Engineering Structures 147 (2017) 652–665

Contents lists available at ScienceDirect

Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct

Vibration of lightweight steel floor systems with occupants: Modelling,

formulation and dynamic properties
Sigong Zhang b, Lei Xu a,b,⇑, Jingwei Qin c
a
Visiting Scholar of Key Laboratory of New Technology for Construction of Cities in Mountain Area of Chongqing University, Chongqing, China
b
Department of Civil and Environmental Engineering, University of Waterloo, Waterloo, ON N2L 3G1, Canada
c
China Electronics Engineering Design Institute, No. 27, Wanshou Road, Haidian District, Beijing 100142, China

a r t i c l e i n f o a b s t r a c t

Article history: Theoretical analyses are presented for investigating the vibration of lightweight steel floor systems with
Received 27 October 2016 human occupants. A damped plate-oscillator model is proposed to obtain the dynamic properties of cou-
Revised 22 May 2017 pled floor-occupant systems. A generalized formulation of the damped plate-oscillator model is pro-
Accepted 4 June 2017
posed. A complex eigenvalue analysis is performed with the use of state-space method and a
validation study is conducted by comparing with other undamped plate-oscillator models. The dynamic
properties obtained from the proposed model are verified by laboratory tests performed on full-scale
Keywords:
lightweight cold-formed steel (CFS) floor systems. The influence of human occupants on the dynamic
Lightweight steel floor systems
Cold-formed steel
properties of lightweight steel floors are investigated in three scenarios: an unoccupied floor, a floor with
Vibration serviceability one standing occupant and a floor with two standing occupants. Four human dynamical models in stand-
Damped plate-oscillator model ing position, two with one degree of freedom (SDOF) and others with two degrees of freedom (2-DOF), are
Human-induced vibration adopted in the proposed plate-oscillator model and the obtained results are compared to the test results.
Human-structure interaction Comparisons are also made between the proposed plate-oscillator model with the integrated 2-DOF
Dynamic properties model for coupled floor-occupant systems. In addition, the need of recalibrating human models for light-
weight floor systems is also discussed.
Ó 2017 Elsevier Ltd. All rights reserved.

1. Introduction lightweight steel floor systems with occupants and predict the
response of such floor systems under human activities.
Over the last several decades, vibration serviceability of floors It is generally recognized that, besides generating loads, human
induced by human occupant activities has become significant in occupants will interact with a structure, and such interaction,
structural design [1], especially for lightweight floor systems [2]. known as human-structure interaction, can be significant if the
As an alternative to lightweight wood construction, lightweight mass of the occupants is comparable to that of the structure [16].
steel floor systems supported by cold-formed steel (CFS) joists pro- For lightweight floors, vibration analysis ought to consider a cou-
vide an efficient and economical structural system. During the past pled system of the floor and occupants because the dynamic prop-
half century, CFS floor systems have been increasingly used in res- erties of the latter may influence the overall response of the system
idential construction and other lightweight framing construction considerably [17]. Significant progress has been made in research-
in North America. Initiated in 1999, multi-phase tests were carried ing human-structure interactions in the floor vibration induced by
out at the University of Waterloo to evaluate the vibration perfor- human activities [18,19]. One widely-known fact is that human
mance of CFS floor systems [3–10]. Although the comprehensive occupants do not act merely as mass on the structure but behave
test results have contributed to better understanding of the perfor- as highly damped dynamical systems (20%–50% damping ratio)
mance of lightweight steel floor systems [11–15], there is still a [20]. Thus, two important issues must be borne in mind. Firstly,
lack of reliable models and adequate design guidelines pertinent human bodies may have a considerable influence on the modal
to the vibration serviceability of lightweight steel floor systems. mass and damping in lightweight floor systems and the dynamic
The objective of this study is, therefore, to propose a damped characteristics therefore change with the location of human walk-
plate-oscillator model for evaluating the dynamic properties of ing [21]. Secondly, the traditional modal analysis where damping is
ignored or assumed to be proportional is not valid [22] because
⇑ Corresponding author at: Department of Civil and Environmental Engineering, floor-occupant systems consist of a lightly damped structure sys-
University of Waterloo, Waterloo, ON N2L 3G1, Canada. tem and human bodies with high damping.
E-mail address: [email protected] (L. Xu).

http://dx.doi.org/10.1016/j.engstruct.2017.06.008
0141-0296/Ó 2017 Elsevier Ltd. All rights reserved.
 S. Zhang et al. / Engineering Structures 147 (2017) 652–665 653

Although human-structure interaction has been investigated for the dynamic response of floors but did not investigate the
extensively and comprehensive understanding has been achieved, response induced by walking, although they recognised that the
most previous studies are based on two-degree-of-freedom (2- use of heel drop impacts to develop design guidelines for light-
DOF) human-structure models [23–26]. Such 2-DOF models were weight floors was questioned by Allen and Rainer [35]. Zhou and
developed to describe coupled vibration of human occupants and Ji [36–38] developed a beam/plate and spring-mass system to rep-
the structure in which the human bodies and the structure were resent a structure occupied by a crowd of people and investigated
simulated as a single-degree-of-freedom (SDOF) model, respec- crowd-structure interaction without considering the damping
tively. Then, the dynamic properties such as natural frequencies associated with crowd and structure. In addition, considerable
and damping ratios could be examined parametrically for a certain research was also conducted to develop the combined vibrational
range of ratios of frequency, mass and damping coefficients of systems for investigation of human-structure interaction in other
SDOF models of human occupants and the structure. The 2-DOF structures such as stadia and footbridges [39].
human-structure models consider only one structural mode based In the present study, a damped plate-oscillator model is pro-
on the rule of superposition of the linear vibration in which the posed to represent lightweight steel floor systems with occupants.
total response can be obtained by summing up the contribution Firstly, the model is validated by other models in [31,34]. Then, the
of each separate mode in modal analysis. However, human occu- dynamic properties obtained from the proposed model are com-
pants may affect all the vibration modes of the structure not just pared with test results. The influence of human occupants is inves-
one. Furthermore, the 2-DOF model is inadequate without taking tigated in three scenarios: an unoccupied floor, a floor with one
into account the spatial variation of human occupants on the struc- stationary occupant and a floor with two stationary occupants.
ture. For instance, the influence of human occupants on floor vibra- Two types of human models are adopted: SDOF and 2-DOF. Several
tion can vary with their locations on the floor. Based on tests of a existing models of a standing human are also examined.
concrete slab occupied by humans in various situations, Sachse
[27] concluded that the location of a human occupant affected 2. Damped plate-oscillator model
the dynamic properties of the test structure and the influence of
the occupant increased with the amplitude of the mode shape at To simplify the presentation, occupants are modeled by SDOF
the occupant’s location. Therefore, it is desirable to develop inte- oscillators and the floor is represented by an orthotropic plate.
grated human-structure models to obtain realistic responses of Then, the coupled floor-occupant system can be simulated by a
structure. damped plate-oscillator model as illustrated in Fig. 1, which is a
Additionally, it is known that the dynamic properties of the rectangular orthotropic plate of constant thickness h connected
human body are strongly related to the intensity of vibration. Thus, to N o linear, damped oscillators at locations of ðni ; gi Þ; i ¼ 1; 2;
the human models used in biomechanics may need to modified . . . ; N o . The dimensions of the plate are 0 6 x 6 a and 0 6 y 6 b.
before being adopted to model human occupants of building and The occupant-induced force, f ðx; y; tÞ, is located at the position of
bridge structures, because the vibration intensities usually encoun- one occupant, and gðtÞ is an external force applied to the oscillator.
tered in such structures are considerably less than those employed
by biomechanics to derive dynamic human models [20,18]. Exist-
2.1. Formulation
ing human models proposed for application in civil engineering
are primarily developed based on the dynamic behaviour of human
Using the dot denoting differentiation with respect to time t,
occupants on a simply-supported beam, one-way slab or a test rig
the governing equation for the orthotropic plate is
under laboratory conditions [16,28,29,18,30]. It is necessary to
recalibrate the parameters of the human models by realistic full- _
r4o wðx; y; tÞ þ cwðx; € y; tÞ ¼ f ðx; y; tÞ
y; tÞ þ qhwðx;
scale test results, and thus to model human occupants on light-
X
Np
weight floor systems to investigate the vibration of such coupled þ fkhi ½zi ðtÞ  wðni ; gi ; tÞ þ chi ½z_ i ðtÞ  wðn
_ i ; gi ; tÞgdðx  ni Þdðy  gi Þ
floor-occupant systems based on the parameters obtained from i¼1
tests of lightweight floors. ð1Þ
Nicholson and Bergman [31] adopted the Green’s function of
the vibrating plate to obtain the natural frequencies and mode where wðx; y; tÞ is the vertical deflection of the plate; c is the viscous
shapes of the undamped plate–oscillator system. The forced damping constant for the plate; q is the mass density; khi ; chi and
response of the combined system is also determined by modal zi ðtÞ are the stiffness, damping constant and displacement of ith
analysis for both proportional damping and general damping. oscillator; d is the Dirac delta function; and r4o is the biharmonic
The same technique was also applied for the vibration analysis of operator for orthotropic plates, which can be expressed as
a class of constrained and/or combined linear dynamical systems
[32]. However, the dynamic properties of the undamped plate- @4 @4 @4
r4o ¼ Dx þ 2H 2 2 þ Dy 4 ð2Þ
oscillator system may not be applicable for the floor-occupant sys- @x 4 @x @y @y
tem. Foschi et al. [33,34] made an early effort to investigate the
in which Dx and Dy are the flexural rigidity of the plate in the x-
combined transient dynamic response of floor systems with occu- 3
pants based on a finite-strip formulation. The floor systems with direction and y-direction, respectively; Dxy ¼ Gxy h =12 is torsional
various complexities commonly applied in construction were mod-
eled by using finite strips combined into T-beam elements, and the
occupants were idealized as damped oscillators. Two human mod-
els were compared: a simple 2-DOF model and a more-detailed
undamped 11-DOF model. Further applications were extended to
develop the design criteria for residential wooden floor systems
and a SDOF human model was proposed [17]. Nevertheless, the
finite-strip formulation might only be applicable for one-way stiff-
ened floor systems without accounting for effects of the transverse
elements such as blocking, bridging and strongbacks. Furthermore,
Foschi et al. [33,34,17] applied the impulse due to heel drop impact Fig. 1. A damped plate-oscillator model.
654 S. Zhang et al. / Engineering Structures 147 (2017) 652–665

