Passive monitoring of actual buildings
Philippe GuéguenISTerre @ Joseph Fourier University Grenoble
Coll. C. Michel (ETHZ), P. Johnson (LANL), P. Roux (ISTerre), T. Kashima (BRI-Japan), ...Students: A. Mikael (former PhD), M.A. Brossault (PhD), A. Astorza, M. Munoz..
Continuous aging and subsequent structural deterioration of a large number of existing structures
Structural Health Monitoring
Motivation
International Energy Agency (IEA 2001) : “in the absence of changes in policy on nuclear energy, the lifespan of plants is the most determining factor for nuclear power for the next decade".
Risk of collapse of the administrative building center in Nice "certainly unpredictable in time or in terms of probabilities, but there exists a real emergency to evacuate
Condition Based Maintenance
Conditional
Planed
Corrective
Regressive
Let it deteriorate
Let it deteriorate
Repair before
failure - Lessdeterioration
Repair before
failure -D
eterioration elim
ination
Action
Man
agem
ent
effe
ctiv
enes
s
Cost
Motivation
To prevent the total interruption of production activity by anticipating the renovation operations engaged on the basis of the observed structural condition
Seismic crisis management
Motivation
Iervolino et al., Earthq. Spectra, 2014
Progressive damage and change of condition of structures during aftershock sequences
MotivationDesign-related uncertainties in Damageand risk prediction
€
P d > ds( ) A[ ] =121+ erf ln(A) − ln(µ)
σ 2$
% &
'
( )
$
% &
'
( )
Motivation
σ2= σ2IM + σ2EDP + σ2Mod σMod = Building-to-building variability within a building class (e.g variability of building properties and design for a given typology)
- Default of design - Aging or damage changing the model
- Boundary condition - Non linear response
Part 1: Motivation
Part 2: Structural Dynamics
Part 3: Experimental assessment of dynamic parameters
Part 4: Passive imaging of buildings
Part 5: Non-linear seismic response of buildings
Part 6: Natural wandering of the building response
Presentation Outline
F = k.Uk = 3.E.I/L3 for fixed base model
E,I
F=m.Ü
U
L
F [N] ForceU [m] Displacementk [N/m] StiffnessÜ [m/s2] Acceleration
E= Young modulus - MaterialI= Inertia momentum - Geometry/shear resistance
Structural dynamics
FU
U F = k.U U=f(t) ü = f(T,earthq.)
k, ζ =f(E,I,H, boundaries conditions)
Static Dynamic Dynamic Free-oscillations Forced Oscillations
ζ
k
F
üg
Structural dynamicsSingle-Degree-of-Freedom - SDOF
natural frequency
% of critical damping
üg: Seismic ground motion (# seismic regulation)
U ü = f(T,earthq.)
k, ζ =f(E,I,L, boundariesconditions)
ζ
k
Equation of motion - SDOF
Structural dynamics
Frequency response of SDOF (with fo const., varying ζo)
0 5 10 15 20−1
−0.5
0
0.5
1
Ampl
itude
2 %
0 5 10 15 20−1
−0.5
0
0.5
1
5 %
0 5 10 15 20−1
−0.5
0
0.5
1
Times − sec
Ampl
itude
10 %
0 5 10 15 20−1
−0.5
0
0.5
1
Times − sec
20 %
Time response of SDOF (with fo const., varying ζo)
1
100
Frequency ratio
Ampl
ifica
tion
fact
or D
0 %5 %10 %20 %100 %
3dB = A/√2
f2f1
Structural dynamics
Wet spongeFlexible-base building
Dry spongeFixed-base building
Structural dynamics
(k , c )hx hx
(k , c )hxry hxry
(k , c )1 1
m J0 0
m J1 1
(k , c )ry ry
xg x1x0
y
H y
H
a) Chargement harmonique (type séisme) b) Chargement ponctuel (type essai de lâcher)
x1x0
y
H y
F0
φ
φ
F=KUK = kstat (kx+iωcx)
m1 : m1(x1 + x0 +H�y
) + c1x1 + k1x1 = �m1xg
m0 : m0x0 + chx
x0 + khx
x0 + chxry
�y
+ khxry
�y
� c1x1 � k1x1 = �m0xg
J0�y
+ cry
�y
+ kry
�y
+ cryhx
x0 + kryhx
x0 �Hc1x1 �Hk1x1 = 0
Xg=input seismic motionX0=relative motion of m0X1=inertial motion of m1Φ=rocking motion
k1,c1: stifness and damping coefficients of the building (fixed-base)
khx,chx: impedance coefficients for translation kry,cry: impedance coefficients for rockingkhxry,chxry: impedance coefficients for translation/rocking coupling
(k , c )hx hx
(k , c )hxry hxry
(k , c )1 1
m J0 0
m J1 1
(k , c )ry ry
xg x1x0
y
H y
H
a) Chargement harmonique (type séisme) b) Chargement ponctuel (type essai de lâcher)
x1x0
y
H y
F0
φ
φ
Structural dynamics
X(ω)
frequency
β1
β3
β2
S-wave velocity β1< β2< β3
{X} (�) = {�m}Xg(�)�2
[Kt]� [M ]�2
1
e!2=
1
!20
+1
!21
+1
!2✓
1e⇣=
⇣!0
e!
