California Geological Survey – “Probabilistic Tsunami Modeling and Public Pol...
Seismic Instrumentation of Structures: What Have We Learned? - Mehmet Çelebi
1. Seismic Instrumentation of
Structures: What Have We Learned?
Mehmet Çelebi, Earthquake Hazards Team,
Menlo Park, CA.
2. OUTLINE
• Statement of Purpose
• US Inventory of Instrumented Structures
• What is not covered?
• What is covered?[5 topics as time allows]
1. Long-period/long-distance effects (Tohoku data)
2. Effectiveness of dynamic response modification
features (Tohoku data)
3. Real-time monitoring advances
4. State-of-the-art hardware/advances in softwares
5. British Columbia (Smart Infrastructure Monitoring
system for Bridges)[from Carlos Ventura]
• Conclusions
2
3. Simple Statement of Purpose for
Structural Instrumentation
The main objective of a
Seismic
Instrumentation
Program for structural
systems is to improve
our understanding of
the behavior and
potential for damage
of structures under the
dynamic loads of
earthquakes 3
4. As a result
• design and construction practices and building
codes can be improved
• life and property losses can be reduced
• informed decisions can be made
– (a) in emergency response situations, and/or
– (b) to retrofit or strengthen the structural system
– (c) to secure the contents within the structures
• The requirement for accomplishing this aim is
to expand and upgrade the existing network of
instrumented structures to record their
responses during earthquakes. 4
5. Structural Monitoring Programs
• State of California (CGS)-CSMIP
• USGS (Cooperative Program with other
agencies, local governments and
organizations)
– DATA & INFO FROM BOTH CGS &
USGS at CESMD (Combined Engineering
Strong Motion Data) web site:
http://www.strongmotioncenter.org
• Private Owners (e.g. Banks, Industry )
5
6. Status-quo Inventory:
USGS-Cooperative Structural Monitoring Program
[USGS,ACOE,VA,GSA,UPR,CITY OF SF.ODAT.MWD,JPL]
[note: 3 channels or more within a structure, updated Dec.20, 2012]
and
CGS(CSMIP) Program [Updated 2012] (info compiled from various CGS
publications & http://strongmotioncenter.org]
6
7. What is NOT covered?
• Use of recorded data to establish code
period formulas (e.g. see Goel-Chopra
papers, Willam Jacobs paper)
• Successes of past analyses of recorded
responses (Transamerica – Rocking
identified?)
• Routine methods to analyze recorded
responses (f, damping, mode shapes).
• Past data analyses from bridge responses
(Golden Gate, Cape Girardeau Bridge)
• Wireless instrumentation (still not viable!).
7
8. What is covered [5 topics]?
1. Long-period/long-distance effects (Tohoku
data) & what we learned!
2. Effectiveness of dynamic response
modification features (Tohoku data)
3. Real-time monitoring advances
4. State-of-the-art hardware/advances in
softwares
5. British Columbia (Smart Infrastructure
Monitoring system for Bridges)[from Carlos
Ventura]
8
9. ITEM 1
LONG PERIOD LONG DISTANCE
EFFECTS & IMPLICATIONS
Japanese Data from Tohoku (3/11/2011)
M=9.0 event
TWO CASES:
Building A: ~770 km from epicenter
Building B:~350-375 km from epicenter
9
10. Background (Long Period/Long Distance Effects) 1/2
• One of the earliest observations in the United States was during the
M=7.3 Kern County earthquake of July 7, 1952, that shook many
taller buildings in Los Angeles and vicinity, about 100-150 km
away from the epicenter
(http://earthquake.usgs.gov/earthquakes/states/events/1952_07_21.php)
• One of the most dramatic examples of long-distance effects of
earthquakes is from the September 19, 1985, Michoacan, Mexico,
M 8.0 earthquake during which, at approximately 400 km from the
coastal epicenter, Mexico City suffered more destruction and
fatalities than the epicentral area due to amplification and
resonance (mostly around 2 sec) of the lakebed areas of Mexico
City (Anderson and others, 1986, Çelebi and others, 1987).
10
11. Why important?
Potential Long Period/Long Distance Effects in the US (2/2)
• Tall buildings in Los Angeles area from Southern
California earthquakes,
• Tall buildings in Chicago from NMSZ,
• Tall buildings in Seattle (WA) area from large Cascadia
subduction zone earthquakes).
