 10 11 12( , )v x y a a x a y a x a xy a y
The displacement compatibility among adjoining elements is satisfied because the three nodes defining adjacent sides define a unique a parabola. The CST and LST triangles are variations of the Pascal triangles as show below.
1
3
2
y
x
6
4
5
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 322
Terms in Pascal Triangle Polynomial Degree
# of Terms
1 0 (constant) 1
x y
1 (linear) 3
x2 xy y2
2 (quadratic) 6
x3 x2y xy2 y3
3 (cubic) 10
The general element displacement functions are:
1
22 2
2 2
11
12
·1 0 0 0 0 0 0·0 0 0 0 0 0 1
aa
x y x xy yx y x xy y
aa
⎧ ⎫⎪ ⎪⎪ ⎪⎪ ⎪⎡ ⎤
Ψ = ⎨ ⎬⎢ ⎥⎣ ⎦ ⎪ ⎪
⎪ ⎪⎪ ⎪⎩ ⎭
or [ ] *M aΨ =
To obtain the values for the a’s substitute the coordinated of the nodal points into the above equations:
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 323
Solving for the a’s and writing the results in matrix forms gives:
or
[ ] 1a x u−=
where [x] is the 12 x 12 matrix on the right-hand-side of the above equation. The “best” way to invert [x] is to use a computer. Note that only the 6 x 6 part of [x] really need be inverted. The general displacement expressions in terms of shape functions and the nodal degrees of freedom are:
[ ] N dΨ =
where [ ] [ ][ ] 1*N M x −
=
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 324
Step 3 - Define the Strain-Displacement and Stress-Strain Relationships Elemental Strains: The strains over a two-dimensional element are:
εε ε
γ
⎧ ⎫∂⎪ ⎪
∂⎪ ⎪⎧ ⎫⎪ ⎪∂⎪ ⎪= =⎨ ⎬ ⎨ ⎬∂⎪ ⎪ ⎪ ⎪
⎩ ⎭ ⎪ ⎪∂ ∂+⎪ ⎪∂ ∂⎩ ⎭
x
y
xy
uxvy
u vy x
or
1
2
11
12
0 1 0 2 0 0 0 0 0 0 0·
0 0 0 0 0 0 0 1 0 0 2·
0 0 1 0 2 0 1 0 2 0
aa
x yx y
x y x yaa
ε
⎧ ⎫⎪ ⎪⎪ ⎪⎡ ⎤⎪ ⎪⎢ ⎥= ⎨ ⎬⎢ ⎥⎪ ⎪⎢ ⎥⎣ ⎦ ⎪ ⎪⎪ ⎪⎩ ⎭
Observe that the strains are linear over the triangular element; therefore, the element is called a linear-strain triangle (LST). The above equation may be written in matrix form as:
[ ] 'M aε =
where [M ’] is based on derivatives of [M*]. If we substitute the values of a’s into the above equation gives:
[ ] B dε =
where [B] is a function of the nodal coordinates (x1, y1) through (x6, y6).
[ ] [ ][ ] 1'B M x −=
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 325
The stresses are given as:
[ ]x x
y y
xy xy
Dσ εσ ετ γ
⎧ ⎫ ⎧ ⎫⎪ ⎪ ⎪ ⎪=⎨ ⎬ ⎨ ⎬⎪ ⎪ ⎪ ⎪⎩ ⎭ ⎩ ⎭
For plane stress, [D] is:
( )⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡
−−
=ν
νν
ν15.0000101
1][
2
ED
For plane strain, [D] is:
( )( )⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡
−−
−
−+=
ννν
νν
νν5.0000101
211][ ED
Step 4 - Derive the Element Stiffness Matrix and Equations The stiffness matrix can be defined as:
[ ] [ ] [ ][ ]T
V
k B D B dV= ∫
However, [B] is now a function of x and y; therefore, we must integrate the above expression to develop the element stiffness matrix. The [B] matrix is:
[ ]1 2 3 4 5 6
1 2 3 4 5 6
1 1 2 2 3 3 4 4 5 5 6 6
0 0 0 0 0 01 0 0 0 0 0 0
2B
A
β β β β β βγ γ γ γ γ γ
γ β γ β γ β γ β γ β γ β
⎡ ⎤⎢ ⎥= ⎢ ⎥⎢ ⎥⎣ ⎦
where the β’s and the γ’s are functions of x and y as well as the nodal coordi-nates. The stiffness matrix is a 12 x 12 matrix and is very cumbersome to com-pute in explicit form. However, if the origin of the coordinates is the centroid of the element, the integrations become more amenable. Typically, the integration are computed numerically.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 326
The element body forces and surface forces should net be automatically lumped at the nodes. The following integration should be computed.
[ ] Tb
V
f N X dV= ∫
[ ] Ts
S
f N T dS= ∫
The element equations are:
[ ] [ ] [ ]
1 11 12 13 1,12 1
1 21 22 23 2,12 1
2 31 32 33 3,12 2
6 12,1 12,2 12,3 12,12 6
12 1 12 12 12 1
x
y
x
y
f k k k k uf k k k k vf k k k k u
f k k k k v
⎧ ⎫ ⎡ ⎤ ⎧ ⎫⎪ ⎪ ⎢ ⎥ ⎪ ⎪⎪ ⎪ ⎢ ⎥ ⎪ ⎪⎪ ⎪ ⎪ ⎪⎢ ⎥=⎨ ⎬ ⎨ ⎬
⎢ ⎥⎪ ⎪ ⎪ ⎪⎢ ⎥⎪ ⎪ ⎪ ⎪⎢ ⎥⎪ ⎪ ⎪ ⎪⎩ ⎭⎣ ⎦⎩ ⎭
× × ×
Step 5, 6, and 7 Assembling the global stiffness matrix, determining the global displacements,
and calculating the stresses, are identical to the procedures used for CST ele-ments.
Example LST Stiffness Determination
Consider a straight-sided triangular element shown below:
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 327
The triangle has a base dimension of b and a height h, with mid-side nodes. We can calculate the coefficients a1 through a6 by evaluating the displacement u at each node.
1 1(0,0)u u a= =
22 1 2 4( ,0)u u b a a b a b= = + +
23 1 3 6(0, )u u h a a h a h= = + +
2 2
4 1 2 3 4 5 6,2 2 2 2 2 4 2b h b h b bh hu u a a a a a a⎛ ⎞ ⎛ ⎞ ⎛ ⎞= = + + + + +⎜ ⎟ ⎜ ⎟ ⎜ ⎟
⎝ ⎠ ⎝ ⎠ ⎝ ⎠
2
5 1 3 60,2 2 2h h hu u a a a⎛ ⎞ ⎛ ⎞= = + +⎜ ⎟ ⎜ ⎟
⎝ ⎠ ⎝ ⎠
2
6 1 2 4,02 2 2b b bu u a a a⎛ ⎞ ⎛ ⎞= = + +⎜ ⎟ ⎜ ⎟
⎝ ⎠ ⎝ ⎠
Solving the above equations simultaneously for the a’s gives:
6 1 2 5 1 31 1 2 3
4 3 4 3u u u u u ua u a ab h
− − − −= = =
( ) ( )2 6 1 1 4 5 64 52
2 2 4u u u u u u ua a
b bh− + + − −
= =
( )3 5 16 2
2 2u u ua
h− +
=
The u displacement equation is:
( )2 6 1 26 1 2 5 1 31 2
2 24 3 4 3( , )u u uu u u u u uu x y u x y x
b h b⎡ ⎤− +− − − −⎡ ⎤ ⎡ ⎤= + + + ⎢ ⎥⎢ ⎥ ⎢ ⎥⎣ ⎦ ⎣ ⎦ ⎣ ⎦
( ) ( )1 4 5 6 3 5 1 22
4 2 2u u u u u u uxy y
bh h⎡ ⎤ ⎡ ⎤+ − − − +
+ +⎢ ⎥ ⎢ ⎥⎣ ⎦ ⎣ ⎦
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 328
The v displacement equation can be determined in a manner identical to that used for the u displacement:
( )2 6 1 26 1 2 5 1 31 2
2 24 3 4 3( , )v v vv v v v v vv x y v x y x
b h b⎡ ⎤− +− − − −⎡ ⎤ ⎡ ⎤= + + + ⎢ ⎥⎢ ⎥ ⎢ ⎥⎣ ⎦ ⎣ ⎦ ⎣ ⎦
( ) ( )1 4 5 6 3 5 1 22
4 2 2v v v v v v vxy y
bh h⎡ ⎤ ⎡ ⎤+ − − − +
+ +⎢ ⎥ ⎢ ⎥⎣ ⎦ ⎣ ⎦
The general form of the displacement expressions in terms of the shape func-tions is given as:
1
11 2 3 4 5 6
1 2 3 4 5 66
6
0 0 0 0 0 00 0 0 0 0 0
uv
N N N N N NuN N N N N Nv
uu
⎧ ⎫⎪ ⎪⎪ ⎪⎡ ⎤⎧ ⎫ ⎪ ⎪=⎨ ⎬ ⎨ ⎬⎢ ⎥
⎩ ⎭ ⎣ ⎦ ⎪ ⎪⎪ ⎪⎪ ⎪⎩ ⎭
where the shape functions are:
2 2 2 2
1 2 32 2 2 23 3 2 4 2 2 21 x y x xy y x x y yN N Nb h b bh h b b h h
= − − + + + = − + = − +
2 2
4 5 62 24 4 4 4 4 4 4xy y xy y x xy xN N Nbh h bh h b bh b
= = − − = − −
The element interpolation functions N have two basic shapes. The behavior of the functions N1, N2, and N3 is similar except referenced at different nodes. The shape function N1 is shown below:
2
1
3 1
N1
y
x
6 4
5
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 329
The second type of shape function is valid for functions N4, N5, and N6. The func-tion N5 is shown below:
The element strain is gives as:
[ ] e B d=
where the [B] matrix is:
[ ]1 2 3 4 5 6
1 2 3 4 5 6
1 1 2 2 3 3 4 4 5 5 6 6
0 0 0 0 0 01 0 0 0 0 0 0
2B
A
β β β β β βγ γ γ γ γ γ
γ β γ β γ β γ β γ β γ β
⎡ ⎤⎢ ⎥= ⎢ ⎥⎢ ⎥⎣ ⎦
where the β’s and the γ’s are:
1 2 34 43 4 0hx hxh y hb b
β β β= − + + = − + =
4 5 684 4 4 4hxy y h y
bβ β β= = − = − −
1 2 34 43 4 0by byb x b
h hγ γ γ= − + + = = − +
4 5 684 4 4 4byx b x x
hγ γ γ= = − − = −
2
1
3 1
N5
y
x
6 4
5
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 330
The stiffness matrix for a constant thickness element can be obtained by substi-tuting the β’s and the γ’s into the [B] and then substituting [B] into the following expression and evaluating the integral numerically.