⌘3 1
⇣0+
⇣!1
e!
⌘3 1
⇣1+
⇣!✓
e!
⌘3 1
⇣✓
Stewart and Fenves, 2005
Structural dynamics
u(t) un(t)
ui(t)
u1(t)
mn
kn cn
mi
ki ci
m1
k1 c1
A more realistic representation of real structures is: a multi-degree-of-freedom system MDOF
Equation of motion of MDOF
n floor system has n natural frequencies (eigenvalues) and n associated modeshapes (eigenvectors)
Courtesy G. Hivin UJF
Structural dynamics
Orthogonality of the modes
Structural dynamics
Structural dynamics
Euler-Bernoulli beam (Cantilever or bending beam)
Shear beam
Timoshenko beam
C =EI⇡2
4KSH2
Constant mass and stiffness
@
2u(x, t)
@x
2=
m
KS
@
2u(x, t)
@t
2
EI
@
4u(x, t)
@x
4+m
@
2u(x, t)
@t
2�m
EI
KS
@
4u(x, t)
@x
2@t
2= 0
Boutin et al., Earthq. Engng. Struct. Dyn, 2005
Part 1: Motivation
Part 2: Structural Dynamics
Part 3: How to we get the dynamic properties ?
Part 4: Passive imaging of buildings
Part 5: Non-linear seismic response of buildings
Part 6: Natural wandering of the building response
Presentation Outline
Asynchronous shaker at the top of the building.
J. L. ALFORD and G. W. HOUSNERA dynamic test of a four-story reinforced concrete building
Bulletin of the Seismological Society of America, Jan 1953; 43: 7 - 16.
How to we get the dynamic properties ?
Bob Barret web page
How to we get the dynamic properties ?
Michel et al., SDEE, 2008
Air blast or explosion
Boutin et al., EESD, 2005
Shocks
How to we get the dynamic properties ?
1906: San Francisco earthquake, California. 1923: Great Kanto earthquake, Japan.
US Intrumentation Programs:Californian Strong Motion Instrumentation Program (1960)National Strong Motion Program (USGS) California Strong Motion Instrumentation Program (CGS)Buildings ~ 225 (1-62 stories high); Code Buildings: instrumented by owner
Hollywood Storage Building:Longest history of recording in U.S. – 80 years. First record in 1933 (So. California earthquake).Factor Building:70 sensors in one building - On-line available data - Continuous recordingsMilikan Library:damaged building after the San Fernando earthquake (1971) + permanent shaker at the top
How to we get the dynamic properties ?
French Intrumentation Program:French Acceletometric Networks (1995)National Building Array Program (2004) 5 Buildings - low seismicity - low sensibility - continuous recordings - 3 buildings in France - 2 in French west indies including one isolating rubber bearing system - 24 channels + free-field + additional sensors
Instrumentation programs in Taiwan, Romania, Greece, Italy, Turkey ... but data are not available online
Japaneese Intrumentation Programs:Building Research Institute BRIBuilding array from 40’1964 Nigata Earthquake - 2012 Tohoku Earthquake74 buildings (top/bottom + boreholes st)
How to we get the dynamic properties ?
Celebi, 1998
How to we get the dynamic properties ?
