Gps in Dynamic Monitoring of Long-Period Structures: M. Cëelebi
Gps in Dynamic Monitoring of Long-Period Structures: M. Cëelebi
Gps in Dynamic Monitoring of Long-Period Structures: M. Cëelebi
www.elsevier.com/locate/soildyn
Abstract
Global Positioning System (GPS) technology with high sampling rates (,10 samples per second) allows scienti®cally justi®ed and
economically feasible dynamic measurements of relative displacements of long-period structuresÐotherwise dif®cult to measure
directly by other means, such as the most commonly used accelerometers that require post-processing including double integration.
We describe an experiment whereby the displacement responses of a simulated tall building are measured clearly and accurately in
real-time. Such measurements can be used to assess average drift ratios and changes in dynamic characteristics, and therefore can be
used by engineers and building owners or managers to assess the building performance during extreme motions caused by earthquakes
and strong winds. By establishing threshold displacements or drift ratios and identifying changing dynamic characteristics, procedures
can be developed to use such information to secure public safety and/or take steps to improve the performance of the building.
Published by Elsevier Science Ltd.
Keywords: Monitoring; GPS; Strong-motion; Building; Structural response; Frequency; Displacement; Acceleration; Drift; Bridge; Long-period structure
2. Technical justi®cation
In the last few years, there have been numerous studies 3. Model tests simulating a tall building
related to the technical feasibility of using GPS to measure
To investigate the feasibility of using GPS technology to
displacements of civil structures. Most of the initial work
monitor tall buildings (and other long-period structures), we
has been accomplished by aerospace atmospheric research-
conducted two experiments. Fig. 1 depicts a photo and the
ers. Studies related to the application of GPS for static or
overall set-up for a simple and inexpensive experiment
dynamic measurements of displacements of structural
designed by selecting a standard stock steel bar to simulate
systems include but not limited to those by Hyzak et al.
a 30±40 story ¯exible building. We selected the length,
[1], Teague et al. [2], Guo and Ge [3], Kondo and Cannon
thickness, and width of the two bar specimens to yield a
[4], Lovse et al. [5], Hudnut and Behr [6], Behr et al. [7] and
fundamental period of approximately 4 s in the weak direc-
Stein et al. [8]. In our study, we direct the efforts to actual
tion. To make things simple, we selected the width and
permanent deployment of GPS units alongside acceler-
thickness of each of two bars with a much weaker axis in
one direction. The width was varied to show the sensitivity
1
For most ¯exible buildings, the fundamental period (T) is approximated of measurements during vibration and at 10 Hz sampling
by 0.1N, where N is the number of ¯oors of a building (even though the rate. Each bar was ®xed at the base and the GPS unit was
fundamental period can vary between 0.05Nand 0.15N depending on the
¯exibility of the building). Therefore, to simulate a 40-story building, we
attached at its tip. By providing an initial displacement
set the period (frequency) equal to 4 s (0.25 Hz) and proportioned the (simply by pulling the top of the bar and releasing), each
length, width and thickness of the cantilever. bar was set into free vibration and its motion was recorded.
M. CËelebi / Soil Dynamics and Earthquake Engineering 20 (2000) 477±483 479
Fig. 2. Particle motion and time-history of relative displacements (NS and EW components) of simulated test specimen.
Table 1
Results of tests with GPS units
Specimen Length [H] ft (m) Width [B] in. (cm) Thickness [t] in. (cm) Measured frequency [f] (Hz) Measured period [T] (s) Damping [j ] (%)
Results are summarized in Table 1. Fig. 2 shows the particle and 0.296 Hz (3.38 s) respectively. Also, a damping percen-
motion and time-history of one of the tests performed. The tage of approximately 2% is extracted. This simple test
axes of the bar were at an angle to the NS (and EW) direc- shows that sampling at 10 Hz with GPS units provides a
tion. Therefore, the NS and EW components of displace- clear and accurate displacement response history from
ments are identical in phase and proportional in amplitude. which drift ratios and dynamic characteristics of the speci-
Also, since the GPS unit is not symmetrically and concen- men can be derived [9±11]. The implications of this go
trically mounted in the weak direction (photo in Fig. 1), the beyond just the measurements. It can be shown that identi-
amplitudes of positive and negative displacements ®cation of dynamic characteristics can be used to prove
measured are not the same. The detection of the effect of possible nonlinearities that occur during vibration (e.g.
