1. Damageandlossevaluationintheperformance-basedwindengineering
Francesco Petrini
Konstantinos Gkoumas
Franco Bontempi
ICOSSAR 2013
11th
International Conference on Structural Safety & Reliability
June 16-20, Columbia University, New York, NY
Sapienza – University of Rome
Francesco Petrini, Ph.D., P.E.
Konstantinos Gkoumas, Ph.D., P.E.
Franco Bontempi, Ph.D., P.E.
Sapienza - University of Rome
Dipartimento di Ingegneria Strutturale e
Geotecnica
Damage and loss evaluation in the performance-
based wind engineering
2. Damageandlossevaluationintheperformance-basedwindengineering
Francesco Petrini
Konstantinos Gkoumas
Franco Bontempi
ICOSSAR 2013
11th
International Conference on Structural Safety & Reliability
June 16-20, Columbia University, New York, NY
Presentation outline
2
• Overview of the Performance Based Wind
Engineering (PBWE) procedure
• Models for tall buildings and the assessment of
occupant comfort:
• Application on a high-rise building
• Assessment of the annual probabilities of exceeding
the human perception thresholds
• Vibration and occupant comfort issues
• Damage analysis
• Loss analysis
• Conclusions and indications for further research
3. Damageandlossevaluationintheperformance-basedwindengineering
Francesco Petrini
Konstantinos Gkoumas
Franco Bontempi
ICOSSAR 2013
11th
International Conference on Structural Safety & Reliability
June 16-20, Columbia University, New York, NY
Presentation outline
3
• Overview of the Performance Based Wind
Engineering (PBWE) procedure
• Models for tall buildings and the assessment of
occupant comfort
• Application on a high-rise building
• Assessment of the annual probabilities of exceeding
the human perception thresholds
• Vibration and occupant comfort issues
• Damage analysis
• Loss analysis
• Conclusions and indications for further research
4. Damageandlossevaluationintheperformance-basedwindengineering
Francesco Petrini
Konstantinos Gkoumas
Franco Bontempi
ICOSSAR 2013
11th
International Conference on Structural Safety & Reliability
June 16-20, Columbia University, New York, NY
Performance-Based Wind Engineering (PBWE)
4
The problem of risk assessment is disaggregated into the following elements:
- site and structure-specific hazard analyses, that is, the assessment of the
probability density functions f(IM), f(SP) and f(IP|IM,SP);
- structural analysis, aiming at the assessment of the probability density function of
the structural response f(EDP|IM,IP,SP) conditional on the parameters characterizing the
environmental actions, the wind-fluid-structure interaction and the structural properties;
- damage analysis, that gives the damage probability density function f(DM|EDP)
conditional on EDP;
- finally, loss analysis, that is the assessment of G(DV|DM), where G(·|·) is a
conditional complementary cumulative distribution function.
G(DV) = ∫…∫ G(DV|DM) · f(DM|EDP) · f(EDP|IM, IP,SP) · f(IP|IM,SP) ·
· f(IM) · f(SP) · dDM · dEDP · dIP · dIM · dSP
Interaction
Parameters
Structural
Parameters
Intensity
measure
IM IP SP
Engineering
Demand
Parameters
EDP
Damage
Measure
DM
Decision
Variable
DV
5. Damageandlossevaluationintheperformance-basedwindengineering
Francesco Petrini
Konstantinos Gkoumas
Franco Bontempi
ICOSSAR 2013
11th
International Conference on Structural Safety & Reliability
June 16-20, Columbia University, New York, NY
PBWE procedure flowchart
5
Petrini, F. & Ciampoli M., 2012, Performance-based wind design of tall buildings, Structure & Infrastructure
Engineering, 8(10), 954-966.
O
f(IM|O)
f(IM) f(IP|IM,SP)
f(IP)
f(EDP|IM,IP,SP)
G(EDP)
f(DM|EDP)
G(DM)
f(DV|DM)
G(DV)
Hazard analysis
Interaction
analysis
Structuralanalysis Damageanalysis Loss analysis
IM: intensity
measure
IP: interaction
parameters
EDP:engineering
demand param.
DM:damage
measure
DV:decision
variable
Select
O, D
O:location
D:design
Environme
nt info
Decision-
making
D
f(SP|D)
f(SP)
Structural
characterization
SP:structural
system parameters
Structural
system
info
Ciampoli M, Petrini, F. & Augusti G., 2011, Performance-Based Wind Engineering: toward a general
procedure, Structural Safety, Structural Safety, 33(6), 367-378.
6. Damageandlossevaluationintheperformance-basedwindengineering
Francesco Petrini
Konstantinos Gkoumas
Franco Bontempi
ICOSSAR 2013
11th
International Conference on Structural Safety & Reliability
June 16-20, Columbia University, New York, NY
6
O
f(IM|O)
f(IM)
f(IP|IM,SP)
f(IP)
f(EDP|IM,IP,SP)
G(EDP)
f(DM|EDP)
G(DM)
f(DV|DM)
G(DV)
Hazard analysis
Aerodynamic
analysis
Struc’l analysis Damage analysis Loss analysis
IM: intensity measure
IP: interaction
parameters
EDP: engineering
demand parameters
DM: damage measures DV: decision variables
Select
O, D
O: location
D: design
Environment
info
Decision-
making
D
f(SP|D)
f(SP)
Structural
characterization
SP: structural system
parameters
Structural
system info
O
f(IM|O)
f(IM)
f(IP|IM,SP)
f(IP)
f(EDP|IM,IP,SP)
G(EDP)
f(DM|EDP)
G(DM)
f(DV|DM)
G(DV)
Hazard analysis
Aerodynamic
analysis
Struc’l analysis Damage analysis Loss analysis
IM: intensity measure
IP: interaction
parameters
EDP: engineering
demand parameters
DM: damage measures DV: decision variables
Select
O, D
O: location
D: design
Environment
info
Decision-
making
D
f(SP|D)
f(SP)
Structural
characterization
SP: structural system
parameters
Structural
system info
O, D
g(IM|O,D)
g(IM)
p(EDP|IM)
P(EDP)
p(DM|EDP)
P(DM)
p(DV|DM)
P(DV)
Hazard analysis Struc’l analysis Damage analysis Loss analysis
IM: intensity
measure
EDP: engineering
demand param.
