The UCLA Louis and Doris Factor building, was instrumented by the U.S. Geological Survey with an embedded 72 channel accelerometer network following the 1994 Northridge earthquake. The accelerometer network is distributed throughout the building and continuously recording building vibrations. In December 2003, the sensor network was upgraded by installing state-of-art data logging equipment and fiber optical network cables. To date, substantial data have been collected from ambient vibrations under different environmental conditions, as well as from low-amplitude vibrations from several earthquakes.

The Factor building, with its embedded sensor network, provides a unique platform for the identification of dynamic characteristics, structural performance monitoring, and damage detection. Establishing a reliable three-dimensional finite element model which accurately represents the stiffness of the structural system is an important step in assessing the structural performance and detecting damage under more significant shaking.

The objective of this site is to provide information on the Factor building, its structural systems, instrumentation, and data acquisition system. In addition, results from system identification, finite element model updating, and response predictions are presented.

1.   Building Description and Instrumentation

1.1.   History

1.2.   Architectural and Structural Systems

1.3    Embedded Sensor Network

2.   Vibration Data

2.1.   Data Acquisition

2.2.   Data Processing

2.3    Sample Data

2.4.   System Identification

3.   Finite Element Modeling

3.1.   Structural Drawings

3.2.   Modeling Assumptions

3.3.   Model Updating

3.4    Response Prediction

4.   1994 Northridge Earthquake

4.1.   Damage from Earthquake

4.2.   Re-evaluation of Modeling Assumptions

4.3.   Linear Dynamic Analysis

5.   References

6.   Downloads

Back to Top

1.   Building Description and Instrumentation

1.1.   History

Louis Factor (1907-1975) was son and successor to the cosmetic magnate Max Factor. For several years, Louis and his wife Doris, regularly attended lectures and classes at UCLA. Before his untimely death, Louis began preparations for a generous donation to the UCLA health sciences. Although he passed away during planning stages, Doris saw his dream fulfilled and made possible the construction of the Doris and Louis Factor Health Sciences Building. Standing at 216.5ft above ground, the Factor building is the tallest structure on campus. The building is considered home to the School of Nursing, the Jonsson Comprehensive Cancer Center, and other biomedical facilities.

            Factor Building Pictures

1.2.   Architectural and Structural Systems

The Factor building's lateral-load resisting system consists of a Type-I steel moment resisting frame (SMF). There are 12 bays in each direction with four of the EW bays having a column rotated 90 degrees and two bays in the NS direction. Column splices are nominally located 4 feet above the floor. The foundation consists of concrete bell caissons and spread footings. There are two basement levels and one level partly below grade; therefore, the ground level is at the first floor on the south side and at the second floor on the north side. The floor area for floors 10 through 16 (roof) increases by approximate 13.5% due to a slight overhang on the east and west faces of the building. For the most part, the building is symmetric about the East-West axis and slightly asymmetric about the North-South axis. Each floor consists of a 6¼" thick lightweight concrete slab on metal decking with a Norman Face brick veneer anchored to the floor slabs and steel studs with "quakeproof" dove tail anchors. A glass curtain wall consists of ¼" spandrel glass supported with an aluminum frame. The fifteenth floor houses mechanical equipment.

Typical Floor Plan

1.3    Embedded Sensor Network

Following the 1994 Northridge earthquake, the U.S. Geological Survey instrumented the structure with 72 uniaxial forced balanced accelerometers with an on-site recording system. Four accelerometers exist at each floor above grade, oriented to record translational motions near the perimeter of the floor (two in each direction). Each of the two basement levels has an accelerometer to record translation in two directions, as well as two accelerometers to record vertical responses.

In December 2003, the building sensor network was upgraded using funds provided primarily by the NSF Science and Technology Center for Embedded Networked Sensing (CENS) headquartered at UCLA. The upgrade consisted of converting all 72 channels to a 24-bit network continuously recording data using Kinemetrics Antelope seismic software. The streaming data is currently viewable via the internet in near real-time. This reconfiguration included rewiring all 72 channels and the installation of nine, eight-channel Quanterra Q1420 A/D data loggers, Figure 2‑4, recording at 100 and 500 samples per second which log data (approx. 3 Gigabytes/day) to a RAID database array. In addition, there is a 1.5 terabyte array for local storage to prevent data loss in the event of power outages.

