176 6 Deflections of Elastic Structures

It is obvious that the effect of geometric incompatibility may be useful for regulation
of the stresses in the structure. Let us consider a three-hinged arch which is loaded
by any fixed load. The bending moments are

M.x/ D M 0  Hy;

where M 0 is a bending moment in the reference beam. If a tie is fabricated longer

then required, then the thrust becomes H D H1 C H2 , where H1 and H2 are thrust
due to fixed load and errors of fabrication, respectively.
Discussion. For computation of displacement due to the settlement of supports
and errors of fabrication, we use the principle of virtual work. The common con-
cept for this principle and for Maxwell–Mohr integral is the concept of generalized
coordinate and corresponding generalized unit force.

6.5 Graph Multiplication Method

Graph multiplication method presents most effective way for computation of any
displacement (linear, angular, mutual, etc.) of bending structures, particularly for
framed structures. The advantage of this method is that the integration procedure
according to Maxwell–Mohr integral is replaced by elementary algebraic procedure
on two bending moment diagrams in the actual and unit states. This method was
developed by Russian engineer Vereshchagin in 1925 and is often referred as the
Vereshchagin rule.
Let us consider some portion AB which is a part of a bending structure; the bend-
ing stiffness, EI, within of this portion is constant. The bending moment diagrams
for this portion in actual and unit state are Mp and MN . Both diagrams for portion
AB are presented in Fig. 6.19. In general case, a bending moment diagram Mp in the
actual state is bounded by curve, but for special cases it may be bounded by straight
line (if a structure is subjected to concentrated forces and/or couples). However, it

Mp(x) dW =Mp(x)dx Centroid of Mp graph

y
Mp (Actual state)

x A dx Ωp B

yc M (Unit state)
a M
O
x
xc

Fig. 6.19 Graph multiplication method. Bending moment diagrams Mp and M in actual and unit
states
6.5 Graph Multiplication Method 177

is obvious that in the unit state the bending moment diagram MN is always bounded
by the straight line. Just this property of unit bending moment diagram allows us to
present the Maxwell–Mohr integral for bending systems in the simple form.
Ordinate of the bending moment in actual state at section x is Mp .x/. Elemen-
tary area of a bending moment diagram in actual condition is d˝ D Mp .x/dx.
Since MN D x tan ˛, then integral in Maxwell–Mohr formula may be presented as
(coefficient 1=EI by convention is omitted)
Z Z Z
Mp MN dx D .x tan ˛/Mp dx D tan ˛ xd˝: (6.19)

R
Integral xd˝ represents the static moment of the area of the bending moment
diagram in actual state with respect to axis Oy. It is well known that a static moment
may be Rexpressed in terms of total area ˝ and coordinate of its centroid xc by
formula xd˝ D ˝p xc . It is obvious that xc tan ˛ D yc . Therefore, the Maxwell–
Mohr integral may be presented as follows
Z
1 ˝p yc
Mp MN dx D : (6.20)
EI EI
R
The procedure of integration Mp MN dx D ˝p yc is called the “multiplication” of
two graphs.
The result of multiplication of two graphs, at least one of which is bounded
by a straight line (bending moment diagram in unit state), equals to area ˝ of
the bending moment diagram M p in actual state multiplied by the ordinate yc
from the unit bending moment diagram M, N which is located under the centroid
of the Mp diagram.
It should be remembered, that the ordinate yc must be taken from the diagram
bounded by a straight line. The graph multiplication procedure (6.20) may be pre-
sented by conventional symbol () as

Z
1 Mp  MN k
kp D Mp MN k dx D : (6.21)
EI EI

It is obvious that the same procedure may be applicable R to calculation

R of similar
integrals, which appear in Maxwell–Mohr integral, i.e., Np NN dx and Qp Qdx. N
If the structure in the actual state is subjected to concentrated forces and/or cou-
ples, then both the bending moment diagrams in actual and unit states are bounded
by the straight lines (Fig. 6.20a). In this case, the multiplication procedure of two
diagrams is commutative. It means that the area ˝ could be calculated on any of
the two diagrams and corresponding ordinate yc will be measured from the second
one, i.e., ˝1 y1 D ˝2 y2 . This expression may be expressed in terms of specific
ordinates, as presented in Fig. 6.20b.
178 6 Deflections of Elastic Structures

