Вы находитесь на странице: 1из 19

Pergamon lnt. J. Mech. Sci. Vol. 38, Nos 8 9, pp.

1017-1035, 1996
Copyright © 1996 Elsevier Science Ltd
Pnnted in Great Britain. All nghts reserved
0020-7403/96 $15.00 + 0.00

0020-7403(95) 00092-5

D E F L E C T I O N S OF T I M O S H E N K O BEAM WITH
V A R Y I N G CROSS-SECTION

F. ROMANO
Dipartimento di Ingegneria Strutturale e Geotecnica, Universita di Palermo, Viale delle Scienze,
90128 Palermo, Italy

(Received 16 September 1992; and in revised form 15 September 1995)

Abstract--Closed form solutions are presented for bending beams with linearly and (in the binomial
form) parabolically varying depth and for bending beams with linearly varying width along the
beam's length. The solutions are developed taking into account the shear deformation of the beam.
The solutions are achieved, in an original way, by transforming the fourth-order differential
equations with variable coefficients into fourth-order differential equations with constant coeffi-
cients. Though the solutions presented refer to three recurrent variations in the beam cross-section
shape, the procedure outlined can be applied to beams with binomial variation (with any exponent)
in the depth or width of the cross-section. Moreover, the solutions can be achieved for polynomial,
exponential and sinusoidal load conditions. The solutions can be utilized to obtain the stiffness
factors and the flexibility coefficients of beams in the analysis of frames. Closed form solutions for
longitudinal displacements are also presented. The analytical solutions are applied to four recurrent
beams commonly used in civil engineering practice and a comparison with a numerical procedure is
made. Copyright ~, 1996 Elsevier Science Ltd.

Keywords: cross-section, curvature, deflections, moment, rotation, shear force, Timoshenko beam.

NOMENCLATURE

A(x), A(x) cross-section area and area below the neutral axis of the beam, respectively, at the
abscissa x
A1, Bt, Dt coefficients given by Eqn (44)
A2, B2, D2, M2 coefficients given by Eqn (64)
A3, B3, Da, L3, M3, N3, Q3 coefficients given by Eqn (82)
b width of the cross-section of a beam with varying depth
b(x), bo width of the cross-section at the abscissa x and at x = 0 for beam with constant
depth
C 1 , C 2 , C 3 , C 4 , C 5 , C 6 integration constants
E,G longitudinal and tangential elastic moduli
F(t) function of t
f(x) function given by Eqn (15)
h depth of the cross-section of a beam with varying width
h(x), ho depth of the cross-section at the abscissae x and x = 0 for beam with constant
width
K polynomial variable of the characteristic equation
kt numerical factor
k2(x) function of the geometrical properties of the beam
l(x) moment of inertia of the cross-section with respect to the neutral axis at the
abscissa x
l half-length of the beam
M(x) moment at the abscissa x
P concentrated transverse load
p(x), Po longitudinal distributed load at the abscissae x and at x = 0
q(x), qo transverse distributed load at the abscissae x and at x = 0
Rt, $1, TI coefficients given by Eqn (46)
R2, $2, T2, V2 coefficients given by Eqn (66)
R3, S~, T3, U3, V3, W3, Ya coefficients given by Eqn (84)
R4, $4, T4 coefficients given by Eqn (89)
Rs, $5, T5 coefficients given by Eqn (90)
r radius of curvature
S(x) first-order moment of the area below the neutral axis
T(x) shearing force at the abscissae x and x = 0

1017
1018 F. R o m a n o

t, Z substitution variables
uIx) longitudinal displacement at the abscissa x
1) abscissa
w(x) function given by Eqn (19)
X abscissa
v(x) deflection at the abscissa x
coefficient of variability of the depth or of the width
coefficients of variability of the distributed load
F shear strain at the centroid axis
rotation of the cross-section
abscissa with origin in the neutral axis
E~ first-order moment of the area eh
normal stress functions at abscissae p and s
Tm average tangential stress at the top boundary of the area cos
(Op, (Os shaded areas at the abscissae p and s
Superscripts
, . ,- IV
first, second, third, and fourth derivative, respectively
Subscripts
p, s x = p and s, respectively

1. I N T R O D U C T I O N

One-dimensional members of variable cross-section along the length are frequently used in
highway bridges, buildings, space and aircraft structures, as well as in many mechanical
components.
The behaviour in the elastic range of bending members subjected to static loads has been
studied in the past by several authors. Het6nyi [1] studied the deflection of beams with
cross-section varying with the step, Fertis and Zobel I-2] and Fertis and Keene [3] studied
the deflection of nonprismatic members by means of the method of equivalent systems; the
actual system is substituted by another system, with concentrated or distributed loads and
constant stiffness (equivalent system), that provides the same curvature as the actual system.
Results of numerical procedures were reported in Ref. [4], in particular some special values
of reactions or deformations of a limited number of beams (beams with a binomial form
variation in the depth) as functions of the ratio of the moments of inertia of the cross-
sections at the free ends of the beams.
The coefficients of the fourth-order differential equation are variable if the geometrical
characteristics are variable along the beam's length. These coefficients are also variable for
a uniform stiffness foundation beam with variable foundation properties along the beam's
length.
Closed form solutions for deflection of members with varying geometrical cross-section
properties along the member's length and for foundation beams with varying foundation
coefficients along the beam's length were reported in Refs [3-1, [5-8] and [9], respectively. In
particular, Fertis and Keene [3] obtained a closed form solution by means of quadratures
for a cantilever beam with linearly varying inertia of the cross-section, such as a beam with
linearly varying width, loaded at the free end. Lowe [5] obtained an analytical solution for
a beam with linear varying depth subjected to end moments by means of quadratures. This
latter solution requires that the moment of inertia be a monomial expression. Thus
a particular reference system must be considered. Peyrot et al. [6] obtained the closed form
solution, after several intermediate steps, for cantilever members having circular cross-
section with linearly varying radius subjected to transverse and axial forces and bending.
However, their solution cannot be applied when the axial force vanishes, since the deflection
expression becomes invalid. Romano and Zingone [7, 8] obtained closed form solutions of
nonprismatic members by solving the fourth-order differential equations with variable
coefficients. In Ref. 1-7] nonprismatic members with circular cross-section were studied
whereas in Ref. [8] nonprismatic members with rectangular cross-section were studied. In
Ref. 1-7] the solutions of the axial displacements of members subjected to axial forces were
also given. Franklin and Scott I-9] obtained the closed form solution for elastically
Deflections of Timoshenkobeam with varyingcross-section 1019

