Engineering Analysis and Design

Lecture: Tuesday, Wednesday and Friday

Tutorial : Thursday
4 pm to 5 pm

Lecture-T2-6-7

Dr. Nidhi Sharma

1
 Contents
Tutorial – 2 ( 20th April )
 Example: Potential Energy
 Example: Galerkin’s Approach
Lecture 6 (22th April )
 Galerkin’s method
 Element Stiffness Matrix in Galerkin’s Method
 Assembly of the Global Stiffness Matrix and Load Vector
Lecture 7 (22th April)
 Element Stiffness Matrix in Galerkin’s Method
 Assembly of the Global Stiffness Matrix and Load Vector
 Example: Assembly of the Global Stiffness Matrix and Load Vector
 Properties of K
 The Finite Element Equations; Treatment of Boundary Conditions
 Elimination Approach: The Finite Element Equations; Treatment of Boundary
Conditions

2
 Example: Potential Energy
Example 1.1
 Figure E1.1a shows a system of springs. The total potential energy is given by

 where δ1, δ2, δ3 and δ4 are extensions of the four springs. Since δ1 = ql - q2, δ2 = q2, δ3 =
q3 – q2 and δ4 = -q3, we have

 where q1, q2, and q3 are the displacements of nodes 1, 2, and 3, respectively

31
 Example: Potential Energy
 For equilibrium of this three degrees of freedom system, we need to minimize П with
respect to q1, q2 and q3. The three equations are given by

 which are

 These equilibrium equations can be put in the form of K q = F as follows:

41
 Example: Potential Energy
 If, on the other hand, we proceed to write the equilibrium equations of the system by
considering the equilibrium of each separate node, as shown in Fig. E1.1b, we can write

 which is precisely the set of equations represented in Eq. 1.29

 We see clearly that the set of equations 1.29 is

obtained in a routine manner using the potential
energy approach, without any reference to the free-
body diagrams. This makes the potential energy
approach attractive for large and complex problems.

51
Example: Galerkin’s Method

61
Example: Galerkin’s Method

71
Example: Galerkin’s Method

81
Example: Galerkin’s Method

91
 Galerkin’s Method
 We introduce a virtual displacement field

 and associated virtual strain

 where Φ is an arbitrary or virtual displacement consistent with the boundary conditions.

Galerkin's variational form, given in Eq. 1.43, for the one-dimensional problem
considered here, is

 This equation should hold for every Φ consistent with the boundary conditions. The first
term represents the internal virtual work, while the load terms represent the external
virtual work. On the discretized region, Eq. 3.39a becomes

101
 Galerkin’s Method
 Note that ε is the strain due to the actual loads in the problem, while is a virtual
strain. Similar to the interpolation steps in Eqs. 3.7b, 3.14, and 3.16 (given in the next
slide), we express

 where represents the arbitrary nodal displacements of element e.A1so, the

global Virtual displacements at the nodes are represented by

Element Stiffness
 Consider the first term, representing internal virtual work, in Eq 3.39b. Substituting Eq.
3.40 into Eq. 3.39b, and noting that ε = Bq, we get

111
Coordinate and Shape Functions

12
 Galerkin’s Method
 In the finite element model (Section 3.2), the cross-sectional area of element Ae, denoted
by Ae, is constant. Also, B is a constant matrix. Further, dx = (le/2)dξ. Thus,

 where ke is the (symmetric) element stiffness matrix given by

 Substituting B from Eq. 3.15, we have

131
 Galerkin’s Method
Force Terms
 Consider the second term in Eq. 3.39a, representing the virtual work done by the body
force in an element. Using and dx = (le/2)dξ noting that the body force in the
element is assumed constant, we have

 where

 is called the element body force vector. Substituting for N1 = (1 - ξ)/2 and N2 = (1 + ξ)/2,
we obtain . Alternatively, is the area under the N1 curve = ½ X 2 X 1
and . Thus

141
 Galerkin’s Method
 The element traction term then reduces to

 where the element traction-force vector is given by

 At this stage, the element matrices ke, fe, and Te have been obtained. After accounting for
the element connectivity (in Fig. 3.3, for example for element 1,
 for element 2, etc.), the variational form

 can be written as

 which should hold for every Ψ consistent with the boundary conditions.
151
Assembly of the Global Stiffness Matrix and Load Vector

161
Assembly of the Global Stiffness Matrix and Load Vector

171
Assembly of the Global Stiffness Matrix and Load Vector

181
Assembly of the Global Stiffness Matrix and Load Vector

191
Assembly of the Global Stiffness Matrix and Load Vector

201
 Assembly of the Global Stiffness Matrix and Load Vector
 We noted earlier that the total potential energy written in
the form

 can be written in the form

Fig 3.2 Finite element modeling of a bar.

