Sunteți pe pagina 1din 240


PB96-194162 Information .

IMPROVED .s our busln.s•.



30 JUN 96

ca 'nformat"Ion Service
PB 96 - 194162


Improved Seismic Design

Criteria for California Bridges:
Provisional Recommendations

Applied Technology Council

Funded by


U.S. Department of CommerceJ
Na.tional Technical Information Service
Springfield, Virginia 22161
Applied Technology Council

The Applied Technology Council (ATC) is a nonprofit, tax-exempt corporation established in 1971 through the
efforts of the Structural Engineers Association of California. ATC is guided by a Board of Directors consisting of
representatives appointed by the American Society of Civil Engineers, the Structural Engineers Association of
California, the Western States Council of Structural Engineers Associations, and four at-large representatives
concerned with the practice of structural engineering. Each director serves a three-year term.

The purpose of ATC is to assist the design practitioner in structural engineering (and related design specialty
fields such as soils, wind, and earthquake) in the task of keeping abreast of and effectively using technological
developments. ATC also identifies and encourages needed research and develops consensus opinions on structural
engineering issues in a nonproprietary format. ATC thereby fulfills a unique role in funded information transfer.

Project management and administration are carried out by a full-time Executive Director and support staff.
Project work is conducted by a wide range of highly qualified consulting professionals, thus incorporating the
experience of many individuals from academia, research, and professional practice who would not be available
from any single organization. Funding for ATC projects is obtained from government agencies and from the
private sector in the form of tax-deductible contributions.

1996-1997 Board of Directors

John C. Theiss, President Douglas A. Foutch

C. Mark Saunders, Vice President James R. Libby
Bijan Mohraz, Secretary/Treasurer Kenneth A. Luttrell
Edwin T. Huston, Past President Andrew T. Merovich
Arthur N. 1. Chiu Maryann T. Phipps
John M. Coil Jonathan G. Shipp
Edwin T. Dean Charles H. Thornton
Robert G. Dean


While the information presented in this report is believed to be correct, ATC and the sponsoring agency assume
no responsibility for its accuracy or for the opinions expressed herein. The material presented in this publication
should not be used or relied upon for any specific application without competent examination and verification of
its accuracy, suitability, and applicability by qualified professionals. Users of information from this publication
assume all liability arising from such use.

California Department of Transportation Disclaimer

The contents of this report reflect the views of the authors who are responsible for the facts and accuracy of the
data presented herein. The contents do not necessarily reflect the official views or policies of the State of
California or the Federal Highway Administration. This report does not constitute a standard, specification, or

Cover Illustration:
New Pescadero Creek Bridge
Photo by Bob Colin, California Department of Transportation
Report Nos: ATC-32
Title: Improved Seismic Design Criteria for California Bridges: Provisional
Date: 30 Jun 96
Performing Organization: Applied Technology Council, Redwood City. CA.
S~onSOring Organization: *California State Dept. of Transportation, Sacramento. Div.
o Structures.
Contract Nos: CALTRANS-59N203
Type of Report and Period Covered: Final rept. 1 May 91-31 Oct 95.
NTIS Field/Group Codes: 50A (Highway Engineering), 500 (Soil &Rock Mechanics), 50C
(Constructlon Equlpment, Materials, &Supplies)
Price: PC A11/MF A03
Availability: Available from the National Technical Information Service, Springfield,
VA. 22161
Number of Pages: 225p
Ke words: *Highway bridges. *Earthquake engineering. *Seismic design. *Structural
re laDl lity. Earthquakes, Seismic effects. Structural response, Damage assessment.
Earthquake damage, Soil-structure interactions. Reinforced concretes. Steel
structures, Foundations(Structures). Load bearing capacity. Load distribution.
Stiffness, Performance evaluation.
Abstract: The ATC-32 report recommends revisions to the California Department of
Iransportation (Caltrans) seismic design standards, performance criteria.
specifications and practices. It is based on recent research in the field of bridge
seismic design and the performance of Caltrans-designed bridges in the 1989 Loma
Prieta and other recent California earthquakes. Specifically, the report provides
recommended revisions to Caltrans current Bridge Deisgn Specifications (BDS)
pertaining to seismic loading, structural response analysis, and component design.
Special attention is given to design issues related to reinforced concrete components,
steel components, foundations, and conventional bearings.

Improved Seismic Design

Criteria for California Bridges:
Provisional Recommendations
555 Twin Dolphin Drive, Suite 550
Redwood City, California 94065

Funded by
P.O. Box 942874
Sacrall1ento,CA 94274-0001
Mohsen Sultan, Contract Manager

Richard V. Nutt


Earth Mechanics, Inc. Ian Buckle, Chairll1an

(Foundation Design) Robert Cassano
Kercheval Engineers Allen Ely
(Bridge Design) Nicholas ForelI, ATC Board Representative
Kleinfelder/Geospectra Jall1es H. Gates
(Seisll1ic Loading) LM.Idriss
Modjeski & Masters, Inc. Roy A. Ill1bsen
(Steel & Bearing Design) Jall1es O. Jirsa
Moehle/Priestley Teall1 Jall1es R. Libby
(Concrete Design & Analysis) Joseph P. Nicoletti
Quincy Engineering Joseph Penzien
(Bridge Design) Maurice S. Power
RDD Consultants Jall1es Roberts
(Publications & Travel)

Technical Report Documentation Page
1. Report No 2. Recipient's Catalog No.
ATC-32 '111Il\ I' 111111'III I11111l\' II'
PB96 -194162
4. Title and Subtitle 5. Report Date

Improved Seismic Design Criteria for Califomia Bridges June 30, 1996
6. Performing Organization Report No.

7. Author(s) 8. Performing Organization Report No.

Applied Technology Council

9. Performing Organization Name and Address 10. Work Unit No. (TRAIS)
Applied Technology Council
555 Twin Dolphin Drive, Suite 550
Redwood City, Califomia 94065 11. Contract or Grant No.

12. Sponsoring Agency Name and Address 13. Type of Report and Period Covered

Califomia Department of Transportation FINAL

Division of Structures 14. Sponsoring Agency Code
Sacramento, Califomia 95807

15. Supplementary Notes

16. Abstract

The ATC-32 report recommends revisions to the Califomia

Department of Transportation (Caltrans) seismic design
standards, performance criteria, specifications and
practices. It is based on recent research in the field of
bridge seismic design and the performance of Caltrans-
designed bridges in the 1989 Lorna Prieta and other recent
Califomia earthquakes. Specifically, the report provides
recommended revisions to Caltrans current Bridge Design
Specifications (BDS) pertaining to seismic loading,
structural response analysis, and component design.
Special attention is given to design issues related to
reinforced concrete components, steel components,
foundations, and conventional bearings.

17. Key Words 18. Distribution Statement

bridges, seismic design standards, performance

criteria, specifications, Bridge Design
Specifications, BDS

19. Security Classif. (of this report 20. Security Classif. (of this page) 21. No. of Pages 22. Price
Unclassified Unclassified 214

FORM DOT F 1700.7 (8-72)


In May 1991, the California Department of Transporta- crete design. They were assisted by Gregory Fenves ofthe
tion (Caltrans) awarded Applied Technology Council University of California, Berkeley. John Kulicki and his
(ATe) a contract to conduct a critical review of the Cal- staff at Modjeski and Masters developed new design cri-
trans Bridge Design Specifications (BDS) related to seis- teria for steel structures and conventional bridge bear-
mic design and to recommend changes where needed. ings.
This contract resulted in the development of the revised Trial designs using the draft BDS were performed by
BDS presented in this ATC-32 report, which provides a two bridge design consultants. John Quincy directed the
number of recommended improvements to bridge efforts of Quincy Engineering and Kosal Krishnan
design practice. A companion document, ATC-32-1, directed those of Kercheval Engineers. Nonlinear
includes additional detailed discussion of these recom- dynamic analysis studies to evaluate near-fault effects
mendations. The recommendations apply to the seismic were performed by Computech Engineering Services
design ofbridges throughout California. under the direction of Ron Mayes. An independent
Because of the broad range of expertise required to external review of the recommendations for structural
develop comprehensive BDS, a 13-member advisory steel was conducted by Ahmad Itani of the University of
Project Engineering Panel (PEP) was assembled to Nevada at Reno.
review recommended changes as they were developed Technical editing and formatting of this report were
and to provide guidance where needed. This panel was performed by Nancy and Rodney Sauer ofRDD Con-
composed ofIan Buckle (Chair), Robert Cassano, Allen sultants and the ATC staff. Their efforts are gratefully
Ely, Nicholas Forell, James Gates, 1. M. Idriss, Roy Imb- acknowledged.
sen, James Jirsa, James Libby, Joseph Nicoletti, Joseph The efforts of several Caltrans personnel are also
Penzien, Maurice Power, and James Roberts. The affilia- gratefully acknowledged. Mohsen Sultan was the Con-
tions of these individuals are provided in the Project Par- tract Manager and coordinated the technical participa-
ticipants list. tion of other Caltrans engineers. Dan Kirkland and Tim
The detailed technical work required for the devel- Leahy served as Contract Administrators. They and their
opment of recommendations was performed primarily staffprovided ATC with invaluable assistance in comply-
by four specialty subcontractors. J.P. Singh and his staff ing with Caltrans requirements. Finally, ATC wishes to
at Kleinfelder/Geospectra were responsible for develop- thank the many Caltrans engineers who have shown an
ing new ARS spectra and other recommendations interest in this project by commenting on draft recom-
related to seismic loading. Po Lam and his staff at Earth mendations and attending PEP and other meetings.
Mechanics, working with Geoff Martin of the University
of Southern California, were responsible for developing Christopher Rojahn,
the foundation design guidelines. Nigel Priestley of the ATC Executive Director
University of California, San Diego and Jack Moehle of
the University of California, Berkeley developed the rec-
ommendations on response analysis and reinforced con-

ATC-32 Preface v

Technical Report Documentation Page iii

Preface v

List of Figures ix

List of Tables xi

Figure Credits xiii

Introduction 1

Summary of Recommendations 5

Revised Bridge Design Specifications 13

Section 3: Loads " 15

Section 4: Foundations 83

Section 8: Reinforced Concrete " 129

Section 10: Structural Steel. 163

References 191

Appendix A: Guide to Bridge Design Specification Modifications 195

Project Participants 203

ATC Projects and Report Information 205

ATC-32 Contents vii

List of Figures

Figure 1 Response modification factor Z 8

Figure RC3-1 Illustrations of full ductility structures and limited ductility structures. . 20
Figure RC3-2 North-South Ground motion recorded at Sylmar, January 17,1994 24
Figure R3-1 Proposed ARS curves for rock (M = 6.50 ± 0.25) 30
Figure R3-2 Proposed ARS curves for rock (M = 7.25 ± 0.25) 31
Figure R3-3 Proposed ARS curves for rock (M = 8.0 ± 0.25) 32
Figure R3-4 Proposed ARS curves for soil type C (M = 6.50 ± 0.25) 33
Figure R3-5 Proposed ARS curves for soil type C (M = 7.25 ± 0.25) 34
Figure R3-6 Proposed ARS curves for soil type C (M = 8.0 ± 0.25) 35
Figure R3-7 Proposed ARS curves for soil type D (M = 6.50 ± 0.25) 36
Figure R3-8 Proposed ARS curves for soil type D (M = 7.25 ± 0.25) 37
Figure R3-9 ProposedARS curves for soil type D (M = 8.0 ± 0.25) 38
Figure R3-1 0 Proposed ARS curves for soil type E (M = 6.5 ± 0.25) 39
Figure R3-11 Proposed ARS curves for soil type E (M = 7.25 ± 0.25) 40
Figure R3-12 Proposed ARS curves for soil type E (M = 8.0 ± 0.25) 41
Figure RC3-3 Effective stiffness of reinforced concrete structure 50
Figure RC3-4 Relationship between cracked-section (Ief[) and gross-section (r g) stiffness values of
reinforced concrete columns 51
Figure RC3-5 Superstructure torsion and column moments for frames under transverse
displacement 57
Figure RC3-6 Model for skewed expansion joint 60
Figure RC3-7 Model for seat-type abutment 61
Figure RC3-8 Load-displacement relationship for circular cross-section, cantilever-reinforced,
concrete column representative of modern Caltrans bridge designs. Column is
subjected to uniaxial lateral load and constant axial load 62
Figure RC3-9 Load-displacement relationship for circular cross-section, cantilever-reinforced,
concrete column representative of modern Caltrans bridge designs. Column is
subjected to biaxial lateral load and constant axial load. . 63
Figure RC3-10 Load displacement relationship for relatively low-aspect-ratio pier wall loaded in its
plane 64
Figure RC3-11 Three-spring model for reinforced concrete pier wall. 65
Figure RC3-12 Measured and idealized load-displacement relationship for restrainer with gap 66
Figure RC3-13 Mean relationships between strength-reduction coefficient (RJl) and displacement
ductility demand (11) .• .••••••••••••••..••.•••........••••..•••.••••••••••••.•.••• 72

ATC-32 List of Figures ix

Figure RC3-14 Shock spectra for a triangular pulse acting on an elastic-perfectly-plastic, single-
degree-of-freedom oscillator 74
Figure R3-13 Force-reduction coefficient, Z 76
Figure RC3-15 Static aspects of gravity load acting through lateral displacement for a cantilever 77
Figure RC4-1 Limits in the gradation curves separating liquefiable and nonliquefiable soils 92
Figure RC4-2 Definition of free face factors, Land H, and ground slope, S, for free-face ground-
spread displacement 94
Figure RC4-3 Definition of ground slope, S, for long, uniform slope ground spread displacement 94
Figure RC4-4 Recommendations for coefficient of variation in subgrade modulus with depth for
sand 105
Figure RC4-5 Recommendations for coefficient ofvariation in subgrade modulus with depth for
clay 106
Figure RC4-6 Lateral stiffness of free-headed piles 107
Figure RC4-7 Coefficient for lateral pile head stiffness (fixed head pile lateral stiffness) 108
Figure RC4-8 Coefficient for pile head rotation 109
Figure RC4-9 Coefficient for cross-coupling stiffness term 110
Figure RC4-1O Comparison of fixed head pile head stiffness at various embedments (0, 5, and 10
feet) III
Figure RC4-11 Comparison of the rotational stiffness coefficient at various embedments (0,5, and
10 feet) 112
Figure RC4-12 Comparison of the cross-coupling stiffness coefficient at various embedments (0, 5,
and 10 feet) 113
Figure RC8-1 Ratio of plastic moment at maximum curvature to design flexural strength 136
Figure RC8-2 Design aid for determining the compression strength of confined concrete 146
Figure RC8-3 Effective joint width for shear stress calculations 156
Figure RC8-4 External vertical joint reinforcement for joint force transfer 158
Figure RC8-5 Locations for vertical joint reinforcement 159
Figure R8-1 Additional cap beam bottom reinforcement for joint force transfer 160
Figure RC8-6 Effective superstructure width resisting longitudinal seismic moments 162
Figure RClO-1 Knee geometry 184
Figure RClO-2 Strength deterioration of knee joint 185
Figure RClO-3 Typical response to cyclic loading 186
Figure RClO-4 Failure modes of box-shaped columns 186

x List of Figures ATC-32

List of TabLes

Table 1 Seismic performance criteria 5

Table 2 Site Characteristics for Standard Design Spectra 6

Table 3 Minimum Required Analysis 6

Table R3-1 Seismic performance criteria 18

Table R3-2 Minimum Required Analysis 25

TableR3-3 Soil Profile Types 42

Table RC3-1 Soil Profile Type Classification 43

Table RC3-2 Values of Site-Amplification Factor Fa as a Function of Soil Profile Types and
Shaking Intensity 45

Table RC3-3 Values of Site-Amplification Factor Fvas a Function of Soil Profile Types and
Shaking Intensity 45

Table RC3-4 Figure Numbers of Appropriate Design Spectra 46

Table RC3-5 Soil Coefficient Pi 48

Table R3-4 Values of T* (in seconds) 71

Table RC4-1 Relationship of Geologic and Water Table Criteria and Liquefaction Susceptibility
(Modified from Tinsley et al., 1985) 90

Table RC4-2 Minimum R for Lateral Ground Spread Models 95

Table RC4-3 Presumptive Pile Stiffness Values (as Derived From Caltrans BDS 104

Table RC4-4 Lateral stiffness and pile capacity 115

Table R10-1 Minimum Material Properties for Structural Steel. 167

Table R10-2 Minimum Material Properties for Pins, Rollers, and Rockers 167

Table RlO-3 Limiting Width-to-Thickness Ratios 180

ATC-32 List of Tables xi

Figure Credits

Geospectra: R3-1 through R3-12 Nishimura, Hwang, and Fukumoto, 1992: RClO-l,
Tsuchida, 1970: RC4-l
MacRae and Kawashima, 1992: RClO-3, RCI0-4
Bartlett and Youd, 1992: RC4-2 and RC4-3
Unknown: 1, RC3-1 through RC3-15, R3-13, RC8-3
Earth Mechanics, Inc.: RC4-4 through RC4-12 through RC8-5, R8-1

ATC-32 Figure Credits xiii

Bridge failures during the October 17,1989 Loma Prieta, form well in earthquakes and meet the seismic safety
California, earthquake demonstrated a clear need for goals established by the Governor. When ATe's pro-
review and revision, as necessary, of the existing seismic posed project was funded by Caltrans in 1991 (ATC-32
design standards and specifications for bridge structures project), the portion of the proposed project pertaining
in California. Thirteen bridges sustained structural dam- to rehabilitation of existing structures was excluded
age severe enough to cause closure for extended periods from the contract and deferred until a later date.
of time and 78 other bridges sustained major damage
(Housner et al., 1990). Damage included collapsed and PROJECT SCOPE
partially collapsed concrete bents; spalled concrete col-
umns; shifted superstructures; anchor bolt and expan- The ATC-32 project team, which consisted of the Project
sion joint damage; damage to bearings, caps, and Manager, Subcontractors, and advisory Project Engi-
earthquake restrainers; large cracks in concrete box cul- neering Panel(PEP), reviewed current Caltrans seismic
vert walls and ceilings; and failure of steel rocker bear- design procedures and recent research in the field of
ings. In addition, the month-long closure of the San bridge seismic design to identify ways to improve Cal-
Francisco-Oakland Bay Bridge, where a link span col- trans' seismic design practice. This work focused on por-
lapsed, and the brief closure of the San Mateo-Hayward tions of the then current Bridge Design Specifications
Bridge, which sustained rocker bearing damage, under- (BDS) pertaining to seismic loading, structural response
scored the need for establishing and implementing seis- analysis, and component design. Special attention was
mic design standards and criteria that will enable critical given to design issues related to reinforced concrete
structures to remain serviceable following severe earth- components, steel components, foundations, and con-
quake-induced ground motions. ventional bearings. In addition, the specifications were
As a result of the effects of the 1989 Loma Prieta revised to more carefully consider displacements in an
earthquake on bridge structures, the Governor of Cali- attempt to satisfy the new performance criteria devel-
fornia appointed a Board ofInquiry to investigate dam- oped by Caltrans during the course of the ATC-32
age resulting from this earthquake and to develop project.
recommendations as to appropriate, necessary actions. Several issues pertaining to earthquake ground
The Board made 52 specific findings and eight recom- motions were considered outside the scope of the
mendations (Housner, et al., 1990). Recommendation 6 project. Caltrans currently has hazard maps that are
calls for ensuring "that Caltrans seismic design policies consistent with safety evaluation under the newly estab-
and construction practice meet the seismic safety policy lished performance criteria. These maps are based on the
and goals established by the Governor". Part A reads: concept of a maximum credible earthquake, which is
"Review and revise standards, performance determined by the location, type, and extent of known
criteria, specifications, and practices to ensure active faults. Seismic hazard is defined in terms of
that they meet the seismic safety goal estab- expected peak rock acceleration values derived from an
lished by the Governor and apply them to the average attenuation of the resulting rock and stiff soil
design of new structures and rehabilitation of motions (determined from published attenuation rela-
existing transportation structures. These stan- tionships). New maps that will consider the effects of
dards, criteria, and specifications are to be thrust faults, added faults, and spectral accelerations are
updated and periodically revised with the assis- currently under development by Caltrans, as a separate
tance of external technical expertise." effort.
Concurrent with the development of the Board of Similarly, it was not within the scope of ATC-32 to
Inquiry's recommendations, Applied Technology Coun- develop seismic hazard maps for functional-evaluation
cil (ATe) submitted a proposal to Caltrans to review and earthquakes. Although the established seismic perfor-
revise as necessary the existing standards, performance mance criteria propose that functional-evaluation earth-
criteria, specifications, and practices for the design and quakes be based on probabilistic principles (e.g., a 60-
construction of new bridge structures and the rehabilita- percent chance of not being exceeded during the life of
tion of existing structures. The intent of the proposed the bridge), the absence of statewide site-dependent seis-
project was to provide criteria and methodology that will mic hazard maps that are consistent with the proposed
ensure that California bridge structures of all types per-

ATC-32 Introduction 1
performance criteria was a factor in the development of b. To develop a design methodology for consider-
the ATC-32 recommendations. ing vertical ground motion.

PROGRAMMATIC AND TECHNICAL DEVELOPMENT c. To develop simplified modeling techniques for

RECOMMENDATIONS capturing the overall seismic response ofbridge
abutments within the overall bridge system,
During the course of the project, the ATC-32 project including the effects of approach embankment
team developed numerous recommendations pertaining response and abutment wall/approach
to the Caltrans BDS as well as to programmatic and embankment interaction. Such techniques
technical development actions that could be imple- should address methods for determining equiv-
mented during or after completion of the project. The alent elastic stiffness, mass, and damping of
technical recommendations are provided later in this various abutment systems.
report. Recommendations pertaining to programmatic
and technical development actions follow. d. To develop and/or verify more realistic, univer-
sal shear capacity models applicable to all prac-
1. The full implication of adopting the recommended tical situations related to reinforced concrete
changes to the Bridge Design Specifications has yet bridge columns. Adoption of the appropriate
to be assessed through extensive trial applications. It model should be achieved through a peer
may be difficult to satisfy these design criteria for review process.
certain bridge configurations under extreme loading
conditions. This may be a signal to designers of e. To establish a quantitative basis (e.g, allowable
potential difficulties with the performance or con- concrete and steel strains) for assessing the
structability of such bridges. Therefore, Caltrans qualitatively defined performance (e.g., repair-
should initiate a trial application period in which the able damage, immediate use) ofbridge col-
impact of adopting the recommended changes to umns. One item of particular concern is the
the Bridge Design Specifications is assessed. prevention oflow cycle fatigue failure oflongi-
tudinal reinforcement in lightly reinforced col-
2. Selected external consultants should participate umns.
during this trial application period in order to assist
Caltrans in making any required modifications to f. To develop nonlinear techniques to distribute
the ATC-32 recommended Bridge Design Specifica- lateral and vertical loads to individual piles in a
tions. pile group.

3. During and following this trial application period, 6. Caltrans should continue to consider and/or
Caltrans should assess the cost impact, design effort, develop new and innovative design strategies to
constructability, and expected performance of meet its challenging seismic design problems.
bridges designed by the ATC-32 recommended
Bridge Design Specifications. TRIAL APPLICATIONS

4. Caltrans should begin developing statewide seismic The recommended Bridge Design Specifications in this
hazard maps for functional evaluation. In addition, document have been reviewed by the ATC-32 Project
existing statewide hazard maps for safety evaluation Engineering Panel (PEP). In addition, bridge design sub-
should be updated to consider the probability of contractors were retained to perform trial designs using
seismic loading in conjunction with the current the draft design specifications. Additionally, experts not
deterministic approach. directly associated with the project were asked to review
portions of these specifications. As with any project of
5. Further development of the bridge design specifica- this type, however, it is not possible to completely evalu-
tions would benefit from additional research. Spe- ate the practical impact of each and every recommenda-
cifically, research should be conducted for the tion in all possible situations. This is why projects of this
following purposes: type and size are traditionally followed by a period in
which the recommendations are applied on a trial basis
a. To develop a design methodology that more to a large number of actual cases. This project is no
accurately reflects the significant characteristics exception, and in some ways such a trial application
of near-fault ground motion and its effect on period is particularly important in this case.
bridge structures.

2 Introduction ATC-32
The principal factor contributing to the increased 3. The state-of-knowledge of seismic design is contin-
need for a trial application period is the time frame in uallyadvancing, and further improvements to the
which these recommendations were developed. The Bridge Design Specifications may become necessary
most critical elements of the recommended design speci- in the future. Nevertheless, the framework of the
fications (e.g., Z factors and rock spectra) were devel- recommended Bridge Design Specifications should
oped first and were subjected to a more thorough readily allow the inclusion of such improvements as
evaluation by trial applications than were the elements they become available.
developed in the second phase of the project (e.g., spec-
tra for soil sites and joint shear requirements). In addi- 4. Some of the safety factors implicit in the current
tion, ongoing laboratory research and the experience of BDS have been replaced with a more direct treat-
the 1994 Northridge earthquake contributed substan- ment of safety margin.
tially to the advancement of knowledge during the
course of this project, resulting in some modification REPORT ORGANIZATION AND COMPANION REPORT
and enhancement of the recommendations. Finally, the
trial applications themselves led to additional modifica- The ATC-32 report includes a summary of the recom-
tions. Some of the modified provisions have not been mended changes to the Caltrans BDS related to seismic
thoroughly tested in trial bridge designs. design, followed by the detailed specifications recom-
It is not unrealistic to expect that future trial appli- mended by the ATC-32 PEP. An article-by-article listing
cations may point out the need for further modification of the entire Caltrans BDS is provided in Appendix A,
of these recommended specifications or associated Cal- followed by a list of project participants and information
trans design procedures. on other ATC projects and reports.
At the request of Caltrans, the detailed recom-
OTHER COMMENTS, FINDINGS, AND CAUTIONS mended specifications have been written in specification
language consistent with the format of the current Cal-
1. Although a critical review of the current Caltrans trans BDS and Commentary. This was done to facilitate
Bridge Design Specifications found numerous evaluation and implementation of these recommenda-
opportunities for improvement, the general concept tions by Caltrans. Although the recommendations are
and format used in these specifications are reason- similar in form to the current Caltrans BDS and retain
able and suitable for further enhancement and many of the same procedures, they differ fundamentally
refinement. Some recommendations of the ATC-32 from the current specifications in that they were devel-
project have already been adopted by Caltrans. oped primarily with displacement response in mind.
Therefore, individual recommendations should not be
2. The recommended changes to the Bridge Design interpreted out of context of the entire document.
Specifications were developed as an integrated pack- The companion document, ATC-32-1 (ATC, 1996),
age. It is the intent that these recommendations, includes additional detailed discussion of the recom-
when properly applied, may result in structures that mendations. It also discusses alternative design methods
satisfy the performance criteria established by Cal- and areas of current research.
trans. If not adopted in their entirety, care should be
exercised when applying any of the recommended
provisions in a piecemeal fashion.

ATC-32 Introduction 3
Summary of Recommendations
Introduction Seismic Performance Criteria

The current California Department of Transportation Recently, Caltrans, with the support of an external Seis-
(Caltrans) Bridge Design Specifications (BDS)(Caltrans, mic Advisory Board and the ATC-32 project team, has
1986) are comprehensive provisions covering all aspects developed a set of seismic performance criteria for new
of bridge design. They are based on the 1983 American bridges. These criteria, which are the basis for the recom-
Association of State Highway and Transportation Offi- mended revisions to the BDS, are summarized in Table
cials (AASHTO) Bridge Specifications (AASHTO, 1983) 1.
and subsequent interim modifications. Caltrans has fur- In these criteria, both safety-evaluation and func-
ther modified these AASHTO specifications to suit its tional-evaluation design earthquakes are defined. The
specific needs, particularly in the area of seismic design. safety-evaluation earthquake, which Caltrans currently
This includes the use of elastic design spectra (ie., ARS defines deterministically as the Maximum Credible
curves) and the introduction of period-dependent Z fac- Earthquake (MCE), has only a small probability of
tors to account for ductility and risk in individual struc- occurring during the useful life of the bridge. A statewide
tural components. The basic earthquake design force is hazard map given in terms of the peak bedrock accelera-
therefore given by tion generated by this level of earthquake has been avail-
able for some time (CDMG, 1992). In the newly defined
EQ = mgARS (1) performance criteria, the safety-evaluation earthquake
may alternately be defined probabilistically as an earth-
where m is the participating mass of the bridge, and g is quake with a 1000- to 2000-year return period. The
the acceleration of gravity. probabilistic safety-evaluation ground motion must be
The recommended changes to the Caltrans Bridge determined on a site-specific basis.
Design Specifications that were developed as part of the The functional-evaluation earthquake is intended to
ATC-32 project deal only with those portions of the cur- represent an event that has a reasonable probability of
rent BDS that are related to seismic design. This involved not being exceeded (approximately 60%) during the life
a complete revision of Article 3.21 dealing with seismic of the bridge. Because no statewide hazard map for these
loads; the addition of Article 4.5, which covers the seis- earthquakes has been developed at this time, the func-
mic design of bridge foundations; and the modification tional-evaluation ground motion must also be deter-
and/or addition of several articles in Sections 8 and 10 mined on a case by case basis through site-specific
that deal respectively with the seismic design of rein- studies.
forced concrete and steel bridge components. Appendix Performance is defined in terms of two criteria: the
A is an article-by article listing ofthe entire Caltrans BDS service level of the structure immediately following the
with those Sections and Articles that were modified as earthquake and the extent (or repairability) of physical
part of the ATC-32 project shown in bold type. This out- damage. Although performance is defined qualitatively,
line is intended to provide a road map to the BDS the recommended revisions to the BDS are based on a
changes described later in this report. The following more quantitative definition established by the ATC-32
paragraphs summarize the recommended changes to the project. Required performance varies for each of the two
current Caltrans BDS. earthquake loadings defined above. Required perfor-
mance also depends on whether a bridge is classified as
Important or Ordinary.

Table 1 Seismic performance criteria

Ground Motion at Site Ordinary Bridges Important Bridges

Functional-Evaluation Service Level- Immediate Service Level - Immediate

Ground Motion Repairable Damage Minimal Damage

Safety-Evaluation Service Level- Limited Service Level- Immediate

Ground Motion Significant Damage Repairable Damage

ATC-32 Summary of Recommendations 5

Structural Action The standard design spectra may also not be appro-
priate for sites adjacent to active faults. At these sites, the
A new requirement ofthe recommended ATC-32 BDS is standard spectra may account for the high spectral accel-
that the designer identify the type of structural action erations, but may not adequately account for the pulse-
desired. Fully ductile behavior assumes that the designer type motion or the differences between fault-normal and
will take maximum advantage of plastic hinging while fault-parallel motions observed in past earthquakes. The
ensuring structural safety. This type of action implies effect of these motions on structural response is most
considerable damage and is reserved for Ordinary accurately determined from an inelastic dynamic analy-
Bridges only. Structural action consistent with limited sis using spectrum-compatible motions that contain the
ductility is recommended for Important Bridges and appropriate velocity pulses. The ATC-32 recommenda-
certain critical foundation components. This type of tions give some guidance for selecting appropriate time
structural action is intended to limit inelastic response to history input motions.
levels consistent with reduced structural damage. Elastic The nature ofvertical earthquake loading is com-
structures carry seismically induced loads elastically and plex: it depends on rupture mechanism, proximity ofthe
thus remain undamaged. Finally, the proposed specifica- earthquake source, local soil conditions, and other fac-
tions recognize the potential use of protective systems tors. The ATC-32 revisions recommend that vertical
that incorporate base isolation, passive energy dissipa- earthquake design loading may be taken as two-thirds of
tion, and other mechanical devices intended to control the horizontal loading spectra for typical sites not adja-
seismic response, although no specific design guidelines cent to active faults. When available, site-specific vertical
are given for these systems. loading spectra are preferred.

Seismic Loading Analysis

Recent studies of strong motion instrumentation results Although the ATC-32 recommendations retain a force-
have yielded information that makes it possible to refine based design approach, some of the inherent shortcom-
the current Caltrans design spectra. Therefore, new ings of this approach have been overcome. This is done
design spectra for three earthquake magnitude ranges through the use of new response modification factors
were developed as part of the ATC-32 project. Because and modeling techniques for analysis that more accu-
some California sites can be adversely affected by Maxi- rately consider seismic displacement. The ATC-32 pro-
mum Credible Earthquakes on a number of different cedures also provide specific means for directly
faults, it may be necessary to design for multiple spectra considering geometric and material nonlinearity in spe-
in some cases. cial cases.
The proposed family of site-dependent design spec- As shown in Table 3, the ATC-32 project has devel-
tra, which vary from the current Caltrans curves, are oped recommended requirements for the minimum
based on four of the six standard sites defined in a
ground motion workshop sponsored by the National Table 3 Minimum Required Analysis
Center for Earthquake Engineering Research (NCEER)
(Martin and Dobry, 1994). These standard sites are pri- Functional Safety
marily characterized by the typical shear wave velocity of Evaluation Evaluation
the upper 100 feet of the soil profile, as shown in Table
2. Spectra for type A (hard rock) and F (poor soils) sites Ordinary Bridge None Required AorB
Type I
as well as type E sites with peak rock accelerations over
0.4 g must be determined on a site-specific basis. Ordinary Bridge None Required B
Type II
Table 2 Site Characteristics for Standard Design Important Bridge AorB AorB
Type I
Important Bridge B BandC
Site Shear Wave Type II
Designation Site Description Velocity Range
A = Equivalent Static Analysis
B Medium rock 2500 to 5000 ft/sec
C Soft rock/Dense soil 1200 to 2500 ft/sec B = Elastic Dynamic Analysis
D Stiff soil 600 to 1200 ft/sec
E Soft soil < 600 ft/sec C = Inelastic Static Analysis (Substitution of Inelastic
Dynamic Analysis is Acceptable)

6 Summary of Recommendations ATC-32

type of analysis that should be used under various cir- reduce design quantities to 80 percent of those deter-
cumstances. The type of analysis depends on whether or mined from elastic analysis. Both geometric and material
not the bridge is classified as Important and on the com- nonlinearity should be considered. In general a lumped-
plexity of the structural configuration (Type I =simple mass "stick" model with five percent of critical damping
and Type II = complex). These analysis types include is appropriate. The maximum response to three repre-
Equivalent Static Analysis, Elastic Dynamic Analysis, and sentative input motions or the average response to seven
Inelastic Static Analysis. Basic requirements for each of such input motions is recommended.
these analysis types are also included. As with current Caltrans practice, the results from
Equivalent Static Analysis allows an equivalent static Equivalent Static Analysis or Elastic Dynamic Analysis
force to be applied to the structure. The magnitude of for orthogonal response spectrum loadings must be
this force is determined from the value of the design combined to obtain design forces and displacements.
spectra at the structure's fundamental period ofvibra- The results for each orthogonal loading are first
tion. This force is applied at the vertical center of mass obtained by combining the maximum modal responses
and distributed in the horizontal plane based on the dis- according to the complete quadratic combination
tribution of mass in the structure or on the product of (CQc) rule. The ATC-32 recommendations then pre-
mass distribution and displacement. scribe the "40 percent rule," as opposed to the "30 per-
Elastic Dynamic Analysis is required when the dis- cent rule" currently used by Caltrans for combining the
tribution of stiffness and/or mass within the structure results for orthogonal loadings. In addition, vertical
and/or the configuration is complex enough to preclude motion is included when it is critical. Therefore, three
the reliable prediction of response without such an anal- design load cases may be considered, each of which
ysis. In most cases a multi-modal response spectrum includes 100 percent of the actions for loading in one of
analysis using a lumped-mass "stick" model will satisfy the orthogonal directions plus 40 percent of the actions
these requirements. It is Caltrans practice to use this type for each of the remaining two orthogonal loadings.
of analysis for most bridges, since the analytical capabili- Alternately, 100 percent of all three orthogonal loadings
ties are readily available to most designers. Member stiff- may be applied simultaneously, and the modal results
ness values that account for cracking of reinforced combined using the square root of the sum of the
concrete members are to be used in both Equivalent squares (SRSS) method. When either elastic or inelastic
Static Analysis and Elastic Dynamic Analysis. This differs time-history analysis is used, 100 percent of the loadings
from the current Caltrans practice of using gross section in each of the orthogonal directions is applied simulta-
properties for force demands. neously, and the resulting maximum actions are taken
Inelastic Static Analysis is required only when the directly from the analysis results.
bridge is classified as Important and it is not simple in The ATC-32 recommendations also provide a
configuration. The analysis, commonly referred to as a method for adjusting the displacement results from an
"push-over" analysis, is done in conjunction with Elastic Elastic Dynamic Analysis to better reflect the actual max-
Dynamic Analysis, and requires a preliminary determi- imum inelastic displacements that are likely to occur
nation of the strength and stiffness of critical members. during an earthquake. The adjustment factor, R d, is
In this analysis, loads are applied incrementally until the given by the following formula:
structure has reached ultimate displacements. At each
step, changes in the structure's characteristics due to
geometric and material nonlinearity are considered. The (2)
effects of gravity loads including dead load and a portion
of the live load are also considered. Results of this analy-
sis are used to confirm that the structure is capable of where
accommodating the displacement demands determined
from an Elastic Dynamic Analysis. A factor of safety of T = natural period of the structure
1.5 for displacement capacity versus displacement T* = predominant period of ground motion
demand is recommended. In general, results of this anal- Z = response modification factor
ysis cannot be used to reduce design quantities deter-
mined from an Elastic Dynamic Analysis. This adjustment factor was derived empirically for typi-
Although Inelastic Dynamic Analysis is not required cal ground motions and may not be appropriate for
for any structure type, the ATC-32 criteria provide near-fault sites where pulse-type motions are likely.
guidelines for conducting such an analysis. This type of Although several simplified methods have been sug-
analysis may be substituted for Inelastic Static Analysis. gested for assessing the impact ofpulse-type motion on
Because member strength and stiffness values are a pre- structural response, Inelastic Dynamic Analysis is still
requisite, this analysis is used primarily for verifying a the most accurate method currently available for this
completed design, although its results may be used to purpose.