DM: damage
measure
DV: decision
variable
Select
O, D
O: location
D: design
Facility
info
Decision-
making
O, D
g(IM|O,D)
g(IM)
p(EDP|IM)
P(EDP)
p(DM|EDP)
P(DM)
p(DV|DM)
P(DV)
Hazard analysis Struc’l analysis Damage analysis Loss analysis
IM: intensity
measure
EDP: engineering
demand param.
DM: damage
measure
DV: decision
variable
Select
O, D
O: location
D: design
Facility
info
Decision-
making
PBWEPBEE
7. Damageandlossevaluationintheperformance-basedwindengineering
Francesco Petrini
Konstantinos Gkoumas
Franco Bontempi
ICOSSAR 2013
11th
International Conference on Structural Safety & Reliability
June 16-20, Columbia University, New York, NY
Presentation outline
7
• Overview of the Performance Based Wind
Engineering (PBWE) procedure.
• Models for tall buildings and the assessment of
occupant comfort
• Application on a high-rise building
• Assessment of the annual probabilities of exceeding
the human perception thresholds
• Vibration and occupant comfort issues
• Damage analysis
• Loss analysis
• Conclusions and indications for further research
8. Damageandlossevaluationintheperformance-basedwindengineering
Francesco Petrini
Konstantinos Gkoumas
Franco Bontempi
ICOSSAR 2013
11th
International Conference on Structural Safety & Reliability
June 16-20, Columbia University, New York, NY
8
Tamura,Y.(2009).Windandtallbuildings,ProceedingsoftheFifth
European&AfricanConferenceonWindEngineering(EACWE5),
Florence,Italy,July19-23,2009..
Vibration frequency
Accelerationthresholdsformotion
perception
w(t;z2)Vm(z2)
Vm (z1)
Vm (z3)
V(t;z2)
v(t;z2)u(t;z2)
X
Z
Y
θ
B1
B2
H
Loss of serviceability
Lossofintegrityof
non-structural
elements
Motionperception
bybuilding
occupants
Displacements
Acceleration
Discomfort level in terms of
perception thresholds
1
9. Damageandlossevaluationintheperformance-basedwindengineering
Francesco Petrini
Konstantinos Gkoumas
Franco Bontempi
ICOSSAR 2013
11th
International Conference on Structural Safety & Reliability
June 16-20, Columbia University, New York, NY
9
Loss of serviceability
Lossofintegrityof
non-structural
elements
Motionperception
bybuilding
occupants
Bashor,R.andKareem,A.(2007)."ProbabilisticPerformanceEvaluationof
Buildings:AnOccupantComfortPerspective",Proc.12thInternational
ConferenceonWindEngineering,1-6July,Cairns,Australia.Available
onlineathttp://www.nd.edu/~nathaz/[Accessed15June2010].
w(t;z2)Vm(z2)
Vm (z1)
Vm (z3)
V(t;z2)
v(t;z2)u(t;z2)
X
Z
Y
θ
B1
B2
H
Discomfort level in terms of
perception thresholds
Usually Across wind vibration
is critical for comfort
The reference period for
comfort evaluation is 1 year
1
2
3 1st
natural frequency is dominant4
Italian Guidelines
f1
Scalarthreshold
Displacements
Acceleration
10. Damageandlossevaluationintheperformance-basedwindengineering
Francesco Petrini
Konstantinos Gkoumas
Franco Bontempi
ICOSSAR 2013
11th
International Conference on Structural Safety & Reliability
June 16-20, Columbia University, New York, NY
10
Case study structure
Structure
•74 floors
•Height H=305m
•Footprint B1=B2=50m (square)
3dframeontheexternalperimeter
centralcore
Bracing system
A steel high-rise building
Finite Element model
FE Model
Approximately
•10,000 elements
•4,000 nodes
•24,000 DOFs
11. Damageandlossevaluationintheperformance-basedwindengineering
Francesco Petrini
Konstantinos Gkoumas
Franco Bontempi
ICOSSAR 2013
11th
International Conference on Structural Safety & Reliability
June 16-20, Columbia University, New York, NY
11
Experimental model of
Actions
SpenceS.M.J.,GioffrèM.,GusellaV.,Influenceofhighermodesonthe
dynamicre-sponseofirregularandregulartallbuildings,Proc.6th
InternationalColloquiumonBluffBodiesAerodynamicsand
Applications(BBAAVI),Milano,Italy,July20-24,2008.
Boundary Layer Wind Tunnel of the
CRIACIV in Prato, Italy
1:500Scalemodel
Response
time history
Time domain structural analyses
(Experimental actions)
Time domain
analyses
Experimental
forces
-30
-20
-10
0
10
20
30
3500 3600 3700 3800 3900 4000
aL, aD
[cm/s2
]
t [s]Along Across
12. Damageandlossevaluationintheperformance-basedwindengineering
Francesco Petrini
Konstantinos Gkoumas
Franco Bontempi
ICOSSAR 2013
11th
International Conference on Structural Safety & Reliability
June 16-20, Columbia University, New York, NY
12
( )
( )
( ) )(),(
),,(exp
1
),(),(
22
212
2
ωχωρ
ωξξ
ωρω
⋅⋅⋅⋅=
=⋅⋅−⋅
⋅⋅⋅⋅=
∫∫
hSVc
dAdAf
A
hSVchS
uumxD
A A
uumxDDD tt
( )
)(),h(S
)(HVc
),h(S)(H),h(S
2
uu
22
mxD
DD
2
rr tttt
ωχω
ωρ
ωωω
⋅⋅
⋅⋅⋅⋅
=⋅=
⋅+
−
⋅
⋅
⋅
=
2
0
2
2
2
0
2
2
0
2
2
41
1
1
)(
ω
ω
ν
ω
ω
ω
ω
m
H
rrm
p
grr σ⋅+= rg
Wind action
spectra
(analytical)
Response spectra
Peak response
Frequency domain response
Response Peak
Factor
Analytical model of the buffeting forces
( ) ( ) ( ) ( )( )ωfexpωSωSωS jkuuuuuu kkjjkj
−=
( )
( )
( ) ( )( )kj
2
kj
2
z
jk
zVzV2π
zzCω
ωf
+
−
=
Cross-spectrum
5.0
0
uu2
x
u
200
300(x)dxR
u
1
L
⋅== ∫
∞
z
where:
( )
( ) [ ]5/3
ju
ju
x2
u
uu
/zLf10.3021ω/2π
/zLfσ6.686
ωS jj
⋅⋅+⋅
⋅⋅⋅
=
( )( ) 2
fri0
0
u
2
u
u1.75)log(zarctan1.16
(n)dnSσ
⋅+⋅−=
== ∫
∞
)z(V2π
zω
f
jm
j
⋅
⋅
=
Autospectrum
( ) 3ew(t)2ev(t)1eu(t))j(zmV)jz(t;jV
⋅+⋅+⋅+=
α
10m
10
z
V(z)V
⋅=
Solari,G.Piccardo,G.(2001).Probabilistic3-Dturbulencemodelingforgustbuffetingof
structures,ProbabilisticEngineeringMechanics,(16),73–86.