[ ] [ ] [ ][ ]T
V
k B D B dV= ∫
Comparison of Elements
For a given number of nodes, a better representation of true stress and dis-placement is generally obtained using LST elements than is obtained using the same number of nodes a finer subdivision of CST elements. For example, a sin-gle LST element gives better results than four CST elements.
Consider the following cantilever beam with E = 30 x 106 psi, ν = 0.25, and t = 1 in.
Table 1 lists the series of tests run to compare results using the CST and LST elements. Table 2 shows comparisons of free-end (tip) deflection and stress for each element type used to model the cantilever beam.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 331
Table 1. Comparison of CST and LST results
Series of Test Runs Number of Nodes
Degrees of Freedom, nd
Number of Elements
A-1 4 x 16 85 160 128 CST A-2 8 x 32 297 576 512 CST B-1 2 x 8 85 160 32 LST B-2 4 x 16 297 576 128 LST
Table 2. Comparison of CST and LST results
Runs nd Bandwidth, nd Tip Deflection
(in) σx
(ksi) Location
(x,y) A-1 160 14 -0.29555 67.236 (2.250,11.250) A-2 576 22 -0.33850 81.302 (1.125,11.630) B-1 160 18 -0.33470 58.885 (4.500,10.500) B-2 576 22 -0.35159 69.956 (2.250,11.250)
Exact Solution -0.36133 80.000 (0,12)
From Table 2, we can observe that: • The larger the number of degrees of freedom for a given type of triangular
element, the closer the solution converges to the exact one (compare run A-l to run A-2, and B-l to B-2).
• For a given number of nodes, the LST analysis yields somewhat better results than the CST analysis (compare run A-l to run B-l).
• Although the CST element is rather poor in modeling bending, we observe from Table 2 that the element can be used to model a beam in bending if suf-ficient number of elements is used through the depth of the beam.
• In general, both the LST and CST analyses yield sufficient results for most plane stress/strain problems provided a sufficient number of elements are used.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 332
Most commercial programs incorporate the use of CST and/or LST elements for plane stress/strain problems although these elements are used primarily as transition elements (usually during mesh generation). Also, recall that finite ele-ment displacements will always be less than the exact ones, because finite ele-ment models are always predicted to be stiffer than the actual structures when using the displacement formulation of the finite element method.
A comparison of CST and LST models of a plate subjected to parabolically distributed edge loads is shown in the figure below. The LST model converges to the exact solution for horizontal displacement at point A faster than does the CST model. However, the CST model is quite acceptable even for modest numbers of degrees of freedom. For example, a CST model with 100 nodes (200 degrees of freedom) often yields nearly as accurate a solution as does an LST model with the same number of degrees of freedom.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 333
The results of Table 2 indicate: • That the LST model might be preferred over the CST model for plane stress
applications when a relatively small number of nodes is used.
• That the use of triangular elements of higher order, such as the LST, is not visibly more advantageous when large numbers of nodes are used, particu-larly when the cost of formation of the element stiffnesses, equation band-width, and overall complexities involved in the computer modeling are consid-ered.
Problems
20. Do problems 8.3, 8.5, and 8.7 on pages 355 - 357 in your textbook “A First Course in the Finite Element Method” by D. Logan.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 335
Development of the Plate Bending Element
Introduction
In this section we will begin by describing elementary concepts of plate bend-ing behavior and theory. The plate element is one of the more important struc-tural elements and is used to model and analyze such structures as pressure vessels, chimney stacks, and automobile parts. A large number of plate bending element formulations exist that would require lengthy chapter to cover. The pur-pose in this chapter is to present the derivation of the stiffness matrix for one of the most common plate bending finite elements and then to compare solutions to some classical problems for a variety of bending elements in the literature.
Basic Concepts of Plate Bending
A plate can be considered the two-dimensional extension of a beam in simple bending. Both plates and beams support loads transverse or perpendicular to their plane and through bending action. A plate is a flat (if it were curved, it would be a shell). A beam has a single bending moment resistance, while a plate re-sists bending about two axes and has a twisting moment. We will consider the classical thin-plate theory or Kirchhoff plate theory.
Basic Behavior of Geometry and Deformation Consider the thin plate in the x-y plane of thickness t measured in the z direc-
tion shown in the figure below:
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 336
The plate surfaces are at z = ±t/2, and its midsurface is at z = 0. The basic ge-ometry of the plate is as follows:
1. The plate thickness is much smaller than its inplane dimensions b and c (that is, ort b c ). If t is more than about one-tenth the span of the plate, then transverse shear deformation must be accounted for and the plate is then said to be thick.
2. The deflection w is much less than the thickness t (than is, / 1w t ).
Kirchhoff Assumptions Consider the differential slice cut from the plate by planes perpendicular to the x axis as show in the figure below:
Loading q causes the plate to deform laterally or upward in the z direction and, the defection w of point P is assumed to be a function of x and y only; that is w=w(x, y) and the plate does not stretch in the z direction. The line a-b drawn perpendicular to the plate surface before loading remains perpendicular to the surface after loading. These conditions are consistent with the Kirchhoff assump-tions:
1. Normals remain normal. This implies that transverse shears strains γyz = 0 and γxz = 0. However γxy does not equal to zero. Right angles in the plane of the plate may not remain right angles after loading. The plate may twist in the plane.
2. Thickness changes can be neglected and normals undergo no extension. This means that εz = 0.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 337
3. Normal stress σz has no effect on in-plane strains εx and εy in the stress-strain equations and is considered negligible.
4. Membrane or in-plane forces are neglected here, and the plane stress re-sistance can be superimposed later (that is, the constant-strain triangle be-havior of Chapter 6 can be superimposed with the basic plate bending ele-ment resistance). Therefore, the in-plane deflections in the x and y direc-tions at the midsurface, t = 0, are assumed to be zero; u(x, y, 0) = 0 and v(x, y, 0) = 0.
Based on Kirchhoff assumptions, at any point P the displacement in the x direc-tion due to a small rotation α is:
wu z zx
α ∂ = − = − ∂
At the same point, the displacement in the y direction is:
wv z zy
α ∂= − = − ∂
The curvatures of the plate are then given as the rate of change of the angular displacements of the normals and defined as:
2 2 2
2 22
x y xyw w wx y x y
κ κ κ∂ ∂ ∂= − = − = −
∂ ∂ ∂ ∂
Using the definitions for in-plane strains, along with the curvature relationships, the in-plane strain/displacement equations are:
2 2 2
2 2 2x y xyw w wz z zx y x y
ε ε γ∂ ∂ ∂= − = − = −
∂ ∂ ∂ ∂
The first of the above equations is used in beam theory. The remaining two equa-tions are new to plate theory.