rigidity; and H ¼ D1 þ 2Dxy is effective torsional rigidity, where Multiplying both sides of Eq. (10) by W m ðx; yÞ, integrating on
D1 ¼ mx Dy ¼ my Dx is defined in terms of the Poisson’s ratios mx and area of the plate and applying the orthogonality relation yields
my of the plate material, respectively. €n ðtÞ þ 2fn xn M n q_ n ðtÞ þ x2n M n qn ðtÞ
Mn q
The equation of motion for each oscillator (i.e., human
occupant) is X
No X
1
þ 2fhi xhi mhi W n ðni ; gi Þ W j ðni ; gi Þq_ j ðtÞ
mhi €zi ðtÞ þ chi z_ i ðtÞ þ khi zi ðtÞ ¼ chi wðn
_ i ; gi ; tÞ þ khi wðni ; gi ; tÞ þ g i ðtÞ i¼1 j¼1

ð3Þ X
No X
1
þ x2hi mhi W n ðni ; gi Þ W j ðni ; gi Þqj ðtÞ
where mhi ; chi and khi are the mass, damping constant and spring i¼1 j¼1

stiffness of the ith human occupant, respectively. Divide both side X

No

of Eq. (3) by mhi and rewrite Eq. (3) in terms of the circular fre-  2fhi xhi mhi z_ i ðtÞW n ðni ; gi Þ
i¼1
quency xhi and the damping ratio fhi of the ith human occupant: Z Z
X
No a b
€zi ðtÞ þ 2fhi xhi z_ i ðtÞ þ x2hi zi ðtÞ ¼ 2fhi xhi wðn
_ i ; gi ; tÞ  x2hi mhi zi ðtÞW n ðni ; gi Þ ¼ f ðx; y; tÞW n ðx; yÞdxdy
i¼1 0 0
1
þ x2hi wðni ; gi ; tÞ þ g ðtÞ ð4Þ ð11Þ
mhi i
where where Mn is the n-th modal mass, represented as
sffiffiffiffiffiffiffiffi Z a Z b
chi khi Mn ¼ qh W 2n ðx; yÞdxdy ð12Þ
fhi ¼ ; xhi ¼ ð5Þ 0 0
2mhi xhi mhi
fn is the n-th modal damping ratio, expressed by
c
2.2. Methodology fn ¼ : ð13Þ
2qhxn
The plate-oscillator system in Fig. 1 can be treated as a plate Introducing the modal mass ratio cni ¼ mhi =M n , Eq. (11) can be
constrained by attached oscillators. For a ‘constrained’ plate per- rearranged as
forming free vibration, the inertia forces of the concentrated
€n ðtÞ þ 2fn xn q_ n ðtÞ þ x2n qn ðtÞ
q
masses and the restoring forces of the translational springs can
be considered as the external exciting forces for the plate [40]. X
No X
1

Thus, the assumed-mode method or eigenfunction expansion þ 2fhi xhi cni W n ðni ; gi Þ W j ðni ; gi Þq_ j ðtÞ
i¼1 j¼1
(i.e., the mode superposition theory) adopted for the forced vibra-
tion of an ‘unconstrained’ plate (without any oscillator attached) X
No X
1
þ x2hi cni W n ðni ; gi Þ W j ðni ; gi Þqj ðtÞ
may be used to determine the natural frequencies and mode i¼1 j¼1
shapes of the ‘constrained’ plate. Therefore, the vertical displace-
X
No
ment of the plate wðx; y; tÞ can be expressed as  2fhi xhi cni z_ i ðtÞW n ðni ; gi Þ
X
1 i¼1
wðx; y; tÞ ¼ W n ðx; yÞqn ðtÞ ð6Þ X
No
n¼1  x2hi cni zi ðtÞW n ðni ; gi Þ ¼ F n ðtÞ ð14Þ
i¼1
where W n ðx; yÞ is the vibration modes of the ‘unconstrained’ plate
with same boundary conditions and qn ðtÞ is the time varying gener- where
alized coordinate. It can be obtained for W n ðx; yÞ that [41] Z a Z b
1
F n ðtÞ ¼ f ðx; y; tÞW n ðx; yÞdxdy ð15Þ
r 4
o W n ðx; yÞ  x qhW n ðx; yÞ ¼ 0
2
n ð7Þ Mn 0 0

where xn is the circular frequency of the n-th mode of the ‘uncon- Similarly, substituting Eq. (6) into Eq. (4), it gives
strained’ plate with same boundary conditions. If all edges are sim- X
1
ply supported, W n and xn can be determined from €zi ðtÞ þ 2fhi xhi z_ i ðtÞ þ x2hi zi ðtÞ  2fhi xhi W j ðni ; gi Þq_ j ðtÞ
j¼1
W n ðx; yÞ ¼ sin ai x sin bj y ð8aÞ X1
1
Xij  x2hi W j ðni ; gi Þqj ðtÞ ¼ g i ðtÞ ð16Þ
x2n ¼ ð8bÞ j¼1
m hi
qh
where Eqs. (14) and (16) can be solved simultaneously and expressed
in a matrix form as follows:
Xij ¼ Dx a4i þ 2Ha2i b2j þ Dy b4j ð9aÞ
€ þ CU_ þ KU ¼ F
MU ð17Þ
ai ¼ ip=a; bj ¼ jp=b ð9bÞ
where M; C and K are the mass, damping and stiffness matrices,
Substituting Eqs. (6) and (7) into Eq. (1) results in _ U€ and F are the displacement, velocity, accel-
respectively; and U, U;
eration and force vectors, respectively. The expressions of the
X
1 X
1 X
1
x2n qhW n ðx; yÞqn ðtÞ þ c W n ðx; yÞq_ n ðtÞ þ qh W n ðx; yÞq
€n ðtÞ matrices and vectors are presented in A.
n¼1 n¼1 n¼1 Since the high damping of human occupants, the damping
( " #
X
No X
1 matrix cannot be expressed as a linear combination of mass and
¼ f ðx; y; tÞ þ khi zi ðtÞ  W n ðni ; gi Þqn ðtÞ stiffness matrices [22]. Thus, the state-space method is employed
i¼1 n¼1
" #) in this study. Eq. (17) can be transformed into the state-space form
X
1
as [42]
þchi z_ i ðtÞ  W n ðni ; gi Þq_ n ðtÞ dðx  ni Þdðy  gi Þ ð10Þ
n¼1 V_ ¼ AV þ B ð18Þ
 S. Zhang et al. / Engineering Structures 147 (2017) 652–665 655

where the state vector V and its time derivative V_ are given by
  ( )
U U_
V¼ V_ ¼ ð19Þ
U_ 2ðNþNo Þ1 U€
2ðNþN o Þ1

In Eq. (18), the state matrices A and B are, respectively,

expressed as
 
0 I
A¼ 1 1 ð20aÞ
M K M C 2ðNþNo Þ2ðNþNo Þ Fig. 2. Occupant-floor systems: MDOF oscillators for human occupant.
 
0
B¼ ð20bÞ      
M1 FðtÞ 2ðNþNo Þ1 mh1 0 ch1 þ ch2 ch2 kh1 þ kh2 kh2
Mh ¼ ; Ch ¼ ; Kh ¼
0 mh2 ch2 ch2 kh2 kh2
For free vibration, F(t) is the (N þ N o )–dimensional null vector 0.  
_ 1 ; f1 ; tÞ þ kh1 wðn1 ; f1 ; tÞ þ g 1 ðtÞ
ch1 wðn
Thus, Eq. (18) becomes Fh ¼
g 2 ðtÞ
V_ ¼ AV ð21Þ     
€z1 z_1 z1
€¼
Z ; Z_ ¼ ; Z¼
Since Eq. (21) is a set of linear ordinary differential equations €z2 _z2 z2
with constant coefficients, a solution can be assumed as ð29Þ
V ¼ Xekt ð22Þ Eq. (11) turns into
in which k is a scalar constant and X is a constant 2ðN þ No Þ vector. €n ðtÞ þ 2fn xn M n q_ n ðtÞ þ x2n M n qn ðtÞ
Mn q
Then, it can be obtained X 1
þ 2fh1 xh1 mh1 W n ðn1 ; g1 Þ W j ðn1 ; g1 Þq_ j ðtÞ
qn ðtÞ ¼ xn ekt ðn ¼ 1; 2; 3; . . . ; NÞ ð23aÞ j¼1
zi ðtÞ ¼ xi ekt ði ¼ N þ 1; N þ 2; . . . ; N þ No Þ ð23bÞ X
1
þ x2h1 mh1 W n ðn1 ; g1 Þ W j ðn1 ; g1 Þqj ðtÞ
where xn and xi are elements of vector X. Substituting Eq. (22) into j¼1
Eq. (21) and canceling the common factor of ekt on both sides of the  2fh1 xh1 mh1 z_ 1 ðtÞW n ðn1 ; g1 Þ
equation, the following eigenvalue problem can be obtained Z a Z b

