Motion-Based Design Approach for a Novel Variable

2 downloads 0 Views 11MB Size Report
Dec 28, 2018 - proposed a novel variable friction cladding connection (VFCC), which enables ... elements and supplemental damping systems to achieve a .... The device consists of two sets of .... the friction element Ff is assumed to be under a constant maximum .... by the first mode when subjected to wind excitations [42].
Motion-Based Design Approach for a Novel Variable Friction Cladding Connection used in Wind Hazard Mitigation Yongqiang Gong1 , Liang Cao2 , Simon Laflamme1,3 , James Ricles2 , Spencer Quiel2 , Douglas Taylor4

Abstract Cladding systems typically serve architectural purposes and protect occupants against the external environment. It is possible to leverage these systems to enhance structural resiliency. A common application is the use of blast resistant panels for enhanced protection against man-made hazards, whereas energy is dissipated through sacrificial elements. However, because these protection systems are passive, their mitigation capabilities are bandwidth limited, therefore targeting single types of hazards. The authors have recently proposed a novel variable friction cladding connection (VFCC), which enables the leveraging of cladding inertia for mitigating blast, wind, and seismic hazards. The variation in the friction force is generated by an actuator applying pressure onto sliding friction plates via a toggle system. Previous work has characterized the dynamic behavior of a VFCC prototype, and established design procedures for blast mitigation applications. Here, work is extended for applications to wind mitigation. An analytical model is developed to characterize the dynamic behavior of the VFCC for wind-induced vibrations. A motion-based design framework is developed to enable an holistic integration of the device within the structural design phase. Numerical simulations are conducted on a 24-story building example to demonstrate the motion-based design methodology. Results show that the semi-active cladding system provides significant reduction in the wind-induced inter-story drift and floor acceleration, therefore demonstrating the promise of the VFCC for field applications. Keywords: Motion-based design, cladding connection, wind mitigation, semi-active control, variable friction, high performance control system

1

1. Introduction

2

Motion-based design (MBD) of civil structures is a design methodology that consists of sizing stiffness

3

elements and supplemental damping systems to achieve a desirable structural performance [1]. However,

4

when multiple excitation inputs are considered either individually or combined, termed multi-hazards, the 1 Department

of Civil, Construction, and Environmental Engineering, Iowa State University, Ames, IA 50011, U.S.A of Civil and Environmental Engineering, Lehigh University, Bethlehem, PA 18015, U.S.A 3 Department of Electrical and Computer Engineering, Iowa State University, Ames, IA 50011, U.S.A 4 Taylor Devices, North Tonawanda, NY 14120, U.S.A 2 Department

Preprint submitted to Journal of Engineering Structures

December 28, 2018

5

MBD approach becomes difficult to implement due to potentially conflicting objectives that may not be

6

simultaneously attained with a traditional structural system. For instance, passive energy dissipation systems

7

have bandwidth-limited mitigation capabilities [2, 3]. A solution is the utilization of high performance control

8

systems (HPCSs) that include active [4, 5, 6], semi-active [7, 8, 9] and hybrid control systems [10, 11, 12].

9

These systems have the particularity of being alterable for adapting to different dynamics, which makes them

10

ideal for multi-hazard mitigation [13, 14].

11

Several types of HPCSs have been proposed of the last decades, with the common objective to provide

12

high mitigation capability on limited power input. Of interest to this paper is the concept of multifunctional

13

cladding systems, where cladding inertia is leveraged to provide vibration mitigation functions. Early work

14

studied blast mitigation capabilities by sacrificing the cladding panel itself. Strategies included double-

15

layer foam cladding [15], sandwich cladding [16, 17], tube-core cladding [18] and metal layer cladding [19].

16

Advanced cladding connections have also been proposed for blast mitigation. For example, Amadio and

17

Bedon [20] developed a viscoelastic spider connector for cable-supported glazing facades. Chen and Hao [21,

18

22] introduced a rotational friction hinge for sandwich panels. Wang et al. [23] experimentally investigated a

19

blast-absorbing cladding connector fabricated from aluminum foam and curved plates. Others have studied

20

wind and seismic mitigation applications. Baird et al. [24] have proposed a U-shape flexural plate dissipator

21

formed by bending a mild steel plate and experimentally investigated its performance at seismic mitigation.

22

Ferrara et al. [25] and Biondini et al. [26] tested bolted friction connectors joining individual cladding

23

elements for energy dissipation. Maneetes et al. [27] studied a spandrel-type precast concrete cladding using

24

supplementary friction devices to establish a lateral force resisting system.

25

It follows that the vast majority, if not all, of the proposed energy dissipation systems leveraging cladding

26

systems are passive strategies. In order to further the applicability to multi-hazard mitigation, the authors

27

have proposed a semi-active cladding connector, termed Variable Friction Cladding Connection (VFCC)

28

[28, 29]. The VFCC is a variable friction device that is engineered to laterally connect cladding elements to

29

the structural system. The variation in the friction force is generated by an actuator applying pressure onto

30

sliding friction plates via a toggle system. Various damping devices based on variable friction mechanism

31

have been studied for structural control applications, including electromechanical- [30, 31], electromagnetic-

32

[32, 33], magnetorheological- [34, 35], hydraulic- [36, 37], and piezoelectric- [38, 39] based technologies. The

33

proposed device is novel by being engineered to leverage cladding panels for vibration mitigation.

34

In previous work, a prototype of the VFCC has been fabricated, experimentally tested, and its dynamic

35

behavior characterized [28]. Afterwards, its passive application for blast energy absorption was studied and

36

a MBD developed [29]. In this paper, the authors extend work to wind-induced vibrations. An analytical

37

model is developed based on cladding-structure transfer functions. It is followed by the development of MBD

38

procedures, which leads to the selection of cladding connection properties, including connection stiffness and

2

39

device damping capacity. While the paper does not consider multi-hazards, results are to be incorporated in

40

future work on multi-hazard applications of the VFCC. Work presented in this paper is different from the

41

investigation on blast loads, because the device is used here in semi-active mode.

42

The paper is organized as follows. Section 2 provides the background on the VFCC device and its

43

dynamics. Section 3 derives a structure-cladding model along with its associated transfer functions to

44

facilitate the MBD procedure. Section 4 develops the MBD procedure used to determine the dynamic

45

parameters of the VFCC. Section 5 demonstrates the VFCC and the MBD procedure through numerical

46

simulations conducted on a 24-story building. Section 6 concludes the paper.

47

2. Variable Friction Cladding Connection

48

The VFCC is a variable friction device engineered to laterally connect cladding elements to the structural

49

system. The device is schematized in Fig. 1 (a) and shown in Fig.1 (b). The device consists of two sets of

50

sliding friction plates upon which a variable pressure is applied by an actuator through toggles. Piezoelectric-

51

based [39] or electromagnetic-based [40] technologies could be good candidates to actuate the device. Figure

52

1 (b) shows a spacer in lieu of the actuator. These spacers were used to emulate actuation during the dynamic

53

characterization process in [28] by maintaining a constant displacement on the toggle. The figure also shows

54

the aluminum frame that was used to mount the device in the testing machine, where the applied force is

55

acting vertically. The blocks are used to prevent the toggles from pushing beyond a vertical alignment.

(a)

(b)

Figure 1: (a) Schematic representation of the VFCC; and (b) annotated picture of a prototype.

3

Figure 2: Example configuration of the VFCC installed in a floor slab (top view) with diagram of forces.

56

Figure 2 illustrates a possible configuration of the device, where it is embedded into a floor slab with the

57

inner plates extended outward to connect to the cladding panel. The actuator force Fa produces an axial

58

force Ft onto the toggles, which in turn generates a normal force N onto the friction plates. A compressive

59

pressure pc is consequently applied and is assumed to be uniformly distributed between both toggles pc =

N

(1)

Ac,max

60

where Ac,max = bp (lp − 2d) is the maximum contact area of the friction plates under the normal pressure, bp

61

and lp are the width and length of the friction plate, respectively, and d is the distance between the toggle

62

and the end of the friction plate, as illustrated in Fig. 2. The effective contact area Ac between friction

63

64

65

66

67

plates varies with their relative displacement y and the generated Coulomb friction force Fc is written    2µc N if 0≤y 1 and Cσ are constants. Table 1 lists parameters of the VFCC prototype characterized in [28].

(a)

(b)

Figure 3: Dynamics of the VFCC under a harmonic excitation of amplitude 13 mm at 0.05 Hz under various levels of actuation capacity: (a) force-displacement loop; and (b) force-velocity loop.

5

Table 1: Parameters of the VFCC prototype [28].

parameter

units

value

lp

mm

165

bp

mm

60

d

mm

45

Cs



1.052



mm

2.185

kN·mm−1

1.147

σ1

N·s·mm

−1

0.200

σ2

N·s·mm−1

0.200

σ0 |Fc0 =0

79

−1

3. Analytical Transfer Functions

80

This section derives the analytical transfer functions of a linear time-invariant structure-cladding system

81

used for the MBD procedure. The equations of motion are first derived, followed by the non-dimensional

82

transfer functions.

