You are on page 1of 12

Measurement 114 (2018) 150–161

Contents lists available at ScienceDirect

Measurement
journal homepage: www.elsevier.com/locate/measurement

Experimental measurement of dynamic properties of composite slabs from MARK


frequency response

Hamed Allahyaria, Iman M. Nikbinb, , Saman Rahimi R.c, Ahmad Allahyarid
a
Department of Civil Engineering, Monash University, Melbourne, Australia
b
Department of Civil Engineering, Rasht Branch, Islamic Azad University, Rasht, Iran
c
Young Researchers and Elite Club, Rasht Branch, Islamic Azad University, Rasht, Iran
d
Department of Civil Engineering, Sari Branch, Islamic Azad University, Sari, Iran

A R T I C L E I N F O A B S T R A C T

Keywords: Nowadays, in order to increase the live load bearing capacity and provide large-scale cost savings associated
Composite deck with construction projects, using structures with a lower dead load and higher strength is extremely common
Modal parameter identification throughout the world. This issue causes structures to be highly susceptible to vibration and, as a consequence,
Damping meeting vibration serviceability dominates design criteria. Hence, identifying dynamic characteristics is crucial
Dynamic identification
to provide a desirable serviceability. Recently, unfilled steel-concrete composite decks with perfobond rib shear
Finite element analysis
connectors are used in buildings and bridges as a novel structural system. A little amount of research has been
reported till date on the dynamic characteristics of this structural system. Thus, this study focuses on the dy-
namic characteristics of unfilled steel-concrete decks, including normal-weight high-strength concrete (HC) and
lightweight high-strength concrete (LHC). Some of the main dynamic characteristics such as damping ratio,
natural frequencies, and frequency response functions (FRFs) assessed by means of non-destructive technique
(NDT) with hammer excitation. Subsequently, the experimental results in terms of natural frequencies were
compared with the finite element model (FEM) predictions. It is concluded that there is good agreement for
natural frequencies with difference of less than 13% and consequently the developed FEM model can be used for
structural performance prediction and damage detection of composite decks with reliable accuracy. The results
show that the damping ratios and natural frequencies of the decks fabricated with LHC (DLHC) and HC (DHC)
decreases in comparison to those of decks fabricate with plain concrete (DPC). The most effective mode was the
first mode with a damping ratio of almost 0.5% for both DHC and DLHC. DPC and DLHC had approximately
similar serviceability, whereas DLHC can be more applicable than DPC due to lower weight.

1. Introduction composite decks composed of perfobond rib shear connectors. Refer-


ences are made to the studies of Leonhardt et al. [4]; Oguejiofort and
Nowadays, using structures with lower dead load and higher Hosaint [5]; Higgins and Mitchell [6]; Machacek and Studnicka [7];
strength in order to augment the live load bearing capacity and reduce Kim and Jeong [8]; Ciutina and Stratan [9]; Ahn et al. [10]; Vianna
the impact of the earthquake has been extremely widespread [1–3]. et al. [11]; Jeong et al. [12]; Kim and Choi [13]; Cho et al. [14].
Steel-concrete composite decks are the examples of these structures that On the other side, the use of high strength concrete has become
have been fostering structural applications as they provide optimization more frequent around the world due to the advances in technology,
of construction time, cost and performances via profiting from benefits which mainly affects intrinsic properties of concrete including tensile
of different materials. In such hybrid structures, the horizontal shear split and flexural strength, stiffness, creep deformation and shrinkage
between steel and concrete slab is transferred by welded connections, [15,16]. Besides these, due to lower porosity, inhomogeneity and micro
for instance studs, steel plates, rebars, and others. In order to transfer cracks in the cement paste, HC is much more qualified compared to PC.
shear forces, using an alternative type of shear connector, called a From the seismic point of view, the main dynamic properties of
perfobond strip, in steel-concrete decks led to introducing this type of structural materials, such as the mass, stiffness, damping ratio, and
deck (Fig. 1). Several researchers studied the static behavior of resonance frequency, dramatically affect the amplitude of seismic force


Corresponding author.
E-mail addresses: allahyari.h@gmail.com, hamed.allahyari@monash.edu (H. Allahyari), nikbin@iaurasht.ac.ir, nikbin_c_eng2007@yahoo.com (I. M. Nikbin),
saman.r.r.ac@gmail.com (S. Rahimi R.), ahmad4039@hotmail.com (A. Allahyari).

http://dx.doi.org/10.1016/j.measurement.2017.09.030
Received 6 July 2016; Received in revised form 17 September 2017; Accepted 19 September 2017
Available online 21 September 2017
0263-2241/ © 2017 Elsevier Ltd. All rights reserved.
H. Allahyari et al. Measurement 114 (2018) 150–161

