Chapter 4

The Direct Stiffness Method_________________________________________

CHAPTER 4

THE DIRECT STIFFNESS METHOD Before going on with the general approach of the FEM, a simple example of DSM application is considered useful to understand this precursory method. The simplest structural element is the plane truss, a bar connecting two pinjoints, capable to take over only the axial force (bending moment and shear force are zero). This type of structural element is commonly used for skeletal structures, being assembled by rotation free pin-joints in rigid triangular panels (here, the rigid behavior takes a particular sense as an opposite to mechanism movement). Each component of the structure is defined by its cross section area A and length L. The stress level in each structural member is presumed to be low enough to provide linear-elastic material state and no large deformations occur. Thus, the Young modulus E and the thermal coefficient are the only material properties. It should be noted that the axially-carrying members and frictionless pins of this structure are only idealizations of a real behavior. Trusses usually have members joined to each other through the use of gusset plates, which are attached by screws or welds. While connections are not frictionless hinges (as in the model) the real members will carry some bending and shear force. However, these values are small enough to be neglected. The example, consisting in a 3 - member truss and its idealized model, is represented in figure 4.1. The main steps of DSM will be followed in order to determine the stress and displacement state of the structure, subjected to known external loads.

34

______________________Basics of the Finite Element Method Applied in Civil Engineering

2 Fx=10

(1) Fy = -20 (3)

(2)

Fig. 4.1 Truss structure with 3 members

In order to simplify the calculations, the main stiffness and geometric characteristics are chosen as follows:
EA(1) = EA( 2) = 100 2 ; L(1) = L( 2) = 10 2 EA(3) = 100 ; L(3) = 20

4.1 JOINT FORCES AND DISPLACEMENTS The geometry of the truss is referred to a Cartesian coordinate system called the global coordinate system. The main parameters of the problem, the forces and displacements at the joints are shown in figure 4.2. They can be expressed by their components and arranged, each of them, in a 6 components column vector as:
f = f x ,1

f y ,1

f x,2

f y,2

f x ,3

f y ,3

= [u1

v1

u2

v2

u3

v3 ]

(4.1)
T

35

Chapter 4

The Direct Stiffness Method_________________________________________

fy,2,v2 L(1)=10 2 EA(1)=100 2

2

fx,2,u2 L(2)=10 2 EA(2)=100 2

(2)

y
fy,1,v1

(1)

fy,3,v3
(3)

fx,1,u1 1

L(3)=20 EA(3)=100

3 fx,3,u3

Fig. 4.2 Force components and member characteristics

where fxi, fyi are the force components of joint i and ui, vi the displacement components of the same joint. The six displacement components are also called degrees of freedom (DOF). The external forces as well as the boundary conditions can act only in joints.

4.2 THE GLOBAL STIFFNESS EQUATIONS

The global stiffness matrix relates the joint displacements with the joint forces f, before specifying the support conditions.
k x1x1 k y1x1 k x 2 x1 k y 2 x1 k x 3 x1 k y 3 x1 k x1 y1 k y1 y1 k x 2 y1 k y 2 y1 k x 3 y1 k y 3 y1 k x1x 2 k y1x 2 k x2x2 k y2x2 k x3x 2 k y3x 2 k x1 y 2 k y1 y 2 k x2 y 2 k y2 y2 k x 3 y 21 k y3 y 2 k x1x3 k y1x3 k x 2 x3 k y 2 x3 k x3 x3 k y 3x3 k x1 y 3 u1 f x1 k y1 y 3 v1 f y1 k x 2 y 3 u 2 f x 2 = k y 2 y 3 v2 f y 2 k x 3 y 3 u 3 f x 3 k y3 y3 v3 f y3

(4.2)

or, in condensed form

K = f
with K the global stiffness matrix.
36

(4.3)

______________________Basics of the Finite Element Method Applied in Civil Engineering

Because the assumed behavior of the truss is linear, the equations must be linear relations. If all displacements vanish, the forces vanish too. The terms of the stiffness matrix can be interpreted by choosing a displacement vector so that all components are zero except the ith one, which is one. The ith column of the stiffness matrix becomes the f vector. For instance,
k x1 y 3 0 k 0 y1 y 3 k x2 y3 0 for = the force vector becomes f = . k y 2 y 3 1 k x3 y 3 0 k y3 y3 0

In this case, each term of K represents the force that would arise on the DOF directions when prescribing a unit vertical displacement of joint 2.
4.3 MEMBER STIFFNESS EQUATIONS IN LOCAL COORDINATES THE BREAKDOWN STEPS 4.3.1 Disconnection

The first step in applying the DSM is to disconnect (or disassemble) the structure into its components (see figure 4.3). For each member (e), (e = 1, 2, 3), a local Cartesian coordinate system x ( e ) , y ( e ) is assigned, with the x axis along the elements and, by convention, the positive direction running from joint i to joint j, i < j. The coordinate system is also called the member attached coordinate system.

