1

Matrix Force/ Flexibility

Method
(Reference: Yuan-yu Hsieh, Elementary Theory of Structures.)

1-1 Treatment of Intermediate Load:

Distributed loads or any other type of member loads acting between the
nodes can be replaced by equivalent loads at the nodes. To illustrate, let us
consider the three-span continuous beam shown in Fig. 1-1(a), whose center
span is subjected to a distributed load. Let us first fix artificially the joints
and of the center span (Fig. 1-1(b)] and then release them (Fig. 1-1(c)). Since
the application and removal of the artificial forces are neutralized, the original
configuration of Fig. 1-1(a) is therefore statically equivalent to the combined
effects of Fig. 1-1(b) and (c). The equivalent nodal loads at and in Fig. 1-
1(c) are the reverse of fixed-end actions (moments and shears) in Fig. 1-1(b).
Note that the final forces and displacements in the loaded member −
must be obtained by superposing the effects of the fixed-end beam and those
resulting from the nodal-force analysis of the original structure.

2
 Figure 1-1

1-2 Division of the Structure:

A structure may be taken as a composition of a number
of three-force members or other convenient units. The
behavior of the subdivided elements, as well as the whole
structure, must satisfy the following basic conditions:
1. Equilibrium of forces.
2. Compatibility of displacements.
3. Force-displacement relationship
specified by the geometric and elastic
properties of the elements.
4
1-3 Loads, Forces, Displacements and
Deformations:
Consider the frame in Fig. 1-2(a), which is composed of three elements (members ,
, and ) subjected to external loads (nodal forces) denoted by , , and with
the corresponding nodal displacements , , and , respectively.
A typical member element is shown in Fig. 1-2(b) and is generally subjected to the
internal forces of the end moments, denoted by and , and the axial forces,
denoted by . The member is also subjected to end shears; however, the end
shears can be expressed in terms of the end moments and therefore are not
considered independent forces. Associated with the end moments and axial forces
are the end rotations and and the axial elongation . The signs shown in Fig.
1-2 are considered as positive. A superscript is used for these quantities to identify
their belonging to a particular member. For instance, , , and indicate the
internal forces for member , and , , and their corresponding deformations.
5

Figure 1-2

6
In matrix representation, we use a column matrix ,

=
.
.
.
to denote all the nodal forces, and a column matrix ,

= .
.
.
to denote the corresponding nodal displacement.
7

Also, we use a column matrix ,

.
.
.
=

.
.
.
8
to denote all the internal end forces for members , , …, and a column
matrix ,

.
.
.
=

.
.
.
to denote the corresponding internal displacements. 9

Frequently involved in the subsequent discussion is the

principle of virtual work, which serves in many cases as an
effective substitute for the equations of equilibrium or
compatibility. The principle simply states that for an elastic
structure in equilibrium, the external virtual work is equal to
the internal virtual work (virtual strain energy). The virtual
work may be the result of either a virtual displacement or a
virtual force. Thus, if virtual displacements are used, we have
= (1 − 1)
On the other hand, if virtual forces are used
= (1 − 2)

10
1-4 Equilibrium, Force Transformation Matrix:
For a statically determinate structure, each of the member forces may
be expressed in terms of the external nodal loads by using the
equilibrium conditions of the system alone. Thus,
= + + ...+
= + + ...+
. (1 − 3)
.
.
= + +...+
11

in which = , = , ..... Observe that , , ..., represent

the total set of applied loads and , , ..., the total set of
member forces. No connection between the subscripts on and is
implied.
The matrix form for Eq. 1-3 is
= (1 − 4)
where
. . .
. . .
. . . .
= (1 − 5)
. . . .
. . . .
. . .
12
is called the force transformation matrix, which relates the internal forces to
the external forces. Matrix is usually a rectangular matrix in which the
typical element is the value of the internal force component caused by
a unit value of external load . Note that is merely an expression of
equilibrium for the system.
As for a statically indeterminate structure, the internal member forces cannot
be expressed in terms of the external loads by equilibrium alone. However, as
previously stated, a statically indeterminate structure can be made
determinate by removing the redundant elements. The statically determinate
and stable structure that remains after the removal of the extra restraints is
called a primary structure. We then consider the original structure as
equivalent to the primary structure subjected to the combined influences of
the applied loads and the unknown redundant forces, thereby treating the
redundants as a part of the external loads of unknown magnitude. In this way,
we can express member forces in terms of the original applied loads and
the redundant forces as
13

= + (1 − 6)
or
= | (1 − 7)

where and are force transformation matrices representing the

separate influences of the known applied loads and the unknown
redundants on the member forces. They are generally rectangular
matrices.

