Вы находитесь на странице: 1из 17

Journal of Sound and Vibration 331 (2012) 2654–2670

Contents lists available at SciVerse ScienceDirect

Journal of Sound and Vibration


journal homepage: www.elsevier.com/locate/jsvi

An approximate method for determining the static deflection and


natural frequency of a cracked beam
Tianxin Zheng n, Tianjian Ji
School of Mechanical, Aerospace and Civil Engineering, The University of Manchester, Manchester M60 1QD, UK

a r t i c l e in f o abstract

Article history: This paper provides an approximate method to determine the stiffness and the
Received 16 May 2011 fundamental frequency of a cracked beam. The cracked beam is first represented as
Received in revised form an un-cracked beam with equivalent reduced sections around the cracks. The effect of
3 December 2011
the cracks is explained, visualised and quantified using the equivalence concept
Accepted 16 January 2012
developed for stepped beams with periodically variable cross-sections. Then an
Handling Editor: I. Trendafilova
Available online 18 February 2012 alternative expression of the improved Rayleigh method is provided to calculate the
natural frequencies of a beam with a variable stiffness distribution along its length. As
the method is insensitive to the assumed mode shapes, it avoids the difficulty in
choosing appropriate mode shapes and yields accurate results. This is shown using
several examples to compare the results determined using the proposed method and
the Finite Element method (FEM). The method greatly simplifies the calculation of
cracked beams with complicated configurations, such as a beam with several cracks, a
cracked beam with concentrated masses, a beam with cracks close to each other, and a
beam with periodically distributed cracks.
& 2012 Elsevier Ltd. All rights reserved.

1. Introduction

The effect of cracks on the dynamic behaviour of a structural member had been investigated worldwide in the last few
decades. Dimarogonas [1] reviewed the published papers from 1971 to 1995 on the dynamic response of beams, rotors and
other structures with cracks. Three methods had been used to deal with the problem: an equivalent reduced section
method, a local flexibility method and a continuous cracked beam method.
The equivalent reduced section method was developed by Kirmsher [2] and Thomson [3]. They simulated the effect of the
cracked region using a local bending moment and a reduced section. It was the first attempt to quantify the stiffness
reduction in a cracked region. Because the theory of cracks and accurate numerical simulation methods were not available
at that time, the stiffness reduction due to a crack was evaluated by experiment.
The local flexibility method considered a cracked beam as a number of beam segments with the same properties connected by
elastic springs at the cracked sections. The stiffness of the springs was determined by estimating the local flexibility of the cracked
region [4]. The flexibility was quantified in the 1950s by Irwin, Bueckner, Westmann and Yang [1]. They related the local
flexibility to the crack Stress Intensity Factor (SIF). SIF [5] is an important concept in fracture mechanics, representing the stress
intensity near the tip of a crack due to remote loads or residual stresses. Using Castigliano’s theorem and fracture mechanics
relations between the strain energy release rate and SIF, Liebowitz et al. [6,7] and Okamura et al. [8] computed the flexibility of
the cracked region of a uniform beam with a transverse surface crack, which was a function of the ratio of depth of the crack to

n
Corresponding author.
E-mail address: t.zheng@ed.ac.uk (T. Zheng).

0022-460X/$ - see front matter & 2012 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jsv.2012.01.021
T. Zheng, T. Ji / Journal of Sound and Vibration 331 (2012) 2654–2670 2655

the height of the cross-section. The function was obtained experimentally according to the measurements provided by Brown and
Srawley, as it is reviewed in Ref. [1]. The factor can also be obtained in Tada et al.’s handbook [5]. The local flexibility method was
first used by Fernandez-saez et al. [9] on a simply supported beam with one transverse crack at an arbitrary location. Fernandez
divided a cracked beam with n 1 cracks into n beam segments. The deflection of each segment was defined as that of an un-
cracked beam plus a polynomial function, which included four unknown coefficients in each segment. The 4n unknown
coefficients in the n segments were determined by: (a) 3(n 1) continuity conditions of deflection, shear force and bending
moment at the cracks; (b) n1 discontinuities of the angle of rotation at the crack sections (discontinuity was determined by the
depth of the crack); and (c) four boundary conditions at the two ends of the beam. Fernandez’s method gave an exact theoretical
solution for the natural frequency of a beam with a single crack. For a cracked beam with other boundary conditions, Fernandez-
saez and Navarro [10] expressed the mode shape of a cracked beam using a flexibility influence function [11,12]. Fernandez’s
method was developed by Zhong and Oyadiji [13] for a cracked beam with a stationary concentrated mass by taking the kinetic
energy of the mass into account. Loya et al. [14] presented a first-order perturbative solution [15] for the natural frequency of a
cracked Timoshenko beam. The method assumed that the mode shape function and the rotation function of a cracked beam were
slightly different to those for an un-cracked beam. The analytical solution of a cracked beam becomes more complicated with the
increase in the number of cracks. Ostachowicz and Krawczuk [16] made a programme to calculate a cantilever with two cracks
(12 unknown coefficients), which was, perhaps, an impractical task for practicing engineers.
The continuous cracked beam method provided another approach to analysing cracked beams. Christides and Barr [17,18]
found that the existence of cracks would lead to substantial changes in stress and strain distributions in the cracked
section and the nearby vicinity. Large stress concentrations would appear near the tip of the cracks. The stress over the
cross-section was distributed nonlinearly. The change in stress and strain distribution in the cracked region was
approximated by an unknown function, which should have a maximum value at the tip, and exponentially decay with
distance from the cracked section. Although boundary conditions and some aspects of stress distribution in the cracked
section and the vicinity nearby are simulated by the function, the discontinuity in the stress distribution was an obvious
weakness of the function, as stress and strain should distribute continuously throughout a structure. The fine-meshed FE
model in Ref. [19] showed that the stress distribution in the vicinity of an abrupt change of cross-section remained
continuous. Another limitation noticed by Dimarogonas [1] was that the method treated the crack as a notch. Experiments
[20] are required to determine stress distribution functions which are then used to construct the equation of motion based
on the extended Hu–Washizu variation principle by integrating over the cross-section. Chondros et al. [21] modelled the
cracked region as a continuous flexible system by means of the displacement field in the vicinity using fracture mechanics,
leading to a better approximation of the stress and displacement distributions.
The above methods provided an accurate theoretical stiffness of a cracked beam. However, the complexity of the
solutions increases greatly when more cracks or additional masses are considered, which may be too complicated for
engineering practice. For example, there are 4(n þ1) undetermined coefficients in Fernandez’s method to a beam with n
cracks. This paper proposes an approximate method to estimate the static deflection and natural frequency of a cracked
beam, in which only one function is used for the whole beam regardless of the number of cracks. This method consists of
two steps; in Section 2, an equivalent un-cracked beam is firstly developed to represent the corresponding cracked beam
by equivalently reducing the cross-section around the cracks. This method was developed based on an equivalent
representation for stepped beams [19]. Then an alternative expression of the improved Rayleigh method in Section 3 gives
a simple approximation for the fundamental frequency of the equivalent beam with variable stiffness distribution along its
length. The method is accurate and not sensitive to assumed mode shapes. Section 4 combines the two methods presented
in the last two sections and provides a reasonable estimation of the dynamic and static stiffness of a simply supported
beam with different types of crack. Numerical verification demonstrates that the proposed method is reasonably accurate.
The conclusions are summarised in Section 5.