Stress/Strain Relationship Based on the third Kirchhoff assumption, the plane stress equations that relate in-plane stresses to in-plane strains for an isotropic material are:
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 338
( ) ( )2 21 1x x y y y x xy xyE E Gσ ε νε σ ε νε τ γν ν
= + = + =− −
The in-plane normal stresses and shear stress are shown acting on the edges of the plate shown in figure below:
Similar to the stress variation in a beam, the stresses vary linearly in the z direction from the midsurface of the plate. The transverse shear stresses τyz and τxz are also present, even though transverse shear deformation is neglected. These stresses vary quadratically through the plate thickness. The bending moments acting along the edge of the plate can be related to the stresses by:
/ 2 / 2 / 2
/ 2 / 2 / 2
t t t
x x y y xy xyt t t
M z dz M z dz M z dzσ σ τ− − −
= = =∫ ∫ ∫
Substituting strains for stresses gives:
( ) ( )/ 2 / 2 / 2
2 2/ 2 / 2 / 21 1
t t t
x x y y y x xy xyt t t
E EM z dz M z dz M zG dzε νε ε νε γν ν− − −
= + = + = − − ∫ ∫ ∫ Using the strain/curvature relationships, the moment expression become:
( ) ( ) (1 )2x x y y y x xy xy
DM D M D M νκ νκ κ νκ κ−= + = + =
where D = Et3/[12(1 - ν2)] is called the bending rigidity of the plate.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 339
The maximum magnitude of the normal stress on each edge of the plate are lo-cated at the top or bottom at z = t/2.
26 x
xMt
σ =
The equilibrium equations for plate bending are important in selecting the ele-ment displacement fields. The governing differential equations are:
0yx QQ qx y
∂∂+ + =
∂ ∂
0xyxx
MM Qx y
∂∂+ − =
∂ ∂
0y xyy
M MQ
y x∂ ∂
+ − =∂ ∂
where q is the transverse distributed loading and Qx and Qy are the transverse shear line loads as shown below.
Substituting the moment/curvature expressions in the last two differential equa-tions list above, solving for Qx and Qy, and substituting the results into the first equation listed above, the governing partial differential equation for isotropic, thin-plate bending may be derived as:
4 4 4
4 2 2 42w w wD q
x x y y ∂ ∂ ∂
+ + = ∂ ∂ ∂ ∂
where the solution to the thin-plate bending is a function of the transverse dis-
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 340
placement w. If we neglect the differentiation with respect to the y direction, the above equation simplifies to the equation for a beam and the flexural rigidity D of the plate reduces to the EI of the beam when the Poisson effect is set to zero. Potential Energy of a Plate The total potential energy of a plate is given as:
( )12 x x y y xy xy
V
U dVσ ε σ ε τ γ= + +∫
The potential energy can be expressed in terms of moments and curvatures as:
( )12 x x y y xy xy
A
U M M M dAκ κ κ= + +∫
Derivation of a Plate Bending Element Stiffness Matrix and Equations Numerous finite elements for plates bending have been developed on the years, references cite 88 different elements. In this section, we will introduce the basic 12-degree-of-freedom rectangular element shown below.
The formulation will be developed consistently with the stiffness matrix and equa-tions for the bar, beam, plane stress/strain elements of previous chapters.
Step 1 - Discretize and Select Element Types Consider the 12-degree-of-freedom plate element shown in the figure above.
Each node has 3 degrees of freedom – a transverse displacement w in the z di-rection, a rotation θx about the x axis, and a rotation θy about the y axis.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 341
The nodal displacements at node i are:
i
xi
yi
wd θ
θ
=
where the rotations are related to the transverse displacements by:
x yw wy x
θ θ∂ ∂= = −∂ ∂
The negative sign on θy is due to the fact that a negative displacement w is re-quired to produce a positive rotation about the y axis.
The total element displacement matrix is:
i
j
m
n
dd
ddd
=
Step 2 - Select Displacement Functions Since the plate element has 12 degrees of freedom, we select a 12-term poly-
nomial in x and y as:
2 2 3 2 21 2 3 4 5 6 7 8 9
3 3 310 11 12
( , )w x y a a x a y a x a xy a y a x a x y a xya y a x y a xy
= + + + + + + + +
+ + +
The function given above is an incomplete quartic polynomial; however, it is complete up to the third order (first ten terms), and the choice of the two more terms from the remaining five terms of the complete quartic must be made. The choice of x3y and y3x ensure that we will have continuity in the displacement among the interelement boundaries. The terms x4 and y4 would yield discontinui-ties along the interelement boundaries. The final term x2y2 cannot be paired with any other term so it is also rejected. The displacement function approximation also satisfies the basic differential equation over the unloaded part of the plate. In addition, the function accounts for rigid-body motion and constant strain in the
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 342
plate. However, interelement slope discontinuities along common boundaries of elements are not ensured.
To observe this discontinuities in slope, evaluate the polynomial and its slopes along a side or edge. For example, consider side i-j, the function gives:
2 31 2 4 7( , )w x y a a x a x a x= + + +
22 4 72 3w a a x a x
x∂
= + +∂
2 33 5 8 11
w a y a x a x a xy
∂= + + +
∂
The displacement w is cubic while the slope /w x∂ ∂ is the same as in beam bending. Based on the beam element, recall that the four constants a1, a2, a4, and a7 can be defined by invoking the endpoint conditions of wi, wj, θi, and θi. Therefore, w and /w x∂ ∂ are completely define along this edge. The normal slope
/w y∂ ∂ is cubic in x: however; only two degrees of freedom remain for definition of this slope while four constant exist a3, a5, a8, and a11. This slope is not uniquely defined and a slope discontinuity occurs. The solution obtained form the finite element analysis using this element will not be a minimum potential energy solu-tion. However, this element has proven to give acceptable results.
The constant a1 through a12 can be determined by expressing the 12 simulta-neous equation linking the values of w and its slope at the nodes when the coor-dinates take their appropriate values.
12 2 3 2 2 3 3 3
22 2 3 2
32 2 2 3
12
10 0 1 0 2 0 2 3 30 1 0 2 0 2 2 0 2
awx y x xy y x x y xy y x y xy a
w x y x xy y x xy ay
x y x xy y x y yw
ax
∂ = ∂ − − − − − − − ∂− ∂
or in matrix form as:
[ ] P aψ =
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 343
where [P] is the 3 x 12 first matrix on the right-hand side of the above equation.
Next, evaluate the matrix at each node point
2 2 3 2 2 3 3 3
2 2 3 2
2 2 2 3
2 2 3 2 2 3 3 3
1
0 0 1 0 2 0 2 3 3
0 1 0 2 0 2 2 0 2
1
0 1 0 2 0
i i i i i i i i i i i i i i i i
i i i i i i i i i
i i i i i i i i i
j j j j j j j j j j j j j j
n n
i
xi
yi
j
yn
x y x x y y x x y x y y x y x y
x y x x y y x x y
x y x x y y x y y
x y x x y y x x y x y y x y x y
x y
w
w
θθ
θ
− − − − − − − −
− − −
=
2 2 3
1
2
3
4
122 2 0 2n n n n n n nx x y y x y y
aaaa
a− − − − −
In compact matrix form the above equations are:
[ ] d C a= Therefore, the constants a can be solved for by: [ ] 1a C d−
= Substituting the above expression into the general form of the matrix gives: [ ] [ ] [ ] 1 orP C d N dψ ψ−
= = where [N] = [P][C]-1 is the shape function matrix.
Step 3 - Define the Strain (Curvature)/Displacement and Stress (Moment)/Curvature Relationships
Recall the general form of the curvatures:
2 2 2
2 22
x y xyw w wx y x y
κ κ κ∂ ∂ ∂= − = − = −
∂ ∂ ∂ ∂
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 344
The curvature matrix can be written as:
4 7 8 11
6 9 10 122 2
5 8 9 11 12
2 6 2 62 2 6 6
2 4 4 6 6
x
y
xy
a a x a y a xya a x a y a xy
a a x a y a x a y
κκκ
− − − − = − − − − − − − − −
or in matrix form as: [ ] Q aκ =
where [Q] is the coefficient matrix multiplied by the a’s in the curvature matrix equations.
1
2
32 2
12
0 0 0 2 0 0 6 2 0 0 6 00 0 0 0 0 2 0 0 2 6 0 60 0 0 0 2 0 0 4 4 0 6 6
ax y xy a
Q x y xy ax y x y
a
− − − − = − − − − − − − − −
Therefore: [ ] [ ] [ ] [ ] 1 orQ a Q C d B dκ κ κ−
= ⇒ = =
where [ ] [ ] [ ] 1B Q C −
= The moment/curvature matrix for a plate is given by:
[ ] [ ] x x
y y
xy xy
MM M D D B d
M
κκκ
= = =
where the [D] matrix for isotropic materials is:
( ) ( )
3
2
1 0[ ] 1 0
12 1 0 0 0.5 1
EtDν
νν
ν
= − −
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 345
Step 4 - Derive the Element Stiffness Matrix and Equations The stiffness matrix is given by the usually form of the stiffness matrix as:
[ ] [ ] [ ][ ]Tk B D B dxdy= ∫∫
The stiffness matrix for the four-node rectangular element is of a 12 x 12.
The surface force due to distributed loading q acting per unit area in the z di-rection is:
[ ] [ ]Ts sF N q dxdy= ∫∫
for a uniform load q acting over the surface of an element of dimensions 2b x 2c the forces and moments at node i are:
3
3
wi
xi
yi
fqcbf c
f bθ
θ
= −
with similar expression at nodes j, m, and n.