experimental tests. They compared the results with some international


codes to assess the serviceability of introduced system. Gandomkar
et al. [35] measured natural frequencies of a Profiled Steel Sheet Dry
Board (PSSDB), both numerically and experimentally. Neves et al. [36]
detected natural frequencies and mode shapes of a composite deck
system in a multi-story building. Da Silva et al. [37] investigated ac-
celeration and vibration of a steel-composite floor in order to determine
its comfortableness. For timber FS, Jarnero et al. [38] experimentally
detected damping ratios, natural frequencies, and mode shapes. Devin
et al. [39] indicated the influence of non-structural partition on modal
properties of a concrete FS.
Hou and Xia [40] analyzed the dynamic characteristics of a simply
supported composite beam, and their results were validated by ex-
perimental results. The prominent feature of steel-concrete composite
decks is the existence of perfobond connectors contributing to the slip
of concrete slab. Thus, perfobond connectors have a potential to reduce
the vibrations of structures, and play a crucial role in reducing the
amplitude of structural vibrations and in increasing the fatigue life of
Fig. 1. A composite bridge deck (The D.S Brown Company, (2007)).
structures [41].
The damping ratio is of particular interest with a central role for
of structures [17,18]. Since the seismic force is a response of inertia of designers, whereas analytical or numerical methods cannot determine
mass, a structure with a lower weight would have smaller seismic force that. Hence, experimental measurements are extremely required
in comparison to a heavy structure. Light weight concrete (LWC) is a [7,42–44].
commonly used material that can effectively replace plain concrete Previous literature has mainly focused on the dynamic properties of
(PC). There are some good examples of bridges and other structures the decks and floors manufactured with PC. Inadequate investigations
where LWC has been used as the main material. Therefore, using LHC in are existed, regarding the natural frequency, the damping ratio, and the
bridge decks to take the advantages of both HC and LWC is of particular mode shapes of the steel-concrete composite deck manufactured with
interest. HC and, specially, LHC. This work, therefore, aims at the impacts of
There are three primary reasons to conduct modal tests of bridges. using HC and LHC, as the concrete part of steel-concrete composite
First, accurate information on the actual performance of bridges un- deck, on the modal parameters of this type of deck using three different
dergoing serviceability conditions can be provided by vibration-based methods by means of NDT technique as a rapid, cost effective, and
evaluation. Second, it provides the identification of main dynamic accurate method. Eventually, the experimental frequencies are com-
characteristics, particularly, damping ratios, natural frequencies, and pared with frequencies derived from FEM. The updated FEM can be
mode shapes. Third, updating and validation of finite element models used to more reliably predict structural performance of composite deck
can be possible through the modal parameters. This can be employed under unusual situations [45].
successfully as a basis for structural health monitoring and damage
detection process [19–21].
Among all modal parameters, damping ratios are of particular in- 2. Theory of methods
terest [22]. Yet, since they cannot determine analytically, measure-
ments from laboratory models are demanded. Nonetheless, they are the Unlike conventional vibration theory associated with the response
most difficult modal parameters to determine accurately. On the other of a dynamic system, modal analysis is related to intrinsic properties of
hand, steel-concrete bridge decks are manufactured as lightweight structures. FRF is one of the most sophisticated methods to examine the
structures with low damping and frequencies. These slender decks are modal analysis [46]. In this NDT, the signal at one of the fixed trans-
more susceptible to dynamic excitation and, as a consequence, affect ducers is applied as a reference to obtain the frequency response
the serviceability. Hence, the vibration trouble pertaining to composite function and the impulse response function. The vibration anaylizer
bridge decks should be analysed with caution, seeking applicable al- software is used to analyze the acquired input provided by impact
ternatives to minimize the human activities vibration effects. hammer and output amplitude signals derived from the accelerometer.
A large number of studies have been done in order to explore the Three supplementary identification methods have been taken into ac-
dynamic characteristics of the structural system with concentrating on count in the present investigation: two of them based on frequency
the natural frequencies and damping ratios [23–26]. Some literature domain analysis and one on time domain analysis.
and studies are available to represent the significance of natural fre-
quencies to reveal floor systems’ serviceability under human activities.
Wiss and Parmelee [27] suggested a response rating relation for floor 2.1. Frequency response function (FRF)
system (FS). Murray [28] suggested a relation in the case of critical
damping ratio, Murray et al. [29] presented a design criteria graph For a harmonic force f (t ) = F (ω). e jωt , the response of a single-
regarding the maximum acceleration and natural frequency of a FS. degree-of-freedom (SDoF) system is another harmonic function
Ellingwood and Talin [30] determined the peak acceleration of a FS. x (t ) = X (ω). e jωt where X (ω) is a complex amplitude. Inserting these
Fukuwa et al. [31] studied dynamic characteristics of a pre- terms into the equation of motion leads to:
fabricated steel structures by obtaining the natural frequency and
damping ratio for various construction steps. El-Dardiry et al. [32] X (ω) 1
=
detected the natural frequencies of a long span concrete floor using FEM F (ω) k−ω2m + jωc (1)
and experimental tests. They modeled a number of FEMs and compared
their results with experimental results. Ferreira and Fasshauer [33] This ratio, often illustrated by α (ω) , is defined as FRF of the system.
carried out a free vibration study on a composite plate through a new This FRF known as receptance FRF uses displacement as the response.
numerical method. Ju et al. [34] introduced a new composite FS and The response can also be velocity (Mobility FRF) or acceleration
measured the natural frequencies and damping ratios by conducting (Acceleration FRF).

151
H. Allahyari et al. Measurement 114 (2018) 150–161

Fig. 4. The dimensions of modeled decks.

estimate the natural frequency (ωr ) for weakly damped structures,


which is equal to the frequency value of the peak point, and damping.
Fig. 2. Half-power bandwidth method for damping measurement.
In order to estimate damping, two points (ωa and ωb ) from each side of
the peak point with the amplitude 3 db = αmax are picked as shown in
2
Fig. 2. By substituting their values into the Eq. (2), it can be estimated
[46].
ωb−ωa
ξ=
2ωr (2)
It is important to note that this method is valid when modes are well
separated. Therefore, it is difficult to determine damping ratio using
this method when modes are very close to each other.