Dynamic and frequency ranges of the typical sensors overlain on a bandpassed signal amplitude plot
➪ the first were analogue / 12bits digitaltriggered accelerometers. Recordingscontinuously misses moderate strong motion and ambient vibrations
Bradford, PhD thesis 2006
Peterson (1992) Low and High noise model
How to we get the dynamic properties ?
Input/output modal analysis!Input force is measured!Output force is measured!Output is related to input through an Impulse Response Filter!IRF is independent of the input force
StructureInput signal Output signal
Time domain : Out(t)=I(t)*h(t)+n(t)Frequency domain : H(ω)=Out(ω)/I(ω)
How to we get the dynamic properties ?
Deconvolution (water-level) Clayton and Wiggins, J. Royal Astr. Soc. 1976
Dynamic and frequency ranges of the typical sensors overlain on a bandpassed signal amplitude plot
➪ the first were analogue / 12bits digitaltriggered accelerometer. Recordingcontinuously misses moderate strong motion and ambient vibrations
➪ 24 bit accelerometer (begining of 90’s) continuoulsy recording can record all ranges of vibration in a structure
How to we get the dynamic properties ?
Bradford, PhD thesis 2006
City-Hall Grenoble
Daily variations of vibration energy
May, 2009
Output-only modal analysis : Ambient vibrations
How to we get the dynamic properties ?
!
Low cost but less sensitive.But all types of recordings: torsion, gyroscope, GPS etc...
Courtesy T. Heaton
Micro-Electro-Mechanical Systems
Episensor noise level
Phidgets noise level
Android Phone noise level
101
100
10-1
10-2
10-3
10-4
10-5
10-6
10-7
10-8
10-3 10-2 10-1 100 101 102
Frequency - Hz
Acc
eler
atio
n - m
/s2
M7.5 max
M7.5
M6.5
M5.
5M
4.5
M3.
5 M2.
5
M1.5
M7.5
M6.5M5.5M4.5M3.5
M2.5
M1.5
Near Source - 10kmRegional Source - 100km
How to we get the dynamic properties ?
Mesure LIDAR des vibrations ambiantes • Portée de 200 m • Cartographie bâtiment par tourelle • Acquisition de 10 min
LIDAR Measurements (optical technology)
Gueguen et al., Bull. Earthq. Engng., 2010Valla et al., ASCE Civil Engng,2014
Spectres de vibration, fréquence dopplerPrincipales fréquences d’intérêt
1.2 km
Long-term monitoring of buildings from one remote reference site: successful measurements up to 5 km!
How to we get the dynamic properties ?
Part 1: Motivation
Part 2: Structural Dynamics
Part 3: How to we get the dynamic properties ?
Part 4: Passive imaging of buildings
Part 5: Non-linear seismic response of buildings
Part 6: Natural wandering of the building response
Presentation Outline
Carder D. S. Observed vibration of buildings. Bulletin of the Seismological Society of America 1936; 26:245–277.
0 50 100 150 200 250 300 350 400−2
0
2x 10
−5
Time − secAm
plit
ud
e −
co
un
t
100 110 120 130 140 150 160 170 180 190 200−2
0
2x 10
−5
Time − sec
10−1
100
101
0
0.5
1
1.5
Frequency − Hz
FF
T
Average FFT: 21 windows 8192 pointsFs=200Hz
Belledonne towerIle Verte
Not a new technique:
Output only - Parametric
Passive imaging of buildings
Φ/Φmax
Mode 1 L-Direction
Mode 3 L-DirectionMode 2 L-Direction
# St
orey
# St
orey
0 1 2 3 4 5 6
4 5 6 7
# St
orey
11 12 13 Frequency - HzFrequency - Hz
f1 = 1.9 Hz - f2 = 6.1 Hz - f3 = 11.4 Hz=== > f2 / f1 = 3.2 - f3 / f1 = 6.0
# St
orey
Passive imaging of buildings: Peak-pickingTimoshenko beam
Experimental modelling
Numerical modelling
Michel et al., Earthq. Engng. Struct. Dyn., 2009
Output-only - non-parametric
Passive imaging of buildings: SV decomposition
0.4 1.2 2.0
Passive imaging of buildings: SV decomposition
Time (s)
Hei
ght (
m)
Model
Passive imaging of buildings: SI by deconvolution
V0 =�i,i+1
�
Snieder and Safak, 2006
Time (s)
Hei
ght (
m)
Model
Building testing
Passive imaging of buildings: SI by deconvolution
H
H
Courtesy of M. Todorovska
Passive imaging of buildings: SI by deconvolution
Part 1: Motivation
Part 2: Structural Dynamics
Part 3: How to we get the dynamic properties ?