83

3.1. Equations of motion

84

Fig. 4 is the diagram of a n-story structure equipped with a cladding system and VFCCs. The primary

85

structure is represented by a lumped-mass shear system and the cladding panel at each floor is modeled as a

86

uniform rigid bar connected to adjacent floors. The lateral cladding connection is represented by a stiffness

87

element kc , a viscous damping element cc0 , and a variable friction element Ff . In this configuration, the

88

VFCC is assumed to provide only lateral resistance. The gravity resistance is considered as decoupled and

89

provided by conventional gravity connections. To develop a mathematically trackable analytical solution,

90

the friction element Ff is assumed to be under a constant maximum capacity (i.e., passive-on configuration)

91

and the corresponding passive friction element is transformed into an equivalent viscous damping element

92

cv [1], yielding a general viscous damping element for the connection cc cc = cc0 +

4Ff πΩˆ x

(9)

93

where Ω is the excitation frequency and x ˆ is the amplitude of structure-cladding displacement due to the

94

dynamic load. Note that this assumption does not hold for the intended semi-active application, and only

95

provides a rough approximation of the friction behavior. The intent is to develop a design procedure that

96

would provide starting points in the sizing of the device. As it will be discussed later, this technique will

97

also require the approximation of the wind load into a harmonic load.

6

(a) Figure 4: Diagram of a n-story structure equipped with a cladding system and VFCCs.

The equations of motion of the linear time-invariant n-story structure-cladding system with equivalent

98

99

viscous damping are given by Ms x ¨s + Cs x˙ s + Ks xs = Ec Cc x˙ c + Ec Kc xc

(10a)

Mc x ¨c + Cc x˙ c + Kc xc = Ep p − Mc ET ¨s cx

(10b)

100

where xs ∈ Rn×1 and xc ∈ R2n×1 are the displacement vectors of the primary structure and of the cladding

101

relative to the primary structure, respectively, p ∈ Rn×1 is the external load input vector, Ep ∈ R2n×n and

102

Ec ∈ Rn×2n are the external loading and cladding location matrices, respectively, and Ms ∈ Rn×n , Cs ∈

103

Rn×n , Ks ∈ Rn×n are the mass, damping, and stiffness matrices of the primary structure and Mc ∈ R2n×2n ,

104

Cc ∈ R2n×2n , Kc ∈ R2n×2n are the mass, damping, and stiffness matrices associated with the cladding

105

system.

106

The mass of the cladding panel at each floor mc is taken to be identical at the preliminary design stage for

107

simplicity. The cladding connection, which includes the lateral stiffness kc and the equivalent viscous damping

108

cc , is assumed to be identical at each floor for simplicity of the design process and field implementation. The

109

mass, damping, and stiffness matrices of the cladding elements, Mc ∈ R2n×2n , Cc ∈ R2n×2n , Kc ∈ R2n×2n ,

110

are element diagonal matrices composed with the cladding mass, damping and stiffness matrices mc , cc and

111

kc :

7

 mc =

mc  3 mc 6



mc 6  mc 3

 cc cc =  0

0 cc

 

 kc and kc =  0

(a)

0 kc

 

(b)

Figure 5: Diagram of the ith cladding element connected to adjacent floors: (a) two DOFs representation; and (b) SDOF representation.

112

113

Fig. 5(a) illustrates the two degrees-of-freedom (DOFs) representation of the ith cladding element connected to adjacent floors. Its equation of motion is written mc x ¨c,i + cc x˙ c,i + kc xc,i = ep pi − mc x ¨s,i xs,i ]T and xc,i = [xc,2i−1

(11)

xc,2i ]T are the displacement vectors of the primary structure

114

where xs,i = [xs,i−1

115

and of the ith cladding element relative to its connected floors, respectively, pi is the external load acting on

116

the ith cladding, and ep = [ 12

117

is connected to the ground and to the first floor, as illustrated in Fig. 4 and xs,0 corresponds to ground

118

motion, which is taken as xs,0 = 0 for wind loading.

1 T 2]

is the load location vector. Note that the first cladding panel (i.e., i = 1)

119

To obtain a mathematically trackable displacement vector xc,i , its dynamics is expressed in terms of the

120

modal vectors Φcr and modal coordinates qcr,i (r = 1, 2) for DOFs r = 1, 2, and is assumed to be governed

121

by the first mode xc,i ≈ Φc1 qc1,i

(12) φc,12 ]T = [1

1]T , obtained by

122

where qc1,i is the modal coordinate of the first mode and Φc1 = [φc,11

123

solving the algebraic equation [kc − ωc2 mc ]Φc1 = 0 [41]. Pre-multiplying Eq. 11 by ΦT c1 and substituting

124

xc,i from Eq.12 yields a linear time-invariant single DOF (SDOF) expression governing the dynamics of the

125

cladding element mce q¨c1,i + cce q˙c1,i + kce qc1,i = pi − mce

126

x ¨s,i−1 + x ¨s,i 2

with the equivalent cladding mass mce , damping cce , and stiffness kce 8

(13)

mce = ΦT c1 mc Φc1 = mc

cce = ΦT c1 cc Φc1 = 2cc

and kce = ΦT c1 kc Φc1 = 2kc

127

where the displacement qc1,i ≈ xc,2i−1 ≈ xc,2i . The SDOF schematic representation is illustrated in Fig.

128

5(b).

129

Similarly, the displacement vector of the primary structure xs is written in terms of its modal vectors

130

Φsk and modal coordinates qsk (k = 1, 2, ..., n), with the assumption that its dynamic behavior is governed

131

by the first mode when subjected to wind excitations [42] xs =

n X

Φsk qsk ≈ Φs1 qs1

(14)

k=1

φs,12

...

φs,1n ]T with φs,1n normalized to unity. Pre-multiplying Eq. 10a by ΦT s1

132

where Φs1 = [φs,11

133

and substituting xs from Eq.14 yields an uncoupled equation of motion for qs1 , mse q¨s1 + cse q˙s1 + kse qs1 = ΦT ˙ c + ΦT s1 Ec Cc x s1 Ec Kc xc

134

(15)

with the equivalent structural mass mse , damping cse , and stiffness kse equal to mse = ΦT s1 Ms Φs1

cse = ΦT s1 Cs Φs1

and kse = ΦT s1 Ks Φs1

135

where qs1 ≈ xsn is the displacement of the SDOF structure and the damping matrix Cs is taken proportional

136

to the stiffness matrix Ks .

137

138

It follows that the governing equations of motion of the linear time-invariant structure-cladding system are reduced to mse q¨s1 + cse q˙s1 + kse qs1 =

n X

αi (kce qc1,i + cce q˙c1,i )

(16a)

i=1

mce q¨c1,i + cce q˙c1,i + kce qc1,i = pi − αi mce q¨s1 139

where αi = 12 (φs1,i−1 + φs1,i ) for i = 2, 3, ..., n and α1 = 12 φs1,1 .

140

3.2. Non-dimensional transfer functions

(16b)

141

To derive the non-dimensional transfer functions, the wind load pi is approximated as a harmonic exci-

142

tation pi (t) = pbi ejΩt . The corresponding steady state response of the structure-cladding system is written

143

as [1] qs1 = qbs1 ej(Ωt+δs ) qc1,i = qbc1,i ej(Ωt+δci )

144

(17)

where the hat denotes an amplitude, j is the imaginary unit, Ω the excitation frequency, and δ the phase.

9

145

Substituting Eq. 17 into Eqs. 16a and 16b yields n X  kse − mse Ω2 + jcse Ω qbs1 ejδs = (kce + jcce Ω) αi qbc1,i ejδci

(18a)

i=1

 kce − mce Ω2 + jcce Ω qbc1,i ejδci = pbi + αi mce Ω2 qbs1 ejδs 146

Multiplying Eq. 18b by αi and summing from i = 1 to n yields n X

αi qbc1,i ejδci =

i=1

147

148

where pbe =

n P i=1

pbe + Γmce Ω2 qbs1 ejδs kce − mce Ω2 + jcce Ω

αi pbi is the amplitude of the equivalent load and Γ =

n P i=1

(19)

αi2 . Substituting Eq. 19 into Eqs.

18a and rearranging Eq.18b yields

kce + jcce Ω qbs1 ejδs = 2 pbe (kse − mse Ω + jcse Ω) (kce − mce Ω2 + jcce Ω) − (kce + jcce Ω) Γmce Ω2 qbc1,i ejδci pbi + αi mce Ω2 qbs1 ejδs = pbe pbe (kce − mce Ω2 + jcce Ω) 149

(20)

Defining the mass ratio µ, tuning frequency ratio f , and excitation frequency ratio λ as µ=

150

(18b)

mce mse

f=

ωc ωs

and λ =

Ω ωs

with r ωs =

kse mse

cse ξs = 2mse ωs

r ωc =

kce mce

and ξc =

cce 2mce ωc

151

yields non-dimensional transfer functions Hs (λ) and Hc,i (λ) that represent the dynamic amplification of the

152

displacement of the SDOF structure and of the displacement of the ith cladding relative to its connected

153

floor, respectively, where qbs1 ejδs f 2 + j2ξc f λ = 2 2 2 2 pbe /kse (1 − λ )(f − λ ) − 4ξc ξs f λ − µΓf 2 λ2 + j [2ξc f λ(1 − (1 + µΓ)λ2 ) + 2ξs λ(f 2 − λ2 )] qbc1,i ejδci pbi /b pe + αi µλ2 Hs (λ) = Hc,i (λ) = 2 pbe /kse µ(f − λ2 ) + j2ξc µf λ Hs (λ) =

154

155

An additional transfer function Ha (λ) representing the dynamic amplification of the structural acceleration is given by Ha (λ) =

156

(21)

b as1 ejδs = −λ2 Hs (λ) pbe /mse

where the amplitude of the acceleration b as1 = Ω2 qbs1 .

