Response Health Monitoring of Buildings Using Threshold Drift Ratios – Now an Established

Method

by

M. Çelebi1

ABSTRACT 1. INTRODUCTION

In earlier papers, we described how observed data 1.1. Background and Rationale
from differential GPS with high sampling ratios
and from classical accelerometer deployed in Following an earthquake, rapid and accurate
structures can be configured to establish seismic assessment of the damage condition or
health monitoring of structures. In these performance of a building is of paramount
configurations, drift ratios 1 are the main importance to stakeholders [owners, leasers,
parametric indicator of damage condition of a permanent and/or temporary occupants, and city
structure or component of a structure. officials and rescue teams that are concerned with
safety of those in the building and those that may
Real-time measurement of displacements are be affected in nearby buildings and
acquired either directly using GPS or by double infrastructures]. Until recently, assessments of
integration of accelerometer time-series data. damage to buildings following an earthquake were
Recorded sensor data is then related to the essentially carried out by inspections of city-
performance level of a building. Performance- designated engineers following procedures similar
based design method stipulates that for a building to ATC-20 tagging requirements (ATC 1989).
the amplitude of relative displacement of the roof Tagging usually involves visual inspection only
of a building (with respect to its base) indicates its and is implemented by colored tags indicative of
performance. potential hazard to occupants - green indicating the
building can be occupied; that is the building does
Experience with both types of sensor deployments not pose a threat to life safety, yellow indicates
(GPS and accelerometers) indicate that they are limited occupation (that is, hazardous to life safety
reliable and provide pragmatic alternatives to alert but not to prevent limited entrance to retrieve
the owners and other authorized parties to make possessions), and red indicating entrance
informed decisions and select choices for pre- prohibited (that is, hazardous to life). However,
defined actions following significant events. one of the impediments to accurately assessing the
Furthermore, recent adoption of such methods by damage level of structures by visual inspection is
financial and industrial enterprises is testimony to that some serious damage may not be visible due
their viability. to the presence of building finishes and
fireproofing. In the absence of visible damage to
KEYWORDS: real-time, health monitoring, the building frame, most steel or reinforced
buildings, GPS, displacement, drift ratio, concrete moment-frame buildings will be tagged
acceleration, accelerometer, internet based on visual indications of building
communication, data acquisition. deformation, such as damage to partitions or
glazing. Lack of certainty regarding the actual
deformations that the building experienced may
1
Drift ratio is defined as relative displacement between any
two floors divided by the difference in elevation of the two 1
Earthquake Hazards Team, USGS (MS977),
floors. Usually, this ratio is computed for two consecutive
floors.
345 Middlefield Rd., Menlo Park, Ca, 94025