Part 4: Passive imaging of buildings
Part 5: Non-linear seismic response of buildings
Part 6: Natural wandering of the building response
Presentation Outline
Strain
Dy= 2.10 -3 - 5.10-3
Hazus (US) methods to predict earthquake-based damage:Dy= 2.10 -3 - 5.10-3
Elastic Plastic
Non linear seismic response
Non linear seismic response
IG−EPN building − Colombia M 7.3 earthquakeFr
eque
ncy
(Hz)
0 100 200 300 400 500 600 700 800 9000.5
1
1.5
2
2.5
3
0 100 200 300 400 500 600 700 800 900−5
0
5 x 104
Time (s)
Acce
lera
tion
(cm
/s2)
IG-EPN building - Colombia M 7.3 earthquake
f0
fmin
Non linear seismic response
Clinton et al., BSSA, 2006
Non linear seismic response
BRI Annex building
In collaboration with Pr Kashima BRI Japan
Non linear seismic response
Steel-frame RC Japanese buildings
Tohoku earthquake
Non linear seismic response
Non linear seismic response
Years1998 2001 2004 2007 2009 2012
Freq
uenc
y by
dec
onvo
lutio
n
0.40.60.8
11.21.41.61.8
22.2
-5
-4
-3
-2
-1
0
1
normalized frequency1998 2001 2004 2007 2009 2012
Vs b
y de
conv
olut
ion
100
150
200
250
-5
-4
-3
-2
-1
0
1
Years
log(Max Acc.)
V0
Year1998 2001 2004 2007 2009 2012
Freq
uenc
y by
dec
onvo
lutio
n
0.5
1
1.5
2
Year09/2010 04/2011 10/2011 05/2012 11/2012 06/2013 01/2014Fr
eque
ncy
by d
econ
volu
tion
0.8
1
1.2
1.4
Non linear seismic responseSampling: N=25 events
Non linear seismic response
Years03/2010 09/2010 04/2011 10/2011 05/2012
Freq
uenc
y by
dec
onvo
lutio
n
1.05
1.1
1.15
1.2
1.25
1.3
1.35
1.4
Frequency by deconvolution1 1.2 1.4 1.6 1.8 2
Top
Acce
lera
tion
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45 ×105
7.3
7.31
7.32
7.33
7.34
7.352012
2009
2007
2004
2001
1998
Sampling: N=25 events
Non linear seismic response
Vo[m/s]400 450 500 550 600
Stra
in
10-8
10-7
10-6
Freq 0.50 @ 7.50 Hertz
µ= 504.79 σ= 11.52 σ/µ=2.3%
Vo[m/s]100 120 140 160 180 200 220
Stra
in
10-7
10-6
10-5
10-4
10-3 Freq 0.30 @ 3.00 Hertz
µ= 161.55 σ= 4.76 σ/µ=2.9%
Michel et al., EESD, submitted
Part 1: Motivation
Part 2: Structural Dynamics
Part 3: How to we get the dynamic properties ?
Part 4: Passive imaging of buildings
Part 5: Non-linear seismic response of buildings
Part 6: Natural wandering and nonlinear elasticity
Presentation Outline
Filtered signal - 1 hour
Time windows - !