2. Equivalent stiffness of the cracked region

A cantilever with one deep and one shallow component was modelled in detail as shown in Fig. 1. A total of 16,000
PLANE42 elements in ANSYS were used. The bending moment at the right end of the cantilever was modelled using
linearly distributed forces; and the left end of the cantilever was fully fixed. It was found [19] that the stress is nonlinearly

Fig. 1. FE model of a stepped cantilever [1].


2656 T. Zheng, T. Ji / Journal of Sound and Vibration 331 (2012) 2654–2670

Fig. 2. Stress distribution in a transition region in a stepped beam [1]: (a) stress distribution, (b) slope line and effective region and (c) calculation model.

Fig. 3. Relationship between a stepped beam and a cracked beam: (a) dimension of a deep and a shallow component in a stepped beam and (b) a stepped
beam reduced to a cracked beam.

distributed in the transition region between the deep and shallow components, as shown in Fig. 2a. The equivalent height
of the transition region is then calculated through the equalisation of the potential energy in the cross-section. Only the
equivalent effective region shadowed in Fig. 2b contributes stiffness to the beam. The exponential curve of the effective
region was then approximated by a straight line with a slope a. The calculation model for the transition region, which is an
equivalent trapezium region, is shown in Fig. 2c; which shows that the ineffective triangular region can be removed from
the calculation model.
A two stepped section is shown in Fig. 3a, where lD, lS, hD and hS are the length and the height of the deep and shallow
components of a stepped beam, respectively. A cracked beam can be considered as a special case of a stepped beam, when
the length of a shallow component lS becomes zero (lS-0). Thus the stepped beam is similar to a cracked beam, as shown
in Fig. 3b. The depth of the crack is hD hS.
Following the same procedure used for the stepped beam, the reduced stiffness in the vicinity of the crack can be
represented by two equivalent trapezium regions connected at the upper part of the section shown in Fig. 4. This
equivalent expression of the stiffness distribution is suitable for the continuous method that considers a cracked beam as
one structure rather than a number of separated segments connected by rotational springs.
The equivalent angle a for a stepped beam is a function of the ratios hD/hS and lD/hS [19]. There are significant
dimensional differences between a cracked beam and a stepped beam, the function a for a cracked beam should be
re-constructed based upon an understanding of stepped beams:

 Ref. [19] showed that a does not change significantly with lD/hS when lD/hS is larger than 5.0, as shown in Fig. 5. For
example, when the y-axis hD/hS is equal to 2 (i.e. the depth of the crack is 50 percent of the height of the beam) and the
x-axis lD/hS is between 5.0 and 50.0, a is almost constant (only slightly changing from 33.51 to 33.61). Thus, a is not
considered as a function of lD/hS. Cracks in a beam are usually well separated from each other, hence lDyhS. Thus a can
be represented as an approximate function of hD/hS:
   
h l h
a D, D a D (1)
hS hS hS

Eq. (1) indicates that the stiffness in a cracked region only depends on the depth of the crack (hD  hS) and the height of
the beam hD. This is consistent with the relationships used in the publications [5,9,10,13].
 It was also demonstrated [19] that a is not sensitive to a change of lS, the length of the shallow part. However, lS in a
cracked beam is equal to zero. Such dramatic change of dimensions may cause a considerable change in the
quantification of a. Fig. 5 is obtained by the calculation and calibration of 221 FE stepped beam models constructed
by PLANE42 element in ANSYS. Each stepped beam has a constant height and length of the shallow component
(hS ¼100 mm, lS ¼100 mm), while the ratio of lD/hS, and hD/hS is varied from 0.5 to 50.0, and 1.25 to 10, respectively.
T. Zheng, T. Ji / Journal of Sound and Vibration 331 (2012) 2654–2670 2657

Fig. 4. The equivalent model of a crack.

Fig. 5. Equivalent angle a determined by hD/hS and lD/hS [1].

Table 1
Comparison of slope a in stepped beam (lD/hS ¼ 10) and cracked beam.

hD/hS a (deg.) Corresponding


crack depth: (hD  hS)/hD
Stepped beam Cracked beam

1.25 26.7 22.4 0.20


1.50 30.0 26.1 0.33
1.75 32.1 28.5 0.43
2.00 33.6 30.3 0.50
2.50 35.6 32.6 0.60
3.00 36.8 34.2 0.67
4.00 38.1 35.8 0.75
5.00 38.8 36.9 0.80
6.00 39.1 37.7 0.83
7.00 39.2 38.3 0.86
8.00 39.3 38.8 0.88
9.00 39.3 39.1 0.89
10.0 39.3 39.4 0.90

A stepped beam contains 30 repeated units (nD ¼ nS ¼30). When lD/hS equals 10, a varies with hD/hS, as shown in the first
column of Table 1 and Fig. 6. Modifying the same stepped beam models (30 deep components and 30 shallow
components) to a cracked beam models (30 cracks, lD/hS ¼10) by reducing lS from 100 mm to 0 mm. Following the same
calculation procedure, the equivalent angle a for a cracked beam is given in the second column of Table 1 and Fig. 6. It is
found that a for a cracked beam is slightly lower than that for a stepped beam.

Once the depth of a crack is known, the angle a can be found from Table 1 and an equivalent beam can be created for
analysis, as shown in Fig. 4.
2658 T. Zheng, T. Ji / Journal of Sound and Vibration 331 (2012) 2654–2670

Fig. 6. Comparison of the slope a in stepped beams and cracked beams.

3. Alternative expression of the improved Rayleigh’s method

Section 2 provides a simple approximate method to represent the stiffness distribution in a cracked beam (see Fig. 4).
The next step is to determine global stiffness of the beam with variable stiffness distribution along its length. The
analytical methods in [8–15] for the dynamic analysis of a beam with a variable cross-section might be too complicated for
use in engineering practice. Thus an approximate approach is proposed in this section.

3.1. Potential energy expressed by bending moment

Consider an Euler–Bernoulli beam under free vibration. The deflection y(x,t) of the beam can be expressed as a product
of the mode shape u(x) and a dynamic component sin ot:
yðx,tÞ ¼ uðxÞ sin ot (2)
The maximum potential energy can be written as a function of the maximum bending moment Mv(x) caused by the
vibration of the beam instead of the second derivative (curvature) of the mode shape u(x):
!2   2
Z L 2 Z L EIðxÞ d2 uðxÞ=dx2 Z L 2
EIðxÞ d uðxÞ M v ðxÞ
V max ¼ 2
dx ¼ dx ¼ dx (3-1)
0 2 dx 0 2EIðxÞ 0 2EIðxÞ

2
d uðxÞ
M v ðxÞ ¼ EIðxÞ (3-2)
dx2
The inertia force distribution along the length of the beam is
pðxÞ ¼ o2 rAðxÞuðxÞ (4-1)
The inertia force p(x) can also be written as the second derivative of the bending moment Mv(x) along the beam [22]:
2
d M v ðxÞ
pðxÞ ¼ (4-2)
dx2
Equating Eq. (4-1) to (4-2) gives
2
d M v ðxÞ
¼ o2 rAðxÞuðxÞ (5-1)
dx2
The first and second integrations of Eq. (5-1) give the distributed shear force Qv(x) and bending moment Mv(x),
respectively:
Z x
dM v ðxÞ
Q ðxÞ ¼ ¼ o2 rAðmÞuðmÞ dm þ C 1 ¼ o2 B0 ðxÞ þC 1 (5-2)
dx 0
Z x Z m
M v ðxÞ ¼ o2 rAðnÞuðnÞ dn dm þ C 1 x þC 2 ¼ o2 BðxÞ þ C 1 x þ C 2 (5-3)
0 0