supported beams of uniform cross-section with linearly varying foundation coefficient by


means of a contour-integral.
The solution by quadratures requires the knowledge of the moment diagram, which in
general is not known for statically indeterminate beams. Though the shape of the moment
diagram (linear, parabolical ...) is known, the parameters of the expression curve are not
known. In any case, for linear, parabolical and polynomial varying moment diagrams, such
as for constant, linear and polynomial varying distributed loads, quadratures require the
solution of Hermite integrals which are more demanding than the standard differential
equation solutions. Moreover, for exponential or sinsoidal distributed transverse loads the
solution by quadratures is very difficult. No difficulties arise in such cases using standard
differential equation solutions.
In all the aforementioned studies the effects of transverse shear deformations are neglect-
ed. Although these effects are negligible for slender beams, for short beams subjected to
concentrated moments or closely spaced concentrated loads alternating in direction, the
error relating to shear force and bending moment distributions, as well as displacements,
can be significant [10].

2. GOVERNING EQUATIONS
For small deflections of one-dimensional nonprismatic bending members the curvature,
taking shear deformation into account, cannot be expressed superimposing shear and
moment effects as in Ref. [11]. To develop deflecton theory for nonprismatic members it is
necessary to start from the compatibility equation (see Fig. 1)

F = 4, + y' (1)
where F = shear deformation of the axis member; 4, = rotation of the axis member and of
the cross-section; and y' = dy/dx (y = deflection) and M and T (see Fig. 1) are, respectively,
moment and shear force. Equation (1) becomes the well known Bernoulli beam formula
when F = 0 (y' = -4,). In Fig. 1 dashed lines refer to deformed member.
With reference to Fig. 2, the equilibrium equation along the x-axis of the shaded areas
~o~and cop of a portion of beam between the abscissae x and x + Ax (subscripts s and p refer
to abscissae x and x + Ax, respectively) is

f,o av(v)dov - f~, a~(v)dco~ - z,.bAx = 0 (2)


p s

where o(v) = stress function of v; b = width of the cross-section; and z,, = average tangen-
tial stresses. Taking into account that a = M~/I (see Fig. 2) (I = moment of inertia of the
cross-section with respect to neutral axis and ~ = abscissa with origin in the neutral axis),
Eqn (2) becomes

f, o -i-/
Mp G O ~ - f~,, ~M~ ¢ , d ~ , - ~bAx = O. (3)

~~,,~ IM + ~T =
Fig. 1. Deformation of an element of beam subjectedto moment and shear.
1020 F. Romano

P AX 1 I b --!

I Ax !

Fig. 2. Momentsand stresses acting on a portion of a beam.

For a beam with uniform width, b, we have (see Fig. 2)

ks
~p=v- ; ~s=v-~-; cop=b(hp-~p); cos=b(hs-~-s) (4)

where ~-p and ~-s = distance from the neutral axis of the top boundary of the shaded areas
cop and cos, respectively. Whence we obtain

,,~p v-~ bay- ~-~ v--~ bdv-zmbAx=O. (5)

where Opand fs -- abscissae of the top boundary of the shaded areas cop and co~,respectively.
Moreover, since

Mp = Ms + AM; I p = I s + A I ; hp=hs+Ah (6)


Eqn (5) becomes

hs2 Ah) bdv


Is + AI v 2 - bdv + Jh~ Is + AI v

h, Ms v
r,. bAx = 0. (7)

For Ax that tends to zero AI, Ah and AM tend to zero, moreover, the second integral of
Eqn (7) becomes immaterial (the integration points are coincident). By neglecting second-
order infinitesimals with respect to first-order ones, we obtain

z m = l i m ih~ AM ( ~ ) ~hsMsAh TsYsMsh'scos


,,40 Jr,, bI~Ax v - dv - J~ 2-MsAx dv - bI~ 2bls (8)

where Es is the first-order moment of the area cos with respect to the centroid axis of the
cross-section (Fig. 2) and h's is the derivative of the height function (dh/dx). Equation (8) is
similar to the formula reported in Ref. [12] for the shear stress of member with linearly
varying height; however, the formula derived here is more general because it takes into
account any height variation of a beam. For uniform member, h; = 0, hence Eqn (8)
provides the well known formula for the shear stress in beams.
With reference to a given cross-section at abscissa x, the shearing strain at the centroid
axis (neutral axis) is obtained from Eqn (8)

r(x)S(x) M(x)h'(x)A(x) kiT(x) k2(x) M(x)


F . . . . (9)
bI(x)G 2bI(x)G GA(x) GI(x)
where S(x) = first-order moment of the area below the neutral axis; A(x) = area below the
neutral axis (half of the area A(x) of the cross-section); G -- tangential elastic modulus;
kl = S(x)A(x)/bI(x) = numerical factor depending only on the shape of the cross-section
(kl = 1.5 for rectangular cross-section); and k2 (x) -- h'(x)fi~(x)/2b = function of the geomet-
rical properties of the beam.
Deflections of Timoshenko beam with varying cross-section 1021

With the same procedure outlined above we obtain the shearing stress of a beam with
varying width and constant depth
T(x)Z(x) M~b'~Z(x)
+ (10)
~'~ - b fl~ b~ I~

