SEISMIC ANALYSIS AND SHAKE TABLE MODELING:

USING A SHAKE TABLE FOR BUILDING ANALYSIS.

by

Sandra Brown

A Thesis Presented to the

FACULTY OF THE SCHOOL OF ARCHITECTURE
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements of the Degree
MASTERS OF BUILDING SCIENCE

May 2007

Copyright 2007 Sandra Brown

 Dedication

This thesis is dedicated to Erik Novales, for all of his

support and encouragement, and to my parents, Stacy

and Joseph Brown.

ii
 Acknowledgements

This thesis would not have been possible without the guidance and dedication of my

thesis committee members. Professor Goetz Schierle provided the inspiration and

the commitment to continue on, and Joseph Pingree provided expertise and

knowledge in subjects foreign to me. Professor Doug Noble kept me on track, and

Professor Marc Schiler kept me honest. From all of these people, I learned a great

deal.

iii
 Abstract

This Thesis is about the process of rehabilitating a shake table for use in seismic

analysis of small-scale models in the School of Architecture. Labview 8.0 Student

Edition was used to write the controlling program for the shake table.

In order to test seismic response of a prototype building, a 7-story reinforced

concrete building was modeled in piano wire and plywood and tested on the shake

table. The shake table recorded data from an accelerometer mounted on the model.

The model was built to have the same resonant frequency as the prototype building.

The model clearly shows modal forms and shows exaggerated deflection, as well as

torsion caused by modeling inconsistencies. Reactions in the model correlate to the

prototype. A model on a shake table is useful to the School of Architecture as a

teaching tool to visually highlight the effect strong ground motion can have on a

building.

Keywords: Shake Table, Labview 8.0, Seismic Analysis, Teaching Tool, Seismic
Modeling.

iv
 Table of Contents

Dedication.................................................................................................ii

Acknowledgements ................................................................................ iii

Abstract....................................................................................................iv

Table of Figures and Tables ................................................................. viii

Chapter One: Introduction.......................................................................1

Thesis Outline ......................................................................................................... 1

1.2 Seismology........................................................................................................ 2
1.3 Lateral Forces................................................................................................... 4
1.4 Damages from Seismic Forces.......................................................................... 5
1.5 Model Testing ................................................................................................... 6
1.6 Shake Tables ..................................................................................................... 7
1.7 Understanding versus Memorization ................................................................ 8
1.8 How to use the results of this research.............................................................. 9
1.9 Definitions of terms and formula .................................................................... 10

Chapter 2: Seismic Forces and Shake Table Analysis ..........................14

2.1 Faults ............................................................................................................... 14

2.2 Seismic Forces ................................................................................................ 17
2.3 Building reaction to Seismic Forces ............................................................... 20
2.4 Model Analysis ............................................................................................... 23
2.5 Shake Tables ................................................................................................... 23
2.6 G G Schierle Shake Table ............................................................................... 24
2.7 Previous Work at USC .................................................................................... 25

Chapter 3: Methodology........................................................................28

3.1 The G. G. Schierle Shake Table...................................................................... 28

3.2 Amplifier ......................................................................................................... 30
3.3 Digital/Analog Converter................................................................................ 34
3.4 Labview 8.0 Student Version.......................................................................... 34
3.5 Shaker.............................................................................................................. 35
v
 3.6 Earthquake Data .............................................................................................. 36
3.7 Building the Model ......................................................................................... 36
3.8 Contingency Plans........................................................................................... 36

Chapter 4: Fixing the Shake Table ........................................................38

4.1 Original Condition .......................................................................................... 38

4.2 Procedure for Fixing the Shake Table............................................................. 39
4.3 Software .......................................................................................................... 50
4.4 Problems.......................................................................................................... 52
4.5 Troubleshooting .............................................................................................. 52

Chapter 5: Building a Test Model .........................................................54

5.1 Types of models already in use....................................................................... 54

5.2 Selecting a Model Type .................................................................................. 55
5.3 Modeling a Real Building ............................................................................... 56
5.4 Symbols........................................................................................................... 59
5.5 Calculations..................................................................................................... 60
5.6 Building the Model ......................................................................................... 63
5.7 Fixing the Model to the Shake Table .............................................................. 65
5.8 Testing............................................................................................................. 66

Chapter 6: Running a shake table test ...................................................68

6.1 Installing the Model ........................................................................................ 68

6.2 Selecting an earthquake data file..................................................................... 71
6.3 Verifying Input and Output results ................................................................. 73
6.4 Running the test .............................................................................................. 74
6.5 Possible Problems ........................................................................................... 77

Chapter 7: Results and Analysis............................................................80

7.1 Introduction of Testing.................................................................................... 80

7.2 Verification Tests ............................................................................................ 82
7.3 Model Testing and Results.............................................................................. 87

vi
Chapter 8: Conclusions........................................................................100

8.1 Correlations to prototype building data......................................................... 101

8.2 Concluding Remarks..................................................................................... 102

Chapter 9: Future Work.......................................................................103

9.1 3-Degree of Motion Shake Table.................................................................. 103

9.2 Different Modeling Techniques .................................................................... 103
9.3 Soils Testing.................................................................................................. 104
9.4 Calibration..................................................................................................... 105
9.5 Better Measurement Devices ........................................................................ 105
9.6 More Earthquake Files .................................................................................. 106
9.7 Earthquake Remediation Strategies .............................................................. 106
9.8 Strategies for use as a teaching tool ........................................................ 107

Bibliography .........................................................................................109

Appendix A : Instructions for Using Labview ....................................111

Appendix B: Index of Videos Submitted .............................................114

vii
Table of Figures and Tables

Figure 1.1: Fatalities of Major Earthquakes 1970-1999 (Schierle 2005 pg 29) ........ 3
Figure 1.2: 10 Freeway at Venice Blvd. after Northridge. ......................................... 5
Figure 1.3: G G Schierle Shake Table with testing model......................................... 7
Figure 1.4: G. G. Schierle Shake Table ................................................................... 11
Figure 1.5 National Instruments USB-6008, Digital Analog Converter,.................. 12
Figure 1.6 Labview Interface Front Screen............................................................... 13
Figure 2.1: Major Tectonic Plate Boundaries .......................................................... 14
Figure 2.2: Types of Faulting in the Earth's Crust ................................................... 15
Figure 2.3: Earthquake Faults in Southern California ............................................. 16
Figure 2.4: Types of Earthquake Waves .................................................................. 19
Figure 2.5: An example of earthquake damage in the 1994 Northridge Earthquake.
........................................................................................................................... 22
Figure 3.1 The G. G. Schierle Shake Table. ............................................................ 29
Figure 3.2: Picture of fuse used to replace broken fuses........................................... 31
Figure 3.3: A Male BNC Connector used in connecting the digital analog converter
to the amplifier. ................................................................................................. 32
Figure 3.4: The back of the amplifier ....................................................................... 33
Figure 3.5: The Shaker Component of the Shake Table. ......................................... 35
Figure 4.1: Block Diagram showing how the shake table components connect to
each other. ......................................................................................................... 40
Figure 4.2: Shaker Component 3-Pin Connector Diagram ...................................... 42
Figure 4.3: Circuit designed by Joseph Pingree for DAC........................................ 43
Figure 4.4: Diagram of the Pin Connections in the TL082 Op Amp that was used to
make the DC to DC converter added to the DAC............................................. 43
Figure 4.5: Image of the DAC inside a custom metal box....................................... 44
Figure 4.6: The author soldering the circuit to convert a positive voltage to a +/-
voltage. .............................................................................................................. 44
Figure 4.7: Functional Block Diagram from Analog, Inc........................................ 45
Figure 4.8: The Pin Diagram for the Accelerometer, from Analog Devices, Inc. ... 46
Figure 4.9: Circuit design for the amplifying circuit of the accelerometer.............. 47
Figure 4.10: The amplifying circuit tested on a breadboard. ................................... 48
Figure 4.11: Oscilloscope, and Joseph Pingree testing the circuit of the
accelerometer and amplifier.............................................................................. 49
Figure 4.12: The amplifier and the accelerometer mounted to the shake table and
connected to the DAC with wires and Molex Connectors................................ 50
Figure 4.13: Labview Front Page............................................................................. 51
Figure 5.1: Locations of the CR-1 recording system in the Van Nuys Building.
Image credited to Prof. Trifunac at USC. ......................................................... 58
Figure 5.2: Compressive Strength of the Columns in the Prototype Building. ....... 61
Figure 5.3: Critical loading for the columns in the prototype building. .................. 62
Figure 5.4: Slenderness ratio for columns in the prototype building........................ 62
Figure 5.5: Model dimensions, section. ................................................................... 64
viii
Figure 5.6: Shake Table Plan for Mounting Models (Units in Inches).................... 65
Figure 5.7: Diagram of Bracing in Prototype Building. (Trifunac 2001 pg. 15) .... 67
Figure 6.1: Screen shot of Simulated File Controls .................................................. 70
Figure 6.2: Lead fishing weights used for model testing. ........................................ 71
Figure 6.3: Data Extracting Program ....................................................................... 73
Figure 6.4: Input and Output graphs ......................................................................... 74
Figure 6.5: Labview File Input ................................................................................ 74
Figure 6.6: Timing Parameters................................................................................. 75
Figure 6.7: Illustration of "Run" Button .................................................................. 76
Figure 6.8: The Stop Button..................................................................................... 76
Table 7.1: Table of Tests run on the Shake Table.................................................... 82
Table 7.2: Accelerometer test 6-8. ........................................................................... 83
Table 7.3: Test 11, paper displacement tests............................................................ 83
Table 7.4: Test of Noise Filters in the Input of the Accelerometer data.................. 84
Table 7.5: Input/Output verification test at 1 Hz ..................................................... 85
Table 7.6: Tests 18, 19, and 20, Roof ...................................................................... 87
Table 7.7: Prototype Building Roof Reaction to Northridge Earthquake 1994....... 88
Table 7.8: Tests 18, 19, and 20 5th Floor................................................................. 90
Table 7.9: Prototype Building 5th Floor Reaction to Northridge Earthquake 1994.91
Table 7.10: Model Test 3rd Floor Reactions ........................................................... 92
Table 7.11: Model Test 2nd Floor............................................................................ 93
Table 7.12: Model Test of Bracing- roof reaction ................................................... 95
Table 7.13: Bracing on the Model- 5th Floor .......................................................... 96
Table 7.14: Model Bracing Test- 3rd Floor ............................................................. 97
Table 7.15: Bracing Test 2nd Floor ......................................................................... 98
Figure 9.4: Types of eccentric braced frames ........................................................ 107
Figure A.1: Labview Front Page............................................................................ 112
Figure A.2: Labview Block Diagram..................................................................... 113

ix
Chapter One: Introduction

The University of Southern California’s School of Architecture G. G Schierle shake

table can be used in a meaningful way to study a building’s seismic response using

existing seismic data. The G. G Schierle Shake Table is an important teaching tool

that is useful to show aspiring architects and engineers how a structure will respond

in a seismic event. With first an understanding of how the shake table was damaged,

as well as an update in the controller interface for the shake table, the Schierle Shake

Table was fixed and is once again operational. Since the shake table is now

operational and updated, models are used to test the shake table and calibrate the

controlling program and devices.

Thesis Outline
In Chapter 2, the causes of earthquakes and the history of seismic modeling on a

shake table will be discussed, as well as previous research using the Schierle Shake

Table.

In Chapter 3, the methodology of modeling a specific building on the shake table

will be examined, as well as an explanation of how the shake table components work

together to simulate a seismic event.