ATC-32 Summary of Recommendations 7

Caltrans currently determines component design nents. Use of capacity design principles allows the
forces by dividing the forces obtained from elastic analy- designer a measure of control over the location of struc-
sis by Z factors to account for ductility and risk. Revised tural damage within the structure so that these locations
Z factors have been developed as part of the ATC-32 can be detailed to provide for ductile behavior. Capacity
project. Nonlinear dynamic analysis studies demon- design is a popular concept that is also included in the
strated that very little, if anything, was lost in using a current Caltrans Bridge Design Specifications and many
simplified Z factor, as opposed to a more complicated other seismic design specifications and guidelines.
factor based on column aspect ratios. Therefore, the Nonlinear analytical studies have been performed to
ATC-32 recommendations include simplified Z factors determine the dynamic effect of P-fl moments. It was
for columns and other components.•The full value ofZ found that a biased response could be prevented ifyield-
applies at a structural period of T*, and the value ofZ ing was limited by keeping the plastic base shear at a high
decreases linearly with period. Z reaches a minimum enough level. The following equation given in terms of
value of 1.0 at a period of zero. Charts showing new Z the ratio of the ultimate displacement, 0u' divided by the
factors, which are typically lower than those defined in average column height, H, was established to prevent
the current Caltrans Bridge Design Specifications, are bridge columns from being significantly affected by P-fl
included in Figure 1. moments.

Full Ductility Structures

5r-----,----"'T':":-:-::---::-r-:-----,--:----, (3)
:ri 4 I----+-+---=r---l----j-----l
§ 3 1----+--7L--+----f-----+----I Because a typical bridge column has a drift value of
g approximately 0.03, this implies a minimum plastic base
ijl 2 f----r-+-t-=oot----I-----+----l shear coefficient of approximately 0.12 to prevent unac-
ceptable P-fl effects. When Equation 3 is not satisfied,
~ 1~ ~=t===I====l=====I====+====l
r- the ATC-32 recommendations require that the response
Brittle elements not designed by capacity design
o 0~---:0:-':.5:-----':1----:-1"::.5--=---'--:-2-=--...:::....-='2.5 analysis directly consider geometric nonlinearities such
Period Ratio, TIT* as the P-fl effect.

Umited Ductility Structures
A great deal of recent research has focused on the design
Well conliJed concrete lOlumns, _

( steel COluins and pile rafts ofvarious structural components. Much of this research
has been aimed at assuring the ductile behavior of these
§ 3 \ components during large earthquakes. This usually
Transversely loaded piers;
g /'( abutment walls and wing walls requires careful attention to detail in the affected mem-

a: bers. When ductile response is not possible, or when it is
~ 1 undesirable to incur the damage that inelastic response
o "- Brittle elements not designed by capacity design implies, capacity design principles are applied to assure a
o 0.5 1 1.5 2 2.5 failure mechanism that protects critical components
Period Ratio, TIT* from inelastic behavior. The ATC-32 recommendations
contain several provisions that address these design
Figure 1 Response modification factor Z issues for reinforced concrete and structural steel com-
ponents. These requirements are discussed briefly in the
The ATC-32 project also reviewed simplified design following paragraphs.
procedures for restrainer cables. Although the current
Caltrans approach is not technically correct, no other Reinforced Concrete
simplified method seems to give better results. Because
of this, and the general feeling that restrainers are sec- Modifications to several aspects of reinforced concrete
ondary to the practice of providing adequate seat widths design have been proposed by the ATC-32 project. These
at expansion joints, no change in the current Caltrans include the design of ductile elements, the design of
method is recommended. nonductile elements and actions using capacity design,
The ATC-32 recommendations provide that, when- and detailing of reinforced concrete for seismic resis-
ever feasible, nonductile components and actions are to tance.
be designed using capacity design principles that con- As with current Caltrans procedures, flexural design
sider the possibility of overstrength in ductile compo- of ductile reinforced concrete columns is based on elas-

8 Summary of Recommendations ATC-32

tic moment demands that are divided by the appropriate shear strength is suggested to be the sum of a contribu-
Z factor. The elastic moment demands are determined tion from concrete, Ve, and a contribution from hori-
from an analysis that reasonably simulates the degraded zontal reinforcing steel, Vs• The nominal concrete
stiffness of the bridge during a large earthquake and thus contribution to shear resistance is given by
uses cracked section properties for the columns. In
determining moment capacity, expected material
strengths are used rather than nominal material
strengths. Although lower Z factors are used in the rec-
ommended design specifications, trial designs and col- (5)
umn design studies using these recommendations
showed that longitudinal column reinforcing steel was
actually less than that required by the current Caltrans within plastic hinge zones, and by
design specifications in most cases. This was primarily
attributed to the use of cracked section properties for
analysis, which resulted in lower elastic force demands (6)
and the use of expected material strengths, which
resulted in higher capacities. The recommended ATC-32
design specifications place a lower limit of 0.01 on the outside of plastic hinges. In both cases Pe is equal to the
longitudinal column reinforcing steel ratio and an upper design axial compressive force in the column, A g is the
limit of 0.04. gross cross-sectional area and A e is the effective shear
An evaluation of the displacement capacity/demand area which is 0.8 A g for columns. Slightly modified ver-
ratio for columns designed by the ATC-32 recommen- sions of these formulas are recommended for columns
dations indicates that their expected performance is subjected to axial tension.
superior to columns designed by the current procedures, The nominal shear contribution from reinforcing is
although expected performance varied significantly from given by
column to column. Improvement was primarily due to
increased displacement capacities resulting from more
stringent confinement requirements. The recommended (7)
requirement for the volumetric reinforcement ratio of
spirally reinforced columns is given by:
for tied rectangular sections, and by

Ps = O.16f ce [ O.5+TA
1.25PJ +O.13[p[-O.Ol] (4)
Tye ce g (8)

for spirally reinforced circular sections. In these equa-
fee = expected concrete strength tions, A v is the total area of shear reinforcement parallel
Ire= expected yield strength of the reinforcement to the applied shear force, A h is the area of a single hoop,
P= column axial load Irhis the yield stress of horizontal reinforcement, D' is
Ag = gross column area the diameter of a circular hoop, and 5 is the spacing of
PI = longitudinal reinforcement ratio horizontal reinforcement along the axis of the member.
Shear demands in ductile columns are higher than
An additional requirement, which is designed to prevent those required by the current Caltrans specifications.
inelastic buckling of the longitudinal reinforcing requires The recommendations call for determining plastic
a volumetric ratio for spirals that is linearly related to the moments using capacity design principles in a manner
number oflongitudinal reinforcing bars. Improved pro- similar to current practice. They differ, though, in that
visions for transverse reinforcement of tied columns and plastic moments are based on expected rather than
piers have also been included. A provision that allows nominal material strengths and a higher overstrength
spirals and hoops to be designed directly using plastic factor of 1.4. Alternate methods are recommended for
moment-curvature analysis considering the required calculating plastic moments, but these also result in high
plastic hinge rotation has also been added to the recom- shear demands. The net result of the recommended
mended design specifications. ATC-32 shear provisions is an increase in the require-
Revised column shear design criteria are recom- ments for column shear reinforcement, although the
mended, which are consistent with the format of current previously defined confinement requirements will usu-
American Concrete Institute (ACI) provisions. Column

ATC-32 Summary of Recommendations 9

ally control the design of horizontal column reinforce- Structural Steel
ment, except for relatively short columns.
New anchorage provisions from the 1995 ACI com- Steel seismic design guidelines, which are absent from
mittee recommendations are adopted as part of the the current Caltrans BDS, have been developed as part
ATC-32 recommended design specifications. These pro- of the ATC-32 project. Construction and detailing
visions have more liberal bar spacing requirements than requirements for steel framing and various types of steel
the 1989 ACI provisions. They also provide a generalized joints likely to be used in bridge work are spelled out.
method for considering non-standard cover, spacing, These requirements are directed toward moment-
and transverse reinforcing steel in tied columns. In addi- resisting beam-to-column connections, diaphragms and
tion, ATC-32 has added a new anchorage equation for cross-bracing, slip-critical bolted connections, concen-
spirally confined column reinforcement. This equation, trically braced steel frames, and stiffened as well as
which is based on recent laboratory experiments, can unstiffened box sections. In addition, recommendations
reduce anchorage length requirements within bent caps for the seismic design of conventional bridge bearings
and footings, thus making it practical to use large- have been developed.
diameter bars (#14 and #18) without hooks. To achieve Moment-resisting beam-to-column joints, which
these reduced lengths, significant confmement reinforce- are typical in building construction, can also be used in
ment is required within anchorage zones unless the bridge structures. In bridges, the strong beam/weak col-
zones are confined by prestress or solid adjacent mem- umn principle is mandated and panel zone detailing
bers. requirements are prescribed that preclude yielding
If sufficient spiral confmement reinforcement is within the joint. The ATC-32 recommendations require
provided, the ATC-32 recommendations also allow a that joint regions be designed to force plastic hinges to
reduction in the splice length of column reinforcement. form in the column at some distance from the joint in
Splices are prohibited within a zone that includes the order to avoid the brittle joint failures experienced in
plastic hinge length plus the column diameter. Although steel moment-resisting building frames during the 1994
the ATC-32 recommendations for splicing were devel- Northridge earthquake. The recommendations include
oped from laboratory testing that included large- additional detailing, slenderness, and compression
diameter bars, lap splicing of #14 and #18 bars is flange bracing requirements that are intended to prevent
discouraged. local buckling and assure ductile behavior.
Longitudinal bar sizes in short columns that are The transmission ofseismically induced forces must
subjected to high moment gradients are limited, due to have a clear load path into the substructure. Typically,
flexural bond requirements. The limitation on bar size is these forces will be transmitted through the deck by dia-
derived from the equations developed for bar splicing. phragm action or upper flange lateral bracing into end
New design requirements for shear and bending cross-bracing or diaphragms, and finally through the
within footing and superstructure joints are also recom- bearings and anchor bolts. Internal cross-bracing will
mended in the proposed ATC-32 design specifications. only be used for seismic resistance if deck diaphragm
These requirements are based on laboratory test results action or upper flange lateral bracing is inadequate. The
and have been used in the design of some Northridge seismically induced forces, which must take into account
earthquake replacement structures. The proposal the concentration of force due to misalignment of bear-
requires that joints be capable of resisting plastic column ings, etc., should generally be resisted elastically. In
moments through a combination of concrete and rein- highly skewed or unusual structures, sophisticated three-
forcing steel action. Joint reinforcing requirements are dimensional analysis techniques may be required to
based on the magnitude of principal tensile stresses determine these forces. Capacity design principles result-
within the joint. When these stresses are below 3.5 Ye, ing from substructure yielding can be used to limit bear-
steel reinforcement is required to carry 50% of these ing forces. In special cases, force reduction will be
stresses. For tensile stresses above 3.5Ye, specific ver- allowed due to bearing movement, provided that the rel-
tical, horizontal, and spiral joint reinforcement is ative movements are within acceptable limits. The use of
required. This reinforcement, which is specified in terms roller and rocker bearings are discouraged, as are pot
of fixed percentages of the longitudinal column steel bearings subjected to high vertical accelerations.
anchored in the joint, must be placed in the cap beam or Concentric bracing, which must be designed to
footing within a distance equal to one-half the column remain elastic, must satisfy specific slenderness require-
diameter. In addition, principal compressive stresses ments. In general, at least 30 percent of the seismically
within the joint are limited to 0.25 f c' induced forces to be resisted by such bracing must be
resisted by members acting in tension.
In order to insure ductile behavior of stiffened and
unstiffened box sections, the revised BDS requires that
details for stiffened columns comply with certain slen-

10 Summary of Recommendations ATC-32

derness requirements and that the amount ofyielding be passive pressure on the sides of the pile cap. Bending
limited through the use of Z factors. Minimum shear strength and stiffness is generally assumed to be attribut-
strength requirements for unstiffened knee joints are able only to the piles themselves. In general, it is appro-
also specified. Many of these requirements are based on priate to use ultimate force and moment capacities that
recent Japanese research on steel bridge columns are consistent with the performance criteria when
(Kawashima et al., 1992). designing pile foundations. Individual piles must gener-
ally be capable of resisting axial loads in both tension and
Foundations compression, as well as lateral loads. The ATC-32 com-
mentary provides an extensive guideline for modeling
Foundation design guidelines that have been developed and designing pile foundations including design charts
as part of the ATC-32 recommendations include provi- for determining pile head stiffness under various condi-
sions for site investigation; determination ofsite stability; tions. Some of these charts account for the soil overbur-
and modeling and designing of abutments and wing- den at a pile foundation, which can often have a
walls, pile and spread footing foundations, drilled shafts, significant impact on pile lateral stiffness and strength.
and earth-retaining structures. For the most part, these Pile shafts are essentially an extension of the bridge
recommendations tend to validate current Caltrans column into the ground and are a special form of pile
practice although there are suggested refinements in foundation used frequently in California. The ATC-32
some cases. They include the latest results of research, recommendations provide guidelines for choosing the
for example, abutment research at University of Califor- correct parameters for considering soil-pile interaction
nia, Davis (Maroneyet al., 1992). It should be pointed when designing and modeling this type of foundation. In
out, however, that many issues related to the effect of addition, a more rational criteria is suggested for deter-
foundations on total system response are still not fully mining the required pile length for lateral stability.
substantiated. Provisions for designing spread footings are also
The recommendations include guidelines for con- included in the ATC-32 recommendations. This type of
ducting geotechnical site investigations when there is a foundation must carry the necessary earthquake-
potential for large earthquake loadings. These guidelines induced loads without excessive settlement or overturn-
outline the information to be collected during such an ing. The allowable area of uplift for different types of
investigation and constitute a standard of practice. footings is specified and some guidance is given for
The ATC-32 recommendations require that poten- determining appropriate settlements under earthquake
tial bridge sites be investigated for possible ground insta- loads.
bilities. Site stability is affected by several factors The ATC-32 recommendations also contain specific
including liquefiable soils, lateral spreading, the presence requirements for the design of earth retaining systems.
of soft clay soils, slope hazards, and surface fault rup- In addition to being structurally capable of resisting
ture. The commentary to the ATC-32 recommendations static and dynamic earth pressures, earth retaining sys-
provides methods for identifying, and guidance on tems should not experience excessive lateral or rota-
quantitatively evaluating, these hazards. In addition, tional movements.
practical methods for mitigating some of these hazards
are discussed. Concluding Remarks
With respect to the passive pressure generated at an
abutment during an earthquake, the ATC-32 recom- The ATC-32 recommendations represent a significant
mendations provide for a uniform ultimate passive pres- step forward in bridge seismic design specifications. They
sure of7.7 kips per square foot, as is current Caltrans are built on the previous efforts of Caltrans and thus
practice. This pressure, which is intended for typical retain many of the features of the current Caltrans
eight-foot high abutment walls, is reduced linearly for Bridge Design Specifications. However, they do differ
shorter abutment walls. This pressure is developed at lat- from the current Caltrans approach in some fundamen-
eral wall displacements of 0.01 to 0.02 times the wall tal ways.
height. Special modeling consideration must be given to It would be a mistake to assume that the ATC-32
expansion gaps, lateral pile stiffness and abutment skew recommendations fully address all issues. Even during
angle. Modeling of abutment stiffness follows the cur- the course of the project, ongoing research efforts and
rent Caltrans procedure in which a trial and error experience from actual earthquakes were advancing the
approach is used to find the appropriate secant stiffness state-of-knowledge, requiring that modifications be
of the abutments. made to draft recommendations on an ongoing basis. As
The ATC-32 recommendations require that pile Caltrans begins to implement these recommendations, it
foundations have sufficient capacity to resist forces and is expected that further modifications and refinements
moments transmitted from the rest of the structure. Lat- may be required.
eral strength and stiffness is provided by the piles and

ATC-32 Summary of Recommendations 11

Revised Bridge Design Specifications

The current California Department of Transportation by-article listing of the entire BDS is included in Appen-
(Caltrans) Bridge Design Specifications (BDS) are a dixA.
modified version of the 1983 American Association of The recommended changes are presented in a two-
State Highway and Transportation Officials (AASHTO) column format with specifications in the left column and
Standard Specifications for Highway Bridges with commentary in the right. Article numbering is consistent
Interim Specifications for 1984, 1985, and 1986. Caltrans with the current BDS, but equation, figure, and table
uses these specifications for seismic design in lieu of the numbers are given consecutively as they appear in the
current AASHTO Division I-A seismic design specifica- recommended revisions. Separate consecutive equation,
tions and the current AASHTO LRFD bridge design figure and table numbering schemes are presented for
specifications. both the specifications and the commentary. This was
The following pages contain the recommended done because, as ofthis writing, Caltrans has not decided
changes and additions to the current Caltrans BDS. Arti- which of the recommendations will be implemented.
cles that are unchanged have not been repeated in order When articles are not modified or when it is recom-
to save space. Because the current Caltrans BDS is a mended that they be deleted, a note indicating this
comprehensive document covering many aspects of action is provided at the location where the article would
bridge design in addition to seismic design, only a rela- normally appear. In addition, the ATC-32 Project Engi-
tively small portion has been revised as a part of the neering Panel (PEP) made several comments regarding
ATC-32 project. To help the reader put these recom- various recommendations, which appear as footnotes
mended changes and additions in perspective, an article- throughout the revised BDS.

ATC-32 Revised Bridge Design Specifications 13

Section 3


Articles 3.1 through 3.20 not modified.


Bridge structures shall be designed according to the Article 3.21 describes design requirements for earth-
provisions of Article 3.21 to resist earthquake motions, quake resistance. The requirements are based on two-
considering the relationship of the site to potentially level performance criteria, intended to preserve func-
active seismic sources, the seismic response of soils at tionality after earthquakes having a reasonable probabil-
the site, and the dynamic response characteristics of the ity of occurring once or more during the design life of
total bridge. the bridge and safety after earthquakes having very low
probability of occurring during the design life of the
bridge. Different requirements exist for Important and
Ordinary Bridges.
The specifications contained in Article 3.21 are con-
sidered to represent minimum requirements for produc-
ing a structure with adequate proportions and details to
enable the structure to resist earthquake effects without
critical loss in strength. The specifications are based on
the assumption that the structure resists the maximum
specified earthquake effects by virtue of the stiffness
reduction and energy dissipation that result from non-
linear response. It is important that the structure be laid
out and proportioned so that a viable load path exists to
transmit inertial loads to the foundation. It is equally
important that structural elements be provided with
details that enable the bridge to respond in a ductile
1. These recommended revisions to the Bridge Design Specifications manner.
are intended to reflect a more realistic assessment ofthe behavior of a
The specifications have been written with the expec-
bridge in a large earthquake. They include significant changes to both
seismic design demands and capacities. Although the net effect of these tation that nonlinear action during a design earthquake
changes does not appear to result in designs that are radically different will be restricted to zones that have been selected and
from current bridge designs in most cases, there may be some bridge specially detailed for ductility by the designer. The
configurations that are significantly affected. Although trial designs and design process is intended to result in inelastic action
other analytical evaluations of these specifications have been per-
restricted to these locations, with other locations pro-
formed, they have been limited in scope and cannot fully evaluate the
impact of these specifications in all cases. Therefore, it is recommended tected from inelastic action. In most cases, inelastic
that further trial applications be conducted to assess the full impact of action should be designed to occur in flexure in support-
implementing these provisions. ing columns and pier walls. The reasons are: (1) cross

ATC-32 8DS Recommendations, Section 3: Loads 15


sections are well defined and procedures for detailing for

ductility are well established, effective, and economical;
(2) damage can often be readily inspected following an
earthquake; and (3) damage can be readily repaired fol-
lowing an earthquake. Except in unusual cases, the
design should explicitly aim to avoid inelastic response in
foundations and superstructures. Two exceptions are
extended pile shafts that are allowed to yield below grade
and abutments and wing walls that are allowed moder-
ate levels of inelastic response. Brittle failures should be
avoided in all members except sacrificial members
designed to act as fuses.
Although not specifically covered in this specifica-
tion, hydrodynamic effects, including drag and added
mass, should be considered where important.
The overall design approach is derived from earlier
Caltrans design specifications. Important modifications
include two-level performance criteria (Article 3.21.2),
restrictions on types of structural action (Article 3.21.3),
restrictions on types of analysis (Article 3.21.4), new ARS
spectra (Figures R3-1 through R3-12), reduced compo-
nent design stiffnesses (Article, inelastic analysis
methods (Articles 3.21.7 and 3.21.8), modified design
displacements (Article 3.21.10), new force reduction fac-
tors Z (Article 3.21.11), and P-t:.. restrictions (Article
3.21.15). The calculated design displacements are likely
to exceed those obtained using the previous specifica-
tions for similar bridge structures.

3.21.1 Notation

A = Estimated mean acceleration at bedrock or

"rocklike" material from the safety-evaluation
A a = Effective peak acceleration-related acceleration
used in Table RC3-2.
A., = Effective peak velocity-related acceleration used
in Table RC3-3.
ARS = Five-percent-damped elastic acceleration
response spectrum at the site, expressed in
terms ofg.
d, = Total thickness of cohesive soil layers in upper
100 ft. at site.
di = Thickness of soil layer i
d s = Total thickness of cohesionless soil layers in the
upper 100 ft. at site.
F = The total uniform force applied to the
superstructure that will cause a one-inch
maximum horizontal deflection in the direction
of loading. This force represents the total
stiffness including the stiffness of the
superstructure, supporting members, and
surrounding soil. Units shall be consistent with

16 BDS Recommendations, Section 3: Loads ATC-32


Fa = Soil amplification factor for the acceleration

controlled part of the rock spectrum (Soil
profile type B).
Fv = Soil amplification factor for the velocity
controlled part of the rock spectrum (Soil
profile type B).
g = Acceleration due to gravity.
H = Maximum height of supporting member for a
frame between superstructure hinges.
Hi = Standard penetration resistance of soil layer i
N = Generalized standard penetration resistance for
upper 100 feet at site (commentary to Article
Nch = Generalized standard penetration resistance for
only the cohesionless layers at site (commentary
to Article
PI = Plasticity index of clay soil.
Rd = Amplification factor applied to elastic modal
spectral displacements to obtain design
R = Five-percent-damped mean elastic acceleration
response spectra on rock (magnitude and
distance dependent).
S = Soil amplification spectral ratio.
Su = Undrained shear strength.
Su = Generalized undrained shear strength of the site
(commentary to Article
Sui = Undrained shear strength of soil layer i
T = Fundamental period of vibration, in seconds, of
the bridge as a whole. For Equivalent Static
Analysis, compute T for entry to the ARS curves
by the expression

T = 0.32JWIF

T = Characteristic ground motion period,

corresponding to the peak of the input energy
spectrum. Values ofT* are given in Table RC3-
V o = Base shear strength of a frame between
superstructure hinges, determined by plastic
Vs = Generalized shear wave velocity for upper 100
ft. at a site (commentary to Article
vsi = Shear wave velocity for soil layer i
W = Dead load of bridge or frame. Units shall be
consistent with F.
Z = Force reduction coefficient, see Figure 3-13
Ou = Maximum design displacement of a frame, from

ATC-32 8DS Recommendations, Section 3: Loads 17


3.21.2 Performance Criteria C3.21.2 Performance Criteria

All bridges shall be designed to meet the seismic perfor- Table R3-1 presents a matrix: of seismic performance
mance criteria given in Table R3-1. Definitions of the objectives defined as a function of ground motion at the
terms in Table R3-1 are given in Articles site and the criticality/importance of the bridge
through structure.

Table R3-1 Seismic performance criteria

Ground Motion at Site Ordinary Bridges Important Bridges

Functional-Evaluation Service Level - Immediate Service Level- Immediate

Ground Motion Repairable Damage Minimal Damage

Safety-Evaluation Service Level- Limited Service Level - Immediate

Ground Motion Significant Damage Repairable Damage Bridge Category C3.21.2.1 Bridge Category

Each bridge shall be classified as either Important or Two bridge categories are defined. A bridge providing
Ordinary, as follows: access to an emergency facility is an example of a bridge
that might be required to provide secondary life safety. A
(a) Important Bridge: Any bridge satisfying one or bridge that serves as a major link in the transportation
more of the following: system is an example of one whose loss (even tempo-
rary) might create a major economic impact. Normally,
.... required to provide secondary life safety it will be the responsibility of the owner to select the
appropriate bridge category.
.... time for restoration of functionality after closure
would create a major economic impact

.... formally designated by a local emergency plan as


(b) Ordinary Bridge: Any bridge not classified as an

Important Bridge. Evaluation Levels C3.21.2.2 Evaluation Levels

(a) Safety-Evaluation Ground Motion: This ground In writing this specification, it has been assumed that
motion may be assessed either deterministically or Ordinary Bridges will automatically meet the perfor-
probabilistically. The deterministic assessment cor- mance criteria for the functional-evaluation ground
responds to the maximum credible earthquake motion if they are designed to meet the performance cri-
(MCE), as defined by the Division of Mines and teria for the safety-evaluation ground motion following
Geology Open File Report 92-1 (CDMG, 1992). A the specifications in Sections 3, 4, 8, and 9. Therefore, an
probabilistically assessed ground motion is one explicit functional evaluation is not required for Ordi-
with a long return period (approximately nary Bridges. Both the functional evaluation and the
1000-2000 years). safety evaluation are required for Important Bridges.
The definition of ground motions for design earth-
For Important Bridges both methods shall be given quakes and the role of probabilistic and deterministic
consideration; however, the probabilistic evaluation methods are subjects of continuing study. Revisions to
shall be reviewed by a Caltrans-approved consensus the definition of ground motion levels, including updat-
group. For Ordinary Bridges, the motions shall be based ing or revision of the Division of Mines and Geology
only on the deterministic evaluation.

18 BDS Recommendations, Section 3: Loads ATC-32


(b) Functional-Evaluation Ground Motion: This is a Open File Report 92-1 (CDMG, 1992), maybe incorpo-
probabilistically assessed ground motion that has a rated in subsequently revised specifications.
60 percent probability of not being exceeded dur- Depending on the seismic activity of a given region,
ing the useful life of the bridge. The determination the deterministic and probabilistic assessments may be
of this event is to be reviewed by a Caltrans- different. For example, the deterministic ground motion
approved consensus group. assessments using the mean ARS spectra for the MCE in
the San Francisco Bay region correspond to return peri-
ods of about 300 to 400 years.
In the future, the role of the two methods in the
design of Ordinary Bridges will be reviewed by a Cal-
trans-approved consensus group. Service Levels and Damage Levels C3.21.2.3 Service Levels and Damage Levels

The following performance levels, expressed in terms of These specifications are intended to produce bridge
service levels and damage levels are defined as follows: designs consistent with these performance levels. How-
ever, the state-of-the-art in seismic design and the gen-
(a) Service Levels eral nature ofthis document are such that it is difficult to
guarantee that the performance levels will be achieved in
.... Immediate: Full access to normal traffic is avail- all cases. Designers should review the specific character-
able almost immediately following the earth- istics of their projects and make a judgment as to
quake. whether additional design features are necessary to
achieve the required performance.
.... Limited: Limited access (e.g., reduced lanes, light With respect to damage levels, the following behav-
emergency traffic) is possible within days of the ior of concrete structures is intended.
earthquake. Full service is restorable within
months. • Minimal Damage: Although minor inelastic
response may occur, postearthquake damage is lim-
(b) Damage Levels ited to narrow cracking in concrete. Permanent
deformations are not apparent.
.... Minimal Damage: Essentially elastic performance.
• Repairable Damage: Inelastic response may occur,
.... Repairable Damage: Damage that can be repaired resulting in concrete cracking, reinforcement yield,
with a minimum risk oHosing functionality.l and minor spalling of cover concrete. The extent of
damage should be sufficiently limited that the struc-
.... Significant Damage: A minimum risk of collapse, ture can be restored essentially to its pre-earthquake
but damage that would require closure to repair. condition without replacement of reinforcement or
replacement of structural members. Repair should
not require closure. Permanent offsets should be

• Signiftcant Damage: Although there is minimum risk

of collapse, permanent offsets may occur and dam-
age consisting of cracking, reinforcement yielding,
and major spalling of concrete may require closure
to repair. Partial or complete replacement may be
1. There is still a need to define repairable damage quantitatively in required in some cases.
terms of allowable material strain or some other measurable physical
characteristic for both steel and reinforced concrete members. To a
certain degree, this becomes a subjective decision since different indi-
viduals view repairability differently. This issue was considered by the
PEP, but the wide range of opinion prevented consensus from being
achieved within the limited time available for this issue. Some discus-
sion of allowable strain levels is included in the companion ATC-32-1
Resource Document.

ATC-32 BDS Recommendations, Section 3: Loads 19


3.21.3 Structural Action C3.21.3 Structural Action

For design purposes, each structure shall be categorized It is intended that the design engineer make explicit
according to its intended structural action under hori- selections regarding the intended structural perfor-
zontal seismic loading. Categories are defined in (a) mance, including locations of inelastic action that might
through (d) below. Important Bridges shall not be be implicit in the design. Furthermore, it is desirable
designed as Full-Ductility Structures. that Important Bridges and bridges having inelastic
action in locations where inspection is difficult be
(a) Full-Ductility Structure designed for limited ductility so that expected damage is
reduced in comparison with Ordinary Bridges in which
Under horizontal loading, a plastic mechanism is full inspection is feasible. It may also be desirable in
intended to develop. The plastic mechanism shall be some exceptional cases to design for elastic response or
defined clearly as part of the design. Intended yielding to use protective systems. Therefore, each structure
shall be restricted to locations that are readily accessible should be categorized according to the classifications (a)
for inspection following a design earthquake. Inelastic through (d) of Article 3.21.3.
action is intended to be restricted to flexural plastic Design force levels for Full-Ductility Structures,
hinges in columns and pier walls and inelastic soil Limited-Ductility Structures, and Elastic Structures are
deformation behind abutment walls and wingwalls. different in this specification (Article 3.21.11). The force
Details and proportions shall ensure large ductility reduction coefficients, Z, are smaller for Limited-Ductil-
capacity under load reversals without significant ity Structures than for Full-Ductility Structures. The
strength loss. force reduction coefficients for Elastic Structures should
be taken equal to unity.
(b) Limited-Ductility Structure Illustrations of Full-Ductility Structures and
Limited-Ductility Structures are given in Figure RC3-1.
Under horizontal loading, a plastic mechanism as
described for Full-Ductility Structures is intended to
(a) Full-Ductility Structure:
develop, but with reduced ductility demands. Yielding
- Ordinary bridge
may occur in areas that are not readily accessible for - Accessible plastic hinge location
inspection. Inelastic action is intended to be restricted
to flexural plastic hinges in columns and pier walls, and
inelastic soil deformation behind abutment walls and
wingwalls. Detailing and proportioning requirements
are the same as those required for Full-Ductility Struc-
(b) Limited-Ductility Structure:
tures. - Important bridge
- Accessible plastic hinge location
(c) Elastic Structure

This is a structure that is intended to remain elastic up

to the design load under combined vertical and hori-
(c) Limited-Ductility Structure:
- Important or ordinary bridge
(d) Structure with Protective Systems - Inaccessible plastic hinge location

This is a structure incorporating seismic isolation, pas-

sive energy dissipating devices, or other mechanical
devices to control seismic response. Under horizontal
loading, a plastic mechanism mayor may not be
intended to form. The occurrence of a plastic mecha-
Figure RC3-1 ILlustrations of fuLL ductility structures
nism shall be determined by analysis.
and Limited ductiLity structures.

Qualitative descriptions follow.

• Full-Ductility Structure: This is a structure that is

detailed for inelastic response. Its design is intended

20 BDS Recommendations, Section 3: Loads ATC-32


to take full advantage of the available ductility

capacity under the safety-evaluation earthquake
loading. Because the inelastic deformation demands
may approach deformation capacities, significant
damage, as described in the commentary to Article
3.21.2, is expected for the safety-evaluation earth-
quake. Given the expected damage, Full-Ductility
Structures should be restricted to Ordinary Bridges,
with inelastic response occurring in locations that
can be inspected and repaired readily following an
earthquake. Classes of structures that should not be
designed as Full-Ductility Structures include
Important Bridges and Ordinary Bridges with
inelastic response occurring below grade, where
inspection is difficult.

Extended pile shafts are often proportioned so that

flexural plastic hinges form below grade. Because
these cannot be inspected readily for damage, such
extended pile shaft structures should not be
designed as Full-Ductility Structures.

Proportions and details for Full-Ductility Structures

should ensure large ductility capacity under load
reversals and/or unidirectional pulse-type motions
from near-fault effects without significant strength
loss. The provisions of Sections 8 and 10 are
intended to satisfy this requirement.