Turbulentwindvelocityspectra
(analytical)
Model of the Vortex shedding forces
(variable with the angle of attack)
1.E+01
1.E+03
1.E+05
1.E+07
1.E+09
1.E+11
0.000 0.001 0.010 0.100 1.000
PSD
n [Hz]
Total Forcespectrum
Turbulenceforcespectrum
Vortexsheddingforcespectrum
15. Damageandlossevaluationintheperformance-basedwindengineering
Francesco Petrini
Konstantinos Gkoumas
Franco Bontempi
ICOSSAR 2013
11th
International Conference on Structural Safety & Reliability
June 16-20, Columbia University, New York, NY
15
Hazard analysis
( ) ( )
θ
−⋅
θ
⋅
θ
θ
=θ
θ−θ
θ
)(
10
1)(
10
10, exp
)(
)(
),(f 10
kk
V
c
V
c
V
c
k
V
The roughness length z0 is characterized by a
lognormal PDF. The mean value μz0 and the
standard deviation σz0 of z0 are expressed as
function of θ (assuming a slight difference between
four sectors, i.e. a mean value of z0 varying
between 0.08 m and 0.12 m and a COVz0 equal to
0.30).
V10 and θ are described by their joint probability
distribution function
θ
V10
IM =
θ
V10
z0
Parameters c(θ) and k(θ) are derived from NIST®
wind speed database.
(Annual occurrence)
Models for tall buildings and the assessment of occupant comfort
O
f(IM|O)
f(IM) f(IP|IM,SP)
f(IP)
f(EDP|IM,IP,SP)
G(EDP)
f(DM|EDP)
G(DM)
f(DV|DM)
G(DV)
Hazard analysis
Interaction
analysis
Structural analysis Damageanalysis Loss analysis
IM: intensity
measure
IP: interaction
parameters
EDP:engineering
demand param.
DM:damage
measure
DV: decision
variable
Select
O, D
O:location
D:design
Environme
nt info
Decision-
making
D
f(SP|D)
f(SP)
Structural
characterization
SP:structural
system parameters
Structural
system
info
16. Damageandlossevaluationintheperformance-basedwindengineering
ICOSSAR 2013
11th
International Conference on Structural Safety & Reliability
June 16-20, Columbia University, New York, NY
Francesco Petrini
Konstantinos Gkoumas
Franco Bontempi
16
Models for tall buildings and the assessment of occupant comfort
Interaction analysis IP =
gr
CD
CL
O
f(IM|O)
f(IM) f(IP|IM,SP)
f(IP)
f(EDP|IM,IP,SP)
G(EDP)
f(DM|EDP)
G(DM)
f(DV|DM)
G(DV)
Hazard analysis
Interaction
analysis
Structural analysis Damageanalysis Loss analysis
IM: intensity
measure
IP: interaction
parameters
EDP:engineering
demand param.
DM:damage
measure
DV: decision
variable
Select
O, D
O:location
D:design
Environme
nt info
Decision-
making
D
f(SP|D)
f(SP)
Structural
characterization
SP:structural
system parameters
Structural
system
info
17. Damageandlossevaluationintheperformance-basedwindengineering
ICOSSAR 2013
11th
International Conference on Structural Safety & Reliability
June 16-20, Columbia University, New York, NY
Francesco Petrini
Konstantinos Gkoumas
Franco Bontempi
17
Interaction analysis IP =
gr
CD
CL
O
f(IM|O)
f(IM) f(IP|IM,SP)
f(IP)
f(EDP|IM,IP,SP)
G(EDP)
f(DM|EDP)
G(DM)
f(DV|DM)
G(DV)
Hazard analysis
Interaction
analysis
Structural analysis Damageanalysis Loss analysis
IM: intensity
measure
IP: interaction
parameters
EDP:engineering
demand param.
DM:damage
measure
DV: decision
variable
Select
O, D
O:location
D:design
Environme
nt info
Decision-
making
D
f(SP|D)
f(SP)
Structural
characterization
SP:structural
system parameters
Structural
system
info
Models for tall buildings and the assessment of occupant comfort
462.2507.1265.0 2
+ξ+ξ−=µ
rg
( )
≤⋅η
>⋅η
⋅η+
−
⋅η
=σ
+
+
+
+
122if650
122if
46
213
45
2
21
.T.
.T
.
)Tln(
.
)Tln(
.
windr,e
windr,e
windr,e
windr,e
gr
( )
<≤
η
<≤
η−
=η +
+
+
1690if
690100if
380631 450
r
r
r
r
.
r
r,e
q.
.q.
.q.
r
r
r σ
σ
=+η
(Obtained from time-domain
analyses)
The peak response factor gr is characterized by a Gaussian distribution function
rgµ
rgµ
Vanmarcke (1975)
The aerodynamic coefficients CD and CL are characterized by Gaussian
distributions. Mean values are expressed as a function of θ, varying from
those corresponding to a square shape (for θ = 0) to those corresponding to
a rhomboidal shape (for θ = 45); the coefficient of variations of CL and CD
are taken equal to 0.07 and 0.05.