In this case the shear strain at the centroid axis of the cross-section is
kxT(x) k2(x)M(x)
F - - - (11)
GA(x) GI(x)
where k2(x) = -b'(x)S(x)/b(x) 2 = function of the geometrical properties of the beam.
The curvature is given by
1 M(x)
- = 4/= (12)
r El(x)
where r = radius of curvature; 0b = dqb/dx; and E = longitudinal elastic modulus.
By differentiating Eqn (1) and taking Eqns (9) (or 11) and (12) into account we obtain
M(x) kl [q(x)A(x) + T(x)A'(x)]
y"(x) =
El(x) GAZ(x)
[k'2(x)M(x)l(x) + k2(x) T(x)I(x) - k2(x)M(x)I'(x)]
- GlZ(x) (13)

where y"(x) = d2y(x)/dx2; A' = dA (x)/dx; and q(x) - distributed transverse load. For pris-
matic members A'(x) = 0, I'(x) = 0, k2 (x) = k~ (x) = 0 hence we obtain the theory of the
Timoshenko beam with uniform cross-section.
Equation (13) can be written as
M (x) = -- E1 (x) y" (x) - f (x) (14)
where
f(x) - k,EI(x) [ q(x)A(x) + TA'(x)q
L J
E[ l'(x) ]
ki(x)M(x) + k2(x)T(x) - T ~ k z ( x ) M ( x ) . (15)
+-G
Differentiating Eqn (14) twice with respect to x we obtain
r (x) = - E1 (x) y'"(x) - El' (x) y"(x) - i f ( x ) (16)
q(x) = EI(x)ylV(x) + 2EI'(x)y'"(x) + El"(x)y"(x) + f"(x) 07)
where y'"(x) = d3y(x)/dx3; yIV(x) = d 4 y ( x ) / d x 4 ; I'(x) = dI(x)/dx; I" (x) = d2 l (x)/dx2;
f'(x) = df(x)/dx; f"(x) = d2f (x)/dx 2.
Equation (17), divided by E, gives

l(x)ylV(x) + 2I'(x)y'"(x) + l"(x)y"(x) - w(~)


E (18)
where
w(x) = q ( x ) - f " ( x ) . (19)
Equation (18) is a fourth-order linear differential equation with variable coefficients that
governs the bending of the Timoshenko beam under the action of a distributed transverse
load. For members with uniform cross-section (I = constant; l'(x) = l"(x) = 0) and without
sliding of adjacent cross-sections due to shearing forces (G = oc; f ' (x) = f " (x) = 0), Eqn (18)
becomes the well known equation for beam bending [11]

y l V ( x ) -= q(x). (20)
E1
1022 F. Romano

The solution of Eqn (18) is given by the sum of the complete primitive (complementary
function) of the homogeneous equation
I ( x ) / V ( x ) + 2I'(x)y'"(x) + l"(x)y"(x) = 0 (21)
and of the particular integral of Eqn (18) [13].

3. ANALYTICAL SOLUTIONS
The analytical solutions of Eqn (18) are provided for either linearly and (in the binomial
form) parabolically varying depth or linearly varying width of members subjected to
parabolically varying distributed load. Solutions of other loading conditions are shown to
be obtained from the ones for a parabolically varying distributed load.
3.1. Linearly varyin 0 depth
Consider a homogeneous beam having cross-section with uniform width, b, and depth,
h(x), linearly varying. The depth h(x) can be written as

h(x) = ho + ~x (22)
where ho = depth of the cross-section at the abscissa x = 0; and ~ = constant.
The moment of inertia of the cross-section with respect to the neutral axis is
I (x) = ~ b(ho + ~x) 3. (23)
The first and second derivatives of Eqn (23) are, respectively,
l'(x) = ¼ b~(ho + ctx)2 (24)
I"(x) = ½b~ 2 (ho + ~x). (25)
Taking Eqns (23-25) into account, Eqn (18) becomes
12w(x)
(ho + ~x)3ylV(x) + 6~(ho + ~x)2y'"(x) + 6c(2(ho + o~x)y"(x) - Eb (26)

By means of the substitution


z = ho + ex (27)
we obtain
dz
- ~ (28)
dx
y'(x) = ~y'(z); y"(x) = ~Zy"(z); y'"(x) = ~3y'"(z); f V ( x ) = ~ 4 f V ( z ) . (29)

By multiplying by (ho + ctx) the left- and right-sides of Eqn (26) and taking Eqns (27) and
(29) into account, we obtain

z4yIV(z) + 6z 3 y'"(z) + 6z2y"(z) = 12zw(z)


~4Eb . (30)

Equation (30) is the Euler's fourth-order differential equation [13], which can be trans-
formed into a linear equation with constant coefficients by means of the substitution
z = e'. (31)

Differentiating Eqn (31) four times, we obtain


y' (z) = e - ' y'(t) (32)
y"(z) = e- 2, [y,, (t) - y'(t)] (33)
y'"(z) = e-3'[y'"(t) - 3y"(t) + 2y'(t)] (34)
yIV(z) = e-4'[yIV(t) - 6y'"(t) + 1 ly"(t) - 6y'(t)]. (35)
Deflections of Timoshenkobeam with varyingcross-section 1023

By means of Eqns (31-35), Eqn (30) becomes


y'V(t) - y'(t) = F(t) (36)
in which F(t) is the right-hand side of Eqn (30) as a function of t. Equation (36) is
a fourth-order linear differential equation with constant coefficients. The homogeneous
equation is
ylV (t) - y"(t) = 0. (37)
The roots of
K 4 -- K 2 = 0 (38)

are: 0; 0; 1; and - 1. Hence, the general solution of Eqn (37) is [13]


y(t) = C1 + C2t + C3e t + C4e -t (39)
where C1, C2, C3, and C4 are integration constants that depend on the boundary condi-
tions (kinematics and mechanics). Taking Eqns (27) and (31) into account, Eqn (39) becomes
C4
y(x) = C1 + C21n(ho + cox) + C3(ho q- ~x) q- (ho + ~x~)" (40)