37

Chapter 4

The Direct Stiffness Method_________________________________________

y2

x 1
(1)

(2)

x2

y
1

y 1

y3
(3)

x3

Fig. 4.3 Local (attached) coordinate systems

4.3.2 Localization

For the generic truss member, the local coordinate system is (x , y ) . The member has four joint force components and four displacement components. The member properties are the length L, the cross section A and the Young modulus E. For an applied axial force F, the member EA (see elongation is d, calculated using the equivalent spring stiffness k s = L figure 4.4).
fy,i,vi fx,i,ui F
i

x
(e)

fy,j,vj
j

fx,j,uj F d

ks=EA/L L

Fig. 4.4 The generic truss member in local coordinates

4.3.3 The member stiffness equations

The force and displacement components are linked by the member stiffness relationship k = f (4.4)
38

______________________Basics of the Finite Element Method Applied in Civil Engineering

Written in full
k xixi k yixi k xjxi k yjxi k xiyi k yiyi k xjyi k yjyi k xixj k yixj k xjxj k yjxj k xiyj ui f xi k yiyj vi = f yi k xjyj u j f xj k yjyj f yj v j

(4.5)

where vectors f and are called member joint forces and member joint displacements and k is the member stiffness matrix. Using the equivalence with the elastic spring, the axial force equation is
F = ks d = EA d L

(4.6)

which can be expressed in terms of joint forces and displacements as:

F = f xj f xi and d = u j ui

(4.7)

These relationships express the force equilibrium and the kinematical compatibility. Combining (4.6) and (4.7) yields
1 EA 0 L 1 0 0 1 0 ui f xi v f 0 0 0 i = yi 0 1 0 u j f xj 0 0 0 f yj v j

from which the member stiffness matrix in local coordinates k is emphasized:

39

Chapter 4

The Direct Stiffness Method_________________________________________

1 EA 0 k= L 1 0
4.4 THE ASSEMBLY PROCEDURE

0 1 0 0 0 1 0 0

0 0 0 0

(4.8)

The assembly involves two sub-steps: back transformation to the global coordinate system and merging of the equations into the global stiffness equations.
4.4.1 Coordinates transformations

This step defines the matrix relationships that connect joint displacements and forces in the global and local coordinate systems (figure 4.5).
ui = ui c + vi s u j = u jc + v j s vi = ui s + vi c

(4.9)
v j = u j s + v j c

with c = cos , s = sin

vj
y
vi ui vi
i

vj uj
j

uj

x
(e)

ui

Fig. 4.5 Oriented generic truss member

In matrix form:

40

______________________Basics of the Finite Element Method Applied in Civil Engineering

ui c s 0 0 ui v i = s c 0 0 vi u j 0 0 c s u j v j 0 0 s c v j

(4.10)

The matrix with the c and s terms is the so called displacements transformation matrix Td. The force transformation matrix Tf is used to express the forces in global coordinate system based on the forces in local coordinate system.

f xi c s 0 0 f xi f yi = s c 0 0 f yi f xj 0 0 c s f xj f yj 0 0 s c f yj

(4.11)

The force transformation matrix is the transpose of the displacement transformation matrix:
Td = TT f =T 4.4.2 Globalization

Working in the global coordinate system, it is necessary to introduce the member index e. The member stiffness equations in global coordinate system will be:
f ( e) = k ( e ) ( e)

(4.12)

( e ) = T ( e ) ( e ) ; f ( e ) = T ( e ) f ( e )

( )

(4.13)

Inserting these matrix expressions into the member stiffness relationship (4.4) and comparing it to (4.12), the member stiffness matrix in the global coordinate system (x, y) can be computed

41

Chapter 4

The Direct Stiffness Method_________________________________________

k ( e ) = T( e ) k ( e ) T( e )

( )

(4.14)

Note that for this particular member the T matrix is square and orthogonal, that is TT = T 1 . Carrying out the matrix multiplication in full, the member stiffness matrix in global coordinates yields
c2 sc c 2 sc s 2 sc s 2 sc c 2 sc c 2 sc 2 sc s2 sc s

k (e) =

E ( e) A( e) L( e)

(4.15)

Thus, the globalized member stiffness matrices are obtained as follows: 2 - member (1), = 45, c = s = 2

0.5 0.5 0.5 0.5 100 2 k (1) = 10 2 0.5 0.5 0.5 0.5 2 member (2), = -45, c = , 2

0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 2 s= 2

(4.16)

k ( 2)