AX ¼ kX ð24Þ  x2h1 mh1 z1 ðtÞW n ðn1 ; g1 Þ ¼ f ðx; y; tÞW n ðx; yÞdxdy

0 0

Since the state matrix A is not a symmetric matrix, the eigenval- ð30Þ
ues and eigenvectors of the matrix are complex valued. The solu-
Consequently, the governing equation, Eq. (17), will be revised
tion of Eq. (24) consists of 2ðN þ N o Þ eigenvalues ki (in complex
accordingly. The expressions of matrices and vectors in Eq. (17)
conjugate pairs) and 2ðN þ N o Þ corresponding eigenvectors Xi (also
are provided in A for a plate having N o damped 2-DOF oscillators.
in complex conjugate pairs). Once the state-space eigenvalue prob-
lem is solved, the modal frequencies and damping ratios can be
2.3. Model validation
determined by [43]
Reðki Þ The damped plate-oscillator model proposed herein is exam-
~ i ¼ jki j; ~fi ¼ 
x ð25Þ
jki j ined by the undamped plate-oscillator system developed in [31].
The frequencies are obtained by Eq. (25) and compared with
The ith mode shape of plate are given by results of [31]. The system consists of a simply supported rectangu-
X
N lar isotropic plate (m ¼ 0:3) with a SDOF undamped oscillator
f i ðx; yÞ ¼
W W n ðx; yÞxn ; x1 ; x2 . . . xN 2 Xi ð26Þ attached to the plate. Damping was considered only for forced
n¼1
response in [31]. The properties of the system provided in [31]
Meanwhile, the mode values of oscillators are taken as are nondimensional and they can be converted in terms of the
xj ; ðj ¼ N þ 1; N þ 2; . . . ; N þ N o Þ in Xi for corresponding parameters defined in present study as:
oscillators. b=a ¼ 0:75; n ¼ 0:225a; g ¼ 0:275a;
The dynamic responses of forced vibrations of occupant-floor
systems can be obtained from Eq. (18) numerically by the mo ko a2
2
¼ 1; ¼ 100;
Runge–Kutta method [43] or the Newmark-b method [44]. Then, qha D ð31Þ
sffiffiffiffiffiffiffiffiffiffiffi sffiffiffiffiffiffiffiffiffiffiffi
the acceleration of occupant-floor systems is determined from D D
[45] co ¼ 0:1mo 4
; c ¼ 0:01qh :
  qha qha4
€ ¼ M1 F  CU_  KU
U ð27Þ
where a; b and h is the length, width and thickness of the plate,
The preceding process adopted the model of SDOF oscillator for respectively; q is the density of the plate; c is the damping constant
human occupants. The extension of the process to multiple degrees of the plate; D is the flexural rigidity of the plate; n and g define the
of freedom (MDOF) oscillators for human occupants is straightfor- location of the oscillator on the plate; and mo ; ko and co is the mass,
ward. For instance, if a 2-DOF oscillator is adopted for a human stiffness and damping constant of the oscillator, respectively. The
occupant as that shown in Fig. 2, Eq. (3) becomes circular frequency of the SDOF oscillator can be obtained as
sffiffiffiffiffiffiffi sffiffiffiffiffiffiffiffiffiffiffi
€ þ Ch Z_ þ Kh Z ¼ Fh
Mh Z ð28Þ ko D
xo ¼ ¼ 10 ð32Þ
in which
mo qha4
656 S. Zhang et al. / Engineering Structures 147 (2017) 652–665

The first six frequencies in terms of frequency parameter

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
4
( D=ðqha Þ) of the plate-oscillator model and plate alone are
listed in Table 1. The present results obtained from the proposed
model agree well with the results in [31]. It should be noted that
the present results are calculated by setting the number of vibra-
tion modes in Eq. (6) as 100.

3. Dynamic properties of coupled floor-occupant systems

In order to assess the proposed plate-oscillator model for its

application on vibration of lightweight steel floor systems with
occupants, the predicted dynamic properties are compared with
the test results of the floors with/without human occupants. A
variety of existing human models used in civil engineering applica-
tion are examined in the proposed plate-oscillator model.

3.1. Laboratory tests Fig. 3. The sandbag drop test with two occupants on the floor.

Full-scale lightweight CFS floor systems with different configu-

rations were constructed and tested in the Structures Lab at the ture is the same as that of a person. As it was expected, the addi-
University of Waterloo by the Canadian Cold-Formed Steel tion of furniture does not improve the damping ratio to any great
Research Group (CCFSRG) from 1999 to 2005. The details on the extent. It can be also found that from test results in Table 3 that
test apparatus and procedure were presented in [3–5,46,6,47,8]. the influence of human occupants on dynamic properties of floors
Both static and dynamic tests were carried out on the floor systems is different from that of the furniture. The presence of furniture
to identify the critical parameters that contribute to the control of results in deceasing the natural frequencies, but the frequencies
floor vibration. The test results were then compared with those of floor occupied by one person are increased compared with the
obtained from different design methods. bare floor. The increase of the natural frequencies observed when
In addition to the published test results reported in [3,6,8], the structures are occupied by humans are also reported by others
influence of occupants was also studied when conducting sandbag [16,27,24].
drop on floors without occupants and with one or two occupants
standing at the center of the floor as shown in Fig. 3. The obtained
natural frequencies and damping ratios are listed in Table 2. The 3.2. Human models
acceleration time-history records of floor A in Table 2 and corre-
sponding Fourier spectrum are illustrated in Fig. 4. It can be found The human body is a complex mechanical system and it is often
from Fig. 4a that floor vibration decays faster when human occu- modelled as a lumped parameter vibratory model consisting of an
pants are standing on the floor. The more human occupants, the assemblage of discrete masses, linear springs, and viscous dashpots
faster the rate of decay. Fig. 4b illustrates that due to human occu- as shown in Fig. 5 [34]. The model has a point contact with the
pant’s presence, the peaks of spectra are significantly reduced and floor and maintains that contact as it walking or stay on the floor.
the damping is increased. The damping ratios are evaluated by the Such a model may vary from a simple SDOF model [17] to a more
bandwidth method. Additionally, the control tests were conducted complex multi-degree-of-freedom model (e.g., 15-DOF in [48]), the
for floors with furniture which have same weight as the human choice of which may be determined in terms of the effects of fre-
occupants. The results of control tests are shown in Table 3. quency, activity and posture when the human body is subjected
Table 2 shows that the frequencies are slightly changed when to certain level of vibration. Identifying human properties and
there are occupants standing at the center of the floor whereas responses associated with various vibrations is complex and chal-
the damping ratios more than doubled. However, the damping lenging. Discrepancies in human properties from measurements
ratios are not increased proportionally with the number of the based on different human models related to issues of vibration
occupants. Furthermore, Fig. 4 shows that the rates of dissipation were reported. Human properties such as mechanical impedance,
of the acceleration for the floor with occupants are much faster or the apparent mass, and the resonance frequency were studied
than that without occupants. As it can be observed in Table 3, from a biomechanical perspective [49,50]; whereas modal mass,
the weight of furniture changes the first natural frequency much frequency and damping ratio of human body are needed herein
more than that by the occupant although the weight of the furni- for the investigation of the lightweight steel floor vibration. The

Table 1
Frequency parameter of simply supported plate coupled to a SDOF oscillator.
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Mode 4
Frequency parameter ( D=ðqha Þ)

Uncoupled plate Plate-oscillator model

Present [31] Present [31]
1 27.41556 27.41556 8.09821 8.09799
2 57.02438 57.02437 30.13916 30.13446
3 80.05346 80.05346 60.46412 60.45673
4 106.37240 106.37239 80.83531 80.83303
5 109.66227 109.66227 107.36724 107.36171
6 159.01029 159.01028 111.57198 111.56776
 S. Zhang et al. / Engineering Structures 147 (2017) 652–665 657

Table 2
Influence of human occupants on dynamic properties of CFS floors [46].