The element equations are given by:
11 12 1,12
21 22 2,12
31 32 3,12
41 42 4,12
12,1 12,2 12,12
wi i
xi xi
yi yi
w j j
yn yn
k k k
k k k
k k k
k k k
k k k
f wfff w
f
θ
θ
θ
θθ
θ
=
The remaining steps of assembling the global equations, applying boundary con-ditions, and solving the equations for nodal displacements and slopes follow the standard procedures introduced in previous chapters.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 346
The figure below shows a number of plate element formulations results for a square plate simply supported all around and subjected to a concentrated vertical load applied at the center of the plate. The results show the upper and lower bound solutions behavior and demonstrate convergence of solution for various plate elements. Included in these results is the 12-term polynomial plate element introduced in this chapter.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 347
The figure below shows comparisons of triangular plate formulations for the same centrally loaded simply supported plate. From both figures, we can observe a number of different formulations with results that converge for above and be-low. Some of these elements produce better results than others.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 348
The figure below shows results for some selected Mindlin plate theory elements. Mindlin plate elements account for bending deformations and for transverse shear deformation.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 349
Computer Solution for a Plate Bending Problem
Consider the clamped plate show below subjected to a 100 lb load applied at the center (let E = 30 x 106 psi and ν = 0.3).
The exact solution for the displacement at the center of the plate is w = 0.0056PL2/D. Substituting the values for the variables gives a numerical value of w = 0.0815 in. The table below shows the results of modeling this plate structure using SAP2000 (the educational version allows only 100 nodes) compares to the exact solution.
Number of square elements
Displacement at the center (in) % error
4 0.09100 11.6
16 0.09334 14.5
36 0.08819 8.2
64 0.08584 5.3
256 0.08300 1.8
1,024 0.08209 0.7
4,096 0.08182 0.3 Exact Solution 0.08154 --
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 350
The figures below show non-node-averaged contour plot for the normal stress σx and σy.
The next set of plots shows the non-node-averaged moments Mx and My.
Both sets of plots seem to indicate good interelement continuity between normal stresses in the x and y direction and bending moments about the x and y axes.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 351
The next set of plots shows the shear stress τxy and the node-average shear stress τxy.
The next set of plots shows the twisting moment Mxy and the node-average twist-ing moment Mxy.
Both sets of plots indicate interelement discontinuities for shear stress and twist-ing moment. However, if the node-average plots are viewed, the discontinuities are smoothed out and not visible.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 352
Problems
21. Do problems 12.1 and 12.6 on pages 455 - 457 in your textbook “A First Course in the Finite Element Method” by D. Logan using SAP2000.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 353
Structural Dynamics
Introduction
This chapter provides an elementary introduction to time-dependent problems. We will introduce the basic concepts using the single-degree-of-freedom spring-mass system. We will include discussion of the stress analysis of the one-dimensional bar, beam, truss, and plane frame.
We will provide the basic equations necessary for structural dynamic analysis and develop both the lumped- and the consistent-mass matrices involved in the analyses of the bar, beam, truss, and plane frame. We will describe the assembly of the global mass matrix for truss and plane frame analysis and then present numerical integration methods for handling the time derivative.
We will provide longhand solutions for the determination of the natural fre-quencies for bars and beams, and then illustrate the time-step integration proc-ess involved with the stress analysis of a bar subjected to a time dependent forc-ing function.
Dynamics of a Spring-Mass System
In this section we will discuss the motion of a single-degree-of-freedom spring-mass system as a introduction to the dynamic behavior of bars, truss, frames. Consider the single-degree-of-freedom spring-mass system subjected to a time-dependent force F(t) as shown in the figure below. The term k is the stiff-ness of the spring and m is the mass of the system.
The free-body diagram of the mass is shown below. The spring force T = kx and the applied force F(t) act on the mass, and the mass-times-acceleration term is
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 354
shown separately.
Applying Newton’s second law of motion, f = ma, to the mass, we obtain the equation of motion in the x direction:
( )F t kx mx− = where a dot over a variable indicates differentiation with respect to time; (·) () /d dt= . The standard form of the equation is:
( )mx kx F t+ = The above equation is a second-order linear differential equation whose solution for the displacement consists of a homogeneous solution and a particular solu-tion. The homogeneous solution is the solution obtained when the right-hand-side is set equal to zero. A number of useful concepts regarding vibrations are available when considering the free vibration of a mass; that is when F(t) = 0.
Let’s define the following term:
2 km
ω =
The equation of motion becomes:
2 0x xω+ = where ω is called the natural circular frequency of the free vibration of the mass (radians per second). Note that the natural frequency depends on the spring stiffness k and the mass m of the body. The motion describe by the homogeneous equation of motion is called simple harmonic motion. A typical displacement/time curve is shown below.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 355
where xm denotes the maximum displacement (or amplitude of the vibration). The time interval required for the mass to complete one full cycle of motion is called the period of the vibration τ (in seconds) and is defined as:
2πτω
=
The frequency in hertz (Hz = 1/s) is f = 1/τ = ω /(2π).
Direct Derivation of the Bar Element
Let’s derive the finite element equations for a time-dependent (dynamic) stress analysis of a one-dimensional bar.
Step 1 - Select Element Type We will consider the linear bar element shown below.
where the bar is of length L, cross-sectional area A, and mass density ρ (with typical units of lb-s2/in4), with nodes 1 and 2 subjected to external time-dependent loads, ˆ ( )e
xf t .
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 356
Step 2 - Select a Displacement Function A linear displacement function is assumed in the x direction.
1 2ˆ ˆu a a x= + The number of coefficients in the displacement function, ai, is equal to the total number of degrees of freedom associated with the element. We can express the displacement function in terms of the shape functions:
[ ] 11 2
2
ˆˆ
ˆx
x
du N N
d
= =
where N1 and N2 are the interpolation functions gives as:
1 2
ˆ ˆ1 x xN N
L L= − =
Step 3 - Define the Strain/Displacement and Stress/Strain Relationships The stress-displacement relationship is:
ˆ ˆ[ ]xdu B ddx
ε = =
where
1
2
ˆ1 1 ˆ[ ]ˆ
x
x
dB d
L L d
= − =
The stress-strain relationship is given as:
ˆ[ ] [ ][ ]x xD D B dσ ε= =
Step 4 - Derive the Element Stiffness Matrix and Equations The bar element is typically not in equilibrium under a time-dependent force;
hence, f1x ≠ f2x. We must apply Newton’s second law of motion, f = ma, to each node. Write the law of motion as the external force fxe minus the internal force equal to the nodal mass times acceleration. Therefore:
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 357
2 21 2
1 1 1 2 2 22 2
ˆ ˆˆ ˆ ˆ ˆe ex xx x x x
d df f m f f mt t
∂ ∂= + = +
∂ ∂
where m1 and m2 are obtained by lumping the total mass of the bar equally at the two nodes such that:
1 22 2AL ALm mρ ρ
= =
In matrix form, the above equations are:
212
1 1 1
222 2 2
2
ˆˆ ˆ 0ˆ ˆ ˆ0
xe
x x
ex x x
df f m t
mf f dt
∂ ∂= +
∂ ∂
If we replace f with [ ] k d we get:
ˆ ˆ ˆ ˆˆ( )ef t k d m d = +
where the elemental stiffness matrix is:
2
2
ˆ1 1ˆ ˆ1 1
dAEk dL t
∂− = = ∂−
and the lumped-mass matrix is:
1 0ˆ2 0 1ALm ρ
=
Let’s derive the consistent-mass matrix for a bar element. The typical
method for deriving the consistent-mass matrix is the principle of virtual work; however, an even simpler approach is to use D’Alembert’s principle. The effec-tive body force is:
ˆeX uρ= −
The nodal forces associated with Xe are found by using the following:
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 358
[ ] Tb
V
f N X dV= ∫
Substituting Xe for X gives:
ˆ[ ]TbV
f N u dVρ= −∫
The second derivative of the u with respect to time is:
ˆ ˆˆ ˆ[ ] [ ]u N d u N d= =
where u and u are the nodal velocities and accelerations, respectively.
[ ] [ ] ˆ ˆˆTb
V
f N N d dV m dρ= − = − ∫
where
[ ] [ ]ˆ T
V
m N N dVρ= ∫
The mass matrix is called the consistent mass matrix because it is derived using the same shape functions use to obtain the stiffness matrix. Substituting the shape functions in the above mass matrix equations gives:
ˆ1 ˆ ˆˆ 1
ˆV
xx xLm dVL Lx
L
ρ
− = −
∫
or
0
ˆ1 ˆ ˆˆ ˆ1
ˆ
Lx
x xLm A dxL Lx
L
ρ
− = −
∫
or
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 359
2
20
ˆ ˆ ˆ1 1
ˆˆ ˆ ˆ
1
L
x x xL L L
m A dxx x xL L L
ρ
− − = −
∫
Evaluating the above integral gives:
2 1ˆ6 1 2ALm ρ
=
Step 5 - Assemble the Element Equations and Introduce Boundary Conditions
The global stiffness matrix and the global force vector are assembled using the nodal force equilibrium equations, and force/deformation and compatibility equa-tions.