a b c e
e
y2 Mp
a c a c
Mp Mp
W1
f f
d d
y1 M b b
M M

W2 l/2 l/2 l/2 l/2

d e

Mp
Mp
W1 W2 W1 W2 W3

y1 y2 y1 y2 y3 M
M

Fig. 6.20 Multiplication of two bending moment diagrams

In this case, the displacement as a result of the multiplication of two graphs may
be calculated using two following rules:
1. Trapezoid rule allows calculating the required displacement in terms of extreme
ordinates
l
D .2ab C 2cd C ad C bc/; (6.22)
6EI
where the crosswise end ordinates has unity coefficients. This formula is precise.
2. Simpson’s rule allows calculating the required displacement in terms of extreme
and middle ordinates

l
D .ab C 4ef C cd /: (6.23)
6EI
Equation (6.23) may also be used for calculation of displacements, if the bend-
ing moment diagram in the actual condition is bounded by a curveline. If the
bending moment diagram Mp is bounded by quadratic parabola (Fig. 6.20c), then
the result of multiplication of two bending moment diagrams by formula (6.23) is
exact; this case occurs if a structure is carrying uniformly distributed load. If the
bending moment diagram Mp is bounded by cubic parabola, then the procedure
(6.23) leads to the approximate result.
If a graph Mp is bounded by a broken line, then both graphs have to be divided
by several portions as shown in Fig. 6.20d. In this case, the result of multiplication
of both graphs is Z
Mp MN dx D ˝1 y1 C ˝2 y2 : (6.24)
6.5 Graph Multiplication Method 179

Sometimes it is convenient to subdivide the curved bending moment diagram by a

number of “good” shapes, for example in Fig. 6.20e. In this case
Z
Mp MN dx D ˝1 y1 C ˝2 y2 C ˝3 y3 : (6.25)

Signs rule. According to (6.21), the displacement will be positive, when the area
of the diagram Mp and the ordinate yc of the diagram MN have the same sign. If
ordinates in (6.22) or (6.23) of bending moment diagram for actual and unit states
are placed on the different sides of the basic line, then result of their multiplication
is negative. The positive result indicates that displacement occurs in the direction of
applied unit load.
Procedure for computation of deflections by graph multiplication method is as
follows:
1. Draw the bending moment diagram Mp for the actual state of the structure.
2. Create a unit state of a structure. For this apply a unit load at the point where the
deflection is to be evaluated. For computation of linear displacement we need
to apply unit force P D 1, for angular displacement to apply unit couple
M D 1, etc.
3. Draw the bending moment diagram MN for the unit state of the structure. Since
the unit load (force, couple) is dimensionless, then the ordinates of unit bending
moment diagram MN in case of force F D 1 and moment M D 1 are units of
length (m) and dimensionless, respectively.
4. Apply the graph multiplication procedure using the most appropriate form:
Vereshchagin rule (6.20), trapezoid rule (6.22), or Simpson’s formula (6.23).
Graph multiplication method requires the rapid computation of graph areas of dif-
ferent shapes and determination of the position of their centroid. Table A.1 contains
the most typical graphs of bending moment diagrams, their areas, and positions of
the centroid. Useful formulas for multiplication of two bending moment diagrams
are presented in Table A.2.

Example 6.14. A cantilever beam AB, length l, carrying a uniformly distributed

load q (Fig. 6.21). Bending stiffness EI is constant. Compute (a) the angle of rotation
A ; (b) the vertical displacement A at the free end.

Solution. Analysis of the structure starts from construction of bending moment dia-
gram Mp due to given external load. This diagram is bounded by quadratic parabola
and maximum ordinate equals ql 2 =2.