Chapter 4 comprised documentation of how the shake table was actually fixed, and

how all the components work together.

1
In Chapter 5, the modeling technique used as well as the guidelines for selecting

modeling materials were explained.

In Chapter 6, the process of running the experiment was described. Possible

problems were listed and potential solutions were listed.

In Chapter 7, the findings of the actual test of the seismic model were documented.

The model were tested using the 1994 Northridge Earthquake data and tested the

case study building’s deflection prior to the seismic bracing and following the

addition of the seismic bracing.

Chapter 8 evaluated the relevance of the data from the test to determine if future

tests in this manner will yield reliable results.

Chapter 9 summarized and concluded the findings.

Chapter 10 comprised suggestions for future work.

1.2 Seismology

Seismology is a relatively recent field of study. Seismology is the study of

earthquakes, which are caused mostly by plate tectonics, the huge pieces of the

earth’s crust as they move relative to each other, which causes strain in the

2
intersecting fault lines. One of the major plate boundaries occurs in California at the

boundaries of the Pacific Plate and the Continental Plate; known as the San Andreas

Fault. This is a strike slip fault, meaning it moves laterally along the fault. The

study of earthquakes is important because earthquakes cause billions of dollars in

damage every year around the world, and thousands of deaths and tens of thousands

of injuries. In the United States, around 1,200 deaths have been recorded since

1900. Many more fatalities occurred in earthquakes elsewhere, see fig. 1.1. Most

deaths in earthquakes are caused when a structure collapses (Congress 1995 pg 6).

Figure 1.1: Fatalities of Major Earthquakes 1970-1999 (Schierle 2005 pg 29)

3
1.3 Lateral Forces

The strong lateral movement caused by a seismic event can cause structural damage

to a building. Buildings are designed for gravity, and thus are usually sufficiently

designed for vertical movement. Lateral movements, however, can cause large

shear forces to the walls, introduce bending to columns, and torsion if the center of

mass and center of resistance are offset. Most structural failure is due to buildings

not having enough bracing or shear resistance or moment resistance. Structures of

brittle material are more vulnerable than structures of ductile material.

More recently, research from the 1994 Northridge event has led to a re-evaluation of

the assumption that lateral forces are the governing forces in terms of building

damage in seismic events. Strong vertical forces over the epicenter of the event

caused damage to freeway pillars on the Santa Monica Freeway, leading to crushing

of the pillars. Strong lateral forces in addition to strong vertical forces that introduce

pounding to columns and can lead to containment issues in concrete beams and

columns. See Figure 1.2 for a picture of the 10 Freeway at Venice Boulevard, just

after the Northridge event.

4
 Figure 1.2: 10 Freeway at Venice Blvd. after Northridge.

1.4 Damages from Seismic Forces

Earthquake forces act similarly to sound waves, in the way they propagate through

the soil. They can be produced at different frequencies and at different amplitudes.

Large earthquakes tend to produce larger amplitude, lower frequency seismic waves,

whereas small earthquakes tend to have smaller amplitudes but higher frequency

waves. This is, however, only a generalization, as each earthquake has a variety of

complex waveforms of various amplitudes and frequencies. The damage done to

structures depends mostly on the interaction between soil and structure and how

these waves hit the structure. Seismic waves can move vertically, horizontally, or a

combination of both, and can come from any direction. Higher frequency
5
earthquakes tend to damage shorter, stiffer structures, and lower frequency

earthquakes tend to damage taller, more ductile structures. Buildings with the same

period of a seismic event tend to resonate and be more damaged. Buildings have a

resonant period of about 0.1 second per story, so a 10-story building would have a

resonant period of 1 second.

1.5 Model Testing

Given the failures due to eccentricity, insufficient strength, etc., it is important that

architects understand the potential hazards in their design and understand where the

building needs to be braced. Testing of models of actual buildings and building

prototypes is one method that is useful in understanding the forces at work. Models

built of plywood and steel wire allow students to see the seismic behavior of a

structure, and to understand how the period of an earthquake if it is resonant with a

building period, will cause the most damage or even collapse. Models may be tested

on a shake table or in a computer program. However, shake tables are more

effective as a teaching tool to simulate seismic behavior because they most closely

simulate human error in the construction process and allow for easy understanding

of the building reaction to the seismic forces.

6
1.6 Shake Tables

The shake table is a device that simulates a seismic event. It can also be used to

create fictional “worst case” scenarios or resonant frequencies. In computer

controlled shake tables a computer program generates a signal, and a digital signal is

sent to a digital/analog converter, which sends a voltage to the amplifier. The

amplifier amplifies the voltage and sends it to the shaker platform to which the

model is attached. The Schierle Shake Table is a one-degree of motion shake table,

meaning that it will move only in one lateral direction.

Figure 1.3: G G Schierle Shake Table with testing model.

7
A model on a shake table with the same stiffness or resonant frequency as the

prototype building, will act in a way similar to that of the actual building.

Mathematical equations and formula alone are not effective to convey seismic

behavior to students. In a hands-on pedagogical method, such as a model on a shake

table, students see the effects of seismic forces on a building and are better prepared

to apply the formula learned to an actual situation. Students then have better

understanding of structures and a greater respect for seismic forces.

1.7 Understanding versus Memorization

In an abstract teaching system, students are given formulae and facts to memorize,

sometimes with a lack of fundamental understanding of the physical laws behind the

formula. In a professional school such as architecture, students are required to take

the knowledge from school and apply that to the creation of a building. As any

architect knows, no two buildings are exactly alike, and each site and building pose

new challenges. Rote memorization does little to prepare a student for application

of knowledge to a unique and specific challenge. Only a fundamental understanding

of the knowledge allows one to apply the correct formula or equation to the right

problem. Eric Mazur, of Harvard University’s Department of Physics, wrote a paper

about physics students’ lack of understanding of memorized problems. He

discovered in his physics classes that peer instruction and physical demonstrations

greatly improved student understanding of physic concepts (Mazur 1995 pg. 5).

Mazur compared the students’ final examination scores from a class he taught

8
conventionally, and the same final examination given to students taught via peer

instruction and physical demonstrations. The students taught with the hands-on

process scored much higher than the conventionally taught students (Mazur 1995 pg.

7). Students in classes taught with peer instruction and in class demonstrations

scored 30 to 70% higher on average than students taught in a traditional class

(Mazur 1999 pg. 3).

1.8 How to use the results of this research

Students in the structural classes in the School of Architecture at the University of

Southern California will be able to build models of their architectural projects for

testing in a seismic event. A student would then be able to see that form (models

are not of the real material) has a large effect on how a building will react to seismic

forces, and then be able to make informed decisions on how best to proceed with

their design.

Students will also be able to build simple models to test the effects of earthquake

bracing on a general building, or test seismic dampeners. Students will gain an

intuitive sense of how a building will react, and can use this teaching tool as a way

to create new and innovative ways of dealing with an unpredictable and damaging

natural disaster.

9
1.9 Definitions of terms and formula

Shake Table: The shake table is the G. G Schierle Shake Table located in the

Masters of Building Science Laboratory consists of a Digital/Analog converter, APS

Dynamics Electro-Seis shaker model 113 and an APS Dynamics amplifier model

114 and a platform suspended from a steel frame. It was originally built in 1981 and

was intended to be used as a teaching tool in the School of Architecture. It has a

steel frame and a wooden platform for the amplifier and the controlling computer.

The shaker is bolted to the base of the frame, and connected to an aluminum

platform with a long bolt. The platform is suspended from the top of the frame with

cables and cross-braced to reduce the introduction of torsion into the model. The

platform has holes drilled 3” on center regularly spaced in a grid, for ease of

securing models for tests. The whole frame is 5’ tall, and 3’ wide.

10
Figure 1.4: G. G. Schierle Shake Table

11
Digital/Analog Converter:

The digital/analog converter is a National Instruments USB-6008 DAC that takes a

digital signal and converts it to a voltage for use with the amplifier.

Figure 1.5 National Instruments USB-6008, Digital Analog Converter,

(http://sine.ni.com/nips/cds/view/p/lang/en/nid/14604)

12
Labview 8.0, Student Version: Labview is a graphic programming language sold by

National Instruments. It uses the C programming language in a graphic way to

control testing instrumentation.

Figure 1.6 Labview Interface Front Screen

The next chapter explores the background of seismic events and modeling.

13
Chapter 2: Seismic Forces and Shake Table Analysis

This chapter will cover the following four things: what earthquake and seismic

events are and how they work, how buildings respond to those forces, what the G. G

Schierle Shake Table is, and the previous research at USC using the shake table.

2.1 Faults
An earthquake occurs when built up energy is released in a sudden slippage of a

fault. Faults, or cracks in the earth’s surface, occur primarily at the edges of tectonic

plates, large pieces of the earth’s crust. (Fig. 2.1). There are three types of faults:

Normal, Reverse, and Strike Slip Faults (Fig. 2.2).

Figure 2.1: Major Tectonic Plate Boundaries

http://www.cev.washington.edu/lc/CEVIMAGES/global-techtonic-plates.jpg

14
Normal faults are faults where the hanging wall, the side of the fault that hangs over

the fault, moves downward in relationship to the footwall. An example of a normal

fault is the Mid-Atlantic Ridge, a spreading plate boundary.

Figure 2.2: Types of Faulting in the Earth's Crust

http://earthquake.usgs.gov/images/faq/3faults.gif

Reverse Thrust Faults are faults where the footwall moves upward in relationship to

the hanging wall. An example of a reverse thrust fault is the Sierra Nevada Fault

zone, which has caused the Sierra Nevada mountain range in California.

15
Strike Slip faults are the faults most relevant to Southern California. A Strike Slip

fault is a fault in which one side of a fault moves horizontally in relationship to the

other side of the fault typically parallel to the fault. The San Andreas Fault System

that forms part of the boundary between the North American Plate and the Pacific

Plate is a Strike Slip fault. Strike Slip Faults are either Right Lateral Faults,

meaning that one side of the fault moves from left to right in relationship to a viewer

on the opposite side of the fault, or Left Lateral Faults, in which the motion of the

fault moves from right to left in relationship to a viewer on the opposite side of the

fault. See Figure 2.3 for more information on the faults specific to Southern

California.

Figure 2.3: Earthquake Faults in Southern California

http://www.earthquakecountry.info/roots/inline/11839sm.jpg

16
The San Andreas Fault, which is responsible for some of the largest earthquakes in

California’s history, is a left lateral strike slip fault. A rupture on the southern part

of the San Andreas Fault could unleash an earthquake upwards of Magnitude 8.0.

There are no measurements of any earthquakes occurring on the lower part of the

San Andreas Fault. The last known earthquake was the 1857 Fort Tejon Earthquake,

with a recorded Modified Mercalli intensity from X to XI, with a peak ground

acceleration of more than 0.60g, where g = gravity acceleration.. Almost all un-

reinforced masonry buildings were destroyed and the ground badly cracked. There

was a 9-meter displacement, and the fault ruptured for 300 miles, from Parkfield to

Wrightwood, California. Two buildings were destroyed and three were heavily

damaged and considered uninhabitable at the army outpost of Fort Tejon.

2.2 Seismic Forces

When a fault ruptures, it releases a large amount of stored energy. This energy

radiates out from the epicenter, the point on the earth’s surface where the rupture

starts. There are four types of waves. Two of the three are called body waves,

which propagate within a body of rock and radiate out from the epicenter of the

earthquake. The two body waves are P-waves, and S-waves. The third and fourth

types of waves are surface waves, the Love wave and the Rayleigh wave.