μCD μCL
D
Cµ
μCD μCL
L
Cµ
rrm
p
grr σ⋅+=
18. Damageandlossevaluationintheperformance-basedwindengineering
ICOSSAR 2013
11th
International Conference on Structural Safety & Reliability
June 16-20, Columbia University, New York, NY
Francesco Petrini
Konstantinos Gkoumas
Franco Bontempi
18
G(EDP) = ∫…∫ G(EDP|IM, IP, SP) · f(IP|IM,SP) · f(IM) · f(SP) · dIP · dIM · dSP
Monte Carlo sim
(5000 runs)
aL
p
Reduced
formulation
O
f(IM|O)
f(IM) f(IP|IM,SP)
f(IP)
f(EDP|IM,IP,SP)
G(EDP)
f(DM|EDP)
G(DM)
f(DV|DM)
G(DV)
Hazard analysis
Interaction
analysis
Structural analysis Damageanalysis Loss analysis
IM: intensity
measure
IP: interaction
parameters
EDP:engineering
demand param.
DM:damage
measure
DV: decision
variable
Select
O, D
O:location
D:design
Environme
nt info
Decision-
making
D
f(SP|D)
f(SP)
Structural
characterization
SP:structural
system parameters
Structural
system
info
Structural analysis
Models for tall buildings and the assessment of occupant comfort
EDP= aL
p
(peak acceleration in the across wind direction)
19. Damageandlossevaluationintheperformance-basedwindengineering
ICOSSAR 2013
11th
International Conference on Structural Safety & Reliability
June 16-20, Columbia University, New York, NY
Francesco Petrini
Konstantinos Gkoumas
Franco Bontempi
19
Risk Curve. EDP= aL
p
= peak acceleration in the across wind direction
The annual probabilities of exceeding the human perception thresholds for
apartment and office building vibrations are 0.0576 and 0.0148 respectively.
aL
p
G(aL
p)
aL
p [mm/s2]
Ciampoli, M. & Petrini, F., 2012, Performance-Based Aeolian Risk assessment and reduction for tall buildings, Probabilistic
Engineering Mechanics, 28 (75–84).
Models for tall buildings and the assessment of occupant comfort
20. Damageandlossevaluationintheperformance-basedwindengineering
Francesco Petrini
Konstantinos Gkoumas
Franco Bontempi
ICOSSAR 2013
11th
International Conference on Structural Safety & Reliability
June 16-20, Columbia University, New York, NY
20
TMD
Design Parameters
γ = mTMD/mtot
β = ωTMD/ ω1
ξ* = damping of TMD
aL
p[mm/s2]
n [Hz]
β = ξ* =
β = ξ* =
β = ξ* =
β = ξ* =
β = ξ* =
β = ξ* =
β = ξ* =
G(aL
p)
aL
p [mm/s2]
Parametric analysis Effects on risk
γ = 1/150
Aeolian Risk reduction using TMD
Models for tall buildings and the assessment of occupant comfort
21. Damageandlossevaluationintheperformance-basedwindengineering
Francesco Petrini
Konstantinos Gkoumas
Franco Bontempi
ICOSSAR 2013
11th
International Conference on Structural Safety & Reliability
June 16-20, Columbia University, New York, NY
Presentation outline
21
• Overview of the Performance Based Wind
Engineering (PBWE) procedure.
• Models for tall buildings and the assessment of
occupant comfort
• Application on a high-rise building
• Assessment of the annual probabilities of exceeding
the human perception thresholds
• Vibration and occupant comfort issues
• Damage analysis
• Loss analysis
• Conclusions and indications for further research
22. Damageandlossevaluationintheperformance-basedwindengineering
Francesco Petrini
Konstantinos Gkoumas
Franco Bontempi
ICOSSAR 2013
11th
International Conference on Structural Safety & Reliability
June 16-20, Columbia University, New York, NY
22
Vibration and occupant comfort issues
Consequences of wind induced vibrations
in high rise buildings
-Fear and alarm
-Discomfort
-Reduced task concentration
-Dizziness, migraine and nausea
Kwok, K.C.S., Hitchcock, P.A. & Burton, M.D., 2009, Perception of vibration and occupant comfort in wind-
excited tall buildings, Journal of Wind Engineering and Industrial Aerodynamics, 97(7-8), 368-380
Wind induced vibration
−Damage analysis
−Loss Analysis
Studies on human perception of vibration and tolerance thresholds
-Field experiments and studies in wind-excited buildings
-Motion simulator tests
-Field experiments conducted in artificially excited buildings
Mitigation measures
-Modifications to the structural system and/or the
aerodynamic shape
-Installation of vibration control devices
- Negative impressions/ publicity
- Eventually they can be an attraction
O
f(IM|O)
f(IM) f(IP|IM,SP)
f(IP)
f(EDP|IM,IP,SP)
G(EDP)
f(DM|EDP)
G(DM)
f(DV|DM)
G(DV)
Hazard analysis
Interaction
analysis
Structural analysis Damageanalysis Loss analysis
IM: intensity
measure
IP: interaction
parameters
EDP:engineering
demand param.
DM:damage
measure
DV: decision
variable
Select
O, D
O:location
D:design
Environme
nt info
Decision-
making
D
f(SP|D)
f(SP)
Structural
characterization
SP:structural
system parameters
Structural
system
info
23. Damageandlossevaluationintheperformance-basedwindengineering
Francesco Petrini
Konstantinos Gkoumas
Franco Bontempi
ICOSSAR 2013
11th
International Conference on Structural Safety & Reliability
June 16-20, Columbia University, New York, NY
23
O
f(IM|O)
f(IM) f(IP|IM,SP)
f(IP)
f(EDP|IM,IP,SP)
G(EDP)
f(DM|EDP)
G(DM)
f(DV|DM)
G(DV)
Hazard analysis
Interaction
analysis
Structural analysis Damageanalysis Loss analysis
IM: intensity
measure
IP: interaction
parameters
EDP:engineering
demand param.