Floor No. of Natural Frequencies Damping ratio

ID Occupants f 1 (Hz) f 2 (Hz) f 3 (Hz) f1 (%)

A 0 12.909 16.434 23.575 1.338

1 13.123 16.632 23.85 5.506
2 13.474 17.09 24.811 8.5
B 0 11.917 15.549 24.857 1.422
1 11.765 15.71 25.238 7.95
2 12.054 16.007 25.452 11.444
C 0 10.849 14.526 20.783 1.276
1 10.849 14.786 20.996 8.687
2 10.757 15.289 22.202 12.085

Fig. 4. Acceleration of the floor A with and without occupants induced by sandbag drop.

modal properties of human body obtained by indirect measure- 6 Hz and a damping ratio ranging from 20% to 50%. Shahabpoor
ments were reported in [51,29,23,52]. Detailed reviews on et al. [19] suggested ranges of 1.85–3.5 Hz and 20–50% for a SDOF
dynamic models of human body in civil engineering can be found model of a walking human. Four typical SDOF models of human
in references [18,19]. walking, sitting and standing are presented in Table 4.
For representing the dynamic properties of the human body, the Furthermore, 2-DOF human models are also commonly used.
most common, convenient and simple reasonable model is a SDOF One of most well-known 2-DOF human models for standing posi-
model [20]. Sachse et al. [18] summarized that vertical vibrations tion might be the one proposed by Coermann [53]. This model
of the whole-body of sitting or standing people are dominated by was adopted by the ISO 5982–1981 [54] and employed by Folz
a heavily damped mode with a natural frequency between 4 and and Foschi [34] to predict the vibration response of wood-framed
658 S. Zhang et al. / Engineering Structures 147 (2017) 652–665

Table 3
Influence of furniture and human occupants on dynamic properties of CFS floors [46]

Floor Furniture No. of Natural Frequencies Damping ratio

ID Occupants f 1 (Hz) f 2 (Hz) f 3 (Hz) f1 (%)

D No 0 13.031 16.571 24.078 1.398

No 1 13.046 16.846 24.643 5.825
Yes 0 11.429 16.266 24.323 2.476
Yes 1 11.398 16.327 25.208 6.447
E No 0 13.519 17.975 24.597 1.088
No 1 13.656 18.265 24.841 4.767
Yes 0 11.932 17.334 24.658 2.154
Yes 1 11.551 17.181 24.78 5.153

Table 5
Properties of 2-DOF models of a standing human body subjected to vertical vibrations.

References Mass Frequency (Hz) Damping ratio

[53,54] mh1 ¼ 62mt =75 f h1 ¼ 5:0 fh1 ¼ 37%
mh2 ¼ 13mt =75 f h2 ¼ 12:5 fh2 ¼ 46%
[55] mh1 ¼ 5:1mt =5:6 f h1 ¼ 6:9 fh1 ¼ 25%
mh2 ¼ 0:5mt =5:6 f h2 ¼ 7:6 fh2 ¼ 31%

mt –total mass of human body.

Fig. 5. Dynamical models of a human occupant.

the joists. The test and present results of frequencies of unoccupied

floors subjected to heel-drop impact. Farah [55] reevaluated the floors are listed in Table 6. Good agreements on the results
experimental data of Coermann [53] and provided the properties obtained from the tests and proposed models can be observed.
of 2-DOF models of standing human. These two 2-DOF human Since test results only provide the damping ratio of the first vibra-
models for standing position are tabulated in Table 5. tion shape, present damping ratios for higher vibration modes of
the unoccupied floors are assumed to be 1.5%.
Then, the dynamic properties of the coupled floor-occupant sys-
3.3. Numerical results
tems are evaluated by the proposed damped plate-oscillator model
with adoption of Brownjohn SDOF human model and Falati SDOF
Configuration details of the floor specimens listed in Table 2
and 3 are provided in Appendix B. The weights of human occupants human model, respectively. In this evaluation, the first 100 vibra-
tion modes are adopted in numerical computations. Table 6 lists
are both 80 kg. The natural frequencies and damping ratios are
the test results and the evaluated results for the floors occupied
obtained by the proposed damped plate-oscillator model for differ-
by one person at the floor center. The evaluated results by the fore-
ent human models and compared with the test results. The human
going different human models shows an additional lower fre-
occupants are modelled to be located at the center of the floor even
though two occupants are not exactly standing at the center of the quency and a corresponding very large damping ratio, which are
floor in the tests (Fig. 3). Four human dynamical models in introduced by human SDOF models and denoted as f 0 and f0 ,
standing position, two SDOF human models presented in [23,29] respectively. These additional vibration modes are associated with
and two 2-DOF human models described in [53,55], are adopted presence of human occupants and might be not observed directly
and the corresponding model properties are listed in Table 4 and from the data measured from floor tests [16,24]. For illustration,
5. The models are denoted in present study as Brownjohn SDOF the first and second mode shapes of floor specimen A obtained
model [23], Falati SDOF model [29], Coermann 2-DOF model [53] by the proposed plate-oscillator model with adoption of Brown-
and Farah 2-DOF model [55]. john SDOF human are presented in Fig. 6. The magnitudes of the
Firstly, natural frequencies are determined for unoccupied complex mode shapes obtained from Eq. (26) are normalized with
floors (i.e., no occupants on the floors). Although only two joist- the mode value of human occupant, and the normalized first and
end edges were supported in the tests, the boundary conditions second mode shapes (j W f 1 ðx; yÞj and j W
f 2 ðx; yÞj) of floor specimen
are assumed as simply-supported for all four edges for the sake A with one occupant are plotted in Fig. 6. It can be observed that
of simplicity in this study. Such simplification is reasonable the first and second modes are dominated by the occupant and
because the tested CFS lightweight floors are one-way floors of the floor, respectively. The extra mode introduced by the human
which the transverse stiffness of the floor is contributed only by occupant may not be detected by accelerometers placed on the
20 mm thick OSB subfloor and is much less than the stiffness of floors in tests, therefore, test results of floors with one person in

Table 4
Properties of SDOF models of a human body subjected to vertical vibrations.

Human models References Mass Frequency (Hz) Damping ratio

Sitting [53]⁄ mt 5.0 32%
Standing [23] mt 5.24  0.40 39%  0.05
[29] mt =3 10.43 50%
Walking [52] mt 2.75v3.00 27.5%v30%
⁄
– unit converted by [18].
mt –total mass of human body.
 S. Zhang et al. / Engineering Structures 147 (2017) 652–665 659

Table 6
Comparison between tested and evaluated dynamic properties of floor specimens with/without human occupants.

Floor Occupant Method Frequency (Hz) Damping ratio (%)

f0 f1 f2 f3 f4 f0 f1 f2 f3 f4

A 0 test / 12.909 16.434 23.575 / / 1.338 / / /

present / 13.781 15.829 22.655 34.936 / 1.338 1.500 1.500 1.500
1 test / 13.123 16.632 23.850 / / 5.506 / / /
present-I 5.124 14.101 15.829 22.670 34.936 33.795 8.038 1.500 6.106 1.500
present-II 10.847 13.388 15.829 22.481 34.936 41.157 9.972 1.500 5.815 1.500
B 0 test / 11.917 15.549 24.857 / / 1.422 / / /
present / 11.596 13.319 19.061 29.396 / 1.422 1.500 1.500 1.500
1 test / 11.765 15.710 25.238 / / 7.950 / / /
present-I 5.129 11.830 13.319 19.093 29.396 33.529 7.401 1.500 5.484 1.500
present-II 11.012 11.104 13.319 18.891 29.396 6.999 43.250 1.500 5.325 1.500
C 0 test / 10.849 14.526 20.783 / / 1.276 / /
present / 11.360 13.570 20.565 32.646 / 1.276 1.500 1.500 1.500
1 test / 10.849 14.786 20.996 / / 8.687 / / /
present-I 5.091 11.682 13.570 20.598 32.646 32.104 9.688 1.500 6.172 1.500
present-II 10.509 11.405 13.570 20.381 32.646 8.538 41.114 1.500 5.954 1.500

/–Not available.
Present-I–Results from the proposed damped plate-oscillator model by applying Brownjohn SDOF human model.
Present-II–Results from the proposed damped plate-oscillator model by applying Falati SDOF human model.