1
typically lead an inspector toward a relatively these methods are reliable and provide requisite
conservative tag. In such cases, expensive and information for owners and other parties to make
time-consuming intrusive inspections may be informed decisions and select choices for pre-
recommended to building owners (e.g., it is known defined actions following significant events.
that, following the [Mw=6.7] 1994 Northridge, CA Furthermore, recent additional adoptions of such
earthquake, approximately 300 buildings ranging methods by financial and industrial enterprises
in height from 1 to 26 stories were subjected to validate its usefulness.
costly intrusive inspection of connections
(FEMA352, SAC 2000)). 1.2. Requisites
The most relevant parameter to
However, as described in this paper, an alternative assess performance is the measurement or
to tagging is now available to owners and their computation of actual or average story drift ratios.
designated engineers by configuring real-time Specifically, the drift ratios can be related to the
response of a structure instrumented as a health performance- based force-deformation curve
monitoring tool. As Porter and others (2006) state, hypothetically represented in Figure 1 [modified
most new methods do not utilize real-time from Figure C2-3 of FEMA-274 (ATC 1997)].
measurements of deformations of a building for When drift ratios, as computed from relative
assessments of building’s performance during an displacements between consecutive floors, are
event with the exception outlined by Çelebi and determined from measured responses of the
Sanli (2002) and Çelebi and others (2004). In building, the performance and as such “damage
these applications, differential GPS (Çelebi and state” of the building can be estimated as in Figure
Sanli, 2002) with high sampling ratios and 1.
classical accelerometer deployed structures (Çelebi
and others, 2004) are configured to obtain data in Measuring displacements directly is very difficult
real-time and compute drift ratios as the main and, except for tests conducted in a laboratory
parametric indicator of damage condition of a (e.g., using displacement transducers), has not yet
structure or one or more components of a been feasibly achieved for a variety of real-life
structure. The rationale here is that a building structures. For structures with long-period
owner and engineers are expected to use the responses, such as tall buildings, displacement
response data acquired by a real-time health measurements using GPS are measured directly at
monitoring system to justify a reduced inspection the roof only; hence, drift ratio then is an average
program as compared to that which would drift ratio for the whole building. Thus, recorded
otherwise be required by a city government for a sensor data is related to performance level of a
similar non-instrumented building in the same building; hence, related to performance-based
area 2 . It is possible that depending on the design which stipulates that for a building the
deformation pattern and associated damage amplitude of relative displacement of the roof of a
indicators observed in a building, it could also be building (with respect to its base) indicates its
possible to direct the initial inspections toward performance.
specific locations in the building that experienced
large and potentially damage-inducing drifts For accelerometer based systems, the
during an earthquake. accelerometers must be strategically deployed at
specific locations on several floors of a building to
Examples of and data from either type of sensor facilitate real-time measurement of the actual
deployment (GPS or accelerometers) indicate that structural response, which in turn is used to
compute displacements and drift ratios as the
2
indicators of damage.
The City of San Francisco, California, has developed a
“Building Occupancy Resumption Program” (BORP, 2001)
whereby a pre-qualified Occupancy decision making process Table 1 shows typical drift ratios for steel
as described in this paper may be proposed to the City as a moment resisting framed buildings. The table is
reduced inspection program but in lie of detailed inspections
by city engineers following a serious earthquake.
developed from FEMA 352 (SAC 2000). For

2
reinforced concrete framed buildings, the lower the differential displacement records from GPS
figures may be more appropriate to adopt. units (Figures 3g-h). Frequency of 0.24-.25 Hz
seen in Sxy plots from both acceleration and
Table 1. Summary of Suggested Typical Threshold
Stages and Ranges of Drift Ratios displacement data belong to the expected
fundamental frequency for a 34-story building. A
Threshold Stage 1 2 3 second frequency at 0.31 Hz (from acceleration
Suggested Typical 0.2-0.3 0.6-0.8 1.4-2.2 data) Hz is belongs to the torsional mode.
Drift Ratios (%)
At the fundamental frequency at 0.24 Hz, the
displacement data exhibits a 0o phase angle;
2. AN TWO APPROACHES FOR MEASURING however, the coherencies are lower (~0.6-0.7).
DISPLACEMENTS The fact that the fundamental frequency (0.24 Hz)
can be identified from the GPS displacement data,
2.1. Use of GPS for Direct Measurements of amplitudes of which are within the manufacturer
Displacements specified error range, and that it can be confirmed
by the acceleration data, is an indication of
Until recently, use of GPS was limited to long- promise of better results when larger
period structures (T>1 s) because differential GPS displacements can be recorded during strong
systems readily available were limited to 10-20 sps shaking.
capability 3 . Currently, the accuracy of 10-20 Hz
GPS measurements is ± 1 cm horizontal and ± 2.2. Displacement via Real-time Double
2cm vertical. Furthermore, with GPS deployed on Integration
buildings, measurement of displacement is
possible only at the roof. A general flowchart for an alternative strategy
based on computing displacements in real-time
A schematic and photos of an application using from signals of accelerometers strategically
GPS to directly measure displacements is shown deployed throughout a building is depicted in
in Figure 2. In this particular case, two GPS units Figure 4, and described by Çelebi and others
are used in order to capture both the translational (2004). Although ideal, deploying multiple
and torsional response of the 34-story building in accelerometers in every direction on every floor
San Francisco, CA. At the same locations as the level is not a feasible approach not only because
GPS antennas, tri-axial accelerometers are of the installation cost, but also from the point of
deployed in order to compare the displacements view of being able to robustly, and in near real-
measured by GPS with those obtained by double- time, (a) stream accelerations, (b) compute and
integration of the accelerometer records. Both stream displacements and drift ratios after double
acceleration and displacement data streaming into integration of accelerations, and (c) visually
the monitoring system is shown also in Figure 2. display threshold exceedences, thus fulfilling the
objective of timely assessment of performance
To date, strong shaking data from the deployed level and damage conditions.
system has not been recorded. However, ambient
data obtained from both accelerometers and GPS A schematic of a recently deployed health
units are analyzed (Figures 3a-d). Sample cross- monitoring system which utilizes these principles
spectra (Sxy) and coherency and phase angle plots is shown in Figure 5 (Çelebi and others, 2004).
of pairs of parallel records (N-S component of The distribution of accelerometers provides data
north deployment [N_N] vs. N-S component of from several pairs of neighboring floors to
south deployment [S_N], from accelerometers are facilitate drift computations. The system server at
shown in Figures 3e-f. The same is repeated for the site (a) digitizes continuous analog data, (b)
pre-processes the 1000 sps digitized data with low-
3
Recently, up to 50 sps differential GPS systems are available pass filters (c) decimates the data to 200 sps and
on the market and have been successfully used (Panagitou et streams it locally, (d) monitors and applies server
al, 2006).