In which:
Z x Z m
BðxÞ ¼ rAðnÞuðnÞ dn dm (6-1)
0 0
T. Zheng, T. Ji / Journal of Sound and Vibration 331 (2012) 2654–2670 2659

Z x
B0 ðxÞ ¼ rAðmÞuðmÞ dm (6-2)
0

Obviously,
Z 0 Z m
Bð0Þ ¼ rAðnÞuðnÞ dn dm ¼ 0 (6-3)
0 0

Z 0
B0 ð0Þ ¼ rAðmÞuðmÞ dm ¼ 0 (6-4)
0

Particularly, if A(n) is a constant A,


Z x Z m Z x Z m
BðxÞ ¼ rAðnÞuðnÞ dn dm ¼ rA ruðnÞ dn dm ¼ ABs ðxÞ (7-1)
0 0 0 0

In which
Z x Z m
Bs ðxÞ ¼ uðnÞ dn dm (7-2)
0 0

Substituting Eq. (5-3) in Eq. (3), the elastic energy is rewritten as


Z L 2 Z L 2
M v ðxÞ ðo BðxÞ þ C 1 x þC 2 Þ2
V max ¼ dx ¼ dx (8-1)
0 2EIðxÞ 0 2EIðxÞ

The two unknown constants C1 and C2 in Eq. (8-1) can be determined using the given boundary conditions for the beam.
The maximum kinetic energy is
Z L
1
T max ¼ rAðxÞu2 ðxÞo2 dx (8-2)
0 2

The proposed method indicates that the elastic energy of a vibrating beam can be expressed by the inertia forces; this
approach follows the principle of the improved Rayleigh method. However, the conventional improved Rayleigh method
calculates potential energy using the product of the inertia force and the static deflection induced by the force. This
requires an accurate mode shape or a calculation for an improved mode shape. The proposed method calculates the
potential (elastic) energy directly from inertial forces using integration which provides more accurate results. By equating
V max and T max , the expression for the natural frequency of the beam is
RL 2 RL
ðM ðxÞ=EIðxÞÞ dx ððo2 BðxÞ þ C 1 x þ C 2 Þ2 =EIðxÞÞ dx
o2 ¼ 0R L v ¼ 0 RL (9)
0 rAðxÞu ðxÞ dx
2
0 rAðxÞu ðxÞ dx
2

It is noted that Eq. (9) is a general expression of the natural frequency; but o2 appears in both sides of the equation. The
direct expression of o2 in Eq. (9) is complex, as the coefficients C1 and C2 are unknown until the boundary conditions are
determined. Some simple and explicit solutions of a beam with different boundary conditions are examined in the
following sub-sections.

3.2. A simply supported beam

Considering the simply supported beam shown in Fig. 7, the bending moment Mv(x) (see Eq. (5-3)) at the left end (x¼0)
of the beam is zero:
Mv ð0Þ ¼ 0 ) o2 Bð0Þ þC 1  0 þ C 2 ¼ 0 (10-1)

C2 ¼ 0 (10-2)

M(x) at the right end (x¼L) is also zero:


M v ðLÞ ¼ 0 ) C 1 L þ o2 BðLÞ ¼ 0 (11-1)

o2 BðLÞ
C1 ¼  (11-2)
L

Fig. 7. A simply supported beam.


2660 T. Zheng, T. Ji / Journal of Sound and Vibration 331 (2012) 2654–2670

By substituting Eq. (10-2) and (11-2) into Eq. (5-3), the bending moment of the beam is
 x 
Mv ðxÞ ¼ o2 BðxÞ BðLÞ (12)
L
By substituting Eq. (12) into Eq. (9), the explicit expression of the natural frequency is obtained:
RL 2
0 rAðxÞu ðxÞ dx
o2 ¼ R L 2
(13)
0 ððBðxÞðx=LÞBðLÞÞ =EIðxÞÞ dx

In the case of a beam with a constant rA(x) and EI(x), and assuming the mode shape function is sinðpx=LÞ, Eq. (12) gives
(p4/L4)(EI/rA), the exact solution of the fundamental frequency of a simply supported uniform beam. If the assumed
deflection is the parabolic function (4x/L)(1 (x/L)) instead of sin ðpx=LÞ, which therefore does not fully satisfy the boundary
conditions, the fundamental frequency obtained using Eq. (13) is (3.14284/L4)(EI/rA), only 0.15 percent larger than the
exact solution. This suggests that the proposed method is not too sensitive to the selection of the assumed mode shape.
Although the second derivative of the assumed mode shape is significantly different from the exact mode shape, the
accuracy of the calculated natural frequency remains reasonable. In contrast, if the parabolic function is used in the
standard Rayleigh Method, the error in the natural frequency is 11.0 percent.
When the beam has a constant mass distribution (rA(x)¼ rA0), such as a uniform beam with cracks, Eq. (13) simplifies
to
RL 2
0 u ðxÞ dx
o2 ¼  RL (14)
rA0 =E 0 ððBs ðxÞðx=LÞBs ðLÞ2 Þ=IðxÞÞ dx
Using the same mode shape of sin ðpx=LÞ and the concept of equivalent second moment area Eq. (14) becomes
RL 2
p4 E 0 sin ðpx=LÞ dx p4 EIeq
o2  R
4 L
¼ (15-1)
rA0 L 0 ðsin ðpx=LÞ=IðxÞÞ dx rA0 L4
2

where
RL
0 sin2 ðpx=LÞ dx
Ieq ¼ R L 2
(15-2)
0 ðsin ðpx=LÞ=IðxÞÞ dx

The natural frequency of higher modes can be obtained using the mode shape of sinðnpx=LÞ, leading to
n4 p4 EIeq,n
o2n ¼ (16-1)
rA0 L4
where Ieq,n is the equivalent second moment of area of the cross-section of the nth mode of the beam
RL
sin2 ðnpx=LÞ dx
Ieq,n ¼ R L0 2 (16-2)
0 ðsin npx=IðxÞÞ dx

The simplified solution to Eqs. (15) and (16) with a variable stiffness but a constant mass distribution will be used in the
calculation for cracked beams studied later.

3.3. A cantilever

For the cantilever shown in Fig. 8, the two unknown coefficients C1 and C2 can be determined using the boundary
conditions at its free end:
The shear force Qv(x) at the free (left) end is zero, thus
Q v ð0Þ ¼ 0 ) o2 B0 ð0Þ þ C 1 ¼ 0 (17-1)

C1 ¼ 0 (17-2)
The bending moment Mv(x) at the free (left) end is also zero, thus
Mv ð0Þ ¼ 0 ) o2 Bð0Þ þC 2 ¼ 0 (18-1)

C2 ¼ 0 (18-2)

Fig. 8. A cantilever and the boundary conditions at the left end.


T. Zheng, T. Ji / Journal of Sound and Vibration 331 (2012) 2654–2670 2661

Substituting Eqs. (17-2) and (18-2) into Eq. (5-3) gives


Mv ðxÞ ¼ o2 BðxÞ (19)
The natural frequency of the cantilever is
RL
rAðxÞu2 ðxÞ dx
o2 ¼ R L0 2
(20)
0 ðBðxÞ =EIðxÞÞ dx

The assumed mode shape function of the fundamental mode of a cantilever should satisfy the following the boundary
conditions: at the fixed end, the deflection and slope equal zero: u(L)¼0 and u0 (L)¼0. The assumed mode shape function is
px
uðxÞ ¼ 1sin (21)
2L
Substituting the mode shape function into Eq. (20), the fundamental frequency of the simply supported beam is then
obtained.