0.5 0.5 0.5 0.5 0.5 0.5 100 2 0.5 0.5 = 0.5 0.5 10 2 0.5 0.5 0.5 0.5 0.5 0.5

(4.17)

member (3), = 0, c = 1 , s = 0

42

______________________Basics of the Finite Element Method Applied in Civil Engineering

k ( 3)

1 100 0 = 20 1 0

0 1 0 0 0 0 0 1 0 0 0 0

(4.18)

4.4.3 Assembly rules

The key operation of the assembly process is the placement of the contribution of each member to the global stiffness equations. The process is called merging. The merging operation can be physically interpreted as reconnecting the members in the fabrication process and is governed by two rules: 1. The compatibility of displacements the displacements of all members meeting at a joint are the same; 2. Force equilibrium the sum of forces exerted by all members meeting a joint balances the external force applied to that joint. As an example of the foregoing rules applied for joint 3,
( 2) ( 3) ( 2) ( 3) u3 = u3 , v3 = v3

(4.19)

2) 3) 1) 2) 3) f x 3 = f x(3 + f x(3 = f x(3 + f x(3 + f x(3

(4.20) f y3 =
2) f y(3

3) f y(3

) f y(1 3

2) f y(3

3) f y(3

1) ) Note that adding the terms f x(3 and f y(1 3 in the relationship 4.20 (terms due

to the first member which is not connected in node 3) has no effect on the overall value, these terms being zero.

4.4.4 Matrix augmentation and merge

The member stiffness relationships should be augmented by adding zero rows and columns in order to complete the force and displacement vectors.

43

Chapter 4

The Direct Stiffness Method_________________________________________

member (1)
1) (1) f x(1 5 5 5 0 0 u1 5 (1) (1) 5 5 5 5 0 0 v1 f y1 (1) f (1) 5 5 5 5 0 0 u2 x2 (1) = (1) 5 0 0 v2 f y 2 5 5 5 (1) f (1) 0 0 0 0 0 0 u3 x3 (1) (1) 0 0 0 0 0 0 f v3 y3

(4.21)

member (2)

0 0 0 0 0 0 member (3) 5 0 0 0 5 0

( 2) f x(12) 0 u1 ( 2) ( 2) 0 0 0 0 0 v1 f y1 ( 2) f ( 2) 0 5 5 5 5 u2 x2 = ( 2) ( 2) 0 5 5 5 5 v2 f y2 ( 2) f ( 2) 0 5 5 5 5 u3 x3 ( 2) ( 2) 0 5 5 5 5 f v 3 y3

(4.22)

( 3) f x(13) 0 0 0 5 0 u1 ( 3) ( 3 ) 0 0 0 0 0 v1 f y1 ( 3) f ( 3) 0 0 0 0 0 u2 x2 ( 3) = ( 3 ) 0 0 0 0 0 v2 f y 2 ( 3) f ( 3) 0 0 0 5 0 u3 x3 ( 3) ( 3 ) 0 0 0 0 0 v3 f y3

(4.23)

According to the first rule, the member index can be dropped in the displacement vectors (left side). The three equations can be represented in matrix form as:
k (1) = f (1) ; k ( 2) = f ( 2) ; k (3) = f (3)
44

(4.24)

______________________Basics of the Finite Element Method Applied in Civil Engineering

According to the second rule, f = f (1) + f ( 2) + f (3) = (k (1) + k ( 2) + k (3) ) = K (4.25)

Hence, to achieve the global stiffness equations, the three stiffness matrices should be added. The merging operation becomes a simple addition. Dropping the member index, the global equation system yields: 10 5 5 5 5 5 5 5 10 5 5 0 5 0 5 0 5 0 3.5 THE SOLUTION In order to solve the global equation system the known and the unknown components of and f should be separated. In the achieved form, the equation system can not be solved because the matrix K is singular (the rows and columns are linear combinations of each other). The physical interpretations stands in the fact that the rigid body motion of the structure is not suppressed (there are no constrains). To eliminate this inconveniency the boundary conditions should be applied:
v1 = u 2 = v2 = 0

5 5 5 0 0 5

u1 f x1 f v1 y1 u 2 f x 2 = 10 5 5 v2 f y 2 5 10 5 u3 f x 3 5 5 5 v3 f y3 0 0 5

(4.26)

The known applied forces are: f x1 = 0 ; f x 2 = 10 ; f y 2 = 20

The simplest way to account for support conditions is to remove those equations associated with known constrains, leading to the reduced equation system. For this example, the second, fifth and sixth rows and columns should be eliminated:

45

Chapter 4

The Direct Stiffness Method_________________________________________

10 5 5 u1 0 5 10 0 u = 10 2 v 5 0 10 20 2

(4.27)