the eccentric mass adds to the assurance that the measure- due to damage and plastic behavior of the structural
ments are accurate and sensitive. These simple tests and members, components and/or joints, or to soil±structure
results were, and can be, duplicated easily and readily. interaction under varying amplitudes of input motions).
Fig. 3 is a plot of NS components of measured relative
displacements and corresponding amplitude spectra of bars
A and B. The ®gure shows the accuracy and sensitivity of 4. GPS ambient test of a 44-story building and strong-
the GPS monitoring technology at 10 sps. The measure- motion acceleration records
ments differentiate between the frequency of the free-vibra-
tion response of the two bars with different dynamic In a second test, we measured ambient vibration (due to
characteristics. From the data, the fundamental frequency winds and traf®c noise) of a 44-story building with a GPS
(period) of the two bars are identi®ed to be 0.245 Hz (4.08 s) unit temporarily deployed on its roof. A reference GPS unit
Fig. 3. Relative displacements of two test specimens (NS components only) in free-vibration and corresponding amplitude spectra identifying the fundamental
frequencies of the test specimens.
480 M. C
Ë elebi / Soil Dynamics and Earthquake Engineering 20 (2000) 477±483
Fig. 4. Ambient test of a 44-story building with GPS technology: Cross-spectrum (left-solid), coherence (left-dashed) and phase-angle (right) of two
orthogonal horizontal motions.
was located within 500 m of the building. The signals were Despite the small displacements (,1 cm) and low signal
very noisy and amplitudes very small; therefore, most to noise ratio of this experiment, the fact that dynamic
common methods to identify structural characteristics did frequencies could be identi®ed indicates that during larger
not work. Only the cross-spectrum of the two orthogonal, displacements, better identi®cation of the dynamic charac-
horizontal, low-amplitude, ambient displacement record- teristics as well as drift ratios can be made with higher
ings (when the signals are coherent and approximately con®dence.
1808 out of phase) were used to identify the fundamental
frequency of the building at 0.23 Hz (Fig. 4), with another
frequency at ,0.3 Hz. Despite the very small signal, these 5. Looking to the future: real-time monitoring
frequencies appear to be reliable when compared with the
0.23 Hz frequency calculated from accelerations recorded 5.1. Perceived real-time monitoring
with a triaxial accelerograph on the 38th ¯oor (accepted
here as the roof response) during a small earthquake Initially, we are planning to deploy the GPS units only on
(Fig. 5). A comparison of these frequencies is provided tall buildings that are already instrumented with acceler-
in Fig. 6. ometers. This will facilitate comparison of absolute and
Fig. 5. Recorded roof (38th ¯oor accepted in lieu of roof) accelerations of a 44-story building and displacements (derived by double integration) and amplitude
spectrum.
M. CËelebi / Soil Dynamics and Earthquake Engineering 20 (2000) 477±483 481
Fig. 6. Comparison of normalized frequencies for ambient GPS recording (from cross-spectrum) and strong-motion recording (from amplitude spectrum).