2.3. Circle fit method

The circle fit method is the most common SDoF modal analysis


method based on frequency domain analysis. It conforms to the circu-
larity of Nyquist plot of an SDoF FRF. The Nyquist plot indicates real
part of an FRF versus its imaginary part on the complex plane.
For an SDoF system, the mobility FRF is given as:

Y (ω) =
k−mω2 + jωc (3)
Fig. 3. The theoretical form for decaying response with linear viscous damping.
Its real and imaginary parts of Eq. (3) are shown as:
ω2c
Table 1 Re[Y (ω)] =
Mix proportion. (k−mω2)2 + (ωc )2 (4)

Materials Concrete series ω (k−mω2)


Im[Y (ω)] =
(k−mω2)2 + (ωc )2 (5)
LHC HC
Subsequently, it is straightforward to demonstrate the following equa-
Water (kg/m3) 150 175 tion mathematically:
Cement (kg/m3) 450 580
2
Normal-weight coarse aggregate (kg/m3) – 940 2
⎛Re ⎡Y (ω)− 1 ⎤ ⎞ + (Im[Y (ω)])2 = ⎛ 1 ⎞
Light-weight coarse aggregate (kg/m3) 300 –
⎝ ⎣ 2c ⎦ ⎠ ⎝ 2c ⎠ (6)
Normal-weight fine aggregate (kg/m3) 740 565
Silica fume (kg/m3) 50 70 The aforementioned equation is representative to a circle which has
Superplasticizer (kg/m3) 4.05 2.85
W/(C + SF) 0.3 0.27
1/2c in diameter. It can be mathematically demonstrated that the
Density of concrete (kg/m3) 1871 2450 natural frequency is at the location where the maximum variation rate
Compressive strength (28-day; Mpa) 66.5 81.9 in arc length occurs on the Nyquist circle. According to the trigono-
Modulus of elasticity (Gpa) 28.7 44.2 metric equations, by dividing the imaginary part by the real part, the
damping ratio can be derived as:

2.2. Half-power bandwidth method ωr2−ω2 Re[Y (ω)]


ξ=
2ωωr Im[Y (ω)] (7)
Half-power bandwidth is one of the simplest SDoF modal analysis
Like half-power bandwidth method, circle fit method is based on the
methods based on frequency domain analysis. In this method, FRF
SDoF assumption; in the vicinity of a resonance, FRF is dominated by
amplitude of the system is determined first. The method can reliably
the impact of that vibration mode, and the influences of other vibration

152
H. Allahyari et al. Measurement 114 (2018) 150–161

Fig. 5. The details of the main bars.

Fig. 6. An assembled steel grid.

Fig. 7. A composite deck prepared to concrete


casting.

Fig. 8. The excitation equipment, (a) an accel-


erometer, (b) an impact hammer. (c) The data
logger and laptop with pulse lab shop software.

modes are negligible. Thus, in case the modes are not close to one 2.4. Log decrement
another, FRF associated with a multi-degree-of-freedom (MDoF) system
or a real structure can be treated as the FRF associated with an SDoF Log decrement is one of the simplest modal analysis methods based
system [46]. on time domain analysis. The theoretical form for decaying response
with linear viscous damping, such as that shown in Fig. 3, is presented

153
H. Allahyari et al. Measurement 114 (2018) 150–161

Fig. 9. A composite deck suspended by springs and ropes.

Exciting points

Fig. 10. Exciting the composite deck by the impact hammer.

Fig. 12. Accelerance FRF of the decks cast with LHC, (a) the modes having maximum
amplitude (b) excited at point 1, (c) excited at point 2.
Fig. 11. The finite elements model.

two amplitudes peaks separated by N cycles, the first at n = 0 (u¨ (t0) )


as: the next at n = N (u¨ (tN ) ). However, precise evaluation of the individual
u¨ = u¨ 0 e−2ξπωn t (8) peaks in digitized measured data demonstrates that the peaks in prac-
tice do not lie on a smooth curve. Due to this fact, Zhu et al. [47]
Traditionally, one uses suggested the use of linear regression to estimate ξ , using the log de-
crement rate of change. Following this methodology, Eq. (9) may be
1 u¨ (t0) ⎞ written in logarithmic form
ξ= ln ⎛ ⎜ ⎟

2nπ ⎝ u¨ (tN ) ⎠ (9)

Which can be obtained from Eq. (8) to estimate ξ using the ratio of any ln(u¨ i ) = ln(u¨ 0)−2πωn ξt = A + Bt (10)

154
H. Allahyari et al. Measurement 114 (2018) 150–161

Fig. 14. Sample FRF showing use of coherence, (a) DLHC (b) DHC.

Fig. 15. Calculation of half-power bandwidth.

Table 2
Fig. 13. Accelerance FRF of the decks cast with HC, (a) the modes having maximum Frequencies estimated from the FRFs.
amplitude (b) excited at point 1, (c) excited at point 2.
Half-power method Circle fit method

where the rate of change B = −2πωn ξ and a least-square fitting can be Deck type Mode 1 Mode 2 Mode 3 Mode 1 Mode 2 Mode 3
used to determine B for a series of measured u¨ (t ) .
DLHC 422.77 587.58 1060.91 422.26 587.45 1060.41
DHC 488.46 731.69 1208.60 488.61 729.82 1207.88
3. Presentation of the experimental program

3.1. Materials and mix proportions relative density (SSD) was 2.67. A commercial lightweight expanded
clay aggregate (Leca) of a single fraction (4.7–12.7 mm) with bulk
Two mixes namely, high strength concrete and lightweight high density of 740 kg/m3 was used in LHC mix as a coarse aggregate. Water
strength concrete were used. The used binders were Type II Portland absorption of 30 min of the Leca was 6%. Based on the water absorp-
cement and Silica fume, respectively, with density of 3150 kg/m3 and tion, Leca was pre-wetted with half of the total water for half an hour so
2000 kg/m3. Natural fine aggregate used was river sand in the pro- that the aggregate was surface saturated before blending with cement
duction of all mixes. The absorption value of the sand used was 3% and and sand. The dosage of superplasticizer (Glenium110) for each mix
its relative density at saturated surface dry (SSD) condition was 2.65 was determined to obtain a slump of 100 ± 20 mm. The mix propor-
with a fineness modulus of 2.82. In HC mix, the coarse aggregate was tions are given in Table 1. Three standard 150 × 300 mm cylinders and
12.5 mm maximum nominal size with 2.5% absorption value and its three standard 100 × 100 × 100 mm cubes were cast from each mix

155
H. Allahyari et al. Measurement 114 (2018) 150–161

Table 3
Damping ratios estimated from the FRFs.