[ ] [ ] ( ) F t K d M d= +
where
[ ] [ ] ( ) ( ) ( )
1 1 1
N N Ne e e
e e eK k M m F f
= = =
= = = ∑ ∑ ∑
Numerical Integration in Time
We now introduce procedures for the discretization of the equations of motion with respect to time. These procedures will allow the nodal displacements to be determined at different time increments for a given dynamic system. The general method used is called direct integration. There are two classifications of direct integration: explicit and implicit. We will formulate the equations for two direct in-tegration methods. The first, and simplest, is an explicit method known as the central difference method. The second, more complicated but more versatile than the central difference method, is an implicit method known as the New-mark-Beta (or Newmark’s) method. The versatility of Newmark’s method is evi-denced by its adaptation in many commercially available computer programs.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 360
Central Difference Method The central difference method is based on finite difference expressions for the
derivatives in the equation of motion. For example, consider the velocity and the acceleration at time t:
1 1 1 1
2( ) 2( )i i i i
i id d d dd d
t t+ − + −− −
= =∆ ∆
where the subscripts indicate the time step for a given time increment of ∆t. The acceleration can be expressed in terms of the displacements (using a Taylor se-ries expansion) as:
1 12
2( )
i i ii
d d ddt
+ −− +=
∆
We generally want to evaluate the nodal displacements; therefore, we rewrite the above equation as:
21 12 ( )i i i id d d d t+ −= − + ∆
The acceleration can be expressed as:
( )1i i id d−= −M F K
To develop an expression of di+1, first multiply the nodal displacement equation by M and substitute the above equation for id into this equation.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 361
( )( )21 12i i i i id d d d t+ −= − + − ∆M M M F K
Combining terms in the above equations gives:
( ) ( )2 21 12i i i id t t d d+ −
= ∆ + − ∆ − M F M K M
To start the computation to determine 1 1 1, , andi i id d d+ + + we need the displacement at time step i-1. Using the central difference equations for the velocity and accel-eration and solving for di-1
2
1( )( )
2i i i itd d t d d−
∆= − ∆ +
Procedure for solution: 1. Given: d0, 0d , and Fi(t)
2. If the acceleration is not given, solve for 0d at t = 0.
( )10 0 0d d−= −M F K
3. Solve for d-1 at t = -∆t
2
1 0 0 0( )( )
2td d t d d−
∆= − ∆ +
4. Solve for d1 at t = ∆t using the value of d-1 from Step 3
( ) ( ) 2 211 0 0 12d t t d d−
− = ∆ + − ∆ − M F M K M
5. With d0 given and d1 determined in Step 4 solve for d2
( ) ( ) 2 212 1 1 02d t t d d− = ∆ + − ∆ − M F M K M
6. Solve for 1d
( )11 1 1d d−= −M F K
7. Solve for 1d using the central difference equation
2 01 2( )
d ddt
−=
∆
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 362
8. Repeat Steps 5, 6, and 7 to obtain the displacement, acceleration, and velocity for other time steps.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 363
Example Problem
Determine the displacement, acceleration, and velocity at 0.05 second time in-tervals for up to 0.2 seconds for the one-dimensional spring-mass system shown in the figure below.
The time-dependent forcing function is given as:
Consider the above spring-mass system as a single degree of freedom problem represented by the displacement d.
Procedure for solution: 1. At time t = 0:
0 00 0d d= = 2. The initial acceleration at t = 0:
( )120 0 0
2,000 100(0) 62.8331.83
ind d s− −
= − = =M F K
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 364
3. Solve for d-1 at t = -∆t
2
1 0 0 0( )( )
2td d t d d−
∆= − ∆ +
2
1(0.05)0 (0.05)0 (62.83) 0.0785
2d in− = − + =
4. Solve for d1 at t = ∆t (0.05 seconds) using the value of d-1 from Step 3:
( ) ( ) 2 211 0 0 12d t t d d−
− = ∆ + − ∆ − M F M K M
( ) ( ) ( ) ( ) ( )( ) 2 21
1 0.05 2,000 2 31.83 0.05 100 0 31.83 0.078531.82
0.0785
d
in
= + − −
=
5. Solve for d2 at t = 0.10 seconds:
( ) ( ) 2 212 1 1 02d t t d d− = ∆ + − ∆ − M F M K M
( ) ( ) ( ) ( ) ( ) ( )( ) 2 22
1 0.05 1,500 2 31.83 0.05 100 0.0785 31.83 031.82
0.274
d
in
= + − −
=
6. Solve for the acceleration 1d at time t = 0.05:
( ) ( )121 1 1
1 1,500 100 0.0785 46.8831.83
ind d s− = − = − = M F K
7. Solve for 1d using the central difference equation
( )2 0
10.274 0 2.74
2( ) 2 0.05d d ind st
− −= = =
∆
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 365
8. Repeat Steps 5, 6, and 7 to obtain the displacement, acceleration, and velocity for the next time step. Repeating Step 5:
( ) ( ) 2 213 2 2 12d t t d d− = ∆ + − ∆ − M F M K M
( ) ( ) ( ) ( ) ( ) ( )( ) 2 23
1 0.05 1,000 2 31.83 0.05 100 0.274 31.83 0.078531.82
0.546
d
in
= + − −
=
Repeating Step 6:
( ) ( )122 2 2
1 1,000 100 0.274 30.5631.83
ind d s− = − = − = M F K
Repeating Step 7:
( )3 1
20.546 0.0785 4.68
2( ) 2 0.05d d ind st
− −= = =
∆
The following table summarizes the results for the remaining time steps as com-pared with the exact solution.
t (s) F(t) (lb) id (in/s2) id (in/s) id (in) id (exact)
0.00 2,000 62.83 0.00 0.000 0.0000 0.05 1,500 46.88 2.74 0.0785 0.0718 0.10 1,000 30.56 4.68 0.274 0.2603 0.15 500 13.99 5.79 0.546 0.5252 0.20 0 -2.68 6.07 0.854 0.8250 0.25 0 -3.63 5.91 1.154 1.132
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 366
Newmark’s Method Newmark’s equations are given as:
1 1( ) (1 )i i i id d t d dγ γ+ + = + ∆ − +
( )2 11 12( ) ( )i i i i id d t d t d dβ β+ +
= + ∆ + ∆ − +
where β and γ are parameters. The parameter β is typically between 0 and ¼, and γ is often taken to be ½. For example, if β = 0 and γ = ½ the above equation reduce to the central difference method. To find di+1 first multiply the above equation by the mass matrix M and substi-tute into this the expression for acceleration. Recall the acceleration is:
( )10 0 0d d−= −M F K
The expression Mdi+1 is:
( ) [ ]2 211 1 12( ) ( ) ( )i i i i i id d t d t d t dβ β+ + += + ∆ + ∆ − + ∆ −M M M M F K
Combining terms gives:
( ) ( )2 2 2 11 1 2( ) ( ) ( ) ( )i i i i it d t d t d t dβ β β+ ++ ∆ = ∆ + + ∆ + ∆ −M K F M M M
Dividing the above equation by β (∆t)2 gives:
1 1' 'i id + +=K F where
21'
( )tβ= +
∆K K M
( ) 211 1 22' ( ) ( )
( )i i i i id t d t dt
ββ+ +
= + + ∆ + − ∆ ∆MF F
The advantages of using Newmark’s method over the central difference method are that Newmark’s method can be made unconditionally stable (if β = ¼ and γ = ½) and that larger time steps can be used with better results.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 367
Procedure for solution of Newmark’s Method:
1. Given: d0, 0d , and Fi(t)
2. If the acceleration is not given, solve for 0d at t = 0.
( )10 0 0d d−= −M F K
3. Solve the displacement d1 at time t = ∆t
1 1' 'd =K F
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 368
4. Solve for 1d (original Newmark equation for 1id + rewritten for 1id + )
( )2 11 1 0 0 022
1 ( ) ( )( )
d d d t d t dt
ββ
= − − ∆ − ∆ − ∆
5. Solve for 1d
1 0 0 1( ) (1 )d d t d dγ γ = + ∆ − +
6. Repeat Steps 3, 4, and 5 to obtain the displacement, acceleration, and velocity for the next time step.
Example Problem
Determine the displacement, acceleration, and velocity at 0.1 second time in-tervals for up to 0.5 seconds for the one-dimensional spring-mass system shown in the figure below.