Fig. 7. Hypothetical thresholds of displacements. The time-history of displacements shown is actually integrated from accelerations recorded at the 38th ¯oor
(accepted in lieu of roof) of a 44-story building. The actual record (in Fig. 5) is ampli®ed by 20 times for illustration purposes.
relative displacements measured by GPS and calculated by Bridge (San Francisco) or Vincent-Thomas Bridge (Los
double integration of accelerations. The GPS units will be Angeles). As in the case of tall buildings, changes in
con®gured to provide data to indicate the real-time average dynamic characteristics after the displacements at critical
drift ratios and changes in the dynamic characteristics of the locations of a bridge have exceeded predetermined thresh-
buildings. This information can be made available to build- olds, can be calculated in near-real-time (Fig. 8). When
ing managers (or interested parties) in real-time or whenever warranted, the management can make decisions to inspect
a predetermined displacement threshold is reached. The the bridge (e.g. decisions can be made to stop the traf®c,
building managers can assess the response of the buildings thus securing the bridge safety, which is one of the objec-
according to (a) different threshold displacements (e.g. A±C tives of lifeline earthquake engineering). With the GPS
as shown in Fig. 7), (b) drift ratios, or (c) changing dynamic technology, we can furnish time-dependent displacements
characteristics. If a situation is serious, the management can for the relative movements of critical locations of structures.
make decisions to evacuate the building for additional For example, for the bridges, GPS units placed at pre-
inspection. Therefore, one by-product of the effort would selected locations of bridge elements can indicate, in real-
be to secure the safety of the occupants and signi®cant time, the amplitude of the displacements of the decks and
contents of the building. Thus, a real-time structural health towers, as well as movements of key bridge elements rela-
monitoring environment will be created. At least three GPS tive to local bedrock reference points. Thus, movements of
units per building are required to monitor a tall building: the piers relative to the abutments, the top of a tower with
two of the units should be deployed on the roof to detect respect to its base, or the span with respect to the ground,
translational and torsional response of the building, and the can be made at a centimeter-level of precision, in real-time.
third unit will serve as a reference ground station to evaluate We will recover both the dynamic motions that accompany
relative displacement. The ground site also needs excellent the earthquake, as well as the static or permanent displace-
sky visibility. ments experienced by the bridge once the shaking has
Similar deployments are being planned for other types of stopped. Such permanent displacements affect the state of
long-period structures. One project in development at this stress of a bridge, and provide evidence for distortion or
time is for deploying GPS units on one of the long-period failure of bridge elements or subsidence of piers due to
suspension bridges such as the Golden Gate Bridge and Bay ground compaction induced by earthquake shaking.
482 M. C
Ë elebi / Soil Dynamics and Earthquake Engineering 20 (2000) 477±483
Fig. 8. Hypothetical thresholds for displacement (from double-integration of recorded acceleration) of channel 21 (vertical at mid side-span) of Vincent
Thomas Bridge (1994 [M 6.7] Northridge earthquake).
5.2. Current deployments on tall buildings Therefore, any deployment had to be planned away from
the parapet wall. To solve this problem, a stiff auxiliary
Currently, deployment of GPS units on the roof of two support frame (Fig. 9) was erected. The ®gure also shows
buildings in Los Angeles has been completed, and we are in the GPS antenna and the radio antenna to communicate with
the process of deploying GPS units on the roof of a building the reference station so that differential displacements are
in San Francisco. Fig. 9 shows the actual deployment on the obtained.
roof of one of the buildings in Los Angeles. As happens with
®eld deployments, physical obstacles and constraints neces- 5.3. Requisite software
sitate alternate approaches. In this case, the window cleaner
machinery on the roof of the 44-story building travels on Requisite software is being developed to assess and miti-
rails near the parapet and uses the parapet continuously. gate the two natural hazards (earthquake and severe wind)
affecting the structures by using the displacements
measured by the GPS units. Fig. 10 shows a screen image
of software being developed to assess displacements
observed with GPS and provide alarm (warning). Such
collected information on the response of the structure during
strong motion events (or strong winds) can be used to make
decisions for further evaluation of the susceptibility to
damage of the structure, and future repair/retro®t schemes
may be developed.
Fig. 10. Software developed to provide alarms at three thresholds of displacement of an instrumented structure with GPS.