The solution of the homogeneous equation (Eqn (40)) does not solve the problem of beams
with linearly varying depth subjected to concentrated loads (q(x) = 0) as in the case of
uniform beams, since the function w(x) depends on f"(x).
The parabolically varying distributed transverse load can be expressed as
q(x) = qo + fix + ?x 2 (41)
where qo = distributed transverse load at x = 0; and fl and ? are constants. The functions
w(z) on the right-hand side of Eqn (30) and F(t) on the right-hand side of Eqn (36) are,
respectively, see Eqn (19),
w(z) = A1 + Bxz + D l z 2 (42)
12et e2~)
F(t) = - K - ~ ( A 1 + Blet + D1 (43)

where
))
A1 = q o - - -~ho+ h2

B1 = fl-- 2 ~ho + ~ k17ho 2 ?ho +

_ ~' ET(~ 2kl) (44)


D1 ~2 G 3 "

A particular integral of Eqn (36) with F (t) given by Eqn (43) is (taking Eqns (27) and (31) into
account)
y(x) = Rl(ho + ~x)ln(ho + ~x) + Sl(ho + otx) 2 q- Tt(ho + ~x) 3 (45)
where
6A1 B1 D1
Rx - o:4Eb; $1 - ~4Eb; T1 - 6~4E b. (46)

The analytical solution of the deflection of beams restrained in any way with linearly
varying depth subjected to a parabolically varying distributed transverse load is given by
the sum of Eqns (40) and (45):
C4
y(x) = C1 + C21n(ho + c~x) + C3(h0 + :~x) + - -
(ho + c~x)
+ Rl(ho + c~x)ln(ho + otx) + Sl(ho + otx) 2 + Tl(ho 4- c(x)3 . (47)
1024 F. R o m a n o

3.2. Binomial form parabolically varyino depth


The binomial form parabolically varying depth can be written as
h(x) = (h 112 q- gX) 2. (48)
The moment of inertia of the cross-section with respect to the neutral axis, and its first and
second derivatives, are
l ( x ) = ~ '1v 1~{1~1/2
,o +~x) 6 (49)
I'(x) = 1 ~,cx
l.-._1,,2
~no + ctx)5 (50)
l"(x) =~oct
s , 2--1/2
tn o +ctx) 4 (51)
Taking Eqns (49-51) into account, Eqn (18) becomes
12w(x)
(h~/2 + ~ x ) 6 y l V ( x ) 4- 12ct(h~/2 + ax)Sy'"(x) + 300~2(h 1/2 + o ~ x ) e y " ( x ) = _ (52)
Eb
By means of the substitution
z = h~/2 + ~x (53)

and taking Eqn (29) into account (since the latter remains valid, h~/z being constant in Eqn
(53)), Eqn (52) provides

z4ytV(z ) + 12z3y,,,(z) + 30z2y,,(z ) = 12w(z)


ct4Ebz 2 . (54)

Equation (54) is the Euler's fourth-order differential equation, which by means of Eqns
(31-35) yield
yW(t) + 6y'"(t) + 5y"(t) - 12y'(t) = F(t) (55)
in which F(t) is the right-hand side of Eqn (54) as a function of t. Equation (55) is
a fourth-order linear differential equation with constant coefficients. The homogeneous
equation is
yW(t) + 6y'"(t) + 5y"(t) - 12y'(t) = 0. (56)
The roots of
K 4 + 6K 3 + 5K 2 - 12K = 0 (57)
are: 0; 1; - 3 ; and - 4 . Hence, the complementary function is
y(t) = C1 + C2e' + Cae -3' + C4e -4t (58)
where C1, C2, C3, and C4 are integration constants. By means of Eqns (31) and (53),
Eqn (58) becomes
Ca C4
y(X) = C l + C 2 ( h 1/e + gx) + (hi~ 2 + ~x)3 -I (hi/2 + o~x)4. (59)

For binomial variation of the depth, the second derivative of the function f ( x ) given by Eqn
(15) is affected by the shear and moment functions (for linear variation depth this does not
occur) so that w(x) and F(t) (see Eqns (19), (54) and (55)) depend on these functions. Since
the shear and moment functions are unknown for statically indeterminate beams, to
determine the particular integral, these functions can be obtained from equilibrium differen-
tial equation of a beam. For the parabolically distributed transverse load given by Eqn (41),
the shear and moment distributions are, respectively,
X2 X3
T(x) = - q o x - fi-~ - 7 - ~ + C5 (60)

X2 X3 X4
M(x) ~- --qo T - fl-;-o -- 7 ~ + Csx + C6 (61)

where C5 and C6 are integration constants.


Deflectionsof Timoshenkobeam with varyingcross-section 1025

The functions w(z) and F(t) are, respectively,


w(z) = A2z 4 + B2z 2 + D2z + M2 (62)
12
F(t) = - ~ (A2e 2t + B2 + D2e-t + M2 e-2~) (63)

where

A2 = - ~ ~ kl -

Bz -- ~~2 - ~E (°~2qo - ~flh~/2 + yho)(3 - kl)

D 2 = ~fl - 2 ~7 h~/2 - -G
E (0~3C5 + o~2qohi~2) -- ~ ho + ~Y h a / 2 ( k l - 6 )

M2 = qo - fl--h~/2 + 2 7 h o
o~

3E ~3Cshl/2 + qoho - h~/2 + ~ hg - ~'C6 . (64)


G
A particular integral of Eqn (55) with F(t) given by Eqn (63) is (taking Eqns (31) and (48) into
account)

y(x) = R2(h 1/2 + ax) 2 + Szln(ho~/2 + c~x) -~ (h~/2T2


+ ~x) + (hlo/2[/2
+ ~x) 2 (65)

where
A2 B2 D2 M2
R2 -- 5~4E b, $2 ~4Eb, T2 ~-T-~, V2 ct4Eb (66)

The analytical solution of the deflection of beams restrained in any way with parabolically
varying depth subjected to a parabolically varying distributed transverse load is given by
the sum of Eqns (59) and (65):
Ca C4
y(x) = C, + C2(h 1/24- o~x) nt- (hi~2 + ax)3 q- (hlo/2 -4- ~xx)4 -F nzln(h~/2 + ~x) 2

-I- S 2 in (hlot2 + o~x) + (hi/2T2 V2 (67)


+ ~x) + (ho~/2 + :~x)2

The constants C5 and C6 appearing in Eqns (60-61) and (64) are unknown for beams
restrained in any way. However, the determination of these constants is simple and the
procedure will be seen in Section 4.