frequencies f 1 of the floor A and B with one occupant evaluated

by Falati SDOF human model are less the ones evaluated for the
floors without occupants, which contradicts the test results shown
in Table 6.
Dynamic properties of floors with two occupants are also eval-
uated by the proposed damped plate-oscillator model with adop-
tion of Brownjohn SDOF human model and Falati SDOF human
model, respectively. Both occupants are modelled individually
and the evaluated results are listed in Table 7. Similar to those in
the Table 6, f 01 and f 02 are two additional frequencies introduced
by human SDOF models and their corresponding damping ratios
are f01 and f02 . However, f 02 and f02 are same as the human models.
The comparison in Table 7 illustrates that the effects of human
occupants is overestimated considerably when modelling each
occupant separately for the case that two occupants closely stand-
ing on the floor. As a results, the obtained damping ratios become
extremely large. Since the damping ratios obtained from tests are
not increased proportionally with the number of the occupants,
it may suggest that closely standing human occupants should be
modeled by one SDOF model instead of modelling each occupant
as a SDOF model individually.
Lastly, 2-DOF human models (i.e., Coermann [53] and Farah
[55]) are adopted into the proposed plate-oscillator model to pre-
dict the dynamic properties of floors with one occupant and corre-
sponding results are shown in Table 8. Comparing the results
shown in Table 6 and 8, it can be found that the results obtained
from the 2-DOF human models are not better than those from
Fig. 6. The normalized first and second mode shapes of floor A with one occupant. SDOF models. The 2-DOF models, however, introduce two extra
vibration modes in the floor-occupant systems. These extra modes
will cause some difficulties to recognise the dominant modes of
floor-occupant systems. Thus, it requires more efforts to identify
Table 6 do not contain f 0 and f0 , but they are included in the
the dominant modes from the results obtained using Coermann
evaluated results from the damped plate-oscillator models.
2-DOF model [53] and Farah 2-DOF model [55] for floor B and C.
Furthermore, from the comparison shown in Table 6, frequen-
Therefore,a SDOF model of human is sufficient to obtain dynamic
cies and damping ratios of the second and fourth vibration modes
properties of floor-occupant systems.
of unoccupied floors (i.e., f 2 ; f 4 ; f2 and f4 highlighted in Table 6) are
not affected by the human occupants. This is because the human
occupants were located at the nodal point of the second and fourth 4. Discussion and remarks
vibration modes, which is the center of the floors. It can be con-
cluded that the influence of the human occupants on the floor 4.1. 2-DOF model of human-structure system
vibration are associated with the human location on the floor.
The dynamic properties of a vibration mode may not much affected An integrated 2-DOF model of a human-structure system was
by human occupants if they are standing at the nodal points of the investigated as a reasonable way to simplify the dynamic analysis
mode. Additionally, it is also observed that the first natural of human-structure interaction [24,25]. The 2-DOF model could be
660 S. Zhang et al. / Engineering Structures 147 (2017) 652–665

Table 7
Comparison between tested and evaluated dynamic properties of floor specimens with two human occupants.

Floor Occupant Method Frequency (Hz) Damping ratio (%)

f 01 f 02 f1 f2 f3 f01 f02 f1 f2 f3

A 2 test / / 13.474 17.090 24.811 / / 8.500 / /

present-I 4.964 5.240 14.790 15.829 22.491 29.096 39.000 14.630 1.500 11.130
present-II 10.382 10.430 14.350 15.829 22.081 25.109 50.000 26.334 1.500 11.213
B 2 test / / 12.054 16.007 25.452 / / 11.444 / /
present-I 4.964 5.240 12.336 13.319 19.013 28.581 39.000 13.455 1.500 9.827
present-II 9.909 10.430 12.629 13.319 18.560 13.325 50.000 36.535 1.500 9.299
C 2 test / / 10.757 15.289 22.202 / / 12.085 / /
present-I 4.868 5.240 12.382 13.570 20.485 26.232 39.000 17.312 1.500 11.297
present-II 9.228 10.430 13.334 13.570 20.004 12.400 50.000 37.824 1.500 10.705

/–Not available.
Present-I–Results from the proposed damped plate-oscillator model by applying Brownjohn SDOF human model.
Present-II–Results from the proposed damped plate-oscillator model by applying Falati SDOF human model.

Table 8
Comparison between tested and evaluated dynamic properties of floor specimens with one human occupant.

Floor Occupant Method Frequency (Hz) Damping ratio (%)

f 01 f 02 f1 f2 f3 f01 f02 f1 f2 f3

A 1 test / / 13.123 16.632 23.850 / / 5.506 / /

present-III 4.394 13.955 14.017 15.829 22.699 29.411 52.914 5.565 1.500 4.822
present-IV 5.802 8.450 14.477 15.829 22.959 18.472 32.327 6.768 1.500 5.436
B 1 test / / 11.765 15.710 25.238 / / 7.950 / /
present-III 4.395 11.793 13.319 13.940 19.112 29.271 4.996 1.500 52.945 4.332
present-IV 5.6790 8.454 12.186 13.319 19.333 18.054 32.011 6.860 1.500 5.038
C 1 test / / 10.849 14.786 20.996 / / 8.687 / /
present-III 4.366 11.623 13.570 13.950 20.619 28.391 6.492 1.500 52.992 4.845
present-IV 5.701 8.419 12.156 13.570 20.883 16.900 31.663 8.790 1.500 5.556

/–Not available.
Present-III–Results from the proposed damped plate-oscillator model by applying Coermann 2-DOF human model.
Present-IV–Results from the proposed damped plate-oscillator model by applying Farah 2-DOF human model.

Table 9
Dynamic properties evaluated by proposed plate-oscillator model and 2-DOF model.

Floor Model Frequencies (Hz) Damping ratio (%)

f 1 (Hz) f 2 (Hz) f1 f2

A Plate-oscillator 5.121 14.108 33.796 9.407

2-DOF 5.171 13.965 35.44 9.738
B Plate-oscillator 5.126 11.838 33.526 8.759
2-DOF 5.176 11.737 35.208 9.049
C Plate-oscillator 5.088 11.692 32.105 11.502
2-DOF 5.149 11.56 33.896 11.466

sufficient for qualitative analysis of human-structure interaction to 4.2. Non-existence of the ‘bubble mode’
explain damping increases, additional vibration modes as well as
the insignificant changes of natural frequencies observed on struc- In 1998, Talja and Kullaa [13,14] performed modal testing on
tures due to human occupancy. However, caution is advised when lightweight steel joist floors with use of a 5 kg impact hammer.
employing the 2-DOF human-structure model to evaluate the The person who applied the hammer impact was resting his knees
dynamic properties and responses of floor-occupant systems on a soft mat near the center of the floor. Surprisingly, results from
because the model considers only one structural mode. Alterna- experimental modal analysis illustrated that the first mode of the
tively, the proposed plate-oscillator model can be reduced to a 2- structure was a full sine wave in the vertical direction instead of
DOF human-structure model through the selection of one certain a half-sine wave (i.e., ‘bubble mode’), a mode that could not be
vibration mode using Eq. (6) instead of adding many modes found from the test results. The so-called ‘bubble mode’ was found
together. For instance, Table 9 shows the results evaluated by the only for floor specimens with concrete topping and the associated
proposed plate-oscillator model and the 2-DOF human-structure damping ratios are abnormally high (i.e., around 10%). However,
model. The Brownjohn SDOF human model is adopted to simulate the reasons for not being able to find the ‘bubble mode’ and the
the presence of humans and 100 vibration modes are considered high damping ratios were unclear.
for the damped plate-oscillator model, whereas only first vibration From the analysis of the damped plate-oscillator model, human
mode is taken for the 2-DOF human-structure model. Differences occupants can increase the first frequency slightly and damping
between the results evaluated by the two models can be observed ratio substantially, but have no influence on vibration modes of
in Table 9. when located at the nodal points. In the tests by Talja and Kullaa
 S. Zhang et al. / Engineering Structures 147 (2017) 652–665 661

[13,14], the presence of the test person at the floor center increased structure interaction also conclude that the frequencies of the
the damping ratios and the first natural frequency, but the second human and unoccupied structure are always between those of
frequency was expected to be unchanged because the person was the coupled human-structure system if only considering the first
located at the nodal point of the second mode. If the first two nat- frequency of the unoccupied structure [16]. The relationship can
ural frequencies of unoccupied floors are closely spaced, the pres- be expressed as
ence of a person on the floor would result in the frequency
f 1 < ðf h ; f s1 Þ < f 2 ð33Þ
associated with the half-sine wave mode (i.e., the first frequency
of the unoccupied floor) being greater than that of sine wave mode in which f h and f s1 are the natural frequency of the human and the
(i.e., the second frequency of the unoccupied floor). Consequently, first frequency of the unoccupied structure, respectively; and f 1 and
the first mode from test results was the one with a full sine wave. f 2 are the first and second frequency of coupled human-structure
system, respectively. However, Eq. (33) may not always be true.
The issue of the ‘‘bubble mode” in Section 4.2 is a notable example.
4.3. Nested frequencies
Sachse [27] (page 107–108) also discussed this Eq. (33) based on the
parametrical study of 2-DOF model of a human-structure system.
It is generally accepted that the natural frequencies of the com-
Based on the findings of Sachse [27], Eq. (33) is studied in more
bined plate-oscillator systems are nested among the natural fre-
detail in this paper, using the proposed damped plate-oscillator
quencies of the plate alone [31,34]. Extensive studies of human-
model. Only the first two frequencies, f 1 and f 2 , of the combined
floor-occupant system are considered, and SDOF human models
with two different modal masses, mh , and their two corresponding
Table 10 mass ratios c ¼ mh =M (M is the mass of the floor) are used, as listed
SDOF human models for parametrical study.
in Table 10. Then, the natural frequencies f 1 and f 2 are determined
Model mh c ¼ mh =M fh f h (Hz) for floor A with a human standing at the floor center. The results
A mt =3 0.038 30% or 50% 0.8f s1 v1.1f s1 are plotted in Fig. 7a and 7b. It is found that f 1 or f 2 is within
B mt 0.115 30% or 50% 0.8f s1 v1.1f s1 the range bound by the natural frequencies of the human model
and structure (i.e., f h and the first frequency of structure f s1 ), which
mt –total mass of human body assumed as 80 kg.
can be described as