DM:damage
measure
DV: decision
variable
Select
O, D
O:location
D:design
Environme
nt info
Decision-
making
D
f(SP|D)
f(SP)
Structural
characterization
SP:structural
system parameters
Structural
system
info
Damage analysis
Probabilistic damage analysis: assign a probability distribution to the perception
thresholds
Procedure: obtain a pdf that assigns at each vibration level a percentage of persons
that experience discomfort
Kwok, K.C.S., Hitchcock, P.A., 2008. Occupant comfort test using a tall building motion simulator. In: Proceedings of Fourth
International Conference on Advances in Wind and Structures, Jeju, Korea, 28–30 May.
Vibration and occupant comfort issues
24. Damageandlossevaluationintheperformance-basedwindengineering
Francesco Petrini
Konstantinos Gkoumas
Franco Bontempi
ICOSSAR 2013
11th
International Conference on Structural Safety & Reliability
June 16-20, Columbia University, New York, NY
24
O
f(IM|O)
f(IM) f(IP|IM,SP)
f(IP)
f(EDP|IM,IP,SP)
G(EDP)
f(DM|EDP)
G(DM)
f(DV|DM)
G(DV)
Hazard analysis
Interaction
analysis
Structural analysis Damageanalysis Loss analysis
IM: intensity
measure
IP: interaction
parameters
EDP:engineering
demand param.
DM:damage
measure
DV: decision
variable
Select
O, D
O:location
D:design
Environme
nt info
Decision-
making
D
f(SP|D)
f(SP)
Structural
characterization
SP:structural
system parameters
Structural
system
info
Loss analysis
Probabilistic loss analysis: assign a cost probability for different damages
Issues: the uncertainty in the cost relies on various factors (e.g. market trend)
DM
Non structural elements
Structural elements
Comfort
Safety
Serviceability
Safety
Serviceability
DV
Direct
Indirect
(As a direct damage to the structure)
(As a consequence of the damaged structure)
IM
SP
IP EDP DM DV
- Direct VS indirect cost that are not
possible to account for in monetary terms.
- Initial VS life-cycle cost. In particular
regarding the evaluation of retrofitting
strategies that could improve the
serviceability performance (e.g. comfort),
by means of vibration mitigation.
Vibration and occupant comfort issues
25. Damageandlossevaluationintheperformance-basedwindengineering
Francesco Petrini
Konstantinos Gkoumas
Franco Bontempi
ICOSSAR 2013
11th
International Conference on Structural Safety & Reliability
June 16-20, Columbia University, New York, NY
25
Vibration and occupant comfort issues
Loss analysis for comfort – concept (1)
Limit states implying damages in structural or non-
structural components
•Direct damages (tangible): costs necessary for
retrofitting the structures
•Direct damages (non tangible): costs due to service
interruption for restoration
Uncertainties are on the unitary costs
26. Damageandlossevaluationintheperformance-basedwindengineering
Francesco Petrini
Konstantinos Gkoumas
Franco Bontempi
ICOSSAR 2013
11th
International Conference on Structural Safety & Reliability
June 16-20, Columbia University, New York, NY
26
Vibration and occupant comfort issues
Loss analysis for comfort – concept (2)
Limit states implying perception of low structural
quality (e.g. discomfort) but not implying damages
Our proposal
Cost necessary for improving the quality to an
acceptable level, e.g.
a.Cost related with a change of the activity in the structure (for
example, change from residential to office)
b.Cost of a TMD installation
• Direct losses: DL= ATMD * Ac + TMD installation cost
• Direct losses due to the activity interruption for retrofitting the TMD’s
• Account for possible future gains due to attraction factor
27. Damageandlossevaluationintheperformance-basedwindengineering
Francesco Petrini
Konstantinos Gkoumas
Franco Bontempi
ICOSSAR 2013
11th
International Conference on Structural Safety & Reliability
June 16-20, Columbia University, New York, NY
27
Vibration and occupant comfort issues
Loss analysis for comfort – concept (3)
INDIRECT DAMAGES
Depending o whether we consider a new or an
existing structure
• Structural design before
the construction
• Existing structure
no indirect costs
costs due to
service interruption
28. Damageandlossevaluationintheperformance-basedwindengineering
Francesco Petrini
Konstantinos Gkoumas
Franco Bontempi
ICOSSAR 2013
11th
International Conference on Structural Safety & Reliability
June 16-20, Columbia University, New York, NY
Wind occurrence
28
Vibration and occupant comfort issues
Lifecycle cost analysis
O
f(IM|O)
f(IM) f(IP|IM,SP)
f(IP)
f(EDP|IM,IP,SP)
G(EDP)
f(DM|EDP)
G(DM)
f(DV|DM)
G(DV)
Hazard analysis
Interaction
analysis
Structural analysis Damageanalysis Loss analysis
IM: intensity
measure
IP: interaction
parameters
EDP:engineering
demand param.
DM:damage
measure
DV: decision
variable
Select
O, D
O:location
D:design
Environme
nt info
Decision-
making
D
f(SP|D)
f(SP)
Structural
characterization
SP:structural
system parameters
Structural
system
info
Loss analysis
Economic
investment
Economic
value
Life cycle
assessment
MC
simulation
With TMD
With TMD retrofitted
Economic
losses
Without TMD
29. Damageandlossevaluationintheperformance-basedwindengineering
Francesco Petrini
Konstantinos Gkoumas
Franco Bontempi
ICOSSAR 2013
11th
International Conference on Structural Safety & Reliability
June 16-20, Columbia University, New York, NY
Presentation outline
29
• Overview of the Performance Based Wind
Engineering (PBWE) procedure.
• Models for tall buildings and the assessment of
occupant comfort
• Application on a high-rise building
• Assessment of the annual probabilities of exceeding
the human perception thresholds
• Vibration and occupant comfort issues
• Damage analysis
• Loss analysis
• Conclusions and indications for further research
30. Damageandlossevaluationintheperformance-basedwindengineering
Francesco Petrini
Konstantinos Gkoumas
Franco Bontempi
ICOSSAR 2013
11th
International Conference on Structural Safety & Reliability
June 16-20, Columbia University, New York, NY
• Occupant comfort is an important issue in the design of
tall buildings. Due to the stochastic nature of wind action
and wind-induced vibration, deterministic analyses are
inadequate for carrying out a comfort assessment.