P-waves, or primary waves, are the fastest waves, and are compression waves,

meaning they have a push and pull type of motion. P-waves act similarly to sound
17
waves, and move through both solid rock and liquid material. S-waves, or

secondary waves, are shear waves that shear the rock sideways at right angles to its

direction of travel. These waves are slower than P-waves, and cannot travel through

liquid. While P-waves act like sound waves, the S-wave acts more like a sine wave

(Fig. 2.4).

18
 Figure 2.4: Types of Earthquake Waves

http://www.darylscience.com/graphics/seiswave.gif

19
When an earthquake occurs, the high-speed P-waves are felt first, in an effect that

rattles windows and can sometimes sound like a sonic boom. The S-waves arrive

with a vertical displacement and a lateral displacement, and are the waves most

likely to damage a building. (Bolt 2004 pg. 20) The time lag between wave arrivals

defines the distance of an earthquake. The distance from three seismic stations

defines the epicenter location. Surface waves travel near the earth surface.

There are two types of surface wave, the Love wave and the Rayleigh wave. Love

waves have a side-to-side movement along the horizontal plane of the earth’s

surface. It has no vertical displacement, and this horizontal shaking is particularly

damaging to a building’s foundations. The Rayleigh wave acts more like an ocean

wave, with an elliptical motion both vertically and horizontally in a vertical plane in

the direction of wave propagation. These surface waves are usually much slower

than the body waves, and in an earthquake, the first moments of shaking are body

waves, followed then by Love waves, which are faster than Rayleigh waves.

2.3 Building reaction to Seismic Forces

Typically, an earthquake can cause four types of damage to a building. A building

can collapse, which can result in the total loss of the building and possibly the lives

of the occupants. A building can suffer structural damage, which leaves the building

standing, but unsafe, and either results in the eventual demolition of the building or

expensive remediation costs to repair the structural damage. A building may also
20
suffer non-structural damage to walls, water pipes, windows, and so forth. These

costs can be expensive to repair, but are preferable to losing lives. Non-structural

damage usually amounts to over 70 percent of total damage (Schierle, 2001, p.1-1).

Lastly, a building might suffer damage to the contents inside, which result from

objects not being properly anchored to walls or otherwise properly secured.

(Congress 1995 pg. 8)

Engineers and architects hope that a building in a seismically active area would

suffer minimal damage, but of the four types, a designer would prefer cosmetic

damage to structural damage or collapse, in an effort to preserve life safety. Of

course, architects would prefer no damage, but due to the nature of seismisivity,

earthquakes are unpredictable, vary in terms of magnitude, strength, period, and

peak ground acceleration. No two earthquakes are alike, and the effects of

earthquake strength can still surprise engineers and seismologists.

21
 Figure 2.5: An example of earthquake damage in the 1994 Northridge Earthquake.

This is a parking structure in Northridge that collapsed, showing the ductility in the concrete.
http://www.calstatela.edu/dept/geology/earthquakes/CSUNParking(2).jpg

Primarily, architects and engineers are concerned with the lateral forces that

earthquakes generate. The rationale is that structural engineers already design for

vertical gravity dead loads and live loads. Because designers include a safety factor

to compensate for unexpected loads in the vertical direction, it is assumed that

vertical forces are not necessarily the problem in an earthquake. Therefore, lateral

forces tend to govern earthquake resistant design in building codes and in practice.

As a note of caution to designers, directly over the epicenter in the Northridge event,

strong vertical acceleration was recorded, and the resulting combination of strong

vertical and lateral forces caused loss of containment in the concrete columns

supporting freeways and buildings.

22
2.4 Model Analysis

Model Analysis is the use of physical or computer models to test shaking or

vibration of an object. Architects and engineers use equations defined by codes to

determine the resonant period of a structure, which is useful to know because a

building will suffer the most physical damage during a seismic event that has the

same frequency as the resonant building frequency. If models are used they must

have similitude to the actual structure that is being studied. A model is said to have

similitude with an actual structure if it has similar geometry, dynamic properties,

and period. Geometric similarity means that the model is a scaled down version of

the actual building, in the same shape. Dynamic similarity means that the ratios of

all forces acting on the building are consistent.

2.5 Shake Tables

Prior to the invention of the seismograph in 1880, there was no way to accurately

test a seismic event. With the invention of the Richter Scale in 1935, there was a

way to measure and compare earthquakes. In the aftermath of the 1906 earthquake

in San Francisco, earthquake research in the United States of America was pushed to

the forefront of geologic studies in California. Two universities in the San Francisco

area, Stanford University and the University of California, Berkeley made major

research gains documenting the effects of the earthquake and the aftermath thereof.

23
2.6 G G Schierle Shake Table
Because architects are primarily concerned with the lateral forces added to a

building in a seismic event, the University of Southern California Chase Leavitt

Graduate Building Science Program has a shake table to visually analyze the effects

of a seismic event to a building model. The Schierle Shake Table is a one-degree of

motion shake table, built by Professor Schierle and students. The shake table

includes:

• Steel frame

• Computer

• Suspended aluminum shaker platform

• APS systems Electro-seis 113 shaker component

• Digital/analog converter

• Amplifier component is Model 114

The Schierle Shake Table accepts electronic input via a digital analog converter,

which is a component in the controlling computer. The amplifier takes +/- 2 volts

from a digital/analog converter and modulates the voltage up to 220 volts,

maximum, and 4 amperes are amplified to 6 amperes. Using Labview 8.0, Student

version, from National Instruments, Inc. to input the data and control the output

voltage, one can send actual earthquake waveform data to the shaker to simulate

building response to the seismic event.

24
Using a shake table for modeling building response during a seismic event is useful

to both teachers and students trying to grasp how structures respond to strong lateral

forces. Students can see the deflection in a model, and see the inflection points in

columns. If one uses fabric in the model to simulate walls, a student can see actual

shear forces acting on the fabric. With shake table simulation students see a

visualization of the forces to help understand the building response.

2.7 Previous Work at USC

Professors G. Goetz Schierle, James Ambrose, and Dimitri Vergun, and students

have previously used the shake table to test models under seismic forces.

The theses regarding the shake table in Professor Schierle’s possession are:

H Iriano (1988) Response of High-rise Structures to Lateral Seismic Ground

Movements S Chong (1993) Investigation of Seismic Isolators as a mass Damper for

Mixed Use Buildings

Chong’s Thesis was the last to use the shake table.

The shake table stopped working sometime thereafter.

The testing was thorough, but lacks the specific information on the shake table to

use to restore it. More thorough documentation on the testing equipment would

have been helpful to the restoration of the shake table. However, the notes on model

building were thorough and helpful.

25
No other work mentioning the Shake Table at USC was available.

However, shake tables have been used before in academic environments. The

Network for Earthquake Engineering Simulation (NEES) at Berkeley, California,

first built their large scale shake table in 1972. This table is 20’ x 20’, and has a 3

degree range of motion. It may be used to subject structures weighing 100,000 lbs

to horizontal accelerations of 1.5g. (http://eerc.berkeley.edu/lab/earthquake-

simulator-lab.html). A 3 degree shake table is more accurate than the 1 degree of

lateral motion, because earthquakes cause 3-degree motions that a 1 degree shake

table cannot simulate.

The University of San Diego has the largest outdoor shake table. This shake table is

only a 1 degree of lateral motion shake table, and is 25’ x 40’ in dimension. San

Diego built a 65’ tall concrete building to test that less reinforcing steel is a more

effective strategy in concrete buildings, defined as ductile design required for

concrete structures since 1976. The researchers then tested the building using the

1994 Northridge earthquake event. Their findings support their hypothesis; that

excessive building strength can actually promote poor structural performance and

non-structural damage

(http://www.jacobsschool.ucsd.edu/news/news_releases/release.sfe?id=508). A

National Geographic Special described one critical flaw in their testing in that they

did not take into account the vertical movements in an earthquake, the containment

26
of the concrete in the columns was inadequate, and would have led to the structural

failure of the columns in the actual event (National Geographic 2006).

Small scale testing has occurred at California State University Northridge. Small

scale testing on a shake table is not as accurate as large scale testing because of the

difficulty of scaling down the material properties of actual building materials.

However, small scale testing is good regarding seismic design concepts, such as

shear cracks, building deflection, torsion, and stiffness of a building. For teaching a

conceptual class, such as in an architecture school, such small-scale tests are more

useful because they can be completed in a short amount of time with a minimum of

model building materials.

In the next chapter, a brief introduction to how the G. G Schierle Shake Table

operates and how models can be built to test conceptual ideas will be discussed.

27
Chapter 3: Methodology

Chapter 2 contained the basics of why earthquakes occur, and why building models

to test on a shake table are useful to examine how a building will roughly respond to

a seismic event.

3.1 The G. G. Schierle Shake Table

The G. G. S. Shake table is composed of four major components:

• Steel frame

• Computer

• Suspended aluminum shake platform

• Digital analog converter (converts a digital waveform file to a voltage)

• Amplifier (amplifies a small voltage of the digital/analog converter outputs

to a larger current to control the shaker)

28
Figure 3.1 The G. G. Schierle Shake Table.

29
3.2 Amplifier

The amplifier takes a voltage of +/- 2 volts maximum input, and amplifies the

voltage up to +/- 220 volts, maximum output needed to activate the shake table. The

amplifier draws 300 Watts of initial Input Power, and amplifies the current from 4

amps to 6 amps. The amplifier draws 1320 maximum Watts. The amplifier, in

order to work properly, needed to be thoroughly dusted and cleaned, and the fuses

that regulate the input voltage and current needed to be checked and replaced. The

amplifier uses two glass fuses, 250V 4 amp. The current fuse used is an SOC SS2

(Figure 3.2). This fuse is a fusible link made of glass and ceramic to protect against

current surge. The amplifier is connected to the Digital Analog Converter using a

BNC connector. The BNC connector is a type of RF connector used for terminating

coaxial cable (Figure 3.3).

30
Figure 3.2: Picture of fuse used to replace broken fuses.

31
 Figure 3.3: A Male BNC Connector used in connecting the digital analog converter to the

amplifier.

http://upload.wikimedia.org/wikipedia/commons/thumb/a/a0/BNC_connector.jpg/639px-
BNC_connector.jpg

The amplifier has two circuit boards inside, two capacitors, and a large heat sink and

fan. Using a voltmeter and a multimeter, the digital analog converter sent a test

voltage to the amplifier using the test functions of the digital analog converter to test

the signal in the amplifier. A multimeter is an electronic measuring instrument that

combines several functions in one unit. The most basic instruments include an

ammeter, voltmeter, and ohmmeter. A voltmeter measures only voltage in a circuit.

32
The test function option will automatically pop up on the computer screen when one

plugs the USB cable of the digital analog converter into the USB port.

Figure 3.4: The back of the amplifier

Note the large cooling fan and the circuit that controls the input and output power. In the

multicolored ribbon cable, the orange, yellow, and brown wires seem to handle the input voltage..

Figure 3.4 shows the large cooling fan and the circuit that controls the input and

output power. In the multicolored ribbon cable, the orange, yellow, and brown wires

seem to handle the input voltage. Follow the orange, yellow, and brown wires in the

33
amplifier circuitry to assure that the voltage is properly flowing from the input and

to the output cables. The color-coding seems consistent throughout the amplifier,

however, not all wires were tested and these colors may not be the only wires to

carry voltage and current. There will be a maximum of 42 volts at the capacitors if

the circuitry is working properly. The amplifier needs be grounded to the casing.