3
triggering threshold criteria and locally records belongs to the second translational mode. At the
(with a pre-event memory) when prescribed right (Figure 7), a 20 s window of computed
thresholds are exceeded, and (e) broadcasts the displacements (20 s into the record) reveals the
data continuously to remote users by high-speed propagation of waves from the ground floor to the
internet. Data can also be recorded on demand to roof. The travel time is extracted as about 0.5
facilitate studies while waiting for strong shaking seconds. Since the height of the building is known
events. (262.5 ft [80m]), travel velocity is computed as
160 m/s. One of the possible approaches for
A “Client Software” remotely acquires detection of possible damage to structures is by
acceleration data to compute velocity, keeping track of significant changes in the travel
displacement and drift ratios. Figure 6 shows two time, since such travel of waves will be delayed if
PC screen snapshots of the client software display there are cracks in the structural system (Safak,
configured to stream acceleration or velocity or 1999).
displacement or drift ratio time series. The
amplitude spectrum for one of the selected 3. MONITORING SINGLE STRUCTURE VS.
channels is periodically recomputed and clearly CAMPUS STRUCTURES
displays several identifiable frequency peaks. In
the lower left, time series of drift ratios are shown. Rather than having only one building monitored,
there may be situations where some owners desire
Corresponding to each drift ratio, there are 4 to monitor several buildings simultaneously, such
stages of colored indicators. When only the as on industrial campus. Figure 8 schematically
“green” color indicator is activated, it indicates shows a campus-oriented monitoring
that the computed drift ratio is below the first of configuration. Depending on the choice of the
three specific thresholds. The thresholds of drift owner and consultants, a campus system may have
ratios for selected pairs of data must also be building specific or central monitoring systems
manually entered in the boxes. As drift ratios and as such is highly flexible in configuration. As
exceed the designated three thresholds, additional can be stipulated, potential variations and
indicators are activated with a different color combinations of alternatives for a campus-wide
(Figure 4). The drift ratios are calculated using monitoring system are tremendous. There can be
data from any pair of accelerometer channels central controlled monitoring as well as building
oriented in the same direction. The threshold drift specific monitoring or both. A wide variety of data
ratios for alarming and recording are computed communication methods can be configured to
and determined by structural engineers using meet the needs (Figure 8).
structural information and are compatible with the
performance-based theme, as previously illustrated 4. CONCLUSIONS
in Figure 1.
An Capitalizing on advances in global positioning
A set of low-amplitude accelerations (largest peak systems, in computational capabilities and
acceleration ~ 1 % g) recorded in the building methods, and in data transmission technology, it is
during the December 22, 2003 San Simeon, Ca. now possible to configure and implement a
earthquake (Mw=6.4, epicentral distance of 258 seismic monitoring system for a specific building
km are exhibited in Figure 7 for one side of the with the objective of rapidly obtaining and
building. The figure (center) also shows evaluating response data during a strong shaking
accelerations at the roof and corresponding event in order to help make informed decisions
amplitude spectra for the (a) two parallel channels regarding the health and occupancy of that specific
(Ch12 and Ch21), (b) their differences (Ch12- building. Using GPS technology and/or double-
Ch21), and (c) orthogonal channel (Ch30). The integrated acceleration, displacements and, in turn,
amplitude spectra depicts the first mode drift ratios are obtained in real-time or near real-
translational and torsional frequencies as 0.38 Hz time. Drift ratios can be related to damage
and 0.60Hz respectively. The frequency at 1.08 Hz condition of the structural system by using

