Overview

The 72-sensor embedded seismic array in the UCLA Factor building is a testbed for a variety of CENS developments including theoretical modeling of building motions which are verified by the high-density data recordings, and in preparation for near-future tests of wireless array developments made by Govindan et al. who are developing new software architecture to perform multi-hop communications among nodes of the tiered embedded network.  Regarding the modeling activities, wave propagation effects have become increasingly useful in determining the system identification of buildings in order to carry out scenario event and damage pattern building response.  Through the modeling, we are trying to determine if the density of measurements will make it possible to observe subtle changes in dynamic characteristics between pairs of floors and to relate the measurements to system damage such as changes in stiffness due to a column failure.  The high dynamic range of the 24-bit digitizers allows both strong motions and ambient vibrations to be recorded with reasonable signal-to-noise ratios.  Regarding the wireless network activities, wireless untethered devices capable of measuring structural vibrations represent an emerging technology that can significantly increase the spatial resolution of structural response to earthquakes. A network of these wireless sensors can be used to instrument structures relatively cheaply and quickly. These sensors have onboard processing capabilities making it possible to intelligently process sensor data within the network. In the absence of this capability, the high sampling rates required for structural monitoring can overwhelm the radio bandwidths of existing sensor platforms.

Approaches

Waveform data from the seismic array in the Factor building are used in comparison with finite element calculations for predictive behavior.  A three-dimensional model of the Factor building has been developed based on structural drawings.  Observed displacements for 20 small and moderate, local and regional earthquakes were used to compute the impulse response functions of the building by deconvolving the subbasement records as a proxy for the free field.  The impulse response functions were then stacked to bring out wave propagation effects more clearly.  The simulation results for travel times, mode shapes, and frequencies of vibration agree to within a few small percent using the impulse response function of the subbasement as ground history input.  It can be shown that small but significant changes in the travel times, mode shapes, and frequencies are observed in the simulation results for strong ground shaking and for modifications to the structural model for hypothesized damage patterns such as broken welds on a particular floor. 

We are working closely with Govindan et al. to prepare for tests of the wireless network in the Factor building.  Our first test consists of determining ambient vibration noise levels recorded by the Mica-Z motes with Crossbow accelerometers placed close to the bottom and close to the top of the building. To simplify the data retrieval for this first test, we will not use wireless communications.  The wired sensors indicate ambient vibration background amplitudes of 0.0001 g for the basement and 0.001 g for the roof.  Our tests will determine the minimum intrinsic recording capabilities of the mote-digitizer box in terms of instrument noise and ambient vibrations.  Once these tests are completed and assuming the instrument noise levels recorded by the motes is not unacceptably high, we will move on to the full wireless test (802.15 platform) of approx. 30 motes distributed throughout the building.

Accomplishments

We have worked with Heaton et al. on constructing the theoretical model of the Factor building.  After numerous comparisons with actual data recording, we are now satisfied that we have a model that very closely approximates the behavior of Factor.  This means that we are now in a position to carry out numerical experiments for scenario earthquake input time history shaking input as well as investigations of building response to realistic damage pattern scenarios such as a significant number of broken welds in specific parts of the building.

We have recently worked with our colleagues at the USGS to upgrade the wired sensors from older-style 1g and 2g FBA-11 force-balance accelerometers to 4g Episensors capable of higher dynamic range and wider frequency response.  The new Episensors are all installed and we are now recording data from them in the same fashion as before: data are sent from the digitizers to our lab at UCLA where Antelope software archives the time series locally and at the IRIS DMC.  Related to this, one of four free-field seismometers was installed at the Geology building on soil to record nearby approximate free-field motions.

We are about to conduct the first of two wireless network tests in Factor, working with our colleagues at USC.  We have just reviewed their side-by-side tests of ambient vibration recordings on several 2g motes and have selected two to deploy in Factor for 12 hours.  They will be deployed on two floors, one near the top and one near the bottom of the building.  This is the first in a multi-step process that has the goal of full 30-mote deployment in Factor.

We were awarded a 2005-2006 USGS ANSS grant for operation and maintenance of the UCLA Factor network: a cooperative project between UCLA and the USGS.  As part of the network operations, quality control procedures will be developed that focus on waveform and spectral amplitude monitoring, metadata maintenance, and station history record-keeping.  We were also awarded a 2006 Southern California Earthquake Center (SCEC) grant to perform experimental end-to-end (earthquake source through to structural response) simulations that will produce the full wavefield at any point in the Los Angeles basin from past and scenario earthquakes in order to estimate the seismic risk of the Factor building and its analogs.

Future Directions

We will be expanding on our wireless network tests as soon as we determine the intrinsic mote noise levels, and as soon as the Govindan et al. completes its development of the software to perform the multi-hop, wireless communications for the Stargate master-Mica mote slave tiered network.  The goal is to deploy the ~30 motes in the Factor distributed throughout the building.  We will especially take advantage of the presence of several vertical-component sensors among his mote array.  These will also be distributed in such a way as to compute relatively dense measurements for rotations (i.e., significant bending, soil-structure interactions, and surface-wave effects).  It remains to be determined what maximum data rate we choose as there is a tradeoff between data rate and the maximum number of components recording data within one mote box.  We may choose to use the motes as a dual-purpose instrument: continuous at one sample rate and triggered at a higher rate.  In the longer-term, we propose instrument development for a large-scale urban experiment using the system that would fully characterize building responses before and after a future damaging earthquake.  In this experiment, we would deploy the wireless system in structures for 24 hours to do complete building identification.  After a future large, potentially damaging earthquake the instruments would be redeployed in order to determine if internal damage has altered the structure’s characteristics in a measurable, statistically significant way.

We plan to continue our collaboration with Heaton et al.  The main objective is to compute real and simulated displacements of the building-soil system using a 3D nonlinear static and dynamic finite element model to simulate building responses to various dynamic loads.  Now that our building model is finalized, we plan to carry out a series of numerical simulations that will predict building motions for scenario, large-amplitude earthquakes, and to test whether we can measure changes in building dynamic characteristics that are due to various types of damage scenarios such as broken columns or broken welds, analogous to those observed in downtown Los Angeles moment steel-frame high rises after the 1994 Northridge earthquake.  Related to this effort we also plan to examine source characteristics of relatively high amplitude shaking observed in during 2004.  In those data we observed temporary decreases in frequencies of modes of vibration that correlate in time with strong wind gusts.  However, the amount of frequency decrease is different depending on the mode examined.  We will examine whether the source spectrum has an effect on the observed changes in building characteristics. 

Finally, we are collaborating with Professor of Statistics Mark Hansen on a smart event-detection algorithm development project.  We have started with a simple elevator motion experiment to determine how to dynamically track locations and motions of vibrational sources such as the elevator cars, motors, and counterweights in the building’s waveform data.

People

Prof. Researcher: M. Kohler
Faculty: Paul Davis
Engineer: Igor Stubailo
Associates: Erdal Safak, USGS, Tom Heaton and Case Bradford, Caltech, Ramesh Govindan and Omprakash Gnawali (USC).