14
1-5 Compatibility:
Compatibility is a continuity condition on the displacements of the structure after
the external loads are applied to the structure. Compatibility must be brought into
the analysis of statically indeterminate structures since the equilibrium equations
alone do not suffice to solve the problem.
If we let denote the prescribed displacement vector corresponding to the
redundant force vector , the compatibility conditions used in the force method for
solving a static structure are that the displacements at all the cuts of redundant
points caused by the original applied loads, and the redundant forces must be made
to be equal to in order that the continuity of the structure can be maintained.
For a loaded structure mounted on rigid supports, the gap in the displacements at
redundant points resulting from applied loads is precisely removed by the redundant
forces. Therefore, the compatibility condition is
= (1 − 8)

15

1-6 Force-Displacement Relationship,

Flexibility Coefficient, Flexibility Matrix:
A flexibility coefficient is the displacement at point due to a unit
action at point , all other points being unloaded. Apparently, the
flexibility coefficient constitutes a relationship between deformation
and force. Applying the principle of superposition, we may express the
deformation at any point of a system caused by a set of forces in terms
of the flexibility coefficients.

16
Our intention is, first of all, to establish the relationship between the member
displacements and the member forces of a structure. Consider a typical
member taken from a plane structure as shown in Fig. 1-3. As before, the
member forces are represented by a column matrix ,

and the corresponding member deformations are represented by a column

matrix ,

Figure 1-3 17

Note that the clockwise end moments and rotations and the tensile axial
forces and elongation are considered as positive.
Using the flexibility coefficient , we may express each of the member
deformations in terms of the separate influences of the whole set of
member forces:
= + +
= + + (1 − 9)
= + +
or in matrix form,
= (1 − 10)

in which = (1 − 11)

18
is defined as the element flexibility matrix. Clearly, the coefficient, for instance
, is given by
= as = = =
The rest can similarly be defined.
The description above refer to an individual element. For a structure
consisting of , , …. Elements, we have
=
=
.
.
.

Let = . and = .
. .
. .
19

These equations can be put in the matrix form

= (1 − 12)

where = . (1 − 13)
.
.

which is a diagonal matrix with element flexibility matrices as its

constituents.

20
Since the flexibility coefficients of Eq. 1-11 serve to relate the member
deformations to the member forces, they are certainly governed by the
geometric and material properties of the member. Suppose that the
member is prismatic with length , cross-sectional area , moment of
inertia , and modulus of elasticity and regarded as simply supported.
The elements in the first column of are, by definition, the member
deformation resulting from = . These are found to be
= rotation of the left end =
= rotation of the right end = −
= elongation of the member =
Note that and can easily be determined by the conjugate-beam
method and that = is apparent. All the other elements can be
obtained similarly.
21

Thus, the member flexibility matrix is given by

= − 1 − 14

Note that the member flexibility matrix is symmetric because of reciprocity.

If the effect of axial forces in the member is neglected, as is usually done in
rigid-frame analysis, then
−
= (1 − 15)
−
For a truss member subjected to axial forces only,
= (1 − 16)
22
We have already established the relationship between the member deformations
and the member forces (the internal relationship). Next we must establish the
relationship between the nodal displacements and the nodal forces (the
external relationship). This may be accomplished by using the technique of virtual
work. We start by Eq. 1-2:
= (1 − 17)
Based on Eqs. 1-4 and 1-12, namely,
= and =
we have =
or = (1 − 18)
and = 1 − 19

23

Substituting Eqs. 1-18 and 1-19 in Eq. 1-17 yield

=
from which = (1 − 20)
If we let = (1 − 21)

being called the total flexibility matrix or flexibility matrix of the structure,
then
= (1 − 22)
For a statically determinate structure Eq. 1-22 gives a direct solution of all the
nodal displacements in terms of the external nodal forces.