 by taking element connectivity into account. This step involves assembling K and F
from element stiffness and force matrices. The assembly of the structural stiffness
matrix K from element stiffness matrices ke will be discussed first.
 Referring to the finite element model in Fig. 3.2b, let us consider the strain energy in
say, element 3. We have

 or, substituting for k3,

211
 Assembly of the Global Stiffness Matrix and Load Vector
 For element 3, we have . Thus, we can write U3 as,

 From the previous equations, we see that elements of the matrix k3 occupy the third and
fourth rows and columns of the K matrix. Consequently, when adding element-strain
energies, the elements of ke are placed in the appropriate locations of the global K
matrix, based on the element connectivity; overlapping elements are simply added. We
can denote this assembly symbolically as

221
 Assembly of the Global Stiffness Matrix and Load Vector
 Similarly, the global load vector F is assembled from element-force vectors and point
loads as

 The Galerkin approach also gives us the same assembly procedure.

231
 Example: Assembly of the Global Stiffness Matrix and Load
Vector
Example 3.2
 Consider the bar as shown in Fig. E3.2.
For each element i, Ai , and li ; are the
cross-sectional area and length,
respectively. Each element i is subjected
to a traction force Ti, per unit length and a
body force f per unit volume. The units of
Ti , f, Ai , and so on are assumed to be
consistent. The Young's modulus of the
material is E. A concentrated load P2 is
applied at node 2. The structural stiffness
matrix and nodal load vector will now be
assembled.

241
 Example: Assembly of the Global Stiffness Matrix and Load
Vector
 The element stiffness matrix for each element i is obtained from Eq. 3.26 as

 The element connectivity table is the following:

 The element stiffness matrices can be "expanded" using the
connectivity table and then summed (or assembled) to obtain

 the structural stiffness

matrix as follows:.

251
 Example: Assembly of the Global Stiffness Matrix and Load
Vector
 which gives

261
 Example: Assembly of the Global Stiffness Matrix and Load
Vector
 The global load vector is assembled as

271
 Properties of K
 The dimension of the global stiffness K is (N x N), where N is the number of nodes. (As
each node has only one degree of freedom.)
 K is symmetric.
 K is a banded matrix. That is all elements outside of the band are zero. K can be
compactly represented in banded form as

281
 Properties of K
 Kbanded is of dimension (N X NBW], where NBW is the half-bandwidth.
 In many one-dimensional problems, the connectivity of element i is i, i + 1 and the
banded matrix has only two columns (NBW = 2).

 The following general formula for the half-bandwidth:

291
 Properties of K
 For example, consider a four-element model of a bar that is numbered as shown in Fig.
3.9a. Using Eq. 3.55, we have

 The numbering scheme in Fig.

3.9a is bad since K is almost
"filled up" and consequently
requires more computer storage
and computation.

 Figure 3.9b shows the optimum numbering for minimum NBW.

 The potential energy or Galerkin's approach is used, along with the boundary conditions
of the problem, to get the finite element / equilibrium equations.
 Solution of these equations gives the global displacement vector Q. The stresses and
reaction forces can then be found out.
301
 The Finite Element Equations; Treatment of Boundary
Conditions
Types of Boundary Conditions
 After using a discretization scheme to model the continuum we have obtained an
expression for the total potential energy in the body as

 where K is the structural stiffness matrix, F is the global load vector and Q is the
global displacement vector.
 K and F are assembled from element stiffness and force matrices, respectively.

 The equations of equilibrium can be obtained by minimizing, the potential energy

П = ½ QT K Q – QT F subject to boundary conditions, with respect to Q,

 The minimum potential energy theorem states that: Of all possible displacements that
satisfy the boundary conditions of a structural system, those corresponding to
equilibrium configurations make the total potential energy assume a minimum value.
311
 The Finite Element Equations; Treatment of Boundary
Conditions
 Boundary conditions are usually of the type

 That is, the displacements along dofs p1, p2, p3 …. pr are specified to be equal to a1,
a2…..ar respectively.
 That is, if there are r number of supports in the structure, with each support node given a
specified displacement. For example, in the bar in Fig. 3.2b, there is only one boundary
condition in this problem, Q1 = 0.
 It is noted here that this treatment of boundary conditions in is applicable to two-
and three-dimensional problems as well. For this reason, the term dof is used here
instead of node, as in a 2D stress problem, each node will have two degrees of freedom.
Also, a Galerkin approach involves same steps for handling boundary conditions as the
energy approach.

321
 Treatment of Boundary Conditions: Elimination Approach
 Improper specification of boundary conditions can lead to erroneous results. Also,
boundary conditions should accurately model the physical system.
Elimination Approach
 Consider the single boundary condition Q1 = a1 . According to minimum potential energy
theorem: the equilibrium equations are obtained by minimizing П with respect to Q,
subject to the boundary condition Q1 = a1. For an N dof structure, we have,

 The global stiffness matrix is of the form

 Note that K is a symmetric matrix.

331