Half-power method Circle fit method Log decrement method

Deck type Mode 1 Mode 2 Mode 3 Mode 1 Mode 2 Mode 3 Mode 1 Mode 2 Mode 3

DLHC 0. 47 0. 40 – 0. 47 0. 39 – 0.448 0.324 0.179


DHC 0. 48 0. 40 – 0.51 0.42 – 0.449 0.295 0.180

Fig. 16. Nyquist plot for mobility FRF of the


decks cast with LHC, (a) excited at point1, (b)
excited at point 2.

Fig. 17. Nyquist plot for mobility FRF of the


decks cast with HC, (a) excited at point1, (b) ex-
cited at point 2.

and cured under water at about 20 °C. These were used to obtain the were inserted in the concrete slab with 10 mm bars (300 Mpa nominal
elastic modulus, Ec, and the compressive strength, fc, of the mixes ac- tensile yield strength) spaced 150 mm on center in the transverse axis
cording to ASTM: C469 [48] and BS EN 12390 [49] respectively. and 16 mm bars (400 Mpa nominal tensile yield strength) spaced
100 mm on center in the longitudinal axis, as shown in Fig. 7. The decks
cured for 28 days. Once cured, the slabs were transferred from the la-
3.2. Structural modeling boratory floor to the testing laboratory using proper supports so as not
to transfer any damage to the deck.
3.2.1. Mechanical and geometrical properties of the composite deck
specimens
In order to investigate the effect of concrete type on dynamic be- 3.3. Experimental setup and test procedure
havior of unfilled steel-concrete composite decks with perfobond shear
connectors, two specimens of unfilled steel-concrete composite decks 3.3.1. Modal test
were prepared for each type of concrete and thus dynamic tests were As shown in Fig. 8, an impact hammer and an accelerometer were
performed on four steel-concrete composite deck with overall dimen- used to excite the Exodermic decks and to record the measured data at
sions of 1000 mm long by 900 mm width, as shown in Fig. 4. The steel the room temperature (25 °C). The weights of the impact hammer and
grid consisted of IPE160 main bars (235 Mpa nominal tensile yield the accelerometer were 230gr and 18gr respectively, whose weight
strength) that were cut into two halves to provide T-shaped sections should not exceed 1/10 wt of plate [50]. A data logger (Fig. 8) and the
spaced 330 mm on center with 25 mm by 8 mm rectangular distribution pulse lab shop software were also used to record the experimental data
bars spaced 102 mm on center, as presented in Fig. 5. One-sided fillet and to import them into the laptop.
welds were created to connect the distribution bars to the main bars. Though the dynamic characteristics of a structure depend upon the
Horizontal Shear transfer mechanism between the steel grid and con- boundary conditions, the influences of the supports were eliminated by
crete slab is assured by 19 mm diameter holes spaced 51 mm on center utilizing two soft springs and ringed ropes (Fig. 9). Such support con-
punched in the web of the T-shaped sections, as shown in Fig. 6, and ditions profit from the benefit of precise indication of structural dy-
friction along the embedded portion (25 mm) of the main bar webs into namic properties when the influence of poorly defined supports is in-
the concrete slab. According to the depth of the shear connectors, the terrupted.
thicknesses of the concrete slabs were selected 94 mm. Reinforcing bars As shown in Fig. 10, five impacts were used on each excited point

156
H. Allahyari et al. Measurement 114 (2018) 150–161

Fig. 20. Variation of the damping ratio and the natural frequency.

Fig. 18. Least squares fit of log decrement, (a) DLHC (b) DHC.

[51] to achieve a high accuracy due to the fact that there were some
noises in the laboratory. Furthermore, in the course of the excitements, Fig. 21. The ratio of Rdf regarding DLHC and DHC.
the accelerometer had been mounted at the point 1; near the nodal line
for the mode 2 illustrated on the FRF graph.

4. Finite element modeling

The primary goal of finite element model updating is to calibrate


and reach an adequate correlation with results derived from measure-
ments on the actual structure. This is attained through matching many
pairs of experimental and analytical vibration modes with respect to the
natural frequencies. As a result, an accurate finite element model is
achievable and would facilitate further studies on the behavior of the
undamaged and damaged decks without carrying out further physical
tests on the actual structure.
As shown in Fig. 11, an FEM conforming with the experimental
specimens was simulated to compare with the experimental results. Fig. 22. Frequencies of three first modes.
A steel-concrete composite deck undergoing the free-free boundary
conditions and boring its own weight was modeled by the FE analysis
package “ABAQUS” [52] in order to obtain all modes and mode shapes.

Fig. 19. Variation of natural frequencies with the maximum response


amplitudes regarding DLHC and DHC.