10

(22)

157

4. Motion-Based Design Procedure

158

This section describes the MBD procedure associated with the semi-active cladding connection used for

159

wind hazard mitigation. This procedure is conducted assuming a passive behavior of the device to provide

160

preliminary design values, including a preliminary value for the VFCC’s friction capacity. To optimize the

161

design parameters, one would be required to conduct simulations with semi-active control rules. Such a

162

design strategy is commonly used in the design of semi-active or hybrid control systems in order to provide

163

a target damping capacity [7, 43, 44].

164

The procedure, diagrammed in Fig. 6, starts by quantifying the design wind load (e.g., return period).

165

After the allowable structural acceleration ap and inter-story drift ratio ∆p are defined, as well as the

166

allowable structure-cladding spacing lc , both based on the design wind load, the dynamic parameters of

167

the cladding connection consisting of stiffness kce and equivalent viscous damping ratio ξc , are selected.

168

Subsequently, three non-dimensional analytical solutions Ra , Rs , and Rc associated with the maximum

169

structural acceleration amax , the maximum drift ratio ∆max , and the maximum structure-cladding spacing

170

lmax , respectively, are used to verify the satisfaction of the performance metrics. If the structural motion

171

is unsatisfactory, the process is iterated by either re-designing dynamic parameters kce and ξc (option 1) or

172

updating the allowable structure-cladding spacing lc (option 2). These MBD steps are described in what

173

follows.

Figure 6: Motion-based design procedure.

11

174

175

4.1. Wind load The time-varying wind load acting on cladding panels at height z is given by [45] Pd (t) =

1 ρCd AVd (t)2 2

(23)

176

where Cd is the drag coefficient, ρ is the air density, A is the cladding area exposed to the wind pressure,

177

and Vd (t) is the time-varying wind speed [45] Vd (t) = V + v(t)

(24)

178

where v(t) is a zero mean fluctuation of the wind flow Vd (t) around its mean wind speed V . The design

179

value for V is computed from a 3-second wind gust speed V0 that can be obtained from wind hazard maps

180

[45, 46] V = 0.13V0

v∗ ln(z/z∗ ) v∗0

(25)

181

where z∗ is the surface roughness length of the building’s terrain, and v∗ and v∗0 are the shear velocities

182

of wind flow of the building and open terrain, respectively. Example values for the ratio v∗ /v∗0 are listed

183

in Table 2. Substituting Eq. 24 into Eq. 23, the wind load is expressed as the sum of a steady wind force

184

component P and a fluctuating wind force component p(t) [45] Pd (t) =

  1 ρCd A V 2 + v 2 (t) + ρCd AV v(t) 2

(26)

≈ P + p(t) 185

where v 2 (t) may be neglected for high-rising buildings or, as done in this paper, approximated into a steady

186

component using its mean value v 2 (t) [45]. The mean square value of the wind fluctuation v 2 (t) can be

187

written [45]  v 2 (t) = βv∗ = β

0.4V ln(z/z∗ )

2 (27)

188

where the constant β depends on surface roughness length at the building terrain, also listed in Table 2. The

189

wind fluctuation v(t) at height z can be modeled as a stochastic process characterized by a two-sided power

190

spectral density function (PSD) at excitation frequency Ω [47], where Sv (z, Ω) =

191

192

1 200 2 z h Ωz i−5/3 v∗ 1 + 50 2 2π V 2πV

(28)

The fluctuating wind forces can be modeled as a zero-mean Gaussian stationary and spatiotemporal field that is characterized by its cross-spectral density function Sp (zi , zl , Ω) [48] p Sp (zi , zl , Ω) = (ρCd Ai Vi )(ρCd Al Vl ) Sv (zi , Ω)Sv (zl , Ω)Coh(zi , zl , Ω)

12

(29)

193

where subscripts i and l refer to different heights, and the coherence function Coh is given by [46] h 10Ω|z − z | i i l Coh(zi , zl , Ω) = exp − π(Vi + Vl )

(30)

Table 2: Terrain exposure constants[45]

194

ocean

open terrain

suburb

city center

downtown

z∗ (m)

0.005

0.07

0.30

1.00

2.50

v∗ /v∗0

0.83

1.00

1.15

1.33

1.46

β

6.5

6.0

5.25

4.85

4.00

4.2. Performance objectives

195

Motion criteria associated with wind-induced vibrations are linked to the average return period of the

196

wind excitation under consideration (e.g., 1, 10, 50 and 475 years). The drift ratio ∆p can be selected on

197

the basis of structural damage minimization. Typical values are within the range 1/750 ≤ ∆p ≤ 1/250

198

corresponding to a serviceability limit state [49].

199

The acceleration criterion ap is typically associated with building serviceability and its performance

200

threshold is left to the designer. Here, the selection of ap is conducted based on Refs. [50, 51] to ensure

201

occupancy comfort. The acceleration threshold ap (m·s−2 ) is function of the average return period Q (yr)

202

and the fundamental frequency of the structure ωs (Hz) p ap = (0.68 + 0.2 ln Q) 2 ln(ωs T )e−3.65−0.41 ln ωs

(31)

203

where T is the observation time of the wind event. A typical value for T is T = 3600 s (1 hr) [50]. Fig. 7

204

plots different values for ap as a function of various ωs and Q.

Figure 7: Acceleration criterion ap as a function of ωs and Q.

13

205

Conventional cladding panels are connected to the structural system using vertical bearing connections

206

to support gravity loads and push-pull (tieback) connections to resist lateral loads [52]. A common push-

207

pull connection is a flexible threaded rod designed to either slide inside a slot (sliding connection) or to

208

bend (flexing connection) [53]. A minimum cladding-structure spacing lmin is required for the installation

209

and maintenance of the lateral connections, and this value can be as much as 15 cm [54]. Here, allowable

210

structure-cladding spacing lc is selected such that lc ≥ lmin .

211

4.3. Connection design

212

A first step in the design of the cladding connection is the selection of the dynamic parameters. This is

213

done through initially selecting connection parameters and obtaining the peak building responses based on

214

three non-dimensional analytical solutions Ra , Rs and Rc , yielding amax , ∆max , and lmax , respectively, to

215

be compared with the performance objectives. These three analytical solutions are derived in what follows,

216

and their solutions verified in section 5.4.

217

4.3.1. Maximum structural response

218

219

Consider the response of a structure exposed to wind where the wind velocity fluctuation is taken as a stochastic stationary Gaussian process. The structural response q(t) is given by [55] q(t) = q¯ + qs (t)

(32)

220

where qs (t) is the zero mean fluctuation of structural displacement q(t) around its mean value q¯. The mean

221

static displacement q¯ can be obtained by applying the equivalent mean wind load component Pe to the

222

structure [56], with Pe =

n X

αi Pi

(33)

i=1 223

yielding q¯ = Pe /kse . Following Davenport’s method [57], the expected maximum displacement qmax over

224

observation time T and the corresponding peak response factor η are given by qmax = q¯ + ησq η=

p

2 ln(ωs T ) + p

0.5772 2 ln(ωs T )

(34)

225

where σq is the standard deviation of qs (t). The maximum structural acceleration amax = ησa [55], where

226

σa is the standard deviation of the wind-induced structural acceleration as (t).

227

The mean square values of the structural displacement and acceleration, σq2 and σa2 , are computed by

228

integrating the PSD based on the product of the fluctuating wind load component and the transfer functions

229

of the structure-cladding model [51]. For a lightly damped structure, the response power spectrum is dom-

230

inated by the contribution of the excitation power spectrum around the natural frequency of the structure 14

231

[58, 59]. It follows that a constant power spectrum of the excitation at the first natural frequency of the

232

structure Spe (ωs ) is used to compute σq2 and σa2 [58]. σq2 = σa2 =

233

ωs 2 kse

ωs m2se

+∞

Z

Spe (Ω)|Hs (λ)|2 dλ ≈

−∞ Z +∞

Spe (Ω)|Ha (λ)|2 dλ ≈

−∞

ωs Spe (ωs ) 2 kse

ωs Sp (ωs ) m2se e

where the PSD of the equivalent fluctuating wind force Spe (Ω) =

+∞

Z

|Hs (λ)|2 dλ

−∞ Z +∞

|Ha (λ)|2 dλ

(35a) (35b)

−∞

n P n P

αi αl Sp (zi , zl , Ω) [51]. A solution for

i=1 l=1 234

the integration of transfer functions,Js and Ja , can be obtained using the integral formula from Gradshteyn

235

and Ryzhik [60] Z

+∞

Js = −∞ Z +∞

Ja =

|Hs (λ)|2 dλ = π

b20 /a0 (a2 a3 − a1 a4 ) − a3 b21 a1 (a2 a3 − a1 a4 ) − a0 a23

(36a)

|Ha (λ)|2 dλ = π

a1 b22 + b23 /a4 (a0 a3 − a1 a2 ) a1 (a2 a3 − a1 a4 ) − a0 a23

(36b)

−∞ 236

where a0 = f 2 , a1 = 2ξc f + 2ξs f 2 , a2 = −4ξc ξs f − [1 + (1 + µΓ)f 2 ], a3 = −2ξc f (1 + µΓ) − 2ξs , a4 = 1,

237

b0 = f 2 , b1 = 2ξc f , b2 = f 2 , and b3 = 2ξc f . Two non-dimensional analytical solutions Ra and Rs are used

238

to compute the maximum acceleration and the maximum interstory drift ratio of the structure, respectively, p η ωs Spe (ωs )Ja amax Ra = = (37a) Pe /mse Pe p η ωs Spe (ωs )Js ∆max =1+ (37b) Rs = φ −φ Pe maxi | s,1i s,1i−1 |Pe /kse hi

239

φ −φ where the maximum inter-story drift ratio ∆max = maxi s,1i his,1i−1 qmax based on the assumption that the

240

structure vibrates at its first modal shape and hi is the inter-story height of the ith floor.