3.3. Linearly varying width


The linearly varying width can be written
b(x) = bo + ~x (68)
where bo = width of the cross-section at the abscissa x = 0; and ct = constant.
The moment of inertia of the cross-section with respect to the neutral axis, and its first
and second derivatives, are
I(x) = ~ (bo + ~ x ) h 3 (69)
I' (x) = a-~ eh 3 (70)
I"(x) = 0. (71)
Taking Eqns (69-71) into account, Eqn (18) becomes

12w(x) (72)
(bo + ~x)yW(x) + 2c~y'"(x) = Eh 3
1026 F. Romano

By means of the substitution


z = bo + ~x (73)
and taking Eqn (29) into account (since the latter remains valid, bo being constant in Eqn
(73)), Eqn (72) provides

z4ytV(z) + 2z3y"'(z) - 12z3w(z)


~¢4Eh3 . (74)

Equation (74) is the Euler's fourth-order differential equation, which by means of


Eqns (31-35) yields the following differential equation with constant coefficients
y ' V ( t ) - 4y'"(t) + 5y"(t) -2y'(t) = F(t) (75)
in which F(t) is the right-hand side of Eqn (74) as a function of t. The homogeneous
equation is
y W ( t ) - 4y'"(t) + 5y"(t) - 2y'(t) = 0. (76)
The roots of
K 4- 4 K 3 + 5K z- 2K = 0 (77)
are: 0; 1; 1; and 2. Hence, the complementary function is
y(t) = C1 + C2 et + C3 te' + C4e 2t (78)
where C1, C2, C3, and C4 are integration constants. By means of Eqns (31) and (73),
Eqn (78) becomes
y ( x ) = C 1 + C2(b o + ~x) + C3(b o + ~x)ln(bo + ~x) + C4(bo + ~x) 2. (79)
For the parabolically varying distributed transverse load given by Eqn (41) the function
w(z) on the right-hand side of Eqn (74) and the right-hand side of Eqn (75) are, respectively,

L3 M3 N3 Q3
w(z) = A3 z2 + B3z + D3 + - - +
z - V -+- U 4- - -
~" (80)

12
F(t) = ~ (A3e 5t + B3 e4t + D3 e3t + L3 e2t + M3e t + N 3 + Q3e -t) (81)
where

A3 0~2

B3 =~-
fl 2 ~bo
7

O~=qo-~-bo +~bo + ~ +~
eEh 2 { 3

M3 - - -

+ ~ b~)

Q3 - 2G - + ~ b4 - Csboct + C6:t 2 ) . (82)


Deflections of Timoshenko beam with varyingcross-section 1027

A particular integral of Eqn (75) with F(t) given by Eqn (81) is (taking Eqns (31) and (73)
into account)
y(x) = R3(bo + ~x) 5 + S3(bo + ~x) ~ + T3(bo + ax) s + U3(bo + ax)21n(bo + ax)

Y3
+ V3(bo + ax)in (bo + ex) + - - (83)
(bo + ax)
where
As 6Bs D3 6L3
R3 - 20a4Eh 3 ; $3 --- 410~4Eh3; T3 = ot4Eh3 ; U3 - e4Eh 3 ;

6M3 6N3 Q3
V3 - c~4Eh3; W3 = ,~Eh 3; Y3 - ~4Eh3. (84)

The analytical solution of the deflection of beams restrained in any way with linearly
varying width subjected to a parabolically varying distributed transverse load is given by
the sum of Eqns (79) and (83)
y(x) = C1 + C2(bo + ~x) + C3(bo + ex)ln(bo + ex) 2 + C4(bo + :~x)2 + Rs(bo + :~x)~

+ S3(bo -t- ~x) a + T3(bo + c~x)3 + (bo + ex)ln(bo + ex)

Y3
x [U3(bo + c~x) + I/3] 4 (bo + ~x)" (85)

4. GENERAL REMARKS
By setting 7 = 0 and fl = 7 --- 0 in Eqns (44), (64), and (82) we obtain the solution for
linearly varying and for uniformly distributed transverse loads, respectively.
The solutions for the Bernoulli beam can be obtained from the previous solutions by
setting G = oc.
The solution relative to more complicated load combinations than those studied here can
be obtained by means of the superposition method.
Reference [7] shows the procedure that must be followed for the determination of the
analytical solution of members with step changes in cross-section shape.
Since the loading condition modifies only the right-hand side of the linear differential
equations, the solution can be achieved for polynomial, exponential and sinusoidal load
conditions; recursive methods exist for the determination of the particular integrals [13].
The previous analytical solutions depend on the four unknown constants Ca, C2, C3, and
C4 which must be determined by imposing the four boundary conditions (mechanical and
kinematical). For beams with parabolically varying depth and for beams with linearly
varying width the solutions also depend on the constants C5 and C6. For statically
indeterminate beams these constants are in general unknown. Hence, for these beams the
problem contains six unknown constants. The six equations that solve the problem are
obtained from the boundary conditions and from the condition that Eqns (16) and (14) give
the same shear and moment of Eqns (60) and (61), respectively, at every cross-section.
In the previous section the procedure to obtain the deflection solution for three shape
variations of the beam cross-section along the beam's length was shown. The procedure
outlined is general and can be applied for any binomial (with any exponent) shape variation
in the depth or width of the cross-section.
The procedure to obtain the solution of the differential equation governing the flexural
deflections of nonuniform members can also be applied to obtain the solution for longitudi-
nal displacement of nonuniform members longitudinally loaded. In Ref. [7] the solutions
for axial displacements were given for members with circular cross-section with varying
radius along the member's length. In this paper the solution for longitudinal displacements
is given for the three shape variations of the cross-section of the beam considered in the
previous section.
1028 F. Romano