With the red probe of the multimeter, switched to voltage, verify that there is 42

volts at the blue capacitors. The black probe should be attached to the case, which is

grounded.

3.3 Digital/Analog Converter

The digital/analog converter used is a National Instruments USB 6008. This device

converts a digital signal, i.e. a waveform, into an analog signal, or a voltage. The

USB device connects to a computer using a USB cable. The device requires a driver

that is available from National Instruments (See Appendix 1). The driver allows one

to test the device and the device is self-calibrating.

3.4 Labview 8.0 Student Version

A computer program called Labview 8.0 Student Version was included with the

USB Digital Analog Converter. Labview 8 includes all of the drivers for the testing

instruments provided by National Instruments, Inc. The program includes various

examples to use in creating a controlling program for the G.G.S. Shake Table.

34
Labview uses the “C” programming language in a graphic way to create a

controlling program for the National Instruments components. Please see appendix

2 for a diagram of the program used to control the shake table and other

documentation.

3.5 Shaker
The shaker component of the G. G. S. Shake Table has been operational from the

beginning. The internal components, one being a rubber band like piece, might need

replacement in the near future, and the components might need to be lubricated. The

shaker component was thoroughly cleaned and dusted. See Figure 3.5.

Figure 3.5: The Shaker Component of the Shake Table.

35
3.6 Earthquake Data

Earthquake data is acquired from the United States Geological Survey, downloaded

from http://quake.usgs.gov/info/data.html. There are three components to each data

file: peak ground acceleration, peak ground velocity, and displacement.

Displacement is the part of the data file used in Labview to run the shake table.

3.7 Building the Model

The model was built based on a prototype building that was instrumented by USGS.

The building has digitized records of many seismic events, and is a good building to

study for testing. Further information about building a model is found in Chapter 5.

3.8 Contingency Plans

Since at the first try, the accelerometers seemed not to be working properly,

measurements of maximum displacement were made by placing a large piece of

paper behind the shake table. A writing implement such as a pencil or crayon was

firmly attached to the model at floor levels corresponding to prototype accelerograph

locations. The pencil had to be in constant contact with the paper to properly record

the displacement. The length of the line on the paper is the maximum displacement

of the building scaled to the model. However, this measurement technique is

questionable because of the torsion of the model while testing. The pencil could not

stay in contact with the paper since the model did not move in a manner parallel to

36
the plane of the paper. The model had torsion because of human error in building

the model, and very slight differences in how the columns were located on the floor

plate. Eccentricities in how the weights were placed on the floor plates might also

have contributed to the torsion in the model.

37
Chapter 4: Fixing the Shake Table

4.1 Original Condition

The last time that the shake table was used was in 1993, when Sammy Chong used

the shake table to test mass dampeners. Since then, it has fallen into disrepair. The

controlling computer, a 1980’s IBM PC, would not turn on. The IBM PC had these

components:

• A digital analog converter

• A hard-drive with earthquake files on it

• A program given to the School of Architecture that would run the shake table

The first step towards fixing the shake table was to take the old IBM PC apart, and

see if any of these components were salvageable. When the computer was opened

up, the digital analog converter was not obvious to spot. This component was a PCI

card that connected to the parallel printer port on another board. It was not

salvageable. The hard-drive was also deemed unsalvageable because of the high

possibility that it had rusted solid.

The shake table platform would move manually if the amplifier was on. Thus, the

amplifier and shaker component were assumed to be in working condition, and the

main problem was the computer that had controlled the shake table.

38
4.2 Procedure for Fixing the Shake Table

After contacting the manufacturer of the shake table and amplifier, APS Dynamics,

an operational manual was acquired. This manual did not include a troubleshooting

section, but did include instructions for cleaning the shaker component. The manual

did give some helpful guidelines, such as the amplifier would only accept a +/- 2

volt input.

Using the above guideline, a USB digital analog converter (DAC) was purchased for

the shake table from National Instruments, part number 779051-01. This device had

2 analog outputs, as well as 6 analog inputs and 8 digital inputs/outputs. The digital

component was not important to the function of the shake table at the moment, but

future improvements may want to incorporate such functions. The analog outputs

on the device provide a 0- 5 volt output.

The next step was to connect the amplifier to the DAC. The amplifier had

previously been connected to the controlling computer with a bayonet Neill-

Concelman (BNC) connector, and then an adapter turned the BNC connector to a

26-pin male connector, which plugged into the serial port in the IBM. A BNC

connector was purchased from Amazon.com and screwed into the DAC.

The DAC came with a utility from National Instruments that allowed the device to

be tested, sending out specified voltages from 0-5 volts. After connecting the DAC

39
to both the computer and the amplifier, test voltages were sent to the amplifier, to

see if anything happened at the shaker. Nothing did. Taking apart the amplifier, it

was discovered that one of the fuses was burned out.

Figure 4.1: Block Diagram showing how the shake table components connect to each other.

The fuse, as described in chapter 3, was replaced, and the amplifier and shake table

were dusted using a can of compressed air. Any method for removing dust would be

acceptable, but the can of compressed air was convenient, fast, and did not damage

the equipment. Using a multimeter, all wires and connectors were traced to try and

identify any problematic connections. No obvious flaws were found, except that the

knob on the amplifier does not seem to work properly, and once set, should not be

turned.

40
The amplifier was then connected to the shaker component. The shaker component

moved when a voltage was run through the amplifier. This proved that the amplifier

and shaker were in working order. However, the shaker only had half of its full

range of motion. In order for the shaker to both push and pull, it requires both a

positive and a negative voltage. The DAC could not supply the negative voltage.

See Fig. 4.1 for a diagram of the pin assignments in the connector cable to the

shaker and how the shaker uses both positive and negative voltage for maximum

displacement. The amplifier might not actually be sending a negative voltage to the

shaker. Often a coil with a center tap is used in what is called a push-pull circuit.

Each of the two outputs goes from zero to some voltage, but are out of phase by 180

degrees. When both voltages are equal, the currents in the two halves of the coil

windings are equal, but in opposite direction, so there is no magnetic field. When

one is larger than the other, some of the current does not cancel out so there is a net

magnetic field and the shaker moves. A push-pull circuit can be economical because

a negative voltage is not required in the power supply.

41
 Figure 4.2: Shaker Component 3-Pin Connector Diagram

Because the amplifier needed both a positive and a negative voltage, a circuit was

required that would convert a positive voltage into a +/- voltage. This circuit was

designed according to Fig. 4.3 with the help of Joseph Pingree . The circuit uses a

TL082 Operational Amp and several resistors soldered to a PC board. The circuit in

figure 4.3 is the dc-to-dc converter that is used to generate -5 volts from the +5 volts

that is provided by the DAC box. Without it, a separate negative power supply

would have been required.. The PC board was placed in a metal box that was

purchased and modified by the author. See Fig. 4.5 for an image of the metal box

with the DAC and circuitry inside. Fig. 4.6 shows the author soldering the circuit

together. A new BNC connector that mounts directly to the metal box was installed,

and the circuit grounds to the box.

42
 Figure 4.3: Circuit designed by Joseph Pingree for DAC.

Figure 4.4: Diagram of the Pin Connections in the TL082 Op Amp that was used to make the

DC to DC converter added to the DAC.

43
 Figure 4.5: Image of the DAC inside a custom metal box.

Figure 4.6: The author soldering the circuit to convert a positive voltage to a +/- voltage.

44
After testing that the shake table was indeed working with this new circuit, and had

both a push and pull component, the DAC was tested using an oscilloscope. This

showed that the DAC was sending out a sine wave as expected.

The next component that needed to be added to the shake table was a way to get

digital input back into the computer, to help calibrate the shake table and to record

deflection and movement in the model in a digital way. A MEMS accelerometer

was purchased from Analog Devices Inc. This accelerometer is an ADXL78 MEMS

accelerometer, and is a very small chip. The chip needed to be connected to a PC

board and then connected into the DAC. See Figure 4.7 for a functional block

diagram of the accelerometer.

Figure 4.7: Functional Block Diagram from Analog, Inc.

45
The accelerometer was soldered to a small piece of insulated board. Then wires

were carefully soldered on to the respective pins. See Figure 4.8 for a pin diagram.

These solder connections are very fragile, and the accelerometer was placed into a

small metal box to protect it from unintentional damage.

Figure 4.8: The Pin Diagram for the Accelerometer, from Analog Devices, Inc.

The accelerometer was tested and the output voltages were very small. In order to

reduce added noise to the output signal, an amplifying circuit was required. See

Figure 4.9 for a diagram of the circuit design. Both op amps are part of another

TL082. The potentiometer on the middle left must be adjusted to set the output to

zero when no acceleration is present.

46
Figure 4.9: Circuit design for the amplifying circuit of the accelerometer.

47
The circuit was first designed on a breadboard, and tested before being hardwired.

See Figure 4.10. Then, the circuit was hooked up to an oscilloscope to verify that it

was indeed working. See Figure 4.11.

Figure 4.10: The amplifying circuit tested on a breadboard.

48
 Figure 4.11: Oscilloscope, and Joseph Pingree testing the circuit of the accelerometer and

amplifier.

The amplifying circuit required these components:

• 1 PC Board

• 1 20K Ten Turn Trim Pot

• Several Pairs of 6 pin Molex Connectors

• 1 TL082 Operational Amp

• Several colors of stranded 26 gauge wires

49
The circuit was soldered together, and placed in another small metal box. The

accelerometer and the amplifying circuit were connected together via wires and

Molex connectors. See Figure 4.12.

Figure 4.12: The amplifier and the accelerometer mounted to the shake table and connected to

the DAC with wires and Molex Connectors.

4.3 Software

The software that came with the DAC was a program provided by National

Instruments called Labview 8.0, student version. This program is a visual

programming language based on the C programming language and is used in both

commercial and academic arenas to build and test components. This program

provides examples that work with the components that National Instruments builds
50
and supplies. One such example worked to run a sine wave on the shake table. This

example was then modified to suit the requirements of the shake table. The

controlling software has two components: Front panel and Block Diagram.

Figure 4.13: Labview Front Page

Front page allows the user to change the constraints of the test. See Figure 4.12.

The block diagram is the part of the program that tells the computer, and the DAC,

what to do. A signal is specified, and sent to the DAC. The signal is scaled so that

it will run on the shake table. The signal goes to the shake table and shown on the

Output Graph, runs the shake table, and then a signal is sent back to the computer

via the accelerometer and then shown on the Input Graph. See Appendix A for a

printout of the block diagram.

51
4.4 Problems

Many problems were encountered while restoring the shake table. While designing

each circuit on a breadboard, the circuits were tested using a voltmeter and a power

supply. Before soldering, the circuits worked. After soldering, often the circuits did

not work. This was due in part to the author’s inexperience with soldering, the fact

that the ground in the circuit might not actually be grounded, and that the

connections were not correct. After discovering the initial problem, isolating the

problem was not hard to do with a voltmeter, and some solder connections needed

repair.

The other problem was in the software. While testing, a 3 second delay was

observed. After examining the block diagram in the program, it was discovered that

what was happening was that only the first 1000 samples of data was being run on

the shake table, and then restarting. After removing that part of the program, the

program would run all of the samples of data.

4.5 Troubleshooting

If the shake table is not running, check first if the problem is in the computer. Make

sure that the earthquake data file is properly scaled, as data won’t run on the shake

table unless it is inside the +/- 2 amplitude requirement.