The time-dependent forcing function is given as:
Consider the above spring-mass system as a single degree of freedom problem represented by the displacement d. Use Newmark’s method with β = 1/6 and γ = ½.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 369
Procedure for solution: 1. At time t = 0: 0 00 0d d= =
2. If the acceleration is not given, solve for 0d at t = 0:
( )1 20 0 0
100 70(0) 56.5 /1.77
d d in s− −= − = =M F K
3. Solve the displacement d1 at time t = 0.1 seconds:
1 1' 'd =K F
2 216
1 1' 70 (1.77) 1,132 /( ) (0.1)
lb intβ
= + = + =∆
K K M
( ) 211 1 0 0 022' ( ) ( )
( )d t d t d
tβ
β = + + ∆ + − ∆ ∆
MF F
( ) ( )21 11 2 621
6
1.77' 80 0 (0.1)0 (0.1) 56.5 280(0.1)
lb = + + + − = F
11
' 280 0.248' 1,132
d in= = =FK
4. Solve for 1d at time t = 0.1 seconds:
( )2 11 1 0 0 022
1 ( ) ( )( )
d d d t d t dt
ββ
= − − ∆ − ∆ − ∆
( )2 1 121 2 621
6
1 0.248 0 (0.1)0 (0.1) 56.5 35.4(0.1)
ind s = − − − − =
5. Solve for 1d
1 0 0 1( ) (1 )d d t d dγ γ = + ∆ − +
( )1 11 2 20 (0.1) (1 )56.5 35.4 4.59 ind s = + − + =
6. Repeat Steps 3, 4, and 5 to obtain the displacement, acceleration, and velocity for the next time step (t = 0.2 s).
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 370
Repeating Step 3:
( ) 212 2 1 1 122' ( ) ( )
( )d t d t d
tβ
β = + + ∆ + − ∆ ∆
MF F
( ) ( )21 12 2 621
6
1.77' 60 0.248 (0.1)4.59 (0.1) 35.4 934(0.1)
lb = + + + − = F
11
' 934 0.825' 1,132
d in= = =FK
Repeating Step 4:
( )2 12 2 1 1 122
1 ( ) ( )( )
d d d t d t dt
ββ
= − − ∆ − ∆ − ∆
( )2 1 122 2 621
6
1 0.825 0.248 (0.1)4.59 (0.1) 35.4 1.27(0.1)
ind s = − − − − =
Repeating Step 5:
2 1 1 2( ) (1 )d d t d dγ γ = + ∆ − +
( )1 12 2 24.59 (0.1) (1 )35.4 1.27 6.42 ind s = + − + =
The following table summarizes the results for the time steps through t = 0.5 sec-onds.
t (s) F(t) lb id (in/s2) id (in/s) di (in)
0 100 56.6 0 0 0.1 80 35.4 4.59 0.248 0.2 60 1.27 6.42 0.825 0.3 48.6 -26.2 5.17 1.36 0.4 45.7 -42.2 1.75 1.72 0.5 42.9 -42.2 -2.45 1.68
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 371
Natural Frequencies of a One-Dimensional Bar
Before solving the structural stress dynamic analysis problem, let’s consider how to determine the natural frequencies of continuous elements. Natural fre-quencies are necessary in vibration analysis and important when choosing a proper time step for a structural dynamics analysis.
Natural frequencies are obtained by solving the following equation:
0d d+ =M K The standard solution for d is given as ( ) ' i td t d e ω= where 'd is the part of the nodal displacement matrix called natural modes that is assumed to independent of time, i is the standard imaginary number, and ω is a natural frequency. Differentiating the above equation twice with respect to time gives:
( )2' i td d e ωω= −
Substituting the above expressions for d and d into the equation of motion gives:
2 ' ' 0i t i td e d eω ωω− + =M K Combining terms gives:
( )2 ' 0i te dω ω− =K M
Since eiωt is not zero, then:
( )2 ' 0dω− =K M
The above equations are a set of linear homogeneous equations in terms of dis-placement mode 'd . There exist a non-trivial solution if and only if the determi-nant of the coefficient matrix of 'd is zero.
2 0ω− =K M
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 372
Example Problem
Determine the first two natural frequencies for the bar shown in the figure be-low. Assume the bar has a length 2L, modulus of elasticity E, mass density ρ, and cross-sectional area A.
Let’s discretize the bar into two elements each of length L as shown below. We need to develop the stiffness matrix and the mass matrix (either the lumped- mass of the consistent-mass matrix). In general, the consistent-mass matrix has resulted in solutions that compare more closely to available analytical and ex-perimental results than those found using the lumped-mass matrix. However, when performing a long hand solution, the consistent-mass matrix is more diffi-cult and tedious to compute; therefore, we will use the lumped-mass matrix.
The elemental stiffness matrices are:
(1) (2)
1 2 2 3
1 1 1 1ˆ ˆ1 1 1 1
AE AEk kL L
− − = = − −
The global stiffness matrix is:
[ ]1 1 01 2 1
0 1 1
AEKL
− = − − −
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 373
The lumped-mass matrices are:
(1) (2)
1 2 2 3
1 0 1 0ˆ ˆ2 20 1 0 1AL ALm mρ ρ
= =
The global lumped-mass matrix is:
[ ]1 0 00 2 0
20 0 1
ALM ρ =
Substituting the above stiffness and lumped-mass matrices into the natural fre-quency equation
( )2 ' 0dω− =K M and applying the boundary condition d1x = 0 (or 1' 0d = ) gives:
22
3
'2 1 2 0 0'21 1 0 1 0
dAE ALdL
ρω −
− = −
Set the determinant of the coefficient matrix equal to zero as:
2 1 2 00
21 1 0 1AE ALL
ρλ−
− = −
where λ = ω2. Dividing the above equation by ρAL and letting 2E
Lµρ
= gives:
20
2
µ λ µλµ µ
− −=
− −
Evaluating the determinant of the above equations gives:
2 2λ µ µ= ± or
1 20.60 3.41λ µ λ µ= =
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 374
For comparison, the exact solution gives λ = 0.616µ, whereas the consistent-mass approach yields λ = 0.648 µ. Therefore, for bar elements, the lumped-mass approach can yield results as good as, or even better than, the results from the consistent-mass approach. However, the consistent-mass approach can be mathematically proven to yield an upper bound on the frequencies, whereas the lumped-mass approach has no mathematical proof of boundedness. The first and second natural frequencies are given as:
1 1 2 20.77 1.85ω λ µ ω λ µ= = = = The term µ may be computed as:
66 2
2 230 10 4.12 10
(0.00073)(100)E sL
µρ
−×= = = ×
Therefore, first and second natural frequencies are:
3 31 21.56 10 / 3.76 10 /rad s rad sω ω= × = ×
In general, an n-degree-of-freedom discrete system has n natural modes and frequencies. A continuous system actually has an infinite number of natural modes and frequencies. The lowest modes and frequencies are approximated most often; the higher frequencies are damped out more rapidly and are usually less important. Substituting λ1 into the following equation
22
3
'2 1 2 0 0'21 1 0 1 0
dAE ALdL
ρω −
− = −
gives:
(1) (1) (1) (1)2 3 2 31.4 ' ' 0 ' 0.7 ' 0d d d dµ µ µ µ− = − + =
where the superscripts indicate the natural frequency. It is customary to specify the value of one of the natural modes 'd for a given µ i or ωi and solve for the remaining values. For example, if (1)
3' 1d = than the solution for (1)2' 0.7d = . Simi-
larly, if we substitute λ2 and let (2)3' 1d = the solution of the above equations gives
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 375
(2)2' 0.7d = − . The modal response for the first and second natural frequencies are
shown in the figure below.
The first mode means that the bar is completely in tension or compression, de-pending on the excitation direction. The second mode means that bar is in com-pression and tension or in tension and compression.
Time-Dependent One-Dimensional Bar Example
Consider the one-dimensional bar system shown in the figure below.
Assume the boundary condition d1x = 0 and the initial conditions d0 = 0 and 0d = 0. Let ρ = 0.00073 lb-s2/in.4, A = a in.2, E = 30 x 106 psi, and L = 100 in. The bar will be discretized into two elements as shown below.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 376
The elemental stiffness matrices are:
(1) (2)
1 2 2 3
1 1 1 1ˆ ˆ1 1 1 1
AE AEk kL L
− − = = − −
The global stiffness matrix is:
[ ]1 1 01 2 1
0 1 1
AEKL
− = − − −
The lumped-mass matrices are:
(1) (2)
1 2 2 3
1 0 1 0ˆ ˆ2 20 1 0 1AL ALm mρ ρ
= =
The global lumped-mass matrix is:
[ ]1 0 00 2 0
20 0 1
ALM ρ =
Substitute the global stiffness and mass matrices into the global dynamic equa-tions gives:
1 1 1
2 2
3 3 3
1 1 0 1 0 01 2 1 0 2 0 0
20 1 1 0 0 1 ( )
x x
x x
x x
d d RAE ALd dL
d d F t
ρ − − − + =
−
where R1 denotes the unknown reaction at node 1.
For this example, we will used the central difference method, because it is easier to apply, for the numerical time integration. It has been mathematically shown that the time step ∆t must be less than or equal to two divided by the highest natural frequency.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 377
2
max
tω
∆ ≤
For practical results, we should use a time step defined by:
3 24 max
tω
∆ ≤
An alternative guide (used only for a bar) for choosing the approximate time step is:
x
Ltc
∆ =
where L is the element length, and xx
Ec ρ= is called the longitudinal wave
velocity. Evaluating the time step estimates gives:
33
3 2 1.5 0.40 104 3.76 10max
t sω
− ∆ = = = × ×
3
6
100 0.48 1030 10
0.00073x
Lt sc
−∆ = = = ××
Guided by these estimates for time step, we will select ∆t = 0.25 x 10-3 s.