Fig. 7. Natural frequencies f 1 and f 2 of floor-occupant systems with SDOF human model.
662 S. Zhang et al. / Engineering Structures 147 (2017) 652–665

Table 11 dicted results obtained by the proposed damped plate-oscillator

Brownjohn SDOF human model with different mass mh . model were compared with results of the laboratory tests of CFS
f0 f1 f0 f1 floor systems with and without occupants. From the results
tested / 13.123 / 5.506 obtained from the proposed damped plate-oscillator model and
tests, following conclusions can be drawn:
mh mt =3 5.206 13.849 37.235 4.239
mt =2 5.187 13.899 36.361 5.575
0:6mt 5.175 13.934 35.839 6.363 1. The influence of the human occupants on the floor vibration are
0:7mt 5.163 13.972 35.322 7.141 dependent on the human location on the floor. The dynamic
0:8mt 5.149 14.014 34.809 7.908 properties of a vibration mode will not be unaffected by pres-
0:9mt 5.136 14.059 34.230 8.664
ence of the occupants if the occupants are located at the nodal
mt 5.121 14.108 33.796 9.407
points of the mode.
=–Not available. 2. It may not be necessary to adopt 2-DOF human models to eval-
mt –total mass of human body.
uate dynamic properties of floor-occupant system as a SDOF
model appears to be sufficient.
3. Falati SDOF model [29] may not be appropriate for simulating
f h < f 1 < f 2 < f s1 ð34Þ
occupants on lightweight steel floor systems. Closely standing
or human occupants should be modeled as one SDOF model
instead of modelling each occupant individually.
f h < f 1 < f s1 < f 2 ð35Þ 4. Human occupants will introduce additional vibration modes
In particular, the frequency relationship of Eq. (34) can be but the frequencies of combined floor-occupant systems are
observed in Fig. 7a for the mass ratio c ¼ 0:038, with a human not always nested among the natural frequencies of the unoc-
damping ratio of 50%. Similarly, the frequency relationship of Eq. cupied floors especially for cases with small mass ratios (i.e.,
(35) can be noticed in Fig. 7b for the mass ratio c ¼ 0:115, with a c < 0:1) and high human damping ratio (fh P 50% in this
human damping ratio of 50%. The nested-frequency relationship study).
(i.e., Eq. (33)) occurs in the case of a human damping ratio of 5. 2-DOF human-structure model could be sufficient for qualita-
30%, as shown in both Fig. 7a and 7b. tive analysis of human-structure interaction but it is not recom-
In summary, the issue of Eq. (33) not always being true occurs mended to be employed evaluate the dynamic properties and
for the case with smaller mass ratios (i.e., c < 0:1) and higher response of floor-occupant systems because the model consid-
human damping ratio (fh P 50% in this study). Therefore, it can ers only one structural mode. The damped plate-oscillator
be concluded that human occupants will introduce additional model provides more accurate results.
vibration modes but the frequencies of combined floor-occupant 6. Brownjohn SDOF human model can be adopted for lightweight
systems are not always nested among the natural frequencies of steel floor systems but the modal mass mh may be less than the
the unoccupied floors especially for the case with smaller mass human total mass mt .
ratios (i.e., c < 0:1) and a higher human damping ratio (fh P 50%
in this study).
Acknowledgement

4.4. Human model in standing position

This work is supported by Visiting Scholar Foundation of Key
Laboratory of New Technology for Construction of Cities in Moun-
Through the numerical results in Section 3.3, the Brownjohn
tain Area of Chongqing University and the Natural Science and
SDOF human model [23] for standing position is verified and can
Engineering Research Council of Canada.
be adopted for the application of lightweight steel floor systems.
However, it is observed that the evaluated damping ratios are lar-
ger than those for the tested results. Thus, it is necessary to recal- Appendix A. Matrices and vectors
ibrate the Brownjohn SDOF human model for its application in
lightweight steel floor systems. Table 11 demonstrates the For SDOF oscillators, M is an identity matrix of size
dynamic properties of floor specimen A by applying Brownjohn ðN þ N o Þ  ðN þ N o Þ. C and K can be represented as
2 3
SDOF human model with different modal mass mh . The frequency cð1;1Þ cð1;2Þ . . . cð1;NÞ cð1;Nþ1Þ . . . cð1;NþNo Þ
f 1 and corresponding damping ratio f1 of the combined floor- 6 cð2;1Þ cð2;2Þ . . . cð2;NÞ cð2;Nþ1Þ . . . cð2;NþNo Þ 7
6 7
occupant system increase along with mh . Same observation was 6 7
6 .. .. .. .. .. .. .. 7
6 . . . . . . . 7
also for other floor specimens. Ji [56] proposed the human body 6 7
6 cðN;2Þ . . . cðN;NÞ cðN;Nþ1Þ . . . cðN;NþNo Þ 7
model in a standing position as a continuous system and derived C ¼ 6 cðN;1Þ 7 ðA:1aÞ
6 7
its modal mass and mh ¼ 2mt =3 was suggested. Therefore, when 6 cðNþ1;1Þ cðNþ1;2Þ . . . cðNþ1;NÞ cðNþ1;Nþ1Þ . . . cðNþ1;NþNo Þ 7
6 7
applying Brownjohn SDOF human model, the mass mh may be less 6 .. .. .. .. .. .. .. 7
6 7
than the human total mass mt . 4 . . . . . . . 5
cðNþNo ;1Þ cðNþNo ;2Þ . . . cðNþNo ;NÞ cðNþNo ;Nþ1Þ . . . cðNþNo ;NþNo Þ
2 3
kð1;1Þ kð1;2Þ . . . kð1;NÞ kð1;Nþ1Þ . . . kð1;NþNo Þ
5. Conclusion
6 k kð2;2Þ . . . kð2;NÞ kð2;Nþ1Þ . . . kð2;NþNo Þ 7
6 ð2;1Þ 7
6 7
In present study, a damped plate-oscillator model is proposed 6 .. .. .. .. .. .. .. 7
6 . . . . . . . 7
to investigate the dynamic properties of lightweight steel floors 6 7
6 7
K ¼ 6 kðN;1Þ kðN;2Þ . . . kðN;NÞ kðN;Nþ1Þ . . . kðN;NþNo Þ 7ðA:1bÞ
with the interaction of between the floor systems human occu- 6 7
6 kðNþ1;1Þ kðNþ1;2Þ . . . kðNþ1;NÞ kðNþ1;Nþ1Þ . . . kðNþ1;NþNo Þ 7
pants. The model can also be used to obtain the dynamic response 6 7
6 .. .. .. .. .. .. .. 7
of lightweight floors subjected to external loads such as human 6 7
4 . . . . . . . 5
heel drop and walking. The influence of stationary occupants on
kðNþNo ;1Þ kðNþNo ;2Þ . . . kðNþNo ;NÞ kðNþNo ;Nþ1Þ . . . kðNþNo ;NþNo Þ
dynamic properties of lightweight floors was investigated. The pre-
 S. Zhang et al. / Engineering Structures 147 (2017) 652–665 663

in which where
X
No X
No
cði;iÞ ¼ 2fi xi þ 2fhk xhk cik W 2i ðnk ; gk Þ cði;iÞ ¼ 2fi xi M i þ 2fhk1 xhk1 mhk1 W 2i ðnk ; gk Þ
k¼1 k¼1
X
No X
No
cði;jÞ ¼ 2fhk xhk cik W i ðnk ; gk ÞW j ðnk ; gk Þ cði;jÞ ¼ 2fhk1 xhk1 mhk1 W i ðnk ; gk ÞW j ðnk ; gk Þ
k¼1 k¼1

cði;NþkÞ ¼ 2fhk xhk cik W i ðnk ; gk Þ cði;Nþ2k1Þ ¼ 2fhk1 xhk1 mhk1 W i ðnk ; gk Þ
cðNþk;iÞ ¼ 2fhk xhk W i ðnk ; gk Þ cðNþ2k1;iÞ ¼ 2fhk1 xhk1 mhk1 W i ðnk ; gk Þ
cðNþk;NþkÞ ¼ 2fhk xhk cðNþ2k1;Nþ2k1Þ ¼ 2fhk1 xhk1 mhk1 þ 2fhk2 xhk2 mhk2
X
No cðNþ2k1;Nþ2kÞ ¼ 2fhk2 xhk2 mhk2
kði;iÞ ¼ x2i þ x2hk cik W 2i ðnk ; gk Þ ðA:2Þ
cðNþ2k;Nþ2kÞ ¼ 2fhk2 xhk2 mhk2
k¼1
X
No cðNþ2k;Nþ2k1Þ ¼ 2fhk2 xhk2 mhk2
kði;jÞ ¼ x2hk cik W i ðnk ; gk ÞW j ðnk ; gk Þ X
No
k¼1 kði;iÞ ¼ x2i M i þ x2hk1 mhk1 W 2i ðnk ; gk Þ ðA:5Þ
kði;NþkÞ ¼ x c hk ik W i ðnk ; k Þ
2
g k¼1
X
No
kðNþk;iÞ ¼ x hk W i ðnk ; k Þ
2
g kði;jÞ ¼ x2hk1 mhk1 W i ðnk ; gk ÞW j ðnk ; gk Þ
kðNþk;NþkÞ ¼ x 2
hk
k¼1

i ¼ 1; 2; . . . ; N i–j6N kði;Nþ2k1Þ ¼ x2hk1 mhk1 W i ðnk ; gk Þ

k ¼ 1; 2; . . . ; No k – l 6 No kðNþ2k1;iÞ ¼ x2hk1 mhk1 W i ðnk ; gk Þ

Note that elements of C and K above not specified are zero. The
kðNþ2k1;Nþ2k1Þ ¼ x2hk1 mhk1 þ x2hk2 mhk2
displacement vector U and the force vector F are given by kðNþ2k1;Nþ2kÞ ¼ x2hk2 mhk2
U ¼ ½q1 ðtÞ; q2 ðtÞ; . . . ; qN ðtÞ; z1 ðtÞ; . . . ; zNo ðtÞ kðNþ2k;Nþ2kÞ ¼ x2hk2 mhk2
 
