11 Engineering/Architecture e, practical coverage of ation, analysis, and ind code requirements of solation systems Design of 1 the concept of reducing seismic demand rather shan inereasing tne ake resistance capacity of structures, seismic isolation is @ surprising approach to earthquake protection. However. proper application oF this >gy within complex seismic design code requiraments is both compli= id difficult. f Seismic Isolated Structures provides complete, up-to-date covelage of isolation, complete with a systematic development of concepts in the- practical application supplemented by numerical examples, This -ROM set helps design professionals navigate and understand the ideas redures involved in the analysis, design, and development of specific ’r seismic isolated structures. It also provides a framework for satisfying wuirements while retaining the favorable'cost-effective and damage aspects of this new technology. An’ indispensable resource for 1g and aspiring engineers and architects; Design of Seismic Isolated 2s includes: ion system components ete coverage of code provisions for seismiciisolation nical characteristics and modeling of isolators 1g and stability of elastomeric isolators les of seismic isolation designs ications for the design, manufacture, and testing of isolation devices he CD-ROM ompanying CD-ROM contains ISOSEL isolation software; SAP2000 i program (student version); and selected suites of earthquake groUnd NAEIM, PRD, SE, is Director of Researeh and Development at Jchn All ssociates, Inc. in Los Angeles, California. JAMES Mz KELLY, Phi, ill Tin the Graduate School ofthe University of ealfartia Febery : Practice =47 4-348 iley & Sons, tne. win nal/rede Ovision [otter wie as | | | | | | Farzad Naeim vee OWE A 2s rt ~ » M. KellS25 CTE DESIGN OF SEISMIC ISOLATED STRUCTURES From Theory to Practice Farzad Naeim, Ph.D., S.E. Director of Research and Development. John A. Martin and Associates, Inc. wutvenszrarca Taenrca Los Angeles, California James M. Kelly, Ph.D. TT Professor in the Graduate School BIBLIOTECA CENTRALA University of California at Berkeley Berkeley, California JOHN WILEY & SONS, INC. New York / Chichester / Weinheim / Brisbane / Singapore / Toronto‘This book is printed on acid-free paper: @ Copyright © 1999 by John Wiley & Sons, Ine. All rights reserve, Published simultaneously in Canada [No par of this publication may be reproduced, stored ina retrieval system of transmited in any form or by any means, cleewonic, mechanical, photocopying, recording, scanning oF ‘otherwise, except as permitted under Sections 107 of 108 of the 1976 United States Copyright ‘Act without either the prior written permission of the Publisher, or authoriation through payment of the appropriate per-copy fee to the Copyright Clearance Center, 222 Rosewood Drive, Danvers, MA 01923, (978)750-6400, fax (978)750-4744. Requests to the Publisher for permission should be addressed to the Permissions Department, John Wiley & Soas, Inc. {603 Third Avenue, New York, NY 10158-0012, (212)850-6011, fax (212)8S0-6008, E-Mail PERMREQ @ WILEY.COM. ‘This publication is designed to provide accurate and authoritative information in regard to the subject matter covered. Ic is sold with the understanding thatthe publisher is not engaged in rencering professional services. If professional advice or other expert assistance fs required, the services of a competent professional person should be sought. Library of Congress Cataloging-n-Publication Data "acim, Farad, Design of seismic isolated structures + from theory to practice / Farzad Nocim, James M. Kel Bem, Includes bibliographical references and index, ISBN 0.471-14921-7 (alk. paper) 1. Earthquake resistant design. I. Kelly, James M, Il Tie TAGSEA4N34_ 1999 624.1°762—de21 se-sises Prined in the United States of America 1098765 To the memory of precious lives lost to earthquakes worldwide| | CONTENTS PREFACE CHAPTER 1. DEVELOPMENT OF SEISMIC ISOLATION Al 12 13 14 1s 16 WORLDWIDE Introduction / 1 Base Isolation in the United States / 6 Base Isolation in Japan / 18 Base Isolation in Europe / 20 Base Isolation in New Zealand / 21 State of Isolation Technology Today / 23 CHAPTER 2, THEORETICAL BASIS OF SEISMIC ISOLATION 24 22 23 Linear Theory / 25 Extension of Theory to Buildings / 31 2.2.1 M-Degree-of-Freedom Equations of Motion / 31 2.2.2 Modal Analysis of M-DOF System / 33 Analysis of Coupled Dynamic Equations / 36 CHAPTER 3. ISOLATION SYSTEM COMPONENTS 3a 32 Introduction / 47 Elastomeric-Based Systems / 47 xl 28 a73.2.1 Low-Damping Natural and