52
Make sure all components are properly hooked up and turned on.

If the amplifier is the problem, make sure that the fuses are all intact. The fuses are

located on the back of the amplifier.

Always follow directions in the manual located in the locker attached to the shake

table and clean both the amplifier and the shaker.

If none of the above solves the problem, use a voltmeter and try to isolate the

problem. Turn on the Labview program, and send a voltage through the system.

Open up the DAC box. Check all the connections, to make sure that the voltage is

moving through the system. If it is not, that might indicate a problem in the circuitry

or soldering.

If the problem is in the data coming into the computer, the problem might be in the

accelerometer because of the delicate nature of the soldering, or in the amplifying

circuit. First open up the amplifying circuit box, and turn the screw on the blue trim

pot. The blue trim pot is actually used to set the output of the accelerometer to zero

when there is no acceleration. If that does not work, the problem may be in the

solder connections on the accelerometer, as they are very delicate. If that is the case,

carefully solder the connections back together, being careful not to cause a short.

This can be difficult because the lid of the package is metal.

53
Chapter 5: Building a Test Model

5.1 Types of models already in use

Modeling a building for seismic simulation is a difficult proposition. In order to

know how a building will respond to a seismic event in a definite sense, all of the

material properties of the building must be used in the model. Hence in the field of

earthquake engineering, computer models and full scale testing models of the same

material as an actual structure can best describe a building reaction to a seismic

event.

Computer programs may not be completely accurate, however. Unexpected

discrepancies in the field, like the quality of the construction, would affect the

building’s ability to withstand a seismic force. For example, during the Northridge

Earthquake, several buildings collapsed due to the lack of required nails in plywood

shear walls- nails purposefully left out in order to save money on construction costs.

For that reason, large-scale models built on large shake tables try to incorporate

modern building practices and seismic hazards in a controlled environment. Such

models may also test flawed construction. Large-scale models can give clues as to

how a building will react at an assumed intensity.

54
Small-scale models are useful, for teaching basic principles of earthquake

engineering and structural concepts. Models can be built of wood and piano wire.

Piano wire provides flexibility to allow for flexibility in the columns and show

deflection. Models can also be built of individual brick-like elements to suggest

how an un-reinforced masonry building would react to a seismic event.

Models can be built out of clear acrylic plastic, as long as the floor plates are stiff

and the columns flexible. Models can be made out of paper, to show modal forms

but not seismic behavior. Models can even be built of plaster, but the stiffness of

plaster may not properly simulate seismic behavior.

Forms of failure can be documented using a video camera to capture the initial

failure. Accelerometers can measure maximum displacement, as can paper and

pencil.

5.2 Selecting a Model Type

On a large shake table, to study material reactions to a seismic event, choose a

building material that most closely resembles the material properties of the real

building. The shake table can be used to test types of joints at a larger scale. A

small shake table is not large or strong enough, however, to properly test weld

strength. Thus, small bricks can be stacked together and held together with mortar

to test the reactions of an un-reinforced masonry wall. Plaster with a small gauge

55
wire mesh might replicate a poured in place concrete wall well enough in a small

scale.

To test deflection, modal reactions, bracing, or seismic remediation techniques like

base isolation, models made of wood and wire are acceptable, and easily modifiable.

Base isolators can be tested using rubber pads. Joints should be modeled as they

would be constructed in the prototype building to be tested. For example, if the

prototype building is a moment frame, joints should be fixed in thick floor plates to

prevent any rotation in the joint. If the prototype has a pin joint, then the model

joint should be a pin joint. Bracing can be attached using wood in the shape and

direction of the prototype bracing, or for testing a remedial bracing strategy.

5.3 Modeling a Real Building

The important factor of the experiments is to try and match the model’s natural

period of vibration with the period of vibration in the actual building. If the two

match, then all building responses will be similar. If the model’s natural period is

shorter, add more mass to lengthen the period of the model to match the existing

building. If the model’s natural period is longer, reduce the mass to shorten the

period to match that of the existing building.

56
A correctly calibrated model can then be adjusted and modified to test mass

dampers, bracing, or other such earthquake remediation devices, and such tests are

contingent on whether or not the model can approximate the real building.

Specific data with which to test and calibrate the shake table comes from an

instrumented building in Van Nuys, California. The building is a Holiday Inn hotel,

a 7-story reinforced concrete structure that has been damaged in several major

earthquakes. Built in 1966, the building had minor structural damage during the

1971 San Fernando Earthquake, was repaired, and then suffered major structural

damage during the 1994 Northridge Earthquake. See Appendix 3 for building

documentation.

The building structure is a reinforced concrete, column and slab structure with shear

walls on the east and west facades. Most of the damage from the Northridge and

San Fernando earthquakes was shear damage to the columns on the North and South

façade. Appendix 3 includes the properties of the construction for the concrete in

the column and slab. Shortly after the Northridge Earthquake, the building was

repaired and retrofitted with steel bracing. Appendix 3 shows the bracing. Since the

majority of the recorded earthquake data exists prior to the addition of the bracing,

the simulation model was initially also without bracing. Additional tests with the

bracing in place verify the actual effectiveness of the bracing.

57
The data exists for 12 major earthquake events for the building. The Strong Motion

Instrumentation Program of the California Division of Mines and Geology operates

the instrumentation. The records for CDMG, and the other half digitized half of the

earthquakes were digitized at USC. Appendix 3 contains tables of the digitized data

available for the Holiday Inn hotel.

The building has 3 AR-20 accelerographs, which recorded the San Fernando

Earthquake, and a 13 channel CR-1 recording system with a SMA-1 acclerograph,

which recorded all earthquakes from the mid 1970’s to 1994. Figure 5.1 shows the

locations of each recording channel in the building.

Figure 5.1: Locations of the CR-1 recording system in the Van Nuys Building. Image credited

to Prof. Trifunac at USC.

58
The shake table was set up to simulate only the lateral ground motion under the

building in the East-West direction, which was assumed to be the least complicated

direction to study. The data from Channel 16 was used, as an assumption that the

data was the actual ground motion. Thus, the data received back with the

accelerometer would roughly match the actual data taken from the upper stories.

The accelerographs of the upper floors and the roof are for comparison of the model

and shake table to the actual response of the building. The data set used for the

calibration testing was the Northridge Earthquake.

5.4 Symbols
The following symbols and definitions shall be used to determine the constraints of

the model.

A= cross sectional area.

D= diameter.

E= Modulus of Elasticity.

I= Moment of Inertia.

L= Unbraced Length of the column or beam.

M= Bending Moment.

P= Axial Load.

R= Radius of Gyration.

T= Period.

59
π= 3.14.

∆= Deflection.

Materials for the model: 1. Plywood sheets.

2. Piano Wires

3. Fishing Weights.

5.5 Calculations

The plan is of fixed dimension. The piano wire columns of .062” in diameter have a

Moment of Inertia of 7.25*10–7 . The greater the diameter of the piano wire, the

stiffer the modeled column will be. I=πD 4 /64 is the equation for the moment of

inertia for circular steel columns. Euler’s equation, Pcr=π2 EI/L 2 , is used to

calculate the maximum weight that the wire column can carry.

The slenderness ratio is defined as KL/r. KL/r must be less than or equal to 120 for

primary members, and 200 for secondary members. R is the radius of gyration,

which is defined as r=(I/A)1/2 .

When the piano wire was .032” in diameter, the following calculations were made:

I= πD4/64

I= (3.14) * (.032”)4/64

I=51.47x10-9

P=(3.14)2 * (29,000,000) * (51.47x10-9)/5

60
P=.6 lbs

When the piano wire was .062” in diameter, the following calculations were made:

I= πD4/64

I= (3.14) * (.062”)4/64

I=725.33x10-9

P=(3.14)2 * (29,000,000) * (725.33x10-9)/52

P=8.3lbs

This is an acceptable maximum weight for the model.

For the prototype building, the moment of inertia for the columns is I=bd3/12.

Therefore, I=14*203/12. I=9,333.333 in3. The Area of the column is 280 square

inches. The compressive strength of the columns are shown in Fig. 5.2.

1st Floor fc'= 5000 psi

2nd Floor fc'= 4000 psi
3rd Floor fc'= 3000 psi
4th Floor fc'= 3000 psi
5th Floor fc'= 3000 psi
6th Floor fc'= 3000 psi
7th Floor fc'= 3000 psi

Figure 5.2: Compressive Strength of the Columns in the Prototype Building.

The critical loading using Euler’s equation, P=π2 EI/L 2 for the columns in the actual

building are shown in Figure 5.3.

61
 Floor L I=bd^3/12 E Pcrit
1 13.5' 9,330 in^4 4.2 x10^6 2.12x10^9 lbs
2 8.67' 9,330 in^4 3.7 x10^6 5.54x10^9 lbs
3 and up 8.67' 9,330 in^4 3.3 x10^6 4.04x10^9 lbs

Figure 5.3: Critical loading for the columns in the prototype building.

For the prototype building, see figure 3.10 for the slenderness ratio. The radius of

gyration r is defined as r=(I/A)1/2 = (9,333/280)1/2 = 5.76 in.

Floor K L r=(I/A)^1/2 KL/r

1 1.2 164" 5.76 34
2 1.2 104" 5.76 12.48
3 and above 1.2 104" 5.76 12.48

Figure 5.4: Slenderness ratio for columns in the prototype building.

These calculations show that the prototype building has a first floor that is twice as

strong as the columns on the floors above. However, because such a discrepancy

will introduce different frequencies in the model, a simpler model with all wire

columns being the same thickness in diameter and the same length was suggested to

minimize modal forms and different frequencies in the building.

The prototype building has a resonant frequency of .7 seconds, or 1.4 hertz. The

model was designed to match that frequency.

62
5.6 Building the Model

The model was built to a reasonable scale of 1/8”= 1’-0” (1:96) to fit on the shake

table. The shake table surface is 2.5 feet square. The Van Nuys building is

approximately 150’ x 63’. Therefore, the model length is 150/96 = 1.56’. The plan

is of fixed dimension. The columns are of stiff piano wire, of.062” diameter. The

greater the diameter of the piano wire, the stiffer the modeled column will be. Floor

plates are of 1/2” thick plywood of quality grade, to minimize knots and

irregularities in mass. Fishing weights are taped to the floor plates to increase the

mass of the building. Floor plate heights are adjusted up and down, to adjust the

building frequency response on particular floors, and then fixed with epoxy once the

desired frequency is met. (See fig. 5.5).

63
 Figure 5.5: Model dimensions, section.

One critical element of the model building is the selection of glue. Use a strong

epoxy. The joints should be clean- apply nothing to the wires or the wood to ease

construction of the model. Wait 24 hours or longer for full strength of the epoxy.

64
5.7 Fixing the Model to the Shake Table

In the metal plate of the shake table, there are holes drilled 3” on center both in the x

and y-axis. Holes are ¼” in diameter. Drill equally spaced holes in the base of your

model that match the holes in the shake table and use ¼” diameter carriage bolts to

bolt the model down to the table. This prevents extraneous shaking due to improper

fixture to the table. See fig. 5.6 for a drawing of the shake table and the holes.

Figure 5.6: Shake Table Plan for Mounting Models (Units in Inches)

65
5.8 Testing

Once the model is built and has the approximate resonant frequency of the prototype

building, place an accelerometer at the location where the actual recording

instrument is on the prototype building. Run the accelerograph from the roof and

gather the data from the accelerometer. Compare the accelerograph from the roof of

the prototype to the data from the shake table, to determine a way to scale

displacement.