157
H. Allahyari et al. Measurement 114 (2018) 150–161

location 1, near the accelerometer, while for Figs. 12(c) and 13(c) was
struck at location 2. From these figures, it is plainly visible that the
location of excitation point did not affect the natural frequencies, and
this fact reflects the accuracy of results. Fig. 14 stands for FRF along
with the coherence function, an indication of how well the output is
linked to the input-a measure of the quality of the FRF. A coherence of 1
shows a linear correlation between output and input, while a coherence
of 0 shows no correlation between them [53]. The coherence function
reveals corresponding drops towards 0 at some anti-resonances. This
behavior can be attributed to energy dissipation via hammer excita-
tion/structure interaction, according to Hickey and Worden [54].
Fig. 15 illustrates an expanded segment of the FRF shown in
Fig. 12(a), in terms of mode1. When a damping ratio is derived from an
FRF using half-power bandwidth, an interpolation is imperative be-
Fig. 23. Natural frequency versus damping ratio of mode 1 for DHC, DLHC, and DPC. tween amplitudes at discrete frequency points to realize the particular
1
frequencies related to the two amplitudes that are 71% , (3db = 2 ) , of
Eventually, the results were validated by the experimental results. that at resonance due to the fact that the data exist as discrete points.
The effects of the springs and the ropes used in the experimental Accordingly, the natural frequencies and the damping ratios estimated
program were ignored in the finite elements method and assumed that from the FRFs. The average results are listed in Tables 2 and 3, re-
their effects were quite negligible. Furthermore, the steel parts, the spectively.
main bars, and the distribution bars were modeled as shell sections. The The FRF provides another method to estimate natural frequencies
solid and the truss sections were also used to model the concrete slab and the damping ratios in frequency domain. Though all the Nyquist
and the rebars. plots (receptance, mobility, and acceleration) are like a circle, only the
Subsequently, the shell-to-solid constraints were used between the mobility FRF is a real one. Hence, the values recorded by the accel-
distribution bars and the concrete slab in order to model the composite erometer have to be divided by jω . Figs. 16 and 17 show Nyquist plots
behavior. The rebars were embedded into the concrete slab afterward. derived from FRFs.
The truss, shell, and solid sections were then discretized by 3-node According to the aforementioned relations in Section 2.3, the nat-
quadratic displacement (T3D3), 8-node doubly curved thick shell ural frequencies and the damping ratios were estimated. The average
(S8R), and 20-node quadratic brick (C3D20R) element types respec- results are given in Tables 2 and 3, respectively.
tively. It is anticipated that half-power bandwidth and circle fit methods
suffer from a drawback in the estimation of damping ratios regarding
5. Results and discussion high frequencies due to the fact that, in higher frequencies, modes are
too close. For this reason, the natural frequencies for mode 3 could not
As shown in Figs. 12 and 13, the amplitude values pertaining to be estimated through half-power bandwidth and circle fit methods.
acceleration were drawn versus the vibration frequencies of FRF to Therefore, the third method (Log decrement) was used to estimate the
extract the natural frequency. The peaks on the FRF correspond to the damping ratios in time domain. The advantage of this method is that
desired natural frequencies of the deck; each peak is associated with a the damping ratios derived from a linear relation, and are not suscep-
mode shapes. For Figs. 12(b) and 13(b), the hammer was struck at tible to this fact that modes are how close to one another. Fig. 18

Fig. 24. Some mode shapes which were obtained via finite elements method, (a) DLHC, (b) DHC.

158
H. Allahyari et al. Measurement 114 (2018) 150–161

Table 4
The results of the FEM analysis.

Mode NO. Frequency (Hz) Mode No. Frequency (Hz) Mode No. Frequency (Hz) Mode No. Frequency (Hz)

DLHC 1 0 8 569.65 15 702.85 22 1260.9


2 0.000106 9 585.55 16 709.14 23 1270.6
3 0.000238 10 625.72 17 856.69 24 1274.1
4 0.001116 11 643.75 18 861.22
5 0.002067 12 644.75 19 866.82
6 13.714 13 674.32 20 1091.2
7 429.84 14 686.01 21 1167.3

DHC 1 0.000247 8 623.64 15 767.84 22 1275.3


2 0.000325 9 630.18 16 819.06 23 1309.8
3 0.000342 10 643.50 17 863.15 24 1347.8
4 0.001069 11 650.84 18 864.68
5 0.001885 12 660.42 19 873.87
6 12.508 13 678.66 20 1275
7 496.65 14 687.43 21 1275.2