4
relevant parameters of the type of connections and SAC Joint Venture, 2000. Recommended Post-
story structural characteristics including its earthquake Evaluation and Repair Criteria for
geometry. Thus, once observed drift ratios are Welded Steel Moment-Frame Buildings,
computed in near real-time, technical assessment prepared for the Federal Emergency
of the damage condition of a building can be made Management Agency, FEMA 352, Washington,
by comparing the observed with pre-determined D.C.
threshold stages. Both GPS and double-integrated
acceleration applications can be used for Porter, K., Mitrani-Reiser, J., Beck, J. L., and
performance evaluation of structures and can be Ching, J., 2006, Smarter Structures: Real- time
considered as building health-monitoring loss estimation for instrumented buildings,
applications. Although. To date, these systems 8NCEE (paper no. 1236), San Francisco, Ca.
were not tested during strong shaking events, April 18-22, 2006.
analyses of data recorded during smaller events or
low-amplitude shaking are promising. Building Occupancy Resumption Program
(BORP), 2001. City and County of San
5. REFERENCES Francisco, Department of Building Inspection,
Emergency Operation Plan, (Rev. 2001).
Çelebi, M., and Sanli, A. (2002). GPS in [www.seaonc.org/member/committees/des_build
Pioneering Dynamic Monitoring of Long-Period .html).
Structures, Earthquake Spectra, Journal of EERI,
Vol. 18, No. 1, pp 47–61, Feb. 2002. Applied Technology Council (ATC). (1997).
NEHRP Commentary on the Guidelines for the
Çelebi, M., Sanli, A., Sinclair, M., Gallant, S., Seismic Rehabilitation of Buildings, prepared
and Radulescu, D. (2004). Real-Time Seismic for the Building Seismic Safety Council,
Monitoring Needs of a Building Owner and the published by the Federal Emergency
solution – A Cooperative Effort, Journal of Management Agency, FEMA 274, Washington,
EERI, Earthquake Spectra, v.19, Issue 1, pp.1- D.C.
23.
Panagitou,M., Restrepo, J.I., Conte, J. P., and
Applied Technology Council (ATC), 1989. Englekirk, R. E., 2006, Seismic Response of
Procedures for Post-Earthquake Safety Reinforced Concrete Wall Buildings, 8NCEE
Evaluation of Buildings, ATC-20, Redwood (paper no. 1494), San Francisco, Ca. April 18-
City, CA. 22, 2006.

FEMA-352: Recommended Post-earthquake Safak, E. (1999). Wave-propagation formulation

Evaluation and Repair Criteria for Welded Steel of seismic response of multistory buildings,
Moment-Frame Buildings (SAC 2000). ASCE, Journal of Structural Engineering, vol.
125, no. 4, April 1999, pp. 426-437.

5
 Figure 1. Hypothetical displacement time-history as related to performance [modified from Figure
C2-3 of FEMA-274 (ATC 1997)].

Figure 2. (Left)- Schematic of the overall system using GPS and accelerometers (San Francisco, CA.):
(Center)- GPS and radio modem antenna and the recorders connected to PC, (Right)- streaming acceleration
and displacement data in real-time.
-3 5 CROSS -SPECTRUM CROSS -SPECTRUM
x 10 0.5 x 10
1.5
[FROM GPS] 3 12 .24
N_N .25 ACCELERATION
1 0 N_N COMPONENT [FROM 10 DISPL.
ACC.[CM/S/S]

ACCELEROMETER]
DISPL (CM)

0.5 [FROM GPS]

2 8
-0.5
Sxy

Sxy

0 N_N vs S_N
.31 6
N_N vs S_N
0.5 -1 1 4
1 [FROM 2
[a] ACCELEROMETER] -1.5 [c] [e] [g]
1.5 0 0
0 20 40 60 0 200 400 600 800 1000 1200 0 0.5 1 0 0.5 1
TIME (S) TIME (S) FREQ (HZ) FREQ (HZ)
-3
x 10 ACC:N_N vs. S_N DISPL.(N_N vs. S_N)
[FROM GPS] 1 1
S_N 150 150
1.5 [FROM 0.5
PHASE ANG.(DEG.)