(a) Angle of rotation at point A. The unit state presents the same beam subjected
to unit couple M D 1 at the point where it is required to find angular displacement;
direction of this couple is arbitrary (Fig. 6.21a). It is convenient that both unit and
actual state and their bending moment diagrams locate one under another.
180 6 Deflections of Elastic Structures

Fig. 6.21 (a) Actual state, a q Actual state

unit state for A and
A B
corresponding bending qA
moment diagrams. (b) Actual
state, unit state for A and c = ql2 /2
Centroid C
corresponding bending e = ql2 /8
moment diagrams a=0 MP
(kNm)
l/2 l/4
W
M=1
Unit state
for qA
f=1 yc
b=1 d=1
M

b q Actual state
A B
DA

Centroid
ql2 /2
ql 2/8
MP
l/ 2 l/4 (kNm)
W
P=1
Unit state
for DA

1⋅l/ 2 yc
1⋅l
M (m)

The next step is “multiplication” of two bending moment diagrams. The area of
square parabola according to Table A.1 is

1 ql 2
˝D   l:
3 2

Centroid of this diagram is located on the distance l=4 from fixed support. Cor-
responding ordinate yc from diagram MN of unit state is 1. Multiplication procedure
is presented in Table 6.5.
This table also contains computation of required displacement using the Simpson
rule (6.23). Ordinates a and b are taken from the bending moment diagrams for
actual and unit states, respectively, at the left end of a beam (point A); ordinates e
and f are taken at the middle of the beam AB, and ordinates c and d at the right end
(point B).
6.5 Graph Multiplication Method 181

Table 6.5 Graph multiplication procedures

General formula (6.20)
1 l
D ˝yc Simpson rule (6.23)  D .ab C 4ef C cd /
Displacement EI 6EI
!
1 1 ql 2 l ql 2 ql 2
(a) Angular A D  l 1 A D 01 C4  1C 1
2 … „ƒ‚…
EI „3 ƒ‚ 6EI „ƒ‚… „ ƒ‚8 … „2ƒ‚ …
Mp  MN yc ab
A D ˝ 4ef cd
EI
ql 3 ql 3
D D
6EI 6EI
!
1 1 ql 2 3 l ql 2 l ql 2
(b) Linear A D  l   1  l A D 0  0 C4 1 C 1l
Mp  MN EI „3 ƒ‚
2 … „4 ƒ‚ … 6EI „ƒ‚… „ 8 ƒ‚ 2… „2 ƒ‚ …
A D ˝ yc
ab
4ef cd
EI
ql 4 ql 4
D D
8EI 8EI

(b) Vertical displacement at point A. The bending moment diagram Mp for actual
state is shown in Fig. 6.21b; this diagram for problems (a) and (b) is same. The unit
state presents the same structure with concentrated force P D 1, which acts at
point A; direction of the unit force is chosen in arbitrary way. The unit state with
corresponding bending moment diagram MN is presented in Fig. 6.21b.
Computation of displacements using Vereshchagin rule in general form and by
Simpson rule are presented in Table 6.5.
Discussion:
1. Elastic curve of the beam is shown by dotted line. The tensile fibers for actual
and unit states are located above the neutral axis of the beam. Bending moment
diagrams are plotted on side of tensile fibers. In the general formula and Simpson
rule we use positive sign, because bending moment diagrams for actual and unit
states are located on the same side of the basic line. Positive signs in the resulted
displacement mean that displacement occurs in the direction of the applied unit
load. The units of the ordinates Mp and MN are (kN m) and (m), respectively.
2. The results, which are obtained by formula (6.20), are precise. Formula (6.23) is
approximate one, but for the given problem it leads to the exact result, because
the beam is loaded by uniformly distributed load, the bending moment diagram
presents quadratic parabola, and total order of curves presenting two bending
moment diagrams in the actual and unit states is equal to three. If the total orders
are more than three, then formula (6.23) leads to the approximate result.
3. The reader is invited to solve the problems above by double integration method,
initial parameters method, conjugate beam method, Castigliano theorem,
Maxwell–Mohr integral, compare their effectiveness with graph multiplication
method, and make personal conclusion about its proficient.
Example 6.15. Design diagram of symmetrical nonuniform simply supported
beam of length l is shown in Fig. 6.22. Bending stiffness equals EI for segments AD
and EB; while kEI for segment DE; parameter k is any positive number. The beam
is carrying force P . Determine the vertical displacement of point C .
182 6 Deflections of Elastic Structures

Fig. 6.22 Design diagram P

of the beam and bending A EI kEI B
moment diagrams for actual Actual
D C E state
and unit states 0.25l 0.25l 0.25l 0.25l
MP
(kN⋅m)
Ω Pl/8
a =Pl/8 c =Pl/4
P =1
Unit
state