The equation of the equilibrium in the longitudinal direction of an elastic member


portion dx of the beam provides

plx)
A(x) u"(x) + A'(x)u'(x) - (86)
E

where p(x) = distributed longitudinal load; u(x) = longitudinal displacement at the abscissa
x; and, of course, the prime denotes differentiation with respect to x.
For beams with linearly and parabolically varying depth subjected to parabolically
varying distributed longitudinal loads the analytical solutions of Eqn (86) are, respectively,

u(x) = C1 + Czln(ho + ax) + R g ( h o + ~x) + S4(h o + ~X) 2 "-~ T4(h o + ~x) 3 (87)

+ :tx) + Rsln (hlo/2 + ~x) + S5(h 1/2 + ~x) + Ts(ho1/2 + ~x) 2


u(x) = C~ + (h~/zC2 (88)

where
R4 1( Po -
~x2Eb ~- ho + ~ h~

$4- 4~ 2 Eb - 2 ~ ho

7 (89)
7"4 - 9~4 E b

Rs = ~
1 Po o~
+ ho

$5- 2~2 E b - 2~ o j

T5 - 1' b
6~4E (90)

and C1 and C2 = integration constants; Po distributed longitudinal load per unit length at
x = 0; and/~ and 7 = coefficients as in Eqn (41).
For a beam with linearly varying width the analytical solution is given by Eqns (87) and
(89) with bo in the place of ho and h in the place of b.
By setting R4 = $4 = 7"4 = 0 and R5 = Ss = T5 = 0 we obtain the solutions for concen-
trated longitudinal loads.
We observe that for e = 0 the analytical solutions here obtained are not valid; however,
as is shown [7] the solutions are strongly convergent to the solution for a uniform beam, as
:¢ tends to zero.

5. APPLICATIONS
The analytical solutions for deflection of Timoshenko beams with varying cross-section
are applied to four beams symmetrical with respect to the midspan, used especially in civil
engineering practice (Fig. 3). Different lengths as well as different midspan heights are
considered. In particular, the beams considered have lengths varying between 1500 mm and
3000 mm; for each length the beams have midspan heights varying between 250 mm and
400 mm with step 50 mm.
Parabolical loading conditions like those utilized to develop the analytical solutions and
concentrated load at midspan are considered. For all beams the constant Cs of Eqn (60) is
given by the expression: C5 = P/2 + qo I + fl12/2 + ~13/3 (C 5 = shear at x = 0 in symmetri-
cal beams with length 2l).
The longitudinal and the tangential elastic moduli are E = 30000 MPa, G = 12500 M P a
(v = 0.2).
Deflections of Timoshenko beam with varying cross-section 1029

I~ 21.=variable =I [~ - - 2L=variable t
I ~x I ~x
a) c)

b) d)
Fig. 3. Beamswith varyingcross-sectionalong the length symmetricalwith respectto the midspan:
(a) linearly varyingsimply supported (b) linearly varyingbuilt-in; (c) parabolicallyvarying simply
supported; (d) parabolicaUyvarying built-in.

A comparison is made between the analytical solution and the classical Timoshenko
solution by subdividing the beam in elements with uniform depth and between the
analytical solution and the Bernoulli beam (by assuming G = oo in the previous equations).

5.1. Beam with linearly varyin 9 depth simply supported at the ends
The boundary conditions are: y(0) = 0; M(0) = 0; ~b(1) = 0; and T(1) = P/2. The latter
three conditions imply, on account of Eqns (1), (14) and (16), that EI(O)y'(O) + f ( 0 ) = 0;
y'(l) - F(0) = 0; and EI(l)y"(l) + EI'(l)y"(1) + f ' ( l ) = - P / 2 .
The solution is given by Eqn (47), the four integration constants being determined by
imposing the above four boundary conditions on this equation and its derivatives.
In Fig. 4(a) are reported the midspan deflections of the beam for h0 = 500 ram;
b = 300mm; P = 5 0 K N , qo = 7 0 N m m -1, fi = - 0 . 0 1 N m m -2 and 7 = 0.
In Fig. 4(b) are reported the percentage errors of the midspan deflections if a Bernoulli
beam is considered instead of a Timoshenko one.
In Fig. 5 the comparison between the analytical solution here developed and the classical
Timoshenko solution by considering a subdivision of the beams in 1400 elements with
uniform depth is made. This figure shows that the numerical curve tends to the analytical
one as the slope of the height function of the member tends to zero. The error with respect to
the analytical solution is not eliminable with this numerical procedure and is due to the fact
that in each element of the subdivision the slope of the height function is zero (elements with
uniform depth). This is the same as neglecting the second term in Eqn (8). This term gives an
increment of the shear stress (or of the shear force, see also Ref. [14]) since for the beams
investigated h' ~ 0.
The procedure utilized to obtain the numerical solution of the deflections of the beams
form part of the Ritz method.
In Fig. 6 the analytical solution and the numerical results of the beams with different
numbers of subdivisions were compared. The error with respect to the analytical solution
remains as the number of the beam subdivision increases (the subdivision of the beam in 600
elements gives the same results of the subdivision of the beam in 1400 elements) for the
above considerations. In the same figure is also reported the comparison between the
Bernoulli beam and the numerical results. In this case the numerical results are convergent
to the exact solution since the shear force is not taken into account in the deflection analysis.