241

4.3.2. Structure-cladding spacing

242

243

The maximum relative displacement between the cladding and the primary structure lmax is estimated using the mean and fluctuating wind force components lmax = maxi |li + qbc1,i |

(38)

244

where li = Pi /kce is the displacement of the ith cladding relative to that of the connected floor due to the mean

245

static wind force component, and qbc1,i is the maximum relative displacement resulting from the fluctuating

246

wind force component p(t). Unlike for the computation of Ra and Rs , the cladding response analysis using

247

PSD of a stochastic wind load excitation could require numerical simulations unless a simplification for

248

the term pbi /b pe is conducted in Hc,i (λ) (Eq. 21). Instead, the forcing fluctuation of the wind flow pi (t) is

249

directly simplified as a harmonic process to obtain a mathematically trackable solution for qbc1,i . Although

250

numerical solutions are more exact, the mathematically trackable solutions enables a quick estimation of the 15

251

structure-cladding spacing and therefore facilitates a selection of design parameters during the structural

252

design phase. The quality of the analytical solution will be discussed in Section 5.4. The approximate

253

harmonic wind velocity fluctuation vi (t) is then written vi (t) = vbi sin(ωt)

254

where vbi is the amplitude of the harmonic excitation. The mean square value of vi (t) is written [61] 1 T0 →∞ T0

vi2 (t) = lim 255

256

0

vbi2 2

(40) √

2βv∗ and the fluctuating

wind load component is taken as p 2βv∗ ρCd Ai Vi sin(ωt)

(41)

giving the amplitude of the equivalent fluctuating wind force component pbe n X i=1

259

2

[b vi sin(ωt)] dt =

Combining Eqs. 27 and 40 yields an amplitude for the wind fluctuation vbi =

pbe = 258

T0

Z

pi (t) = pbi sin(ωt) = 257

(39)

αi pbi =

n X

αi

p

2βv∗ ρCd Ai Vi

(42)

i=1

It follows that the maximum cladding-structure displacement qbc1,i attributed to the fluctuating wind loads is written qbc1,i =

pbe b Hc,i kse

(43)

260

b c,i = max|Hc,i (λ)| represents the maximum value of the amplitude of the transfer function Hc,i (λ) where H

261

over the excitation frequency ratio λ. The non-dimensional analytical solution Rc is used to compute the

262

maximum structure-cladding spacing Rc =

263

P b c,i | maxi |Pi /kce + pbe /kse H pbe b lmax i = = maxi 2 + Hc,i Pe /kse Pe /kse µf Pe Pe

(44)

4.3.3. Dynamic parameters for cladding connection

264

The dynamic parameters for the cladding connection are selected on the basis of all three analytical

265

solutions Ra , Rs , and Rc . First, the structure-cladding displacement ln (i.e., ln ≤ lc ) due to the mean static

266

wind force component Pn at the top floor is assigned, yielding kce = Pn /ln

267

Second, an initial stiffness value kce is selected, and the tuning frequency ratio f =

(45) q

kce µkse

obtained.

268

Third, the damping ratio of the connection ξc is selected through a minimization of the structural acceleration,

269

because acceleration is mostly related to serviceability under wind hazards. Fourth, using Eq. 35b, an

16

270

estimated value for ξc is obtained by setting ∂Ja /∂ξc = 0. For simplicity, structural damping is taken as

271

ξs = 0 and Ja (Eq. 36b) reduces to Ja =

272

π µΓ



f3 + 2ξc f 2ξc



and applying ∂Ja /∂ξc = 0 yields ξc = 0.5f

(46)

273

where note that structural damping is not considered to be negligible (ξs 6= 0) in the computation of the

274

peak building responses. The resulting peak structural responses are then compared against the performance

275

objectives. If amax ≤ ap , ∆max ≤ ∆p , and lmax ≤ lc , the design is completed. Otherwise, an additional

276

iteration is required, where the design parameters kce , ξc , and/or lc are altered until the process is completed.

277

The friction damping capacity Fcp at each connection is then obtained using Eq. 9 with the VFCC arbitrarily

278

designed for a harmonic excitation acting on the first natural frequency of the structure ωs and the amplitude

279

of structure-cladding displacement due to dynamic load taken as x ˆ = lmax − ln , Fcp =

1 πmce ωc ωs (ξc − ξc0 )(lmax − ln ) 4

(47)

cc0 2mce ωc .

280

with ξc0 =

281

5. Numerical Simulations

282

Numerical simulations are conducted on a 24-story office occupancy building located in Los Angeles, CA

283

[62]. The office tower is a steel moment-resisting frame structure with six bays in the North-South direction

284

and eight bays in the East-West direction. Each bay is 9 m wide and each floor is 3.9 m high, except for

285

the first floor that is 4.5 m high. The building is simulated as a lumped-mass shear system in the East-West

286

direction. The dynamic properties of the primary structure are listed in Table 3. The structural damping

287

ratio is assumed to ξs = 1%. The cladding is taken as concrete panels with 30% window opening area and

288

the thicknesses of the concrete and glass panels are 15 cm and 0.6 cm, respectively [53]. The densities of

289

the concrete and glass are taken as 2400 kg/m3 and 2800 kg/m3 , respectively, giving a cladding mass of

290

5.51 × 104 kg at each floor, except at the first floor where its mass is 6.36 × 104 kg.

17

Table 3: Dynamic properties of building.

floor

291

mass

stiffness

damping

mass

stiffness

damping

3

(10 kg)

(kN/m)

(kN·s/m)

12

2011

622953

9879

1993

11

2011

668616

10603

177431

2814

10

2011

713378

11313

2011

224212

3556

9

2011

761873

12082

20

2011

271092

4299

8

2011

818123

12974

19

2011

315938

5010

7

2011

913068

14480

18

2011

359101

5695

6

2011

102413

16241

17

2011

398817

6325

5

2011

111175

17631

16

2011

436525

6923

4

2011

119869

19010

15

2011

473719

7513

3

2011

131956

20927

14

2011

512098

8121

2

2011

156541

24826

13

2011

560090

8882

1

2041

172255

27318

3

(10 kg)

(kN/m)

(kN·s/m)

24

1423

67355

1068

23

2011

125646

22

2011

21

floor

The equation of motion for the 24-story building has the form M¨ x + Cx˙ + Kx = Ep + Ef F

(48)

292

where x ∈ R72×1 is the displacement vector, p ∈ R24×1 is the wind force input vector, F ∈ R48×1 is the

293

control input vector, E ∈ R72×24 and Ef ∈ R72×48 are the wind loading and control input location matrices,

294

respectively, and M ∈ R72×72 , C ∈ R72×72 , K ∈ R72×72 are the mass, damping and stiffness matrices of the

295

building, respectively.

296

The state-space model of Eq. 48 for simulation is then given by ˙ = AX + Bp p + Bf F X

297

where X = [x

(49)

x] ˙ T ∈ R144×1 is the state vector and the constant coefficient matrices are defined as follows   0 I  A= (50) −M−1 K −M−1 C 144×144





0  Bp =  M−1

(51)

144×24





0  Bf =  M−1 Ef

144×48

18

(52)

298

The numerical algorithm follows the discrete form of the Duhamel integral [1]: X(t + ∆t ) = eA∆t X(t) + A−1 (eA∆t − I)[Bp p(t) + Bf F(t)]

(53)

299

where ∆t = 0.001 s is the simulation time interval and I ∈ R144×144 is the identity matrix. This discrete

300

state-space linear formulation has been used in literature to simulate the dynamic response of a linear

301

structural system with nonlinear damping devices [63, 64]. Of interest to this paper, the authors used the

302

same method in previous work to simulate a different variable friction device [7]. Three simulations cases

303

are conducted to investigate the performance of the VFCC device, namely the semi-active, passive-on and

304

uncontrolled cases. The semi-active scheme is the VFCC used with a full-state feedback control rule based

305

on a linear quadratic regulator (LQR) computing the required control force Freq for each device Freq = −Gf X

306

(54)

where Gf ∈ R48×144 is the control gain matrix, tuned to minimize a performance objective index JLQR Z 1 ∞ T JLQR = (X Rx X + FT Rf F)dt (55) 2 0

307

where Rx ∈ R144×144 is the regulatory weight matrix and Rf ∈ R48×48 is the actuation weight matrix. Note

308

that the design and optimization of the controller is out-of-the-scope of this work. In the later numerical

309

simulation, the regulatory and actuation weight matrices Rx and Rf are pre-tuned as Rx = diag[I24×24

0.5I10×10

10I10×10

15I28×28

50I15×15

200I5×5

1000I4×4

I48×48 ]

310

and Rf = 10−10 I48×48 . The performance of the semi-active VFCC is compared against that of a passive-on

311

friction case where the VFCC is used under a constant maximum capacity, and of an uncontrolled case with

312

a conventional stiffness connection. In the uncontrolled case, a set of conventional stiffness connectors are

313

used to laterally connect the cladding elements to the structural system with 24 tie-back connectors at the

314

top and 24 bearing connectors at the bottom of cladding panels for each floor. The lateral stiffnesses of each

315

tie-back connector and bearing connector are taken as 39 kN/mm and 2335 kN/mm, respectively, based on

316

Ref. [53]. The stiffness element of the lateral connection for the lumped model is taken as the sum of the 24

317

connectors.