Procedure for solution: 1. Given: d1x = 0 (fixed end), all nodal displacements, velocities are zero at time t = 0, d0 = 0 and 0d = 0, also 1xd = 0 at all times.
2. Solve for 0d at t = 0.
( )10 0 0d d−= −M F K
122
03 0
0 0 2 1 020 1 1,000 1 1 0
x
x t
d AEdAL Ld ρ
=
− = = − −
Applying the boundary conditions d1x = 0 and 1xd = 0 and simplifying gives:
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 378
220
3 0
0 020001 27,400
x
x
d ind sALd ρ = = =
3. Solve for d-1 at t = -∆t
2
1 0 0 0( )( )
2td d t d d−
∆= − ∆ +
Applying the initial conditions for d0 and 0d and 1xd from Step 2 gives:
3 22 3
33 1
0 0(0.25 10 )0 (0.25 10 )(0)2 27,400 0.856 10
x
x
din
d
−−
−−
×= − × + = ×
4. Solve for d1 at t = ∆t using the value of d-1 from Step 3
( ) ( ) 2 211 0 0 12d t t d d−
− = ∆ + − ∆ − M F M K M
( )1 222 3
3 1
0 0 2 02 2(0.073)0.25 100.073 20 1 1,000 0 1
x
x
dd
− = × +
( ) ( )23 43
2 1 0 2 0 00.0730.25 10 30 1021 1 0 0 1 0.856 10
−−
− − × × − − ×
Simplifying the above equation gives:
122
3 33 1
0 0 020.073 0 1 0.0625 10 0.0312 10
x
x
dd − −
= − × ×
The nodal displacements at t = 0.25 x 10-3 are:
23
3 1
00.858 10
x
x
din
d −
= ×
5. With d0 given and d1 determined in Step 4 solve for d2
( ) ( ) 2 212 1 1 02d t t d d− = ∆ + − ∆ − M F M K M
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 379
( )1 222 3
3 2
0 0 2 02 2(0.073)0.25 100.073 20 1 1000 0 1
x
x
dd
− = × +
( ) ( )23 43
2 1 0 2 0 00.0730.25 10 30 1021 1 0.858 10 0 1 0
−−
− − × × − − ×
Simplifying the above equation gives:
−
− −
× = − × ×
3122
3 33 2
0 0 0.0161 1020.073 0 1 0.0625 10 0.0466 10
x
x
dd
The nodal displacements at t = 0.5 x 10-3 are:
32
33 2
0.221 102.99 10
x
x
din
d
−
−
×=
×
6. Solve for 1d
( )11 1 1d d−= −M F K
12 42
33 1
0 0 2 1 02 (30 10 )0.073 0 1 1000 1 1 0.858 10
x
x
dd −
− = − × − ×
Simplifying the above equation gives:
22
3 1
3,52620,345
x
x
d insd
=
7. Solve for 1d using the central difference equation
2 01 2( )
d ddt
−=
∆
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 380
( )
3
32
1 33 1
00.221 1002.99 10 0.442
5.982 0.25 10x
x
d ind sd
−
−
−
× −
× = ⇒ = ×
8. Repeat Steps 5, 6, and 7 to obtain the displacement, acceleration, and velocity for other time steps. Repeating Step 5:
( ) ( ) − = ∆ + − ∆ − 2 21
3 2 2 12d t t d dM F M K M
( )1 222 3
3 3
0 0 2 02 2(0.073)0.25 100.073 20 1 1000 0 1
x
x
dd
− = × +
( ) ( )−
−−−
− × − × × − − ××
323 433
2 1 2 0 00.221 10 0.0730.25 10 30 1021 1 0 1 0.858 102.99 10
Simplifying the above equation gives:
3122
3 33 3
0 0 0.080 1020.073 0 1 0.0625 10 0.135 10
x
x
dd
−
− −
× = + × ×
The nodal displacements at t = 0.75 x 10-3 are:
32
33 3
1.096 105.397 10
x
x
din
d
−
−
×=
×
Repeating Step 6:
( )12 2 2d d−= −M F K
312 42
33 2
0 0 2 1 0.221 102 (30 10 )0.073 0 1 1000 1 1 2.99 10
x
x
dd
−
−
− × = − × − ×
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 381
Simplifying the above equation gives: =
22
3 2
10,5004,600
x
x
d insd
Repeating Step 7:
−=
∆3 1
2 2( )d dd
t
( )
3
332
2 33 2
01.096 100.858 105.397 10 2.192
9.0782 0.25 10x
x
d ind sd
−
−−
−
× − × × = ⇒ = ×
Beam Element Mass Matrices and Natural Frequencies
We will develop the lumped- and consistent-mass matrices for time-dependent beam analysis. Consider the beam element shown in the figure below.
The basic equations of motion are:
[ ] [ ] ( ) F t K d M d= +
where the stiffness matrix is:
2 2
3
2 2
1 1 2 2
12 6 12 66 4 6 2ˆ12 6 12 66 2 6 4
d dy y
L LL L L LEIk
L L LL L L L
φ φ
− − = − − − −
and the lumped-mass matrix is:
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 382
1 1 2 2
1 0 0 00 0 0 0ˆ
2 0 0 1 00 0 0 0
d dy y
ALm
φ φ
ρ =
The mass in lumped equally into each transitional degree of freedom; however, the inertial effects associated with any possible rotational degrees of freedom is assumed to be zero. A value for these rotational degrees of freedom could be assigned by calculating the mass moment of inertia about each end node using basic dynamics as:
3
24ALI ρ
=
The consistent-mass matrix can be obtained by applying
[ ] [ ]ˆ T
V
m N N dVρ= ∫
[ ]1
21 2 3 4
30
4
ˆ ˆL
A
NN
m N N N N dA dxNN
ρ
=
∫ ∫
where
( ) ( )3 2 3 3 2 2 31 23 3
1 1ˆ ˆ ˆ ˆ2 3 2N x x L L N x L x L xLL L
= − + = − +
( ) ( )3 2 3 2 23 43 3
1 1ˆ ˆ ˆ ˆ2 3N x x L N x L x LL L
= − + = −
Substituting the shape functions into the above mass expression and integrating gives:
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 383
2 2
2 2
156 22 54 1322 4 13 3ˆ[ ]
420 54 13 156 2213 3 22 4
L LL L L LALm
L LL L L L
ρ−
− = − − − −
Example Problem
Determine the first natural frequency for the beam shown in the figure below. Assume the bar has a length 2L, modulus of elasticity E, mass density ρ, and cross-sectional area A.
Let’s discretize the beam into two elements each of length L. We will use the lumped-mass matrix. We can obtained the natural frequencies by using the fol-lowing equation.
2K M 0ω− =
The boundary conditions are d1x = d3x = 0 and φ1 = φ3 = 0. Therefore the global stiffness matrix is:
2 2
3 2
24 0K
0 8
yd
EIL L
φ
=
The global lumped-mass matrix is:
2 0M
2 0 0ALρ
=
Substituting the global stiffness and mass matrices into the global dynamic equa-tions gives:
23 2
24 0 1 00
0 8 0 0EI ALL L
ω ρ
− =
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 384
Dividing by ρAL and simplify
24
24EIAL
ωρ
=
or
24.90 EIL A
ωρ
=
The exact solution for the first natural frequency is:
25.59 EIL A
ωρ
=
Example Problem
Determine the first natural frequency for the beam shown in the figure below. Assume the bar has a length L = 30 in, modulus of elasticity E = 3 x 107 psi, mass density ρ = 0.00073 lb-s2/in, and cross-sectional area A 1 in2, moment of inertia I = 0.0833 in4, and Poisson’s ratio ν = 0.3.
Let’s discretize the beam into two elements each of length L = 15 in. We will use the lumped-mass matrix. We can obtained the natural frequencies by using the following equation.
2K M 0ω− =
The problem is similar to the previous problem. The solution for the first natural frequency is:
23.148 EI
L Aω
ρ=
The exact solution for the first natural frequency is:
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 385
23.516 EI
L Aω
ρ=
According to vibration theory for a clamped-free beam, the higher natural fre-quencies to the first natural frequency is given as:
32
1 1
6.2669 17.5475ωωω ω
= =
The figure below shows the first, second, and third mode shapes corresponding to the first three natural frequencies for the cantilever beam.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 386
The table below shows various finite element solutions compared to the exact so-lution.