5.2. Beam with linearly varying depth built-in at the ends


We assume that the cross-section at x = 0 cannot warp, hence ~b(0)= 0 [11]. The
boundary conditions are: y ( 0 ) = 0; ~b(0)= 0; q~(l)= 0; and T ( l ) = P/2. The three latter
conditions imply that y'(0) - F(0) -- 0; y'(l) - F(I) = 0; and EI(l)y"'(l) + EI'(l)y"(l) +
f ' (1) = - P/2. It is pointed out that axial displacements are allowed by the built-in ends, and
hence arching action is not allowed.
1030 F. Romano

5.00

4.00 Timoshewko b e a ~ / /
Bernoulli beam / /
I:1 /,
/
Ix /
hot500 .~ ' / /

// ~Y / ..
2.00 // "/ O..C'."
./"/ , .d..
/: " sgv,f.,
0

"el 1.00

O.Oft~ 5 0 i , L i I , , i L I , , , ,
1000 1250 1500
half-length of the beam (ram)
30

b)
201
q~

t~

l i I i l l i i i I i l i i

750 1000 1250 1500


half-le~gth of the be~7~ (ram)
Fig. 4. Simply supported beam with linearly varying depth symmetrical with respect to the midspan:
(a) midspan deflections versus half-length of the beam; and (b) percentage deflection errors of the
Bernoulli beam.

5.00

4.00 Analytical solution / ~


Numerical solution / /

5.00
,. oo / /
2.00

<=
~0 1.00

0.00 r J r
750 1 obo . . . . 12so . . . . 1500
h~lf-length of the beam (ram)
Fig. 5. Comparison between analytical and numerical results of a simply supported beam with
linearly varying depth symmetrical with respect to the midspan.
Deflectionsof Timoshenkobeam with varyingcross-section 1031

5.00

•,,• 4.00 7~moshenko beam


- - - Jgerrto~ll~ beam J
?
g
hofffi 500
~3.00

l~umm'~oJ200.000.fSOOe~a~'a~~ c,. /
2.00
0

1.00 J f" / "

~ I :nak#Cio,~aad,--al
,....

0.00
750 ' ' ' ' 1000
~ ' ' ' ' 1 2~50 ' ' ' ' 1500
half-length of the beam (ram)

Fig. 6. Numerical results for differents subdivision of the beam in uniform elements.

The solution is given by Eqn (47), the four integration constants being determined by
imposing the above four boundary conditions on this equation and its derivatives.
For the same values of the loads and geometrical properties as in the previous example, in
Fig. 7(a) are reported the midspan deflections, in Fig. 7(b) the restrained end moments and
in Fig. 7(c) the percentage errors of the midspan deflections if a Bernoulli beam is considered
instead of a Timoshenko one.

5.3. Beam with parabolically varying depth simply supported at the ends
The boundary conditions are the same as those of the beam with linearly varying depth
simply supported at the ends.
The solution is given by Eqn (67), the four integration constants (since C5 and C6 are
known, C6 = 0 in particular) being determined by imposing the four boundary conditions
on this equation and its derivatives. In Fig. 8(a) are reported the midspan deflections
for ho = 500mm; b = 300mm; P = 50kN, qo = 7 0 N m m -1, fl = -0.01 N m m -2 and
7 = - 1 . 1 . 1 0 -5 N m m -3. In Fig. 8(b) are reported the percentage deflections errors if
a Bernoulli beam is considered instead of a Timoshenko one.
In Fig. 9 the comparison of the analytical solution and the numerical one by subdividing
the beams in 1400 elements is reported. In this case the error with respect to the analytical
solution is greater than that for a beam with linearly varying depth (see Fig. 5) since the
derivative of the height function is greater.

5.4. Beam with parabolically varying depth built-in at the ends


The boundary conditions are the same as those of the beam with linearly varying depth
built-in at the ends. However, since in this case the constant C6 of Eqn (61) is unknown, the
constants to be determined are five. As well as for the beam of Section 5.2, the axial
displacements are allowed by the built-in ends, and hence arching action is not allowed.
The solution is given by Eqn (67), the five integration constants being determined by
imposing, in addition to the previous four boundary conditions on this equation and its
derivatives, the condition that in whatever cross-section, the moment given by Eqn (61)
must be equal to the one given by Eqn (14).
For the same values of the loads and geometrical characteristics as for the beam in
Section 5.3, in Fig. 10(a) are reported the midspan deflections, in Fig. 10(b) the restrained
end moments and in Fig. 10(c) the percentage deflection errors if a Bernoulli beam is
considered instead of a Timoshenko one.
1032 F. Romano

1.00

~.~ 0.80 TimoshewJco beam


• -- -- l~errcoulli beam
g
"¢~0.60
0
ho= 500 / / ~ / ~

~ 0.40 17~ /j
0

*~ 0.20

t
O'On50'~ ' ' " ' 10001 , = L , 12501 , , , 1 ,00
half-length of the beam (ram)
-20

0 -- 60 =

- - Timosh~.nko beam
-- -- Ber'r~oulli beam

-1 O,0 I I
• I I
1000
( I t t .... t
1250
1, i I I l
1500
half-length of the beam (turn)
6O

1 t I I I I J I l,, I ,~ [ , I I I I

1000 1250 1500


half-length of the becLm (ram)

Fig. 7. Built-in beam with linearly varying depth symmetrical with respect to the midspan:
(a) midspan deflections versus half-length of the beam; (b) restained end moments; and (c) percent-
age deflection errors of the Bernoulli beam.