Preferably, inelastic response will be restricted to

column and pier wall plastic hinges and inelastic
soil deformation behind abutment walls and wing-
walls. Inelastic response of superstructure elements
may cause damage that is difficult to inspect and
costly to repair. Furthermore, extensive plastic hing-
ing in superstructure elements tends to result in
residual deformations that may make the structure
unserviceable and unrepairable. Therefore, super-
structure hinging under design horizontal earth-
quake forces should be avoided, preferably by using
the capacity design approach of Article 3.21.14. A
moderate level of inelastic response under vertical
input motions may need to be accepted for eco-
nomic reasons.

• Limited-Ductility Structure: This is a structure that

is detailed for inelastic response in the same manner
as a Full-Ductility Structure, but whose design is
not intended to take full advantage of the available
ductility capacity under the safety-evaluation earth-
quake loading (see Article 3.21.11.). Because the
intended inelastic deformation demands do not
approach deformation capacities, significant dam-
age, as described in the commentary to Article

ATC-32 BDS Recommendations, Section 3: Loads 21


3.21.2, is not expected for the safety-evaluation

earthquake. Instead, repairable damage, as
described in the commentary to Article 3.21.2, is
expected. The reduced damage expectation makes
this designation appropriate for Important Bridges
and for any bridge in which inelastic response is
expected in an inaccessible location. Included in
this latter category are bridges with extended pile
shafts, where inelastic action is expected below

Ordinary Bridges with accessible plastic hinge loca-

tions, which normally are designed as Full-Ductility
Structures, may be designed as Limited-Ductility
Structures if it is desired to reduce the damage level.
This applies particularly to structures located in the
near-source region that could be subjected to ener-
getic, unidirectional pulse-type motions. Where
reduced damage is the objective, the design engi-
neer should consider checking deformation capaci-
ties using Inelastic Static Analysis, as described in
Article 3.21.7.

Because it is desirable that Limited-Ductility Struc-

tures have large reserve inelastic deformation capac-
ity, the proportioning and detailing requirements
are the same as those for Full-Ductility Structures.
Therefore, the standard details specified in Sections
8 and 10 may be used without modification. For
Important Bridges, the results of Inelastic Static
Analysis may indicate the need for enhanced details.

As with Full-Ductility Structures, it is preferred that

inelastic response be restricted to column and pier
wall plastic hinges, and to inelastic soil deformation
behind abutment walls and wingwalls. Superstruc-
ture hinging under design horizontal earthquake
forces should be avoided, preferably by using the
capacity design approach of Article 3.21.14. A mod-
erate level of inelastic response under vertical input
motions may need to be accepted for economic rea-

• Elastic Structure: This is a structure that is intended

to remain elastic under the safety-evaluation earth-
quake loading. For this reason, proportioning and
detailing requirements may be relaxed somewhat
from those required for Full-Ductility Structures
and Limited-Ductility Structures. However, because
inelastic response may occur for loadings exceeding
the design earthquake loading, the structure should
be provided with considerable ductility capacity,
and the capacity design approach should be used to
prevent brittle failure modes from occurring. Spe-

22 BDS Recommendations, Section 3: Loads ATC-32


cific design proportioning and detailing require-

ments for Elastic Structures have not been
established. Proportioning and detailing criteria
should be established for individual bridges. These
criteria should be independently reviewed.

It is emphasized that the ARS spectra used to estab-

lish the safety-evaluation loading represent mean
response spectra, rather than upper bound spectra.
Furthermore, it is unlikely that all potential earth-
quake sources are identified on current hazard
maps, and it is also unlikely that the potential
ground motions associated with known sources are
completely defined by current knowledge. There-
fore, response amplitudes exceeding the calculated
design values are not unexpected. This is why struc-
tures designated as Elastic Structures should be
designed to have some ductility capacity, even
though the intention is for effectively elastic struc-
tural performance.

• Structure with Protective Systems: The specification

allows for design of structures with protective sys-
tems, including seismic isolation devices, passive
energy dissipating devices, and other mechanical
devices to control seismic response. This document
does not include explicit provisions for structures
with protective systems. Criteria should be estab-
lished for individual bridges. These criteria should
be independently reviewed.

3.21.4 Structural Design Requirements C3.21.4 Structural Design Requirements

A structure may be designed by any approved method General Considerations

satisfying the requirements of statics and kinematics if
the structural actions of Article 3.21.3 are identified Articles 3.21.4 through 3.21.15 present design specifica-
clearly and if experimental evidence and analysis dem- tions that are considered adequate for design of typical
onstrate that the performance criteria of Article 3.21.2 structures intended to meet the performance criteria of
are satisfied. Article 3.21.2. These specifications were developed on
Except for sites close to potentially active seismic the basis of analysis, experimentation, and experience.
sources, sites with unusual geologic conditions, and Alternate approaches to design are permitted where it is
unusual structures, the requirements of the preceding demonstrated by experimental evidence and analysis
paragraph may be satisfied as follows: that the performance criteria of 3.21.2 are satisfied, and
where the alternate approaches are approved by a Cal-
(a) The initial design is to be based on either Equiva- trans-approved consensus group. Alternate approaches
lent Static Analysis (Article 3.21.5) or Elastic may be particularly desirable for unusual and complex
Dynamic Analysis (Article 3.21.6), as indicated in structures, for unusual geologic conditions, and for sites
Table R3-2. These analysis methods use linear elas- adjacent to active faults.
tic analysis of the bridge structure to determine the Design of structures near active faults requires con-
design displacements (Article 3.21.10) and design sideration of the effects of near-source ground motions.
forces (Article 3.21.11). Actions on restraining fea- At least four effects are prominent. The first of these is
tures (Article 3.21.12) are also determined. Stan- high effective peak ground acceleration, which is
dard member details are adopted, as specified in accounted for to some degree by the ARS curves.
Sections 8 and 10.

ATC-32 BDS Recommendations, Section 3: Loads 23


800 MAX· 8?6.c

-800 L-.l..-.l..-.L-.L-...l--'--'---L---L-L-L-L----'---l.---L--l---JI....-I.-L.-.I..-..l----'--_..J._~
I i
120 ".r . -128.9 l

MAX • ~32.5

-30 L-L.-.l...-.L-..llJ--..l--'--'---L---L-L-L-L----'---l.---L--l---JI.-I.-l...-.::':-.L--'---'--'-.:!c!--l.--L--L--L-::J
o 15 20 25 30

Figure RC3-2 North-South Ground motion recorded at

SyLmar, January 17, 1994.
The second is that near-source ground motions may
(b) Inelastic Dynamic Analysis (Article 3.21.8) may be contain large amounts of energy in long-duration, unidi-
used to refine design requirements determined in rectional pulses (Singh, 1981; Singh, 1985). The pulse-
Article 3.21.4(a), except that design displacements, type motions were first observed in near-fault recordings
design forces, and quantities of transverse rein- of the 1966 Parkfield earthquake. A more recent exam-
forcement shall not be less than 80 percent ofval- ple, from the 1994 Sylmar record of the Northridge
ues required by Elastic Dynamic Analysis. earthquake is given in Figure RC3-2. The magnitude-
and distance-dependent ARS curves for acceleration val-
(c) Either Inelastic Static Analysis (Article 3.21.7) or ues of 0.5, 0.6, and 0.7 g do contain some pulse-related
Inelastic Dynamic Analysis (Article 3.21.8) shall be long-period energy.
used to verify deformation capacity ofthe structure The third effect is that source directivity causes dif-
in the Safety Evaluation of the Important Bridge ferences between fault-normal and fault-parallel
Category having Configuration Type II (Table R3- motions at periods longer than one-half second (Singh,
2). Inelastic Static Analysis shall not be used as the 1981; Singh, 1985; Singh, 1995; and Somerville, et al.,
basis for reducing design quantities required by 1995). The magnitude- and distance-dependent ARS
Elastic Dynamic Analysis. curves are for average near-fault conditions and can be
modified to fault-normal and fault-parallel conditions
In Table R3-2, a Configuration Type I bridge is one using the factors proposed by Sommerville et al. (1995).
with continuous superstructure, well-balanced spans, The fourth effect is that the ratio ofvertical to hori-
supporting bents with approximately equal stiffness, zontal spectra at short periods is much larger than the
and insignificant vertical response. Bridges in this cate- commonly assumed ratio of two-thirds. Some additional
gory may include one- and two-span bridges with short discussion of the effects of near-source ground motions
spans, with small skew, and without intermediate on structural response is included in the commentary to
superstructure hinges. A Configuration Type II bridge is Article 3.21.10.
one with intermediate superstructure hinges, irregular
configuration, bents ofnonuniform stiffness, significant Description of Recommended Design Procedure
skew, or spans likely to be excited by vertical input
motion. Table R3-2 describes the minimum analysis require-
ments. The requirements vary with bridge category, con-
figuration type, and evaluation level (functional or
safety). The two bridge categories include Ordinary

24 8DS Recommendations, Section 3: Loads ATC-32


Table R3-2 Minimum Required Analysis Bridges and Important Bridges, as described in Article
3.21.2. Configuration Type I is intended to include
Functional Safety bridges with simple response characteristics, including
Evaluation Evaluation bridges with continuous superstructure, well-balanced
spans, supporting bents with approximately equal stiff-
Ordinary Bridge None Required AorB ness, and insignificant vertical response. Configuration
Type I Type II is intended to include bridges with more com-
Ordinary Bridge None Required B plex response characteristics that are unlikely to be rep-
Type II resented well by Equivalent Static Analysis, including
Important Bridge AorB AorB bridges with intermediate superstructure hinges, irregu-
Type I lar configuration, bents of nonuniform stiffness, signifi-
Important Bridge B BandC cant skew, or spans likely to be excited by vertical input
Type II motion. Bridges with such irregularities may also be
more vulnerable to near-fault motions.
Analysis method "A" is Equivalent Static Analysis (Arti- According to Table R3-2, the design of a bridge is to
cle 3.21.5); analysis method "B" is Elastic Dynamic be based on either Equivalent Static Analysis or Elastic
Analysis (Article 3.21.6); and analysis method "c" is Dynamic Analysis procedures, depending on the config-
Inelastic Static Analysis (Article 3.21.7) or Inelastic uration type and importance. According to these proce-
Dynamic Analysis (Article 3.21.8)1 dures, a linear elastic model of the bridge is analyzed for
the ARS or site-specific spectra to determine forces and
displacements. Design forces in plastic hinge regions are
taken equal to forces obtained from the elastic analysis
divided by the force reduction coefficient Z obtained
from Article 3.21.11. Forces outside plastic hinge regions,
and shears in plastic hinges, are determined using the
capacity design approach, as specified in Article 3.21.14.
Design displacements are taken as equal to the displace-
ments obtained from the elastic analysis factored by Rd,
as specified in Article 3.21.10.
For Important Bridges, Inelastic Static Analysis or
Inelastic Dynamic Analysis is required to verify the
deformation capacity of the structure for the safety-eval-
uation earthquake. In most cases, Inelastic Static Analy-
sis will be used. Where Inelastic Static Analysis or
Inelastic Dynamic Analysis indicate that the deformation
capacity is inadequate, the structure is to be modified (by
changing stiffness, strength, details, configuration, or
some other parameters) until all deficiencies are elimi-
nated. Where Inelastic Dynamic Analysis shows that the
bridge is overdesigned, design quantities may be reduced
by up to 20 percent, provided that the revised design is
adequate according to the Inelastic Dynamic Analysis.
Inelastic Static Analysis may not be used as the sole basis
for reducing design quantities.
Although inelastic analysis is required only for
Important Bridges, Inelastic Static Analysis is encour-
1. Caltrans has indicated its intent to use nonlinear static analysis in
conjunction with elastic dynamic analysis as a routine design proce- aged for all bridges because of the insight that it may
dure. The PEP endorses this approach since it gives the designer provide into the behavior of the structure and its design
greater insight into the potential seismic behavior of the bridge being requirements.
designed. Also, adopting a two-step design approach at this time will
make it easier to implement a true two-level design approach in the
future. A true two-level design approach, which many PEP members
feel is a worthwhile goal for Caltrans, would involve force design at the
functional-evaluation level and a displacement design check at the
safety-evaluation level.

ATC-32 8DS Recommendations, Section 3: Loads 25


Considerations Leading to Recommended Design


Several considerations influenced the recommendations

for minimum analysis requirements in Table R3-2. Some
of these are summarized below.
For one- or two-span bridges with monolithic abut-
ments, the superstructure is likely to respond effectively
as a rigid body. Furthermore, for structures in which the
skew is small, the lateral stiffness is likely to be con-
trolled by the abutments. A moderately sophisticated
Elastic Dynamic Analysis model will not provide much
insight into response beyond that which may be
obtained by Equivalent Static Analysis. For this reason,
Equivalent Static Analysis is specified as a minimum for
Type I bridges.
Bridge dynamic response is influenced by skew.
Therefore, for short-span bridges with significant skew
(exceeding 30 degrees), Equivalent Static Analysis is not
allowed. Elastic Dynamic Analysis is required instead.
For multi-span, continuous bridges having uniform
support conditions and small skew, static analysis is
likely to provide an adequate measure of expected
response. However, Elastic Dynamic Analysis is pre-
ferred for the purpose of assessing lateral displacements
and the effects of higher modes. The abutment stiffness
is likely to dominate response for many of these struc-
tures. Therefore, for Elastic Dynamic Analysis it is essen-
tial to correctly model the abutment stiffness and mass
For skewed bridges, curved bridges, and bridges
with intermediate superstructure hinges, static methods
are not likely to provide a realistic assessment of ex-
pected response. Elastic Dynamic Analysis including all
significant vibration modes is preferred as a minimum.
Variation of subsurface conditions along the length
of a bridge may result in significant variations in ground
motions at different supports. Similarly, long bridges are
subjected to spatial and temporal variations of ground
motion along the length even when subsurface condi-
tions are constant. Elastic Dynamic Analysis is preferred
for analyzing such structures. The commentary to Arti-
cle 3.21.6 describes a simple and approximate approach
to deal with these problems.
For long-span bridges and bridges with outriggers
or C-bents, vertical response may be significant. In the
design of such bridges, vertical response should be con-
sidered directly in a response-spectrum analysis of an
elastic model that includes all significant vibration
modes. Bridges with unbalanced spans may be prone to
global torsional responses that result in increased flex-
ural deformation demands on some elements. Elastic
Dynamic Analysis is more appropriate for these struc-
tures than is Equivalent Static Analysis.

26 8DS Recommendations, Section 3: Loads ATC-32


Bridge response to the safety-evaluation earthquake

and, in particular, to the near-source motions of this
event is likely to involve significant levels of inelastic
response. Linear elastic analysis provides only an
approximation of the expected response, and could be
seriously in error in certain cases. This is especially true
for skewed bridges, bridges with in-span expansion
joints, and highly irregular bridges. Inelastic analysis
methods are likely to provide an improved representa-
tion of actual inelastic response. Therefore, for Impor-
tant Bridges of configuration Type II it is required to
carry out either an Inelastic Static Analysis or Inelastic
Dynamic Analysis to check that the final design is ade-
quate. Inelastic response analysis is likely to provide an
improved set of design values for Ordinary Bridges as
well; for these bridges it is encouraged but not required.
While inelastic analysis is likely to better represent
response than linear elastic analysis, the engineer must
understand that all analysis procedures are approximate.
The inherent uncertainties in the specification of the
ground motion, behavior ofthe soil and foundation, and
the anticipated behavior of the structural components
make the analysis results uncertain. Although current
analysis and design procedures do not explicitly account
for the uncertainty, the engineer must recognize it as a
fundamental characteristic of the design problem.
Assumptions and estimates in the model must be judged
against the uncertainty in the complete problem. Fur-
thermore, design proportions and details must be judged
against these uncertainties as well, with final selections
making allowance for the possibility that the analysis
results might err nonconservatively. Experience with
modern bridges in California indicates that current
strength requirements and detailing practices produce
structures that perform adequately despite all the uncer-
tainties in the design environment. It is for this reason
that the specification does not permit more than a 20-
percent reduction in the standard requirements even
when Inelastic Dynamic Analysis indicates a greater
reduction is warranted.
When Inelastic Static Analysis is carried out, the tar-
get displacement is commonly gauged from results of a
linear elastic analysis model. Several uncertainties are
therefore introduced, including ground motion repre-
sentation, displacement estimation, and phasing of
orthogonal responses. For this reason, the specification
does not permit reductions from the standard require-
ments based solely on Inelastic Static Analysis.
The specifications were written to allow for both a
functional evaluation and a safety evaluation for any
bridge, although the functional evaluation is not
required for Ordinary Bridges. For the safety-evaluation
earthquake, the designer must consider the following:

ATC-32 80S Recommendations, Section 3: Loads 27


• Overall response amplitudes must be controlled to

preclude pounding between adjacent structures of
different height (pounding at expansion joints is
considered acceptable and pounding of adjacent
structures of equal height may be acceptable) and
instability due to P-Ll effects.

• Local plastic deformation demands in members

selected for inelastic response must be less than
plastic deformation capacities. For Ordinary
Bridges, the specifications do not require that any
damage be "repairable." For Important Bridges,
replacement of a structure may involve delays. For
these structures, the plastic demands must be con-
trolled sufficiently so that members can be
"restored" to near their original condition.

• Other members must be protected from inelastic

response, and nonductile failure modes should be
avoided in all members.

For the functional-evaluation earthquake, the anal-

ysis is applied to a structure for which a safety evaluation
would also be required. Therefore, it would not be nec-
essary to check pounding (between adjacent structures),
instability, and demands in capacity-protected mem-
bers. Instead, the analysis would be limited to checking
demands at locations where plastic hinges are allowed to
form in the safety-evaluation earthquake. Some inelastic
response is acceptable so long as damage requiring repair
is avoided.

3.21.5 Equivalent Static Analysis (3.21.5 Equivalent Static Analysis Application of Lateral Loads The specification permits use of Equivalent Static Analy-
sis for one- and two-span continuous structures with
Seismic load shall be assumed as an equivalent static small skew, even though it is recognized that dynamic
horizontal force applied to individual frames. The total response will occur during an earthquake. The rationale
applied force shall be equal to the product of ARS and is that, in most cases, moderately sophisticated dynamic
lv, but not less than 0.4W. The lateral force may be analysis will not provide significant additional insight
applied at the vertical center of mass, and shall be dis- into behavior, and will not in general result in additional
tributed in the horizontal plane in proportion either safety. The engineer should recognize that the Equiva-
with the mass distribution or with the product of the lent Static Analysis method is best suited for structures
mass distribution and displaced shape. with well-balanced spans and supporting elements of
approximately equal stiffness. For these structures,
response is primarily in a single mode and the lateral
force distribution is simply defined. For unbalanced sys-
tems, or systems in which vertical accelerations may be
significant, the Elastic Dynamic Analysis method of Arti-
cle 3.21.6 should be used.
Two options in applying Equivalent Static Analysis
are allowed. In the first option, lateral load is distributed
to the superstructure in proportion with the mass distri-
bution. This distribution is consistent with the assump-

28 BDS Recommendations, Section 3: Loads ATC-32


tion that the superstructure displaces as a rigid body. In

reality, the superstructure deforms under the action of
lateral load, resulting in lateral inertial forces that are not
precisely in proportion with the mass distribution. The
second method, which is more theoretically rigorous,
specifies that the lateral load should be applied in pro-
portion with the product of the mass distribution and
the displaced shape. A simple means of employing the
second method is as follows: (1) Apply lateral load to the
superstructure in proportion with the mass distribution,
and monitor the resulting displaced shape of the super-
structure. (2) Redefine the lateral load to be in propor-
tion with the product of superstructure mass and the
displaced shape obtained from step (1). This second
approach is recommended in the AASHTO guide speci-
fications. Although it is more theoretically rigorous, the
AASHTO approach is not required as part of this specifi-
cation because it is believed that the added rigor does not
add appreciably to the design outcome for this class of
The minimum lateral force ofOAWis unchanged
from the value contained in the current Caltrans Bridge
Design Specifications. Seismic Loading C3.21.5.2 Seismic Loading

Five-percent-damped elastic ARS response curves from Some aspects of the ground motion representation are
Figures R3-1 through R3-12, or from equivalent site- presented below.
specific elastic response spectra, shall be used as the
static loading. Standard soil profiles in Figures R3-1 Ground Motion Representation
through R3-12 shall be as defined in Table R3-3.
The new procedure for developing seismic loading
maintains the deterministic ARS approach.
A: Peak Rock Acceleration. This procedure still uses
the deterministic A values obtained from the CDMG
Open File Report 92-1 entitled "Peak Acceleration from
Maximum Credible Earthquakes in California Rock and
Stiff Soil Sites." The peak acceleration values reported in
these maps are mean values obtained using the CDMG
(1992) attenuation relationships. It is understood that
Caltrans is in the process of updating the acceleration
attenuation relationships in order to produce a new map
of peak acceleration values.
R: Rock Spectra. The existing Caltrans R curve for
rock (depth of alluvium 0-10 feet) are magnitude-inde-
pendent. The new rock spectra R have been modified to
become magnitude- and distance-dependent. The spec-
tral shapes for acceleration values between 0.1 and O.7g
(in 0.1 g increments) for three magnitude groups
(6.5±0.25, 7.25±0.25, and 8.0±0.25) are shown in Fig-
ures R3-1 through R3-3. These spectra are for Califor-
nia-type rock and correspond to NEHRP Soil Profile
Type B. These curves are a reasonable upper bound of

ATC-32 BOS Recommendations, Section 3: Loads 29

2.0 r--,--..,.--....---....---..,...-.,---.,---..,-----r---,--,...--.,......--r---r--...,....-....,

MAGNITUDE: 6.5± 0.25

-< 1.2
;i 0.8


2 3 4
PERIOD (sec)

50 I I I I I I I I I

f- -
40 I-- -

f- -
30 I-- -
f- -
-< 20 I-- -
f- -
10 I--
0.50 -


o 2 3 4
PERIOD (sec)

Figure R3-1 Proposed ARS curves for rock (M = 6.50 ± 0.25)

30 BDS Recommendations, Section 3: Loads ATC-32

2.0 ....---..,..--....,--""'"T""---.-----,--,..--..,---.--......,....-----r-----,,...--,....--....,--""'"T""---,----,



1.6 MAGNITUDE: 7.25± 0.25

« 1.2
~ 0.8


2 3 4
PERIOD (sec)

50 ....---..,..--....,--""'"T""---.-----,--,..--..,---.--......,....-----r-----,--,....--....,--""'"T""---,----,

w 30
Q. 0.79
« 20

e::: 0.69
10 0.49
0 2 3 4
PERIOD (sec)

Figure R3-2 Proposed ARS curves for rock (M = 7.25 ± 0.25)

ATC-32 BDS Recommendations, Section 3: Loads 31

2.0 .--.,...--.,...--...,...--...,...---r---...,...--..,..--..,..--.,...--.,...---,----,----,----,---..,----,



MAGNITUDE: 8.0± 0.25

<: 1.2
<i 0.8


o 2 3 4
PERIOD (sec)

50 .---.,...--.,...--..,..--..,..---r---..,..--..,..--.,...--.,...---,----,----,----,---..,--..,--..,

z 0.79
w 30

<: 20



o 2 4
PERIOD (sec)

Figure R3-3 Proposed ARS curves for rock (M = 8.0 ± 0.25)

32 BDS Recommendations, Section 3: Loads ATC-32

2.0 ,----r---,----,.--,--,-----r-----,--,...---r-----r-----,.--..,...---r----r--r-----,


1.6 MAGNITUDE: 6.5± 0.25
Note: Peak ground acceleration values
z not in parentheses are for rock (Soil
0 Profile Type 8) and peak ground
« 1.2 acceleration values· in parentheses
w are for Soil Profile Type C.
« 0.8


0 2 3 4
PERIOD (sec)

50 I I I I

f- -

40 I-- -

f- -
30 - -
- -
« 20
-l - -
- 0.6g (0.6g) -
0.5g (0.5g)
10 -
0.3g (0.33g)


- I I
0.2g (0.24g)

0.10 (0.120)
PERIOD (sec)

Figure R3-4 Proposed ARS curves for soil type C (M = 6.50 ± 0.25)

ATC-32 BDS Recommendations, Section 3: Loads 33

2.0 .----...,.....--...,...---,---,.---.,..--...,..---,---,.---.,..-----r---,....--...,.....--...,..---,---,.---,


1.6 MAGNITUDE: 7.25± 0.25


Note: Peak ground acceleration values

0 not in parentheses are for rock (Soil
f= Profile Type B) and peak ground
<l: 1.2
a:: acceleration values in parentheses
w are for Soil Profile Type C.
<l: 0.8


2 3 4
PERIOD (sec)

50 .----.,..--...,..----r--,.---.,..--.....,....----r--,.---.,..--......,...--,....--.,..--.....,....----r--,.---.,



z 0.7g (0.7g)
::2 30
0 0.6g (0.6g)
--l 20
u 0.5g (0.5g)
0.4g (O.4g)
10 0.3g (0.33g)
0.2g (0.24g)
0.1g (0.12g)

0 2 3 4
PERIOD (sec)

Figure R3-5 Proposed ARS curves for soil type C (M = 7.25 ± 0.25)

34 8DS Recommendations, Section 3: Loads ATC-32

2.0 ,..--...,---r--,--;---.,.---r----r--,....--.,...---,---;---.,.---.--....,.--,........---,


1.6 MAGNITUDE: 8.0± 0.25
"--' Note: Peak ground acceleration values
z not in parentheses are for rock (Soil
i= Profile Type B) and peak ground
« 1.2 acceleration values in parentheses
w are for Soil Profile Type C.
« 0.8


2 3 4
PERIOD (sec)

50 .----,........-,........-..,.--,--,.--.....,.--,---,----,---,;---,........-,........-..,.--..,---.---,


~ 30
;i 20
0.29 (0.249)
0.19 (0.129)

o 2 3 4
PERIOD (sec)

Figure R3-6 Proposed ARS curves for soil type C (M = 8.0 ± 0.25)

ATC-32 BDS Recommendations, Section 3: Loads 35

2.0 ,..--..,..---;---r----r--,..--..,..--...,.---,---r----r--,..--.,.--...,.---,---r----r


MAGNITUDE: 6.5± 0.25
Q'\ Note: Peak ground acceleration values
not in parentheses are for rock (Soil
Z Profile Type B) and peak ground
;::: acceleration values in parentheses
« 1.2 are for Soil Profile Type D.
« 0.8


0 2 3
PERIOD (sec)

50 ,...--...,......--,-----,---r---r----,-----,r---..,..---r----r--,..----r---,-----r--,..--,


w 30
« 20

f- 0.6g
w 0.5g
10 0.3

2 3 4
PERIOD (sec)

Figure R3-7 Proposed ARS curves for soil type D (M = 6.50 ± 0.25)

36 BDS Recommendations, Section 3: Loads ATC-32

2.0 .---.,..----,-----,--,....--.,....---,-----,--,....---,....---.---,----r---,-----r--,-----.


1.6 MAGNITUDE: 7.25± 0.25
C;; Note: Peak ground acceleration values

z not in parentheses are for rock (Soil

0 Profile Type B) and peak ground
<t 1.2 acceleration values in parentheses
0:: are for Soil Profile Type D.
<t 0.8


2 3 4
PERIOD (sec)

50 r----r--...,....----,.--,..---r---,-----,.--,---..,.----,-----..,--,....---,....---,-----..,---,


0.79 (0.79)

w 30
0... 0.69 (0.69)
<t 20
0:: 0.59 (0.59)
w 0.49 (0.449)
0.39 (0. 369)
0.29 (0. 28 9)

0.19 (0. 169

0 2 3 4
PERIOD (sec)

Figure R3-8 Proposed ARS curves for soil type 0 (M = 7.25 ± 0.25)

ATC-32 BDS Recommendations, Section 3: Loads 37

2.0 .---.,..--...,......--,..-.....,...---,.--,..--.,..---,----,..-.....,...---,--,--...,......-...,..-.....,---,


1.6 MAGNITUDE: 8.0± 0.25
Note: Peak ground acceleration values
nat in parentheses are for rock (Soil
z Profile Type B) and peak ground
i= acceleration values in parentheses
« 1.2 are for Soil Profile Type D.
~ 0.8

0.0 2 4
PERIOD (sec)

50 ,..---r---r---r---r-----r-----r-----r-----r-----r----.,.--.,.--.,.--.,.--.,...--.,...----,

::;; 30
« 20


OL-_~:::::::L__..L. _ _L.._ _.L_ ___l._ _L...-_..I.-_...J-_....J..._---J._--l_ _.l..__..l-_....J..._ __1

o 2 3 4
PERIOD (sec)

Figure R3-9 Proposed ARS curves for soil type D (M = 8.0 ± 0.25)

38 BDS Recommendations, Section 3: Loads ATC-32

2.0 .---,...------,----,--,--.-......,.-.....,.-.....,.--,---,-,--,--,--,----r--,

MAGNITUDE: 6.5± 0.25
Note: Peak ground acceleration values
not in parentheses are for rock (Soil
1.2 Profile Type B) and peak ground
acceleration values in parentheses
...J are for Soil Profile Type E.
w 0.39 & 0.49 (0.369)

0.0 ot_.L_-L_-l__L====::::~==2C=~~~~S~~f3~SE~E~~~a4
PERIOD (sec)



20 0.49 (0.369)
u 0.39 (0.369)

10 0.29 (0.349)

0.19 (0.259)

00 2 3 4
PERIOD (sec)

Figure R3-10 Proposed ARS curves for soil type E (M = 6.5 ± 0.25)

ATC-32 BDS Recommendations, Section 3: Loads 39

2.0 r--,...-..,...---r---r--r--.,--...,..---r--,-----;--,---.,...--,---r--,---,


1.6 MAGNITUDE: 7.25± 0.25
Note: Peak ground acceleration values
z not in parentheses are for rock (Soil
o Profile Type B) and peak ground
F= acceleration values in parentheses
« 1.2
e::: are for Soil Profil e Type E.
<i. 0.8


o 2 3 4
PERIOD (sec)

50 r--,---.,--.,...-..,.---,--,---,--,----;--r--,--...,.--....---,--...,...---,


w 30
--' 20
O.4g (0.36g)
t- 0.3g (0.36g)
V1 0.2g (0.34g)

0.' (0.25g)

3 4

Figure R3-11 Proposed ARS curves for soil type E (M = 7.25 ± 0.25)

40 BDS Recommendations, Section 3: Loads ATC-32

2.0 ,..--...,..--...,----r--,..---.--"""T'"----,.--..,.--..,..--.....,..----,..--..,.--..,..--.....,..----,----,


MAGNITUDE: 8.0± 0.25
Note: Peak ground acceleration values
:§ not in parentheses are for rack (Soil
z Profile Type B) and peak ground
f= acceleration values in parentheses
« 1.2 are for Soil Profile Type E.
« 0.8


0.0 L..-_..J..__....l..._.......l._ _" - - _ - ' - _........_ - - '_ _.l.-_...J..._--J.._--''--_..l-_...J..._--J.._--'_---'

o 2 3 4
PERIOD (sec)

50 ...--...,......-"""T'"----.,.---r---...,------r-----,..--...,..--...,----r--,..--...,..---,----r--,..----,


w 0.49 (0.369)
w 30
0... 0.39 (0.369)
0.29 (0.349)
« 20

0.19(0. 259)

OL..-_e=:;;;"".........l__........L_ _ _ _.l.__....l..._.......l._ _l..__..J..__.......l__

.l..-_~_.....l.._..-J __l._ _ '--___I
o 2 3 4
PERIOD (sec)

Figure R3-12 Proposed ARS curves for soil type E (M = 8.0 ± 0.25)

ATC-32 8DS Recommendations, Section 3: Loads 41


the mean spectral values obtained using various spectral

Table R3-3 Soil Profile Types a S: Site Modification Factors. New S factors have
Soil been developed using the soil profile types and soil
Profile amplification factors developed at a workshop on how
Type Soil Profile Description site response should be reflected in seismic code provi-
sions (Rinne, 1994; Martin and Dobry, 1994). Table
A Hard rock with measured shear wave velocity RC3-1 summarizes the new soil profile types, which are
vs > 5,000 ft/s (1,500 m/s) the same as those adopted in the 1994 NEHRP Provi-
sions (BSSC, 1994).
B Rock with shear wave velocity 2,500 < vs ~
The following steps have been recommended for
5,000 ft/s (760 mls < vs~ 1,500 m/s) classifying a site according to these new soil profile types:
C Very dense soil and soft rock with shear wave
velocity 1,200 ft/s < V s ~ 2,500 ft/s (360 mls < 1. Determine whether the site fits into one of the four
Vs ~760 m/s) or with either standard penetra- Type F categories that require site-specific evalua-
tion resistance N > 50 or undrained shear tion. If so, classify the site as a type F profile and
strength su~ 2,000 psf (100 kPa) carry out the required evaluation.
D Stiff soil with shear wave velocity 600 ft/s < v,
.s;. 1,200 ftls (180 mls <vs.s;. 360 m/s) or with 2. Determine whether there is a soft clay layer that is
either standard penetration resistance 15 .s;. N over ten feet (three meters) thick. Soft clay is
.s;. 50, or undrained shear strength 1,000 psf.s;. defined as having an undrained shear strength
su < 2,000 psf (50 kPa.s;. Su < 100kPa) Su < 500 psf (25kPa), moisture content ~40 percent,

E A soil profile with shear wave velocity Vs < 600 and plasticity index PI > 20. If this criterion is satis-
ftl s (180ml s) or any profile with more than 10 fied, classify the site as a type E profile.
ft (3 m) of soft clay, defined as soil with plas-
ticity index PI > 20, water content w ~ 40 per- 3. If the site cannot be classified as soil profile type E
cent, and undrained shear strength Su < 500 or F by steps 1 and 2, categorize the site by perform-
psf(25 kPa) ing one or more of the following three calculations
F Soils requiring site-specific evaluation: and using the results to select the appropriate soil
profile type from Table RC3-1 (see the definitions
1. Soils vulnerable to potential failure or that follow for details):
collapse under seismic loading; ie., lique-
fiable soils, quick and highly sensitive a. calculate vfor the top 100 feet (30 meters)
clays, collapsible weakly-cemented soils
2. Peat andlor highly organic clay layers b. calculate N for the top 100 feet (30 meters)
more than 10 ft (3 m) thick
c. calculate Nch for cohesionless soil layers
3. Very high-plasticity clay (PI> 75) layers (PI < 20) and Su for cohesive soil layers
more than 25 ft (8 m) thick (PI ~ 20) in the top 100 feet (30 meters)
4. Soft-to-medium clay layers more than
120 ft (36 m) thick Definitions
a. The soil profile types shall be established through The defmitions given below apply to the upper 100 feet
properly substantiated geotechnical data.
(30 meters) of the site profile. Profiles containing dis-
tinctly different soil layers shall be subdivided into lay-
ers, each designated by a number that ranges from 1 (at
the top) to n (at the bottom), where there are a total of n
layers in the upper 100 feet (30 meters). The symbol i in
the following expressions refers to anyone of the layers
between 1 and n.