3.4. A propped cantilever

The propped cantilever shown in Fig. 9 is an indeterminate structure. The calculation of C1 and C2 is more complicated
than for the previous cases.
The bending moment Mv(x) at the left end is zero, thus
Mv ð0Þ ¼ 0 ) o2 Bð0Þ þC 1  0 þ C 2 ¼ 0 (22-1)

C2 ¼ 0 (22-2)
The deflection at the left end is also equal to zero. The deflection u(x), x¼ 0 at the left end is calculated in Fig. 10 and
Eq. (23):
Z L
Mv ðxÞMF ðxÞ
uð0Þ ¼ dx ¼ 0 (23)
0 EIðxÞ
M F ðxÞ is the bending moment of a cantilever subjected to a unit force at the left end:
M F ðxÞ ¼ Fx ¼ x ðF ¼ 1Þ (24)
By substituting Eqs. (5-3) and (24) into Eq. (23), C1 is obtained:
RL
ðlBðlÞ=EIðlÞÞ dl
C 1 ¼ o2 0R L 2 (25)
0 ðl =EIðlÞÞdl

Substituting Eqs. (22-2) and (25) back to (5-3):


RL !
2 0 ðlBðlÞ=EIðlÞÞ dl
M v ðxÞ ¼ o BðxÞx RL 2 (26)
0 ðl =EIðlÞÞ dl

The natural frequency of the propped cantilever is


RL 2
0 rAðxÞu ðxÞ dx
o2 ¼ R L  RL RL 2  (27)
0 ðBðxÞxð 0 ðlBðlÞ=EIðlÞÞ dl= 0 ðl =EIðlÞÞ dlÞ2 =EIðxÞ dx

The assumed mode shape function of the fundamental mode of the propped cantilever should satisfy the following
boundary conditions: (a) the deflections at both ends are equal to zero: u(0) ¼0, u(L) ¼0; (b) the curvature at the left end is

Fig. 9. A propped cantilever.

Fig. 10. A cantilever subjected to a unit force at the left end.


2662 T. Zheng, T. Ji / Journal of Sound and Vibration 331 (2012) 2654–2670

equal to zero: u00 (0) ¼0; and (c) the slope at the right end also equals zero: u0 (L)¼0. The assumed mode shape function is
uðxÞ ¼ xðLxÞ2 ðLþ 2xÞ=L4 (28)

3.5. A fix-supported beam

The beam and its boundary conditions are shown in Fig. 11:
The transverse deflection (Fig. 12a) and slope (Fig. 12b) at the left end is calculated in Eqs. (29-1) and (30-1),
respectively. MM ðxÞ is the bending moment of a cantilever subjected to a unit bending moment M¼1 at the left end
Z L
M v ðxÞM F ðxÞ
uð0Þ ¼ dx ¼ 0 (29-1)
0 EIðxÞ

M F ðxÞ ¼ Fx ¼ x ðF ¼ 1Þ (29-2)
Z L
Mv ðxÞM M ðxÞ
u0 ð0Þ ¼ dx ¼ 0 (30-1)
0 EIðxÞ

MM ðxÞ ¼ 1 ðM ¼ 1Þ (30-2)

Substituting Eqs. (5-3) and (29-2) into Eq. (29-1); and substituting (5-3) and (30-2) into Eq. (30-1), leads to
RL RL RL RL
ð1=EIðlÞÞ dl 0 ðlBðlÞ=EIðlÞÞ dl 0 ðl=EIðlÞÞ dl 0 ðBðlÞ=EIðlÞÞ dl
C 1 ¼ o2 0 R L 2 RL RL RL (31-1)
0 l =ðEIðlÞÞdl 0 1=ðEIðlÞÞ dl 0 ðl=EIðlÞÞ dl 0 ðl=EIðlÞÞ dl

RL RL RL 2 RL
0 ðl=EIðlÞÞ dl 0 ðlBðlÞ=EIðlÞÞ dl 0 ðl =EIðlÞÞ dl 0 ðBðlÞ=EIðlÞÞ dl
C 2 ¼ o2 RL RL RL 2 RL (31-2)
0 ðl=EIðlÞÞ dl 0 ðl=EIðlÞÞ dl 0 ðl =EIðlÞÞ dl 0 ð1=EIðlÞÞ dl

The assumed mode shape of the fix-supported beam should satisfy the following boundary conditions: (a) the
deflection at both ends is equal to zero: u(0)¼0, u(L)¼0; (b) the slope at both ends is equal to zero too: u0 (0) ¼0, u0 (L) ¼0.
The mode shape is
2px
uðxÞ ¼ 1cos (32)
L

3.6. Numerical verification

Two single span Euler–Bernoulli beams with variable cross-sections are used to verify the accuracy of the formulae
proposed in Sections 3.2–3.5. The first beam has a symmetrically distributed cross-section, as shown in Fig. 13a and the
variable height h(x) of the beam is shown in Eq. (33-1). The other is asymmetric and has a rectangular cross section with a
linearly variable height shown in Fig. 13b and Eq. (33-2). R in the equations is the difference between the largest and
smallest heights of the cross-section and indicates how quickly the height varies. It ranges from 2 to 10 with an increment
of 1 in this study
 px
Model I ðsymmetric modelÞ : hðxÞ ¼ 1 þ ðR1Þ sin b (33-1)
L

Fig. 11. A fixed supported beam.

Fig. 12. Calculation model for a fix-supported beam: (a) a cantilever subjected to a unit force at the free end and (b) a cantilever subjected to a unit
bending moment at the free end.
T. Zheng, T. Ji / Journal of Sound and Vibration 331 (2012) 2654–2670 2663

Fig. 13. Two simply supported beams with variable cross-sections: (a) Model I (symmetric model) and (b) Model II (asymmetric model).

Table 2
Comparison of the fundamental frequency of FEA and the proposed method.

R SS CA PC FS

FEA Eq. (13) Err. (percent) SRM Err. (percent) FEA Eq. (21) Err. (percent) FEA Eq. (27) Err. (percent) FEA Eq. (31) Err. (percent)

Model I (symmetric model)


2.0 9.33 9.34 0.1 9.75 4.5 2.39 2.41 0.8 12.1 12.1 0.2 15.4 15.5 0.7
3.0 13.1 13.1 0.2 14.4 9.9 2.60 2.64 1.3 15.8 15.9 0.5 19.0 19.3 1.2
4.0 16.6 16.7 0.3 19.0 14.5 2.71 2.76 1.6 19.3 19.5 0.7 22.6 22.9 1.6
5.0 20.0 20.1 0.4 23.7 18.5 2.78 2.83 1.8 22.8 23.0 0.9 26.0 26.5 1.9
7.0 26.5 26.7 0.5 33.0 24.5 2.86 2.92 2.0 29.4 29.7 1.2 32.8 33.6 2.3
10.0 35.8 36.0 0.6 47.0 31.3 2.92 2.98 2.2 38.9 39.5 1.5 42.5 43.7 2.8

Model II (asymmetric model)