• Let us remember that the recent M=5.8 Virginia
earthquake of August 23, 2011 was felt in 21 states of
the Eastern and Central U.S., that include large cities
such as New York and Chicago
(http://earthquake.usgs.gov/earthquakes/eqinthenews/20
11/se082311a/#summary, July 15, 2011).
11
12. What we learned from Tohoku EQ?
Why important?
Discussion of INTERRUPTED functionality of
buildings at low ground level input motions caused by
event at far distances (e.g. 770km). Possible hidden
damages?
Discussion of what may happen at larger input motions
with similar frequency content.
Discussion of DRIFT RATIOS w.r.t. codes (Japan,
USA, Chile)
Two cases (others are not presented):
Building A (770 km from epicenter)
Building B (350-375 km from epicenter)
12
13. Building A; in Osaka Bay ~770 Km from
epicenter of March 11, 2011 main-shock
• 256 m tall (55 stories+3 story basement)
• Construction finished in 1995 (pre-1995 code,
pre-(KIK-NET/K-NET). Vertically irregular,
steel, moment-frame (rigid truss-beams/10
floor). No shear walls around elevator shafts
13
• 60-70 m long piles below foundation
15. Closest Free-Field Station:
OSKH02 (KIK_NET)
Record indicates
[(a) site frequency from
actual data (~.14-.18 Hz)
& (b) shaking duration]
15
16. Site Info from OSKH02 and building site
indicate similarities [resulting in similar site
frequency transfer function] and also similar to
that from strong shaking data (previous slide]
[f(site)~0.13-0.17 Hz]. Also expected to be
similar to freq of 55 story bldg..RESONANCE!
16
17. ACCELERATIONS
RECORDED
(MAIN-SHOCK)
Note long duration
of record and strong
shaking
17
22. AVERAGE DRIFT RATIO
• Why average drift ratio?
• Sparse instruments
• ~.005 (or ~.5%) drift ratio
(X-Dir)
• Implications(!!): 3%g
input motion, ~.5% drift
ratio: not-acceptable
22
23. Building A:CONCLUSIONS AND REMEDIES
• Building lost functionality for many days due to elevator cable entanglements
and other problems.
• Resonance occurred and still occurs because f(building)~f(site)
• Damping is too low [ambient tests could have provided some clues]
• (average) Drift Ratios are high for a 3% g input motion. What if input a>.2g?
• Sparse instrumentation
• Implications (US): tall buildings in Chicago, NY, Boston from far sources
• Structural Response Modification Technologies (being designed – see below)
23
24. Building B: 55-Story Shinjuku Center Building in
Shinjuku, Tokyo (~350-375 km from Epicenter)
According to EERI Special Earthquake Report (EERI Newsletter, 2012), the 54-
story Shinjuku Center Building was constructed in 1979. The report states:
“The structure’s height is 223m, and the first natural period of the structure is
5.2 and 6.2 seconds in two perpendicular directions. The dampers were
calculated to have reduced the maximum accelerations by 30% and roof
displacement by 22% “.
25. Building B: Shinjuku Center Building
• Figure courtesy of J. Moehle and Y. Sinozaki (Taisei Corp)
• Recorded first story acceleration (BLUE~max ~0.15g).
• Roof level displacement time history (RED : max~54 cm) Most notable
is the long duration motion over 10 minutes.
• AVERAGE DRIFT RATIO: 54/21600 = ~0.25% < 1% according to
Japanese practice.
• However: the actual drift ratios computed from relative displacements
divided by story heights between some of the pairs of two consecutive
floors are certainly to be larger than the average drift ratio computed
using the maximum roof displacement divided by the height of the 25
building
26. CONCLUSIONS [what we learned?] (1/2)
• 1. For small ground level input ground motions as in the
two cases presented herein, these two tall buildings
deformed significantly to experience sizeable drift ratios.
• 2. Collection of such data is essential (a) to assess the
effect of long period ground motions on long period
structures caused by sources at large distances, and (b) to
consider these effects and discuss whether the design
processes should consider reducing drift limits to more
realistic percentages (c) finally, further applications of
unique response modification features are feasible to
reduce the drift ratios.
26
27. CONCLUSIONS [what we learned?] (2/2)
• 3. Behavior and performances of these particular tall
buildings far away from the strong shaking source of the
M9.0 Tohoku earthquake of 2011 and large magnitude
aftershocks should serve as a reminder that, in the United
States as well as in many other countries, risk to such
built environments from distant sources must always be
considered.