42 BDS Recommendations, Section 3: Loads ATC-32


s is the generalized shear wave velocity for the
upper 100 feet of the soil profile defined as

Table RC3-1 Soil Profile Type Classification

Soil Profile Type vs Nor Nch

E < 600 ftis « 180 m/s) < 15 < 1,000 psf ( < 50 kPa)
D 600-1,200 ft/s 15-50 1,000-1,999 psf
(180-360 m/s) (50-99 kPa)
C 1,200-2,500 ft/s >50 L2,OOO psf
(360-760 m/s) (100 kPa)

d. ~
if'I 1
vs =n- -d.
i = 1 vsi

where.r d i is equal to 100 feet (30 meters)
1 = 1

Vsi is the shear wave velocity oflayer i in feet per second

(meters per second).

dj is the thickness of any layer i between 0 and 100 feet

(30 meters)
N is the generalized standard penetration resistance of all
soils in the upper 100 feet (30 meters) of the soil profile
defined as

- ~ d.1
N - 1
- --;;cr.

i= IN i

where.r d i is equal to 100 feet (30 meters)
1= 1

N i is the standard penetration resistance oflayer i (ASTM

DI586-84), not to exceed 100 blows per foot, as
directly measured in the field without corrections.

Nch is the generalized standard penetration resistance

for only the cohesionless soil layers ofthe soil profile
defined as


ATC-32 BDS Recommendations, Section 3: Loads 43


ds is the total thickness of cohesionless soil layers in the

top 100 feet (30 meters)
n d·
i= INi
incluges cohesionless soil layers only when calculating

su is the generalized undrained shear strength of the

upper 100 feet of the soil profile defined as

i = 1 Sui


de is the total thickness (l00 - d s ) of cohesive soil layers

in the top 100 feet (30 meters).

Sui is the undrained shear strength oflayer i in psf (kPa),

not to exceed 5,000 psf (250 kPa), as determined by
ASTM 2166-91 or D2850-87.

n d.
I, --.: includes cohesive soil layers only
i = 1 Sui

The plasticity index PI is determined according to

ASTM D4318-93.
Moisture content is determined according to ASTM
}lote: if calculation c as defined above is used and
the Neh and sui criteria differ, select the site classifica-
tion with the softer soils (for example, E rather than D).
The shear wave velocity for rock, Soil Profile Type
B, shall be either measured on-site or estimated by a geo-
technical engineer or engineering geologist/seismologist
for competent rock with moderate fracturing and
weathering. Softer and more highly fractured and weath-
ered rock shall either be measured for shear wave veloc-
ity or classified as profile type C.
A classification as hard rock, profile type A, shall be
supported by shear wave velocity measurements either
on-site or on profiles of the same rock type in the same
formation with an equal or greater degree of weathering
and fracturing. Where hard rock conditions are know to
be continuous to a depth of 100 feet (30 meters), surficial
shear wave velocity may be extrapolated to assess s • v
A site shall not be classified as a rock site (profile
types A or B) ifthere is more than ten feet (three meters)
of soil between the rock surface and the bottom of the
spread footing or mat foundation.

44 8DS Recommendations, Section 3: Loads ATC-32


Profile type A corresponds to hard rock conditions

found most commonly in the northeastern United
States. Rocks corresponding to profile type B are more
comparable to the rock types found in California. The R
curves shown in Figures R3-1 through R3-3 correspond
to California-type rock (profile type B). Curves for pro-
file type A are not provided.
The values of the soil amplification factors Fa and Fv
for the acceleration- and velocity-controlled parts of the
spectrum for different levels of shaking for the various
soil profile types are given in Tables RC3-2 and RC3-3,
Table RC3-4 shows which figures display the correct
ARS curves for profile types C, D, and E over three dif-
ferent magnitude ranges.

Table RC3-2 Values of Site-Amplification Factor Fa as a Function of Soil Profile Types and Shaking Intensity
Shaking Levels"

Soil Profile Type Aa~O.lg A a = O.2g A a = O.3g A a = OAg Aa'LO.5g

A 0.8 0.8 0.8 0.8 0.8

B 1.0 1.0 1.0 1.0 1.0
C 1.2 1.2 1.1 1.0 1.0
D 1.6 1.4 1.2 1.1 1.0
E 2.5 1.7 1.2 0.9
a. Site-specific geotechnical investigations and dynamic site response analysis shall be performed to determine the seis-
mic coefficient for profile types F and E when effective peak acceleration-related accelerations A a exceed 0.4.

Table RC3-3 Values of Site-Amplification Factor Fv as a Function of Soil Profile Types and Shaking Intensity
Shaking Levels"

Soil Profile Type A v ::; O.lg A v =O.2g A v =O.3g A v =OAg A v ;;:: O.5g

A 0.8 0.8 0.8 0.8 0.8

B 1.0 1.0 1.0 1.0 1.0
C 1.7 1.6 1.5 1.4 1.3
D 2.4 2.0 1.8 1.6 1.5
E 3.5 3.2 2.8 2.4
a. Site-specific geotechnical investigations and dynamic site response analysis shall be performed to determine the seis-
mic coefficient for profile types F and E when effective peak acceleration-related accelerations A a exceed 0.4.

ATC-32 80S Recommendations, Section 3: Loads 45


Table RC3-4 Figure Numbers of Appropriate Design Spectra

Soil Profile Type 6.5 ± 0.25 7,25± 0.25 8.0± 0.25

C R3-4 R3-5 R3-6

D R3-7 R3-8 R3-9
E R3-10 R3-11 R3-12

The deterministic assessment of ground motion for

the safety-evaluation earthquake using the ARS curves
described above is about the mean values. Reasonable
mean-pIus-one sigma values can be obtained by multi-
plying the mean spectral ordinates by a factor of 1.5.
Site-specific equal-hazard spectra for a desired
return period can be generated using the seismic hazard
analysis for the appropriate region. Distribution of Seismic Force

The distribution of the seismic force to individual

members shall reflect the stiffness of the superstructure
and supporting bents or piers, including restraint at the
abutments. Combination of Effects

Responses in multiple directions shall be determined

according to Article 3.21.9.

3.21.6 Elastic Dynamic Analysis (3.21.6 Elastic Dynamic Analysis General C3.21.6.1 General

Seismic response shall be determined as structure Seismic design ofmost bridge structures will normally be
displacements and individual member forces using carried out using linear Elastic Dynamic Analysis. For
dynamic analysis techniques considering stiffness, safety evaluation, linear analysis of response will usually
damping, and mass of the structure and soil. indicate stress above the limits oflinear behavior in some
elements. The presence of such stress in the linear elastic
model signals that nonlinear response is likely to occur.
As a structure responds in the nonlinear range, effective
member stiffness values change, internal forces redis-
tribute, energy dissipation characteristics vary, and over-
all response amplitudes deviate from those indicated by
an elastic analysis. When nonlinear response is indicated
by a linear response analysis, the engineer should recog-
nize that the results of linear response analysis are not
wholly correct and must be interpreted to achieve useful
and reliable conclusions for design.
Sources of nonlinear response include: the soil, the
behavior of which is strongly dependent on the strain
level; cyclic yielding of structural components; opening
and closing (pounding) of decks at expansion joints;
engagement, yielding, and'release of restrainers; and the
complex behavior of abutments. The extent of the non-

46 BDS Recommendations, Section 3: Loads ATC-32


linear behavior that is permitted depends on the bridge

category (Ordinary Bridge versus Important Bridge),
accessibility of the inelastic region for inspection, and the
design earthquake (safety-evaluation ground motion or
functional-evaluation ground motion).
Although the actual response may be nonlinear, a
linear analysis model can provide useful insights into
expected behavior, including indications of the total dis-
placement amplitude and local ductility demands.
Expected displacement amplitude is discussed in the
commentary of Section 3.21.10. Expected ductility
demands are discussed in the commentary to Section
3.21.11. Analysis Model C3.21.6.2 Analysis Model

A modal spectral analysis based on the application of a The bridge analysis is normally carried out using modal
response spectrum of ground acceleration to a lumped- spectral analysis of a linear model of the bridge. If
mass space frame model of the structure is recom- response history analysis is used, it is recommended that
mended. The number of degrees of freedom and the several ground motions are used. Specific recommenda-
number of modes considered in the analysis shall be tions for response-history analysis are given in Section
sufficient to include all critical response modes.
A linear model of a bridge is constructed using the
finite element method, in which the assemblage of the
elements represents the characteristics of the system.
Most earthquake analyses ofbridges can be performed
using models consisting of three-dimensional frame ele-
ments, or so-called "stick" models. Columns and bent
caps usually can be adequately modeled by frame ele-
ments; there may be significant approximations in mod-
eling ofbridge decks or pier walls by equivalent frame
elements. l
The analysis model must adequately represent the
mass of the bridge. The decks and supporting girders are
usually the largest mass in a typical bridge. The mass of
other structural components such as bents, piers, and
footings should be included in the model, but they are
usually a small percentage of the total mass. Nonstruc-
tural mass, including pavement topping, railings, side-
walks, catwalks, and signage, should be included. It is
not necessary to include the mass oflive loads. Studies of
short bridge overcrossing response during recent earth-
quakes indicate that response may be driven largely by
movement at the abutment; the analysis model must
attempt to represent not only the abutment stiffness but
the abutment mass (Werner, 1993).

1. Simplified abutment modeling (including the effect of approach

fills) by using elastic analysis techniques (ie., equivalent stiffness, mass,
and damping) is a subject that is not fully understood and requires
additional study. The latest state-of-the-art related to this subject is dis-
cussed further in the Commentary to proposed changes to Section 4 of
the recommended BDS and Chapter 4 of the companion ATC-32-1
Resource Document.

ATC-32 80S Recommendations, Section 3: Loads 47


The mass distribution in a stick model is determined

by the number of elements used to represent individual
components. The mass distribution must be able to rep-
resent the vibration modes of the components that con-
tribute to the earthquake response of the system. It is
usually sufficient to model a single span using five ele-
ments. Such a model captures the rigid-body modes of
the span, and the first symmetric and antisymmetric
modes in the vertical and transverse directions. Many
vibration modes may contribute significantly to the
response oflong spans, in which case more elements
should be used to capture more modes of the span.
A more refined judgment about the mass discretiza-
tion can be based on an estimate of the vibration periods
of the span. The vibration period (in seconds) of a single
span can be expressed as:

where L is the span length, m is the mass per unit length,

and EI is the flexural rigidity of the span. Depending on
the support conditions at the ends of the span, the coef-
ficient Pi is bounded by the simply supported case and
the fixed-support case for each mode of vibration, as
given in Table RC3-5.

Table RC3-S Coefficient Pi

Pi for Simple Supports Pi for Fixed Supports

Mode Type of Mode (upper bound on period) (lower bound on period)

1 first symmetric 2.00 1.3

2 first antisymmetric 1.00 0.80
3 second symmetric 0.67 0.57
4 second antisymmetric 0.50 0.44
higher modes alternate symmetric and 2/ i 2
antisymmetric (2i + 1)

Five elements per span are sufficient for a good rep-

resentation ofthe first three vibration modes of a span. If
the higher vibration periods of a span are within the
range of the amplified-acceleration portion of the earth-
quake response spectrum or near the lower vibration
periods of the entire bridge, it is necessary to include
more elements to capture the higher vibration modes. If
the contribution of the ith vibration mode is to be
included in the analysis, the span should be modeled by
2i -1 elements over the length of the span.
Along with translational inertias, the rotational
moment of inertia of the superstructure should be

48 80S Recommendations, Section 3: Loads ATC-32


lumped at the nodes, particularly for spans supported by

single-column bents and C-bents.
A column can generally be modeled with a single
element because the mass participation ofthe columns is
relatively small compared with the deck, and column
vibration modes have short vibration periods. For col-
umns taller than 15 meters, several elements should be
used at intervals no greater than eight meters. Cap
beams and outriggers themselves have short vibration
periods compared with the bridge, so one element is
usually sufficient to capture the rigid-body modes of
these elements. However, more elements may be neces-
sary to represent the stiffness and connectivity of these
Other large masses are the pile caps, which should
be included as nodal masses because they are generally
assumed to be rigid for earthquake analysis. ModeL Stiffness C3.21.6.3 ModeL Stiffness

The structural model shall include the effects of crack- Representing the linearized structural stiffness of a com-
ing on stiffness of reinforced concrete members and plex bridge system responding nonlinearly involves sig-
shall include the restraint of the surrounding soil. nificant approximations. Two general approaches are
recognized here. The first approach is to construct a lin-
earized model whose stiffness approximates the stiffness
of the bridge as it approaches the displacement at which
significant yielding occurs. The second approach is to
construct a linearized model with stiffness that approxi-
mates the secant stiffness of the bridge at the maximum
anticipated displacement level. Conventional Caltrans
practice is to use a combination ofthese two approaches,
with framing member stiffness taken equal to the stiff-
ness near yield and abutment stiffness taken equal to a
secant value. Only this approach is described here.
Where seismic isolation or other protective systems are
used, the effective stiffness values should be derived con-
sidering the characteristics of the system, and the results
should be reviewed independently.

Framing Member Stiffness

An objective of the analysis is to estimate inelastic

response quantities using results from the elastic analysis
model. This is possible using relationships established
for simple oscillators whose initial stiffness is linear to
the yield point.
The real bridge is likely to display nonlinear
response before yielding, as a result of concrete cracking
and nonlinear soil response. Ifwe are to use the relation-
ships established for the simple oscillators, it is necessary
to select a single-valued stiffness that corresponds to an
effective stiffness near "yield" of the bridge, as shown in
Figure RC3-3. For structural steel members, it is appro-
priate to use the elastic stiffness. For reinforced concrete

ATC-32 BDS Recommendations, Section 3: Loads 49


Gross-Section Effective
Stiffness Stiffness


Bilinear Response Envelope

RC Member Response Envelope

Figure RC3-3 Effective stiffness of reinforced concrete structure.

structures, different assumptions apply depending on

the extent of cracking expected in the various members,
as well as the expected effect of cracking on stiffness.
For reinforced concrete structures in which plastic
hinges are expected to form in the columns, the follow-
ing assumptions regarding element stiffness are usually

• For columns, the fully-cracked flexural stiffness is

appropriate. The stiffness should represent effects of
reinforcement slip from adjoining footings, pile
caps, or bent-cap joints. For structures supported
on columns ofnearly equal length and cross section,
it is appropriate to approximate the column
cracked-section stiffness, including reinforcement
slip, by half the gross-section stiffness. Where col-
umns vary, the cracked-section stiffness can be
approximated using Figure RC3-4. To account for
reinforcement slip from adjoining footings, pile
caps, or bent-cap joints, the cracked-section stiff-
ness obtained using the stiffness reduction factors of
Figure RC3-4 should be modified by the factor
(1- 0.811 ), where I is in meters and represents the
length from point of maximum moment to the
point of zero moment or contraflexure.

• For box-girder spans, longitudinal framing stiffness

is reduced owing to shear lag effects. For typical
box-girder construction, an effective width of box-
girder should not exceed the width of the column

50 BOS Recommendations, Section 3: Loads ATC-32


Jii 0.70 _ _- - - - - - - - - - - - - - - - - - Ast/A g = .04
a: _ _- - - - - - - - - - - - - - - - - - A s t / A g =.03
- _ _- - - - - - - - - - - - - Ast/A g =.02
u. ~_-------Ast/Ag =.01
en 0.-'0
..J 0.30

'P'f"'I'....,.-............~...............,.-......,....~- .......- ............-~.................

C..20 ~......
0.90 O.C~ 0.10 O.:~ 0.20 0..25 0..!5
Figure RC3-4 Relationship between cracked-section (l eff) and gross-section (1 9) stiffness values of rein-
forced concrete columns. See Section 8 for definition of other variables.

plus a width equal to twice the beam depth of the

cap on both sides of the column. Furthermore, the
flexural moment of inertia of a reinforced concrete
span should be reduced to three-fourths of the gross
moment of inertia to account for cracking. For pre-
stressed spans, no stiffness reduction due to crack-
ing is required.

• Where member torsion plays a significant role in

response, effective torsional stiffness of reinforced
concrete beams should be taken as equal to the
cracked-section value, and may be approximated as
being equal to KtPtn, where Kt = uncracked tor-
sional stiffness, Pt = volume ratio of transverse rein-
forcement, and n = ratio of Young's modulus of
steel to that of concrete. Because significant levels of
cracking are not expected in most cases in a box-
girder span, the torsional stiffness of the span may
be assumed to be equal to the gross-section value.
Torsional stiffness of prestressed sections may be
taken equal to the gross-section value.

• Shear and axial stiffness values of columns, piers,

bent caps, box girders, and other similar compo-
nents should be based on the gross-section values,
without a stiffness reduction to account for crack-

• Rigidity of column-cap joints should be taken into

account. In most cases, it is sufficient to model the
joint as a rigid block.

ATC-32 80S Recommendations, Section 3: Loads 51


Expansion Joint Modeling1

The dynamic response of multiple-span bridges is com-

plicated by the expansion joints separating the frames.
The restrainers at expansion joints also affect dynamic
response. Furthermore, the effectiveness of the restrain-
ers depends on the initial gap. Because these effects are
nonlinear in nature, linear dynamic analysis procedures
can be expected to produce only approximate estimates
of actual response.
The earthquake response ofbridges with expansion
joints can be bounded approximately by two linear mod-
els, referred to as the "tension model" and the "com-
pression model". In the tension model, spans on either
side of the expansion joint are joined by an element hav-
ing axial stiffness equal to the tension stiffness of the
restrainers. In the compression model, spans on either
side of the expansion joint are joined by an element hav-
ing infinite axial stiffness. In both cases, rotation at the
hinge is unrestrained.
For most structures, the tension model produces
results similar to those that are obtained for analysis of
the isolated frames.
Analysis for forces on and displacements of a frame
may be carried out by modeling the subject frame plus
any adjacent frames within two-frames distance from
the subject frame (and abutments if any are within two
frames). It is not necessary to use boundary frames
beyond the actual frames included in the model. There-
fore, a model for a straight or curved bridge may consist
ofbetween three and five frames in a structure. A model
for an interchange may include more frames because of
the more complicated geometry created by intersecting
frame lines. The use of significantly more than the rec-
ommended number of frames is discouraged because
larger models do not necessarily result in a better repre-
sentation of the dynamic response of the system with
expansion joints. Also, there is a greater possibility that
important vibration modes will be overlooked in the
analysis oflarge models.
Variation of subsurface conditions along the length
of a bridge may result in significant variations in ground
motion along the length. Currently available computer
programs do not facilitate use of multiple support input
motions. For structures supported on soils whose prop-
erties vary significantly along the length, the following
procedures are recommended. (1) Construct a "com-
pression model" of the bridge using the number of
frames recommended in the preceding paragraph. Ana-

1. Peer Review Panels on some past Caltrans retrofit projects have

required a "stand alone" analysis of each frame of a multi-frame
bridge. Designers may wish to consider the benefits of such analyses on
future bridge design projects.

52 8DS Recommendations, Section 3: Loads ATC-32


lyze the model assuming uniform support input motion,

using the ARS spectrum appropriate for that length of
the bridge. (2) Construct an isolated model for each
frame. Analyze each frame for the ARS spectrum most
appropriate for that frame.
Long bridges are subjected to spatial and temporal
variations of ground motion along their length. Along
with those conditions identified in the preceding para-
graph (and which tend to be the most significant where
they occur), traveling waves also affect the response of
long structures. The assumption of uniform ground
motion appears conservative because it forces the entire
bridge to vibrate in phase, whereas spatially varying
ground motion produces out-of-phase dynamic
response, which in most cases tends to cancel the energy.
For long structures with intermediate expansion joints,
the advantages of out-of-phase dynamic response are
commonly lost because out-of-phase movement is taken
up in the expansion joints. Therefore, the assumption of
uniform ground motion may be reasonable. Addition-
ally, in some cases out-of-phase ground motion may
cause amplified response because of impact of adjacent
frames. Another aspect to consider is that spatial varia-
tion of the support displacements produces so-called
pseudo-static stresses. These stresses do not appear to be
significant for bridges with a large number of relatively
short spans interconnected by expansion joints. They
may be of some consequence in long, continuous
bridges. Furthermore, some classes of structures (e.g.,
suspension bridges and cantilever bridges) may be sensi-
tive to spatially varying ground motion because the non-
uniform motion induces vibration modes not excited by
uniform ground motion, such as rocking of the piers.
These aspects should be considered when selecting the
ground motion representation.

Soil Stiffness

Although it is not practical to include all the effects of

the soil and foundation on the earthquake response of a
bridge, it is important to recognize that soil-structure
interaction introduces flexibility and energy dissipation.
The stiffness and damping properties of a foundation
depend on the characteristics of the soil, piles, and the
connections between the piles and pile cap. For use in
the Elastic Dynamic Analysis procedure, the founda-
tion/soil rotational and translational flexibilities for col-
umns and pier walls should be secant values based on
the maximum loads expected from the column or pier
wall. Soil springs at abutments and wing walls should be
input as a secant value consistent with the expected level
of deformation. Refer to Section 4 for detailed modeling
procedures for the soil-foundation system.

ATC-32 BDS Recommendations, Section 3: Loads 53


Five-percent-damped elasticARS response curves from Modal spectral analysis of a bridge structure normally
Figures R3-1 to R3-12 or from equivalent site-specific uses ARS response curves for five percent of critical
elastic response spectra shall be used as the horizontal damping.
response-spectrum loading. Where applicable, vertical Vertical response may be significant for certain
response-spectrum loading shall be taken as two-thirds bridge structures, especially those having long spans,
of the horizontal spectrum curves, except where site- outriggers, cantilevers, or C-bents. In these cases, the
specific evaluation is used to define vertical response- effects of vertical ground motion input should be consid-
spectrum loading. Standard soil profiles in Figures R3-1 ered. The relationship between horizontal and vertical
through R3-12 shall be as defined in Table R3-3. In input ground motion characteristics is a complex one
cases where soil andlor special mechanical devices influ- depending on rupture mechanism, rupture proximity,
ence energy dissipation appreciably, and where justified local soil conditions, and other factors. In general, both
by experimental evidence and analysis, damping the spectral ordinates and the spectral shapes vary for
exceeding five percent ofthe critical value is allowed. vertical and horizontal motions at a given site. The sim-
plistic specification ofArticle regarding vertical
response spectrum loading should be replaced by a more
appropriate representation whenever better information
is available.
See commentary Article for a discussion of
ground-motion representation.


The damping for a dynamic analysis using the response

spectrum method must be specified by modal damping
ratios. Studies of flexible reinforced concrete structures
founded on firm materials indicate that reasonable
response quantities can be obtained using viscous damp-
ing equal to five percent of the critical value combined
with the effective stiffness described in Article 3.21.6.
Therefore, the ARS spectra were derived for this damp-
ing ratio.
An exception is short bridges for which the response
may be dominated by the behavior of the abutment. In
such cases, most of the damping is due to energy dissipa-
tion in the abutments. Based on the information from
Seed, et al. (1984) for cohesionless soils, it is reasonable
to use a viscous damping ratio of 8 percent to 20 percent
for abutment fills in which the maximum shear strain
ranges between 0.05 percent and 0.5 percent, respec-
tively. The design spectrum must be modified for these
higher levels of damping. Field studies reported by Tsai
and Werner (1993) indicate that these approaches may
be warranted for some bridge designs. However, use of
five-percent damping is usually a conservative alterna-
Higher damping ratios may also be appropriate for
bridges with seismic isolation or supplemental damping.
These should be evaluated on a case-by-case basis, and
the conclusions independently reviewed.

54 BDS Recommendations, Section 3: Loads ATC-32


Responses in multiple directions shall be determined

according to Section 3.21.9.

3.21.7 Inelastic Static Analysis (3.21.7 Inelastic Static Analysis GeneraL C3. 21. 7.1 GeneraL

Seismic response shall be determined as local displace- Inelastic Static Analysis is a procedure to check inelastic
ments, individual member deformations, and individ- behavior of a bridge when subjected to lateral displace-
ual member forces using Inelastic Static Analysis ment amplitudes consistent with those expected during
techniques, considering nonlinear stiffness properties the design earthquake. The analytical model represents
of the structure and soil. the nonlinear load-deformation behavior of the compo-
nents, including the soil. Because the analytical model
accounts for the redistribution of internal actions as
components including the soil respond inelastically,
Inelastic Static Analysis is expected to provide a more
realistic measure of behavior than can be obtained from
elastic analysis procedures. Inelastic Static Analysis (or
alternately Inelastic Dynamic Analysis) is required for
the safety evaluation of Important Bridges. Use of
Inelastic Static Analysis for Ordinary Bridges and for the
functional evaluation is optional. As noted in Section
3.21.4(c), Inelastic Static Analysis may not be used to
reduce the requirements indicated by the Equivalent
Static Analysis or Elastic Dynamic Analysis methods. AnaLysis Procedure C3.21.7.2 AnaLysis Procedures

A step-by-step lateral-displacement response analysis of The analysis model should in general be a three-dimen-
a space-frame model of the structure is recommended. sional space-frame model of the bridge, including the
The number of degrees of freedom considered in the soil-foundation system. In most cases, it will be suffi-
analysis shall be sufficient to represent all critical cient to model individual frames between in-span super-
response modes. Gravity loads shall include dead loads. structure hinges; in many cases it will be sufficient to
Live loads shall also be considered where their effects model individual bents. Frame or individual bent mod-
are significant. Seismic loads may be assumed to act in els often are preferred because greater detail in analytical
one horizontal direction only. Nonlinear effects of grav- modeling and interpretation of results is possible. When
ity loads acting through lateral displacements shall be individual frame or bent models are used, care must be
included where significant. taken to represent the interaction effects among adjacent
frames and bents.
The analytical model should be developed to ade-
quately represent important flexural, shearing, torsional,
and axial force deformabilities and strengths. Columns
(including extended pier shafts), bent caps, and outrig-
gers can commonly be represented using line elements
with nonlinear response represented by concentrated
plastic hinges at critical locations. Fiber models are also
suitable. Pier walls can be modeled using line elements,
truss elements, or planar finite elements. The super-
structure, if included in the model, can be represented
by line elements. Where the analysis results indicate that
superstructure strength is approached under the design
loading, the model of the superstructure should be suffi-

ATC-32 8DS Recommendations, Section 3: Loads 55


ciently detailed to indicate local distributions of internal

moments and forces. In many cases a grid model can be
used for this purpose. The model should include
increased rigidity of cap-column joints.
VVhere inelastic response may occur along the span
of a member (usually a bent cap or superstructure), the
member should be modeled in a manner that will repre-
sent these effects. It may be appropriate to place nodes at
several locations along the member length for this pur-
pose. Static shear, moment, and torsion diagrams should
be compared with strength values along the full span
length to ensure that all sources of nonlinearity are
The analysis model should be capable of represent-
ing the effects of gravity loads on behavior. In general,
because response is dependent on load history, the grav-
ity load effects should be in place before applying lateral
loads. In most cases, it is sufficient to represent only
dead loads, including the superstructure, bents, and
nonstructural mass. Live load normally need not be con-
sidered, except for cases where live loads represent a sig-
nificant portion of total superstructure loading. VVhere
live load is included, it is normally adequate to represent
it as 15 percent of the AASHTO design lane live loading,
but other patterns ofloading should be considered
where they are likely to occur and may be critical.
Seismic loads may be applied in one horizontal
direction at a time. Normally, two load cases will be con-
sidered, one transverse to the alignment of the super-
structure and the other parallel to that alignment. For
longitudinal loading it is likely to be necessary to model
the entire frame including the superstructure. For trans-
verse loading, it is often feasible to model individual
bents. However, for curved frames and frames sup-
ported by columns of variable length or variable bound-
ary conditions subjected to transverse displacements, it
should be recognized that kinematics may produce tor-
sion in the superstructure and reversed curvatures in
supporting columns (Figure RC3-5). These interactions
should be taken into account.
Explicit consideration of concurrent loading in two
orthogonal directions is not required, but in the assess-
ment of capacities it should be considered that orthogo-
nal response exists. For example, when designing a
structure for displacement response in one direction, it
should be recognized that simultaneous displacements
are likely in the orthogonol direction, and the simulta-
neous occurrence may reduce the deformation capacity. StructuraL ModeL C3.21. 7.3 StructuraL ModeL

The structural model shall include the effects of con- Framing member models should include at least a bilin-
crete cracking and other material nonlinearities on ear load-deformation relation to represent response of
stiffness of members, and shall include the restraint of the member both before and after yielding, although

56 BDS Recommendations, Section 3: Loads ATC-32


the surrounding soil. Inelastic response characteristics

of the analysis model shall be justified by experimental • superstructure torsion
evidence. movement


(a) Curved Superstructure

superstructure twist
due to varying pier rotation




(b) Varying Pier Heights

Figure RC3-5 Superstructure torsion and coLumn

moments for frames under transverse dispLacement.

more realistic representations are encouraged. VVhere

strain-hardening models are used, results should be
checked to ensure that calculated actions do not exceed
strengths. Models capable of realistic modeling of
unloading stiffness are encouraged, although not
required. Refer to the commentary to Article for
additional details on framing-member models.
Models for seat-type abutments should be capable
of representing nonlinear response properties associated
with the bearings, the gap between the superstructure
and abutment, the passive resistance of the soil behind
the abutment backwall, the passive resistance behind the
footing/pile cap, the pile group stiffness, and other
aspects, as appropriate. Bearings may be either linear or
nonlinear depending on deformation levels and bearing
properties. Passive resistance of the soil should be mod-
eled using compression-only elements. Details of
soil/foundation models are presented in Section 4.
Models for integral abutments should be capable of
representing nonlinear response properties associated
with the passive resistance of the backfill soil, the pile
group stiffness, wingwall effects, and other aspects as
appropriate. Details of soil/foundation models are pre-
sented in Section 4.
Other foundation elements should be modeled to
represent their nonlinear response characteristics.
VVhere significant nonlinearities are not anticipated, the
soil/foundation system may be modeled using linear

ATC-32 BOS Recommendations, Section 3: Loads 57


elastic translational and rotational springs having secant

stiffness values corresponding to values expected for the
actions applied by the framing members. Where nonlin-
ear response is expected, including uplift, the soil/foun-
dation model shall be nonlinear. Details of
soillfoundation models are presented in Section 4. Distribution of Loading C3.21.7.4 Distribution of Loading

The center of mass of the superstructure shall be dis- The analysis may be carried out under either displace-
placed in steps to displacement amplitudes derived from ment control or force control. The displacement or force
dynamic response analyses according to Section increments should be sufficiently small that the develop- In multi-level structures supported by com- ment of inelastic response, including force redistribu-
mon elements, except where otherwise justified, lateral tion, can be correctly represented. Some computer
forces shall be applied to the center of mass of the major analysis packages will produce erroneous results if large
elements of the superstructure in proportion with the increments are used. The user should verify that the
product of their mass and centroidal height above the increments are adequate to correctly model the
base, and the upper level shall be displaced in steps to dis- behavior.
placement amplitudes indicated by dynamic response Although it is necessary to carry out the analysis
analyses. Local displacements and individual member only to the target displacement level, useful information
deformations and forces shall be monitored at each step. about ultimate behavior, including margins against col-
lapse, can be obtained by carrying out the analysis to
larger lateral displacements. Structural Capacity C3.21.7.5 Structural Capacity

Member forces and flexural plastic hinge rotations Member forces and flexural plastic hinge rotations
obtained from Article shall not exceed capaci- obtained from the Inelastic Static Analysis are to be
ties calculated according to Sections 8 and 10. checked against available capacities. The analysis should
include checks of actions (forces, moments, deforma-
tions, etc.) in plastic regions as well as forces outside the
plastic regions to verify that the inelastic action does not
occur in unintended locations. The evaluation shall
include framing members and their connections, rein-
forcement anchorage, foundations, and all other vulner-
able components.

3.21.8 Inelastic Dynamic Analysis (3.21.8 Inelastic Dynamic Analysis General C3.21.8.1 General

Seismic response shall be determined as structure dis- Inelastic Dynamic Analysis is a procedure to check
placement and individual member forces using inelastic behavior of a bridge when subjected to input
dynamic analysis techniques that consider nonlinear ground motions consistent with those expected during
stiffness, damping, and mass properties of the structure the design earthquake. The analytical model represents
and soil. the nonlinear load-deformation behavior of the compo-
nents, including the soil. Because the analytical model
accounts for the redistribution of internal actions as
components including the soil respond inelastically,
Inelastic Dynamic Analysis is expected to provide a more
realistic measure of behavior than can be obtained from
elastic analysis procedures. Inelastic Dynamic Analysis
(or alternately Inelastic Static Analysis) can be used to
satisfy the required analysis for the safety evaluation of

58 BDS Recommendations, Section 3: Loads ATC-32


Important Bridges. Use of Inelastic Dynamic Analysis

for Ordinary Bridges and for the functionality evalua-
tion is optional. As noted in Section 3.21.4(b), Inelastic
Dynamic Analysis may be used to reduce by up to 20
percent the requirements indicated by the Equivalent
Static Analysis or Elastic Dynamic Analysis methods Response-History AnaLysis C3.21.8.2 Response-History AnaLysis

A response-history analysis of a lumped-mass space- Dynamic response-history analysis is normally carried

frame model of the structure is recommended. The out using numerical step-by-step integration of the
number of degrees of freedom considered in the analy- equations of motion. As applied by commonly available
sis shall be sufficient to excite all critical response computer codes, the response history is divided into a
modes. Inertial mass shall consider dead loads only. sequence of short intervals, and response is calculated
Gravity loads shall include dead loads. Live loads shall during each step for a linear model having properties
also be considered as gravity loads where their effects equal to those at the beginning of the time step. The
are significant. Nonlinear effects of gravity loads acting properties are updated at the end of the interval, or the
through lateral displacements shall be included where interval may be subdivided automatically if significant
their effects are significant. events (nonlinearities) occur during the interval. The
integration method and the length of the time step
should be investigated to demonstrate accuracy and sta-
bility of the integration procedure.
The model should contain sufficient detail that a
realistic representation of response is obtained. As a
minimum, analysis for forces and displacements of a
frame may be carried out by modeling the subject frame
plus any frames within two-frames distance of the sub-
ject frame (and abutments if any are within two frames).
Models including more frames are acceptable. In select-
ing the size of the model, the analyst should be aware of
tradeoffs between longer frame models with limited
local modeling details and shorter frame models with
greater local modeling details. The number of degrees of
freedom and placement of nodes should be sufficient to
represent actual behavior of the bridge. Refer to the
commentary to Article 3.21.6 for discussion of selection
of nodes and nodal masses.
Typical elements to be included in the model are:
Superstructure-In most cases it will be adequate to
model concrete box-girder superstructures using linear,
elastic, three-dimensional line elements. More detailed
models may be required for other superstructure ele-
ments, or in cases where nonlinear superstructure
response is anticipated.
Bent caps -In most cases these can be modeled
using linear, elastic, three-dimensional line elements.
More detailed models may be required in unusual situa-
tions and where nonlinear response is anticipated.
Columns--Usually columns will be modeled using
three-dimensional line elements having inelastic
response properties with a yield surface described by the
interaction between axial load and biaxial bending. In
most cases, torsion and shear actions can be modeled
using linear elastic properties.

ATC-32 8DS Recommendations, Section 3: Loads 59


Pier walls-Usually pier walls will be modeled using

either three-dimensional line elements, truss elements,
or planar finite elements with a yield surface described
by the interaction between axial load and biaxial bend-
ing. Nonlinear shear deformations may need to be mod-
eled. Usually, torsion can be modeled using linear elastic
Cap-Column joints-The rigidity ofjoints should be
modeled. In many cases, it is appropriate to assume the
joint volume to be rigid, although the possibility of flex-
ible joints (especially in steel construction) should be
Expansion joints-These usually should be modeled
as three-dimensional nonlinear hinge elements that
account for tension-only nonlinear restrainer action,
compression-only impact effects occurring at the edges
(not centerline) of the deck, hinge seat gaps, supereleva-
tion in the deck, skewed hinge geometry, and transverse
shear keys. Figure RC3-6 depicts an example of a model
of a skewed expansion joint (Imbsen, 1994).
Impact Spring

Rigid Bar

Vertical Spring.
Figure RC3-6 Model for skewed expansion joint.