2.0 7.52 7.55 0.4 7.99 6.3 4.04 4.04 0.0 13.0 13.1 1.0 17.3 17.5 1.5
3.0 9.50 9.58 0.9 10.9 14.7 6.38 6.38 0.1 17.5 17.9 2.3 22.2 23.0 3.6
4.0 11.3 11.5 1.4 13.9 23.0 8.82 8.84 0.1 21.7 22.5 3.5 26.8 28.3 5.5
5.0 13.0 13.2 1.8 16.9 30.0 11.3 11.4 0.2 25.9 27.0 4.5 31.3 33.5 7.1
7.0 16.1 16.5 2.4 23.0 42.9 16.5 16.6 0.4 33.9 36.0 6.1 39.8 43.7 9.8
10.0 20.5 21.2 3.1 32.2 57.1 24.5 24.6 0.5 45.6 49.1 7.9 52.2 58.8 12.7

SS: Simple-supported beam; CA: Cantilever; PC: Propped cantilever; FS: Fix-supported beam, SRM: Standard Rayleigh’s Method.

 x
Model II ðasymmetric modelÞ : hðxÞ ¼ 1þ ðR1Þ b (33-2)
L
The fundamental frequencies of the beams with different boundary conditions are firstly calculated using the FEM. The
FE models of the beams are constructed using BEAM4 element in ANSYS; each of the beams is divided into 100 units along
the length of the beams; the material being steel (E: 210 GPa, density: 7850 kg/m3). The fundamental frequencies of the
beams are also estimated using the proposed alternative expression of the improved Rayleigh’s method (Eqs. (13), (20),
(27) and (31)) and the assumed mode shapes in Eqs. (21), (28) and (32). In addition, the natural frequency of a simply
supported beam is calculated using the standard Rayleigh’s method. A comparison of the three methods is shown in
Table 2. It is found that: (a) the fundamental frequencies predicted by the proposed method agree well with the FEA results
in most cases; and (b) the assumed mode shapes are far different from the true ones for non-uniform beams functions, but
they do not significantly affect the estimation. As expected, the accuracy of the proposed method reduces with the increase
of the degree of non-uniformly and unsymmetrical distribution of the cross-section. (c) The accuracy of the proposed
method is clearly better than that of the standard Rayleigh’s Method.

4. Equivalent dynamic and static stiffness of simple-supported cracked beams

4.1. A beam with a single crack

A simply supported beam with one transverse crack can be represented by four segments, A, B, C and D, connected in
series, as shown in Fig. 14a. The Segments A and D have the height hD of the beam. The stiffness reduction induced by the
transverse crack is considered using two symmetric trapezoidal components of Segments B and C. The effective height of
the two segments reduces from hD to hS along a length of ðhD hS Þ=tan a, as shown in Fig. 14b.
The second moment of area of the cross-section of Segments A and D is I¼IA(x)¼ID(x)¼(h3b/12). The variable height
and the second moment of the Segments B and C are:
Segment B:
 
hD hS
hðxÞ ¼ hS þðcxÞ tan a c oxoc (34-1)
tan a

hðxÞ3 b ðhS þ ðcxÞ tan aÞ3 b


IB ðxÞ ¼ ¼ (34-2)
12 12
2664 T. Zheng, T. Ji / Journal of Sound and Vibration 331 (2012) 2654–2670

Fig. 14. Equivalent model of a cracked beam: (a) four segments in the calculation model of a beam with a crack and (b) the variable equivalent height in
the cracked region.

Segment C
 
hD hS
hðxÞ ¼ hS þ ðxcÞ tan a c o x o c þ (35-1)
tan a

hðxÞ3 b ðhS þ ðxcÞ tan aÞ3 b


IC ðxÞ ¼ ¼ (35-2)
12 12
The mode shape function of the simple-supported  beam is assumed to be sinðpx=LÞ and the corresponding fundamental
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
frequency o is calculated using the formula p2 =L2 ð EIeq =rAÞ. The equivalent second moment of area of the beam is
obtained by substituting Eqs. (34-2) and (35-2) into Eq. (16-2). Assuming that the crack occurs at the centre of the span, Ieq
becomes
R L=2 R L=2
sin2 ðpx=lÞ dx 2
0  sin ðpx=LÞ dx
Ieq ¼ R L=2 0  ¼ R ðL=2Þððh h Þ=tan aÞ  R L=2   (36)
0 sin2 ðpx=LÞ=IðxÞ dx 0
D S
sin2 ðpx=LÞ=IA dx þ L=2 ððh h Þ=tan aÞ sin2 ðpx=LÞ=IB ðxÞ dx
ð Þ D S
The bending moment varies along the length of the beam, but the variation of the bending moment in Segments B and C
is small and can be considered constant, since the lengths of Segments B and C are very small in comparison with the
length of the beam:
 
px p L hD hS L
sin2 ffi sin2 ¼ 1  ox o (37)
L 2 2 tan a 2
The second integration in the denominator of Eq. (36) is thus simplified to
Z L=2 Z L=2 2 2
sin2 ðpx=LÞ 1 6ðhD hS Þ
dx ffi dx ¼ 2 2
(38)
ðL=2ÞððhD hS Þ=tan aÞ IB ðxÞ ðL=2ÞððhD hS Þ=tan aÞ IB ðxÞ hD hS b tan a
Substituting Eq. (38) into Eq. (36), gives
3 2
bhD hS L
Ieq ¼ 2 3 2
(39-1)
a
12ðhS L þ2hD cot 2hD hS cot aÞ
In the case of a crack in arbitrary location c (see Fig. 14b), Ieq is
RL 2 RL
u ðxÞ dx sinðxÞ2 dx
Ieq ¼ P3n þ 1 R0 ¼ P4 R0 2
(39-2)
2
i¼0 Li ðu ðxÞ=Ii ðxÞÞ dx i ¼ 0 Li ðsinðxÞ =Ii ðxÞÞ dx

In which
uðxÞ ¼ sinðxÞ (39-3)
The calculation of Ieq of a cracked beam using Eq. (39-2) is straightforward:

(a) The assumed mode shape function u(x) of a simply supported beam is sinðxÞ;
(b) The denominator is split into several piecewise integration. There are total 3n þ1 piece of integration in a beam with n
cracks. Each piece of the integration represents the effect of a segment on the cracked beam.
(c) The range of each segment is determined by the locations and depth of the adjacent cracks. In Fig. 14b, the range is
        
0,c ðhD hS Þ=tan a , cððhD hS Þ=tan aÞ,c , c,c þððhD hS Þ=tan aÞ and c þ ððhD hS Þ=tan aÞ,L for Segments A–D.
(d) The stiffness of the each segment is also clear: the second moment of area of the segments far away from the crack
(Segments A and D) equals to that of the uncracked beam, while that of the segments near the crack (Segments B and
C) is determined by the depth of the crack, as shown in Eqs. (34-2) and (35-2).