24
1-7 Analysis of Statically Determinate
Structures by the Matrix Force Method:
As developed in sec. 1-4, for a statically determinate structure, the
internal forces can be solved by equilibrium alone:
=
See Eq. 1-4.

Also the nodal displacements can be solved by

=
See Eq. 1-20.

25

Assume that the purely statical task of evaluating the force transformation matrix
is not difficult, although this phase of analysis may cost a considerable amount of
labor in complicated problems.
The procedure for analyzing a statically determinate structure by the force method
is as follows:
1. Define the external nodal loads .
2. Define the internal member forces .
3. Determine the force transformation matrix .
Consider the elements of the first column of . If we let
= = =⋯= =
it is readily seen from equation 1-3 that , , …., are the elements. of the
first column. The rest can be obtained similarly.
4. The internal member forces are then solved by
=

26
5. Determine individual element flexibility matrices , ,….
according to Eq. 1-15 or 1-16, and assemble them as diagonal
matrix,

= .
.
.
6. Compute the flexibility matrix of the structure,
=
7. Find the nodal displacements ,
=

27

Example 1-1
Find the bar forces of the truss shown in Fig. 1-4. Find also the deflections
corresponding to the applied loads and . Assume that / = for all
members.
The load matrix is =

Figure 1-4
28
The truss has five bars designated by , , , and . The member-
force matrix is

The force transformation matrix is given by

= −

29

in which the first column contains the bar forces of the truss in Fig. 1-4
in the order , , , , , resulting from = , = . The second
column contains the corresponding bar forces resulting from = ,
= . From equilibrium

= −

For individual members the flexibility matrices are found to be

= = = = =

30
since ⁄ = for all members. Thus, the diagonal matrix is

= =

The total flexibility is then determined:

=
−
−
= −
−

=
31

The nodal displacements are solved by

=

or = =

32
Example 1-2
Find the deflections corresponding to the applied loads for the cantilever beam
shown in Fig. 1-5(a). Assume constant .
Since the loaded point of must be considered as a nodal point, it divides the
beam into two segments, designated as member and member in Fig. 1-5(b). The
internal member forces are shown by dashed lines. From equilibrium
= = =

− −( + ) −
= =
− −

Figure 1-5 33

Note that the elements of the first column of matrix are the member forces
caused by = , = = for the beam shown in Fig. 1-5(b). This gives
=− = = =

The second column of matrix contains the member forces resulting from
= , = = . Thus,
=− + = =− =

And the third column of matrix contains the member forces due to a unit
couple applied only at the free end of the beam (i.e., = , = = ).
This gives
=− = =− =

The individual member flexibility matrices are

− −
= =
− −

34
from which
−
−
=
−
−

35

The total flexibility matrix is obtained from

−
= = − + −
− −
− − − + −
−
.
− − −
−
+

+ + +
=

+ +
36
Thus,
+

+ + +
=

+ +

or
( + )
= + + (1 − 23)

( + ) ( + ) ( + )
= + + (1 − 24)

( + ) ( + )
= + + (1 − 25)
37

As a particular problem, find the vertical deflection and the rotation at the free end
of the loaded cantilever beam shown in Fig. 1-6.

Figure 1-6
To do this, we set = , = = in Eq. 1-24 to obtain
( + )
=
which is the resulting vertical deflection of the end of the beam, and we set = ,
= = in Eq. 1-25 to obtain
=
which is the resulting rotation of the end of the beam.

38
 1-8 Subsequent Derivations for Indeterminate
Structures
If the structure is statically indeterminate, then the external work must
include the work caused by the redundant forces; also, Eq. 1-6 must be
used instead of Eq. 1-4. To relate the displacements to the
corresponding forces, we begin the derivation by
+ = 1 − 26
where and are displacements corresponding to nodal forces
and redundant forces , respectively. Using Eq. 1-6 and virtual force, we
have
= +
so that
= + (1 − 27)
39