Foundations-These usually will be modeled as

either nonlinear elements or as equivalent linear ele-
ments having translational and rotational stiffness val-
ues compatible with the expected actions during the
design earthquake, taking into account deformations of
the soil, stiffness of pile groups, passive resistance of
footings and pile caps, and possible uplift.
Abutments-These will usually be modeled as non-
linear elements that represent the tension-only nonlin-
ear restrainer action, compression-only impact effects at
the edges of the superstructure, compression-only yield-
ing of the backfill soil, stiffness of pile groups, nonlinear
response of wingwalls, seat gaps, and skewed hinge

60 BOS Recommendations, Section 3: Loads ATC-32


geometry. Figure RC3-7 depicts an example of a model

of an abutment (Imbsen, 1994).

Wall stiffness spring

Footing stiffness spring


Pile stiffness spring

Abutment Bridge Deck

Figure RC3-7 Model for seat-type abutment.

Variation of subsurface conditions along the length

of a bridge may result in significant variations in ground
motion along the length. Wave passage effects may also
be important for some bridges. It may be possible to
input different ground motions at different support
locations; for such an analysis an effort should be made
to represent coherency and site conditions. Alternately,
analyses can be carried out for segments of the bridge for
ground motions representative of individual segments.
Typically, a segment will include a central frame for
which design actions are being sought plus any frames
and abutments within two frames of that frame.
The analysis model should be capable ofrepresent-
ing the effects of gravity loads on behavior. This will usu-
ally include only dead loads, as described in the
commentary to Article The action of gravity
loads acting through lateral displacements (the P-l:.
effect) will reduce lateral load strength ofthe system, the
reduction increasing with increasing lateral displace-
ment. A possible consequence of the P-l:. effect is a grad-
ual or sudden accumulation oflateral displacements in
one direction, leading to increases in demands or in
some cases leading to collapse. It is not sufficient to sim-
ply adjust the strength of the bridge or bridge compo-
nents to approximate the P-l:. effect. Instead, the
Inelastic Dynamic Analysis is required to include the P-l:.
effect explicitly as part of the resistance function for the
bridge model. Most computer codes for Inelastic
Dynamic Analysis are designed to include the P-l:. effect. Structural Model C3.21.8.3 Structural Model

The structural model shall include the effects of con- The linear and nonlinear properties of the components
crete cracking and other material nonlinearities on the of the bridge should be adequately represented in the
stiffness of members, and shall include the restraint of analysis model. For nonlinear elements, hysteresis rela-
the surrounding soil. Inelastic response characteristics tions under reversed cyclic loading should be consistent
of the analysis model shall be justified by experimental with observations from experiments. Where simplified
evidence. Viscous damping equal to five percent of the strain-hardening models are used, results should be

ATC-32 BDS Recommendations, Section 3: Loads 61


critical value or less shall be assumed for all critical checked to ensure that calculated actions do not exceed
response modes in addition to inelastic energy dissipa- strength values. Additional details of the nonlinear mod-
tion, except that higher viscous damping values are els are provided below.
allowed where justified by experimental evidence and Superstructure and Bent Caps--In most cases the
analysis. design should avoid nonlinear response in these compo-
nents. Therefore, it is appropriate to model these with
linear elastic elements. Stiffness assumptions for the lin-
ear elastic models should be consistent with the stress
levels and anticipated cracking, as described in the com-
mentary to Article
Columns--Usually columns will be designed to
develop flexural plastic hinges at one or both ends.
Therefore, it is necessary to model the column elements
using inelastic models that properly represent load-
deformation behavior under inelastic, cyclic deforma-
tion and force reversals. Figure RC3-8 plots the load-dis-
placement response measured in a laboratory test on a





0 0.0
Il. -50.0


-8.0 -4.0 0.0 4.0 8.0

Figure RC3-8 Load-dispLacement relationship for circuLar cross-section, cantilever-reinforced, concrete column
representative of modern Caltrans bridge designs. Column is subjected to uniaxial lateraL load and constant
axial load.

circular cross-section, cantilever, reinforced concrete

column representative of modem Caltrans bridge
designs (Chai, 1993). Axial load was constant, and lateral
load was applied along one axis only.
Characteristics of the behavior include (1) initial
stiffness degradation due to concrete cracking and rein-
forcement slip from the foundation block, (2) yielding
in the load-displacement relation due to flexural yield-
ing at the fixed end, (3) moderate strain hardening fol-
lowing yield, (4) unloading slope that decreases with
increasing displacement amplitude, and (5) a moderate

62 BDS Recommendations, Section 3: Loads ATC-32


amount of pinching around the point ofload reversal.

Strength degradation is expected to be minimal in a col-
umn satisfying the modern Caltrans requirements. Fig-
ure RC3-9 plots the behavior of a column under biaxial
lateral loading with axial load that was equal to a con-
stant plus a linear function of the lateral load in one
direction (Mazzon~ 1995). Note that behavior on the
tension and compression sides varies and that resistance
along one lateral axis is a function of the lateral load
applied along the orthogonal lateral axis.

80 1--,----,--,---,----r---,---.-----,,--~--_r356

60 -j---t------i---t---t---+---t---f---J---+---l-267

40 -t---t-----j---t---+--+-;;~~"rl~+-++---;f_---+---+ 178

~ 20 -t----j----i---t---+----r-=:JI 89 .--,

'-' ~

~ 0 0
~ ~
00. -20 -t---+----if---t7L----h#HI-i. -89 00.

-40 +--+--+---t4-:JIf~~-i----+--I--~-+--I- -178

-60 +---j---+--+--t---t--+--+---1---+---.f- -267

-80 +-----;I----+--+--+---t--+--4---1---+---I- -356

-12.5 -10 -7.5 -5 -2.5 0 2.5 5 7.5 10 12.5
Drift (%)

Figure RC3-9 Load-dispLacement relationship for circuLar cross-section, cantilever-reinforced, concrete coLumn
representative of modern CaLtrans bridge designs. CoLumn is subjected to biaxiaL LateraL Load and constant axial
The behaviors described in Figure RC3-9 can be
approximated using either fiber models or concentrated
plastic hinge models with stiffness and strength values
calculated according to conventional procedures. Fiber
models subdivide the column cross-section into steel,
plain concrete, and confined concrete fibers, each having
representative hysteretic material properties, which are
subsequently integrated to compose the section load-
deformation behavior. Concentrated plastic hinge mod-
els represent the yielding region with a concentrated
plastic hinge having appropriate hysteretic properties.
Fiber models tend to be better able to represent triaxial
behavior of the column under the relatively random
load histories to which the column is subjected, but are
relatively computationally inefficient. Concentrated
plastic hinge models may not be able to adequately
model triaxial interaction effects, but are more computa-
tionally efficient. Whatever model is used, the analyst

ATC-32 BOS Recommendations, Section 3: Loads 63


should verify with component studies that the model

adequately represents the column hysteretic behavior.
For reinforced concrete columns it is usually considered
sufficient for the model to represent linear behavior to
yielding, modest strain-hardening after yielding, stiff-
ness degradation upon load reversal, and interaction
with axial load and transverse lateral loading. Behavior
before cracking and pinching are commonly ignored.
Bilinear, non-stiffness-degrading models usually should
not be used for reinforced concrete columns.
Pier walls-Usually pier walls will be designed to
develop flexural plastic hinges at one or both ends.
Under out-of-plane loading, flexural response is similar
to that for a column. For in-plane loading, the nonlinear
response can be dominated by either flexural or shearing
action, depending on the level of shear stress. Figure
RC3-10 plots the load-displacement response measured

, •• a
.. .. , .. ,

~ 100
~ 50
~ Rexural Crack
.... -50
.... ,.,
.. , .... .. , •
-0.8 -0.6 -0.4 -0.2 0.0 0.2 0.4 0.6 0.8
Lateral Deflect/on at Load PoInt, In.

Figure RC3-l0 Load displacement relationship for relatively low-aspect-ratio pier waLL loaded in its plane.
in a laboratory test on a wall having aspect ratio near
unity, for which shear cracking preceded flexural yield-
ing (Sozen, 1993). Axial load was constant, and lateral
load was applied in the plane of the wall only. Character-
istics of the behavior are similar to those identified for
columns, although there is a tendency for greater stiff-
ness degradation at higher shear stress levels. Although
not clear in the figure, studies indicate that slip of rein-
forcement from the foundation may cause displace-
ments equal to those caused by conventional flexure
before yielding (Sozen, 1993). Interactions between
biaxial lateral loads have not been investigated in this
The behavior of a pier wall subjected to out-of-
plane lateral loading can be modeled using the proce-
dures described for reinforced concrete columns under

64 BDS Recommendations, Section 3: Loads ATC-32


uniaxial lateral loading. Behavior under in-plane loading

may be modeled using line elements, truss models, or
planar finite elements. A simple approach is to model
the wall as three springs in series, as suggested by Figure
RC3-11 (Sozen, 1993). The flexural spring has flexural
stiffness and strength values calculated according to con-
ventional procedures for columns. The shear spring has
initial stiffness and strength representing pre-cracking
response, with parameters calculated according to con-
ventional procedures. Following cracking, the shear
spring stiffness is reduced to approximate post-cracking
stiffness. The slip spring has a linear elastic stiffness cal-
culated based on an assumed bond stress-slip relation.
Details of the stiffness calculation are given in the figure.
Hysteretic response for the flexure and shear springs can
be modeled using stiffness-degrading rules.

Flexural Sheor Slip

Deformation Deformotion Deformotion

Figure RC3-11 Three-spring modeL for reinforced con-

crete pier waLL.

Cap-Column joints-Reinforced concrete joints

should be designed to remain essentially elastic in
response to worst-case loadings. In reinforced concrete
construction, it is usually appropriate to assume that the
joint volume is rigid. In steel construction, the joint flex-
ibility should be modeled. Shear yielding should be
modeled if it is possible for the joint to yield in shear
under worst-case loadings. Simple bilinear models usu-
ally are adequate for modeling the nonlinear behavior of
yielding steel joints.
Expansion joints-Details of modeling expansion
joints depend on the details of the expansion joint itself.
Figure RC3-6 depicts an example for reference. In this
model, tie stiffness springs represent the restrainers.
Restrainer properties can be based on the stress-strain
relations reported in Caltrans Memo to Designers 20-3.
The example in Figure RC3-12 suggests that a bilinear
relation with strain hardening and a gap will adequately
represent the restrainer. Impact springs, vertical springs,
and shear springs can be simple linear springs with high
Foundations-Modeling of foundations will depend
on the stress levels and foundation details. See Section 4
for detailed modeling procedures.
Abutments-Details of the abutment model will
incorporate aspects of expansion joint modeling and

ATC-32 BDS Recommendations, Section 3: Loads 65


o 25.4 50.8 76.2 101.6 127.0 152.4 1n.8 203.2 228.6 254.0 304.8
200 889
I V4-' btlr ASTM A -722 'willi su."p'."""'tlr, '.f1ui,."",n,':h.
r y Y

r .".- ~
-r I


125 556

~IOO 445 z
c o
3 75 I I . I 334 ~

I ....-~ -6XJ9 C'tlb1fFed.Spec·tH-w-4IOC) ..J
I ~
50 222
~ ~ V
/} I
f4 II I V

, .
, I itlg• '·jg'1 S i 1/4·

2 3 4 567 8 9 10 II

(Caltrans Memo to Designers 20-3, May 1994)


fye-----/i -

gc gt
gc = Impact gap
gt = restrainer gap
dy = restrainer yield
(Compression) fy = restrainer yield
Figure RC3-12 Measured and idealized load-displacement relationship for restrainer with gap.

66 BOS Recommendations, Section 3: Loads ATC-32


foundation modeling. See relevant discussion of these

subjects in the commentary to Article
Energy dissipation in the analytical model occurs
both by viscous damping and by hysteretic energy dissi-
pation. It is recommended to use viscous damping equal
to five percent ofthe critical value. The reason is that this
will produce results consistent with results from the
Elastic Dynamic Analysis for low response amplitudes,
and because viscous damping at or around this level
does not significantly affect inelastic response where the
behavior is dominated by hysteretic response for large-
response amplitudes. Although some computer codes
enable the analyst to select five percent of critical damp-
ing for all modes, most do not. It is considered accept-
able to use Raleigh damping (mass and stiffness
proportional damping) that has five-percent damping at
the longest period and at the shortest period likely to
contribute significantly to the displacement response. By
so doing, intermediate periods will be underdamped
(conservative) and shorter periods that do not contrib-
ute significantly will be overdamped (Clough, 1993).
Higher damping is likely to be appropriate only for
cases where special energy dissipating devices are used.
For short structures where abutments control the
response, it has been shown that effective damping is
higher than five percent of the critical value. However,
the nonlinear response model should represent this
action through abutment hysteretic response rather than
through increased viscous damping. Seismic Time-History Loading ( Seismic Time-History Loading

Where Inelastic Dynamic Analysis is used to reduce There is no unique ground motion for a given site. For
design requirements as allowed in Section 3.21.4(b), or safety or functional evaluation, the possible ground
to satisfy the analysis requirements for Important motions can vary widely depending on rupture mecha-
Bridges as allowed in Section 3.21.4(c), the following nism, location, propagation path, and geologic condi-
requirements shall be satisfied. The bridge shall be ana- tions. Bridge response may vary widely to these possible
lyzed for an ensemble of ground motions along each ground motions. Therefore, it is preferable to evaluate
principal direction whose characteristics bound the the bridge response for several (rather than a single)
expected design ground-motion parameters. Ampli- ground motions that are comparable with the smooth
tude, frequency content, long-period wave forms design spectra. Maximum design response for three
(velocity and displacement time histories), and duration spectra-compatible ground motions or the mean
of each motion shall be consistent with the site condi- response for seven such motions is a standard criterion
tions and evaluation type (functional evaluation or in structural design practice. The specific requirement of
safety evaluation). Vertical ground motion should be this section is that this is the minimum number of
considered where important. Design actions shall be ground motions to consider when Inelastic Dynamic
taken to be equal to either the maximum values calcu- Analysis is used to satisfy requirements for Important
lated for three ground motions in each principal direc- Bridges or to reduce design requirements for any bridge
tion, or the mean values calculated for seven ground below the standard requirements. When Inelastic
motions in each principal direction. Dynamic Analysis is used only to gain improved per-
spective on response, and not to satisfy specific require-
ments of this document, fewer than the minimum
number of ground motions may be considered. How-
ever, the analyst should recognize the limitations of

ATC-32 8DS Recommendations, Section 3: Loads 67


using fewer than the recommended number of ground

The character of near source time histories is greatly
dependent upon the point(s) of rupture initiation and
direction(s) of rupture propagation (Singh, 1985). The
energetic unidirectional pulse-type motions (also
referred to as "fling") are present in the direction of
propagation. When time histories are selected for
response spectral matching for the near-field conditions
it is important that they contain proper pulse-type
motions of the forward direction and the waveforms of
the fault-normal or fault-parallel ground motion as the
question may be. It should be pointed out that the spec-
tral matching process cannot build a rupture directivity
pulse or the wave form offault-normal and fault-parallel
motions where these features are not present to begin
Ground motions are amplified in basin structures
due to trapping and focusing of seismic energy within
dipping layers of the near-surface sediments. As wave
energy enters the thickening margin of a basin, the waves
can become trapped as surface waves which propagate
laterally across the basin with slow apparent velocities.
Since the surface waves decay slowly with distance, this
leads to amplified motions and extended durations of
shaking in the basin sites. This suggests that care must be
taken when selecting ground motion time histories for
sites located in basin environments that include proper
long-period content, waveforms, and durations for
reflective basin effects. Combination of Effects C3.21.8.5 Combination of Effects

Responses in multiple directions shall be determined Except in unusual cases, ground motions will be input
according to Article 3.21.9. simultaneously in either two horizontal directions or
two horizontal directions plus the vertical direction.
This being the case, it is not necessary to use the combi-
nation rules of Article Instead, the design
response quantities will be governed by Article; that is, the design ground motion is taken to
be equal to the maximum results obtained for three
ground motions or the mean for seven ground motions
(see Section Structural Capacity

Member forces and flexural plastic hinge rotations

obtained from Article shall not exceed capaci-
ties calculated according to Sections 8 and 10.

68 BOS Recommendations, Section 3: Loads ATC-32


3.21.9 Combination of Effects C3.21.9 Combination of Effects General C3.21.9.1 General

Earthquake actions shall be determined for seismic Design should consider the effects of ground motions in
input in at least two orthogonal directions (usually the two orthogonal horizontal directions in all cases. Verti-
horizontal longitudinal and transverse axes of the cal input should be considered for unusually long spans,
bridge). The longitudinal axis of a curved bridge may for outriggers, for "c" bents, and for cantilevers.
be represented by a chord connecting the two abut- For modal spectral analysis, as commonly used with
ments. Seismic input along the vertical axis is to be con- Elastic Dynamic Analysis, the structural model should
sidered where its effects are significant. be analyzed for the spectra applied separately along the
orthogonal axes. The results are combined according to
Section, which is intended to provide a rea-
sonable estimate of the multiaxial actions to be used for
For response-history analysis, as commonly used
for Inelastic Dynamic Analysis, the structural model
should be analyzed for simultaneous ground motions in
the two horizontal (or two horizontal plus one vertical)
directions. By applying the motions simultaneously, a
range of simultaneous design actions can be obtained
for comparison with an interaction surface. An example
is a reinforced concrete bridge column, for which it is
necessary to determine biaxial bending moments and
axial load at discrete times for comparison with the P-
Mx-Myinteraction surface. This is the approach of Arti-
Common practice is to apply one of the horizontal
ground motions in the longitudinal direction, defined
parallel to a chord connecting the ends of the bridge,
and to apply the other horizontal ground motion in the
transverse direction. Studies have shown that this is not
necessarily the most critical set of directions along which
to apply the ground motion representation. However,
within the overall uncertainty of the design problem,
this design simplification is considered adequate. Design Procedures C3.21.9.2 Design Procedures

Earthquake actions determined according to Article There are two ways of considering the combination rules shall be combined as follows: for design. If the objective is to find the maximum
response to multi-component ground motions for a sin-
(a) For structures designed using Equivalent Static gle response quantity, a preferred approach is to use the
Analysis or modal spectral dynamic analysis, seis- square root of the sum of squares (SRSS) combination
mic effects shall be determined for the following rule (Note that the complete quadratic combination, or
three load cases, except that Seismic Load Case 3 CQC, does not apply). On the other hand, if the objec-
may be ignored where vertical seismic effects are tive is to locate the response to multi-component
not significant: ground motion on a failure surface (such as a P-MrMy
interaction diagram for a column), alternate approaches
Seismic Load Case 1: Combine the actions resulting may be preferred. Considering the latter to be the objec-
from the transverse loading with 40 percent of the cor- tive, one approach based on minimizing the worst error
responding actions from the longitudinal and vertical for an elliptic failure surface is to use the combination
loadings. rule identified in Article

ATC-32 BDS Recommendations, Section 3: Loads 69


Seismic Load Case 2: Combine the actions resulting To clarify the intention of the combination rule of
from the longitudinal loading with 40 percent of the Article, consider an example of a reinforced
corresponding actions from the transverse and vertical concrete bridge column design. Under longitudinal
loadings. loading, denote axial load, moment about x axis, and
Seismic Load Case 3: Combine the actions resulting moment about y axis as pL, MxL, and M/, respectively.
from the vertical loading with 40 percent of the corre- Under transverse loading, similarly use pT, M x T, and
sponding actions from the transverse and longitudinal My T, and under vertical loading use pV, M x v, Myv. Then
loadings. for Seismic Load Case 1, the simultaneous design actions
are axial load, P, moment about x axis M", and moment
(b) For structures designed using response-time-his- about y axis, My, where P = pT + O.4(pL + pV), M x = Mx T
tory analysis, the input motions in orthogonal + Oo4(MxL + M xv), and My = M/ + Oo4(M/ + M/).
directions shall be applied simultaneously, and These design actions are compared with the biaxial
individual responses shall be monitored directly. bending and axial load interaction diagram for the col-
Where this is not feasible, analysis may be for indi- umn. Similar results and comparisons are obtained for
vidual input motions, and responses may be com- Seismic Load Cases 2 and 3, and the worst case is used
bined according to paragraph (a) of this article. for design.
The coefficient of 0.4 for horizontal ground motion
loading (that is, the specification to use 40 percent of the
response due to loading in the orthogonal directions)
differs from the current Caltrans specification, which
uses a coefficient of 0.3 (or 30 percent). The coefficient
0.3 was derived assuming that the response-spectrum
loading represented the maximum principal direction,
with the orthogonal ground motion intensity being 85
percent of this value. For the case where the response
spectrum represents the average ground motion inten-
sity, as is the case in this specification, the correct coeffi-
cient is 004.
The current Caltrans specification considers only
combination of actions due to horizontal input motions.
The extension to include vertical input motions was
done without extensive study. The coefficient 0.4 applied
to load combinations involving three components of
motion is done on an interim basis. In the interim, it
provides a simple approach to a very complicated non-
linear interaction surface. This rule is the subject of con-
tinuing study.

3.21.10 Design Displacements C3.21.10 Design Displacements

Displacements of a structure during an earthquake may

be of equal or greater importance than the forces that
develop in the structure. For any structure, lateral dis-
placements determine whether adjacent frames or struc-
tures impact. For Full-Ductility and Limited-Ductility
Structures, lateral displacements determine inelastic
deformation demands in plastic hinges and other ductile
regions. Results of nonlinear dynamic analysis may be
used directly to estimate expected displacements. Results
oflinear dynamic analysis require some interpretation if
the structure responds to the design earthquake in the
nonlinear range

70 BOS Recommendations, Section 3: Loads ATC-32

BRIDGE DESIGN SPECIFICATIONS COMMENTARY Adjustment of Elastic Displacement C3.21.10.1 Adjustment of Elastic Displacement

Horizontal displacements calculated from Elastic A common approximation for design purposes, known
Dynamic Analysis shall be multiplied by the factor Rd as the equal-displacement rule, is that the peak displace-
to obtain design displacements. ment amplitude for a structure responding inelastically
is equal to the peak displacement amplitude calculated
R d =(1- ~) ~ + ~ ~ 1 R3-1 for the same structure (same initial period and viscous
damping ratio) responding elastically. It should be noted
The value of Z used shall be taken equal to the max- that the equal-displacement rule is not theoretically
imum value of Z used in the design of that frame. Values based; rather, it is an observation made from experimen-
of y* are given in Table R3-4. tal and analytical studies. The equal-displacement rule

TableR3-4 Values of y* (in seconds)

M= 6.5± 0.25 M = 7.25 ± 0.25 M= 8.0± 0.25

g B C D E B C D E B C D E
--- --- ---
0.1 0.32 0.45 0.46 0.44 0.41 0.53 0.56 0.56 0.51 0.69 0.71 0.71
0.2 0.37 0.44 0.49 0.64 0.42 0.53 0.55 0.74 0.47 0.61 0.65 0.85
0.3 0.35 0.43 0.50 0.73 0.38 0.51 0.55 0.76 0.48 0.64 0.65 0.98
0.4 0.39 0.47 0.50 0.87 0.42 0.56 0.59 0.93 0.46 0.62 0.66 1.04
0.5 0.37 0.46 0.50 0.42 0.53 0.62 0.45 0.59 0.70
0.6 0.35 0.44 0.50 0.43 0.54 0.64 0.46 0.60 0.76
0.7 0.50 0.66 0.76 0.54 0.71 0.80

holds only in an average sense, and even then only for a

restricted period range. For short-period structures, lin-
ear response models tend to underestimate inelastic dis-
placement amplitudes. Article attempts to
establish an estimate of inelastic displacement amplitude
from the amplitude calculated assuming elastic
Equation R3-1 produces a displacement amplifica-
tion factor Rd = 1 for Y~ Y*. In effect, this is stating that
the equal displacement rule is adequately valid if the
effective initial period T of the structure is equal to or
exceeds the characteristic ground motion period, Y*.
The period y* corresponds to the peak of the input
energy spectrum, and may be taken as the intersection of
the nearly constant velocity and nearly constant acceler-
ation ranges of the elastic response spectrum. Values of
y* are indicated in Table R3-4. For soft soil sites, values
of T exceeding 1.0 seconds are not unusual, and site-
specific analysis is required.
For Y < Y*, Equation R3-1 produces a displacement
amplification factor Rd greater than unity, reflecting the
observation that inelastic displacement amplitudes tend
to exceed amplitudes calculated with a linear-elastic
response model. The form of the equation was selected
to approximate results of analytical studies of single-
degree-of-freedom systems (Miranda, 1991). These

ATC-32 BDS Recommendations, Section 3: Loads 71


results indicated that mean displacement ductility

demands could be held approximately constant if
strength was equal to the elastic strength demand for
zero period, equal to the elastic strength demand divided
by the target displacement-ductility ratio for T =1'*, and
varied linearly between these two points (Figure
RC3-13). Algebraic manipulation results in Equation

8.0 ROCK
....... " ..


........ Statistical study [21]
- - Miranda (Eqs. 38 & 39)
0.0 1.0 2.0 3.0
PERIOD [sec]
(b) .'

..... -_ .....

....... -_ - ..

•••••••. Statistical study [21]
- - Miranda (Eqs. 38 & 40)
0.0 1.0 2.0 3.0
PERIOD [sec]

Figure RC3-13 Mean relationships between strength-

reduction coefficient (R,J and displacement ductility
demand (/l).

The observation that linear response analysis can be

used to estimate overall nonlinear response amplitude
does not carry over directly to local displacements and
deformations. Where it is critical to identify local nonlin-
ear deformations, for example in Important Bridges,
either Inelastic Static Analysis or Inelastic Dynamic
Analysis is recommended.
The displacement amplification factor, Rd , does not
account for effects of near-source ground motions. I
Design of structures near active faults requires special
considerations with respect to energetic long-duration
pulse-type loadings not taken into account in the ARS
spectra or in developing the factor Rd.

72 BDS Recommendations, Section 3: Loads ATC-32


A structure that yields during a long-duration pulse

(impulse loading) may experience very large displace-
ment and ductility demands, and it may sustain large
residual deformations. The extent of these actions
depends on the strength and initial period of the struc-
ture, as well as the amplitude, duration, and shape of the
pulse. Some theoretical solutions of structural response
to impulsive loadings are available (Biggs, 1964). For
example, the shock spectra in Figure RC3-14 indicate the
effect of pulse duration on ductility demand for the case
of a simple triangular pulse acting on an elasto-perfectly-
plastic single-degree-of-freedom structure (Bertero et
al., 1991). Studies of simple bridge structures subjected
to near-source ground motions have also been con-
ducted (Mayes, 1995). When designing structures adja-
cent to potentially active faults, the nature of potential
ground motions should be investigated and the likely
impact of those ground motions on structural response
should be gauged, taking into consideration the effects of
impulsive motions on a yielding system. Additionally,
the design should be modified from the standard design
outlined in this specification with due consideration of
the importance of the bridge. Inelastic Displacements C3.21.10.2 Inelastic Displacements

Displacements calculated from Inelastic Dynamic Anal- Inelastic Dynamic Analysis, as specified in Article 3.21.8,
ysis may be used directly in design, but shall not be less takes nonlinear response characteristics into account
than 80 percent of the values determined from Article directly, so results do not need to be modified according to the requirements of Article Because of sig-
nificant uncertainties in ground motion definition and
structural modeling, it is recommended that no less than
80 percent of the standard requirements be used even if
Inelastic Dynamic Analysis indicates less than that level
is acceptable.

1. The issue of near-source ground motions with high velocity pulses

arose late in the ATC-32 project and there was not enough time to
arrive at a consensus approach to this problem. Three basic approaches
were considered. The first is simply to conduct an inelastic dynamic
analysis of all structures in near-fault regions. Acceleration histories
used for these analyses should contain identifiable near-fault motion
effects. The second method is the use of a modified Z factor at near-
fault locations. Something similar to this has been proposed by the
Structural Engineers Association of California (SEAOC) for its lateral
force provisions. but some PEP members felt the SEAOC modification
factors were inappropriate for the higher peak bedrock accelerations
used by Caltrans. Caltrans may wish to consider slightly lower modifi-
cation factors on an interim basis. A third method would involve a sep-
arate pulse loading for structures near faults. Because near-source
effects are currently not considered. Caltrans should develop interim
design guidelines for considering this problem. even if they are not vig-
orously correct. In addition, Caltrans should conduct further studies to
refine these guidelines based on the latest research into this topic. Fur-
ther information on the issue of near-source motion may be found in
ATC-32-1. the Resource Document.

ATC-32 8DS Recommendations, Section 3: Loads 73



11.. /1';.0.2 ~~0.*?1= 0.5 C"O.6
., 'I
/ / I I
I i I / 1/ 1/ / II I
I ·1 I If IJ

r - -~
,.....- -; V- -V I/ -/-ij ~+ bl- - II ,- -~-


! !I V 71-
l4 ~l
,- l-


I - 1-

~ --{-

8 / / ./
I I I 1.- ,..-- II
I V ./1 /
1/ :I / V /
V I I 1/ 0.9-l--

I Iii v v
J 1/ ~ I j :J I
~;+-/ -f l-) k':V
P\t ......... v ,....- ,-
~ --,-

t/j/ v!/
')'/ V
~ ~r- ~V 1--- -i-

~ v-- vI--'
' - 1-

1.0 /
/ 0/
/ ./
J l./
V ......
"- r-
[Vyr\ / r\ 1/1\
f"\ I.OC-r-

0.8 t.20+--
./ ./ / /./ r-....
/y A ./ .- ~ "./ '\./" 1.60~

0.5 V /

V /t/
V/V ~ I "l I
V;(- ~..l.
I t
VI I ~ r.-j r,l till.
~-+- ~+ Resistance Displacement ~-
I I Triongulor pulse
tunct>()tl function I
I I ; lo~d I I . I I I I I I I I I
0.1 0.2 0.5 0.8 1 2 5 8 10 20

Figure RC3-14 Shock spectra for a trianguLar pulse acting on an elastic-perfectly-plastic, singLe-degree-of-
freedom osciLLator. Target DispLacement Capacity C3.21.10.3 Target DispLacement Capacity

The target displacements for Inelastic Static Analysis The Inelastic Static Analysis procedure of Section 3.21.7
shall be 1.5 times the displacements obtained from is required to be used to check deformation capacities
Article for the safety-evaluation earthquake. versus deformation demands for the safety evaluation of
Important Bridges. In establishing the criteria for its use,
the variabilities associated with Inelastic Static Analysis
need to be considered. Data available at the present time,
combined with the crudity of the basic analytical model,
do not justify a detailed probabilistic analysis. Rather,
engineering judgment has been applied, considering
known variabilities, to establish the requirement of this
section. The following aspects were considered:
Analyses of available test data for columns satisfying
the recommended criteria indicate that the ratio
between available displacement ductility capacity and
capacity calculated according to the procedures of Sec-

74 BDS Recommendations, Section 3: Loads ATC-32


tion 8 ranges from about 1.3 to 1.9, with a mean of

about 1.5.
The response modification factor, R d , given by
Equation R3-1 represents a mean relation between
inelastic and elastic responses. Mean-plus-one-stan-
dard-deviation values are about 1.5 times larger than the
mean values.
Ground motion representations are inherently
uncertain. The ARS response curves represent mean
spectral responses. Mean-plus-one-standard-deviation
ARS values are about 1.5 times the mean values.
Although site-specific response curves are required for
Important Bridges, and therefore ARS response curves
normally are not to be used for these bridges, the level of
uncertainty in any design ground motion should be duly

3.21.11 Design Forces (3.21.11 Design Forces

The following requirements apply depending on the The requirements are organized according to whether
intended structural action, as defined in Article 3.21.3. they apply to ductile structures, elastic structures, or
protected structures. FuLL-Ductility Structures and Limited- C3.21.11.1 FuLL-Ductility Structures and Limited-
Ductility Structures. Ductility Structures

For functional evaluation of Important Bridges, mem- Direct design for the functional-evaluation earthquake
ber forces and moments calculated from Articles 3.21.5 is required only for Important Bridges (Table R3-2). Full
or 3.21.6 shall not exceed member design strengths, service access for Important Bridges is required almost
except that larger calculated actions are permitted if immediately after this event. In addition, minimal struc-
analysis demonstrates that the functionality require- tural damage should be experienced. On the basis of
ments are satisfied. required performance, there should be no crushing of
For safety evaluation, locations of inelastic action the concrete, and residual crack widths should be
are to be identified clearly. Design forces and moments acceptably small so that remedial action is not required.
associated with those actions shall be at least equal to Calculations based on an acceptable crack width of one
the forces and moments obtained from Articles 3.21.5 millimeter at rest after the earthquake suggest that at the
or 3.21.6 divided by the force reduction coefficient, Z, maximum response, the tensile steel strains should not
which is interpolated from Figure R3-13. Full-Ductility exceed about 0.01. To avoid concrete crushing, concrete
values apply only to Ordinary Bridges and only when strain at the maximum response should not exceed
intended inelastic action forms in accessible locations. about 0.004. A direct evaluation of whether a structure
Otherwise, Limited-Ductility values shall be used. satisfies the functionality criteria is possible using either
Design forces and moments associated with loca- (a) Inelastic Dynamic Analysis or (b) Inelastic Static
tions of inelastic action shall be increased to include the Analysis, where the structural model is displaced to
effects of gravity loads acting through the lateral dis- amplitudes expected for the functional-evaluation
placements (P-L1 effects), as required by Article 3.21.15. earthquake. It is appropriate to assume that the displace-
Plastic hinge design shear strength and design ment amplitude for the functional-evaluation earth-
strengths of members resisting the plastic hinge quake is equal to the displacement calculated using
moments shall be determined from the capacity design elastic analysis, without modification.
procedures of Article 3.21.14. For superstructure ele- For Important Bridges, Inelastic Static Analysis is
ments, the design forces shall also include forces from required for safety evaluation. Therefore, for these
vertical seismic input motions where they are impor- bridges, a direct check of functionality by nonlinear
tant. Design forces in restraining elements shall be analysis does not require a significant amount of effort
determined according to Article 3.21.12. beyond that required for safety evaluation. However, if
the functionality check reveals inadequacies in the struc-

ATC-32 BDS Recommendations, Section 3: Loads 75


Full Ductility Structures tural system, the entire system must be reproportioned,
5.-------.------,----,------,.---..., and a significant amount of design effort may be lost. For
N this reason, and in the interest of simplicity, Article
;i41----t---f-----:..r----t----t-----f permits the functionality check to be carried
8 out using elastic analysis. Although some nonlinear
§ 3 1----t--7'f<..---t-::----I---:----:--+----f
:g response is permissible, it is difficult, in simple terms, to
~ 2 1---r--t--\-:::;;;;;ooI----I----+----f express the permissible amount in relation to the elastic
Gl analysis. For this reason, when elastic analysis is used as
If~ 1 I-~=.=*===I===*===*====l the sole means of checking functionality, Article
o L -_ _---I...B_ritt
_·_le_e_le_m~e_nts_n_ot_d_es_'ig~n_ed_by:..capa_.:.~c_ity:.._de_s...::ig:....n..J requires that the structure remain fully elastic.
o 0.5 1 1.5 2 2.5
A rational analysis, involving Inelastic Static Analysis or
Period Ratio, TfT*
Inelastic Dynamic Analysis is permitted and may result
in considerable economy in some structures.
Umited Ductility Structures For the safety-evaluation earthquake, it will usually
not be practical to design the structure to remain elastic.
Well COnfi~ed
concrete Llumns, _ In a structure designed according to these recommenda-
§ 3
( steel cOluins and pile rafts
tions, inelastic action can be expected to be predomi-
\ Transversely loaded piers; nantly in the form of flexural plastic hinge rotations
~( abutment walls and wing walls occurring in preselected locations. In most structures, it
~ ~ will be most practical to select plastic hinges to form in
~ 1 the columns. Using the standard design procedures of
o "- Brittle elements not designed by capacity design Article 3.21.4, design moments at plastic hinges will be
equal to moments calculated from the Equivalent Static
o 0.5 1 1.5 2 2.5
Period Ratio, T/T* Analysis or Elastic Dynamic Analysis divided by the
force-reduction coefficient, Z, interpolated from Figure
Figure R3-13 Force-reduction coefficient, Z. RC3-13. Note that different coefficients apply to Full-
Ductility and Limited-Ductility Structures.
For periods T greater than r (see discussion in the
commentary to Article 3.21.lO.1), the values ofZ for
Full-Ductility Structures are approximately equal to cal-
culated displacement ductility capacities, with some
allowance for judgment considering redundancy and
conservatism in the predictive equations. For Limited-
Ductility Structures, values were reduced to provide for
increased serviceability and a greater margin of safety
against collapse. Values of Z decrease linearly from the
maximum values to unity as period decreases to zero.
This decrease has been shown to be necessary to main-
tain approximately constant displacement ductility
demands that are equal to the value ofZ at periods
Gravity loads acting through lateral displacements
affect the distribution of moments in a structure. For a
bridge column, the effect is illustrated in Figure RC3-15.
There is an increase in the base moment relative to the
value due to lateral load alone, and the moment distri-
bution is changed slightly. For practical cases, the shift in
the moment distribution in a column is not important
and can be ignored. Where a column responds inelasti-
cally, the P-A effect does not add to the base moment,
because the moment is limited by the moment capacity.
Instead, the presence of P-A moment reduces the lateral-
load resistance. The reduction increases with increasing
lateral displacement, which may in some cases result in

76 BDS Recommendations, Section 3: Loads ATC-32


(a) Structure (b) Moment (c) Moment (d) Combined

and Loading due to V due to P Moments

'{, +---
-- ---- --- -- --
v =Yield Shear (P- Ll not considered)

Vp =Yield Shear (P- Ll considered)

0 =Yield displacement

Figure RC3-15 Static aspects of gravity load acting

through lateral displacement for a cantilever.

instability. Section 3.21.15 describes an approximate

procedure for taking this effect into account in design.
Design forces outside yielding regions, and design
shear forces within flexural plastic hinges, are calculated
using the capacity design principles of Article 3.21.14,
except for restrainers, which are covered in Article
3.21.12. Elastic Structures C3.21.11.2 Elastic Structures

Design strength values shall be at least equal to the In designing an Elastic Structure, it may be anticipated
forces and moments obtained from Articles 3.21.5 or that elastic or nearly elastic response will result. How-
3.21.6. Where inelastic response is likely to occur at ever, it must be realized that a great deal of uncertainty
greater than the design loading, the capacity design exists in the definition of the design ground motions and
approach of Article 3.21.14 will be applied to avoid analytical models. For this reason, it is prudent to
nonductile response modes if possible. Design forces in assume the possibility of some inelastic response.
restraining elements shall be determined according to Aspects of capacity design may be appropriate, and
Article 3.21.12. moderate levels of ductility should be provided by
proper detailing and proportioning.