318

5.1. Verification of SDOF simplification for cladding system

319

In the derivation of the transfer functions in Section 3, an assumption was made that a linear time-

320

invariant 2DOF cladding system could be reduced to a linear time-invariant SDOF representation. In what

321

follows, the quality of this assumption is verified on a linear and time-invariant 4DOF representation of the

322

building. To create a realistic system, the 4DOF is created by lumping the 24 structural mass elements into

323

two floors of identical masses and the 24 cladding mass elements into a single cladding mass, as illustrated 19

324

in Fig. 8(a). The dynamic properties of the 4DOF system are listed in Table 8. The dynamic parameters

325

at each floor are taken as identical and such that the 4DOF’s first structural frequency is equal to the first

326

natural frequency of the building (0.2 Hz), with the structural damping ratio set at ξs = 1%. Note that this

327

4DOF system is only used to verify the SDOF simplification for a 2DOF cladding system and it does not

328

necessarily represent the dynamics of the 24-story building system. Following the methodology discussed in

329

Section 3, the 4DOF system is reduced into its equivalent 2DOF representation, illustrated in Fig.8(b) with

330

its dynamic properties listed in Table 4. The amplitude of the equivalent harmonic loading pˆe acting on the

331

2DOF system is taken as pˆe =

332

φs,11 + φs,12 pˆ 2φs,12

where pˆ is the amplitude of the harmonic load acting on the 4DOF system.

(a)

(b)

Figure 8: Representations of the building for verifications of the assumptions: (a) 4DOF model; and (b) 2DOF model.

Table 4: Dynamic properties of the 4DOF system and its equivalent 2DOF system.

4DOF

2DOF

parameter

value

unit

ms

23848

103 kg

ks

98595

kN/m

cs

1569

kN·s/m

mc

1332

103 kg

mse

32958

103 kg

kse

52045

kN/m

cse

828

kN·s/m

mce

1332

103 kg

20

333

The transfer functions Hs (λ), Ha (λ) and Hc (λ) for the equivalent 2DOF system are written f 2 + j2ξc f λ (1 − λ2 )(f 2 − λ2 ) − 4ξc ξs f λ2 − µΓs f 2 λ2 + j [2ξc f λ(1 − (1 + µΓs )λ2 ) + 2ξs λ(f 2 − λ2 )] 1 1 − λ2 + j2ξs λ Hc (λ) = √ 2 2 2 2 µ Γs (1 − λ )(f − λ ) − 4ξc ξs f λ − µΓs f 2 λ2 + j [2ξc f λ(1 − (1 + µΓs )λ2 ) + 2ξs λ(f 2 − λ2 )]

Hs (λ) =

(56)

Ha (λ) = −λ2 Hs (λ) φs,11 +φs,12 2 ) . 2φs,12

The verification of the model reduction assumption is conducted through the

334

where Γs = (

335

comparison of transfer functions derived from the equivalent 2DOF system with those computed numerically

336

on the 4DOF system. In the numerical simulation, each friction damping element is simulated as passive

337

Coulomb friction force Fc of capacity Fc =

338

element of damping ratio ξv under harmonic loading. The initial viscous damping ratio of the lateral

339

connection is taken as ξc0 =

340

Hc (λ).

√ cc0 2 kce mce

π 4 µf λξv |Hc (λ)|,

which is equivalent to a linear viscous damping

= 1% and |Hc (λ)| denotes the magnitude of the transfer function

341

The magnitudes of transfer functions are plotted in Figs. 9 to 11, with tuning frequency ratio f = 3 which

342

exhibits a typical behavior, and two representative damping cases ξv = 10% and ξv = 150% to investigate the

343

effect of the connection’s damping. Note that ξv = 150% represents an overdamped case; it is investigated

344

because the design procedure is likely to yield overdamped strategies based on Eq. 46. In general, the

345

analytical solutions (2DOF) can track the numerical solution (4DOF). There is a disagreement between

346

both solutions for Hs and Ha that occurs around the excitation frequency ratio λ = 2.5, in particular under

347

lower damping (ξv = 10%). The reason is that the second mode of the primary structure, which corresponds

348

to the excitation frequency ratio λ = 2.5, is not part of the analytical solution that is based on an SDOF

349

representation. For Hc , there is a good agreement between the analytical and numerical solutions at low

350

damping (ξv = 10%), but a significant disagreement for the overdamped connection (ξv = 150%), observable

351

in Fig. 11(b). This is likely attributed to the discontinuous motion from the stick-slip phenomenon that is

352

amplified with large friction force.

21

(a)

(b)

Figure 9: Transfer function Hs at f = 3: (a) ξv = 10%; and (b) ξv = 150%.

(a)

(b)

Figure 10: Transfer function Ha at f = 3: (a) ξv = 10%; and (b) ξv = 150%.

22

(a)

(b)

Figure 11: Transfer function Hc at f = 3: (a) ξv = 10%; and (b) ξv = 150%.

353

In order to further investigate the effect of dynamic parameters, the analytical solutions of the transfer

354

functions (Eq. 56) are used to conduct a parametric study of the mass ratio µ (Fig. 12), tuning frequency

355

ratio f (Fig. 13), and damping ratio ξc (Fig. 14, showing ξce ). A study of the transfer functions shows that

356

adding mass to the cladding (i.e., increasing µ) improves the mitigation performance for both the inter-story

357

displacement (Hs ), absolute acceleration (Ha ), and structure-cladding displacement (Hc ). Variation in the

358

cladding mass can generally be attained through using different materials (e.g. glass, masonry or precast

359

concrete). A typical cladding-structural system mass ratio ranges from 0.01 to 0.1 [65, 66]. Augmenting

360

the connection’s stiffness (i.e., increasing f ) has the converse effect on Hs and Ha , yet still decreases Hc .

361

Also, using a more flexible connection leads to a behavior analogous to that of a tuned mass damper,

362

observable under f = 1.5 and f = 1.8. However, this added performance comes at the cost of a much

363

larger structure-cladding displacement which will unlikely meet the design target. A similar performance is

364

observed by decreasing ξc , although one must be careful in evaluating the behavior under high damping given

365

the lower accuracy of the transfer functions to represent structural behavior. Once again, higher mitigation

366

(i.e., lowering ξc ) comes at the cost of higher structure-cladding displacement that may not be possible to

367

achieved. It follows that the VFCC is expected to be used under low stiffness and high damping.

23

(a)

(b)

(c) Figure 12: Plot of transfer functions vs various mass ratios µ at f = 0.9 and ξc = 20%: (a) Hs ; (b) Ha ; and (c) Hc .

24

(a)

(b)

(c) Figure 13: Plot of transfer functions vs various tuning frequency ratios f at µ = 0.05 and ξc = 10%: (a) Hs ; (b) Ha ; and (c) Hc .

25

(a)

(b)

(c) Figure 14: Plot of transfer functions vs various connection damping ratios ξc at µ = 0.05 and f = 0.8: (a) Hs ; (b) Ha ; and (c) Hc .

368

5.2. Demonstration of MBD procedure

369

The proposed MBD procedure is demonstrated on the 24-story office tower building. A basic design

370

wind speed with a return period of Q = 50 years is selected with a 3-second wind gust speed V0 at reference

371

height 10 m of 38 m/s based on the wind hazard map in ASCE 7-10 (2010) [67]. The cladding connection

372

parameters are designed based on the MBD non-dimensional analytical solutions Ra , Rs and Rc .

373

Wind Hazard Quantification

374

The wind loading parameters are determined based on the building location’s terrain with the ratio of

375

shear velocity of wind flow v∗ /v∗0 = 1.15, surface roughness length z∗ = 0.3 m and its corresponding value

376

β = 5.25, and the drag coefficient Cd = 1.4 based on building cross-section [45]. The cladding area exposed

377

to wind pressure at each floor is A = 210 m2 , except at the first floor A = 243 m2 . Using Eq. 25 to Eq. 27,

26

378

the mean static wind load at the top floor is obtained as Pn = 1.93 × 105 N and the equivalent mean static

379

wind load is calculated as Pe = 1.82 × 106 N using Eq. 33. The design parameters are listed in Table 5.

380

Performance Objectives

381

The performance objectives are then quantified. Using the average return period Q = 50 years and the

382

first nature frequency of the structure 0.2 Hz, the acceptable peak acceleration for occupancy comfort is

383

calculated as ap = 27 mg based on Eq. 31. The allowable lateral drift ∆p = 1/250 and a design value of the

384

allowable structure-cladding spacing lc is set to 0.5 m for the preliminary design, corresponding to half the

385

allowable distance (1 m) reported in Ref. [66]. The performance objectives are listed in Table 5.