damping ratios to decrease. The phenomenon that specimens reflected


the property that damping ratio decreases with an increase in natural
frequency may be called damping-frequency dependent rule as pre-
viously reported by Yan et al. [55]. Damping ratios decreased by 60%
from mode 1 up to mode 3, while the natural frequency increased by
150% and 148% regarding DLHC and DHC, respectively.
Fig. 21 illustrates the ratio of damping to frequency (Rdf). According
to the figure, the difference between DLHC and DHC in terms of Rdf is
much lesser for mode 3 in comparison to the first two modes. The figure
reveals that the values of Rdf decrease in higher modes, showing that
lower modes tend to decay much earlier. Thus, the effect of material
property on the damping ratio is much evident in lower modes.
Moreover, Rdf of DLHC is higher than that of DHC, indicating better
serviceability for DLHC.
Fig. 22 represents how frequencies are associated with one. Overall,
the frequencies of DHC are higher than that of DPC [56] and DLHC.
Also the frequencies of DPC are almost near to that of DLHC; however,
the frequencies of DLHC are slightly lower than that of DPC. Clearly,
there is a larger variation in frequencies for modes 2 and 3, particularly
in terms of DHC. As [57] previously reported, natural frequencies are
directly related to modulus of elasticity (ωn ∝ EI ), and according to
Table 1, DLHC has the least modulus of elasticity, justifying why DLHC
has the least frequencies and DHC has the highest frequencies. In the
other words, the higher stiffness, the higher frequency.
Fig. 23 stands for natural frequency versus damping ratio of mode 1
for DHC, DLHC, and DPC. Both DHC and DLHC have a lower damping
ratio in comparison to DPC. With respect to natural frequency, DLHC
has the least frequency while DHC has the highest one. From this figure,
DHC has undesirable serviceability than DPC although could provide
higher stiffness and bearing capacity. From Figs. 21 and 23, it is an-
Fig. 25. Experimental and theoretical results of modal frequencies. ticipated that DLHC and DPC has comparable serviceability, whereas
DLHC has some advantages in comparison to DPC, and could profits
illustrates the peaks of acceleration time histories similar to that of from lower weight and higher bearing capacity.
Fig. 3. They have been obtained, converted to natural logarithms, and Fig. 24 and Table 4 show the results of the FEM analysis in the form
the least square-fit derived using that feature in a spreadsheet for DHC of graphical and numerical, respectively. Structures with free-free
and DLHC. Following that, the damping ratios obtained using Eq. (10). boundary conditions illustrate rigid body motion at low frequencies
The results are listed in Table 3. that are close to zero and do not undergo bending or flexure [58].
Fig. 19 shows the relation between the natural frequencies and the According to the Table 4, the frequency values regarding the first six
maximum response amplitudes. As shown in the figure, the amplitude modes were negligible; as a result, these modes were rigid body modes.
value encounters a downward trend while the frequency is increasing. Obviously, the first three modes had rigid translation motion and, as
The higher the natural frequency, the lower the maximum response shown in Fig. 24, no deformation was seen in the shape of the steel-
amplitude. The degree of reduction in frequency for various types of concrete deck while the stress contour extends from one edge to the
concrete was different. The slope values show that the reduction rate in opposite edge. In the same way, the second three modes had rigid ro-
DHC was higher than DLHC. In the other words, the greater stiffness, tational motion. From the 7th mode to the end, all the modes were
the higher tendency to lose the maximum amplitude. flexural since the frequencies were notable, and the deformations were
Fig. 20 shows the relation between the damping ratio and the nat- large enough. In the case of Table 4, the rows 7, 8, and 20 associated
ural frequency. Clearly, increasing the natural frequency causes the with DLHC, also the rows 7, 8, and 23 associated with DHC were
equivalent to the modes 1, 2, and 3 of the FRF charts respectively.

159
H. Allahyari et al. Measurement 114 (2018) 150–161

Fig. 25 indicates a comparison between frequencies derived from [9] A.L. Ciutina, A. Stratan, Cyclic performances of shear connectors, Compos. Constr.
Steel Concr. (2008) 5–5. 10.1061/41142(396)5.
FRF, Nyquist circle, and FEM. Overall, the natural frequencies derived [10] J. Ahn, S. Kim, Y. Jeong, Shear behaviour of perfobond rib shear connector under
from FRF and Nyquist circle were almost close to one another, and the static and cyclic loadings, Mag. Concr. Res. 60 (2008) 347–357, http://dx.doi.org/
natural frequencies derived from FEM were just over the experimental 10.1680/macr.2007.00046.
[11] J.D.C. Vianna, L.F. Costa-Neves, PCGDS Vellasco, S a L. de Andrade, Structural
natural frequencies. However, in terms of DHC, the natural frequencies behaviour of T-Perfobond shear connectors in composite girders: an experimental
derived from FEM were just lower the natural frequencies. approach, Eng. Struct. 30 (2008) 2381–2391, http://dx.doi.org/10.1016/j.
According to Table 4 and Fig. 25, the maximum difference between engstruct.2008.01.015.
[12] Y.-J. Jeong, H.-Y. Kim, H.-B. Koo, Longitudinal shear resistance of steel–concrete
the experimental frequencies and the numerical results was less than composite slabs with perfobond shear connectors, J. Constr. Steel Res. 65 (2009)
13%, and the minimum difference associated with mode 2 was almost 81–88, http://dx.doi.org/10.1016/j.jcsr.2008.01.031.
1%; this issue could be attributed to the location of the accelerometer in [13] S.-H. Kim, J.-H. Choi, Experimental study on shear connection in unfilled composite
steel grid bridge deck, J. Constr. Steel. Res. 66 (2010) 1339–1344, http://dx.doi.
the vicinity of the nodal line. The mentioned errors stand for a reliable
org/10.1016/j.jcsr.2010.05.008.
accuracy of FEM and, therefore, these models could be comfortable for [14] J.-R. Cho, S.Y. Park, K. Cho, S.T. Kim, B.-S. Kim, Pull-out test and discrete spring
users. model of fibre-reinforced polymer perfobond rib shear connector, Can. J. Civ. Eng.
39 (2012) 1311–1320, http://dx.doi.org/10.1139/cjce-2011-0573.
[15] Y.S. Cho, Non-destructive testing of high strength concrete using spectral analysis of
6. Summary and conclusions surface waves, NDT & E Int. 36 (2003) 229–235.
[16] C.-K. Ma, A.Z. Awang, W. Omar, Flexural ductility design of confined high-strength
In this research, four specimens of steel-concrete composite decks concrete columns: theoretical modelling, Measurement 78 (2016) 42–48.
[17] N.-H. Lee, K.-B. Song, Seismic capability evaluation of the prestressed/reinforced
were fabricated (two DHC and two DLHC). Subsequently, some of the concrete containment, Yonggwang nuclear power plant Units 5 and 6, Nucl. Eng.
modal parameters of them were investigated and compared to those of Des. 192 (1999) 189–203.
DPC. The outstanding results of this study are as follows: [18] M. Wesolowski, E. Barkanov, Air damping influence on dynamic parameters of
laminated composite plates, Measurement 85 (2016) 239–248.
[19] C. Gentile, G. Bernardini, NDT & E International Output-only modal identification
• In higher frequencies, due to the fact that resonances were too close, of a reinforced concrete bridge from radar-based measurements, NDT & E Int. 41
(2008) 544–553, http://dx.doi.org/10.1016/j.ndteint.2008.04.005 (in press).
log decrement method was more reliable than half-power band-
[20] T.-H. Yi, H.-N. Li, Innovative structural health monitoring technologies,
width and circle fit methods.