Synthetc Rubber Bearings / 48 3.2.2 Lead-Plug Bearings / 49 3.2.3. High-Damping Natural Rubber Sysems (HDNR) / 50 3.3. Isolation Systems Based on Sliding / 52 33.1 Blectricité-de-France System / 55 33.2 EERC Combined System / $5 333 The TASS System / 56 3.3.4 Resilient Tviction Base Isolation Systems / 56 33.5 Friction Pendulum System / 57 3.4 Spring-Type Systems / $8 3.5. Sleeved-Pile Isolation System / 58 3.6 Rocking Systems / 61 CHAPTER 4. CODE PROVISIONS FOR SEISMIC ISOLATION 63 4.1 Inoduetion / 63 42 Seismic Hazard Level / 64 43° Design Methods / 65 44 Static Analysis / 65 44.1 Seismic Zone Factor Z / 67 44.2 Site Soil Profile Type / 68 44.3 Seismic Source Types: A, B, and C / 68 444 —Near-Source Factors: Na and Ny / 68 4s 46 a7 44.5 MCE Response Coefficient Myy / 70 44.6 Spectral Seismic Coefficients: Cyp, Cv and Cap, Cam / 70 44.7 Damping Coefficients: Bp and By / 71 4.4.8 Effective System Vibration Periods: T'p and Ty / 73 449 Total Design Displacements: Dro and Dry / 74 4410 Design Forces / 75 44,11 Vertical Distribution of Force / 76 4.4.12 Drift Limits / 77 Dynamic Analysis / 78 4.5.1 Time History Analysis / 78 4.5.2 Scaling / 78 Other Requirements for Nonstructural Components / 80 Peer Review / 80 48 49 410 411 Design and Testing Requirements for Isola‘ors / 81 OSHPD-96 Requirements / 82 ‘Other Considerations / 84 Step-by-Step Procedure for UBC-97 Compliant Design / 85 4.11.1 Preliminary Design Steps / 85 4.11.2 Final Design Steps / 87 4.11.3 Design Example / 89 CHAPTER 5. MECHANICAL CHARACTERISTICS AND sa 52 33 3a 35 56 MODELING OF ISOLATORS 93 Introduction / 93 Mechanical Characteristics of Blastomeric Bearings / 93 Mechanical Characteristics of Lead-Plug Bearings / 100 Mechanical Characteristics of Friction Pendulum System / 101 Modeling of Isolation Bearings by Bilinear Modeling / 104 Implications of Bilinear Modeling / 107 5.6.1 Energy Dissipation in High-Damping Natural Rubber Bearings / 111 5.6.2 Adjustments to the Model to Account for High-Strain Stiffening / 115 5.6.3 Comparisons with Experimental Data / 117 CHAPTER 6. BUCKLING AND STABILITY OF ELASTOMERIC ISOLATORS 121 61 Introduction / 121 62 Stability under Large Lateral Displacement / 126 63 Rollout Stability / 133 CHAPTER 7. DESIGN EARTHQUAKE GROUND MOTIONS 137 7.1 Introduction / 137 7.2 Characteristics of Earthquake Ground Motions / 139 7.3 From Response Spectra to Design Spectra / 143 7.4 Earthquake Energy Content and Energy Spectra / 149 7.8 Various Ground Motion Predictive Formulations / 150 75.1 Source Characterizations / 150 7.5.2 Attenuation Relations / 155 7.6 Deterministic and Probabilistic Approaches / 158 76.1 Deterministic Seismic Hazard Evaluation / 1581 78 ‘CONTENTS 7.62 Probabilistic Seismic Hazard Evaluation / 159 7.6.3 Limitations of Deterministic and Probabilistic Seismic Hazard Analysis / 166 7.64 Seismic Hazard Maps / 170 Code Interpretations of Design Ground Motions / 172 Application of Earthquake Time Histories / 174 7.8.1 UBC-94 Provisions / 174 7.8.2 UBC-97 Provisions / 175 OSHPD-91 Provisions / 176 OSHPD-96 Provisions / 176 ‘Time-Domain Scaling of Time Histories / 176 Frequency-Domain Scaling of Time Histories / 180 Analytic Dilemma of Designing by Time History Analysis / 183 CHAPTER 8. DESIGN EXAMPLES 185 a1 82 Design Example for a HighDamping Rubber Bearing / 185 8.L1 Bearing Stiffnesses / 186 First Estimate of Design Displacement Dp / 187 ‘Actual Bearing Stiffness / 188 Composite Stiffness / 188 Composite Damping / 188 Allowance for Torsion / 189 Elastic Base Shear from Code / 190 Bearing Detail / 191 Buckling Loads: Safety Factor / 193 Calculation of MCE Displacement Dyy / 195 Rollout Displacement / 196 Design Example for a Lead-Plug Bearing / 196 ‘CHAPTER 9. COMPUTER APPLICATIONS 203 94 92 Introduction / 203 Preliminary Selection of Design Loads and Isolator Sizes / 203 9.2.1 Overview / 203 9.22 Selection of Design Loads by Linear Regression and Least-Squares Analysis Techniques / 204 9.2.3 The ISOSEL Solution Strategy / 206 9.2.4 Application of the ISOSEL Computer Program / 207 9.3 Computer Programs for Analysis of Seismically Isclated Structures / 216 93.1 N-PAD / 217 3D-BASIS / 217 ETABS / 218 SAP-2000 Nonlinear / 218 General Nonlinear Three-Dimensional Analysis Programs / 220 9.4 Nonlinear Dynamic Time History Analysis / 220 ‘CHAPTER 10. SPECIFICATIONS FOR DESIGN, MANUFACTURING, AND TESTING OF ISOLATION DEVICES 10.1 Introduction / 233 10.2 Sample Specifications / 234 APPENDIX. COMPANION SOFTWARE AND EARTHQUAKE DATA FILES. REFERENCES INDEX 233 255 273PREFACE Seismic isolation is