ACCELEROMETER] S_N COMPONENT 0.8 0.8

100 100
ACC.[CM/S/S]

1
COHERENCE

0
DISPL (CM)

0.6 50 0.6 50
0.5 0
-0.5 0
0
0.4 0.4
-50 -50
-.5 -1
0.2 -100 0.2 -100
-1 [d]
[b] -1.5 [f ] -150 [h] -150
-1.5 0 0
0 20 40 60 0 200 400 600 800 1000 1200 0 0.5 1 0 0.5 1
TIME (S) TIME (S) FREQ(HZ) FREQ(HZ)

Figure 3. [a,b] 60 second windowed accelerations and [c,d] 1200 second windowed GPS displacement
data in the north-south orientation and at N (North) and S (South) locations. Cross-spectra (Sxy) and
associated coherency and phase angle plots of horizontal, and parallel accelerations [e,f] and GPS
displacements [g,h]. [Note: In the coherency-phase angle plots, solid lines are coherency and dashed lines
are phase-angle].

6
Figure 4. Flow-chart for observation of damage levels based on threshold drift ratios.

Figure 5. Schematic of real-time seismic monitoring of the building.

7
 Figure 6. Screen snapshots of sample client software displays: (left) acceleration streams and
computed amplitude and response spectra, and (right) displacement and corresponding drift ratios and
alarm systems corresponding to thresholds.

ACCELERATIONS PARALLEL TO FIRST STREET EQ: DEC. 22, 2003 EQ: DEC. 22, 2003 DISPLACEMENTS: PARALLEL TO FIRST STREET
60 15000 CH12 (SOLID) CH30 (ROOF)
140 .38
CH30 (ROOF) CH12 CH21 (DASHED) 8
10000
1.08 CH29 (23RD.) FL.
120 50
CH29 (23RD. FL.) 5000 7

CH28 (18TH.)FL.
ACCELERATION (CM/S/S)

0
ACCELERATION (CM/S/S)

100 CH28 (18TH. FL.) 40 0 0.5 1 1.5 2 6

DISPLACEMENT (CM)
CH21
AMPLITUDE (CM/S)

CH27 (17TH. FL.) CH27 (17TH.) FL.

1.08 5
30 2000 CH12 - CH21
80
CH26 (14TH. FL.) CH26 (14TH.) FL.
1.60
.60 4
1000
60 CH25 (12TH. FL.) 20 CH25 (12TH.) FL.
CH12 - CH21
0 3
0 0.5 1 1.5 2
CH24 (7TH. FL.) CH24 (7TH.)FL.
40 10
2
CH30 CH30 1.08
CH23 (6TH. FL.) 10000
CH23 (6TH.)FL.
20 0 .38 1
CH22 (EL. 25FT.) CH22 (EL. 25) FT.
5000
CH1 (GR.) FL.
CH1 (GR. FL.) 0
0 -10
0 20 22 24 26 28 30 32 34 36 38 40
0 20 40 60 80 100 120 0 20 40 60 80 100 120 0 0.5 1 1.5 2
TIME (S ) TIME (S) FREQ (HZ) TIME (S)

Figure 7. Accelerations (left) at each instrumented floor on one side of the building [San Simeon EQ. of
12/22/2003], (center) from parallel channels (CH12, CH21) at the roof, their difference (CH12-CH21),
and orthogonal CH30, and corresponding amplitude spectra indicate fundamental frequency at 0.38 Hz.
(right) A 20-s window starting 20-s into the record of computed displacements shows propagating waves
with travel time of ~ 0.5 s (onsets indicated by dashed line) from the ground floor to the roof.

8
Figure 8. A schematic of campus oriented monitoring system. Each building within a campus may have its
own monitoring system or there may a central monitoring unit.