3.21.12 Restraining Features (3.21.12 Restraining Features

Positive longitudinal restraint shall be provided Procedures for the design of restraining features are the
between adjacent sections of superstructure at all inter- subject of current research. At the time of this writing,
mediate expansion joints. Restraint shall be provided no consensus on design approach has been reached. The
by hinge restrainers or other flexible, single-direction proposal is to retain the current Caltrans procedure,

ATC-32 BDS Recommendations, Section 3: Loads 77


restraining devices that limit superstructure displace- pending future research findings and consensus-build-
ment. ing efforts.
Additionally, designs that rely on restrainers only as
a secondary system are encouraged. These include
superstructures capable of cantilevering from support-
ing piers rather than relying on support at the seat and
superstructures with conservatively long seats that rely
on seat length rather than restrainer resistance as the
primary means of avoiding unseating. Restrainer Design Forces

Seismic forces in hinge restrainers and other similar

devices shall be determined using the Equivalent Static
Analysis method (Article 3.21.5). The longitudinal stiff-
ness of one adjacent superstructure frame including any
restraint at the abutment, gaps at joints, and gaps in
restrainers shall be considered when determining the
total stiffness of the frame moving away from the joint.
Only one span at a time shall be considered when ana-
lyzing multiple simple spans. Shear Key Design Forces

Seismic forces in shear keys and other "fixed" restrain-

ing devices shall preferably be determined using the
Elastic Dynamic Analysis method (Article 3.21.6). Single-Span Bridges

Restraining features for one-span bridges will not nor-

mally require detailed analysis. However, any connec-
tions between the span and the abutment shall be
evaluated by the Equivalent Static Analysis method
(Article 3.21.5). One-span superstructures fixed to the
abutment in the transverse direction may be assumed
to have a zero period of vibration in that direction. Vertical Restrainers

Hold-down devices shall be provided at all supports

and intermediate hinges where the vertical seismic force
opposes and exceeds 50 percent of the dead-load reac-
tion. In this case, the minimum seismic design force for
the hold-down device shall be the greater of

(a) 10% of the dead-load reaction or

(b) 1.2 times the net uplift force.

78 BDS Recommendations, Section 3: Loads ATC-32


3.21.13 Seismic Design of Bent and Pier C3.21.13 Seismic Design of Bent and Pier
Foundations Foundations General General

Bent and pier foundations shall be designed for the Design should account for at least the minimum of (1)
lesser of forces resulting from seismic plastic hinging the actions associated with plastic hinging in the fram-
(Article 3.21.14) or dead loads plus the elastic ARS ing members and (2) the elastic forces calculated for the
forces. design loading. Where elastic forces are used, the
designer is cautioned that actions larger than design Pile Deformations actions are possible if ground motions exceed the design
ground motion or if the analysis model is in error. The
Pile design shall consider the consequences of deforma- ARS spectra represent mean response quantities, so
tions due to seismic ground distortions. larger values should be anticipated. Approximations
required in modeling bridge structures do not necessar- Other Design Requirements ily produce conservative results. Some judgment in
deciding foundation design actions is necessary. It is
The requirements of Section 4 shall be satisfied. generally preferable to design for the plastic hinging
forces, except in unusual cases.
Article specifies that the relevant specifica-
tions of Section 4, Foundations, shall be satisfied in
addition to the specific requirements of Article 3.21.13.

3.21.14 Capacity Design C3.21.14 Capacity Design

Whenever feasible, structural elements and actions Capacity design is a procedure to control the locations of
shall be designed to remain in the nearly elastic range inelastic action in a structure. The procedure involves
under extreme loading conditions by considering the several steps, as follows:
inelastic force and moment capacity of the ductile ele-
ments in the structural system. 1. Locations where inelastic response is intended to
occur are identified. These locations are commonly
plastic hinges in columns. This step is required by
Article 3.21.3.

2. Plastic hinges are proportioned for the design loads.

This step is required by Article 3.21.11.

3. Plastic hinges are detailed to ensure ductile

response. This step is required by Articles 8.17 and
8.18 and Section 10.

4. Plastic moment strengths are calculated considering

actual proportions and expected material over-
strengths. This step is required by Articles

5. The structure is reanalyzed to determine the inter-

nal forces that will develop throughout the structure
when the plastic moment strengths are reached.
This step is required by Article

6. Design strengths of members and their connections

outside plastic hinges, and design shear forces in
plastic hinges, are taken as equal to the forces from

ATC-32 BDS Recommendations, Section 3: Loads 79


step 5, and these parts are designed to have reliable

strength exceeding the design strengths. This step is
required by Article

The procedure relies on adequate knowledge of

plastic hinge strengths, lateral and vertical load distribu-
tions, and reliable strengths to resist design actions cor-
responding to plastic hinging. While there are
uncertainties associated with each of these, capacity
design is believed to provide an adequate measure of
safety against nonductile or cosdy damage associated
with reaching strength in unintended locations of the
structure. AppLicability C3.21.14.1 AppLicability

Capacity design shall be applied to Full-Ductility Struc- Full-Ductility Structures and Limited-Ductility Struc-
tures and Limited-Ductility Structures. Capacity design tures are expected to develop lateral load strengths in
shall be considered for Elastic Structures in which load- plastic hinge regions and to require ductile response.
ings greater than the design loading are likely to result Capacity design is required for these structures to pro-
in inelastic action. vide reasonable assurance that a predictable and ductile
response mechanism will develop in the design event.
Given uncertainties in definition of the ground motion
and analysis model, it is possible that structures
designed for elastic response will be loaded beyond their
strength and require ductile response. Therefore, capac-
ity design should be considered for these structures as
well. Capacity Design Forces C3.21.14.2 Capacity Design Forces

The structure shall be analyzed under lateral forces that The analysis associated with capacity design should con-
produce likely plastic mechanisms of the structure. sider all reasonable load combinations. For single-level
Gravity loads shall include dead loads. Live loads shall structures, the lateral force distribution usually can be
also be considered where their effects are significant. assumed to be a simple pattern of loads applied at the
Locations of plastic hinges shall be clearly identified superstructure level. Gravity loads should be in place for
and shall be consistent with the intended structural the analysis. The analysis should consider the likelihood
action, as defined in Article 3.21.3. Probable plastic that biaxial lateral loading is present. In general, it is
moment strength values in columns, pile shafts, and acceptable to assume that while the structure is displaced
pier walls shall be defined according to Article in one direction to the maximum displacement it is dis-
Inelastic Static Analysis, as described in Article 3.21.7, placed to about 60 percent of the maximum value in the
can be used to satisfy the requirements of this orthogonal direction. The extent to which biaxial lateral
paragraph. action should be considered will depend on the unique
characteristics of the bridge. Design Strengths C3.21.14.3 Design Strengths

Design strength values of members and their connec- It is common to make simplifying assumptions when
tions outside plastic hinges, and design shear forces in analyzing a structure for the plastic mechanism. For
plastic hinges, shall be equal to the forces and moments example, it is common to assume that columns in sin-
obtained from the analysis described in Article gle-column bents respond as ideal cantilevers under transverse loading. As shown in Figure RC3-5, the actual
conditions may be different, and they may pose a more

80 BDS Recommendations, Section 3: Loads ATC-32


critical loading. These aspects should be investigated


3.21.15 P-f). Effects (3.21.15 P-f). Effects

Dynamic effects of gravity loads acting through lateral Tendencies toward tall bridge structures supported on
displacements shall be included in the design by use of relatively slender columns have led to concerns about
Inelastic Dynamic Analysis as described in Article dynamic stability because ofthe so-called P-f). effect. Tall
3.21.8, except the effects may be ignored where the fol- structures have relatively long periods, and therefore
lowing relation is satisfied. tend to have relatively low lateral-load strengths and rel-
Vo au atively large lateral displacements under earthquake
w ~4 H R3-2 loading. Because the lateral load strength is low and the
lateral displacements are high, these structures are
in which Vo = base shear strength of the frame obtained thought to be particularly vulnerable to the effects of
from the plastic analysis of Article 3.21.14, W = dead gravity loads acting through lateral displacements; that
load ofthe frame, au =maximum design displacement is, the P-f). effect.
of the frame from Article, andH= the maxi- Figure R3-15 defines the static aspects of the P-f).
mum height of the frame. effect. For P > 0, the axial load P acting through the lat-
eral displacement produces a moment at the base of
the cantilever equal to the product pa. For a column
with base moment strength equal to Mp, the lateral load
strength in the presence of gravity load P is given by
M p pa
Vp = - - -
It is seen from this expression that the lateral load
strength is decreased by the P-f). effect. It may also be
seen that P-f). effects are greatest for structures with
small base-shear strength (MpIL) and structures with
large lateral-drift ratios (aiL).
The P-f). effect tends to weaken a structure as dis-
placements increase. Thus, there is a tendency, once a
structure yields in one direction, for it to continue to
yield in that direction, leading to progressively worsen-
ing damage. Near-fault, pulse-type or long-duration
ground motions can be particularly damaging because
both conditions can accumulate inelastic deformations
in one direction.
Studies demonstrate that P-f). effects are strongly
dependent on the hysteretic model (Mahin, 1991; Priest-
ley, 1993). These studies conclude that steel structures,
which tend to have bilinear response characteristics, are
more susceptible to P-f). effects than are reinforced con-
crete structures with stiffness-degrading characteristics.
Structures with inherent material strain-hardening tend
to be less susceptible to P-f). effects than structures with
non-strain-hardening behavior.
It may be shown that P-f). effects are worse for struc-
tures supported on flexible foundations. Therefore, it is
important to include soil/foundation flexibility in the
P-f). evaluation.
Equation R3-2 presents a simple procedure for
checking if P-f). effects are likely to be significant. It is
based on results of numerical studies (Mahin, 1991;

ATC-32 BOS Recommendations, Section 3: Loads 81


Priestley, 1993) that indicate negligible P-L1 influence

when the inequality is satisfied. The results were
obtained using single-degree-of-freedom oscillators.
The extension of these results to multiple-degree-of-
freedom bridge structures is based on judgment. When
the inequality of Equation R3-2 is not satisfied, either
the design should be modified to satisfy it, or Inelastic
Dynamic Analysis should be used to directly evaluate the
P-L1 influence.

82 BOS Recommendations, Section 3: Loads ATC-32

Section 4



Unless otherwise noted, capacities and loads referred to

in Articles 4.1 through 4.4 are for Service Load Design
for nonseismic loads. Article 4.5 describes the require-
ments for seismic loads (Group VII Loads).

Articles 4.1 through 4.3.3 not modified.

4.3.4 Load Capacity of Piles

Article not modified. The values of each of these cases shall be

determined by making subsurface investigations or
tests and by referring to other available information.
Consideration shall also be given to:

(1) The difference between the supporting capacity of

a single pile and that of a group of piles.

(2) The capacity of the underlying strata to support

the load of the pile group.

(3) The effects on adjacent structures of driving piles.

(4) The possibility of scour and its effect.

(5) The transmission of forces from consolidating


ATC-32 BDS Recommendations, Section 4: Foundations 83


(6) The reduction in soil-bearing capacity due to cyclic

degradation effects, especially at poor soil sites.
(See Commentary 4.5.5 for additional discussions).

Articles through not modified. Uplift C4.3.4.6 Uplift Friction piles may be considered to resist an In the design of pile footings, the possibility of uplift on
intermittent but not sustained uplift. Resistance for any pile should be investigated.
standard piles may be equivalent to 40 percent of both End bearing piles should not be considered to resist
the allowable and the ultimate compressive load capac- an uplift force in excess of their weight unless special
ity, except that for seismic loads, 50 percent of the ulti- provisions are incorporated at the pile tip or along the
mate compressive load capacity may be considered. length of the pile, such as belling or socketing the end of
Design uplift capacities exceeding those above the pile or installing shear lugs along the length of the
must be demonstrated by a comprehensive site-specific pile. The uplift capacity of any such special provisions
analysis that considers the structural capacity of the should be verified by an uplift pile-load test.
piles and anchorage as well as the soil capacity (espe- Friction piles derive a major portion of their axial
cially uplift capacity based on skin friction). The maxi- load resistance incrementally along the length of the pile
mum uplift capacity for design shall be the lower of the by friction. Uplift capacity of a pile shall be determined
structural and the soil capacities. by a geotechnical engineer using site-specific soil and
pile data. The capacity value should be checked against
the structural capacity of the pile as well as the capacity
Articles through 4.4.9 not modified. of the connection detail. The details of Caltrans standard
Class 45 and 45C piles, Class 70 and 70C piles, and 16-
inch cast-in-drilled-hole piles are adequate for an uplift
force equal to 50 percent of the ultimate compressive
axial load capacity for earthquake loads. When the 50-
percent value is exceeded, the structural capacity of the
pile and the connection details must be designed on a
project-specific basis. If necessary, an uplift pile-load test
can be conducted to determine the capacity.
The details for Caltrans standard Class 45 and 45C
piles, Class 70 and 70C piles, and 16-inch cast-in-
drilled-hole piles are adequate for an uplift force of 40
percent of the ultimate compressive axial load capacity
for sustained loading and 50 percent for short duration
earthquake loading. When this uplift force is exceeded,
these piles must be specially designed, including the
design of the connection to the footing.
Eccentric load tests on groups of piles capped with a
rigid footing have indicated that the exterior piles have
higher reactions than do the interior piles and that cor- .
ner piles have the highest reactions. Because it is not
practical to control the maximum uplift resistance pro-
vided by soil friction (i.e., build in a load fuse), it is
important that in the design of a pile footing, the
demand for uplift resistance at any pile be limited to the
structural capacity of the pile and its connection.
Because of various uncertainties including the magni-
tude of earthquake load and actual soil capacities, a

84 BDS Recommendations, Section 4: Foundations ATC-32

Bridge Design Specifications Commentary

higher factor of safety should be adopted in the design of

connection details so that the probability of achieving
the desired soil failure mode is increased (see Commen-
tary to Article 4.5.1).


4.5.1 Seismic Design Philosophy C4.5.1 Seismic Design Philosophy

The effect of foundation and abutment stiffness and The basic process in foundation design involves first esti-
capacity, based on the best estimate of site conditions mating the forces and displacements on a specific foun-
and soil parameters, shall be considered in analyzing dation component, then ensuring that the component
overall bridge response and the relative distribution of has the capacity to accomodate the forces and displace-
earthquake effects to various bridge components. ments. In the case of earthquakes, the most difficult
The unreduced ultimate capacity of foundations aspect relates to determining the appropriate level of the
and abutments, consistent with the performance crite- resulting forces and displacements. Unlike static loads,
ria described in Article 3.21.2, may be used to resist where forces and displacements are readily determined,
safety-level earthquake loading. response to earthquake loads depends on the dynamic
Seismic design requirements for typical short-span response characteristics of the overall bridge, which in
bridge foundations are described in Sections 4.5.4 turn is affected by the foundation stiffness.
through 4.5.6. For unstable soil sites (as defined in In many cases, the seismic demand of the founda-
4.5.3) or for unusual or major bridges, special studies tion obtained from an analysis is an artifact of the
and a more detailed site investigation program are dynamic model. For example, as is the case in structural
required. These studies shall include an evaluation of design, a linear response spectrum analysis using initial
seismic hazards and potential site instabilities, as well as foundation stiffness often predicts unrealistically high
soil-structure interaction studies to evaluate foundation foundation forces. Response analysis for foundation
kinematic interaction, differential ground motion, and design should include sensitivity analyses to evaluate the
cyclic degradation effects. potential variations in soil behavior that can occur dur-
ing the duration of an earthquake.
Another aspect ofthe current design process is that it
does not consider some important loading mechanisms,
namely those associated with ground movements rather
than inertial loading of the structure. Review of perfor-
mance of foundation systems in past earthquakes (Lam,
1994) suggests that bridge foundations have performed
well during moderate (up to magnitude 7) earthquakes,
with relatively few cases ofbridge collapse due to failure
of foundations. However, past performance records also
indicate that foundation failures were the main causes of
bridge failure for very large earthquakes (Magnitude
above 7), as evident from the 1964 Alaskan earthquake,
the 1964 Niigata earthquake, the 1990 Philippines earth-
quake, and the 1991 Costa Rica earthquake. In all these
earthquakes, foundation failure related to lateral spread-
ing and loss of foundation bearing capacity associated
with soil liquefaction were the principal cause of bridge
collapse. Such a load case is not accounted for in present
design practice.
A discussion is presented in the ATC-32-1 docu-
ment (ATC, 1996) on an alternate foundation design
philosophy that emphasizes designing the foundation to
ensure a preferred mode of failure in case of overload
and the need to evaluate displacement aspects rather
than the magnitude of force. This approach is basically

ATC-32 BDS Recommendations, Section 4: Foundations 85


compatible with present-day practice of applying ductil-

ity design principles to superstructures. In addition to
improved performance, such an approach leads to more
economical foundations in high seismicity areas such as
California. Designing to ensure the preferred mode of
failure is very important due to various uncertainties
(i.e., with respect to loading as discussed above).
Irrespective of the design philosophy, a design pro-
cedure must address the following three requirements:

1. Structural Details. The foundation must be detailed

to prevent failure and preferably to achieve the
desirable failure mode, which in most cases is fail-
ure in the soil rather than in connections and struc-
tural components. Connections and structural
details must be adequate to provide a load path to
transfer the load to the soil. This aspect has added
significance at poor soil sites, where the potential
for ground movement is much more significant.
Improving the connection details and providing for
a more flexible foundation type improves the
chance of a bridge surviving soil liquefaction.

2. Bearing Capacity. Adequate bearing capacity must be

ensured to prevent excessive settlements. Cyclic deg-
radation effects must be included for the earthquake
loading condition. Past experience suggests that at
normal soil sites (i.e. other than poor soil sites as
defined in 4.5.3 Commentary), foundations designed
to the traditional factors of safety have sufficient
reserve to account for cyclic degradation effects.
However, foundations at river crossings and poor soil
sites (liquefiable and soft, sensitive clay sites) have
experienced bearing capacity failure during large
earthquakes. Cyclic degradation effects need to be
accounted for when assigning the soil capacities as·
part of foundation design for poor soil sites.

3. Tolerable Displacement. Attention needs to be placed

on displacement aspects in addition to force and
capacity issues. Some criteria have been developed
by Moulton et al. (1985) and Duncan and Tan
(1991) for service-level loads. They can serve as a
conservative presumptive criteria for earthquake
loads. Higher values can be used, based on evalua-
tions conducted for a specific bridge. On the basis
of the work reported by Moulton et al. and Duncan
and Tan, the presumptive tolerable foundation
movement criteria are as follows:

... Angular distortion: up to 0.008 radian

... Lateral deflection: up to two inches

86 BDS Recommendations, Section 4: Foundations ATC-32


Displacement criteria developed from an analysis of

site-specific structural capacities and configurations shall
be used where possible, in place of the above criteria.

4.5.2 Site Investigation C4.5.2 Site Investigation

A site investigation program shall be conducted to pro- Present Caltrans practice involves site-specific investiga-
vide adequate information for proper foundation tions for every bridge project. Some general information
design. on the basic requirements of geotechnical programs can
be found inAASHTO (1978) and FHWA (1982).
The importance of the site investigation program
becomes more critical when seismic considerations are
included in the foundation design. The following com-
ments discuss the elements of a thorough site investiga-
tion, with special reference to current Caltrans practice.
Presenting an accurate, standardized description of
the site soil conditions on the log-of-test-boring (LOTB)
is the first step in proper foundation design. The LOTB
should describe soils according to the Unified Soil Clas-
sification System presented in ASTM D-2487. The first is
a field visual inspection, which is followed by conduct-
ing index tests (i.e., grain size and Atterberg limits) in
the laboratory. Data from these tests are used to correct
the soil description on the field boring log. Such a labo-
ratory index test program should be conducted to
ensure proper classification of the soil type on the LOTE.
LOTBs for older bridges designed by Caltrans very
often provide only a generic description of the soil type
without blowcount data. The LOTBs for bridges built in
recent years most often have adopted the Unified Soil
Classification. However, it sometimes appears that the
LOTBs are based on visual inspection and lack a parallel
laboratory program to verify the soil type description.
At liquefiable and soft-soil sites (see Commentary
to 4.5.3), extra effort is required to ensure that the data
obtained from the site investigation program are mean-
ingful. The following recommendations cover the spe-
cial considerations for such sites.

• Using the appropriate drilling equipment is most

important for liquefiable and soft clay sites. As dis-
cussed more fully below, rotary-wash boring with
drilling mud is the most reliable drilling method for
such sites.

• For loose silts and soft clay sites, the use of a thin-
wall, selby-tube push sampler can enhance the
chance of recovering more undisturbed samples for
laboratory testing.

• Other less conventional tests, such as cone-pene-

trometer and geophysical shear-wave tests should be
considered for such sites.

ATC-32 BDS Recommendations, Section 4: Foundations 87


• There is more of a need to combine the expertise of

both a geologist and a geotechnical engineer to pro-
duce the LOTB for such sites, to include both the
geologic history and the mechanical-property infor-
mation of the soil layers.

Site investigations shall be conducted by qualified geo-

technical engineers and geologists. A foundation report
shall be prepared that addresses the following seismic
design issues:

• Active earthquake faults affecting the site

• The potential for surface fault rupture

• The extent of variation in soil conditions along the


• Site classification in relation to standardized elastic

response spectra or the appropriate site-specific
response spectra, as defined in Article

• Potential for ground liquefaction

• Potential for ground setdement

• Potential for lateral ground movement

• Slope stability

• Ground water conditions

For liquefiable or soft soil sites (see Commentary 4.5.3

for definitions of these sites), the geotechnical report
shall also include the following information:

• Carefully conducted standard penetration tests

(SPTs) shall be carried out using proper equipment
and procedures with blowcount measurements at
five-foot intervals.

• Depth at which ground water is encountered shall

be measured, where applicable.

• Liquefaction strengths of saturated cohesionless

soils (silts and sands) shall be determined from nor-
malized SPT blowcount correlations.

• Soil index tests, including grain size distribution

and Atterberg limit tests shall be conducted.

• Where potentially liquefiable, saturated, nonplastic

silts are encountered, cyclic triaxial or cyclic simple
shear tests on high-quality, thin-wall tube samples

88 BDS Recommendations, Section 4: Foundations ATC-32


shall be conducted, where feasible, to determine liq-

uefaction strengths.

• Laboratory measurement of both the peak and the

residual undrained shear strength of saturated, soft,
cohesive soils shall be conducted, where feasible.

• In situ tests including cone-penetrometer tests and

geophysical shear-wave velocity measurements shall
be conducted at poor soil sites, where feasible.

. 4.5.3 Site Stability (4.5.3 Site Stability

Bridge sites shall be evaluated for potential instabilities There are four categories of sites that warrant additional
related to (l) soil liquefaction, (2) the presence of soft efforts to address earthquake-hazard issues. The follow-
clays, (3) slope hazards, and (4) fault crossing. When ing definitions apply to both the specifications and the
such potential instability exists, special studies shall be commentary:
conducted and structural and/or site-enhancement
measures shall be implemented to mitigate the effect of • Liquefiable sites-sites that contain one or more
these instabilities to the extent that the performance identifiable layers of potentially liquefiable deposits
criteria of Article 3.21.2 will be satisfied.
• Soft clay sites-sites that contain sensitive, organic,
very high plasticity, or soft/medium-stiff clays

• Slope hazard sites-sites for which known landslide

hazards exist, as inferred from past historical and
geological information or observable landslide scars

• Fault-rupture special study zone sites-sites where

there is a potential for surface ground displacement
due to fault rupture

• Unstable sites-sites with one or more of the above

four site conditions

• Poor soil sites-sites classified as either liquefiable

or soft clay sites

Liquefiable Sites

Table RC4-1 provides a set of criteria that is appropriate

for preliminary screening of liquefiable versus non-liq-
uefiable sites. With the exception of those sites that fall
into the very low liquefaction potential category, further
efforts are necessary to evaluate the liquefaction hazard
of the site by more detailed analyses. Past case histories
(Barlett and Youd, 1992) suggest that liquefaction can
extend to a depth of 60 feet. Therefore, emphasis should
be placed on liquefaction analysis for the upper 60 feet
of the soil profile.
When liquefiable sites (Holocene sand and silt sites)
are encountered, the key information to be developed
during the site investigation includes:

ATC-32 BOS Recommendations, Section 4: Foundations 89


Table RC4-1 Relationship of Geologic and Water Table Criteria and Liquefaction Susceptibility (Modified from
Tinsleyet al., 1985)

Depth to groundwater, in meters

Sedimentary Unit
0-3 3-10 10-15 >15
Latest ......................... Very hifh Moderate2 Low Very low
to high

Earlier ...................... High Moderate Low Very low

Late ........................ Low Low Very low Very low

Middle and early ........................ Very low Very low Very low Very low

Tertiary and pre-Tertiary ............... Very low Very low Very low Very-low

1 Areas are mapped as having very high susceptibility if fluvial channel and levee deposits are known to
be present; sediment deposited in other sedimentary environments is considered to have high susceptibility.

2 fluvial deposits having high suceptibility occur rarely and are not widely distributed; other sediments
are moderately susceptible to liquefaction.

• geologic information

• ground water data

• grain size information

• blow count data

The subject of liquefaction has been extensively

researched in the past 30 years. Detailed discussions on
liquefiable soil types can be found in Ishikara (1985);
Seed, Idress, and Arango (1983); and Seed et al. (1985).
Soil Borings. Properly recorded blowcount data are
key for assessment of the liquefaction potential of a
given sandy site. The use of proper drilling equipment
and standardized blowcount procedures are very impor-
tant for liquefiable sites. Rotary-wash borings are pre-
ferred over auger borings for such sites. The use of a
rotary-drill rig in conjunction with casing or drilling
mud has been found to be the most reliable method to
prevent cave-in problems and to enhance the chance of
recovering undisturbed soil samples and reliable blow-
count measurements. If a hollow-stem auger is used,
maintaining a water column inside the hollow stem
above the water table is very important in preventing
soils from running up into the stem when the center rod

90 BDS Recommendations, Section 4: Foundations ATC-32


is removed to insert the soil sampler and/or to record the

blowcount. In addition, the use of standardized SPT
equipment (e.g., the use of safety hammers) and proce-
dures are all key to obtaining meaningful blowcount
data to address the liquefaction problem. In situ tests
such as cone-penetrometer soundings are also valuable
for defining the soil layering and therefore the extent of
the liquefiable soil layers.
Geologic Information. Most of the sites that have liq-
uefied in past earthquakes contained Holocene soil
deposits. The soil layers should be correlated with the
geologic units on the LOTB for liquefiable sites. Table
RC4-1 presents some criteria for relating geologic
soil units and ground water data to liquefaction
Grain Size. Figure RC4-1 summarizes the grain-size
distribution of the soil types that are vulnerable to lique-
faction. Grain-size distribution curves should be devel-
oped for liquefiable layers. As shown in the figure, fine
sands (particle size ranging from 0.075 to 0.425 mm)
comprise the majority of the most liquefiable soils.
Medium sands (particle size ranging from 0.425 to 2
mm) and nonplastic silts (particle size less than 0.075
mm) are also relatively vulnerable to liquefaction.
Coarse sands (particle size ranging from 0.425 to 4.75
mm) are relatively less likely to liquefy than other sands
and nonplastic silts. Gravels and clays are generally less
prone to liquefaction.
Although grain size information is very important
for a liquefaction evaluation, additional information is
necessary if the soil has a high fines content (i.e., parti-
cles less than 0.075 mm in diameter). Both silts and clays
are classified as fines, but they have very different physi-
cal behavior and liquefaction characteristics. The
proportion of clay in the fines is very important for
determining the liquefaction potential of the soil. Seedet
al. (1983) discussed this aspect of soil analysis in detail
and recommended various rules for the conditions
under which silts and clays are liquefiable. For these
types of soils, Atterberg limit tests are very useful. Gen-
erally, if the soil exhibits sufficient plastic behavior such
that the plasticity index versus the liquid limit is plotted
above the ''A'' line, the soil can be regarded as non-lique-
fiable. Atterberg limit data are also required in the Seed
et al. (1983) procedure for screening liquefiable versus
non-liquefiable fines. Readers are alerted to new lique-
faction hazard screening criteria by Youd and Gummow
Reduction in Foundation Capacity. One of the major
damaging aspects ofliquefaction is the loss of founda-
tion capacity. Such loss in capacity should be accounted
for in design.
Lateral Ground Spread. In recent years, in addition
to determining whether the soil at a given site is liquefi-

ATC-32 BDS Recommendations, Section 4: Foundations 91


: Sands :
Clays and Silts ~-------------------,-----------------r--------~ Gravels
: Fine : Medium: Coarse:
O.075mm 0.425mm 2mm 4.75mm

/- r
/ ' ,
1 '
/ : / : X
" /' I , , / ' ,

/ I ' ': I:,,' : I I
··· ······ ·· . f I
· . · · · · · · · . · ····:·1..· · · · . · . · · · . · . · · . · ;· · · ·
' " ,
· ···;········.. ··/· · :· ·
I '.
'I' I ,
I ,
' V I

::c: : ," I l ::
: ,: Boundaries: /'[ : /
I ' for most: : : I
~ 0 I , . liquefiable sc}il ' I I

, -:
.. ..I• L
: /...
I 1
II '
, ' II II '


....Z : I " I
I ',
II , ,I
, " I
··· --;··.. ······ · · . ····.. ·········f··f ····l·····..····..··..······..·········· ·· ·····-[···..····.. ·j ········l·· ········/··!··.. ············.. ······· [

' I'

J:a B d' ,
p.. 'i:

oun ane~
for potentially


./.. : liquefiable soil!
/... : .1..

I ,
/ '
~ ,II
,,' / , /:1, I II
/ V :
;, I
. I

10- 1


Figure RC4-1 Limits in the gradation curves separating liquefiable and unliquefiable soils.

92 BDS Recommendations, Section 4: Foundations ATC-32


able, emphasis has been placed on evaluating the lateral

ground displacement potential associated with the liq-
uefied soil. Past case histories strongly suggest that most
bridge damage at liquefied sites has been related to
excessive lateral ground displacement. A number of
researchers have studied the lateral ground displacement
problem and empirical equations are emerging for
design applications (e.g., Bartlett and Youd, 1992).
These studies indicate that the average grain size, the
average fines content (proportion of the soil mass by
weight with particle size finer than 0.075 mm), blow-
count, earthquake magnitude, and site-to-earthquake-
source distance are important parameters affecting the
magnitude of lateral ground displacement. In addition,
it has also been found that the magnitude of ground dis-
placement is also very sensitive to the proximity of a free
face (such as a river bank) and the overall slope angle of
the ground (even a few percent ground slope can have
an effect). Figures RC4-2 and RC4-3 provide some indi-
cation of the topographic features that should be docu-
mented and measured as part of the site investigation.
For sites that have a factor of safety less than one with
respect to liquefaction, the extent of lateral ground dis-
placement should be assessed using the following equa-

Free face model:

Log(DH + 0.01) = -16.366 + 1.178 M - 0.927 Log R

- 0.013 R + 0.657 Log W
+ 0.348 Log TIS
+ 4.527 Log(100 - PIS) - 0.922 D50 ls

Ground slope model:

Log(DH + 0.01) = -15.787 + 1.178 M - 0.927 Log R

- 0.013 R + 0.429 Log S
+ 0.348 Log TIS
+ 4.527 Log( 100 - PIS) - 0.922 D50 ls

where log is common log (base 10)

DH = the horizontal displacement (m)

M = the moment magnitude of earthquake
R = nearest horizontal distance to the seismic energy
source (km)
W = the free-face ratio; i.e., 100 HIL (%)
S = the ground slope (%)
TIS = the thickness of saturated sands with blowcount
(NI)60::; 15 (m)
PIS = average fines content in TIS particle size < 0.075
mm (%)
D50 1S =the average grain size (mm)

ATC-32 80S Recommendations, Section 4: Foundations 93



W - FREE FACE RATIO - 100 HIL. (LIl percant)





W - FREE FACE RATIO -100 HIL. (1Cl percent)



Figure RC4-2 Definition of free face factors, Land H, and ground slope,S, for free-face ground-spread displace-



. ~-,+--"""'-;:~
Y1 CRE&f y :~

TOE '--- x -J i~
I llXl"'Yl 031'!