1 1 ðA:3Þ kðNþ2k;Nþ2k1Þ ¼ x2hk2 mhk2
F ¼ F 1 ðtÞ; F 2 ðtÞ; . . . ; F N ðtÞ; g 1 ðtÞ; . . . ; g No ðtÞ
mh1 mhNo ðNþN o Þ
i ¼ 1; 2; . . . ; N i–j6N
k ¼ 1; 2; . . . ; No k – l 6 No
For 2-DOF oscillators, M; C and K are matrices of size
ðN þ 2N o Þ  ðN þ 2N o Þ and can be represented as Elements in C and K which are not specified above are zero. The
2 3 displacement vector U and the force vector F are given by
M1 0 ... 0 0 0 ... 0
6 7 U ¼ ½q1 ðtÞ; q2 ðtÞ; . . . ; qN ðtÞ; z11 ðtÞ; z12 ðtÞ; . . . ; zNo 1 ðtÞ; zNo 2 ðtÞðNþ2No Þ
6 0 M2 ... 0 0 0 ... 0 7
6 7

6 .. .. .. .. .. .. .. .. 7 F ¼ F 1 ðtÞ; F 2 ðtÞ; . . . ; F N ðtÞ; g 11 ðtÞ; g 12 ðtÞ; . . . ; g No 1 ðtÞ; g No 2 ðtÞ ðNþ2No Þ

6 . . . . . . . . 7
6 7
6 . . . MN ... 7 ðA:6Þ
6 0 0 0 0 0 7
6 7
M¼6
6 0 0 ... 0 mh11 0 ... 0 7
7 ðA:4aÞ in which
6 7
6 0 0 ... 0 0 mh12 ... 0 7 Z Z
6 7 a b
6 .. .. .. .. .. .. .. .. 7 F n ðtÞ ¼ f ðx; y; tÞW n ðx; yÞdxdy ðA:7Þ
6 . . . . . . . . 7
6 7 0 0
6 7
4 0 0 ... 0 0 ... mhNo 1 0 5
0 0 ... 0 0 ... 0 mhNo 2 Appendix B. CFS floor construction data
2 3
cð1;1Þ cð1;2Þ ... cð1;NÞ cð1;Nþ1Þ ... cð1;Nþ2No Þ Layout details of floor specimens are illustrated in Fig. 8. Config-
6 c cð2;2Þ ... cð2;NÞ cð2;Nþ1Þ ... cð2;Nþ2No Þ 7
6 ð2;1Þ 7 uration details and structural properties of the floor specimens
6 7
6 .. .. .. .. .. .. .. 7 listed in Table 2 and 3 are provided in Table 12 where designations
6 . . . . . . . 7
6 7 (i.e., fl-6.114-2-6”-1/5-B0-S6-2b) are customized to identify the
6 c 7
C ¼ 6 ðN;1Þ cðN;2Þ ... cðN;NÞ cðN;Nþ1Þ ... cðN;Nþ2No Þ 7ðA:4bÞ
6 7 different floor configurations. For instance, the designation of fl-
6 cðNþ1;1Þ cðNþ1;2Þ . . . cðNþ1;NÞ cðNþ1;Nþ1Þ . . . cðNþ1;Nþ2No Þ 7
6 7 6.114-2-6”-1/5-B0 represents the floor assembly with a joist span
6 .. .. .. .. .. .. .. 7
6 7
4 . . . . . . . 5 length of 6.114 m, only two joist-end edges were supported, solid
cðNþ2No ;1Þ cðNþ2No ;2Þ . . . cðNþ2No ;NÞ cðNþ2No ;Nþ1Þ . . . cðNþ2No ;Nþ2No Þ blocking was a 6” channel section, the blocking was at every sixth
joist spacing, and no restraining beams were used. In addition, B2
2 3
kð1;1Þ kð1;2Þ ... kð1;NÞ kð1;Nþ1Þ ... kð1;Nþ2No Þ indicates restraining beams were used at the edges; S6 represents
6 k kð2;2Þ ... kð2;NÞ kð2;Nþ1Þ ... kð2;Nþ2No Þ 7 screw spacing being 6” at the edges and 12” in the interior area of
6 ð2;1Þ 7
6 7
6 .
. .. .. .. .. .. .. 7 OSB subfloor; 2b describes two rows of strapping at every 1/3 span
6 . . . . . . . 7
6 7 of the joists; Ce is defined for gypsum board ceiling; and g is used
6 7
K ¼ 6 kðN;1Þ kðN;2Þ ... kðN;NÞ kðN;Nþ1Þ ... kðN;Nþ2No Þ 7ðA:4cÞ
6 7 for OSB floor panel glued with CFS joists.
6 kðNþ1;1Þ kðNþ1;2Þ . . . kðNþ1;NÞ kðNþ1;Nþ1Þ . . . kðNþ1;Nþ2No Þ 7
6 7 CFS floor properties can be calculated from equivalent rigidities
6 .. .. .. .. .. .. .. 7
6 7 provided in [57] as
4 . . . . . . . 5
kðNþ2No ;1Þ kðNþ2No ;2Þ . . . kðNþ2No ;NÞ kðNþ2No ;Nþ1Þ . . . kðNþ2No ;Nþ2No Þ
664 S. Zhang et al. / Engineering Structures 147 (2017) 652–665

Fig. 8. A typical CFS floor specimen layout (adapted from [4]).

Table 12
CFS floor configurations.

Floor Floor configuration Span Width Density Dx Dy H

(m) (m) (kg=m2 ) (Nm) (Nm)
00
A fl-6.114–2-6 -1/5-B0-S6-2b 6.114 4.5 25.22 2:655  106 1:694  104 0
00
B fl-6.114–2-6 -1/5-B0-S6-2b-Ce 6.114 4.5 35.62 2:656  106 1:694  104 0
00
C fl-6.754–2-6 -1/5-B0-S6-2b-g 6.754 4.5 25.11 2:655  106 1:533  104 0
00
D fl-6.114–2-6 -1/5-B0-S6-2b-g 6.114 4.5 25.22 2:655  106 1:694  104 0
00
E fl-6.114–2-6 -1/5-B2-S6-2b-g 6.114 4.5 25.22 2:655  106 1:694  104 0

EIeff [8] Xu L, Tangorra F. Experimental investigation of lightweight residential floors