After the model has been calibrated and responds like the prototype, build bracing

out of wood and place in the model scaled to the actual prototype. See fig. 5.7. Run

the same record again, and record the displacement. The difference should be

compared to the previous displacement, to see if the bracing that was installed after

Northridge stiffened the building and reduced the amount of deflection on the roof.

66
Figure 5.7: Diagram of Bracing in Prototype Building. (Trifunac 2001 pg. 15)

67
Chapter 6: Running a shake table test

The following set of instructions is a description of the procedure to be followed for

testing a model.

6.1 Installing the Model

Drill ¼” diameter holes in the model base to attach it to the shake table. Use ¼” in

diameter carriage bolts, 1.5” long as appropriate for the size of the model. Use

washers for spacers if necessary, and use nuts to secure the model.

Place the accelerometer and the amplifying circuit on the model, approximately

where the recording devices are in the prototype building. If the prototype building

had no recording devices, place the accelerometer on the model, in the center of the

top-most floor plate. After each test, move the accelerometer down one level, and

carefully save the data in a file indicative of the location. Affix the accelerometer

with double-stick tape. Carefully install the accelerometer in the direction marked

on the top of the aluminum box. The accelerometer is a single axis MEMS

accelerometer and will only record in the direction marked by the arrows.

A less accurate measuring device can also be used in conjunction with the

accelerometer. Place a large sheet of paper on the white plywood board mounted on

the shake table frame, using masking tape or some other drafting tape. Tape pencils

68
onto the model, in the center of each floor plate, directly in contact with the paper

behind the model. In theory, when the test is run, the pencils will mark the

maximum deflection of the model. In practice, the model has torsion, and the

pencils may not stay in direct contact with the paper. This method does, however,

give a general record of how the model reacted.

A laser instead of pencil would be better to record the model drift.

Assuming that the prototype building has a recorded natural period, the model

should be calibrated to match the prototype period. Run a sine wave in Labview to

match the period of the prototype building. If the model’s natural period is shorter,

add more mass to lengthen the period of the model to match the existing building. If

the model’s natural period is longer, reduce the mass of the model. Labview enters

frequencies in Hertz, so the natural period must be converted into Hertz. One Hertz

is one cycle per second; therefore the frequency of the building is the inverse of the

natural period. See Fig. 6.1 for a screen shot of the Labview program where the

frequency of the prototype building is entered.

69
 Figure 6.1: Screen shot of Simulated File Controls

Mass is added to the model in the form of fishing weights. These weights can be

taped on to the floor plates with double-sided tape, which is easy to apply and

remove. The weights are purchased at a fishing store and are made of lead. Lead is

hazardous in large quantities and careful handling is required. See Fig 6.2 for an

image of the lead fishing weights used. Be certain to place the lead at the centroid

of the floor plan, or at least on the central axis in the direction of motion.

70
 Figure 6.2: Lead fishing weights used for model testing.

Once the model is properly calibrated, meaning that the model responds with the

strongest shaking at the desired frequency, the model is then ready to test using an

actual earthquake file.

6.2 Selecting an earthquake data file

Earthquake data files are recorded in a string algorithm. In computer programming,

a string algorithm is a way to organize data – numbers are separated by spaces,

typically 5 numbers per line. Many programs, including Labview, can read this

71
data. However, this data needs to be reformatted into a single column of data, rather

than five columns of data.

The first step is to choose the earthquake data. Data is available from USGS, but

typically there is a multitude of data for each seismic event. A simpler site to

navigate is COSMOS, the Consortium of Organizations for Strong Motions

Observation Systems. The data is organized by country, state, and recording station.

Any recorded building should be available on the site. U.S. Geological Survey,

California Geological Survey, U.S. Army Corps of Engineers, and U.S. Bureau of

Reclamation, as well as many universities throughout the world run the site.

Download the pertinent data file, and then open the file in Notepad or another word

processor. Some files include several types of data: the best data to use on the

shake table is displacement. Select the displacement data, and copy it into the

Labview program called “Extracting Data”. See Fig. 6.3 for a screenshot of the

program. Follow the instructions to reformat the data into a file format that the

shake table program can use.

72
 Figure 6.3: Data Extracting Program

6.3 Verifying Input and Output results

A simple test to verify input and output results is to place the accelerometer on the

shake table, input the desired data file, and run the shake table without a model. Run

a sine wave at 1Hertz. The output file, the data file the computer sends to the shake

table, should match the input file on the Labview front page. See fig. 6.4 for a

screenshot of the data display.

73
 Figure 6.4: Input and Output graphs

6.4 Running the test

After verifying that the earthquake input and output match, load the earthquake file

in the box shown in Fig. 6.5.

Figure 6.5: Labview File Input

74
Load the earthquake file in the box titled “File Name”. The file location is browse-

able. The minimum and maximum value of the measurement file must be entered in

order to scale the data file properly. The amplitude of the wave must fall between

+/- 2, otherwise the shake table will not run.

Once the file is loaded, the switch located on the front page must be switched from

“simulate wave form” to “input from file” (see Fig. 6.6). Also the timing parameters

must be specified. The timing parameter will matter on the data file chosen, and

should be denoted in the file similarly to this “3000 POINTS OF DISPL DATA

EQUALLY SPACED AT .020 SEC (cm units)”. In the example test, the timing

parameter was 20 millimeters/second.

Figure 6.6: Timing Parameters

When the data is properly uploaded, the model is secure to the shake table, and the

accelerometer is in the proper location, the shake table is ready to run the earthquake

file. Make sure all components are on; the amplifier is plugged in, turned on, and

turned to “current”. The DAC should be connected to the computer being used to

75
control the shake table and the accelerometer should be hooked up to the DAC.

When all the connections have been checked, press the “run” button, which is a

white arrow located in the upper left of the screen.

Figure 6.7: Illustration of "Run" Button

Figure 6.7 shows the arrow button that will start the shake table running.

Labview will open up a prompt to save the data file from the accelerometer. Save

the file in an easily accessible location. The data will collect until the “Stop” button

is pressed. See Fig. 6.8 for an illustration of the stop button.

Figure 6.8: The Stop Button

76
The shake table shaking can be documented by using a video camera, and test pencil

or laser in addition to the data from the accelerometer.

6.5 Possible Problems

It is possible that during or after repeated tests, the model may become damaged.

This may be due to glue or connection failure, or permanent deformation of

columns. One common sense strategy is to only use the model on the shake table

when calibrating the model or running a test. Excessive shaking may cause damage

to a model. Hot glue or reapplying epoxy glue can help to fix joints that have been

damaged as a stopgap measure, but the best strategy is to build another model when

the first model is too damaged to use.

If the shake table becomes damaged, first check to see that all the connections are

undamaged. The fuses on the back of the amplifier might have broken, and if that is

the case, simply replace the fuses. If the damage is not in the connections or in the

amplifier, contact APS Dynamics in Carlsbad, California, manufacturer of the

shaker unit.

77
APS Dynamics

5731 Palmer Way, Suite A

Carlsbad, California 92120

760-438-4848

apsdynamics@att.net

The shake table needs to be calibrated so that the data from the shake table can be

published. However, there may be inaccuracies in the shake table data. One that

was noticed was that while running a simple sine wave, the shake table’s

displacement started out small and got larger before settling on about 3.5 centimeters

from it’s natural resting point at 1 Hertz. The shake table did seem to have the same

amplitude in both the positive and negative directions, so the change in amplitude

might have more to do with inertia than an inaccuracy in the shake table.

The shake table should be able to run any desired earthquake file. If the file will not

run, check to make sure that the maximum and minimum values have been properly

input, because improper scaling can cause problems. If the file still does not run,

turn the amplitude knob on the amplifier. This is not recommended for regular use,

as the amplitude knob seems to be problematic. If neither of the above solutions

work, then check to make sure that the file is the correct data. If running a simulated

78
waveform does not work as well, then the problem may be in the shake table itself,

and outside help may be required.

79
Chapter 7: Results and Analysis

7.1 Introduction of Testing

The following table is a record of all the tests that were run on the shake table and

how the test was recorded.

How the Test

Test # Earthquake/Frequency Results/Comments
was recorded

Test recorded push motion of shake

1 5 Hz Video
table. 9/20/06

Test recorded push/pull motion of shake

2 5 Hz Video
table 10/11/06

1.3 Hz to 5 Hz, adjusted to Testing of variable floor plate height

3 2.8 Hz for resonant Video model and resonant frequency of
frequency of model. model. 10/18/06

Test of fixed floor plates with soft story

on first floor. Discovered that soft story
4 10 Hz to 1.4 Hz Video added extra frequencies into the model,
and thus made it unsuitable for testing.
10/25/06

Resonant Frequency Hand Test to determine resonant frequency of

5 Video
Test new model with no soft story. 11/15/06

Accelerometer Test with Mass on

6 1 Hz Excel/ Notepad
Shake Table 11/29/06

Accelerometer Test with No Mass on

7 1 Hz Excel/ Notepad
Shake Table 11/29/06

Accelerometer Test after reattaching

8 1 Hz Excel/ Notepad
the accelerometer 11/29/06

80
 Test recorded that new model would
9 1.4 Hz Video resonate at the same frequency of the
prototype building. 11/29/06

Northridge at Van Nuys, Test recorded how the model

10 Video
Channel 1 responded to Northridge. 12/2/06

Test recorded maximum deflections on

Northridge at Van Nuys,
11 Paper Roof, 6th, 3rd, and 2nd floors of model.
Channel 1
12/2/06

Test recorded maximum deflections on

Northridge at Van Nuys,
12 Paper Roof, 6th, 3rd, and 2nd floors of model.
Channel 1
12/2/06

Test recorded maximum deflections on

Northridge at Van Nuys,
13 Paper Roof, 6th, 3rd, and 2nd floors of model.
Channel 1
12/2/06

Test recorded the data coming in from

14 1 Hz Excel/ Notepad the accelerometer to determine if it was
working properly. 12/09/06

Input/Output test- verified 3 times on

15 1 Hz Excel/Notepad
1/23/07.

Resonant Frequency Calibration

16 1.4 Hz Excel/Notepad
1/25/07

Test recorded maximum deflections on

Northridge at Van Nuys,
17 Excel/Notepad Roof, 6th, 3rd, and 2nd floors of model.
Channel 16
1/25/07

Test recorded maximum deflections on

Northridge at Van Nuys,
18 Excel/Notepad Roof, 6th, 3rd, and 2nd floors of model.
Channel 16
1/25/07

Test recorded maximum deflections on

Northridge at Van Nuys,
19 Excel/Notepad Roof, 6th, 3rd, and 2nd floors of model.
Channel 16
1/25/07

Test with bracing on model, maximum

Northridge at Van Nuys,
20 Excel/Notepad deflections on Roof, 6th, 3rd, and 2nd
Channel 16
floors of model. 1/25/07

Test with bracing on model, maximum

Northridge at Van Nuys,
21 Excel/Notepad deflections on Roof, 6th, 3rd, and 2nd
Channel 16
floors of model. 1/25/07

81
 Test with bracing on model, maximum
Northridge at Van Nuys,
22 Excel/Notepad deflections on Roof, 6th, 3rd, and 2nd
Channel 16
floors of model. 1/25/07

Table 7.1: Table of Tests run on the Shake Table

Table 7.1 shows a table of all the tests run on the shake table. Results recorded on

Video will be attached in a supplementary appendix to the thesis. Any results

recorded in Excel and/or Notepad will be shown as a graph in this chapter, as seen in

Table 7.4.