The format of the calculation of the equivalent stiffness depends on the boundary condition; they can be deduced from
Eqs. (13), (20), (27) and (31). It is noted that the integration of the segment becomes more complicated when the fixed
boundary conditions are considered.
T. Zheng, T. Ji / Journal of Sound and Vibration 331 (2012) 2654–2670 2665

Eqs. (39-1) and (39-2) are based on the equalisation of the fundamental frequency. Ieq will have a different value for
different equivalence criteria. Considering a concentrated force F applied at the centre of the beam, the static stiffness K is
defined by F divided by the maximum deflection D at the mid-span:
F F 1
K¼ ¼ R L=2 ¼ R L=2 (40-1)
D 2 ðFx2 =4EIðxÞÞ dx 2 ðx2 =4EIðxÞÞ dx
0 0

Similarly, the static stiffness Keq for the equivalent uniform beam is

F F 1
K eq ¼ ¼ R L=2 ¼ R L=2 (40-2)
Deq 2 ðFx2 =4EIeq Þ dx 2 ðx2 =4EIeq Þ dx
0 0

By substituting Eqs. (34-2) and (35-2) into Eq. (40-1) and equating the two static stiffnesses (or the maximum
deflections), Ieq is obtained:
R L=2 2 R L=2 2
x dx x dx
Ieq ¼ R L=2 0  ¼ R L=2ððhD hS Þ=tan aÞ  
0
R L=2   (41-1)
x2 =IðxÞ dx x 2 =I dx 2
0 0 A þ L=2ððhD hS Þ=tan aÞ x =IB ðxÞ dx

RL R L=2 R L=2
u2 ðxÞ dx x2 dx þ ðLxÞ2 dx
Ieq ¼ P3n þ 1 R0 ¼ 0
P4 R
0
(41-2)
i¼0 Li ðu2 ðxÞ=I i ðxÞÞ dx i¼0
2
Li ðx =Ii ðxÞÞ dx

In which:
)
uðxÞ ¼ x, x r L=2
(41-3)
uðxÞ ¼ Lx, x 4 L=2

Eqs. (39-1) and (41-1) are the simplified expressions of Eqs. (39-2) and (41-2), respectively, using the symmetric
characteristics of a beam with a crack at the centre. Two simply supported cracked beams with one crack are used for
verification. Both beams have a length of 0.2 m, height of 0.01 m and width of 0.1 m. The location of the crack is c ¼0.20 m
(see Fig. 14b) in Model 1 and c¼0.15 m in Model 2, respectively. The depth of the crack varies between 0 and 90 percent of
the beam height with increments of 10 percent. The results calculated by the proposed method is compared with those
from the FEM, Fernandez’ analytical method [9]. A FE model of the beams is constructed using 32,000 PLANE42 elements in
ANSYS [23]: 40 elements through the height of the cross-section and 800 elements along the length. The material of the
cracked beam is steel (E: 210 GPa, density: 7850 kg/m3).
The fundamental frequencies and the maximum deflections due to a 1 kN concentrated force applied at the mid-span of
the two beams are calculated using FEA, Fernandez’s method and the proposed method. The results are compared in
Tables 3 and 4 show that:

(a) The fundamental frequencies of the cracked beams calculated by the proposed method agree well with the FEA results.
Although the maximum error is 6.3 percent in Model 2 with a crack of 90 percent the depth of the beam, the errors in
the proposed method are basically less than 2.0 percent;
(b) The error of the proposed method rises with increasing depth of the crack; the proposed method performs better in
Model 1 than in Model 2, when the ratio of (hD  hS)/hD is bigger than 0.6. This is because the difference between the
assumed shape function sin ðpx=LÞ (from uniform simply supported beam) and the true one becomes larger with the
increase of the degree of unsymmetrical stiffness distribution. Furthermore, since the assumed mode shape is different
from the exact mode shape, it is equivalent to applying additional constraints to the beam, which force the beam to
deform in the assumed shape and provide additional potential energy to the system. Thus the estimated natural
frequencies are higher than the exact ones;

Table 3
Fundamental frequency and static deflection of Model 1.

Depth of crack: Fundamental frequency (Hz) Max. deflection (mm)


(hD  hS)/hD
FEA Eq. (39-1) Err. Fernandez [9] Err. (percent) FEA Eq. (41-1) Err. (percent)

0.0 579 586 1.2 586 1.2 0.096 0.095  1.0


0.1 576 584 1.4 583 1.2 0.097 0.096  1.0
0.2 568 575 1.2 575 1.2 0.102 0.101  1.0
0.3 554 562 1.4 561 1.3 0.109 0.108  0.9
0.4 532 540 1.5 539 1.3 0.122 0.120  1.6
0.5 500 506 1.2 506 1.1 0.143 0.142  0.7
0.6 454 459 1.1 456 0.5 0.184 0.182  1.1
0.7 386 390 1.0 385  0.2 0.270 0.269  0.4
0.8 289 294 1.7 288  0.3 0.515 0.509  1.2
0.9 160 164 2.5 158  1.0 1.786 1.755  1.7
2666 T. Zheng, T. Ji / Journal of Sound and Vibration 331 (2012) 2654–2670

Table 4
Fundamental frequency and static deflection of Model 2.

Depth of crack: (hD  hS)/hD Fundamental frequency (Hz) Max. deflection (mm)

FEA Eq. (39-2) Err. (percent) Fernandez [9] Err. (percent) FEA Eq. (41-2) Err. (percent)

0.0 579 586 1.2 586 1.2 0.096 0.095  1.0


0.1 578 585 1.2 585 1.2 0.096 0.095  1.0
0.2 577 581 0.7 581 0.7 0.097 0.096  1.0
0.3 566 573 1.2 573 1.2 0.099 0.099 0.0
0.4 554 561 1.2 561 1.3 0.103 0.102  0.8
0.5 535 542 1.3 541 1.1 0.108 0.107  0.9
0.6 504 512 1.6 507 0.7 0.119 0.118  0.8
0.7 451 459 1.9 451 0.0 0.141 0.140  0.1
0.8 359 372 3.6 358  0.3 0.202 0.202  0.2
0.9 210 223 6.3 207  1.3 0.518 0.518 0.0

Fig. 15. A cracked beam with a stationary mass.

(c) The accuracy of the proposed method is close to that of Fernandez’ method in most cases. Because no assumed mode
shape is required in Fernandez’s method, it performs better than the proposed method in the extreme cases, such as
when the depth of the cracked is more than 80 percent of the beam, in which case the assumed mode shape is
significantly different from the exact one. However, the complexity of the proposed method does not increase
remarkably with the increase of the number of the cracks, while the solution of the Fernandez’s method becomes very
complicated for a beam with several cracks, as the number of the unknown coefficients in Fernandez’s method is
4(n þ1), in which n is the number of cracks;
(d) Since there is no assumed shape required in the static equivalence, the maximum deflection obtained by the proposed
method is more accurate.

4.2. A beam with several cracks and masses

The proposed method can be extended to a cracked beam with stationary masses. Assume a stationary mass is placed at
location d and a crack occurs at location c of a simply supported beam, as shown in Fig. 15.
The maximum potential energy of a beam can be expressed in terms of the maximum bending moment of the beam:
Z L 2 Z L
M ðxÞ ðM beam ðxÞ þ Mmass ðxÞÞ2
V max ¼ dx ¼ dx (42-1)
0 2EIðxÞ 0 2EIðxÞ

Mbeam(x) is the bending moment caused by the motion of the beam, which has been given previously (see Eq. (10)):
 x 
M beam ðxÞ ¼ o2 BðxÞ BðLÞ (42-2)
L
Mmass(x) is the bending moment caused by the motion of the mass:
( 2
o muðdÞ Ld L x ð0 o x o dÞ
M mass ðxÞ ¼ (42-3)
o2 muðdÞ dL ðLxÞ ðd ox o LÞ

The maximum kinetic energies of the beam and the mass are :
T max ðxÞ ¼ T beam ðxÞ þT mass ðxÞ (43-1)
Z l
rAðxÞo2
T beam ¼ u2 ðxÞ dx (43-2)
0 2

1 1
T mass ¼ mv2 ¼ mo2 u2 ðxÞ (43-3)
2 2
T. Zheng, T. Ji / Journal of Sound and Vibration 331 (2012) 2654–2670 2667

Table 5
Fundamental frequency of Model 3.