~r- 1lXl"V1 :;3

~: ~1 ~ I

~i TOE~ y
Ii:'"----~L:;;~=DlSPLNElENT==-- x~ TOE

S (%) -100(Y/Xl • (")-'llll(Y1/X1)



Figure RC4-3 Definition of ground slope, S, for long, uniform slope ground spread displacement.

94 BDS Recommendations, Section 4: Foundations ATC-32


The lateral ground spread models have been developed

using data from stiff soil sites in the western United
States and Japan. They are considered appropriate for
seismic sources and ground motion attenuation charac-
teristics of the western United States. The equations are
considered invalid for a site-to-seismic source distance
below a threshold value, given in Table RC4-2.

Table RC4-2 Minimum R for Lateral Ground Spread


M R(km)

6.5 0.25
7.0 1
7.5 5
8.0 10
8.5 25
9.0 50

Bartlett and Youd's lateral spread prediction model

is based largely on empirical data. One of its drawbacks
is that a very large statistical scatter was observed in the
empirical data. The presented equations merely repre-
sent the mean value prediction. The database upon
which these equations were developed indicate that dis-
placement magnitude can vary by a factor of two
between the mean value and the 92-percent confidence
level. Other researchers (e.g. Hamada, 1992) have
offered different lateral-spread prediction models based
more on mechanistic theory than on empirical data. The
Barlett and Youd models are relatively simple and
account for most appropriate parameters that appear to
affect lateral spreading.
Dynamic Settlement. The potential for dynamic set-
tlement should be evaluated at loose sand sites. Guid-
ance on dynamic settlement considerations is given in
Tokimatsu and Seed (1987) and Pyke, Seed, and Chan
Liquefaction Mitigation. A number of methods are
available to mitigate liquefaction problems. These

• Removal or replacement of liquefiable soils.

• Dynamic compaction. This method entails drop-

ping weights of 10-50 tons from heights of 50-150
feet. The weights impact grids ranging in size from
7 x 7 to 25 x 25 feet. The effective depth that can be
compacted depends on the size of the weights, the
height of the drop, and the soil type. The effective
depth is typically limited to 40 feet. Generally this

ATC-32 BDS Recommendations, Section 4: Foundations 95


method is suitable for clean sands and gravels at

sites free from existing structures.

• Vibro-compaction. This method involves inserting

and withdrawing a large vibrating probe. The probe
is inserted to the depth for which compaction is
required. Sometimes granular backfill is added and
densified during withdrawal of the probe. This
method is most effective for particle sizes larger
than about 0.2 mm.

• Vibro-replacement (Stone column). This method is

very similar to vibro-compaction, but gravel or
crushed stone is backfilled and compacted during
withdrawal of the vibrator to leave "stone columns"
about three feet in diameter. The columns are
placed in a grid with spacings ranging from 6 to 11
feet. In addition to the densification, the stone col-
umns also improve drainage and reinforce the soil.
This method can be used to treat soils with smaller
particle size that are not suitable for vibro-compac-
tion alone. It can also be used for silts and clays.

• Compaction grouting. For sites where vibratory

densification may be impractical because of poten-
tial damage to nearby structures, compaction grout-
ing can be used to inject a stiff soil-cement-water
mixture to form grout bulbs in the soil, which dis-
place and densify the ground.

Soft Clay Sites

New soil site classifications have been developed for the

National Earthquake Hazards Reduction Program
(NEHRP) provisions. A preliminary version of the new
provisions identified a Class F site condition for which
site-specific studies to determine the site-specific design
spectrum are mandatory. These Class F sites include a
number of soil conditions, as follows:

• Soils vulnerable to potential failure or collapse

under seismic loading, including liquefiable soils,
quick and highly sensitive clays, collapsible weakly-
cemented soils, etc.

• Sites with peats and/or highly organic clays exceed-

ing ten feet in thickness

• Sites with a major layer of very high plasticity clays

with PI >75%, exceeding 25 feet in thickness

• Sites with layers of very soft or medium-stiff clays,

exceeding 120 feet in thickness

96 BDS Recommendations, Section 4: Foundations ATC-32


Site Response. The potential for amplification of

ground motions from a distant earthquake at soft clay
sites was first recognized in the 1985 Mexico earthquake
and was seen again during the 1989 Lorna Prieta earth-
quake. A mechanistic explanation for the large site-
amplification effects observed for soft clay sites was
given by Bielak and Romo (1989). They attributed the
effect to the relatively linear stress-strain behavior and
small damping exhibited by clays for strain levels during
freefield site response. Another factor contributing to
the large amplification effects in 1985 and 1989 is that at
both Mexico City and the San Francisco Bay Area, the
soft clay is often underlain by bedrock at a relatively
shallow depth (approximately 200 feet). This configura-
tion gives rise to a large impedance contrast, which leads
to a pronounced site period and results in a higher
degree of resonance. Site-specific response analyses shall
be conducted at Class F sites. Data typically required for
site response analyses are shear-wave velocity profiles,
soil moduli, and soil damping versus strain amplitude
curves. Therefore, in situ tests such as geophysical tests
for shear-wave velocity would be desirable for such soil
sites. Collecting undisturbed samples for consolidation
tests to determine past stress history, index tests for
Atterberg limits, and initial and remolded (or residual)
shear strength and dynamic tests (cyclic triaxial and sim-
pIe shear tests) of the soft clays are important for charac-
terizing soft clay sites.

Slope Hazard Sites

Earthquakes have been recognized as major causes of

landslides. It is essential to make use of available geo-
logic information to evaluate potential slope hazards.
Geologic literature is often the source of information of
past landslide activities, geologic units that are landslide
prone, etc. Sites that have experienced landslide prob-
lems from other loading conditions (e.g., static gravity
and ground water) would also be vulnerable to landslid-
ing during earthquakes. Therefore, existing geologic
data should be used to identify whether there are land-
slide hazards at a given bridge site. At those sites that
have been known for landslide hazards, slope stability
analyses should be conducted. Seismic slope stability can
be evaluated using conventional pseudo-static slope sta-
bility analyses. The magnitude of earthquake-induced
slope movements can be addressed using Newmark's
sliding block analysis. Simplified equations (e.g., Frank-
lin and Chang, 1977) can be used to quantify the magni-
tude of slope movements in Newmark's sliding block
As discussed earlier, empirical equations have been
provided by researchers (e.g., Bartlett and Youd, 1992)
for determining the lateral ground displacements at liq-

ATC-32 BDS Recommendations, Section 4: Foundations 97


uefied soil sites. Such empirical procedures are consid-

ered reasonable to complement the above-discussed
deterministic slope stability procedure (Newmark's slid-
ing block in conjunction with pseudo-static slope stabil-
ity analysis). The deterministic procedure allows the
analysis to reflect site-specific soil data (e.g., residual
strength parameters) and the local slope configuration
and detailed ground motion information. It is more
refined than the empirical approach (e.g., Bartlett and
Youd, 1992). The empirical approach is appropriate for
addressing the lateral ground-spread problem for very
gently sloping ground at a site not immediately in the
vicinity of the slope embankment (i.e., several hundreds
of feet from a river bank). The deterministic approach
would be more reliable in the immediate vicinity of the
slope or embankment. However, as discussed earlier,
there is a significant level of overall uncertainty in all
current approaches used to estimate the magnitude of
lateral and vertical ground displacement. Also, there is a
significant uncertainty with respect to the appropriate
soil strength parameter (e.g., residual shear strength of
soils) to be used in such ground-deformation evalua-
tions. More research in the area of ground deformation
is needed. More detailed background information on
the subject has been extracted and included in the
ATC-32-1 document (ATC, 1996).

Fault Rupture Hazard Sites

There is California state legislation (the Alquist-Priolo

Special Study Zone Act) that forbids constructions and
development activities at sites that are known to have
fault rupture hazards from active faults, unless the
project is adequately investigated and evaluated. The
California Division of Mines and Geology (CDMG)
publishes and maintains a set of maps that designate
zones as having fault rupture hazards. Although Caltrans
is exempt from this legislation, it would still be appro-
priate to make use of the CDMG information to evaluate
the potential for fault rupture at a given bridge site. At
sites that are identified as special study zones, it would
be appropriate to conduct more detailed geologic map-
ping and investigations to clarify the relative location of
the fault trace in relation to the bridge structure. Infor-
mation developed from such investigation programs
should be provided to CDMG to enhance the database
of that agency.

4.5.4 Abutments and Wingwalls (4.5.4 Abutments and Wingwalls

The participation of abutment walls in the overall Prior to an earthquake, abutments and wingwalls
dynamic response of bridge systems to earthquake function as fill-retaining systems. Therefore, they are
loading and in providing resistance to seismically generally designed as retaining walls, based on the
requirements set forth in the Caltrans Bridge Design Aid

98 BOS Recommendations, Section 4: Foundations ATC-32


induced inertial loads shall be considered in the seismic Articles 1.1 through 1.6 and the earth-pressure load as
design of bridges. defined in Article 3.20 of the Caltrans Bridge Design
Damage to abutment walls that is allowed to occur Specifications. Although retaining walls are designed for
during earthquakes shall be consistent with the perfor- relatively small active earth pressures (say a static lateral
mance criteria described in Section 3.21.2. earth-pressure coefficient of 0.3), most free-standing
Abutment participation in the overall dynamic retaining walls not associated with other structures have
response of bridge systems shall reflect the structural performed well during past earthquakes (see further dis-
configuration, the load-transfer mechanism from cussion in 4.5.8). On the other hand, certain bridge
bridge to abutment system, the effective stiffness and abutments, especially skewed abutments, have been
force capacity of wall-soil systems, and the level of known to be highly prone to damage during earth-
expected abutment damage. quakes. The drastic difference in performance between
The capacity of abutments to resist the bridge iner- retaining and abutment walls can be attributed to the
tialload shall be compatible with the structural design fact that free-standing retaining walls are not prevented
of the abutment wall (i.e., whether part of the wall will from moving in ways that relieve the earthquake-
be damaged by the design earthquake) as well as the soil induced soil pressure, whereas abutment movements are
resistance that can be reliably mobilized. The soil capac- typically restricted to some degree by the bridge struc-
ity shall be evaluated based on an applicable passive ture. In addition, the inertial load of the bridge forces
earth-pressure theory. the wall to move into the backfill soil, creating a passive
earth-pressure loading condition on the abutment walls.
The magnitude of the passive earth pressure is generally
extremely high as compared to the active pressure (pas-
sive-pressure coefficient is generally over 30 times that of
the active-pressure coefficient). If a wall is designed for
the lower active pressure only, it would be highly vulner-
able to damage when subjected to the passive pressure
loading condition. Caltrans has adopted a design philos-
ophy that controlled abutment damage is acceptable.
Experience from past earthquakes indicates that follow-
ing such damage, emergency repairs are effective in
restoring the bridge to a usable condition within a short
Given that it is impractical to design abutments for
no damage in an earthquake, the most immediate need
regarding abutment design is to characterize the abut-
ment stiffness for dynamic response analysis of the
bridge, in order to capture the overall bridge displace-
ment amplitude and the load distribution to columns or
piers so that the integrity of the overall bridge can be
evaluated. The following aspects should be considered in
characterizing the abutment stiffness in the overall
bridge model.

• The designer should examine the connection

between the bridge and abutment walls. Many wing
walls are not tied structurally to the abutment back-
wall and therefore cannot mobilize any wing-wall
stiffness in the bridge model.

• The designer should evaluate whether the abutment

wall would be damaged during the design earth-
quake, to determine the portion of the wall height
that the bridge can rely on to mobilize the backfill

ATC-32 BDS Recommendations, Section 4: Foundations 99


• Flexibility of the abutment walls should be consid-

ered in determining the extent to which the wall
would be effective in mobilizing backfill resistance.
Simplified models of a cantilever beam on elastic
springs indicate that an 18-inch thick concrete wall
cannot mobilize the soil resistance beyond five feet
from the point of support. For typical box-girder
bridges, the Caltrans practice oflimiting the soil
resistance at the back wall to an eight-foot wall
height is reasonable. Soil resistance mobilized
beyond the effective height should be ignored in
bridge response analysis. For other bridge types
(e.g., slab bridges), a smaller wall height (i.e., the
effective depth below the soffit) would be appropri-
ate. Similarly, only the effective width of the wing
wall should be accounted for in developing soil
resistance to the inertial load in the transverse direc-
tion. The soil resistance of deeper back walls or
longer wing walls can be used if the structural con-
figuration justifies the assumption (e.g., non-canti-
lever walls).

• The designer should estimate the backfill pressure

capacity. Research at the University of California at
Davis (Maroney et al., 1994) showed that 7.7 ksf
average soil pressure capacity is reasonable for an
eight-foot wall height. The average unit soil pres-
sure capacity (i.e., 7.7 ksf) should be reduced lin-
early in proportion to wall height for wall heights
less than eight feet, in accordance with the effective
overburden pressure of frictional backfills. These
are some typical rule-of-thumb parameters for
assessing the backfill capacity for abutment design.
Although these parameters have been found to pro-
vide reasonable designs, more rigorous analysis pro-
cedures that can implement more site-specific soil
and structural data have been developed from
recent research. Some information from this
research is presented in ATC-32-1 (ATC, 1996).

• Site-specific backfill properties can be used to deter-

mine abutment capacity in lieu of the presumptive
values based on appropriate passive earth-pressure
theory. Some discussions on passive earth-pressure
theories are provided in ATC-32-1 (ATC, 1996).

• The designer should estimate the magnitude of dis-

placement required to mobilize the ultimate pas-
sive-pressure capacity. Clough and Duncan (1991)
suggested movements ranging from 0.0 1 to 0.02 of
wall height. The Ue. Davis data suggested a move-
ment of 0.006 of effective wall height. A wall move-
ment of 0.01 of wall height to mobilize the full

100 BDS Recommendations, Section 4: Foundations ATC-32


ultimate passive pressure would be reasonable for

most typical conditions.

• The displacement to mobilize the soil resistance

should be added to the size of the movement to
close the gap at expansion joints for seat-type abut-
ments when determining effective abutment stiff-
ness in bridge response analysis.

• Analysis results should be checked to see whether

the load distributed to the abutment has exceeded
the estimated soil capacity for the abutment. If nec-
essary, the abutment stiffness should be reduced
and the dynamic analysis repeated until the results
reflect the proper load distribution between the
abutments and the bridge columns.

• Similar procedures can be used to evaluate the

transverse stiffness of wing walls. Factors to adjust
for the flexibility of the wing wall and to account for
the combined effects of a pair of wing walls at each
abutment can be found in Caltrans Bridge Design
Aids 14-1 through 14-10 (Caltrans, 1989). Effective
wing wall width should be limited to five feet as dis-
cussed previously, unless a different value can be
justified for the specific structural configuration
(e.g., non-cantilevered wing wall).

• Past earthquake performance reveals that skewed

abutments are highly vulnerable to damage, espe-
cially at the wing wall that forms an acute angle with
the back wall (i.e., the acute wing wall). Therefore
the wing wall stiffness should be either reduced for
the acute wing wall, or ignored for highly skewed
abutments. In addition to reducing the effective
abutment stiffness in the transverse loading direc-
tion, the skewness of the abutment introduces sig-
nificant differences in the transverse abutment
stiffness in the positive versus the negative loading
direction. This effect could induce torsional
response in the bridge, leading to a significantly
higher level of displacement at the acute wing wall
than at the obtuse wing wall, and hence a higher
level of damage at the acute wing wall. The skew
angle at abutments should be reduced, even at the
expense of increasing the bridge length. This is
especially pertinent for long, curved connectors for
which large earthquake displacements and forces
are anticipated at the abutments. Adding CIDH
piles at the acute wing wall can also be used to min-
imize damage to skewed bridges.

• The Caltrans Design Aids Articles 14-1 through 14-

10 (Caltrans, 1989) recommend typical stiffness val-

ATC-32 BDS Recommendations, Section 4: Foundations 101


ues for standard piles of 40 kips/in per pile, with a

lateral capacity of 40 kips at abutments. This value
is reasonable and the complexities of most abut-
ment configurations render further refinements on
the lateral pile stiffness at abutments meaningless in
practice. Stiffness of the piles is generally very small
compared to the backwall and should be neglected
in the longitudinal abutment stiffness calculations
to minimize complexity in the analysis (e.g., adjust-
ments of pile stiffness for gaps at expansion joints,
accounting for breakoffs of part of the wall, or when
the wall is on bearing pads and not directly con-
nected to the pile foundation). The pile stiffness is
significant when compared to the transverse abut-
ment stiffness and should be included in the calcu-
lations where appropriate (e.g., where there are
shear keys to mobilize the foundation stiffness or
the wall is monolithic with the foundation).

The above requirements address the need for abut-

ment modeling to evaluate the integrity of the bridge
structure. They do not address the integrity of the abut-
ment system itself with respect to the design earthquake
load. For Important Bridges, where functionality of the
bridge must be ensured during and immediately follow-
ing an earthquake, the structural component of the
abutments must be designed to accommodate the
required earthquake load and the passive-pressure con-
ditions. Detailed guidelines on designing abutments to
prevent abutment damage require further development
and most of all, innovation in abutment configurations.
The use of structural approach slabs would increase the
chance of the bridge remaining functional following
earthquakes. Implementation of ongoing research find-
ings is encouraged to improve the performance and sur-
vivability of bridge abutments in earthquakes.

4.5.5 PiLe Foundations (4.5.5 PiLe Foundations

The following design requirements shall apply to Cyclic Degradation on Bearing Capacity. Adequate bear-
bridges on deep foundations including large diameter ing capacity must be ensured to prevent bearing capacity
drilled shafts, driven concrete and steel piles, driven failure of the pile foundation. The traditional safety fac-
steel shells filled with concrete, and cast-in-drilled-hole tor of two for service-level loads appears to be adequate
piles: for most sites without poor soil. There have been many
case histories of bearing capacity or excessive settlement
(a) Lateral foundation design forces associated with failure from past earthquakes at poor soil sites. This
Group VII Loading shall be based on either (i) plas- indicates that the traditional static factor of safety of two
tic hinging of the bridge column, (ii) linear may not be adequate to allow for cyclic degradation
dynamic response analyses using the appropriate effects. Therefore cyclic degradation effects should be
elastic response spectrum (e.g., ARS), or (iii) more explicitly taken into account in pile capacity evaluations
advanced nonlinear or linear dynamic response at poor soil sites. The Caltrans BDS pile design criterion
analyses. The governing foundation design forces should be interpreted that "after allowance for cyclic
shall be based on the maximum forces that can be degradation effects, the ultimate pile capacity should be
two times that of the compressive load required to resist

102 80S Recommendations, Section 4: Foundations ATC-32


transmitted to the foundation (See Commentary the demand for service-level load cases (ie., dead weight
3.21.7). plus live load)". Experience suggests that explicit cyclic
degradation evaluations should be carried out for the
(b) The capacity of pile foundations and their individ- following soil/pile conditions
ual components to resist seismic loading shall be
based on ultimate structural and soil capacities, • At river crossings or foundations in open water
consistent with the safety-evaluation earthquake
performance criteria described in Article 3.21.2. • At liquefiable sites

(c) Effects of loading from earth pressure generated by • At soft clay sites (sensitivity of the clay is four or
lateral ground displacements and dynamic settle- greater).
ment associated with liquefaction or soft-soil
response shall be accounted for at poor soil sites. • For friction piles where the pile tip is not embedded
in bedrock, or where a high proportion (over 50
(d) Stronger connection details and the choice of more percent) of the ultimate capacity would come from
compliant pile types, and detailing of piles for skin friction rather than from end-bearing.
potential in-ground hinges (e.g., at boundaries of
liquefied or soft soil layers) shall be evaluated at • For long and slender piles (pile length over 50 feet)
such poor soil sites. where the cyclic pile top displacement amplitude is
sufficiently large to initiate cyclic degradation ofskin
(e) The use of batter pile groups shall be based on friction (i.e., zero-to-peak cyclic pile top displace-
load-deformation analysis of the pile group config- ment in excess ofO.s inch). Past cyclic loading pile
uration. test data suggest that the mechanism of shear stress
reversal (or plastic slippage) at the soil-pile inter-
(f) Foundation stiffness shall be accounted for in the face, induced by rocking motion of the superstruc-
dynamic response analysis of the overall bridge ture is the key mechanism of cyclic degradation of
(See Commentary 4.5.5). the skin-friction component of pile capacity.

(g) Rotational and lateral displacements at founda- Lam (1994) presented some procedures for soil-pile
tions shall be consistent with the performance cri- interaction analysis, including aspects for rate and cyclic
teria described in Article 3.21.2 (See Commentary degradation effects that can be used in a comprehensive
4.5.1). soil-pile interaction analysis. However, such analyses
might not be practical for common usage. In lieu of soil-
pile interaction analysis, the factor of safety can be
increased to three to arrive at the allowable compressive
pile load from the ultimate pile capacity as determined
from conventional pile capacity procedures. Some infor-
mation soil-pile analysis is included in ATC-32-1 (ATC,
Uplift Capacity. While the presumptive values on
uplift pile capacities (based on structural capacity) typi-
cally assumed by Caltrans (ie., 50 percent ofthe ultimate
compressive capacity, as stated in Caltrans BDS Com-
mentary, Article is reasonable, the uplift capacity
of a pile can vary significantly. Therefore, site-specific
evaluation must be conducted to determine the uplift
pile capacity. The uplift soil capacity must then be
checked against the capacity of the pile connection
details and the structural capacity of the pile.
Connection Details. As stated in Caltrans BDS Com-
mentary, the details for the standard Class-45 and 45C
piles, Class-70 and 70C piles, and 16-inch cast-in-
drilled-hole standard Caltrans piles are adequate for an
uplift force equal to 50 percent of the ultimate compres-

ATC-32 BOS Recommendations, Section 4: Foundations 103


sive axial load capacity (two times the allowable pile

loads of 45 and 70 tons). When the 50 percent value is
exceeded, the structural capacity of the pile must be cal-
culated from a project-specific analysis. Connection
details must be designed on a project-specific basis as
well, and shown on the design plan. As discussed in the
section on displacement-based foundation design in
ATC-32-1, the connection details should be strength-
ened whenever possible to exceed the uplift soil capacity.
This enhances the chance that a ductile soil failure mode
will occur.
Lateral Pile Stiffness. The subject oflateral pile stiff-
ness has been heavily researched during the past 20
years. There is no lack of procedures for conducting soil-
pile interaction analysis. However, many procedures are
not sufficiently simple for practical applications. The
state-of-practice is to use beam-column models sup-
ported by nonlinear lateral springs (p-ycurves). The
most widely adopted p-y procedures are those developed
by Reese et al. (1974) for sand and Matlock (1970) for
clays. Linear subgrade stiffness values, as recommended
by Terzaghi, have also been widely used by geotechnical
In the past, Caltrans has recommended some pre-
sumptive stiffness values (BDS Article dated Jan-
uary, 1993) for service load design at a pile deflection of
one-quarter inch. These stiffness values can also be used
for earthquake design, as summarized in Table RC4-3.

Table RC4-3 Presumptive Pile Stiffness Values (as

Derived From Caltrans BDS

1/4 in. defl. Stiffness

Pile Type Load (kips) (kip/in)

16" CIDH 13 52
IS" driven concrete 13 52
12" driven concrete 5 20
12" or 10" steel flange 5 20
8" steel flange 4 16
Timber 5 20

The tabulated stiffness values were based largely on

pile load tests. They tend to underestimate the pile stiff-
ness in view of the fact that most pile footings are
embedded some distance below grade (with the pile top
typically under five feet of overburden).
Lam has developed a series oflinear pile-head stiff-
ness design charts to expedite soil-pile analysis, as shown
in Figures RC4-4 through RC4-12. The stiffness charts
arise from the basic beam-on-elastic subgrade reaction
theory first proposed by Terzaghi (1955). Such proce-
dures have been widely used for pile design for over 30

104 BDS Recommendations, Section 4: Foundations ATC-32

28· 29'" 30" 38·



Recommended by Terzaghi, 1955

(After O'Neill and Murchison, 1983)

oOJ ~------J----__L . - - - - - - + _ - _ ----t----r------t

-f---------i-------if--~,;£--- ""L------+-----i

a 4------+----~~---_+----...;..----...,

o 20 YO 60 80 100

Rel.ative Density (Percent)

Figure RC4-4 Recommendations for coefficient of variation in subgrade modulus with depth for sand.

ATC-32 BOS Recommendations, Section 4: Foundations 105


0 2 4 8 15 30





Cl \Q
II: 2
III ////
Z ,./
0 Q
i= ,,"'bOW
II: /""..~~;9
« ~~
U. Q

o 1 2 3 4 5

Figure RC4-5 Recommendations for coefficient of variation in subgrade modulus with depth for clay.

106 80S Recommendations, Section 4: Foundations ATC-32


I.' ...-:
j..- ~ r..,....-
Z v I---' ::,....-" V
::::::: b;:::j;.o
co ~ ~ V I--'"


..........-:: ~ l--:
~ ~ l\J,..
~J,.. V
V I---'

----- ~ ~ .....

~ .....


.,..- ~
.,..- .....
V" ...-":r\
~ '"\ \ 1\[\
\\' ~\ 1\/\ ~
.....-V ......

.,..- ~
I--' V
~ ....... j..-

a: f 200
0 f 150
u. ...... 1./ I,..- ....... I" \
zu. ...-
.,..- !-'" -- .....
........... 1\ \ \
f = 100
f = 80
f =60

i-"'" f =40
C"l ........- V
u. .............. ............... f =20
C/) ~ f = 10
...J f =5
<{ f =1
a: l..-""'" \
f =0.5
f= 0.1
N Coeff. of Variation of Soil Reaction
Q Modulus with Depth, f (LB/IN 3)
'" I
10 10 1011 1012

=KIi - ~
,I Ka
= 0.41
I ~1\1/5
T= \T7

Figure RC4-6 Lateral stiffness of free-headed piles.

ATC-32 BDS Recommendations, Section 4: Foundations 107

l.----: t::~
f = 200 ..... ~~
~ b:=:
~~ ~ ...
V I--" ~
?""4 -:
. f
150 ~
100 ~ ./ ./'
:c f = 80 ~ V-~ t...-"'" ~ ~

0 f = 60 "- "
~ ~ ...... V v .,.,... ....
w V ~ i::=-I,.o v V
~ I,.o~
L,......-:: t::::: l...-
~ t:::: i--'"
......~ ."
f-'f-' t>< vf"-, ~

...... ~

u. .,......

i= ~
L...- ~
.,.... K" p..: D
.,....K f"-,i" .....
i"-~ V
."V ~

V .....
« I-"'" j;'
i--'" I" = 40
z V
0 ~ V V i--'"
I..-' J>" f = 20
f = 10
f = 5
-I ~ f = 1
"" "
CJ) ~ J..-"
Z .,.... f = 0.5
" f = 0.1
V Coeff. of Variation of Soil Reaction
t') V
V .
Modulus with Depth, f (LB/IN3 )
, •• I .
10 10 1011 10 13

Pt =KS'5+ Kss·e
, /
Mt = K 5+ se Ke' e
K =1.0765·E·1
T =\ f I

Figure RC4-7 Coefficient for lateral pile head stiffness (fixed head pile lateral stiffness).

108 80S Recommendations, Section 4: Foundations ATC-32

~ v~t::
« "¢ ~~ ~ ,,-~~

"l~ '"

~ ~~ ..-
~ I
~~ "'->' ~
f 200
z~ r/ "- f 100
~ 'l "'- "- "'- f'f = 60
Cf) 'l ""
:;-' ~ ~ f'f = 10
...J V.h :;-' ,/ ~ f'f = 5
« =
~ ~
V f'-f 1
z I"~~ =
H 0.5
0 ~
- f = 0.1
O i'7/ VV
a: ' / p /'
/' Coeff. of Variation of Soil Reaction
Modulus with Depth, f (LB/IN 3)

10 10 1011 10 12

Pt =K6-0+ KIl9-a
Mt =KooO+ Ke-a
K =1.499- E·I
I T3
I ~1/5
T =\ f I

Figure RC4-8 Coefficient for pile head rotation.

ATC-32 BDS Recommendations, Section 4: Foundations 109

...J "l'"A
~ 1''''-:::

~ [...01..-

co .-I
..... /'
w ~A ~ 1./ ......
Z v KP. ...... V .;-
~ ~ ~~
fQ$ ~ Kt'-- k" .;- .....
::J 7 ~ V.) '.:::: f 100
D. ~ ~~ V/, ...... 1A" V f = 80
k:%:; f¢ r;:; 'i ..... [...0 )
~'"i"- "'"
~" f = 60

I..-'" [...oJ.-

= ~ r::
V V f = 40
/ . r:::
CI) \C 1/ [...oJ.- i"- f = 20
.-I f = 10
a: f = 5
() =
/' 1/ 1"-. "-. f 1
f = 0.5
./ ./
v./ [...0
f = 0.1

V Coeff. of Variation of Soil Reaction

./ Modulus with Depth, f (LB/IN 3 )
• II 'I I

10 10 1011 10 12

Pt = Ko 5+ Koeoe

Mt =K065+ Keoe

Koe =0.999 Eol 0

l.lli1/ 5
T =\ f J

Figure RC4-9 Coefficient for cross-coupLing stiffness term.

110 BDS Recommendations, Section 4: Foundations ATC-32

............ 0
............ O'
.-J - - - - S'
------...0 --- 10' L-- .- !-
0 l---
.- '-
- V
-- vv
0 ~
- f = 100
::x: ~~..... -- -
-- :.-- -- ""- ./

-- -- -- ----
"--" I,{) -- .- .-- ~
;:; -::::
---- --
i.-' .-
-- V


'" I,....-,
l- t...--
L-- . .-1- f'-, k::::'";: vI-'
.- ~ -- vl--"
.:::,~ ~
« 0 l--- I-
-i f = 10 f-

« ..- I-"
f =1
" "- f = 0.1
:z: ,.-1-
« ,.,., .-
------ vI-" Coeff. of Variation of Soil Refction
I- a
V "'adulus wifh Depth, f (LB/IN )
10 10 1011 10 12

PI = K& . 0 + K 6 .·8 u ·& +K.·8



Figure RC4-10 Comparison of fixed head pile head stiffness at various embedments (0, 5, and 10 feet).

ATC-32 BDS Recommendations, Section 4: Foundations 111

, >

:z: - - - - 5' -6 !.t
---10' ~p ~~ ~
Cl 0

~V ~ V v
<! -<
~ 0
ro .......
:z: .~v ~, /' "
/ / / Ih
v J? i'-~

::::" "'~r-..
V') (])
:z: ......
L.... -'/ f 100
l.L.. '/.~/ "- "-
V/ #/
.,... ~ "'-- 'f = 10
~ '/.fJ .L. ~
« ~. 'f = 1
.,,:, l% ~~ ~ ~
f = 0.1 r:-
<X: v'/ ih'.
I- ./
0 V
a::: 1/// h- I--
l'- ~ Coeff. of Variation of Soil Refclion
0 l..Iodulus with Depth, f (LB!IN )
....... , 'I

10 10 10 11 10 12

PI = Ki · 8 + KH·e

~= Ki, ·8 + K.·e

Figure RC4-11 Comparison of the rotational stiffness coefficient at various embedments (0, 5, and 10 feet).

112 80S Recommendations, Section 4: Foundations ATC-32

- - - - S'
--- 10' ,- i--v
CD a::>
..0 ./--/
~ 1/ __ '/
1:72 /. ~
V y
./ "<: h
~- /
::;..... "" '>;

I-- .......
/ "-
......- --;;; -- 1/..- / ~
f = 100
z /' 1..- / .... ~
-l ..-[7 L....- v . ..-: I"..... v 4) ~ r--.
=> co
v- /'
),,-- b V , - ~

~ f = 10 -
v..-- /
'/ "-
f =1
..... / "-...
U ./ ..... ~,-/ f = 0.1
ID :/" Coeff. of Variation of Soil Refclion
a Modulus with Depth, f (lB/IN )
....... I I I

10 10 10 11 10 12

Pt= K,' 8 + K,.·e

~ = K,.' 8 + K. ,e

C- pt .&

Figure RC4-12 Comparison of the cross-coupling stiffness coefficient at various embedments (O, 5, and 10 feet).