Dx ¼ ðB:1aÞ supported by cold-formed steel C-shape joists. J Constr Steel Res 2007;63
s
(3):422–35.
EIb
Dy ¼ Dsy þ cb ðB:1bÞ [9] Parnell R, Davis B, Xu L. Vibration performance of lightweight cold-formed
a=2 steel floors. J Struct Eng 2009;136(6):645–53.
[10] Xu L. Floor vibration in lightweight cold-formed steel framing. Adv Struct Eng
Gp t3s 2011;14(4):659–72.
H¼ ðB:1cÞ
12 [11] Linehan PW, Kudder RJ, Wiss JF. Dynamic and human response behavior of
cold-formed steel joint residential floor systems. In: Fourth International
Speciality Conference on Cold-Formed Steel Structures, vol. II, St. Louis, USA;
where EIeff is the effective flexural stiffness of a joist with OSB sub-
1978. p. 615–45.
floor which can be calculated based on [58]; s is the joist spacing, [12] Kraus CA. Floor vibration design criterion for cold-formed C-Shaped supported
Dsy is the flexural stiffness of OSB in the direction perpendicular residential floor systems Master’s thesis. Blacksburg, VA, USA: Virginia
Polytechnic Institute and State University; 1997.
with the joists; EIb is the flexural stiffness of blocking; cb ¼ 1=4 is
[13] Talja A, Kullaa J. Vibration tests for lightweight steel joist floors–dynamic
the coefficient for discrete blockings; a is the floor span; Gp is the properties and vibrations due to walking. Finnish R&D conference on steel
shear modulus of the OSB; and t s is the thickness of the OSB. Since structure (Teräsrakenteiden tutkimus-ja kehityspäivät), Lappeenranta,
Finland; 1998.
the shear modulus and the thickness of the OSB are very small, val-
[14] Kullaa J, Talja A. Vibration tests for lightweight steel joist floors–subjective
ues of H in Table 12 are ignored. perceptions of vibrations and comparisons with design criteria, Finnish R&D
conference on steel structure (Teräsrakenteiden tutkimus-ja kehityspäivät),
Lappeenranta, Finland; 1998.
References [15] Rack W, Lange J. Human induced vibrations of lightweight steel floor systems.
In: Proceedings of the eleventh Nordic Steel Construction Conference,
NSCC2009, Malmö, Sweden; 2009. p. 502–09.
[1] Bachmann H, Ammann W. Vibrations in structures: induced by man and
[16] Ellis BR, Ji T. Human–structure interaction in vertical vibrations. Proceedings
machines. IABSE; 1987. vol. 3.
of the Institution of Civil Engineers – Structures and Buildings 1997;122
[2] Smith I, Chui YH. Design of lightweight wooden floors to avoid human
(1):1–9.
discomfort. Can J Civ Eng 1988;15(2):254–62.
[17] Foschi RO, Neumann GA, Yao F, Folz B. Floor vibration due to occupants and
[3] Xu L, Ling Z, Xie W-C, Schuster RM. Dynamic behaviour of floors with cold-
reliability-based design guidelines. Can J Civ Eng 1995;22(3):471–9.
formed steel joists. In: 15th International Specialty Conference on Cold-
[18] Sachse R, Pavic A, Reynolds P. Human-structure dynamic interaction in civil
Formed Steel Structures, St. Louis, Missouri, USA; 2000. p. 377–92.
engineering dynamics: a literature review. Shock Vib Digest 2003;35(1):3–18.
[4] Xu L. Dynamic behaviour of residential floor systems using cold-formed steel
[19] Shahabpoor E, Pavic A, Racic V. Interaction between walking humans and
joists, Phase I, Final Report Prepared for the Canadian Sheet Steel Building
structures in vertical direction: a literature review. Shock Vibr 2016. Article ID
Institute. Waterloo, ON, Canada: University of Waterloo; 2000.
3430285.
[5] Xu L. Dynamic behaviour of residential floor systems using cold-formed steel
[20] Griffin MJ. Handbook of human vibration. Academic Press; 1990.
joists, Phase II, Interim Report Prepared for the Canadian Sheet Steel Building
[21] Smith I, Chui YH. Design of lightweight wooden floors to avoid human
Institute. Waterloo, ON, Canada: University of Waterloo; 2001.
discomfort: Reply. Can J Civ Eng 1989;16(2):202–3.
[6] Tangorra FM, Xu L, Xie WC. Vibration characteristics of lightweight floors using
[22] Ji T. Understanding the interactions between people and structures. Struct Eng
cold-formed steel joist. In: 16th International Specialty Conference on Cold-
2003;81(14):12–3.
Formed Steel Structures, Orlando, Florida, USA; 2002. p. 573–88.
[23] Zheng X, Brownjohn JMW. Modeling and simulation of human-floor system
[7] Xu L. Site investigation of the dynamic behaviour of the iSPAN floor
under vertical vibration. In: SPIE’s 8th Annual International Symposium on
system. Waterloo, ON, Canada: University of Waterloo; 2005.
 S. Zhang et al. / Engineering Structures 147 (2017) 652–665 665

Smart Structures and Materials, International Society for Optics and Photonics; quasi-analytical and numerical combined methods. Int J Numer Methods Eng
2001. p. 513–20. 1997;40(12):2171–93.
[24] Sachse R, Pavic A, Reynolds P. Parametric study of modal properties of damped [41] Leissa AW. Vibration of plates, NASA SP-160. Washington, D.C.: National
two-degree-of-freedom crowd–structure dynamic systems. J Sound Vib Aeronautics and Space Administration; 1969.
2004;274(3):461–80. [42] Balachandran B, Magrab EB. Vibrations. 2nd ed. Cengage Learning; 2009.
[25] Shahabpoor E, Pavic A, Racic V. Using MSD model to simulate human-structure [43] Inman DJ. Vibration with control. John Wiley & Sons Ltd; 2006.
interaction during walking. In: Topics in Dynamics of Civil Structures. [44] Clough RW, Penzien J. Dynamics of structures. 3rd ed. Computers & Structures
Proceedings of the 31st IMAC, A Conference on Structural Dynamics, vol. Inc; 2003.
4. Springer; 2013. p. 357–64. [45] Qin J, Law S, Yang Q, Yang N. Pedestrian–bridge dynamic interaction, including
[26] Zhou D, Han H, Ji T, Xu X. Comparison of two models for human-structure human participation. J Sound Vib 2013;332(4):1107–24.
interaction. Appl Math Model 2016;40(5):3738–48. [46] Liu W. Vibration of floors supported by cold-formed steel joists Master’s
[27] Sachse R. The influences of human occupants on the dynamic properties of thesis. Waterloo, ON, Canada: University of Waterloo; 2001.
slender structures (Ph.D. thesis). University of Sheffield; 2002. [47] Tangorra FM. A design procedure for floor supported by cold-formed steel
[28] Brownjohn JMW. Energy dissipation in one-way slabs with human joists Master’s thesis. Waterloo, ON, Canada: University of Waterloo; 2005.
participation. In: Proceedings of the Asia-Pacific Vibration Conference, [48] Nigam S, Malik M. A study on a vibratory model of a human body. J Biomech
Nanyang Technological University, Singapore, 11–13 December, vol. 1; 1999. Eng 1987;109(2):148–53.
p. 155–160. [49] Matsumoto Y, Griffin M. Dynamic response of the standing human body
[29] Falati S. The contribution of non-structrual components to the overall exposed to vertical vibration: influence of posture and vibration magnitude. J
dynamic behaviour of concrete floor slabs (Ph.D. thesis). University of Sound Vib 1998;212(1):85–107.
Oxford; 1999. [50] Subashi G, Matsumoto Y, Griffin M. Apparent mass and cross-axis apparent
[30] Zhang Q. Models of a standing human body in structural vibration Ph.D. mass of standing subjects during exposure to vertical whole-body vibration. J
thesis. Manchester, UK: University of Manchester; 2013. Sound Vib 2006;293(1):78–95.
[31] Nicholson JW, Bergman LA. Vibration of damped plate-oscillator systems. J Eng [51] Ji T, Ellis R, Beak M. Indirect measurement of human whole-body frequency.
Mech 1986;112(1):14–30. In: The 28th UK Group Meeting on Human Response to Vibration,
[32] Mcfarland DM, Bergman L. Free vibration of constrained plates and plate- Farnborough, UK, September 20–22; 1993.
oscillator systems. In: 30th Structures, Structural Dynamics and Materials [52] Shahabpoor E, Pavic A, Racic V. Identification of mass–spring–damper model
Conference; 1989. p. 1248. of walking humans. Structures, vol. 5. Elsevier; 2016. p. 233–46.
[33] Foschi RO, Gupta A. Reliability of floors under impact vibration. Can J Civ Eng [53] Coermann RR. The mechanical impedance of the human body in sitting and
1987;14(5):683–9. standing position at low frequencies. Human Factors: J Human Factors Ergon
[34] Folz B, Foschi RO. Coupled vibrational response of floor systems with Soc 1962;4(5):227–53.
occupants. J Eng Mech 1991;117(4):872–92. [54] ISO 5982. Vibration and shock – Mechanical driving point impedance of the
[35] Allen DE, Rainer JH. Discussion: design of lightweight wooden floors to avoid human body. 1st ed. Geneva, Switzerland: International Organization for
human discomfort. Can J Civ Eng 1989;16(2). 202–202. Standardization; 1981.
[36] Zhou D, Ji T. Dynamic characteristics of a beam and distributed spring-mass [55] Farah A. Human response: a criterion for the assessment of structural
system. Int J Solids Struct 2006;43(18):5555–69. serviceability Ph.D. thesis. Ontario, Canada: University of Waterloo; 1977.
[37] Zhou D, Ji T. Free vibration of rectangular plates with continuously distributed [56] Ji T, Ellis B. A continuous model for the vertical vibration of the human body in
spring-mass. Int J Solids Struct 2006;43(21):6502–20. a standing position. In: UK Informal Group Meeting on Human Response to
[38] Zhou D, Ji T. Free vibration of rectangular plates with attached discrete sprung Vibration, Silsoe, UK; 1995. p. 1–12.
masses. Shock Vib 2012;19(1):101–18. [57] Zhang S, Xu L. Fundamental frequency of lightweight cold-formed steel floor
[39] Živanovic S. Modelling human actions on lightweight structures: experimental systems. In: Dynamics of Coupled Structures, Volume 4, Proceedings of the
and numerical developments. MATEC Web of Conferences, vol. 24. EDP 34th IMAC, A Conference and Exposition on Structural Dynamics, Springer,
Sciences; 2015. Orlando, FL, USA; 2016. p. 137–45.
[40] Wu J-S, Luo S-S. Free vibration analysis of a rectangular plate carrying any [58] Allen DE, Onysko DM, Murray TM. Minimizing floor vibration, ATC Design
number of point masses and translational springs by using the modified and Guide 1. Redwood City, California: Applied Technology Council; 1999.