The comparison between the Timoshenko and Bernoulli beams shows that the error in
the deflection is quite big for statistically indeterminate beams (20-30%) of usual length
(3000 mm). The deflection error increases rapidly as the length of the beam decreases either
for statically indeterminate beams or for statically determinate beams. For beams having
a length 3 times the end height ho the error is of the order of 20% for statically determinate
beams and of the order 40-55% for statically indeterminate beams.
5.00

~ 4.00

l
T£moshe~aleo beam
Bernoulli becLm / / " ///
/
/

g //
,-c~3.00 / /
ho=500
2.00

"~ 1.00

0.00 , , , , l , , , ' 250 1 I I I


750 1000 1 1500
h a l f - l e n g t h of the beam (rnm)
50
b-
o
L b)
2O

4 ~Oo

0
750 ' ' 10=00 . . . . 1250
I . . . . 1500
h a l f - l e n g t h of the beam (rnrn)

Fig. 8. Simply supported beam with parabolically varying depth symmetrical with respect to the
midspan: (a) midspan deflections versus half-length of the beam; (b) percentage deflection errors of
the Bernoulli beam.

5.00

~ 4.00
Analytical solution
N~me'r'i,cu2 solution A
/
//
g /.
~3.00
//
h.= 500 //~/////
2.00
0
-ra
¢,,;I / / / i.

~ 1.00

0.00
750 , , , , 1000
i , , , 1 2~50 , , , , 1500
half-length of the beam Cram)

Fig. 9. Comparison between analytical and numerical results of a simply supported beam with
parabolically varying depth symmetrical with respect to the midspan.
1034 F. Romano

I .00

~)

0.80 - - Ti~oshe~lco beam


- - - Bernoulli beam
i
"¢:~0.60
I h o= 500 ~.eb~Tz///
"i

0
0.40

~ 0.20

00% i I i i
1000
i I I I I
1250
L i i I I
1500
half-length of the beam (mm)

-2G

-60

----- Timoshenko beam "~0~


¢o
- -- B e r n o u l l i b e a m
-

-10~i 0
1000 1250 1500
half-length of the beam (rnm)

6O
t-
o
50

40

~ 30

~ 20

I I I I I I I J I I I 1
1 '
~5o 1000 1250 1500
half-length of the beam (ram)

Fig. 10. Built-in beam with parbolically varying depth symmetrical with respect to the midspan:
(a) midspan deflections versus half-length of the beam; (b) restrained end moments; and (c) percent-
age deflection errors of the Bernoulli beam,

6. C O N C L U S I O N S

The closed form solutions of the deflections of Timoshenko beams with linearly and (in the
binomial form) parabolically varying depth and with linearly varying width have been
presented. The solutions for the axial displacement have also been presented. The solutions
for Bernoulli beams can be obtained from the ones for Timoshenko beams by setting
G=oo.
Deflections of Timoshenko beam with varying cross-section 1035

T h o u g h the analytical p r o c e d u r e has been a p p l i e d to achieve the closed form s o l u t i o n s


for three shape v a r i a t i o n s of the cross-section of the beam, the p r o c e d u r e is valid for v a r y i n g
d e p t h o r width in b i n o m i a l form, with any exponent. M o r e o v e r , the a n a l y t i c a l s o l u t i o n can
be o b t a i n e d for p o l y n o m i a l , s i n u s o i d a l a n d e x p o n e n t i a l d i s t r i b u t e d transverse loads. F o r
these l o a d s s t a n d a r d m e t h o d s to o b t a i n the p a r t i c u l a r integrals exist.
T h e s o l u t i o n s can be utilized to d e t e r m i n e the stiffness factors a n d the flexibility coeffi-
cients of b e a m s a n d the fixed ends m o m e n t s for the analysis of frames.
By m e a n s of the s u p e r p o s i t i o n m e t h o d , m o r e c o m p l e x l o a d i n g c o n d i t i o n s t h a n those
c o n s i d e r e d here can be analysed.
T h e c o m p a r i s o n of the a n a l y t i c a l s o l u t i o n with n u m e r i c a l s o l u t i o n b a s e d on the subdivi-
sion of the b e a m in elements with u n i f o r m d e p t h show t h a t the e r r o r of the n u m e r i c a l
s o l u t i o n d e p e n d s o n the derivative of the height function. The greater is the derivative of the
height function, the larger is the error.

Acknowledgement--The author gratefully acknowledges a reviewer for the substantial comments.

REFERENCES

1. M. Het6nyi, Deflections of beams of varying cross section. J. Appl. Mech. 4(2), 49 (1936).
2. D. G. Fertis and E. C. Zobel, Equivalent systems for the delection of variable stiffness members. J. Struet. Div.,
ASCE 84(ST6), 1820 (1958).
3. D.G. Fertis and M. E. Keene, Elastic and inelastic analysis of nonprismatic members. J. Struct. Engng, ASCE
116(2), 475 (1990).
4. R. J. Roark and W. C. Young, Formulas for Stress and Strain. McGraw-Hill, Tokyo (1975).
5. P. G. Lowe, Classical Theory of Structures. Cambridge University Press (1971).
6. A. H. Peyrot, M. E. Criswell, M. D. Folse and J. P. Aznavour, Reliability analysis of wood transmission poles.
J. Struct. Div., ASCE I08(ST9), 1981 (1982).
7, F. Romano and G. Zingone, Deflections of members with variable circular cross-section. Int. J. Mech. Sci.
34 (6), 419 (1992).
8. F. Romano and G. Zingone, Deflections of beams with varying rectangular cross section. J. Engng Mech.,
ASCE 118(10), 2182 (1992).
9. J.N. Franklin and R. F. Scott, Beam equation with variable foundation coeflicent. J. Engng Mech. Div., ASCE
105(EM5), 811 (1979).
10. L. M. Shirima and M. W. Giger, Timoshenko beam element resisting on two-parameters elastic foundation.
J. Engng Mech., ASCE 118(2), 280 (1992).
11. S. Timoshenko, Strength of Materials, Part I. Macmillan, London (1930).
12. R. Park and T. Pauley, Reinforced Concrete Structures. John Wiley, Toronto (1975).
13. E. L. Ince, Ordinary Differential Equations. Longmans, Green, London (1927).
14. F. R. Stanley, Mechanics of Materials. McGraw-Hill, New York (1967).

Вам также может понравиться