386

Connection Design

387

The equivalent mass and stiffness of the primary structure is mse = 1.48×107 kg and kse = 2.36×107 N/m,

388

yielding the mass ratio between cladding and the primary structure µ = mce /mse = 0.37%. The structure-

389

390

cladding displacement ln due to the mean static wind force is first set to 0.3 m, yielding the equivalent q kce = 2.7, and lateral connection stiffness kce = Pn /ln = 6.43 × 105 N/m, a tuning frequency ratio f = µk se

391

a damping ratio ξc = f /2 = 1.35. However, these values give a maximum structure-cladding displacement

392

lmax = 0.52 m based on Rc = 6.73 (Eq. 44) that exceeds the allowable value lc = 0.5 m. An iteration is

393

conducted by decreasing the ln value to 0.28 m, yielding an equivalent connection stiffness kce = 6.89 × 105

394

N/m, a tuning frequency ratio f = 2.8, and a damping ratio ξc = f /2 = 1.40. This non-dimensional

395

analytical solutions Ra , Rs and Rc are 2.166, 3.060, and 6.278, respectively. The corresponding maximum

396

structural acceleration and the maximum inter-story drift ratio are amax = Pe /mse Ra = 27 mg 6 ap and

397

∆max = Pe /kse Rs = 1/320 6 ∆p , respectively, which meets the performance objectives. The total structure-

398

cladding spacing at each floor is within the allowable spacing with the maximum value lmax = 0.48 m

399

occurring at the top floor. The friction capacity of the cladding connection at each floor is designed for

400

Fcp = 54.2 kN (Eq. 47). The dynamic parameters are listed in Table 5.

27

Table 5: Cladding connection design parameters.

parameter

variable

wind load

motion criteria

cladding connection

401

402

403

value

units

note

Pn

193

kN

Eqs. 25 - 27

Pe

1820

kN

Eq. 33

ap

27

mg

Eq. 31

∆p

1/250





lc

0.5

m



Ra

2.166



Eq. 37a

Rs

3.060



Eq. 37b

Rc

6.278



Eq. 44

kce

689

kN·m−1

Eq. 45

Fcp

54.2

kN

Eq. 47

5.3. Simulated wind model The time series of wind fluctuation vi (t) is simulated as a multivariate stochastic process with crossspectral density matrix S(Ω) with its element expressed as [47]   Sv (zi , Ω) if i = l Sil (Ω) = p   Sv (zi , Ω)Sv (zl , Ω)Coh(zi , zl , Ω) if i 6= l

(57)

404

with the two-sided PSD Sv (zi , Ω) in Eq. 28 and the coherence function Coh(zi , zl , Ω) from Eq.30. To simulate

405

the stochastic process, the power density matrix S(Ω) is first decomposed into the following product [68]: S(Ω) = Λ(Ω)Λ

406

Λ11 (Ω)

   Λ21 (Ω) Λ(Ω) =   ..  .  Λn1 (Ω)

408

(Ω)

(58)

where subscript asterisk denotes the complex conjugate and Λ(Ω) is a lower triangular matrix 

407

T∗

0

...

0



Λ22 (Ω) .. .

... .. .

0 .. .

      

Λn2 (Ω) . . .

(59)

Λnn (Ω)

Once the matrix S(Ω) is decomposed, the stochastic process of wind fluctuation vi (t) is generated by the following series [68] vi (t) = 2

NΩ i X X

√   |Λiq (Ωqτ )| ∆Ω cos Ωqτ t − θiq (Ωqτ ) + δqτ

q=1 τ =1

28

(60)

409

where δqτ is a random phase uniformly distributed between 0 and 2π; and the phase θiq (Ωqτ ) is given by θiq (Ωqτ ) = tan−1

410

n Im[Λ (Ω )] o iq qτ Re[Λiq (Ωqτ )]

(61)

The double-indexing frequency Ωqτ is defined as: Ωqτ = (τ −

i−q )∆Ω i

(62)

411

with the frequency step ∆Ω = Ωu /NΩ , where Ωu is an upper cutoff frequency, taken as Ωu = 20π, and

412

the total number of frequency points NΩ = 213 [69]. Lastly, the simulated wind load Pd,i (t) acting on the

413

cladding panels at the ith floor is generated Pd,i (t) =

1 2 ρCd Ai [Vi + vi (t)] 2

(63)

414

where Vi is the mean wind speed at building height zi obtained using Eq. 25 and the air density is taken

415

as ρ = 1.225 kg/m3 . A typical 10-minute duration wind excitation at the top floor is plotted in Fig. 15.

416

Fig. 16(a) shows a comparison of the power spectral density function of the wind fluctuation at the top

417

floor between the simulated wind data and the analytical model (Eq. 28). The turbulence intensity of the

418

generated wind velocity fluctuation is verified against the analytical model, illustrated in Fig. 16(b). These

419

comparisons show a good match between the analytical and computational wind model.

(a)

(b)

Figure 15: Typical realization of a wind hazard time series over a 10-minute duration (a) wind speed and (b) wind load.

29

(a)

(b)

Figure 16: Comparison of the analytical wind model and the simulated wind data (a) power spectral density function and (b) turbulence intensity.

420

5.4. Simulation results

421

The maximum response profiles of the building is plotted in Fig. 17, showing the maximum inter-story

422

displacement (Fig. 17(a)), the maximum absolute acceleration (Fig. 17(b)), and the maximum structure-

423

cladding displacement (Fig. 17(c)). Results show that the semi-active controlled VFCC significantly reduces

424

the structural response compared with the passive-on case and the uncontrolled case. The maximum reduc-

425

tion of the inter-story displacement and acceleration reached 16.4 % and 31.4 %, respectively. In particular,

426

the semi-active VFCC brought the acceleration under the limit threshold, while the passive-on and uncon-

427

trolled cases were unsatisfactory. In all cases, the structure-cladding spacing remains under the threshold

428

value of 0.5 m, but the additional mitigation performance provided by the semi-active scheme does come at

429

the cost of higher structure-cladding displacement.

30

(a)

(b)

(c) Figure 17: Maximum building response profiles: (a) inter-story displacement; (b) absolute acceleration; and (c) claddingstructure displacement (uncontrolled case not shown).

430

It can also be observed in Fig. 17 that the passive design indicated that the passive-on strategy was within

431

the acceptable acceleration bounds. This discrepancy is mainly attributed to the non-negligible errors in

432

the analytical solutions arising from the assumptions made to reach a mathematically trackable solution,

433

for instance the negligence of the higher structural modes, the simplification of the wind spectral density

434

function, and the modeling of the friction element through equivalent viscous damping. Nevertheless, the

435

objective is to obtain a connection that would perform appropriately under control, which objective has been

436

attained given the higher teachability of the device under its semi-active mode.

437

To further investigate the quality of the analytical solutions, Fig. 18 plots three non-dimensional analyt-

438

ical solutions (Rs , Ra and Rc ) at the designed connection stiffness under various damping capacities. These

31

439

analytical solutions are compared against the simulation results, using performance metric R∗ : R∗ =

Rmodel − Rsimulation × 100% Rmodel

440

where the R refers to Rs , Ra and Rc and the subscript ‘model’ represents the analytical solutions and

441

‘simulation’ the simulation results of passive-on case or semi-active case. Results show a non-negligible dis-

442

agreement between the passive-on (i.e., numerical solution) and all three analytical solutions. In particular,

443

Rs and Rc are overestimated, while Ra is underestimated by the analytical solutions, consistent with the

444

results from Fig. 17. The large discrepancies are attributed to the assumptions made to develop a mathe-

445

matically trackable solution, whereas 1) higher modal responses of the structure were neglected; 2) the wind

446

spectrum was taken as constant; and 3) friction was simplified into an equivalent viscous system ignoring

447

high nonlinearities. However, in all cases, the analytical solutions overestimate the responses with respect

448

to the semi-active case, demonstrating that the design methodology provides a conservative design for the

449

VFCC used in its semi-active mode. This finding also agrees with results from Fig. 17.

32

(a)

(b)

(c) Figure 18: Comparison of the non-dimensional design factors (a) Rs , (b)Ra , and (c) Rc .

450

6. Conclusions

451

A novel variable friction cladding connection (VFCC) has been previously proposed by the authors to

452

leverage a structure’s cladding system for multiple hazard mitigation. This friction device provides a lateral

453

connection between the cladding element and the structural system. Its variation in the friction force is

454

provided by an actuator that applies a variable pressure onto sliding friction plates via a toggle system. In

455

this paper, a motion-based design (MBD) approach to the design of the VFCC is developed for wind hazard

456

mitigation. While this paper investigates a single hazard, results are to be integrated in future work for

457

multi-hazard applications.

458

A set of analytical transfer functions were derived to enable the MBD approach, based on a simplification

459

of the structural-cladding system into a two degrees-of-freedom (2DOF) system. These transfer functions

33

460

were used to estimate the maximum building responses under a quantified wind load, which responses in-

461

cluded inter-story displacement, absolute acceleration, and structure-cladding displacement. These analytical

462

solutions were numerically verified, and show good agreement between the numerical and analytical solu-

463

tions, except when the connection was overdamped where the equivalent viscous damping representation of

464

the friction element failed at accurately modeling the friction behavior by ignoring important nonlinearities,

465

such as the stick-slip motion.