ω1, (rad/s) ω2, (rad/s) Exact Solution 228 1,434
Finite Element Solution Using 2 elements 205 1,286 Using 6 elements 226 1,372
Using 10 elements 227.5 1,410 Using 30 elements 228.5 1,430 Using 60 elements 228.5 1,432
Truss and Plane Frame Analysis
The dynamics of trusses and plane frames are preformed by extending the concepts of bar and beam element. The truss element requires the same trans-formation of the mass matrix from local to global coordinates as that used for the stiffness matrix given as:
ˆT TTm m=
Truss Elements Since the motion of the element is now in two- or three-dimension, the bar ele-ment mass matrix must be reformulated to account for the axial and transverse inertial properties in the x and y directions.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 387
Considering two-dimensional motion, the axial and the transverse displacement are given as:
1
1
2
2
ˆ
ˆˆ ˆ ˆ0 01ˆˆ ˆ ˆ0 0ˆ
x
y
x
y
d
du L x xLv L x x d
d
−
= −
The shape functions for the matrix are:
[ ]ˆ ˆ0 01
ˆ ˆ0 0L x x
NL L x x
− = −
The consistent-mass matrix can be obtained by applying:
ˆ[ ] [ ] [ ]T
V
m N N dVρ= ∫
2 0 1 00 2 0 1ˆ
6 1 0 2 00 1 0 2
ALm ρ =
The lumped-mass matrix for two-dimensional motion is obtained by simply lump-ing mass at each node and remembering that mass is the same in both the x and y directions, The lumped-mass matrix is:
1 0 0 00 1 0 0ˆ
2 0 0 1 00 0 0 1
ALm ρ =
Frame Elements The plane frame element requires combining the bar and beam elements to
obtain the local mass matrix. There are six degrees of freedom associated with a plane frame element.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 388
The plane frame analysis requires first expanding and then combining the bar
and beam mass matrices to obtain the local mass matrix. The bar and beam mass matrices are expanded to a 6 x 6 and superimposed. Combining the local axis consistent-mass matrices for the bar and beam elements gives:
22
2 2
2 16 6
156 54 1322420 420 420 420
13 322 4420 420 420 420
1 26 6
54 13 156 22420 420 420 420
13 3 22 4420 420 420 420
0 0 0 00 00 0ˆ[ ]
0 0 0 00 00 0
LL
L LL L
L L
L L L L
m ALρ
−
−
−
− − −
=
The resulting lumped-mass matrix for a plane frame element is give as:
ˆ ˆ ˆ ˆˆ1 1 2 2 21
1 0 0 0 0 00 1 0 0 0 00 0 0 0 0 0ˆ[ ]
2 0 0 0 1 0 00 0 0 0 1 00 0 0 0 0 0
d d d dx y x y
ALm
φ φ
ρ
=
The global mass matrix for the plane frame element arbitrarily oriented in x-y co-ordinates is transformed by:
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 389
[ ]0 0 0
0 0 0i j m
i j m
N N NN
N N N
=
ˆT TTm m= where the transformation matrix is given as:
−
−
=
1000000000000000010000000000
CSSC
CSSC
T
Long-hand solution to the truss and frame problem are quite tedious and lengthy; therefore, we will use a computer problem to generate approximation for the mo-tion of truss and frame structures.
Plane Stress/Strain Elements The plane stress/strain constant-strain triangle consistent-mass matrix is ob-tained using the shape functions given below as:
The consistent-mass matrix can be obtained by applying: ˆ[ ] [ ] [ ]T
V
m N N dVρ= ∫
where dV = tdA The CST global consistent-mass Matrix is:
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 390
2 0 1 0 1 00 2 0 1 0 11 0 2 0 1 0
[ ]12 0 1 0 2 0 1
1 0 1 0 2 00 1 0 1 0 2
tAm ρ
=
Example Problem
Determine the motion of the frame structure shown below.
Assume the modulus of elasticity E = 3 x 107 psi. The mass densities ρ are ob-tained by dividing the total mass of each floor by the cross-sectional area and length the element. For example, consider the element 6:
( )( )( )2
266
104 30 15121
386.4 ins
psf ft ftW lb sM ing⋅= = =
22
46 2121 0.0136
(24.7 )(360 )
lb sin lb s
inin inρ
⋅⋅= =
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 391
Use Newmark’s method with β = ¼ and γ = ½. The following is the input file for WinFElt.
problem description title=îdynamic frame analysisî nodes=8 elements=9 analysis=transient
analysis parameters beta=0.25 gamma=0.5 alpha=0.0 duration=0.8 dt=0.05 nodes=[8,6,3] dofs=[Tx] mass-mode=lumped
nodes 1 x=0 y=0 constraint=fixed 2 x=360 y=0 3 x=0 y=180 constraint=free force=f1 4 x=360 5 x=0 y=300 force=f2 6 x=360 7 x=0 y=420 force=f3 8 x=360
beam elements 1 nodes=[1,3] material=wall_bottom 2 nodes=[3,5] material=wall_top 3 nodes=[5,7] 4 nodes=[7,8] material=floor_top load=top_wt 5 nodes=[5,6] material=floor_bottom load=bottom_wt 6 nodes=[3,4] load=bottom_wt 7 nodes=[8,6] material=wall_top 8 nodes=[6,4] 9 nodes=[4,2] material=wall_bottom
material properties wall_bottom A=13.2 Ix=249 E=30e6 rho=0.0049 wall_top A=6.2 Ix=107 E=30e6 rho=0.0104 floor_top A=12.3 Ix=133 E=30e6 rho=0.01315 floor_bottom A=24.7 Ix=237 E=30e6 rho=0.0136
distributed loads top_wt direction=perpendicular values=(1,-62.5) (2,-62.5) bottom_wt direction=perpendicular values=(1,-130) (2,-130)
forces f1 Fx=1000*(t < 0.2 ? 25*t : 5) f2 Fx=800*(t < 0.2 ? 25*t : 5) f3 Fx=500*(t < 0.2 ? 25*t : 5)
constraints fixed Tx=c Ty=c Rz=c free Tx=u Ty=u Rz=u
end
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 392
The following is the WinFElt output
------------------------------------------------------------------ time Tx(8) Tx(6) Tx(4) ------------------------------------------------------------------ 0 0 0 0 0.05 0.0054775 0.0046834 0.0047332 0.1 0.032795 0.028894 0.026946 0.15 0.10231 0.092341 0.078059 0.2 0.23314 0.21232 0.16267 0.25 0.43808 0.39636 0.27818 0.3 0.71526 0.63623 0.41441 0.35 1.0528 0.91484 0.56253 0.4 1.4335 1.2132 0.71742 0.45 1.8341 1.5124 0.87255 0.5 2.2269 1.7954 1.0185 0.55 2.5809 2.0464 1.1461 0.6 2.8674 2.2528 1.2504 0.65 3.0641 2.4041 1.3297 0.7 3.1589 2.4924 1.3824 0.75 3.1498 2.511 1.4031 0.8 3.0433 2.4556 1.3844
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 393
Example Problem
Determine the motion of the frame structure shown below. This problem is the same as the previous example, except for the loading function F(t) and the time duration.
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 394
The following is the WinFElt output
------------------------------------------------------------------ time Tx(8) Tx(6) Tx(4) ------------------------------------------------------------------ 0 0 0 0 0.05 0.0054775 0.0046834 0.0047332 0.1 0.032795 0.028894 0.026946 0.15 0.10231 0.092341 0.078059 0.2 0.21123 0.19359 0.14374 0.25 0.32881 0.29952 0.18933 0.3 0.43722 0.38244 0.20996 0.35 0.52949 0.43491 0.22408 0.4 0.59182 0.45828 0.23644 0.45 0.61612 0.45612 0.23879 0.5 0.59851 0.42714 0.22158 0.55 0.53509 0.37319 0.1881 0.6 0.42314 0.30076 0.1489 0.65 0.27506 0.21612 0.11208 0.7 0.11427 0.11995 0.07504 0.75 -0.041956 0.012366 0.028051 0.8 -0.18758 -0.10174 -0.034383 0.85 -0.31656 -0.21548 -0.10737 0.9 -0.41638 -0.31995 -0.17587 0.95 -0.48337 -0.40479 -0.22728 1 -0.52583 -0.45861 -0.25632 1.05 -0.55228 -0.47809 -0.26688 1.1 -0.55737 -0.46626 -0.26286 1.15 -0.5321 -0.42613 -0.24354 1.2 -0.4741 -0.3581 -0.20324 1.25 -0.38481 -0.26715 -0.13927 1.3 -0.26441 -0.16487 -0.062339 1.35 -0.11615 -0.058684 0.0067415 1.4 0.046209 0.05035 0.051067 1.45 0.20774 0.16045 0.076603 1.5 0.35642 0.26056 0.10478 1.55 0.48124 0.34101 0.15095 1.6 0.56682 0.40072 0.20559 1.65 0.60441 0.44251 0.24564 1.7 0.59778 0.46276 0.25798 1.75 0.55525 0.45409 0.24836 1.8 0.48211 0.41509 0.22947 1.85 0.38454 0.3486 0.20213 1.9 0.27609 0.25766 0.15744 1.95 0.16472 0.14464 0.088264 2 0.046915 0.019038 0.0033795
CIVL 7117 Finite Elements Methods in Structural Mechanics Page 395
Problems
22. Do problems 16.5 and 16.11 on pages 611-613 in your textbook “A First Course in the Finite Element Method” by D. Logan.
23. Do problems 16.14 and 16.16 on pages 613 - 614 in your textbook “A First Course in the Finite Element Method” by D. Logan using WinFElt.