• The insertion of passive control devices can reduce
the vibration perception of building occupants. But the
effectiveness of the device must be evaluated in terms of
cost (by computing the probability of exceeding
acceptable values of an appropriate DV).
• Damage and loss analysis of wind-induced vibrations
will be based on corroborated literature studies that
provide statistics on the occupant comfort.
30
Conclusions and indications for further research
31. Damageandlossevaluationintheperformance-basedwindengineering
Francesco Petrini
Konstantinos Gkoumas
Franco Bontempi
ICOSSAR 2013
11th
International Conference on Structural Safety & Reliability
June 16-20, Columbia University, New York, NY
Thank you for your attention
31
Francesco Petrini, Konstantinos Gkoumas, Franco Bontempi
Sapienza - University of Rome, Dipartimento di Ingegneria Strutturale e Geotecnica
Acknowledgements:
Prof. Marcello Ciampoli, Prof. Giuliano Augusti
This study is partially supported by StroNGER s.r.l. from the fund “FILAS - POR FESR LAZIO
2007/2013 - Support for the research spin-off”.
33. Damageandlossevaluationintheperformance-basedwindengineering
ICOSSAR 2013
11th
International Conference on Structural Safety & Reliability
June 16-20, Columbia University, New York, NY
Francesco Petrini
Konstantinos Gkoumas
Franco Bontempi
33
Models for tall buildings and the assessment of occupant comfort
Interaction analysis IP =
gr
CD
CL
O
f(IM|O)
f(IM) f(IP|IM,SP)
f(IP)
f(EDP|IM,IP,SP)
G(EDP)
f(DM|EDP)
G(DM)
f(DV|DM)
G(DV)
Hazard analysis
Interaction
analysis
Structural analysis Damageanalysis Loss analysis
IM: intensity
measure
IP: interaction
parameters
EDP:engineering
demand param.
DM:damage
measure
DV: decision
variable
Select
O, D
O:location
D:design
Environme
nt info
Decision-
making
D
f(SP|D)
f(SP)
Structural
characterization
SP:structural
system parameters
Structural
system
info
rrm
p
grr σ⋅+=
)T(log
.
)T(log
winde
windegr
⋅η
+⋅η=µ
2
5770
2 Davenport
(1983)
Reliable results for a broad
range of processes
34. Damageandlossevaluationintheperformance-basedwindengineering
ICOSSAR 2013
11th
International Conference on Structural Safety & Reliability
June 16-20, Columbia University, New York, NY
Francesco Petrini
Konstantinos Gkoumas
Franco Bontempi
34
Interaction analysis IP =
gr
CD
CL
O
f(IM|O)
f(IM) f(IP|IM,SP)
f(IP)
f(EDP|IM,IP,SP)
G(EDP)
f(DM|EDP)
G(DM)
f(DV|DM)
G(DV)
Hazard analysis
Interaction
analysis
Structural analysis Damageanalysis Loss analysis
IM: intensity
measure
IP: interaction
parameters
EDP:engineering
demand param.
DM:damage
measure
DV: decision
variable
Select
O, D
O:location
D:design
Environme
nt info
Decision-
making
D
f(SP|D)
f(SP)
Structural
characterization
SP:structural
system parameters
Structural
system
info
rrm
p
grr σ⋅+=
)T(log
.
)T(log
winde
windegr
⋅η
+⋅η=µ
2
5770
2 Davenport
(1983)
Reliable results for a broad
range of processes
- In the Davenport formulation the peak factor does not depend on the bandwidth of the stochastic process.
- Alternative formulations consider this dependence.
( )
≤⋅η
>⋅η
⋅η+
−
⋅η
=σ
+
+
+
+
122if650
122if
46
213
45
2
21
.T.
.T
.
)Tln(
.
)Tln(
.
windr,e
windr,e
windr,e
windr,e
gr
)ln(2
577.0
)ln(2
,
,
windre
windreg
T
Tr
⋅η
+⋅η=µ
+
+
Vanmarcke
(1975)
Models for tall buildings and the assessment of occupant comfort
35. Damageandlossevaluationintheperformance-basedwindengineering
ICOSSAR 2013
11th
International Conference on Structural Safety & Reliability
June 16-20, Columbia University, New York, NY
Francesco Petrini
Konstantinos Gkoumas
Franco Bontempi
35
Interaction analysis IP =
gr
CD
CL
O
f(IM|O)
f(IM) f(IP|IM,SP)
f(IP)
f(EDP|IM,IP,SP)
G(EDP)
f(DM|EDP)
G(DM)
f(DV|DM)
G(DV)
Hazard analysis
Interaction
analysis
Structural analysis Damageanalysis Loss analysis
IM: intensity
measure
IP: interaction
parameters
EDP:engineering
demand param.
DM:damage
measure
DV: decision
variable
Select
O, D
O:location
D:design
Environme
nt info
Decision-
making
D
f(SP|D)
f(SP)
Structural
characterization
SP:structural
system parameters
Structural
system
info
rrm
p
grr σ⋅+=
)T(log
.
)T(log
winde
windegr
⋅η
+⋅η=µ
2
5770
2 Davenport
(1983)
Reliable results for a broad
range of processes
- In the Davenport formulation the peak factor does not depend on the bandwidth of the stochastic process.
- Alternative formulations consider this dependence.
( )
≤⋅η
>⋅η
⋅η+
−
⋅η
=σ
+
+
+
+
122if650
122if
46
213
45
2
21
.T.
.T
.
)Tln(
.