466

The MBD procedure was demonstrated on a 24-story building, and compared to the performance of the

467

VFCC under semi-active control with the passive-on mode and the uncontrolled structure. Simulation results

468

show that the VFCC was capable of reducing the response of the uncontrolled structure under the prescribed

469

performance objectives. However, while the design of the VFCC under the MBD approach assumes a passive-

470

on mode, the passive-on scenario underestimated the maximum absolute acceleration. This was attributed

471

to an underestimation of the acceleration response in the analytical solutions due to the various assumptions

472

in the model. Nevertheless, it was also found that the analytical solutions always overestimate structural

473

response with respect to the semi-active mode of the VFCC, therefore demonstrating that the MBD approach

474

yields a conservative design of the semi-active VFCC. Overall, this study demonstrated the promise of the

475

VFCC at mitigating wind-induced vibrations in structures by leveraging the cladding system, and provided

476

an MBD approach enabling its holistic integration during the structural design phase.

477

Acknowledgments

478

This material is based upon the work supported by the National Science Foundation under Grant No.

479

1463252 and No. 1463497. Their support is gratefully acknowledged. Any opinions, findings, and conclusions

480

or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the

481

views of the National Science Foundation.

482

References

483

[1] J. Connor, S. Laflamme, Structural motion engineering, Springer, 2014.

484

[2] S. Laflamme, J. E. Slotine, J. Connor, Self-organizing input space for control of structures, Smart

485

486

487

488

489

Materials and Structures 21 (11) (2012) 115015. [3] T. E. Saaed, G. Nikolakopoulos, J.-E. Jonasson, H. Hedlund, A state-of-the-art review of structural control systems, Journal of Vibration and Control 21 (5) (2015) 919–937. [4] F. Ubertini, Active feedback control for cable vibrations, Smart Structures and Systems 4 (4) (2008) 407–428.

34

490

491

[5] A. L. Materazzi, F. Ubertini, Robust structural control with system constraints, Structural Control and Health Monitoring 19 (3) (2012) 472–490.

492

[6] I. Venanzi, F. Ubertini, A. L. Materazzi, Optimal design of an array of active tuned mass dampers for

493

wind-exposed high-rise buildings, Structural Control and Health Monitoring 20 (6) (2013) 903–917.

494

[7] L. Cao, S. Laflamme, D. Taylor, J. Ricles, Simulations of a variable friction device for multihazard

495

mitigation, Journal of Structural Engineering 142 (12) (2016) H4016001. doi:10.1061/(asce)st.

496

1943-541x.0001580.

497

URL https://doi.org/10.1061/(asce)st.1943-541x.0001580

498

499

[8] A. Downey, L. Cao, S. Laflamme, D. Taylor, J. Ricles, High capacity variable friction damper based on band brake technology, Engineering Structures 113 (2016) 287–298.

500

[9] M. Amjadian, A. K. Agrawal, Vibration control using a variable coil-based friction damper, in: Active

501

and Passive Smart Structures and Integrated Systems 2017, Vol. 10164, International Society for Optics

502

and Photonics, 2017, p. 101642J.

503

[10] J. Love, M. Tait, H. Toopchi-Nezhad, A hybrid structural control system using a tuned liquid damper

504

to reduce the wind induced motion of a base isolated structure, Engineering Structures 33 (3) (2011)

505

738–746.

506

[11] Y.-H. Shin, S.-J. Moon, J.-M. Kim, H.-Y. Cho, J.-Y. Choi, H.-W. Cho, Design considerations of linear

507

electromagnetic actuator for hybrid-type active mount damper, IEEE Transactions on Magnetics 49 (7)

508

(2013) 4080–4083.

509

510

511

512

513

514

515

[12] J. Høgsberg, M. L. Brodersen, Hybrid viscous damper with filtered integral force feedback control, Journal of Vibration and Control 22 (6) (2016) 1645–1656. [13] B. Spencer Jr, S. Nagarajaiah, State of the art of structural control, Journal of structural engineering 129 (7) (2003) 845–856. [14] L. Cao, S. Laflamme, Real-time variable multidelay controller for multihazard mitigation, Journal of Engineering Mechanics 144 (2) (2018) 04017174. doi:10.1061/(asce)em.1943-7889.0001406. [15] G. Ma, Z. Ye, Energy absorption of double-layer foam cladding for blast alleviation, International

516

Journal of Impact Engineering 34 (2) (2007) 329–347. doi:10.1016/j.ijimpeng.2005.07.012.

517

URL https://doi.org/10.1016/j.ijimpeng.2005.07.012

518

[16] R. Alberdi, J. Przywara, K. Khandelwal, Performance evaluation of sandwich panel systems for blast

519

mitigation, Engineering Structures 56 (2013) 2119–2130. doi:10.1016/j.engstruct.2013.08.021.

520

URL https://doi.org/10.1016/j.engstruct.2013.08.021 35

521

522

[17] C. Naito, M. Beacraft, J. Hoemann, J. Shull, H. Salim, B. Bewick, Blast performance of single-span precast concrete sandwich wall panels, Journal of Structural Engineering 140 (12) (2014) 04014096.

523

[18] M. Theobald, G. Nurick, Experimental and numerical analysis of tube-core claddings under blast loads,

524

International Journal of Impact Engineering 37 (3) (2010) 333–348. doi:10.1016/j.ijimpeng.2009.

525

10.003.

526

URL https://doi.org/10.1016/j.ijimpeng.2009.10.003

527

[19] G. Langdon, G. Schleyer, Deformation and failure of profiled stainless steel blast wall panels. part III:

528

finite element simulations and overall summary, International Journal of Impact Engineering 32 (6)

529

(2006) 988–1012. doi:10.1016/j.ijimpeng.2004.08.002.

530

URL https://doi.org/10.1016/j.ijimpeng.2004.08.002

531

[20] C. Amadio, C. Bedon, Viscoelastic spider connectors for the mitigation of cable-supported fa¸cades

532

subjected to air blast loading, Engineering Structures 42 (2012) 190–200. doi:10.1016/j.engstruct.

533

2012.04.023.

534

URL https://doi.org/10.1016/j.engstruct.2012.04.023

535

[21] W. Chen, H. Hao, Numerical study of blast-resistant sandwich panels with rototional friction dampers,

536

International Journal of Structural Stability and Dynamics 13 (06) (2013) 1350014. doi:10.1142/

537

s0219455413500144.

538

URL https://doi.org/10.1142/s0219455413500144

539

[22] W. S. Chen, H. Hao, Preliminary study of sandwich panel with rotational friction hinge device against

540

blast loadings, Key Engineering Materials 535-536 (2013) 530–533. doi:10.4028/www.scientific.

541

net/kem.535-536.530.

542

URL https://doi.org/10.4028/www.scientific.net/kem.535-536.530

543

[23] Y. Wang, J. R. Liew, S. C. Lee, X. Zhai, W. Wang, Crushing of a novel energy absorption connector

544

with curved plate and aluminum foam as energy absorber, Thin-Walled Structures 111 (2017) 145–154.

545

doi:10.1016/j.tws.2016.11.019.

546

URL https://doi.org/10.1016/j.tws.2016.11.019

547

[24] A. Baird, R. Diaferia, A. Palermo, S. Pampanin, Parametric investigation of seismic interaction between

548

precast concrete cladding systems and moment resisting frames, in: Structures Congress 2011, American

549

Society of Civil Engineers, 2011. doi:10.1061/41171(401)114.

550

URL https://doi.org/10.1061/41171(401)114

551

[25] L. Ferrara, R. Felicetti, G. Toniolo, C. Zenti, Friction dissipative devices for cladding panels in precast

552

buildings. an experimental investigation, European Journal of Environmental and Civil engineering 36

553

15 (9) (2011) 1319–1338. doi:10.3166/ejece.15.1319-1338.

554

URL https://doi.org/10.3166/ejece.15.1319-1338

555

[26] F. Biondini, B. Dal Lago, G. Toniolo, Experimental and numerical assessment of dissipative connections

556

for precast structures with cladding panels, in: 2nd European conference on earthquake engineering and

557

seismology (ECEES), Istanbul, Turkey, Vol. 2168, 2014, pp. 25–29.

558

[27] H. Maneetes, A. Memari, Introduction of an innovative cladding panel system for multi-story buildings,

559

Buildings 4 (3) (2014) 418–436. doi:10.3390/buildings4030418.

560

URL https://doi.org/10.3390/buildings4030418

561

[28] Y. Gong, L. Cao, S. Laflamme, S. Quiel, J. Ricles, D. Taylor, Characterization of a novel variable

562

friction connection for semiactive cladding system, Structural Control and Health Monitoring 25 (6)

563

(2018) e2157. doi:10.1002/stc.2157.

564

[29] L. Cao, S. Lu, S. Laflamme, S. Quiel, J. Ricles, D. Taylor, Performance-based design procedure of a novel

565

friction-based cladding connection for blast mitigation, International Journal of Impact Engineering 117

566

(2018) 48–62. doi:10.1016/j.ijimpeng.2018.03.003.