M
(m)
yC l/8
b =l/8
d =l/4

Solution. Bending moment diagrams in actual and unit states are presented in
Fig. 6.24. For computation C D .Mp  MN /=EI, we have to subdivide bending
moment diagrams on the parts, within which the bending stiffness is constant. These
parts for the left half of the beam are AD D l=4 and DC D l=4. The Vereshcha-
gin rule for multiplication of diagrams Mp and MN within portion AD leads to the
following result

1 1 Pl l 2 l P l3
C1 D      D : (a)
EI 2 ƒ‚
„ 8 …4 „ƒ‚…
3 8 768EI
˝ yc

For portion CD the trapezoid rule is applied. According to (6.22) we get

2 3
l=4 6 7 3
62  P l  l C 2 P l  l C P l  l C P l  l 7 D 7 P l : (b)
C 2 D
6  k EI 4„ ƒ‚ 8 … 8 „ 4ƒ‚ … 4 „8ƒ‚ … 4 „4ƒ‚ … 85 768 k EI
2ab 2cd ad cb

Finally, the vertical displacement at C becomes

 
P l3 7 P l3 P l3
C D 2 C D ;
768EI 768 k EI 48EI

where D .1=8/ C .7=8k/; factor 2 takes into account symmetrical part CEB of the
beam. If k D 1, then D 1.
Example 6.16. A portal frame is subjected to horizontal force P as shown in
Fig. 6.23. The bending stiffness for each member is shown on design diagram. Cal-
culate (a) the horizontal displacement at the rolled support B and (b) the angle of
rotation at the same point B.
6.5 Graph Multiplication Method 183

a Design diagram Actual state Unit state for Dhor

P C EI2 D P P⋅h Centroid 1⋅h 1⋅h
• y1
W2 •
W1 y2
h EI1 EI3
M
A B A MP A B
B
H=P H=1 P =1
l l
RA RB

b Design diagram Actual state Unit state for qB

P C EI2 D P P⋅h Centroid 1
1
W2 yC
h W1
EI1 EI3
A B A MP M M=1
B A B
H=P
l l
RA RB 1/l
1/l

c P C D

qB
A B

ΔB

Fig. 6.23 (a) Portal frame. Actual and unit states and corresponding bending moment diagrams.
(b) Portal frame. Actual and unit states and corresponding bending moment diagrams. (c) Design
diagram and elastic curve for portal frame

Solution. As usual, the analysis starts from construction of bending moment dia-
gram in actual state.
Reactions of supports are H D P , RA D RB D P h= l. The real directions
of reactions are shown in Fig. 6.23a. The tensile fibers on elements CD and AC are
located below and right from the neutral lines of the elements, respectively. Bending
moment ordinates at point C for vertical and horizontal members are Ph.
(a) Horizontal displacement at B. For required displacement hor , the unit state
presents the same frame with horizontal force P D 1, which is applied at point B.
Direction of the unit force is chosen in arbitrary way. Only horizontal reaction
H D 1 is induced. The tensile fibers are located outdoor of the frame. The bending
moments at rigid joints C and D and within cross bar equal to 1  h.
Multiplication of the bending moment diagrams should be performed for mem-
bers AC, CD, and DB separately. For horizontal member CD, the area of the bending
moment diagram in actual state is ˝1 D .1=2/P h  l and corresponding ordinate
from unit state is y1 D 1  h. We assume that horizontal portion CD of both bending
184 6 Deflections of Elastic Structures

moment diagrams is located one under the other and vertical portion AC is located
one besides the other. For vertical member AC, the area of the bending moment
diagram in actual state is ˝2 D .1=2/h  P h. Corresponding ordinate from unit
bending moment diagram is y2 D .2=3/  1  h (Fig. 6.23a). Using Vereshchagin rule
and taking into account the different flexural rigidities for the vertical and horizontal
members, we find the required displacement:

X 1 Z
s
1 1 2 1 1
B D Mp MN ds D   h  Ph  1  h  P h  l  „ƒ‚…
1h
EI EI 1 2 3
„ ƒ‚ … „ ƒ‚ … EI 2 2
„ ƒ‚ … yc
0
˝ yc ˝
„ ƒ‚ … „ ƒ‚ …
AC element CD element