7.2 Verification Tests

Table 7.2 shows the results of tests 6 through 8. The concern was that the

accelerometer was not properly recording the data, and thus that the accelerometer

was not working. It was discovered that the accelerometer was set up to record 3000

data points per second, and thus, the resulting graph was not longer than .01 seconds

worth of data due to human error. The variations in the line of data are a result of

noise transmitted through the long wires into the DAC.

82
 Accelerometer Tests 6,7,8

5
Test after refixing the
accelerometer
4
Test with old model on
table
3
Test with no mass on table

0
1
8
15

22
29
36

43
50
57
64

71
78
85

92
99
Data Points

Table 7.2: Accelerometer test 6-8.

Maximum Maximum Maximum Maximum

Displacement of Displacement Displacement of Displacement of
Floor Location Prototype of Model #1 Model #2 Model #3
Roof (channel 3) 25.042 cm 13 cm 10.5 cm 14.5 cm
6th Floor (Channel 4) 22.749 cm 11 cm 9 cm 10 cm
3rd Floor (Channel 5) 14.101cm 9.5 cm 7.5 cm 9 cm
2nd Floor (Channel 8) 14.307 cm 9.5 cm 7.5 cm 9 cm
Ground Floor 12.305cm 7 cm 7 cm 7 cm
Table 7.3: Test 11, paper displacement tests

Table 7.3 shows the displacement measurements of the model while running the data

from Channel 1. This data is suspect because the model was oriented in the wrong

direction for the earthquake data. The data is also suspect because the pencil did not

remain in constant contact with the paper throughout the test due to torsion in the

83
model. A laser would have been better with a high speed camera to capture the

results.

Tables 7.4 shows a test of the noise filters that were introduced to the controlling

program to make the input data more clear, and with less extraneous data points.

Test of Noise Filters

2.5
Amplitude

2
1.5
1 Accelerometer
0.5
0
1
1227
2453
3679
4905
6131
7357
8583
9809
11035
12261
13487
14713
15939
17165
18391
19617
Data Points (1000 points = 1 sec)

Table 7.4: Test of Noise Filters in the Input of the Accelerometer data.

Table 7.5 shows the results of the first Input/Output verification test. A sine wave of

1 Hz was run on the shake table, and these graphs show the data that comes from the

accelerometer mounted on the table. The first test, 15.1, shows that the noise filters

need to be adjusted. The vertical lines every 5000 data points represent a stop point

automatically placed in Labview when it writes the data file, where it places a

marker in the file to denote the timing of the next 5000 samples.

84
The same test was run three times for verification of data input. The data

corroborates the 1 Hz frequency that the shake table was operating at, thus the shake

table seems to be roughly calibrated and responding properly to the wave files.

Tests 15.2 and 15.3 were run with better noise filters, hence the cleaner lines on the

graphs.

Input/Output Verification Test 15

4 Test 3
Test 2
Test 1
3

0
10001

11001

12001

13001

14001

15001

16001

17001

18001

19001
1001

2001

3001

4001

5001

6001

7001

8001

9001
1

Table 7.5: Input/Output verification test at 1 Hz

Table 7.6 shows the results of three model tests of the model reaction to the 1994

Northridge event, as recorded by Channel 16 of the recording devices stationed in

the prototype building. The accelerometer recorded 68 seconds of data – only 10

85
seconds of greatest reaction is compiled in the figure to condense the data. 10

seconds was also the Northridge Earthquake intense shaking limit. Three tests, with

all variables remaining the same, were run to verify the output results and rule out

the possibility of aberrations. The second and third tests start off the record being

extremely similar, then after a few seconds, the reactions diverge. This is probably

due to the model, and not due to differences in how the shake table ran the file.

Tests 2 and 3 remain similar throughout the rest of the record. Test 1, however, does

not correspond to the last 2 tests, for reasons unknown.

Table 7.6 shows the prototype building’s reaction to Northridge, during the same

recording time as the record used to test the model. The maximum displacement of

the prototype building does not correspond to the data taken from the model. The

actual record of the building does not resemble the data taken from the model, thus

no numerical comparisons between the model and the building are possible.

Table 7.6 plotted the output of the accelerometer vs. time. This is not the same as the

displacement. To calculate the displacement from the acceleration the acceleration

needs to be integrated twice. The acceleration data has a DC offset, thus the mean

of the acceleration must be subtracted before doing the integration. This may explain

why the model data does not look like that from the real building. The comparisons

between accelerations on different floors, or between two different models are still

valid.

86
7.3 Model Testing and Results

Table 7.6: Tests 18, 19, and 20, Roof

87
This graph shows model response to the Northridge event on the roof of the model. The graph shows

that the largest displacement occurred on the roof, although the reason for the variations between

test 1, and tests 2 and 3, are unknown.

Prototype Building Roof Data

25
20
15
10
Drift in CM

5
0 Channel 9 Data
-5
-10
-15
-20
-25
1001
1101
1201
1301
1401
1501
1601
1701
1801
1901
2001
2101
2201
2301
2401
2501
2601
2701
2801
2901
3001
101
201
301
401
501
601
701
801
901
1

Data Points over time

Table 7.7: Prototype Building Roof Reaction to Northridge Earthquake 1994

Generalizations correlating the model and the building can be made: the building

had the greatest deflection on the roof. The model had the greatest deflection on the

roof. The model reacted similarly to the prototype building in all of the tests run:

Therefore, anything done to the model in terms of earthquake remediation, like the

88
bracing that was added to the prototype building after the Northridge Earthquake,

should give general indications of how the prototype building would respond.

89
 Table 7.8: Tests 18, 19, and 20 5th Floor

This graph shows model response to the Northridge event on the5th floor of the model. The graph
shows more correlation between the tests in the initial seconds, but then the displacement values
diverge. The displacement on the 5th floor is less than that on the roof.

90
 Prototype Building 5th Floor

20
15
Drift in CM 10
5
0
-5 Channel 10 data
-10
-15
-20
-25

1001
1101
1201
1301
1401
1501
1601
1701
1801
1901
2001
2101
2201
2301
2401
2501
2601
2701
2801
2901
3001
101
201
301
401
501
601
701
801
901
1

Data Points over time

Table 7.9: Prototype Building 5th Floor Reaction to Northridge Earthquake 1994.

Table 7.8 shows a greater correspondence in the early part of the data records. All

three tests match for the first few seconds, and then diverge. The wave amplitudes

differ, but the waves seem to be peaking at similar times. The timing might be an

issue in the record, but the recording of the data is not human controlled, therefore

probably not an issue. The computer should begin recording the input data as soon

as the data is sent out from the computer.

Table 7.9 shows the prototype building’s reaction to the earthquake. Similarly, the

graphs of the model and the prototype do not correspond. The model and the

building cannot correspond unless the model perfectly mimics the actual building,

and since the actual building’s materiality is very different from the model, and the

building has varying stiffness from the ground floor to the roof, as well as eccentric

loads and live loads.

91
 Table 7.10: Model Test 3rd Floor Reactions

This graph shows model response to the Northridge event on the3rd floor of the model. The graph
shows more correlation between the tests in the initial seconds and then remains similar throughout
the rest of the tests. This indicates that the upper floor variations were most likely due to unintended
model variations.
92
 Table 7.11: Model Test 2nd Floor

This graph shows model response to the Northridge event on the 2nd floor of the model. The graph
shows more correlation between the tests in the initial seconds and then remains similar throughout

93
the rest of the tests. This indicates that the upper floor variations were most likely due to unintended
model variations.

Table 7.10 shows the model reactions to the Northridge earthquake file on the third

floor. All three of the files correspond more closely to each other. Thus the model

is probably reacting more regularly at the bottom of the model than at the top of the

building. There might be a more conflicting frequency at the top of the building, or

interference as one waveform hits the top of the model and starts to reverberate back

down.

Table 7.11 shows the model reactions to the Northridge earthquake file on the

second floor. All three tests closely resemble one another in terms of displacement.

The model is reacting predictably, and variations in how the model is built and how

the model responds to the earthquake file are causing the discrepancies at the top of

the model.

After these tests were run, four more tests were run that attempted to model the

bracing of the prototype building during the earthquake retrofit after the damage that

Northridge and the aftershocks caused to the building. The bracing was previously

discussed in Chapter 5. The model bracing was made out of basswood strips, ¼”

wide and approximated the shape of the prototype bracing. The bracing was

fastened with hot glue, and greatly increased the stiffness of the building. While the

bracing reduced the movement of the model in response to the Northridge file, there

is some concern that the bracing would lead to increased damage at the bracing
94
joints due to the stresses that the stiffness would put on the joints. Table 7.12 shows

the model reaction after bracing was introduced, at the roof.

Table 7.12: Model Test of Bracing- roof reaction

This graph shows model response to the Northridge event on the roof of the model, after bracing was
introduced. The graph shows that building displacement was reduced 120%, compared to the
building without bracing.
95
 Table 7.13: Bracing on the Model- 5th Floor

This graph shows model response to the Northridge event on the 5th floor of the model, after bracing
was introduced. The graph shows that building displacement was reduced 120%, compared to the
building without bracing.

96
 Table 7.14: Model Bracing Test- 3rd Floor

This graph shows model response to the Northridge event on the3rd floor of the model, after bracing
was introduced. The graph shows that building displacement was reduced 120%, compared to the
building without bracing.
97
 Table 7.15: Bracing Test 2nd Floor

This graph shows model response to the Northridge event on the2nd floor of the model, after bracing
was introduced. The graph shows that building displacement was reduced 120%, compared to the
building without bracing.

98
The bracing consistently reduced the deflection and dampened the impact of the

earthquake wave. Damage to the building might be reduced due to the lessening of

sharp movements in the building. However, due to the increased stiffness, damage

might be concentrated at the joints of the bracing. There was no damage to the

model after the initial tests.

After running several tests with the bracing on the model, however, there was

damage to the joints where the bracing connected to the floor plates, showing that

either the glue used to fasten the bracing to the model was inadequate, or that the

forces concentrated at the joints was so great to cause damage. This should be

considered while examining the actual building’s structure, and has implications for

the actual connections.

99
Chapter 8: Conclusions

From the test data, it can be seen that the shake table is restored to working order.

The shake table can run both sine waves and earthquake waveforms. When running

a sine wave at 1 Hz, the accelerometer on the shake table sends a 1 Hz waveform

back into the computer.

The test data also shows that the shake table seems to run consistently. This is more

evident in the data files taken from the lower 2nd and 3rd floors, where there is less

variation in the waveforms, and the three tests seem to more closely match. The

shake table also can create repeatable tests.

Each level of the model reacted predictably in each test. There was greatest

deflection at the upper levels and smaller deflection at the lowest level. The model

reacted as expected when bracing was installed on the model. The model’s reactions

are similar to the prototype building’s reactions.

The bracing reduced the acceleration and displacement on all of the floors of the

model and increased the stiffness of the model by 130%. However, increased

stiffness will increase the stress at the joints of the bracing, and might lead to more

future damage in the prototype building. The model’s shaking was greatly reduced

compared to the tests without the bracing. The bracing increased stiffness by 120%.

100
8.1 Correlations to prototype building data

The model can be used to make general conclusions about the behavior of a building

in response to an assumed earthquake. The building had the greatest deflection on

the roof, as did the model. The least amount of deflection in the prototype building

was on the second floor, as in the model. After bracing was added, the building was

stiffer. After adding bracing to the model, the model was much stiffer. While

numerical data taken from the model applied to the building would be questionable,

generalizations about building behavior based on shape and stiffness are valid.