Depth of crack: M ¼1.0 kg M ¼ 3.0 kg


(hD  hS)/hD
FEA Eq. (44) Err. (percent) FEA Eq. (44) Err. (percent)

0.0 384 388 0.9 264 266 0.8


0.1 382 386 0.9 262 264 0.8
0.2 375 379 0.9 257 259 0.8
0.3 364 367 0.9 248 250 0.8
0.4 347 350 0.9 236 238 0.8
0.5 322 325 1.0 218 220 0.9
0.6 287 291 1.3 194 196 1.2
0.7 240 242 1.0 160 162 0.8
0.8 176 178 1.5 117 118 1.3
0.9 96.6 97.4 1.9 63 64 1.4

Table 6
Fundamental frequency of Model 4.

Depth of crack: M ¼1.0 kg M ¼ 3.0 kg


(hD  hS)/hD
FEA Eq. (44) Err. (percent) FEA Eq. (44) Err. (percent)

0.0 445 453 1.8 326 334 2.3


0.1 445 453 1.8 326 333 2.3
0.2 442 450 1.8 325 332 2.2
0.3 438 446 1.7 323 330 2.2
0.4 432 439 1.7 319 326 2.1
0.5 421 428 1.7 313 320 2.0
0.6 403 410 1.8 303 309 2.0
0.7 370 377 1.8 284 289 1.9
0.8 306 316 3.3 243 250 2.7
0.9 186 199 6.7 156 164 5.6

Equating the maximum potential energy to the maximum kinetic energy, using the assumed mode shape of sin ðpx=LÞ,
the fundamental frequency of the beam is
RL 2 2 
0 rAsin ðpx=LÞ dx þ msin pc=L
o2 ¼ R   2 R

 2
 (44)
c L2 pd =EIðxÞ dx þ L L2
0 p2 sinðpx=LÞ þ mxððLdÞ=LÞsin L c p2 sin ðpx=LÞ þmðLxÞðd=LÞ sin pd=L =EIðxÞ dx

The masses of the two cracked beams are 1.56 kg and the attached masses are 1 kg and 3 kg, respectively, at simply
supported beam using in Models 1 and 2. The location of the mass and the crack (see Fig. 15) is c¼0.1 m, d ¼0.1 m in
Model 3, and c¼0.05 m, d¼0.15 m in Model 4, respectively. Tables 5 and 6 compare the fundamental frequencies of the
two beams obtained using FEA and Eq. (44). The results of the proposed method are reasonably accurate in most cases.
Similar to the observations from Table 4, the error obtained using the proposed method increases with the increase of the
depth of the crack and with the un-symmetric stiffness distribution. However, at the worst situation, where the crack
locates at a quarter span and the length of the crack reaches 90 percent of the height of the beam, the error is 6.7 percent.
The concentrated masses are large in comparison with the mass of the beam, but they do not obviously affect the accuracy
of the proposed method.
The calculation procedure can be extended to a beam with several cracks and stationary masses. Fig. 16 shows two
simply supported beams with two cracks and three masses at different positions. The results given in Table 7 demonstrate
the natural frequencies calculated by the proposed method agree well with those obtained using FEA, although the same
mode shape sin ðpx=LÞ is used. The maximum error is 3.0 percent.

4.3. A beam with cracks close together

Cracks in most cracked beams are well separated, thus the interaction effects between adjacent cracks is negligible.
However, the effect becomes large when the cracks are close to each other, as shown in Fig. 17a. Two cracks divide the
beam into three segments: the two segments outside the two cracks have length of c1 and c2, respectively; and
the segment between the two cracks has a length of e that is less than the height of the cracks. Between the two cracks, the
two equivalent slope lines might intersect within the cross-section. Since the ratios of e/hS,1 and e/hS,2 are probably less
2668 T. Zheng, T. Ji / Journal of Sound and Vibration 331 (2012) 2654–2670

Fig. 16. Two simply supported beams with two cracks and three masses: (a) Model 5 and (b) Model 6.

Table 7
Fundamental frequency of Models 5 and 6.

Depth of crack: Model 5 Model 6


(hD  hS)/hD
FEA Eq. (51-2) Err. (percent) FEA Eq. (51-2) Err. (percent)

0.0 235 237 0.9 235 237 0.9


0.1 233 235 1.0 234 236 1.0
0.2 228 230 0.9 231 233 1.0
0.3 219 221 0.9 225 227 1.0
0.4 206 207 0.9 217 219 1.0
0.5 188 189 0.8 205 207 1.0
0.6 164 165 0.7 186 188 1.1
0.7 134 135 0.7 159 161 1.5
0.8 96.1 97.0 0.9 120 122 2.1
0.9 51.2 51.8 1.2 65.9 67.7 3.0

Fig. 17. Calculation model for a beam with two cracks close to each other: (a) two cracks in a beam and (b) equivalent model.

than 5, angles b and g should therefore be determined using the two variable relationship hD/hS and lD/hS by the contour in
Fig. 5.
Assuming hS,2 4hS,1 and the two lines between the cracks intersect within the cross-section, the equivalent second
moment of area of the cross-section in Fig. 17 is
3
hD b
IðxÞ ¼ ð0 o x o c1 ðhD hS,1 Þ cot aÞ (45-1)
12

ðhS,1 þ ðc1 xÞ tan aÞ3 b


IðxÞ ¼ ðc1 ðhD hS,1 Þ cot a o x o c1 Þ (45-2)
12
 
ðhS,1 þ ðxc1 Þ tan bÞ3 b ðhS,2 hS,1 Þ cot2 b þ d cot b
IðxÞ ¼ c1 ox oc1 þ (45-3)
12 cot b þ cot g

 
ðhS,2 þ ðc1 þdxÞ tan gÞ3 b ðhS,2 hS,1 Þ cot2 b þ d cot b
IðxÞ ¼ c1 þ ox oc1 þe (45-4)
12 cot b þ cot g

ðhS,2 þ ðxc1 dÞ tan yÞ3 b


IðxÞ ¼ ðc1 þ e ox o c1 þ e þðhD hS,2 Þ cot yÞ (45-5)
12

3
hD b
IðxÞ ¼ ðc1 þ eþ ðhD hS,2 Þ cot y o x o lÞ (45-6)
12
The equivalent Ieq can be determined based on Eqs. (45-1)–(45-6). In Model 7, two cracks are placed at the middle of
the simple supported beam, c1 ¼c2 ¼0.097 m, d ¼0.006 m (see Fig. 17). The comparison in Table 8 demonstrates that the
T. Zheng, T. Ji / Journal of Sound and Vibration 331 (2012) 2654–2670 2669

Table 8
Fundamental frequency and static stiffness of Model 7.

Depth of crack: (hD  hS)/hD Fundamental frequency (Hz) Max. deflection (mm)

FEA Prop. Err. (percent) FEA Prop. Err. (percent)

0.0 579 586 1.2 0.096 0.095  1.0


0.1 574 581 1.3 0.099 0.097  1.9
0.2 559 567 1.4 0.106 0.105  1.4
0.3 536 544 1.4 0.119 0.117  1.6
0.4 503 510 1.5 0.141 0.138  1.9
0.5 456 464 1.8 0.179 0.175  2.3
0.6 394 402 2.1 0.252 0.245  2.7
0.7 315 323 2.7 0.415 0.402  3.2
0.8 220 227 3.2 0.877 0.855  2.6
0.9 116 119 2.6 3.257 3.226  1.0

Fig. 18. Calculation model for a beam with evenly distributed cracks: (a) cracks distributed evenly in a beam and (b) equivalent model.