)Tln(
.
windr,e
windr,e
windr,e
windr,e
gr
)ln(2
577.0
)ln(2
,
,
windre
windreg
T
Tr
⋅η
+⋅η=µ
+
+
Vanmarcke
(1975)
Models for tall buildings and the assessment of occupant comfort
rR1
2
rB
2
rR2
2
n*Srr
n (Hz)
rR1
2
rB
2
rR2
2
n*Srr
n (Hz)
Background
(broad band process)
Resonant
(narrow band
process)
Lightly damped
buildings
Highly damped
buildings
36. Damageandlossevaluationintheperformance-basedwindengineering
ICOSSAR 2013
11th
International Conference on Structural Safety & Reliability
June 16-20, Columbia University, New York, NY
Francesco Petrini
Konstantinos Gkoumas
Franco Bontempi
36
Interaction analysis IP =
gr
CD
CL
O
f(IM|O)
f(IM) f(IP|IM,SP)
f(IP)
f(EDP|IM,IP,SP)
G(EDP)
f(DM|EDP)
G(DM)
f(DV|DM)
G(DV)
Hazard analysis
Interaction
analysis
Structural analysis Damageanalysis Loss analysis
IM: intensity
measure
IP: interaction
parameters
EDP:engineering
demand param.
DM:damage
measure
DV: decision
variable
Select
O, D
O:location
D:design
Environme
nt info
Decision-
making
D
f(SP|D)
f(SP)
Structural
characterization
SP:structural
system parameters
Structural
system
info
Models for tall buildings and the assessment of occupant comfort
rR1
2
rB
2
rR2
2
n*Srr
n (Hz)
rR1
2
rB
2
rR2
2
n*Srr
n (Hz)
Background
(broad band process)
Resonant
(narrow band
process)
Lightly damped
buildings
Highly damped
buildings
Therefore, the bandwidth
parameter, and also the response
peak factor must depend on the
structural damping
37. Damageandlossevaluationintheperformance-basedwindengineering
ICOSSAR 2013
11th
International Conference on Structural Safety & Reliability
June 16-20, Columbia University, New York, NY
Francesco Petrini
Konstantinos Gkoumas
Franco Bontempi
37
Interaction analysis IP =
gr
CD
CL
O
f(IM|O)
f(IM) f(IP|IM,SP)
f(IP)
f(EDP|IM,IP,SP)
G(EDP)
f(DM|EDP)
G(DM)
f(DV|DM)
G(DV)
Hazard analysis
Interaction
analysis
Structural analysis Damageanalysis Loss analysis
IM: intensity
measure
IP: interaction
parameters
EDP:engineering
demand param.
DM:damage
measure
DV: decision
variable
Select
O, D
O:location
D:design
Environme
nt info
Decision-
making
D
f(SP|D)
f(SP)
Structural
characterization
SP:structural
system parameters
Structural
system
info
Models for tall buildings and the assessment of occupant comfort
462.2507.1265.0 2
+ξ+ξ−=µ
rg
( )
≤⋅η
>⋅η
⋅η+
−
⋅η
=σ
+
+
+
+
122if650
122if
46
213
45
2
21
.T.
.T
.
)Tln(
.
)Tln(
.
windr,e
windr,e
windr,e
windr,e
gr
( )
<≤
η
<≤
η−
=η +
+
+
1690if
690100if
380631 450
r
r
r
r
.
r
r,e
q.
.q.
.q.
r
r
r σ
σ
=+η
(Obtained from time-domain
analyses)
The peak response factor gr is characterized by a Gaussian distribution
function
rgµ
rgµ
Vanmarcke (1975)
Notas del editor
- First, I will provide an overview of the PBWE procedure, as it has been defined in several studies and journal papers by my co-author Dr. Petrini In the second part of my presentation I will briefly discuss models for the occupant comfort assessment in high-rise buildings Finally, I will introduce some considerations for the extension of the application of the PBWE framework to the Damage and Loss analysis from vibration discomfort in high-rise buildings. This is a work in process by me with my co-authors.
The structural risk is conventionally measured by the probability of exceeding a relevant value of the corresponding DV . A simplification is introduced: If the performance is expressed by the fulfillment of a limit state , and the limit state condition in terms of an EDP , the whole procedure can be simplified assuming DV = EDP .
The assessment of the serviceability of high-rise buildings under wind actions is usually carried out considering the peak values of the horizontal displacements, some measure of the acceleration . Horizontal displacements shall be limited to prevent loss of integrity to cladding and partitions ; the acceleration measure and the building natural frequencies are essential to determine the level of perception of motion , and in general the habitability issue under building vibrations. the occupant motion perception can be related to body sensation and/or visual cues; in general, the perception related to body sensation is dominant in case of low frequency vibrations (less than 2 Hz) , while the perception related to visual cues is dominant in case of relatively high frequency vibrations (greater than 2 Hz). In the right figure from Tamura ed al, comfort curves are shown (ISO and Japanese) Percentage of people that experience discomfort, confronted with literature (at any given frequency corresponds a perception threshold) Figura Tamura, curve di comfort, percezione da giaponesi e ISO Fa vedere I risultati (percentualle di persone che sentono discomfort), confrontate con letteratura, a seconda della frequenza ce soglia di percezione
The Italian code adopts the Japanese curves. What you do is enter the graph with the first natural frequency of the structure and see the acceptance rate of the accelerations.
The examined high-rise steel building has a square plan (side length: B = 50 m) and a total height of 305 m; the number of floors is 74. The main structural system is composed by a central core (a 3D frame with 16 columns) and a 3D frame on the external perimeter (28 columns). The two substructures are connected at three levels (at the height of 100 m, 200 m and 300 m); the stiffening systems are extended for 3 or 2 floors . The structural analyses have been carried out on a FE model of the building implemented in ANSYS V11; the FE model is composed by 7592 BEAM4 elements and 2680 LINK8 elements Linear dynamic analyses , assuming rigid diaphragms
Forse togliere The time series of the floor forces have been obtained by wind tunnel tests on a 1:500 scale rigid model (Fig. left ), that have been carried out at the Boundary Layer Wind Tunnel of the CRIACIV (Inter-university Research Centre on Buildings Aerodynamics and Wind Engineering) in Prato, Italy. One the other hand, in this study well consolidated analytical models has been adopted in order to carry out structural analyses in frequency domain Left: Floor forces in the along and across wind direction, evaluated at the top of the building, by scaling the experimental measures Right: Acceleration at the top of the building for a mean wind velocity at the top of the building for a V=35 m/sec (DESIGN RETURN PERIOD OF 1 YEAR according to the ITALIAN CODE/CNR 2008) – V0=20.25 m/sec
In this slide you can see the analytical model of the buffeting forces and the Vortex Shedding model . In particular, the VS effect is strong when the wind is orthogonal to the building side. The VS energy is higher. La frequenza di VS ha piu’ energia delle altre Rosso: turbolenza normale Nero VS ipotizzato tarato Blu: totle sovraposizione dei due For the examined building, time series of the floor forces were available , as derived by experimental tests (Spence et al 2008a, 2008b). However, in order to carry out probabilistic calculations, in the linear range it is preferable (at least, to reduce the computational burden) to analyze the structural response in the frequency domain , by adopting the previously introduced analytical model of the wind action.