Also, because of Eqs. 1-12 and 1-6,

= = + 1 − 28
Substituting Eqs. 1-27 and 1-28 in Eq. 1-26 yields
+ = + +
or
+ = + + +
Comparing the virtual forces on the left and the right sides of the preceding
equation, we have
= +
= +
These may be arranged as
= + (1 − 29)
= + (1 − 30)
40
 if we let
= =
(1 − 31)
= =
Eqs. 1-29 and 1-30 may be put in matrix form as
= 1 − 32

For structures on rigid supports Eq. 1-32 become

= 1 − 33
=
The compatibility condition is, therefore,
+ = (1 − 34)
From which =− (1 − 35)
Eq. 1-35 expresses the solution for the redundents. 41

Substituting Eq. 1-35 in Eq. 1-29, we finally relate the unknown nodal displacements to
the corresponding applied nodal forces, in a statically indeterminate structure, covering
the effects of the redundant forces.
= −
or simply = (1 − 36)
if we let = − (1 − 37)
being the flexibility matrix of the indeterminate structure.
With the redundants found to be − , the member forces are then solved by
equilibrium:
= + = −
Let = − (1 − 38)
then, = 1 − 39
′ being the force transformation matrix of the indeterminate structure, relating directly
the member forces to the applied nodal loads covering the effects of the redundants.

42
An alternative form of ′ may be obtained in terms of ′:
= (1 − 40)
since = −
= − = ( − )

which yields Eq. 1-40. Note that Eq. 1-37 for finding is quite general but that
the alternative form given by Eq. 1-40 is more convenient if is first determined.
The following identity is useful for checking results:
= (1 − 41)
This can easily be proved as follows:
= − = −
= − =

43

1-9 Analysis of Statically Indeterminate

Structures by the Matrix Force Method
As developed in Secs. 1-4 to 1-6, the procedures for analyzing a statically
indeterminate structure by the force method are given as follows:
1. Define the external loads .
2. Define the internal member forces , and specify the redundants .
3. Calculate the force transformation matrices and from equilibrium:
= |
4. Determine the individual element flexibility matrices , ,.., and assemble
them to obtain :

= .
.
.
44
5. Calculate :
=
6. Calculate :
=
7. Find the inverse of .
8. Solve the redundants by
=−
and substitute in the equilibrium equation to obtain the member forces .
9. Alternatively, we may find ′ by
= −
and obtain the member forces by
=
10. If the nodal displacements are desired, calculate ′ by
= ′
and find by =
As seen in the latter part of Example 1-2, if the points where
the displacements are desired are not actually loaded, then we
must apply fictitious loads of zero value at these points in order
to carry out the procedures listed above.
45

Example 1-3
Find the bar forces of the truss in Fig. 1-7(a) by the force method. Also find the nodal
displacement corresponding to the applied load. Assume that = , ⁄
and ( )/ ( . ) = for all members.
The truss shown in Fig. 1-7(a) is statically indeterminate to the first degree. Let us select
bar as the redundant and denote the external load of by , as shown in
Fig. 1-7(b). The bar forces are denoted by , , ..., . From equilibrium based on the
primary structure of Fig. 1-7(b),

Figure 1-7
46
 = =
−
⎡ ⎤
⎢ − ⎥
⎢ ⎥
−
=⎢ ⎥
⎢ −
⎥
⎢ ⎥
⎢ ⎥
⎢ ⎥
⎣− ⎦

47

since ⁄ = for all members

Thus,
=

.
= − − − − =−

−
48
 =
−

= − − =
− − −

=
49

The redundant force is then solved by

=−

.
=− − = .

Substituting in the equilibrium equation, we obtain

.
.
.
= =
. − .
.
− .
50
Alternatively, we find
= −
−

− .
.
. .
= − − − =
− .
.
− − .

51

and obtain by = :

. .
. .
. .
= =
− . − .
. .
− . − .
To find , we first calculate the flexibility matrix of structure ′:
=
.
.
. .
= − =
− .
.
− .
52
The displacement is then solved:
=

. .
= = = .
,
in the direction of the applied load.