The model cannot be used to make direct numerical conclusions because of

variables beyond the modeling capabilities available in a small-scale model.

Building reaction to a seismic event is dependent on many different variables: soil

type, connection to the foundation, material properties, cladding, partitions, etcetera,

all of which are difficult to model or scale. Modal analysis can occur on a small-

scale model, and generalizations on whether or not an earthquake remediation

attempt will have the desired dampening effect will be possible on a small-scale

model.

101
8.2 Concluding Remarks

The shake table is restored to working order and can be used by the School of

Architecture to teach students basic concepts and building response to seismic

events. The shake table runs consistently and can run an earthquake file.

A model on the shake table can show modal form and building displacement. The

model will react similarly to the prototype building if the model’s resonant

frequency matches the prototype building’s frequency. Earthquake remediation

techniques can be applied to the model to draw general conclusions about the

prototype building. The Van Nuys building’s earthquake retrofit increases the

stiffness of the building, which may lead to more building damage in a future

seismic event.

The University of Southern California’s School of Architecture G G Schierle shake

table can be used in a meaningful way to study a building’s seismic response using

existing seismic data. The G G Schierle Shake Table is an important teaching tool

that is useful to show aspiring architects and engineers how a structure will respond

in a seismic event.

102
Chapter 9: Future Work

9.1 3-Degree of Motion Shake Table

The G. G Schierle Shake Table is a 1-degree of motion shake table, meaning that it

only moves back and forth in one direction. Earthquakes do not move in just one

direction - earthquakes have movement on the x, y, and z-axis. Therefore, in order

to properly study the effects of an earthquake on a building, it is necessary to

incorporate the effect of these movements on the structure. Studying the effects of

only one component of the earthquake is effective for conceptual study but not

specific enough to accurately predict the response a building will have to an

assumed earthquake.

9.2 Different Modeling Techniques

There are many different types of models that can be built to test structures on the

shake table. This thesis only tested one type, the wood and piano wire models

described in Chapter 5. An in-depth study of modeling techniques would be

beneficial to determine whether or not wood and piano wire models are accurate

enough. Acrylic plastic might be an appropriate modeling material, as might metal.

Wood and piano wires do not accurately model the materiality of concrete, and

cannot study masonry buildings.

103
Basswood might be used to model a typical Type V construction project. The wood

pieces can be bought to scale and can be framed out. With wood framing, it might

be possible to get a more accurate small-scale model in terms of material properties

and construction techniques.

Plaster might be used to test poured in place concrete buildings. Actual concrete

might also be used to simulate the material properties of a concrete building. A

study to determine how best to study the materiality of concrete would be quite

interesting and useful to engineers and architects.

Sugar cubes and peanut butter have been used to study masonry buildings in

elementary schools throughout California to build model missions. A similar

modular building material might be able to simulate the dangers of unreinforced

masonry buildings in a seismic event. Small bricks could be made into walls using a

mortar of some kind. It would be important to study the relationship between the

ductility of the “mortar” and the “block” when making conclusions about the mode

of failure.

9.3 Soils Testing

The soil upon which the building is built is a large part of how the earthquake will

affect the building. Softer soils can lead to greater shaking, whereas harder, rockier

soils might reduce the amount of shaking. Earthquake waves travel fastest through

104
dense materials, and in a basin area like Los Angeles, the waves can reverberate in

the soft soils of the basin, reflecting off the rock in the mountain, amplifying the

earthquake waves and causing more damage.

Thus, a way to simulate the soils type that a building is built upon and test the effect

that that has on the building would be very useful. Testing foundation systems

would also be a part of such research. Such testing could occur in a large box

mounted onto the shake table, in which would be the soil type to be tested, some

moisture to hold the ground together, and the model to be tested, anchored into the

soil in the way it was in the prototype. Tests could also study the effect liquefaction

has on a building, and the types of damage that can occur.

9.4 Calibration
The shake table needs to be properly calibrated before it can be used to record

numbers for scientific and publishing use. Using the accelerometer, calibrating the

shake table should be a simple prospect. Contact the USC Civil Engineering

Department for more guidance, and perhaps comparison with the already properly

calibrated shake tables in the Civil Engineering Department.

9.5 Better Measurement Devices

The accelerometer is a single axis accelerometer. There are multiple axis

accelerometers, and there are also 6 analog inputs into the DAC. There could be up
105
to 6 accelerometers hooked up to the DAC, taking measurements from 6 different

levels on a model at one time.

Professor Dimitri Vergun recommended a miniature seismograph created using a

rotating drum and a small rotating motor that could be designed to record on paper

the movements of the model. This could be designed and built and implemented for

an alternate way to record model deflection.

9.6 More Earthquake Files

For this thesis, only the Northridge earthquake of 1994 was used for earthquake

data. Earthquake data from all over the world, and from other instrumented

buildings across the United States and from around the world could be tested on the

shake table. The data needs to be found and converted into a form that could be

used by the shake table.

The prototype building studied in this thesis could be used to study how different

earthquakes would affect the same building. Different prototype buildings could

also be studied.

9.7 Earthquake Remediation Strategies

106
Using the same prototype building, different remediation techniques could be

studied to determine whether or not they positively affect the building’s response to

a seismic event. Different bracing types could be tested, as well as the effectiveness

of base isolation. See fig. 9.1 for some examples of eccentric braced frames that

could be tested.

Figure 9.4: Types of eccentric braced frames

(http://www.cen.bris.ac.uk/projects/eqteach97/images/frame8.gif accessed 1/24/07)

9.8 Strategies for use as a teaching tool

The shake table is a valuable teaching tool for use in both the building science

program, both graduate and undergraduate, and in the school of architecture.

Strategies and curriculum should be developed to use the shake table in the

education process. Possible strategies include building different types of models and

running them with different major worldwide earthquakes, such as wood models and

brick models. Another strategy would be to incorporate Architecture students

design models on the shake table to see how the design would withstand earthquakes

typical to the site.

107
There are many further avenues of explorations for the enterprising student using the

shake table as a valuable research or educational tool. The author has only

suggested a few possibilities that were discussed during the course of the Thesis.

The G. G. Schierle Shake Table is a flexible and useful tool that can provide

meaningful results in many fields of research.

108
 Bibliography

Bolt, B, 2004, Earthquakes, W. H. Freeman and Company, New York.

California University Los Angeles, 2006, viewed 1 November 2006,

<http://www.calstatela.edu/dept/geology/earthquakes/CSUNParking(2).jpg>

Chong, L, 1993, Investigations of Seismic Isolators as a Mass Damper for Mixed-

Use Buildings, University of Southern California, Los Angeles.

Congress of the United States, Office of Technology Assessment, 1995, Reducing

Earthquake Losses, U.S. Government Printing Office, Washington DC.

Crouch, C, Fagan, A, Mazur, E, 2002, Peer Instruction: Results from a Range of

Classrooms, The Physics Teacher, Volume 40, April 2002.

Daryl Science 2006, viewed 18 October 2006,

<http://www.darylscience.com/graphics/seiswave.gif>

Earthquake Country Alliance 2006, viewed 18 October 2006,

<http://www.earthquakecountry.info/roots/inline/11839sm.jpg>

G. G. Schierle, 2003, Northridge Earthquake Field Investigations: Statistical

Analysis of Woodframe Damage, The CUREE- Caltech Woodframe Project,
Consortium of Universities for Research in Earthquake Engineering,
Richmond CA.

Mazur, E, 1995, Understanding or Memorization: Are we teaching the right thing?,

Department of Physics, Harvard University, Cambridge MA.

Missouri Department of Natural Resources 2006, viewed 20 October 2006,

<http://www.dnr.mo.gov/geology/images/Richter_Mercalli.jpg>

Moore, F, 1999, Understanding Structures, The McGraw-Hill Companies, Inc.,

Boston.

National Geographic, Anatomy of an Earthquake, viewed Sunday, October 15, 2006.

National Instruments, 2006, viewed 26 August 2006,

<http://sine.ni.com/nips/cds/view/p/lang/en/nid/14604>

National Oceanic and Atmospheric Administration, viewed Oct. 22, 2006,

<http://www.ngdc.noaa.gov/seg/hazard/slideset/22/22_slides.shtml>
109
Trifunac, M, Hao, T, 2001, 7-storey reinforced concrete building in Van Nuys,
California: photographs of the damage from the 1994 Northridge
earthquake, USC Report CE 01-05, University of Southern California, Los
Angeles, CA.

Todorovska, M, Trifunac, M, 2006, Impulse response analysis of the Van Nuys 7-

story hotel during 11 earthquakes and earthquake damage detection,
Structural Control and Health Monitoring, University of Southern
California, Los Angeles, CA.

Todorovska, M, Trifunac, M, 2006, Impulse response analysis of the Van Nuys 7-

story hotel during 11 earthquakes (1971-1994): one-dimensional wave
propagation and inferences on global and local reduction of stiffness due to
earthquake damage, USC Report CE 06-01, University of Southern
California, Los Angeles, CA.

Todorovska, M, Trifunac, M Ivanovic, S, 1999, Instrumented 7-storey reinforced

concrete building in Van Nuys, California: description of damage from the
1994 Northridge earthquake and strong motion data, USC Report CE 99-02,
University of Southern California, Los Angeles, CA.

United States Geological Survey 2006, viewed 20 October 2006

<http://earthquake.usgs.gov/regional/states/events/1857_01_09.php>.

United States Geological Survey 2006, viewed 20 October 2006,

<http://earthquake.usgs.gov/images/faq/3faults.gif>

United States Geological Survey 2006, viewed 28 October 2006

http://earthquake.usgs.gov/regional/states/events/1857_01_09.php

University of California, Berkeley, 2006, viewed 1 November 2006,

<http://eerc.berkeley.edu/lab/earthquake-simulator-lab.html>

University of California San Diego, 2006, viewed 1 November 2006,

<http://www.jacobsschool.ucsd.edu/news/news_releases/release.sfe?id=508>

Washington University, 2006, Viewed on 16 October 2006,

<http://www.cev.washington.edu/lc/CEVIMAGES/global-techtonic-
plates.jpg>

Wikipedia 2006, viewed 20 October 2006,

<http://upload.wikimedia.org/wikipedia/commons/thumb/a/a0/BNC_connect
or.jpg/639px-BNC_connector.jpg>
110
Appendix A : Instructions for Using Labview

Instructions for using Labview.

1. Select the Physical Channel to correspond to where your signal is output on the

DAQ device.

2. Enter the Minimum and Maximum Voltage Ranges

3. Set the loop rate.

4. Run theVI.

5. Stop the VI when desired.

Block Diagram Steps:

1. Create an Analog Output Voltage channel.

2. The Array Data is a sinewave with 1000 points, generates 5 cycles, and has an

amplitude of 2.

3. Call the Start VI

4. Write one data point from the array (modulo indexed to loop count) until the user

hits the stop button or an error

occurs. The loop rate is settable to 1 millisecond.

5. Call the Clear Task VI to clear the Task.

6. Use the popup dialog box to display an error if any.

111
Figure A.1: Labview Front Page

112
Figure A.2: Labview Block Diagram

113
Appendix B: Index of Videos Submitted

List of Videos submitted:

Shaker Push-Pull

Shaking (Old Model)

Shaking (Torsion)

Shaking (Soft Story)

Shake Table Verification

Resonant Frequency Shaking 1

Resonant Frequency Shaking 2

Model Hand Test

114