• 4. The risk from closer large-magnitude earthquakes that
could subject the buildings to larger peak input motions
(with similar frequency content) should be assessed in
light of the substantial drift ratios under the low peak
input motions experienced during and following the
Tohoku earthquake of 2011. 27
28. DRIFT RATIO (DR): CODES
• JAPAN: Max DR=1% for collapse protection
level motions (level 2 used for buildings 60 m or
taller [The Building Center of Japan, 2001a and
2001b]).
• UNITED STATES: MAX DR=2% for tall
buildings for Risk Category 1 or 2. (Table 12.12,
ASCE7-10, 2007).
• CHILE: MAX DR=0.2% [The Chilean Code for
Earthquake Resistant Design of Buildings
(NCh433.Of96, 1996) [effective since 1960’s]
28
29. What is the risk to tall buildings from earthquakes
originating at NMSZ or Charleston, SC or Cascadia
Subduction Zone? Should we consider what may happen to
tall buildings in Chicago, New York, Boston or Seattle?
29
30. Tall Buildings in Chicago, Boston and Seattle
According to Wikipedia:
In Chicago: 72 bldgs taller than 555ft (168 m) [37-108] stories
In Boston :27 buildings taller than 400 feet (120 m). “
In Seattle, 15 buildings >400 ft(122m), 24 buildings>400 ft under constrruction
• How will they perform during a strong event from distant sources??? [pictures from Wikipedia]
30
31. ITEM 2
SUCCESSES OF STRUCTURAL
DYNAMICS MODIFICATION FEATURES
(in light of data from Tohoku earthquake)
(a) Base-isolated
(b) Dampers
31
32. Seismic Base-Isolation
• Since 1980’s… [similar to soft first story]
• US (today) 125 bldgs, ~ dozen bridges
• Japan (today) 6500 bldgs and ~ 6500 bridges
(from Taylor & Aiken, Structure [ASCE], March
2012 issue)
• In Tokyo alone, ~1500 tall buildings, ~1000 base-
isolated buildings: Performances: Excellent! 32
33. In the USA (best data to date from):
USC_BASE ISOLATED HOSPITAL
[from CGS network]
33
34. THE RECORDS RETRIEVED
FROM THE BASE-ISOLATED USC
HOSPITAL [LARGEST ACCELERATIONS
AND DISPLACEMENTS EXPERIENCED,
TO DATE, FOR A BASE-ISOLATED BUILDING]
CONFIRMED THE EFFECTIVENESS OF
BASE-ISOLATION TECHNIQUES
(0.21 g)
(0.13 g)
(0.37g)
(0.49 g)
34
35. PERFORMANCE OF BASE-ISOLATED USC
HOSPITAL DURING THE NORTHRIDGE EQ.:
AT DESIGN LEVEL EQ., ONLY 10% OF
DISPLACEMENT CAPACITY OF ISOLATORS
REACHED. NO DAMAGE
35
36. From Japan:
Base-isolated buildings (Tohoku Event M=9)
(a) 7-story bldg in Tsukuba [~350 km from
epicenter]
[BI: ~.33g, AI=.09g, 6th Fl=.125g]
36
41. CONCLUSIONS [what we learned?]
• IN US, with example of USC Hospital data
during shaking at design level, and
• IN JAPAN: Numerous examples with base
isolation and other features
reduced responses significantly to result in
excellent performance of the buildings.
41
42. ITEM 3
NEW DEVELOPMENTS:
REAL-TIME MONITORING
USING IP (INTERNET PROTOCOL) :
Using masured data from
accelerometers &
computing displacements and
drift ratios in near real-time
42
43. THE PROBLEM:
• The owner needs reliable and
timely expert advice on whether
or not to occupy the building
following an event.
• Data gathered will enable the
owner to assess the need for post-
earthquake connection
inspection, retrofit and repair of
the building.
43
44. Drift vs. Performance
• The most relevant parameter to assess performance is the
measurement or computation of actual or average story drift
ratios. Specifically, the drift ratios can be related to the
performance- based force-deformation curve hypothetically
represented in Figure 1 [modified from Figure C2-3 of
FEMA-274 (ATC 1997)]. When drift ratios, as computed
from relative displacements between consecutive floors,
are determined from measured responses of the building,
the performance and as such “damage state” of the building
can be estimated as in the figure (below).