In February 2005, a 100m deep borehole seismometer was installed approximately 50m away from the building. The level of instrumentation provided in and around the Factor building makes it one of the most (if not the most) densely permanently instrumented buildings in North America.
 

Back to Top

2.   Vibration Data

2.1.   Data Acquisition

  . . . Under Construction . . .

2.2.   Data Processing

Once the raw data is obtained, there are essentially three processing steps; detrending, calibrating, and decimating.

Detrending the raw data removes any linear trends or offsets from the data signals. In addition, detrending yields data in relative volts as opposed to absolute volts in which case the actual calibrated values would not make sense. In other words, the building’s natural motion is oscillating about zero not about some offset.

Converting the acceleration data; a, from counts to more useful units such as g’s is done using the following equation

where, CF is the calibration factor that was measured for each channel and SR is the sensor range, both of which are listed in the table of sensors:

Excel Spreadsheet:  Sensors.xls
Delimited Text:        Sensors.txt

The building motion is sampled at a rate of 100 or 500 samples per second (sps). Decimating at a lower sample rate not only reduces the data to a more manageable size, it reduces the effects of unavoidable noise and filters out high frequency content. For example, if a signal sampled at 100sps is resampled to 20sps, or a time step of 0.05s, then all noise above the nyquist frequency of 10Hz are removed. And, for a 60 sec sample time, that signal originally contains 6000 data points, but after decimation, is reduced to 1200 data points, significantly increasing computational efficiency.

Once the data processing is finished, the story accelerations can be calculated for each floor. The processed data collected from the four uniaxial accelerometers (u₁, u₂, v₁, and v₂) are used to compute the story accelerations (u_o, v_o, and θ_o) at the selected reference point. Coordinates of each sensor location are needed for these calculations and are provided in the table of sensors.

There are some errors as of (6/05) in the orientation and naming of several sensors that need to be taken in account.

1. Roof sensors E-GS and N-GE are switched: u₂↔v₂
2. 10^th floor sensor N-AE is reversed: v₂ = -v₂
3. 2^nd floor sensors E-2S, N-2W, and N-2E are switched: u₂→v₁, v₁→v₂, and v₂→u₂
4. Basement level A sensors N-XE and Z-XE are switched: u₁ ↔ z₁ (z is in the up direction)
5. Basement level B sensor Z-YW is unavailable.

2.3    Sample Data

Data collected on 04/29/2004 (Thursday) at 3:00 am are presented as an ambient vibration data set. This date and time were chosen as to minimize the effects of traffic in and outside the building. The recent Parkfield, CA earthquake (Mw = 6.0) occurred on 9/28/2004 17:15:24 UTC (10:15:24 AM local PDT). According to the USGS Earthquake Hazards Program, the epicenter was located at 35.815˚N & 120.374˚W, or approximately 163 miles from the Factor building (34.067˚N & -118.442 ˚W). Peak accelerations of  0.0025g were recorded at the roof of the building.

Below are data files in .mat form. They include the vibration data for all floors (except ground floor in case of Ambient vibrations) for EW, NS, and torsion. The time step is 0.05s.An FFT of each direction is displayed and gives some idea of the natural frequencies of the building.

Parkfield.mat         Ambient.mat

Fast Fourier Transforms

2.4.   System Identification

It is the goal of system identification to identify the relevant properties of an unknown system from measurements of the output(s) and input(s) of the given system. In regards to this project, the unknown system is the Factor building structure, the measured outputs are the story accelerations, and the measured input is the ground motion (no measured input in case of ambient vibrations). The inputs and outputs are recorded by the embedded sensor network. For a detailed discussion on system identification procedures, see Skolnik (2005). The results of system identification can be seen below.