a simple structural design approach to mitigate or reduce earthquake damage potential. Seismic isolated structures are currently difficult to analyze, design, and implement, however, due to complex code requirements. This book serves as a guide to help the reader navigate and understand the con- cepts and procedures involved in analysis, design, and development of specifi- cations for seismic isolated structures. The book is intended as a reference for practicing engineers and architects as well as a text for graduate-level courses on seismic isolation. It provides a complete and up-to-date coverage of the subject and numerical examples and systematic development of the concepts in theory and practical application. The book is partially developed from lecture notes generated over the period of two decades of teaching graduate classes at the University of Cali- fornia at Berkeley. The material is complemented by practical insights obtained from many years of involvement in analysis, design, and review of major seis mic isolated projects in the United States. In response to the damage generated by recent earthquakes in densely pop- ulated areas, seismic design codes for the design of buildings, bridges, and industrial facilities changed with the intention of leading to better seismic per- formance. This process has been repeated in response to all recent damaging. earthquakes, such as the 1989 Loma Prieta, 1994 Northridge, and 1995 Kobe events. In addition, damaging earthquakes in Mexico, Australia, Armenia, Iran, and India have all caused changes in the seismic design codes and procedures. Consequently, all structural engineers designing earthquake-resistant structures have had to contend with increasingly complex code requirements. Seismic isolation is an approach to carthquake-resistant design that is basedxiv PREFACE con the concept of reducing the seismic demand rather than increasing the earth- ‘quake resistance capacity of the structure. Proper application of this technology leads to better performing structures that will remain essentially elastic during large earthquakes. Surprisingly, the basics of this approach are rather simple. Unfortunately, however, the seismic design code provisions for design of seismic isolated structures are complicated and difficult for the engineer to apply. This textbook is intended to clarify for the practicing engineer, as well as engineering student, the basic concepts of seismic isolation and the funda- ‘mental principles of seismic isolation design. It provides a framework for the design professional to satisfy the code requirements while retaining the favor- able, costffective, damage control aspects of this new technology. It is the authors’ contention that, once the intimidating barriers of code application are bypassed, the use of seismic isolation will become a routine procedure for many types of structures. ‘The theoretical underpinning of seismic isolation has been firmly established and the technology has been verified by extensive experimental work over the past quarter century. The relevant material is covered in many technical reports, archived publications, and journal articles. But no general collection of this research in a form accessible to the practicing engineer has been available until this time. This textbook will bring much of this information to the engineering profession and will compliment it with knowledge gained by the authors from years of experience as designers, consultants, and reviewers of mary seismic isolated projects. ‘The authors would like to express their gratitude to the following individu- als and institutions: Ashraf Habibullah and Computers and Structures, Inc., for ‘kindly permitting the educational version of the SAP2000 computer program to ‘de used as a companion to this book; Syed Hasanain and Iqbal Suharwardy for their valuable assistance in this regard; Claire Johnson for her expert typing of the manuscript; Roy Lobo for coding the optimization algorithm for the com- panion ISOSEL computer program; lan Aiken and Fredrick Tajirian of Seismic Isolation Engineering, Inc., for providing the sample specification included in Chapter 10 of the book; and last but not least, Mark Day, Andrew Besirof, Evita Oseguera, and Nicki Hyde for their valuable assistance with the artwork and logistics during the course of preparation of the manuscript. FARZAD NAEIM Los Angles, California JAMES M. KELLY Berkeles, California CHAPTER 1 DEVELOPMENT OF SEISMIC ISOLATION WORLDWIDE 4.1 INTRODUCTION In August 1905 J. A. Calantarients, a