44
50. Commercial Versions installed at some banks
[3 components: (1) sensors, (2) DAQ’s &
Processors, (3) Display&Alarm system for
informed decisions]
(Figure from D. Sokolnik)
50
51. CONCLUSIONS [what we learned?]:
• Measured data from carefully crafted
monitoring are being successfully used (in
US, Japan and other countries) to make
informed decisions in near-real time about
behavior, performance and occupancy of
buildings.
• Not covered here: Some bridges in Korea.
51
52. ITEM 4
NEW TECHNOLOGY & BENEFITS
• Hardware (new developments)
• Software (real-time technology, and
developments in improved analyses methods)
• Ambient Data acquisition issues & Comment
about low-amplitude/strong shaking
52
53. HARDWARE (INDUSTRY STANDARD)
• 24 bit almost universally available & feasable
• High sampling rate (1 sps to 2000 sps): USGS
(200sps)
Multiple Data formats (COSMOS, EVT, SAC,
MATLAB, ASCII, MiniSEED) & Telemetry
Protocols (Antelope)
Extensive SOH monitoring; input and system
voltage, internal temp, humidity, comm link
diagnostics
• On-demand ambient data acquisition issues
53
54. SOFTWARE (for Analyses)
• Well-known spectral analyses (auto/cross
spectra, amplitude spectra, response spectra,
phase and coherency analyses)
• Less well-known : sophisticated system
identification analyses –many new developments
• Good data is essential for analyses
• High sampling rate (at least 100sps, preferably
200sps)
• Strong-shaking data/low-amplitude shaking data
54
55. SOME EXAMPLES :
Existing Output-Only Identification Techniques
(from E. Taciroglu, 2012)
Technique Input Type Domain Reference
Ibrahim Method Ambient Time Ibrahim SR, and Mikulcik EC, 1973
Frequency Domain
Ambient Frequency Brincker R, et al., 2000
Decomposition
Enhanced Frequency Domain
Ambient Frequency Brincker R, et al., 2001
Decomposition
Random Decrement Technique Ambient Time Asmussen JC, 1997
Stochastic Subspace
Ambient Time van Overschee P, and Moor BD, 1993
Identification
Eigensystem Realization
Juang JN, and Pappa RS, 1985
Algorithm + Natural Excitation Ambient Time
James G, et al., 1995
Technique
56. OTHER EXAMPLES
Summary of System Identification Methods (from R. Omrani, 2012)
Identified
Method Domain Model Data
Features
CEM (1) Time ARMA ω, ζ IO
RPEM (2) Time ARMA ω, ζ IO
UMP (3) Time & Freq. ARMA ω, ζ, φ IO
ERA (4) IO w/ IE
ERA-DC Time State Space System Matrices OO
ERA-OKID IO
N4SID (5) Time State Space System Matrices IO & OO
(1) Cumulative Error Minimization (Ljung, 1987)
(2) Recursive Prediction Error Minimizatin (Ljung 1987)
(3) Unified Matrix Polynomial (Allemang, 1994)
(4) Eigensystem Realization Algorithm, w/ Data Correlations or Observer/Kalman Filter
Identification (Juang et al., 1985, 1988, 1993)
(5) Numerical Algorithm for Subspace State Space System Identification (N4SID) (Van
Overschee & De Moor, 1994)
57. EXAMPLE: One Rincon Tower: Building Info
• Construction completed in
2008
• Designed according to 2001
SFBC
• 64-story
• 188.31m (617.83ft) tall
• Design includes outrigger
columns, and
• Unique dynamic response
modification features:
• BRB (buckling
restrained braces), and
• TSD (tuned liquid
sloshing Pile
foundation
57
58. ONE RINCON TOWER (SF):
RECENTLY COMPLETED COOPERATIVE
PROJECT BETWEEN CGS (CSMIP) and USGS
(TSD) Tuned Sloshing Dampers and BRB(Buckling Restrained Dampers -
also known as unbonded diagonal bracing)
From MKA document
BRB
TSD
58
59. One Rincon Tower: Structural Monitoring
System (www.strongmotioncenter.org)
59
64. CONCLUSIONS [what we learned?]:
• Ambient data analyzed yields repeatable /consistent frequencies
• No indication that BRB’s and TSD “kicked in” to alter dynamic
characteristics
• As expected low-amplitude ambient data yield periods (frequencies)
that are shorter (longer) than DBE or MCE analyses which reflect
larger level shaking for design
• The low-amplitude shaking dynamic characteristics will serve as
base-line for comparison with those that will be obtained from future
strong shaking data.
• This building percentage of core shear wall area/floor area compares
well with those that are practiced in Chile and has performed well.