567

URL https://doi.org/10.1016/j.ijimpeng.2018.03.003

568

569

570

571

572

[30] S. Narasimhan, S. Nagarajaiah, Smart base isolated buildings with variable friction systems: H controller and saivf device, Earthquake engineering & structural dynamics 35 (8) (2006) 921–942. [31] Y. Kawamoto, Y. Suda, H. Inoue, T. Kondo, Electro-mechanical suspension system considering energy consumption and vehicle manoeuvre, Vehicle System Dynamics 46 (S1) (2008) 1053–1063. [32] M. Lorenz, B. Heimann, V. H¨ artel, A novel engine mount with semi-active dry friction damping, Shock

573

and Vibration 13 (4-5) (2006) 559–571. doi:10.1155/2006/263251.

574

URL https://doi.org/10.1155/2006/263251

575

[33] H. Dai, Z. Liu, W. Wang, Structural passive control on electromagnetic friction energy dissipation

576

device, Thin-Walled Structures 58 (2012) 1–8. doi:10.1016/j.tws.2012.03.017.

577

URL https://doi.org/10.1016/j.tws.2012.03.017

578

579

580

581

[34] M. A. Karkoub, M. Zribi, Active/semi-active suspension control using magnetorheological actuators, International journal of systems science 37 (1) (2006) 35–44. [35] S. Bharti, S. Dumne, M. Shrimali, Seismic response analysis of adjacent buildings connected with mr dampers, Engineering Structures 32 (8) (2010) 2122–2133.

37

582

[36] L. Cao, A. Downey, S. Laflamme, D. Taylor, J. Ricles, Variable friction device for structural control based

583

on duo-servo vehicle brake: Modeling and experimental validation, Journal of Sound and Vibration 348

584

(2015) 41–56. doi:10.1016/j.jsv.2015.03.011.

585

URL https://doi.org/10.1016/j.jsv.2015.03.011

586

587

[37] M. Mirtaheri, A. P. Zandi, S. S. Samadi, H. R. Samani, Numerical and experimental study of hysteretic behavior of cylindrical friction dampers, Engineering Structures 33 (12) (2011) 3647–3656.

588

[38] D. Zhao, H. Li, Shaking table tests and analyses of semi-active fuzzy control for structural seismic

589

reduction with a piezoelectric variable-friction damper, Smart Materials and Structures 19 (10) (2010)

590

105031. doi:10.1088/0964-1726/19/10/105031.

591

URL https://doi.org/10.1088/0964-1726/19/10/105031

592

593

594

595

596

597

598

599

600

601

602

[39] J. Pardo-Varela, J. Llera, A semi-active piezoelectric friction damper, Earthquake Engineering & Structural Dynamics 44 (3) (2015) 333–354. [40] M. Amjadian, A. K. Agrawal, A passive electromagnetic eddy current friction damper (pemecfd): Theoretical and analytical modeling, Structural Control and Health Monitoring 24 (10). [41] A. K. Chopra, Dynamics of structures: theory and applications to earthquake engineering. 2007, Google Scholar (2007) 470–472. [42] K. S. Moon, Vertically distributed multiple tuned mass dampers in tall buildings: performance analysis and preliminary design, The Structural Design of Tall and Special Buildings 19 (3) (2010) 347–366. [43] J. Scruggs, W. Iwan, Control of a civil structure using an electric machine with semiactive capability, Journal of Structural Engineering 129 (7) (2003) 951–959. [44] G. J. Hiemenz, W. Hu, N. M. Wereley, Semi-active magnetorheological helicopter crew seat suspension

603

for vibration isolation, Journal of Aircraft 45 (3) (2008) 945–953.

604

[45] E. Simiu, R. H. Scanlan, Wind effects on structures, Wiley, 1996.

605

[46] R. H. S. Emil Simiu, Wind effects on structures, John Wiley and Sons, Inc., 1986.

606

[47] J. C. Kaimal, J. Wyngaard, Y. Izumi, O. Cot´e, Spectral characteristics of surface-layer turbulence,

607

Quarterly Journal of the Royal Meteorological Society 98 (417) (1972) 563–589.

608

[48] Q. Li, X. Zou, J. Wu, Q. Wang, Integrated wind-induced response analysis and design optimization of

609

tall steel buildings using micro-ga, The Structural Design of Tall and Special Buildings 20 (8) (2011)

610

951–971.

38

611

[49] C.-M. Chan, K.-M. Wong, Structural topology and element sizing design optimisation of tall steel

612

frameworks using a hybrid OC–GA method, Structural and Multidisciplinary Optimization 35 (5) (2007)

613

473–488. doi:10.1007/s00158-007-0151-1.

614

URL https://doi.org/10.1007/s00158-007-0151-1

615

[50] W. Melbourne, T. Palmer, Accelerations and comfort criteria for buildings undergoing complex motions,

616

Journal of Wind Engineering and Industrial Aerodynamics 41 (1-3) (1992) 105–116. doi:10.1016/

617

0167-6105(92)90398-t.

618

URL https://doi.org/10.1016/0167-6105(92)90398-t

619

[51] Q. Li, J. Wu, S. Liang, Y. Xiao, C. Wong, Full-scale measurements and numerical evaluation of wind-

620

induced vibration of a 63-story reinforced concrete tall building, Engineering structures 26 (12) (2004)

621

1779–1794.

622

623

[52] J. P. Hunt, Seismic performance assessment and probabilistic repair cost analysis of precast concrete cladding systems for multistory buildings, University of California, Berkeley, 2010.

624

[53] E. Pantoli, T. C. Hutchinson, Experimental and analytical study of the dynamic characteristics of

625

architectural precast concrete cladding, in: Improving the Seismic Performance of Existing Buildings and

626

Other Structures 2015, American Society of Civil Engineers, 2015. doi:10.1061/9780784479728.046.

627

URL https://doi.org/10.1061/9780784479728.046

628

629

630

[54] J.-P. Pinelli, J. I. Craig, B. J. Goodno, Energy-based seismic design of ductile cladding systems, Journal of Structural Engineering 121 (3) (1995) 567–578. [55] G. Solari, Gust buffeting. II: Dynamic alongwind response, Journal of Structural Engineering 119 (2)

631

(1993) 383–398. doi:10.1061/(asce)0733-9445(1993)119:2(383).

632

URL https://doi.org/10.1061/(asce)0733-9445(1993)119:2(383)

633

[56] G. Huang, X. Chen, Wind load effects and equivalent static wind loads of tall buildings based on

634

synchronous pressure measurements, Engineering Structures 29 (10) (2007) 2641–2653. doi:10.1016/

635

j.engstruct.2007.01.011.

636

URL https://doi.org/10.1016/j.engstruct.2007.01.011

637

638

[57] A. G. Davenport, Note on the distribution of the largest value of a random function with application to gust loading., Proceedings of the Institution of Civil Engineers 28 (2) (1964) 187–196.

639

[58] K.-W. Min, J.-Y. Seong, J. Kim, Simple design procedure of a friction damper for reducing seismic

640

responses of a single-story structure, Engineering Structures 32 (11) (2010) 3539–3547. doi:10.1016/

39

641

j.engstruct.2010.07.022.

642

URL https://doi.org/10.1016/j.engstruct.2010.07.022

643

[59] S. H. Crandall, W. D. Mark, Random vibration in mechanical systems, Academic Press, 2014.

644

[60] I. S. Gradshteyn, I. M. Ryzhik, Table of integrals, series, and products, Academic press, 2014.

645

[61] R. G. Brown, P. Y. Hwang, Introduction to random signals and applied kalman filtering: with matlab

646

exercises and solutions, Introduction to random signals and applied Kalman filtering: with MATLAB

647

exercises and solutions, by Brown, Robert Grover.; Hwang, Patrick YC New York: Wiley, c1997.

648

[62] Y. Gong, L. Cao, L. Micheli, S. Laflamme, S. Quiel, J. Ricles, Performance evaluation of a semi-

649

active cladding connection for multi-hazard mitigation, in: A. Erturk (Ed.), Active and Passive Smart

650

Structures and Integrated Systems XII, SPIE, 2018. doi:10.1117/12.2295933.

651

652

[63] G. Yao, F. Yap, G. Chen, W. Li, S. Yeo, Mr damper and its application for semi-active control of vehicle suspension system, Mechatronics 12 (7) (2002) 963–973.

653

[64] C.-C. Lin, L.-Y. Lu, G.-L. Lin, T.-W. Yang, Vibration control of seismic structures using semi-active

654

friction multiple tuned mass dampers, Engineering Structures 32 (10) (2010) 3404–3417. doi:10.1016/

655

j.engstruct.2010.07.014.

656

URL https://doi.org/10.1016/j.engstruct.2010.07.014

657

658

659

660

661

662

663

664

[65] K. S. Moon, Tall building motion control using double skin fa¸cades, Journal of architectural engineering 15 (3) (2009) 84–90. [66] T. S. Fu, R. Zhang, Integrating double-skin fa¸cades and mass dampers for structural safety and energy efficiency, Journal of Architectural Engineering 22 (4) (2016) 04016014. [67] ASCE, Minimum Design Loads for Buildings and Other Structures,American Socitey of Civil Engineering, Reston, VA, asce/sei 7-10 edition., 2010. [68] G. Deodatis, Simulation of ergodic multivariate stochastic processes, Journal of engineering mechanics 122 (8) (1996) 778–787.

665

[69] F. Ubertini, F. Giuliano, Computer simulation of stochastic wind velocity fields for structural response

666

analysis: Comparisons and applications, Advances in Civil Engineering 2010 (2010) 1–20. doi:10.

667

1155/2010/749578.

668

URL https://doi.org/10.1155/2010/749578

40