Measurement 88 (2016) 343–344.
In a given range of frequency, the DHC had a higher ratio of max- [21] T. Türker, A. Bayraktar, Structural safety assessment of bowstring type RC arch
imum amplitude to natural frequency in comparison to DLHC. bridges using ambient vibration testing and finite element model calibration,

• From mode 1 to mode 3, the damping ratios for both DLHC and DHC Measurement 58 (2014) 33–45.
[22] J. Chang, W. Liu, H. Hu, S. Nagarajaiah, Improved independent component analysis
decreased by 60% while the natural frequencies increased by ap- based modal identification of higher damping structures, Measurement 88 (2016)
proximately 150% for both DLHC and DHC. 402–416.

• In higher frequencies, the ratio of damping ratio to natural fre- [23] R. Velmurugan, G. Balaganesan, Modal analysis of pre and post impacted nano
composite laminates, Lat. Am. J. Solids Struct. 8 (2011) 9–26.
quency was almost equal for both DLHC and DHC, whereas this ratio [24] M. Bayat, G.R. Abdollahzade, Analysis of the steel braced frames equipped with
was slightly higher for DLHC in comparison to DHC. ADAS devices under the far field records, Lat. Am. J. Solids Struct. 8 (2011)

• The natural frequencies of DHC were higher than both DLHC and 163–181.
[25] A.V. de A Mello, JGS Da Silva, PCG da S Vellasco, SAL De Andrade, LRO De Lima,
DPC, whereas the damping ratios of DPC were higher than both DHC Modal analysis of orthotropic composite floors slabs with profiled steel decks, Lat.
and DLHC. Am. J. Solids Struct. 5 (2008) 47–73.
• According to Rdf ratios for mode 1, DLHC and DPC had approxi- [26] P. Castellini, G.M. Revel, L. Scalise, Measurement of vibrational modal parameters
using laser pulse excitation techniques, Measurement 35 (2004) 163–179.
mately similar serviceability, whereas DLHC can be more applicable [27] J.F. Wiss, R.A. Parmelee, Human perception of transient vibrations, J Struct. Div.
than DPC due to lower weight. 100 (1974) 773–787.
• According to FEMs, the first and the second three modes were rigid [28] T.M. Murray, Acceptability criterion for occupant-induced floor vibrations, Eng. J.
18 (1981) 62–70.
translational and rotational modes respectively, whereas the other [29] T.M. Murray, D.E. Allen, E.E. Ungar, Floor vibrations due to human activity, 11th
modes were flexural. Steel Des. Guid. Ser. Am. Inst. Steel Constr., Chicago, 1997.
• The errors of the frequencies derived from FEMs ranged from 0.01 to [30] B. Ellingwood, A. Tallin, Structural serviceability: floor vibrations, J. Struct. Eng.
110 (1984) 401–418.
0.13. Therefore, these models have a reliable accuracy, and could be
[31] N. Fukuwa, R. Nishizaka, S. Yagi, K. Tanaka, Y. Tamura, Field measurement of
comfortable for users. damping and natural frequency of an actual steel-framed building over a wide range
• According to damping ratios, the most effective mode was the first of amplitudes, J. Wind Eng. Ind. Aerodyn. 59 (1996) 325–347.
[32] E. El-Dardiry, E. Wahyuni, T. Ji, B.R. Ellis, Improving FE models of a long-span flat
mode with a damping ratio of almost 0.5% for both DHC and DLHC.
concrete floor using natural frequency measurements, Comput. Struct. 80 (2002)
2145–2156.
References [33] A.J.M. Ferreira, G.E. Fasshauer, Analysis of natural frequencies of composite plates
by an RBF-pseudospectral method, Compos. Struct. 79 (2007) 202–210.
[34] Y.K. Ju, S.D. Kim, Y.K. Lee, Technical papers: dynamic characteristics of the new
[1] S. Schubert, D. Gsell, R. Steiger, G. Feltrin, Influence of asphalt pavement on composite floor system, Int. J. Steel Struct. 8 (2008) 347–356.
damping ratio and resonance frequencies of timber bridges, Eng. Struct. 32 (2010) [35] F.A. Gandomkar, M. Danesh, fundamental natural frequency of chromite composite
3122–3129, http://dx.doi.org/10.1016/j.engstruct.2010.05.031. floor, System 9 (2015) 1272–1276.
[2] Å. Bolmsvik, A. Brandt, Damping assessment of light wooden assembly with and [36] L.F. Costa-Neves, J.G.S. da Silva, L.R.O. de Lima, S. Jordão, Multi-storey, multi-bay
without damping material, Eng. Struct. 49 (2013) 434–447, http://dx.doi.org/10. buildings with composite steel-deck floors under human-induced loads: The human
1016/j.engstruct.2012.11.026. comfort issue, Comput. Struct. 136 (2014) 34–46.
[3] H. Tanyildizi, The investigation of microstructure and strength properties of [37] J.G.S. da Silva, S.A.L. de Andrade, E.D.C. Lopes, Parametric modelling of the dy-
lightweight mortar containing mineral admixtures exposed to sulfate attack, namic behaviour of a steel–concrete composite floor, Eng. Struct. 75 (2014)
Measurement 77 (2016) 143–154. 327–339.
[4] E.F. Leonhardt, W. Andra, H.P. Andra, W. Harre, New improved shear connector [38] K. Jarnerö, A. Brandt, A. Olsson, Vibration properties of a timber floor assessed in
with high fatigue strength for composite structures. Neues Vor Verbundm Fur laboratory and during construction, Eng. Struct. 82 (2015) 44–54.
Stahlverbund-Tragwerke Mit Hoher Danerfestigkeit), Beton-Und Stahlbetoubau 12 [39] A. Devin, P.J. Fanning, A. Pavic, Modelling effect of non-structural partitions on
(1987) 325–331. floor modal properties, Eng. Struct. 91 (2015) 58–69.
[5] E.C. Oguejiofort, M.U. Hosaint, Numerical analysis of push-out specimens perfo- [40] Z. Hou, H. Xia, Y. Zhang, Dynamic analysis and shear connector damage identifi-
bond rib connectors, Comput. Struct. 62 (1997). cation of steel-concrete composite beams, Steel Compos. Struct. 13 (2012) 327–341.
[6] C. Higgins, H. Mitchell, Behavior of composite bridge decks with alternative shear [41] B. Singh, B.K. Nanda, Dynamic analysis of damping in layered and welded beams,
connectors, J. Bridg. Eng. 6 (2001) 17–22. Eng. Struct. 48 (2013) 10–20, http://dx.doi.org/10.1016/j.engstruct.2012.09.037.
[7] J. Machacek, J. Studnicka, Perforated shear connectors. Steel, Compos. Struct. 2 [42] X. Li, Q. Liu, S. Pei, L. Song, X. Zhang, Structure-borne noise of railway composite
(2002) 51–66. bridge: numerical simulation and experimental validation, J. Sound Vib. 353 (2015)
[8] H.Y. Kim, Y.J. Jeong, Experimental investigation on behaviour of steel-concrete 378–394, http://dx.doi.org/10.1016/j.jsv.2015.05.030.
composite bridge decks with perfobond ribs, J. Constr. Steel Res. 62 (2006) [43] K. Wei, W. Yuan, N. Bouaanani, Experimental and numerical assessment of the
463–471, http://dx.doi.org/10.1016/j.jcsr.2005.08.010.