• Unique design building monitoring yield high-quality data that
enables extraction of at least 5 modes and associated
frequencies/damping in each direction. MANY NEW SID
METHODS! 64
65. In addition: Low-amplitude vs strong shaking
Millikan Library Pacific Park Plaza
from E. Taciroglu(2012) from Çelebi, 2009
[attributed to J. Clinton (2006)
65
66. ITEM 5:
Seismic and Structural Health
Monitoring in British
Columbia, Canada
from
Carlos Ventura
• A new network in Canada
• Data will be available openly (per C. Ventura)
66
67. UBC Earthquake
Engineering
Seismic and Structural Health
Monitoring in British Columbia,
Canada
68. Background
The Geologic Survey of
Canada maintains a strong
motion network
The BC Ministry of
Transportation (MoT) has
become involved adding
many strong motion sites
around the province
MoT has been involved in
structural monitoring since
the late 1990’s, in
collaboration with UBC
The last 5 years has seen
their integration into a
comprehensive online
network
69. Bridge Seismic Engineering
MOT Post Earthquake Response
What is damaged – how bad?
Staffing
Inspections
Route condition
Prioritization
Changing plans
Risks/decisions
What does MOT need?
•Fast, accurate field intelligence
•Speed of initial response
•Effective risk assessment and decision making
70. A way to address concerns
Have Strong Motion Network put on
“the net”.
Have a GIS layer for Disaster Response
Routes (DRR) & bridges
Have a GIS layer for other vital points.
Read the values affecting DRR’s/bridges
Prioritize inspection deployment.
Collaboration between UBC and MOT Bridge
Seismic Engineering
71. Smart Infrastructure
Monitoring System (BCSIMS)
The goals of this Project are to:
1. develop and implement a real-time seismic structural response
system to enable rapid deployment and prioritized inspections
of the Ministry’s structures; and
2. develop and implement a health monitoring program to address
the need for safe and cost-effective operation of structures in
British Columbia
This system will include decision-making tools that will expedite
the process of prioritization, risk assessment and damage
evaluation to assist decision-makers with post earthquake
response and recovery options.
72. Smart Infrastructure
Monitoring System (SIMS)
The objectives of this Project are to:
1. develop and implement a cost-effective, reliable Structural
Health Monitoring (SHM) technology that makes effective use of
sensors and broad band digital communications for remote
monitoring of structures subjected to dynamic loads;
2. identify and implement effective algorithms for system
identification, model updating and damage detection suitable for
remote monitoring of structures subjected to seismic forces and
condition changes due to deterioration or impact;
3. develop an integrated decision support system that incorporates
geographical information and information about ground shaking
and structural performance of a portfolio of remotely monitored
structures distributed throughout the province.
73. Seismic SHM Solution
Implement damage detection algorithms to
identify loss of structural integrity
Damage from impact, deterioration or seismic
Provide a dense ground motion network
Provide “Internet” and “local” data access and
information display
Provide a trigger response mechanism
Instrument key bridges
Bridge Seismic Engineering
80. Current and Future Works
Installation of updated monitoring hardware on three structures
Review of monitoring systems for the new structures
Implementation of existing systems into BCSIMS
Deploying 20 new strong motion stations
Seismic performance modelling of several bridges
Further implementation of damage detection algorithms – ISMS
Project
Upgrading of current BCSIMS system components
Work with inspectors/bridge engineers to determine best format of
information, creation of operational guidelines
Collaboration with other Government agencies and groups;
Ministry of Education, Emergency Management BC, BC Housing,
Municipalities
83. Remember Virginia Eq (Mw=5.8) August 23, 2011:
No deaths. Few injured. Felt in 21 states.
National Cathedral and Washington Monument damaged
84. What do we remember
about these events most?
WASHINGTON
MONUMENT
base 55 feet 1½ inches
wide on each side
(reduces to 34-feet 5⅝
inches wide at the 500-
foot level).
85. PERFORMANCE OF BASE-ISOLATED USC
HOSPITAL DURING THE NORTHRIDGE EQ.
THE RECORDS RETRIEVED
FROM THE BASE-ISOLATED USC
HOSPITAL [LARGEST ACCELERATIONS
AND DISPLACEMENTS EXPERIENCED,
TO DATE, FOR A BASE-ISOLATED BUILDING]
CONFIRMED THE EFFECTIVENESS OF
BASE-ISOLATION TECHNIQUES
85