P h3 P lh2
D  : (a)
3EI 1 2EI 2

Each term of the expression for horizontal displacement has negative sign, because
the bending moment diagrams for actual and unit states are located on different
sides of the basic line of the frame. The result of multiplication of diagrams within
the vertical member BD equals to zero, since in actual state the bending moments
within the member BD are zeros. A final negative sign means that assumed unit
force produces a negative work along the real horizontal displacement B .
(b) Angle of rotation at B. The bending moment diagram for actual state is shown
in Fig. 6.23b; this diagram for problems (a) and (b) is same. For required displace-
ment B , the unit state presents the same structure with concentrated couple M D 1,
which acts at point B; direction of the unit couple is chosen in arbitrary way. In the
unit state, only vertical reactions 1= l arise. The extended fibers are located indoor of
the frame. The centroid and area ˝ of bending moment diagram in actual state and
corresponding ordinate yc from bending moment diagram for unit state are shown
in Fig. 6.23b.
The result of multiplication of diagrams within two vertical members AC and BD
equals to zero. Indeed, for these portions the procedure Mp MN D 0 because MN D 0
for member AC, and Mp D 0 for member BD. For member CD, the Vereshchagin
rule leads to the following result

X 1 Z
s
1 1 1 P lh
B D Mp MN ds D  Ph  l   1 D : (b)
EI EI 2 2 ƒ‚ … „ƒ‚…
„ 3 6EI 2
0
˝ yc
„ ƒ‚ …
CD element

The positive sign is adopted because the bending moment diagrams for actual and
unit states are located on one side of the basic line CD of the frame. A final positive
sign means that assumed unit couple produce a positive work along the real angular
displacement B . Or by other words, the actual direction of angular displacement
coincides with the chosen direction for unit couple M , i.e., the section at support B
rotates counterclockwise.
6.6 Elastic Loads Method 185

Discussion. In actual state, the bending moment along the right vertical bar does
not arise; as a result, multiplication of bending moment diagrams for this element
for both problems (a) and (b) equals to zero. Therefore, the final result for problems
(a) and (b) does not contain a term with stiffness EI 3 . This happens because in actual
state the element BD does not subjected to bending, i.e., this member has displace-
ment as absolutely rigid body. Elastic curve of the frame is shown in Fig. 6.23c.

6.6 Elastic Loads Method

Elastic load method allows simultaneous computation of displacements for set of

points of a structure. This method is based on conjugate beam method. The method
especially effective for computation of displacements for set of joints of the truss
chord; for trusses this method leads to the precise results.
Elastic loads W are fictitious loads which are applied to the conjugate structure.
Bending moments of the conjugate structure and displacements of the real structure
at the point of application of elastic loads coincide. A final expression for elastic
load at joint n of the truss may be calculated by formula

X N n  Np  l
Wn D (6.26)
EA

This formula uses the following notation: Np is internal forces due to given load and
NN n is internal forces in all members of the truss in the unit state.
The right part of the formula (6.26) is similar to formula (6.12), however left part
of 6.26 is elastic load, while in Maxwell-Mohr formula - left part is displacement.
Computation of displacements procedure is as follows:
1. Calculate the axial forces Np in all elements of the truss caused by given load.
2. Calculate the elastic load at a joint n. For this:
a. Show a fictitious truss. If a real truss is simply supported then the fictitious
truss is also simply supported.
b. Apply two unit couples M D 1 to members, which are adjacent to the joint n.
Present each couple using forces Fn1 D 1=dn1 for span dn1 and Fn D
1=dn for span dn , as shown in Fig. 6.24.
c. Calculate the axial forces NN n in all elements of the truss caused by forces in
Fig. 6.24.
d. Calculate the elastic load Wn at the joint n by formula (6.26).
3. Calculate the elastic loads W for remaining joints of the truss chord, as explained
in pos. 2.
4. Show the fictitious simply supported beam subjected to all elastic loads W . If
the elastic load is positive, then it should be directed downward, i.e., in the same
direction as the adjacent forces of neighboring couples.