Blu: torque Red: forces in the x and y axis
Displacements
The analysis steps are as follows: First hazard analysis: The intensity measure vector contains the following random elements V10 (10 minute velocity) Theta (the direction of the wind velocity) Z0 (roughness length) Joint Probability Density Function of the MEAN VELOCITY Finally, all previously introduced parameters has been considered random. Here the structural response in terms of across wind peak accelerations is still represented as function of the V10. I passi dell’analisi Distribuzione V10 di cui parametri dipendono da theta Dal database di NIST, Dal mare coef rug basso Z0 dipende da theta
For the interaction analysis vector, 3 parameters are chosen The peak response factor And two aerodynamic coefficients
The response parameter is given as a sum of the average value plus the peak response factor times the variance
Riassunto Edp accelerazione di picco across wind
Figure Complementary cumulative function of the EDP
Metto TMD tre parametri beta rapporto di frequenze Analisi parametrica variando I 3 rapporti per scegliere Risultati per gamma fissato, alvariare di beta… risposte massime Basso a destra come cambia la curva di rischio TAPEI 101 In order to optimize the structural response, the insertion of a TMD can be considered. Here the TMD has been inserted at the top of the building and the structural response is represented in terms of occupant comfort. The TMD produces two main effects: the variation of the structural natural frequency, and a reduction in maximum the response. Here the maximum across wind acceleration obtained for different sets of TMD design parameters are shown and compared with the comfort thresholds. Each figure represents a different mass ratio “gamma”, different markers represent different natural frequency ratios “beta” and different points with the same marker represent different damping ratios “csi”. The maximum effect is obtained by “gamma” equal to 1/150, “Beta” equal to 1 and “csi” equal to 10%
In this part I will provide some considerations for the application of the PBWE framework to the Damage and Loss analysis due to the occupant discomfort in high-rise buildings
So far, the research focused on the Hazard, Interaction and Structural analysis for the occupant comfort. In order to apply the PBWE framework to the damage and loss analysis from vibration comfort, some additional considerations are necessary. In this slide are reassumed the major issues (from a paper by Kwok et al) Mitigation measures in particular are proposed in literature in the form of among else the installation of vibration control devices
Left Distribution of comfort ratings from occupant comfort tests conducted in motion simulator Right Comparison of occupant comfort serviceability criteria for 1 year return period wind storm . The uncertainty can be modeled considering the different density of the lines at different frequencies.
Direct costs were in the past considered the tangible costs. Now also business/service interruption should be considered. Indirect costs should consider indirect consequences (in the supply chain) Need to focus on a case-by-case basis.
Questo si fa di solito ma non si puo fare in questo caso For limit states that damage at physical components doesn’t occur, as fo example discomfort in building occupants, the big issue is how to calculate the losses
Se la qualita’ non e’ accetabile per residenziali ma lo e’ per uffici Se non e’ accetabile neanche per uffici o se non voglio cambiare la destinazione d’uso, installo TMD Indirect
So far, the research focused on the Hazard, Interaction and Structural analysis for the occupant comfort. In order to apply the PBWE framework to the damage and loss analysis from vibration comfort, some additional considerations are necessary. In this slide are reassumed the major issues (from a paper by Kwok et al) Mitigation measures in particular are proposed in literature in the form of among else the installation of vibration control devices For direct cost, the surface of the floor thatare occupied by the TMD should be considered. Direct: differenza di superficie In the case of retrofitting: direct losses are in terms of m2 times m2 cost + cost of instalation perdite dirette differenza di metri quadri e costo di istalazione ( Sismica: perdita dirette sono calcolate con il costo degli elementi struturali e non rotti Vento: se non c-e rottura di niente? Faber e Ciampoli: LQI Concetto: non riusciamo a calcolare le perdite dirette usualmente si basano su un danno fisico… Quinon c’e nessun danno fisico. Se ci sono come si fa a calcolare? Giorni di chiusura per installare il TMD o per attuare la soluzione del programma, costo del TMD: associ il costo della cura. Della soluzione piu economca possibile (e.g. liquid mass damper). Io calcolo le perdite dirette come spesa per togliere quel effetto.
So far, the research focused on the Hazard, Interaction and Structural analysis for the occupant comfort. In order to apply the PBWE framework to the damage and loss analysis from vibration comfort, some additional considerations are necessary. In this slide are reassumed the major issues (from a paper by Kwok et al) Mitigation measures in particular are proposed in literature in the form of among else the installation of vibration control devices For direct cost, the surface of the floor thatare occupied by the TMD should be considered. Direct: differenza di superficie In the case of retrofitting: direct losses are in terms of m2 times m2 cost + cost of instalation perdite dirette differenza di metri quadri e costo di istalazione ( Sismica: perdita dirette sono calcolate con il costo degli elementi struturali e non rotti Vento: se non c-e rottura di niente? Faber e Ciampoli: LQI Concetto: non riusciamo a calcolare le perdite dirette usualmente si basano su un danno fisico… Quinon c’e nessun danno fisico. Se ci sono come si fa a calcolare? Giorni di chiusura per installare il TMD o per attuare la soluzione del programma, costo del TMD: associ il costo della cura. Della soluzione piu economca possibile (e.g. liquid mass damper). Io calcolo le perdite dirette come spesa per togliere quel effetto.
Media datta da Davenport. ‘E affidabile solo per processi stocastici a banda larga