53

Example 1-4:
Find the member forces (end moments) of the rigid frame in Fig. 1-8(a) by
the force method. is constant.
The frame shown in Fig. 1-8(a) is statically indeterminate to the second
degree. It may be made determinate by inserting two pins as in Fig. 1-
8(b). Then the structure is subjected to the original applied loads denoted
by and together with the redundant couples and . The
member forces (end moments) in Fig. 1-8(b), , , ... , are shown by
dashed lines.
The force transformation matrix is obtained by considering the influences
of, = , = , = , and = successively and separately, as
shown in Fig. 1-9.

54
Figure 1-8 55

= = = =

−
−

= —

56
From individual member flexibility matrices, we form

−
−
−

−
=
−

−
−
−
57

Figure 1-9
58
 Using , and found previously, we obtain

−
= =
− −

= =
and
−
= =
−
59

The force transformation matrix of the indeterminate structure is

= −
−
−

−
−
= − .
−
− −
−

60
 − − − −
− −
− −
− − . − . −
= − =
. − ,
− −
− −

The end moment are then solved by = :

−
−
−
= − . −
.
− −

61

Using = and = = , we obtain

=− -
=− =− . -
=− =− . -
=− = . -
=
To check, we find that the identity
=
is satisfied by substituting in the values of , and ′ previously
found. The answer diagram for the end moments together with the
reactions at the supports found by statics is shown by the dashed line
in Fig. 1-8(c). 62
 Example 1-5
Find the end moments for the rigid frame shown in Fig. 1-10(a) by the
force method. Assume constant .
The equivalent form of the given loaded frame is shown in Fig. 1-10(b).
Because of symmetry, the vertical reaction at each support of the
frame is known to be acting upward, as indicated. If only flexural
deformation is considered, then the nodal axial forces, shown in the
frame in Fig. 1-10(b), only increase the compression in the two columns
but cause no effect on the end moments of the frame and can
therefore be neglected in the nodal-force analysis for obtaining end
moments. The primary structure may be chosen as the one shown in
Fig. 1-11, subjected to nodal moments and and redundant
reaction components of the left support, denoted by and . Those
shown by dashed lines are member end moments , , .... They can
be expressed in terms of and as 63

Figure 1-10 64
Figure 1-11 65

= = = =

−
−
= —
− −
− −
− −
From the member flexibility matrices, we form
−
−
−
=
−
−
−
66
Thus,

−
−
− − −
= .
− − − − −
− −
− −
−
=
−

67

= −
−

Similarly, from which

= =
( )

68
the force transformation matrix of the indeterminate structure is
= −

−
− −
= −
− − −
− −
− −

= − =
−

69
−

The end moments based on the nodal-force analysis are then

solved by = :

= = -

70
The final result is obtained by adding fixed-end moments [see upper part of
Fig. 1-10 (b)] to the end moments of member . Thus,

= = -

71

Example 1-6
Use the matrix force method to find the reaction at support , and the
deflection and slope at , for the beam shown in Fig. 1-12(a). Assume
that the spring flexibility is .
We consider the spring as a member and therefore the system is
composed of the beam portion, denoted as member , and the spring,
denoted as member . The whole can be separated into two parts: the
fixed-end beam under a uniform load and the system subjected to nodal
forces , at and redundant reaction at as shown in Fig. 1-
12(b). To obtain the nodal displacements at and reaction at , it is
necessary only to analyze the nodal-load system of Fig. 1-12(b).

72
 Figure 1-12 73

We first relate the end moments , of member , and the spring force to
the external loads , and as
−
= −

The flexibility matrix of member is

−
=
−
The flexibility of member (spring) is . Therefore, the assembled flexibility is

−
=
−
74
With , and obtained, we have
= = −

= = +

=
⁄ +
The redundant force , which is equal to the spring force , is determined by
=−

=− ⁄
−

= ⁄
Apparently, if = ,
=
75

To find the deflection and slope at , we use

= +
Now since
−
= =
−

−
= =

We reach

− −
= +
−
76
Using
= = = ⁄
We obtain

= − ⁄

= ⁄ ⁄

= + ⁄

= − + ⁄ ⁄
77

As a check, if = (rigid support), we have

= =
if = ∞ (free end),
= =−
The foregoing procedure for fixing a loaded beam is not limited to
the case of distributed loads. The procedure can also be applied to
members subjected to a set of concentrated loads, if reducing the
number of nodes is desirable.

78