The coefficient matrix is no longer singular and the algebraic equation system can be solved. Solving the reduced equation system yields:

u1 1 u = 0.5 2 v 2 . 5 2

(4.28)

called the reduced displacement solution. They are the primary unknowns of the problem. The reduced vector is expanded to six components including the constrained (zero) values. u1 1 v 0 1 u 2 0.5 = = v2 2.5 u 3 0 0 v3

(4.29)

3.6 THE DERIVED QUANTITIES

Quantities as internal forces and stresses or reaction forces along the constrained DOF are called derived quantities or derived unknowns. They can be recovered from the displacement solution. The internal forces and stresses are of utmost importance in the design process because they determine the necessary cross section area of the members and the joint assembling details. For this type of structure, the internal forces are only axial member forces, here denoted as p(1), p(2),and p(3) (see figure 4.6). The axial force p(e) of a member (e) can be obtained by

46

______________________Basics of the Finite Element Method Applied in Civil Engineering

extracting its displacements from the global solution. Then, the joint displacements in local coordinates are recovered by multiplying with the displacement transformation matrix T(e):
(e) = T (e) (e)
2

(4.30)

p(1)

p(2)

p(3) Fig. 4.6 Axial internal member forces

The relative displacement is than computed as d ( e ) = ui( e ) u (je) and the axial force is recovered from the equivalent spring constitutive law p (e) = The ratio ( e ) = d (e) L( e) E ( e ) A( e ) ( e ) d L( e )

is the axial strain of the member, while the axial

stress is ( e) = E ( e) ( e ) .

3.7 PRESCRIBED (FORCED) DISPLACEMENTS

The prescribed (or forced) displacements are typical for stress assessment in various civil engineering structures due to foundation settlements, but also in analysis of motion-driven machinery components. They are called nonhomogenous boundary conditions.

47

Chapter 4

The Direct Stiffness Method_________________________________________

Suppose that for the previous example we prescribe the following displacements to joints 1 and 2 (pin-joint 1 moves upwards and pin-joint 2 to the right): v1 = 0.5; u3 = 1.5; v3 = 0 Inserting the new values in the global stiffness system, yields
10 5 5 5 5 5 5 5 10 5 5 0 5 0 5 0 5 0

0 u1 0 f 0 5 0 0.5 y1 0 5 5 u 2 10 = 10 5 5 v2 20 5 10 5 1.5 f x3 5 5 5 f x3 0
5 5

(4.31)

As before, the lines of the stiffness matrix corresponding to the known degrees of freedom are removed but the columns are transferred at the right hand side of the system. The reduced stiffness matrix is the same as before, while the right hand term is now the modified node forces vector. By solving the reduced system yields
u1 0.5 10 5 5 5 5 0 0 5 5 10 0 5 5 u 2 = 10 (4.32) v 2 20 5 5 0 10 5 5 1.5 0 10 5 5 u1 0 5 0.5 + (5) 1.5 + (5) 0 5 5 10 0 u = 10 (5) 0.5 + (5) 1.5 + 5 0 = 20 2 v 5 0 10 20 ( 5 ) 0 . 5 + 5 1 . 5 + ( 5 ) 0 25 2

(4.33) and the unknown displacements are

48

______________________Basics of the Finite Element Method Applied in Civil Engineering

u1 0.5 u = 2.25 2 v 2 . 25 2

Going further with the reaction forces and internal member forces calculation, one can notice that the same results will be found. The fact is due to the static determination of the initial structure.

3.8 THERMAL EFFECTS

In the previous truss structure example the displacements, stresses and strains are due to the applied forces and boundary conditions. In such a case, the structures members are considered having a constant and equal temperature (the so called reference temperature). Changes of temperature for some (or all) members of the structure with respect to the reference lead to the following possible effects: changing of displacement state of the structure due to the thermal expansion or contraction with no effect on the stress state, if the constrains allow free deformation (static determination conditions); changes in the stress state of the members due to the constrains when the structure is statically undetermined.

Thermal effects can be considered in the framework of initial stresses. Other sources of initial stresses can be moisture, residual welding stresses or lack of fit. Generally, they are considered as components of an initial force vector that has to be added to the external loads. Supposing the previous truss member at the reference temperature Tref (conventionally chosen such as the structure displacements, strains and stresses are zero if no external loads are applied). While the truss has norestricted displacements, for a temperature change of T the members length changes with uT = LT

49

Chapter 4

The Direct Stiffness Method_________________________________________

where is the thermal expansion coefficient, assumed to be uniform along the truss, as well as the temperatures change T. The thermal strain yields

T = uT /L = T

If the member truss is also subjected to an axial forces F, the corresponding axial stress and axial strain are = F/A and M = /E respectively. The total strain is the sum of both mechanical and thermal effects:

= M + T =

+ T

On the other hand, if the member ends are fixed against axial displacement, u = 0, = 0 and M = -T. If T > 0 the member goes in compression because = -ET = - ET < 0.

50