RDT signature
N windows having the same initial condition intial condition found in the signal
0 10 20 30 40 50 60
Am
plit
ud
e -
co
un
ts
Time - s
2
1
0
-1
-2
x 10-5
1+...+ 6+...+N
[500T - 1000T]
Random Decrement Technique RDTstationnary and white noise conditions of ambient vibrationsTo stack a large number of short windows having the same triggering conditionMikael et al., BSSA, 2013
- Band-pass filtering Butterworth (order 4)[f1 - f2] at ΔA > 3dB
- N windows <=> 500-1000 T - 20 minutes- τ= 10T - 5 secondes- Triggering conditions: v(t)=0, u(t)>0
Natural wandering
Mikael et al., BSSA, 2013
Natural wandering: example 1 City Hall Grenoble
Guéguen et al.,SCHM, submitted
OGH1OGH3
OGH4OGH6
OGH2
OGH5
44 m
49 m
13 m
1
e!2=
1
!20
+1
!21
+1
!2✓
Natural wandering: example 1 City Hall Grenoble
Guéguen et al.,SCHM, submitted
ῶ redω black
The Anne Paul’s buildings
Natural wandering: example 2 Anne Paul building
Quasi-static events: environmental loading
L
T
Days07-May09 15-Aug09 23-Nov09 03-Mar10 11-Jun10 19-Sept10 28-Dec10 07-Apr11
2.75
2.70
2.65
2.60
2.55
Freq
uenc
y - H
z
3.38
3.37
3.36
3.35
3.34
Freq
uenc
y - H
z
Δf +/-4% L Δf +/-0.6 T
Natural wandering: example 2 Anne Paul building
Summer time
Non-linear elasticity of building response to slow loadingTe
mpe
ratu
re °C
Frequency - Hz
Natural wandering: nonlinear elasticity
Long-term (seasonal) variations - Smoothed with a sliding windows of 120 hours length
Quasi-static events: environmental loading
2.75
2.70
2.65
2.60
2.55
Freq
uenc
y - H
z
Days
07-May09 15-Aug09 23-Nov09 03-Mar10 11-Jun10 19-Sept10 28-Dec10 07-Apr11
2.2
2.0
1.8
1.6
1.4
1.2
1.0
0.8
0.6
Dam
ping - %
Natural wandering: example 2 Anne Paul building
Strain = max[Utop
(t)� Ubot
(t)
H]
Factor building (IRIS web service)
Data from Sept. to Dec. 2004
2004/09/28 - Parkfield Earthquake Mw=6.0 200 km far from the building
Top and bottom recordings
Natural wandering: example 3 factor Building (CA)
0.55
0.6
Freq
. [Hz
]
0
4
8
Dam
p. [%
]
02040
Tem
p. [°
C]
050
100
Hum
. [%
]
Oct. 04 Nov. 04 Dec. 040
1632
Days in 2004
Win
d. S
p.
[km
/h]
UCLA Factor building
Meteorological Station: DOWNTOWN L.A./USC CAMPUS
Event 428/11/2004
Event 506/12/2004
Event 331/10/2004
Event 128/09/2004
Event 223/09/2004a)
c)
d)
e)
b)
Dynamic - Event 1 - 2004/09/28 - Parkfield Earthquake Mw=6.0 - R=200 kmStatic - Events 2 to 5 - Elastic variation controlled by atmospheric conditions Clinton et al., BSSA, 2006; Todorovska and Al Rjoub, SDEE, 2006; Todorovska, BSSA, 2009; Mikael et al., BSSA 2013
Guéguen, Johnson, Roux, will be published next decade
Dynamic Event Event 1 Top disp: 2 mm Drift: 3 10-5
Static Event Event 3 Top Disp: 0.5 mm Drift: 4 10 -6
Natural wandering: example 3 factor Building (CA)
Guéguen, Johnson, Roux, will be published next decade
Hazus : Yield capacity Dy=2.10 -3 - 5.10-3
Nonlinear elasticity at Factor building compared to laboratory scale resultsNatural wandering: example 3 factor Building (CA)
Modified fromJohnson and Sutin , JASA 2005
Guéguen, Johnson, Roux, will be published next decade
Hazus : Yield capacity Dy=2.10 -3 - 5.10-3
−0.1 −0.05 0
10−7
10−6
10−5
10−4
∆f/fapp
∆/H
0 2 4 6
10−7
10−6
10−5
10−4
∆ζ/ζapp
∆/H
0 2 4 6x 10−3∆ζ/ζapp
−10 −5 0x 10−3∆f/fapp
Event1Event2Event3Event4Event5
PySPAMQG
a) b) c) d)
Natural wandering: example 3 factor Building (CA)Nonlinear elasticity at Factor building compared to laboratory scale results
Mod
ified
from
John
son
and
Sut
in ,
JAS
A 20
05
Guéguen, Johnson, Roux, will be published next decade
Conclusions
Building testing thanks to new instrumentation
- physical origin of frequencies and damping in actual buildings- for SHM and CBM
NL elasticity and slow dynamic
- probing the invariant scale of NL elasticity (laboratory mm Johnson, building m, earth km Brenguier)
- related to the structural health
Philippe Guéguen ISTerre/[email protected]