ATC-32 BDS Recommendations, Section 4: Foundations 113

Bridge Design Specifications Commentary

years, and guidance is available regarding the input soil

parameters to use in the design charts (e.g., NAVFAC,
1986). The pile stiffness values presented in the design
charts compare favorably with the presumptive stiffness
values tabulated in Table RC4-3. The design charts pro-
vide stiffness values for various pile-head embedment
and boundary conditions (these factors significantly
affect pile stiffness). The lateral pile stiffness should be
compatible with the pile-cap connection. For example,
an assumption of a free pile head may be appropriate for
timber and steel piles, whereas an assumption of a fixed
head may be appropriate for concrete piles, based on the
Caltrans standard pile details. In addition to the lateral
stiffness, the charts can be used to estimate pile head
moment versus lateral load (e.g., for a fixed-head pile).
Pile Analysis for Liquefied Soil. Recent unpublished
centrifuge tests conducted by Ricardo Dobry at the
Rensselaer Polytechnic Institute indicates that fully liq-
uefied sands (from freefield liquefaction effects) has a
residual strength of about 10 percent of the initial p-y
curve resistance, as determined by Reese's p-y proce-
dure. The 10 percent residual strength is appropriate for
fine sands. Other soil types that are less prone to lique-
faction are expected to have higher residual strength val-
Lateral Pile Capacity. The subject oflateral pile
capacity requires an appreciation ofboth geotechnical
and structural engineering principles. The soil resistance
over the entire pile length generally far exceeds the
demand of the lateral load on the pile foundation. The
issue is not whether there is adequate soil resistance, but
whether the resistance can be mobilized before struc-
tural failure of the pile and before deflections that would
cause distress to the bridge become excessive. A rigorous
process to evaluate lateral pile capacity would involve a
soil-pile interaction analysis to determine the axial load,
bending moment, and shear load along the pile length,
which can then be checked against the structural capac-
ity of the pile. The results should also be checked to
determine if the deflection is excessive. Table RC4-4 pro-
vides Caltrans guidelines for lateral pile capacity and
corresponding stiffness that were used in retrofitting the
San Francisco double-deck viaduct. The guidelines were
based on pile load tests, as interpreted and summarized
by Caltrans. They are applicable for normal soil sites.
Allowable loads and displacements are based on
both pile and pile-cap interaction with surrounding soils
and reflect current Caltrans standard pile details.
The previously presented values on pile stiffness
and capacity represent typical rule-of-thumb values that
can be used for normal soil sites and typical pile footing
configurations. The effects of pile embedment at most
pile footings in the constructed condition would justify a
higher capacity and stiffness, which can be developed

114 BDS Recommendations, Section 4: Foundations ATC-32


TableRC4-4 Lateral stiffness and pile capacity

Ultimate able
Capacity Displ. Stiffness
Pile Type Soil Type (KlPile) (inch) (klin)

Steel Dense 100 3 35

Steel Loose 75 3 25
Steel Soft Cohe- 60 2 30
Concrete Dense 40 1 40
Concrete Loose 40 2 20
Concrete Soft 40 2 20

from a site-specific soil-pile interaction analysis. Figures

RC4-4 through RC4-12 can be used to facilitate the stiff-
ness comparison for pile embedments of 0, 5, and 10
feet. Pile stiffness and shear capacity for poor soil sites
(liquefiable and soft clay sites, especially if the pile top is
submerged or within five diameters to ground water)
should be determined from site-specific analysis.
In recent years, significant advancements have been
made in understanding the aspects oflateral stiffness
and capacity of pile footing foundations for highway
bridges. Some discussions on these aspects will be pro-
vided in ATC-32-1. Some of these aspects include:

• Interaction between the pile cap and the supporting


• The depth of ground cover above the pile founda-


• Ductility capacity of the pile member

Refinements in considering these aspects have

become more common in conventional design practice.
Designers now often attempt to incorporate the effects
of specific soil conditions, pile footing configuration,
and mechanical behavior of the specific pile types in
design. In conducting project-specific analyses, either to
provide for lateral capacities or to account for softer soil
characteristics at poor soil sites, proper accounting ofthe
structural and pile connection details is necessary.
Determining the structural capacity of Caltrans standard
concrete piles can follow the procedure described
in Calrans Memos to Designers 22-1 (July, 1989). Inter-

ATC-32 80S Recommendations, Section 4: Foundations 115


action diagrams for standard reinforced concrete piles

are available in Bridge Design Aid 16-9 (October, 1990). Pite Shaft Foundations C4.5.5.1 Pile Shaft Foundations

Pile extensions and column shafts, where piles are This type of bridge structure includes bridges in which
extended above ground to directly support the super- the bridge column is extended into the ground as a
structure without a pile cap, shall be detailed in accor- large-diameter drilled shaft (typically several feet in
dance with the requirements for ductile columns. diameter) and bridges in which conventional smaller-
diameter driven piles (say 16-inch piles) are extended
above ground to support the bridge deck. As discussed
in the specification, the overall dynamic response of this
type of bridge structure is very sensitive to the stiffness
characteristics of the pile foundations, especially rota-
tion of the pile.
Past earthquake performance indicates that whereas
large-diameter drilled shafts appear to have performed
adequately, pile-extension bridges (e.g., the Struve
Slough bridge, which collapsed during the Lorna Prieta
earthquake) appear to be more vulnerable to earthquake
damage. It appears that the reinforcing details for many
existing pile extensions (even the unsupported portion)
resemble those for piles. Therefore, they contain less
transverse reinforcement than typical reinforced con-
crete columns. This reduced transverse reinforcement
may have contributed to the relatively poor performance
of pile-extension bridges. However, it should be pointed
out that this design deficiency exists only in older Cal-
trans standards. Current practice and standards require
ductile detailing of the unsupported portion of the pile
extension. Therefore, pile-extension structures built
using the new seismic design criteria would probably
perform substantially better.
Proper modeling of pile-shaft foundations is
required to capture the overall response of pile-shaft
structures. Caltrans procedures for pile-shaft design as
outlined in Bridge Design Aids 12-30 through 12-49
(Caltrans, 1986) provide a good framework for the
design of drilled shafts. The following sections provide
some specific comments in relation to the design proce-

P-y Curves for Large Diameter Shafts

The conventional approach of soil-pile interaction anal-

ysis tends to underestimate the subgrade resistance for
large-diameter shafts, because most lateral soil support
(p-y) criteria are based on data from pile load tests using
24-inch-diameter piles. Evidence from testing oflarger
diameter shafts such as those on the Century Freeway
(FHWNCNSD-88) suggests significantly higher soil
resistance for these piles. The higher resistance has been
attributed to diameter effects (Pender, 1990 and Stevens
and Audibert, 1979). Lam and Martin (1986) attribute

116 BDS Recommendations, Section 4: Foundations ATC-32


the higher resistance oflarge-diameter shafts to the

mode ofsoil resistance associated with the shaft rotation,
which is not accounted for by the translational mode of
soil (p-y) resistance. Also, the difference in construction
methods (casting concrete in a drilled hole versus pile
driving) could alter the soil (p-y) behavior. Irrespective
of the reason for the higher resistance, it is evident that
the present p-y criteria underpredict the soil resistance
for large-diameter shafts. Modification ofthe commonly
used Reese and Matlock p-y criteria is recommended for
large-diameter drilled shafts.
In Reese's p-y criteria for sands, the input soil
parameters include the friction angle <1>, which controls
the ultimate resistance of the p-y curves and an initial
modulus of subgrade-reaction coefficient k Ob/in 3),
which controls the initial tangent stiffness of p-y curves.
To modify the p-y curves for the apparent diameter
effects, the k value can be increased in linear proportion
to the drilled shaft diameter in excess of two feet. For
example, for a medium-dense sand, a k value of 80 pci
would normally be used to develop p-y curves in Reese's
procedures. For a four foot diameter shaft, the appropri-
ate k value would be 160 pci The friction angle and the
ultimate soil resistance on the p-y curves should there-
fore remain unchanged.
Similar adjustments in Matlock's p-y criteria for
clays can be made to develop p-y curves for large-diame-
ter shafts. In Matlock's criteria for clay, the input soil
parameters include cohesion c ,which controls the ulti-
mate resistance of the p-y curves and a soil-strain value
e" defined as the strain amplitude that occurs at one-half
the maximum stress on laboratory undrained compres-
sion tests of undisturbed soil samples. A typical value of
e, is about 0.01. For a four foot diameter shaft, the
appropriate e, value would be 0.005. The cohesion, and
the ultimate soil resistance on the p-y curves should
therefore remain unchanged.
A similar adjustment procedure can be adopted
when using Figures RC4-4 through RC4-12 for large-
diameter drilled shafts. The subgrade stiffness parameter
f can be increased in proportion to pile-shaft diameters
for diameters larger than 24 inches.
The above adjustments are justified only for certain
situations such as large-diameter shafts supporting high-
way bridges where both the shear and the moment load
produce pile deflection in the same direction and the pile
is constructed by casting concrete in a drilled hole. For
other structures, such as offshore platforms or concrete
pile footings, the pile top at the mudline would experi-
ence a negative moment due to the constraint ofthe steel
frame. For such configurations, adjustments to the p-y
curves to account for the apparent diameter effects
would not be justified. The diameter effect is evident

ATC-32 80S Recommendations, Section 4: Foundations 117


from pile load tests, only because most pile load tests are
free-head load tests.

Equivalent Cantilever Length

The equivalent cantilever concept has been commonly

used to extend the point of fixity of the shaft to about
three to four diameters below grade to account for the
flexibility of the embedded shaft. The three to four
diameters value arose from linear beam-on-elastic-sub-
grade analyses that do not account for nonlinear
moment-curvature behavior of the shafts. Research data
from D.C. San Diego (Budek, 1994) indicate that the
plastic hinge would develop at shallower depths due to
such structural nonlinearity. In addition, current sub-
grade theories tend to underpredict the soil stiffness at
grade. Furthermore, concrete sidewalks or pavements
often exist around drilled shafts. The shear resistance
due to pavements has typically been neglected in analy-
sis. All these factors can lead to an overly large equiva-
lent cantilever length. The overestimate on the
equivalent cantilever length can lead to underprediction
of shear load corresponding to the flexural moment
capacity of the shaft. This is nonconservative and there
have been a number of incidents to indicate that the
presence of concrete sidewalks or floor slabs contributed
to column shear failures in past earthquakes (e.g., Olive
View Hospital in the 1971 San Fernando earthquake and
the Imperial County Services Building failure in the 1979
Imperial Valley earthquake). The equivalent column
length for shear load determination associated with plas-
tic moment load should be assumed to be no deeper
than two shaft diameters and should be assumed to be at
grade when concrete pavement is present, unless a gap in
the pavement is provided around the shaft.

Cracked Sectional Modulus

Recent test data further show that minor cracking occurs

at a relatively low nominal moment value on large-
diameter shafts and therefore cracked sectional proper-
ties should be used in dynamic response analyses of
structures supported by drilled shafts.

Minimum Pile Length

Current Caltrans practice involves pile length sensitivity

studies using nonlinear load-deformation analysis for
determining the critical pile length that ensures that
minimum stability ratio criteria are met. Whereas the
concept of ensuring some level of stability (safety mar-
gin) in the design is quite rational, the parameter called
the "stability ratio" has no physical meaning and can
lead to unreasonable designs in many cases. For exam-

118 BDS Recommendations, Section 4: Foundations ATC-32


pIe, for the same soil condition and the same lateral load
demand, the stability ratio concept requires a larger pile
length for a larger diameter pile, even though the larger
pile would have a higher lateral soil capacity and there-
fore a higher factor of safety in relation to lateral load
A more rational approach would be to ensure that
the pile length is adequate to provide stable load-deflec-
tion characteristics (e.g., meet an absolute deflection
limit or a deflection limit that is a proportion of pile
diameter) for an overload condition (e.g., a factored
load of two times the demand level). The approach
could further require that P-L1 effects be incorporated in
the pile solutions. Such an approach, which recognizes
the safety margin in relation to the loading condition,
are better than the stability-ratio concept, which overly
penalizes large-diameter shafts with respect to the pile
length requirement. This penalty is unreasonable and
often leads to complexities in construction (e.g., the tip
of the shaft needs to be extended below the ground water
table to meet the stability-ratio criteria). The stability-
ratio criteria are also unreasonable for many retrofit sit-
uations in which large-diameter drilled shafts are used at
abutments, primarily to provide additional lateral stiff-
ness, not to support the weight of the bridge structure. Pile Footing Foundations C4.5.5.2 Pile Footing Foundations

Pile footings shall be designed to resist shear and So far as resistance of pile footings is concerned, the
moment from Group VII loads. At normal soil sites, the overturning moment is resisted primarily by axial pile
ultimate lateral resistance of pile caps acting against soil capacity, whereas the lateral shear load is resisted prima-
may be included in the forces resisting the relative hori- rily by the lateral force capacity of piles. Since the axial
zontal movement of the foundation. At liquefiable and pile capacity is largely provided by soil resistance at
soft clay sites, the pile-cap resistance shall be neglected. depth and lateral capacity by soil resistance at very shal-
When yielding of piles will occur below the pile low depth, there is very little cross-coupling between the
footing, pile/footing connection details shall be suffi- moment and the lateral load capacities for pile footings.
cient to prevent pile pull-out, and transverse reinforce- Therefore, soil-pile interaction can be evaluated inde-
ment in concrete piles at the pile head shall be in pendently with respect to the two modes ofloading. The
accordance with Section following comments are provided regarding design
When reliable uplift pile capacity from skin-fric- aspects for overturning moment and lateral load at pile
tion is present, and when the pile/footing connection footings.
detail and structural capacity of the pile are adequate,
uplifting of a pile footing is acceptable, provided that Rotational Stiffness and Capacity
the magnitude of footing rotation will not result in
unacceptable performance. There is ample evidence to suggest that the rotational
stiffness of a pile footing has more significant influence
on the overall bridge response than does the lateral stiff-
ness. The rotational stiffness and moment capacity of
pile footings are largely related to the axial pile stiffness
and the ultimate compressive and uplift pile capacities.
The axial stiffness of a pile-soil system can be developed
by computer beam-column analyses (Lam and Martin,
1986) or by simplified graphical methods (Lam and
Martin, 1984). The ultimate and allowable compressive

ATC-32 BDS Recommendations, Section 4: Foundations 119


and uplift pile capacities should be documented in the

The bending stiffness at individual pile heads can
contribute rotational stiffness to the pile footing, in
addition to the axial pile stiffness. However, this compo-
nent of stiffness is usually small and unreliable because
oflocal concrete cracking at the pile top or pinned pile-
head connection detail. The bending stiffness at the pile
top should hence be neglected, or a pinned head condi-
tion be assumed when estimating the rotational stiffness
of pile footings.

Lateral Stiffness and Capacity

The lateral stiffness of a pile is very sensitive to the pile

head connection details as well as the depth of embed-
ment of the pile head. These factors should be properly
taken into account in evaluating the lateral stiffness of
the pile footing. The soil resistance at the pile cap can
contribute a significant lateral stiffness to the pile foot-
ings (Abcarius, 1991). At normal, stable soil sites, the
resistance of the pile cap can be included in calculations
of the lateral shear force resistance. However, at poor soil
sites (liquefiable and soft clay sites), the potential loss of
bearing capacity of the surficial soils could be a concern,
and the pile cap resistance should be ignored. However,
the designer should be aware that it is generally uneco-
nomical to allow the shear load to control the number of
required piles, considering that the pile is effective in
mobilizing soil resistance at only about the upper five
pile diameters. Other design strategies can be used to
resist the shear load, including:

• Use of thicker or larger footings and including the

pile cap resistance at normal, stable soil sites. The
procedure discussed for abutment backwall stiffness
and capacity can also be used for pile caps

• Use of deeper pile footing embedment, which

would increase the resistance of both the pile and
pile cap

• Modification of the pile top connection detail to

achieve a greater degree of pile head fixity (e.g.,
embedding the pile top deeper into the pile cap)

• Strengthening the structural capacity of the pile at

approximately the upper ten pile diameters

• Use of more ductile pile types that can develop soil

resistance to a higher amplitude of pile deflection

• Soil improvement at shallow depths around the pile

footing and pile head at poor soil sites

120 BDS Recommendations, Section 4: Foundations ATC-32


The source of the lateral capacity (in terms of pile

versus pile cap) and the magnitude of deflection to
develop the capacity should be documented in the geo-
technical report.

Structural Types Consideration

The general philosophy in foundation design is to

include a load fuse located in the bridge column. There
is significant merit in this concept due to the many
uncertainties in the magnitude of earthquake load and
the desire to restrict damage to those components above
ground that can be easily inspected and repaired. For
column footings, the load can be effectively limited by
the design of plastic hinging at the column, in which
case the chance of overload in the foundation would be
low. However, the flexural strength of pier walls in the
strong bending direction of the wall would be very high
and the plastic hinge could be forced to form in the pile
head-pile cap connection. A stronger connection detail
and a more ductile pile would enhance the chance of
improved performance of the overall system.

Pile Foundations at Poor Soil Sites

Stronger connection details and more ductile pile types

should be used at poor soil sites (i.e., liquefiable sandy,
soft clay, and landslide-prone sites). At these sites, large
freefield ground displacements would be likely for the
large safety-level earthquake design condition, in addi-
tion to the inertial load of the superstructure. The load-
ing condition associated with ground displacement is
difficult to analyze and design for, due to uncertainty in
the magnitude and mode of ground displacement as well
as the soil property itself. However, the use of stronger
connection details and more ductile pile types will allow
the foundation system to accommodate ground dis-
placement and improve the chance of survivability.

Batter Piles

There has been controversy regarding the use of batter

piles, based on their performance in past earthquakes.
Before deciding to use batter piles, the benefits of these
piles should be weighed against the additional complex-
ity in design, difficulty in construction, and potential
reduced performance. The major benefit of using bat-
tered piles relates to mobilizing larger axial stiffness to
increase the lateral stiffness of the pile group. However,
construction practice generally limits the batter angle.
Although the axial stiffness is relatively large as com-
pared to the lateral pile stiffness, it is finite, and therefore
a realistic assessment of the potential increase in lateral
stiffness by pile battering must be evaluated by a load-

ATC-32 BDS Recommendations, Section 4: Foundations 121


deformation analysis to account for the pile configura-

tion in conjunction with realistic axial and lateral pile
stiffness characteristics. Experience indicates that in
many cases, the benefit of the increase in the lateral stiff-
ness by pile battering is relatively minor, especially at
poor soil sites, which usually require the use oflong fric-
tion piles.
In the past, designers often implicitly assumed that
the axial stiffness of batter piles was infinite relative to
the lateral stiffness, which directly led to the assumption
that all the lateral force in a batter pile group will be
resisted by the axial pile force and that therefore, a batter
pile need not be designed for bending moment. Such an
assumption is probably the cause of poor performance
of batter piles in past earthquakes (e.g., at Port of Oak-
land during the 1989 Loma Prieta earthquake). All the
damaged batter piles had poor connection details, as
well as inadequate transverse steel for resisting the flex-
ural moment at the pile head. As discussed above, the
assumption of an infinite axial stiffness is invalid. Expe-
rience in load-deformation analyses of batter pile groups
(especially in the post-elastic range) indicates that batter
piles experience a large bending moment that, in most
cases, is of about the same magnitude as that experi-
enced by a corresponding vertical pile group. Therefore,
batter piles must be detailed for moment and shear load.
Soil-pile interaction must also be considered in the
design of batter pile groups. Most conventional design
analyses only address inertial loading from the super-
structure and not the loading arising from lateral
ground displacement (e.g., lateral spread ofliquefied soil
or lateral embankment movement). For such condi-
tions, the stiffer batter pile groups attract very large
forces and do not perform well compared to the more
compliant vertical pile groups. Batter pile groups should
be avoided at poor soil sites (liquefiable and soft soil sites
or sites that are known to be unstable), unless detailed
analyses are conducted to address all the above issues by
personnel experienced in soil-pile interaction analysis.

Group Effects

For typical pile footings (i.e., fewer than 20 piles at three

diameter center-to-center spacing), group effects can be
ignored in considering the rotational response of pile
groups, because the response of individual piles within
the group are not in phase. At a given time, some of the
piles are compressed whereas others are uplifted from
earthquake-induced moments. There could be some
group effects in regard to lateral loading due to pile
shadowing effects. However, for a typical footing, group
effects are relatively small (about 20 percent). Most
available pile-group test data are from monotonic load-
ing tests or from simplified analyses that tend to over-

122 BDS Recommendations, Section 4: Foundations ATC-32


dramatize group effects. Bogard and Matlock (1983)

offered a practical method to account for group effects
based on limited cyclic pile-group test data. These
authors indicate that group effects would be reduced
under cyclic loading conditions. This is particularly valid
for softer soils, where cyclic loading tends to remold a
zone immediately around the pile, with the weakened
soil becoming less effective in transferring induced
stresses to the neighboring piles. In general, group effects
would be more significant for stiffer soil and for static
loading conditions. However, such situations are gener-
ally less critical design concerns. In view of the overall
uncertainties, it is recommended that group effects be
neglected for earthquake loading at three-diameter cen-
ter-to-center spacing or higher.
Group effects can become very important for large
pile groups, such as those at major bridges crossing
water (e.g., the San Francisco-Oakland Bay Bridge).
Such structures can have several hundred piles in a
group. The configuration changes drastically from long
slender piles to a reinforced soil-mass system in which
the overall configuration of the pile group rather than
individual piles can become the governing mechanism.
For cases in which group effects can be important, they
shall be properly accounted for in the analysis. Such sys-
tems warrant finite-element analysis to account for
potential group effects.

4.5.6 Spread Footing Foundations C4.5.6 Spread Footing Foundations

Spread footings shall be designed to resist shear and The traditional procedure to determine the size of
moment from Group VII loads. The seismic design spread footings is based on the use of s~rvice-Ievelloads
requirements for spread footing foundations are essen- along with allowable bearing pressures for specific soil or
tially similar to the requirements for pile footings with rock types. The allowable bearing pressure is most often
respect to stiffness modeling and the guidelines for tol- based on presumptive values specified in design codes
erable foundation displacements. Additional require- rather than on fundamental soil mechanics bearing
ments for the design of spread footings are as follows: capacity theories and soil strength parameters. After ini-
tially sizing the footing, current Caltrans design practice
(a) Spread footing foundations shall be designed for requires a check of the footing for the seismic (Group
proper performance under earthquake group loads VII) loading. This involves conducting a statically deter-
(Group VII Load). The design capacity shall reflect minant analysis (using the combination of axial load and
the capacity of the foundation soil, the structural moment associated with the Group VII loads) of a rigid
capacity of the footing, and the connection details footing model to determine the maximum soil pressure
between the column and the footing. Ultimate at the edge of the footing and the proportion of the foot-
bearing capacity may be used for seismic design. ing uplifted from the soil surface. The maximum soil
pressure demand is then compared against a maximum
(b) The effect of overturning moment (eccentricity ultimate bearing pressure recommended by the geotech-
loading) and lateral loading (inclined loading) on nical engineer. A factor of safety of three is commonly
bearing capacity shall be considered in the seismic used in relating the allowable bearing pressure to the
design of spread footings (See Commentary 4.5.6). ultimate bearing pressure. As discussed above, the allow-
able bearing pressure is generally based on rather con-
(c) Total foundation settlement and differential settle- servative presumptive values in design codes, rather than
ments between adjacent bents shall not result in more basic bearing capacity theories.The geotechnical

ATC-32 BDS Recommendations, Section 4: Foundations 123


unacceptable performance of the bridge (See Com- report and the design plan for the safety-evaluation
mentary 4.5.1). earthquake shall document both the allowable capacities
for service-level loads and the ultimate capacities for
(d) The area of uplift of spread footings due to over- both compressive and uplift loading conditions for
turning moment for single-column bents shall not earthquake loads.
exceed 0.25 of the width of the footing. The eccen- Although there is little evidence to suggest that the
tricity shall not exceed 0.33 of the width of the foot- above practice has led to poor performance of spread
ing for multiple-column bents. footings in past earthquakes, there are some develop-
ments in the AASHTO code that could lead to adopting
more basic bearing capacity theories in the design of
future bridges. In the interest of a simple design proce-
dure, the service-load design method relies heavily on
the presumptive allowable bearing pressure rather than
on the more basic bearing capacity theories that are the
cornerstones of soil mechanics theories. The effects of
the lateral shear load are ignored in the traditional
design procedure. Unlike pile foundations for which the
moment-plus-dead load can be uncoupled from the
shear load in soil capacity determinations, dead weight,
moment, and shear on a spread footing will be resisted
by the same soils atshallow depths. Therefore, it is
invalid to assume that the soil capacities are independent
of the various modes of loading.
Theoretically, the dead load, moment, and lateral
shear need to be simultaneously considered in bearing
capacity evaluations for spread footings. The classic
bearing capacity theory can be used to determine the
ultimate bearing capacity for such a combined loading
condition, with the moment and shear load on the foot-
ing represented by a statically equivalent eccentric and
inclined load. There are relative merits in both
approaches. The traditional allowable pressure approach
is more simple and practical, whereas the classical bear-
ing capacity theory is more rigorous and represents the
future trend for spread footing design, as evident in the
recently adopted load-factor design procedure (NCHRP
Report 343).
Unlike a pile footing, where the uplifted pile would
provide a restoring force on the footing, the portion of
the uplifted spread footing area would lead to significant
geometric nonlinearity and can alter drastically the rota-
tional stiffness of the footing. Therefore, there is an
added incentive to provide an additional conservatism
in limiting the uplift area in the design of spread foot-
It is recommended that the traditional service-load
design approach be used for service-load design to deter-
mine the footing size initially. However, in the course of
checking the design for Group VII loads, the classic bear-
ing capacity theory should be used for checking the ade-
quacy of the footing, with due consideration of the
combined effects of dead, moment, and lateral loads. A
factor of safety of 1.0 is adequate for this analysis.
Detailed procedures are included in NCHRP Report 343.

124 BDS Recommendations, Section 4: Foundations ATC-32


4.5.7 Retaining Structures (4.5.7 Retaining Structures

Lateral and rotational movement of earth retaining There are two basic tasks in designing retaining struc-
structures during an earthquake shall be limited to val- tures:
ues that satisfy the requirements of the performance
criteria described in Section 3.21.2. In addition, retain- 1. Determining the overall size and configuration.
ing walls shall be structurally capable of withstanding This is generally controlled by the overall stability
the static and dynamic earth pressures generated dur- (referred as external stability) of the retaining sys-
ing the design earthquake. tem.
Type-selection studies for earth retaining systems
shall consider the historic performance of these systems 2. Structural design of the retaining system to with-
in seismically active regions and at poor soil sites. stand a given earth pressure on the retaining wall.
Design earth pressures (static and dynamic) shall
consider the effect of restrained movement on the earth Overall Stability
retaining system.
Review of past practice suggests that most conventional
retaining structures are designed for service level (non-
earthquake) loads in conjunction with an adopted factor
of safety. In the overall stability evaluation, the factor of
safety generally varies from 1.5 to 2. In general, service-
level loads are based on static, active earth-pressure con-
ditions (e.g., the lateral earth-pressure coefficient of 0.3
in Caltrans practice for retaining walls).
Dynamic, active earth-pressure requirements have
been introduced into the AASHTO Bridge Design Speci-
fications in the context of requirements for estimating
potential movement of the retaining wall in seismic
design. However, many designers consider the require-
ment too complex. In manycases, the designer assumes
that the inherent reserve in the static design (static factor
of safety) would be adequate to limit the displacement to
an acceptable level for earthquake loading. In past earth-
quakes, free-standing retaining walls (not associated
with other structures) appear to have performed well,
even though most retaining walls have been designed
only for the relatively low static, active earth-pressure
Although there are some case histories of earth-
quake damage to certain type of retaining walls (e.g.,
crib walls) in past earthquakes, the good performance of
retaining walls probably indicates that when allowed to
yield, excessive soil pressure on the wall is relieved, and a
small amount of movement is of little consequence.
Most of the case histories of retaining wall failure
(mostly unrelated to earthquakes) appear to be associ-
ated with clay soils, either as retained fill or as founda-
tion soils.
From past performance histories, it is suggested that
typical retaining walls (typically less than 30 feet high
and not associated with adjacent structures) should be
designed using static earth-pressure theories without
considering earthquake loads. However, it is suggested
that the adopted factor of safety should be 1.5 for sandy
soils (both backfill and foundation soils are cohesion-

ATC-32 8DS Recommendations, Section 4: Foundations 125


less), whereas the factor of safety be increased to 2 for

clay soils (clay used either as backfill or found in founda-
tion soils).
Additional materials have been included in
ATC-32-1 (ATC, 1996), which provides some informa-
tion on active and passive earth-pressure theories,
including dynamic, active pressure theories. Design
charts have also been developed to facilitate the determi-
nation ofearth-pressure coefficients. Furthermore, some
design charts have been provided for permanent dis-
placement solutions of the Newmark sliding-block
model to simplify the process of applying dynamic,
active earth-pressure theories and permanent displace-
ment evaluation procedures for earthquake loads.

Structural Design

It should be recognized that the static, active earth-pres-

sure condition represents the minimum earth pressure
that would be exerted on the wall, and that the wall must
be allowed to move to relieve excessive earth pressure for
the low-pressure assumption to be valid. There are many
scenarios in which the earth pressure would exceed the
static, active earth-pressure coefficient. Whereas the
static, active earth pressure may be adequate for evaluat-
ing the overall stability of the retaining wall, it may not
provide an adequate margin of safety for structural
design. Various factors, including unexpected restraint
ofthe wall from nearby constructed facilities and uneven
earth-pressure distribution could result in localized
higher soil pressure on the wall that should be accounted
for in structural design. Although there are load factors
and material strength-reduction factors inherent in
structural design codes to provide for the needed margin
of safety for unusual conditions (ie., wall height higher
than 30 feet or adjacent to other structures), it is recom-
mended that the structural design of the retaining wall
be based on other more refined approaches or a more
conservative basis.
One alternative is to design for a dynamic active
earth pressure condition in conjunction with using the
appropriate horizontal acceleration coefficient (say 0.5 of
the peak ground acceleration coefficient) as suggested in
the 1992 AASHTO Bridge Design Specifications (Section
6 in Division I-A for abutment design). Many designers
consider the dynamic, active earth-pressure theory
overly complex and problematic. In such a case, a more
simple approach would be to design the retaining wall to
a higher static earth-pressure coefficient in the service-
load design scenario; say to the at-rest earth-pressure
condition (coefficient of about 0.5 in conjunction with
the appropriate factor of safety).

126 BDS Recommendations, Section 4: Foundations ATC-32


Poor Soil Sites

Because the availability of conventional nonproprietary

and proprietary retaining wall designs has grown expo-
nentially in recent years, many alternatives are now
available to the designer. Therefore, selection of the
appropriate type of retaining wall has become a major
design consideration. At poor soil sites, in view of the
significantly higher potential for ground displacements
(i.e., total and differential horizontal displacement and
vertical settlement), use of a retaining wall that can tol-
erate ground displacements (e.g., MSE walls) shall be
considered. Such retaining walls also have good inherent
earthquake performance.

ATC-32 BDS Recommendations, Section 4: Foundations 127

Section 8
Reinforced Concrete



8.1.1 General

The specifications of this section are intended for

design of reinforced (non-prestressed) concrete bridge
members and structures. Bridge members designed as
prestressed concrete shall conform to Section 9.

8.1.2 Notation

The following notation is in addition to that shown in

the current BDS.

A bI = area of longitudinal reinforcing bar being

spliced (Article
A. = effective shear area of columns or beams
A hb = area of hoop or spiral bar
Ajv = area of reinforcing required to provide tie force
Ts (Article C8.34.4.2)

1. This section on reinforced concrete design includes new methods

for calculating flexural and shear strength, anchorage and splice
lengths, transverse confining reinforcement and joint reinforcement.
Although these methods may be an improvement over current prac-
tice, many of them are based on an interpretation of recent research
that has not been subjected to the traditional peer review required by
the American Concrete Institute (ACI) prior to code adoption. The
PEP did not feel it had the depth and/or breadth of expertise to ade-
quately fill this role of peer review. For this reason, some subcontractor
recommendations that would have tended to make designs less conser-
vative have not been included, and there is a concern that designs may
be overly conservative in some areas. The PEP recommends that Cal-
trans conduct additional trial applications of these recommendations.
The PEP also recommends that the new requirements be subjected to
the traditional peer review process for reinforced concrete design pro-
visions. These recommendations also apply to some of the alternative
methods such as the shear capacity model described in the accompa-
nying resource document, ATC-32-1, even though these methods have
not been adopted into the recommendations.

ATC-32 BDS Recommendations, Section 8: Reinforced Concrete 129


A st = total area of column reinforcement anchored in

the joint
bje = effective joint width
D = diameter of circular columns
D' = diameter between centerlines ofperipheral hoop
or spiral
d bb = effective diameter of bundled bars as defined in
dbl = nominal diameter of column longitudinal
reinforcement being anchored or spliced
(Article 8.33)
Eds = "double modulus" of steel, defined in C18.2.2.2
Ei = initial modulus of elasticity of longitudinal
Esu = effective secant modulus of elasticity of
longitudinal reinforcement measured from
current stress Isu to ultimate stress Iu
Et = initial modulus of elasticity of transverse
f'cc = compression strength of confined concrete
f'ce = expected concrete compression strength
f'co = maximum feasible concrete compression
Ih = average normal stress in the horizontal direction
within moment resisting connection
Ie' = equivalent uniform confinement stress as
defined in Equation 8-24
Is = maximum tensile stress in column longitudinal
reinforcement (Article
Isb = axial stress in reinforcing bar at inelastic
Iu = ultimate stress in reinforcing steel
Iv = average normal stress in the vertical direction
within moment resisting connection
ire = expected yield stress of column longitudinal
irh = specified yield stress of transverse reinforcement
iro = maximum feasible yield stress of column
longitudinal reinforcement
Iyv = yield stress of vertical stirrup or tie (Article
H = length of pile shaft!column from ground surface
to point of zero moment above ground
h b = cap beam or footing depth
h c = lateral column dimension (as defined in
hs = superstructure depth
i! ac = length of column bar embedment into cap beam
or footing (Articles &
i! b = length used for flexural bond requirements
(Article 8.33.3)
i! c = length of column between point of maximum
moment and point of zero moment

130 BOS Recommendations, Section 8: Reinforced Concrete ATC-32


i 0 = length of the plastic end region requiring special

i p = plastic hinge length
is = splice length
nb = number of column longitudinal reinforcing bars
distributed around section that are subject to
inelastic buckling if cover concrete spalls
Pc = principal compression stress within moment
resisting connections (Article
Pt = principal tensile stress within moment resisting
connections (Article
Tb = horizontal tie force required for joint shear force
Tc = Tension force in column reinforcement to be
transferred to joint
Ts = vertical tie force required for joint shear force
vhv = average shear stress in the hv plane within
moment resisting connection (Article
Ccu = required compression strain for confined
Cs = reinforcing steel strain
<1>0 = overstrength ratio (Mp/Md)
a p = required plastic hinge rotation
PI = column longitudinal reinforcement ratio
Pv = minimum volumetric ratio ofvertical stirrups in
footing (Article
J.lw = section curvature ductility capacity
'If = curvature
'lfp = plastic curvature
'Ifu = ultimate curvature
'lfy = yield curvature

Articles 8.2 through 8.15 not modified.



8.16.1 Strength Requirements Required Strength C8.16.1.1 Required Strength

Except for Group VII loads, the required strength is the It is emphasized that all forces acting on a structure must
strength necessary to resist the factored loads and forces be in equilibrium at all times. This also applies to the
applied to the structure in the combinations stipulated determination of appropriate forces for design under
in Article 3.22. Group VII loads. Thus, in the preliminary design of duc-
For Group VII loads, the required strength of plas- tile columns, where required strength is calculated by
tic hinges is the strength necessary to resist the factored reducing the results from a dynamic elastic analysis by a
loads and forces applied to the structure in the combi- force-reduction factor Z, this factor Z must initially be

ATC-32 8DS Recommendations, Section 8: Reinforced Concrete 131


nations stipulated in Article 3.22. Additionally, the applied to both seismic moments and seismic axial force.
required strength of members outside the plastic hinges The final design will be based on axial forces in equilib-
is the strength necessary to equilibrate the forces associ- rium with gravity loads and the nominal flexural
ated with development of maximum plastic moment in strength of the plastic hinges.
potential plastic hinges, in accordance with Article 'When determining response under overstrength conditions, where the plastic regions develop plastic
All sections of structures and structural members moment capacity (see Article, the column axial
shall have design strengths at least equal to the required force resulting from seismic response will need to be
strength. increased by the overstrength factor, and all elements of
the structure that are to be protected against inelastic
action must be designed for required strength. The
strength is calculated by combining actions due to grav-
ity loads and the column overstrength forces, now con-
sidered as actions applied at plastic hinge locations. In
this analysis, gravity loads do not induce additional
moments at the plastic hinges, which can thus be con-
sidered as perfect hinges for the gravity load analysis. Design Strength C8.16.1.2 Design Strength The design strength provided by a member The coefficient l/J provides for the possibility that where
or cross section in terms of load, moment, shear, or strength is the prime concern, imperfections in the
stress shall be the nominal strength calculated in accor- equations for nominal strength or small adverse varia-
dance with the requirements and assumptions of the tions in material strength, workmanship, and dimen-
strength design method, multiplied by a strength- sions, while individually within acceptable tolerances
reduction factor l/J. and limits of good practice, may combine to result in
understrength. The strength-reduction factors l/J shall be as A strength-reduction factor l/J = 1.0 is applied to the
follows: flexural design of ductile columns for Group VII loads.
This is because the actual flexural strength is expected to
(a) Flexure, or axial tension and flexure be developed in the design earthquake. Design using a
(except for Group VII column design) l/J = 0.90 strength-reduction factor l/J < 1.0 is not specified because
flexural understrength only marginally increases the
(b) Shear l/J=0.85 ductility demand on the column. On the other hand, use
of a flexural strength-reduction factor will result in a
(c) Axial compression and flexure proportionate increase [i.e., l/l/J] in the required
(except Group VII columns): strength of all capacity-protected actions and members,