160
H. Allahyari et al. Measurement 114 (2018) 150–161

three-dimensional modal dynamic response of bridge pile foundations submerged in 1987.


water, 120921235409005, J. Bridg. Eng. 18 (2012), http://dx.doi.org/10.1061/ [51] I. Mishra, S.K. Sahu, An experimental approach to free vibration response of woven
(ASCE)BE.1943-5592.0000442. fiber composite plates under free-free boundary condition, Int. J. Adv. Technol. Civ.
[44] C. Yang, G. Jin, Z. Liu, X. Wang, X. Miao, Vibration and damping analysis of thick Eng. 1 (2012).
sandwich cylindrical shells with a viscoelastic core under arbitrary boundary con- [52] Version A, 6.12 Documentation Collection, ABAQUS/CAE User’s Man 2013.
ditions, Int. J. Mech. Sci. 92 (2015) 162–177, http://dx.doi.org/10.1016/j.ijmecsci. [53] A.G.A. Rahman, Z.C. Ong, Z. Ismail, Enhancement of coherence functions using
2014.12.003. time signals in modal analysis, Measurement 44 (2011) 2112–2123.
[45] H. Hashim, Z. Ibrahim, H.A. Razak, Dynamic characteristics and model updating of [54] D. Hickey, K. Worden, Comparing the bispectrum, coherence function and Hilbert
damaged slab from ambient vibration measurements, Measurement 46 (2013) transform as methods of nonlinear identification, Proc. ISMA (2010).
1371–1378. [55] L. Yan, C.H. Jenkins, R.L. Pendleton, Polyolefin fiber-reinforced concrete compo-
[46] J.H. Zhi-Fang Fu, Modal analysis, Butterworth-Heinemann, 2001. sites: Part I. damping and frequency characteristics, Cem. Concr. Res. 30 (2000)
[47] G.H. Zhu, M.J. Crocker, M.D. Rao, Data processing and accuracy analysis of 391–401.
damping measurements, J. Acoust. Soc. Am. 85 (1989) 171–177. [56] H. Allahyari, M. Dehestani, M.H.A. Beygi, B.N. Neya, E. Rahmani, Mechanical be-
[48] ASTM C469/C469M-14, ASTM C469/C469M-14, Standard Test Method for Static havior of steel-concrete composite decks with perfobond shear connectors, Steel
Modulus of Elasticity and Poisson’s Ratio of Concrete in Compression 2014. Compos. Struct. 17 (2014) 339–358.
[49] EN BS. 12390-3, Testing hardened concrete, Compressive Strength Test Specimens [57] I.E.J. Henderson, X.Q. Zhu, B. Uy, O. Mirza, Dynamic behaviour of steel–concrete
2009,19. composite beams with different types of shear connectors. Part II: modelling and
[50] M. Senldge, T.R. Llcht, Piezoelectric Accelerometers and Vibration Preamplifiers comparison, Eng. Struct. 103 (2015) 308–317.
Piezoelectric Accelerometer Aild Vibration Preamplfier Hal{dbook. Brüel & Kjær, [58] C. Sujatha, Vibration And Acoustics, Tata McGraw-Hill Education, 2009.

161