Table 9
Fundamental frequency and static stiffness of Model 8.

Depth of crack: Ieq (10  9 m4) Fundamental natural frequency (Hz) Max. deflection (mm)
(hD  hS)/hD
FEA Prop. Err. (percent) FEA Prop. Err. (percent)

0.0 83.2 579 586 1.2 0.096 0.095  1.0


0.1 79.6 568 573 0.9 0.101 0.100  0.8
0.2 69.6 532 536 0.7 0.115 0.114  1.1
0.3 56.8 481 484 0.6 0.142 0.140  1.3
0.4 43.2 421 422 0.4 0.186 0.183  1.3
0.5 30.7 355 356 0.2 0.261 0.258  1.0
0.6 20.0 287 287 0.1 0.402 0.397  1.2
0.7 11.5 217 218 0.1 0.699 0.690  1.4
0.8 5.24 147 147 0.1 1.538 1.515  1.5
0.9 1.39 75.4 75.7 0.4 5.882 5.765  2.0

fundamental frequencies and the static stiffness obtained by the proposed method reasonably match those obtained
using FEA.

4.4. A beam with evenly distributed cracks

Consider a beam with evenly distributed cracks (Fig. 18) which has the following characteristics:

 The beam is divided into several equal segments by the cracks. The bending moment within each segment can be
considered approximately constant.
 The cracks have the same depth.

A beam with evenly distributed cracks has a number of equal segments; each segment can be considered as a
calculation unit, as shown in Fig. 18b. The stiffness of the whole beam is equal to that of the calculation unit [19]. If the two
lines in a calculation unit do not intersect within the cross-section, it consists of two trapezium transition segments and a
rectangular region. According to Ref. [19], Ieq of the three parts connected in series is
P 2 2  
L bhD hS LD 2ðhD hS Þ
Ieq ¼ P i ¼ lD  40 (46-1)
Li =Ii 2 3 2 2
12ðh LD þ h cot a3hD h cot a þ 2h cot aÞ tan a
S D S S
2670 T. Zheng, T. Ji / Journal of Sound and Vibration 331 (2012) 2654–2670

If the two sloping lines intersect and the rectangular region vanish, Ieq of the two trapezium regions becomes:
2  
ð2hS þ lD tan aÞ2 hS 2ðhD hS Þ
Ieq ¼ lD  o0 (46-2)
12ð4hS þlD tan aÞ tan a
Consider the same simple supported beam that is divided by 9 cracks into 10 even segments (lD ¼0.02 m in Fig. 18).
Because of the periodical stiffness distribution of the cracked beam, the same Ieq is used for both dynamic and static
calculation. The comparison in Table 9 shows a good agreement between the results obtained from the proposed method
and those using FEA.

5. Conclusions

By considering that a crack in a beam is a special case of a stepped beam when the length of a shallow segment
becomes zero, the proposed method provides a model of un-cracked beam to represent the cracked beam. As the
equivalent beam has a variable stiffness distribution along its length, it leads to difficulties for determining its mode shape.
To overcome the difficulties the alternative expression of the improved Rayleigh’s method is developed to estimate the
fundamental frequency and static stiffness of a cracked beam. The method is not sensitive to the mode shapes selected and
the predictions are accurate enough. The method is verified using a number of cracked beams by comparison with results
obtained using FEA and Fernandez’ method [9].
The proposed method can simplify the calculation of complicated cases of cracked beams, such as a beam with a
number of cracks and a number of stationary masses, a beam with cracks close together, and a beam with evenly
distributed cracks.

References

[1] A.D. Dimarogonas, Vibration of cracked structures: a state of the art review, Engineering Fracture Mechanics 55 (5) (1996) 831–857.
[2] P.G. Kirmsher, The effect of discontinuity on natural frequency of beams, Proceedings of the American Society of Testing and Materials 44 (1944)
897–904.
[3] W.J. Thomson, Vibration of slender bars with discontinuities in stiffnes, Journal of Applied Mechanics 17 (1943) 203–207.
[4] A.D. Dimarogonas, Vibration Engineering, West Publishers, St Paul, 1976.
[5] H. Tada, P. Paris, G. Irwin, The Stress Analysis of Cracks Handbook, Research Corporation, St Louis, MO, 1985.
[6] H. Liebowitz, W.D. Claus Jr, Failure of notched columns, Engineering Fracture Mechanics 1 (2) (1968) 379–383.
[7] H. Liebowitz, H. Vanderveldt, D.W. Harris, Carrying capacity of notched columns, International Journal of Solids and Structures 3 (4) (1967) 489–490
IN1–IN2, 491–500.
[8] H. Okamura, et al., A cracked column under compression, Engineering Fracture Mechanics 1 (3) (1969) 547–564.
[9] J. Fernandez-saez, L. Rubio, C. Navarro, Approximate calculations of the fundamental frequency for bending vibrations of cracked beams, Journal of
Sound and Vibration 225 (2) (1999) 345–352.
[10] J. Fernandez-saez, C. Navarro, Fundamental frequency of cracked beams in bending vibrations: an analytical approach, Journal of Sound and Vibration
256 (1) (2002) 17–31.
[11] L. Meirovitch, Analytical Methods in Vibrations, Macmillan Publishing, New York, 1967.
[12] J.E. Penny, J.R. Reed, An integral equation approach to the fundamental frequency of vibrating beams, Journal of Sound and Vibration 19 (4) (1971)
393–400.
[13] S. Zhong, S.O. Oyadiji, Analytical predictions of natural frequencies of cracked simply supported beams with a stationary roving mass, Journal of
Sound and Vibration 311 (1–2) (2008) 328–352.
[14] J.A. Loya, L. Rubio, J. Fernandez-saez, Natural frequencies for bending vibrations of Timoshenko cracked beams, Journal of Sound and Vibration 290
(3–5) (2006) 640–653.
[15] T. Kato, A Short Introduction to Perturbation Theory for Linear Operators, Springer-Verlag, New York, 1982.
[16] W.M. Ostachowicz, M. Krawczuk, Analysis of the effect of cracks on the natural frequencies of a cantilever beam, Journal of Sound and Vibration 150
(2) (1991) 191–201.
[17] S. Christides, A.D.S. Barr, One-dimensional theory of cracked Bernoulli–Euler beams, International Journal of Mechanical Sciences 26 (11–12) (1984)
639–648.
[18] S. Christides, A.D.S. Barr, Torsional vibration of cracked beams of non-circular cross-section, International Journal of Mechanical Sciences 28 (7) (1986)
473–490.
[19] T. Zheng, T. Ji, Equivalent representations of beams with periodically variable cross-sections, Engineering Structures 33 (3) (2011) 706–719.
[20] A.D.S. Barr, An extention of the Hu–Washizu variational principle in linear elasticity for dynamic problems, Journal of Applied Mechanics 33 (2)
(1966) 465.
[21] T.G. Chondros, A.D. Dimarogonas, J. Yao, A continuous cracked beam vibration theory, Journal of Sound and Vibration 215 (1) (1998) 17–34.
[22] R.W. Clough, J. Penzien, Dynamics of Structures, second ed, McGraw-Hill Book Co., New York, 1993.
[23] K. Peter (Ed.), ANSYS, Theory Manual, 12th ed. SAS IP, Inc., 1994.

Вам также может понравиться