medical doctor from the northern English city of Scarborough, wrote a letter to the Director of the Seismological Service of Chile in Santiago calling his attention to a method of building construction that he had developed whereby “substantial buildings can be put up in earth- quake countries on this principle with perfect safety since the degree of severity of an earthquake loses its significance through the existence of the lubricated ‘free joint.” Calantarients had submitted a patent application to the British patent office for his construction method, which proposed that the building be built on his “free joint” and a layer of fine sand, mica, or talc that would allow the ‘building to slide in an earthquake, thereby reducing the force transmitted to the. building itself. ‘What the dector was prescribing was an early example of an earthquake- resistant design strategy known as base isolation or seismic isolation. Many mechanisms heve been invented over the last century to try to achieve the ‘goal of uncoupling the building from the damaging action of an earthquake, for example, rollers, balls, cables, rocking columns, as well as sand. Buildings ‘have been built on balls, including a building in Sevastopol, Ukraine, and a. {five-story school in Mexico City. At least one building, a four-story dormitory ‘for the State Seismological Observatory in Beijing, has been built on a sand ‘ayer between the building and the foundation specifically designed to slip in ‘the event of an earthquake. Dr. Calantarients mentions in his letter that, “T made the experiment with ‘balls many years before it was done in Japan, or at any event before any amount 12 DEVELOPMENT OF SEISMIC ISOLATION WORLOWIDE of it appeared in the papers about 25 years ago.” The reference is almost cer- tainly to the Englishman John Milne, who was a Professor of Mining Engineer- ing in Tokyo in the years 1876 through 1895. During this period Milne became very interested in earthquake phenomena, and he devised and improved a num- ber of seismoscopes and seismographs. Milne carried out pioneering research ‘on seismology, so much so that he is often referred to as the “Father of Modern Seismology.” He also gave muich thought to the design of buildings in seismi- cally active areas and published rules for earthquake-resistant construction that are still valid today. ‘While at the University of Tokyo, he built an example of anisolated building. ‘The structure was built on balls in “cast-iron plates with saucer-like edges on the heads of piles. Above the balls and attached to the buildings are cast-iron plates slightly concave but otherwise similar to those below.” The building was instru- ‘mented and apparently experienced earthquake movement. In 1885 he described this experiment in a report to the British Association for the Advancement of Science. Apparently he was not completely satisfied with the performance of his building under wind load, so he changed the isolation system, and in 1886 he described the new version in a follow-up report to the association. The first balls had been 10 inches in diameter; 8-inch balls were tried, then 1-inch balls. Finally the house was rested at each of its piers on a handful of cast-iron shot, each 1/4 inch in diameter. By this means the building became stable against ‘wind loads. The final design was evidently successful under actual earthquake action. The concept of seismic isolation has become a practical reality within the last 20 years with the development of multilayer elastomeric bearings, which are made by vulcanization bonding of sheets of rubber to thin steel reinforcing plates, These bearings are very stiff in the vertical direction and can carry the vertical load of the building but are very flexible horizontally, thereby enabling the building to move laterally under strong ground motion. Their development was an extension of the use of elastomeric bridge bearings and bearings for the vibration isolation of buildings. In recent years other systems have been developed that are modifications of the sliding approach. The concept of base isolation is now widely accepted in earthquake-prone regions of the world for protecting important structures from strong ground motion, and there are now many examples in the United States and Japan. A smaller number of base-iso- lated buildings have been built in New Zealand and in Italy. mainly for large ‘and important buildings. Demonstration projects that apply low-cost base iso- lation systems for public housing in developing countries have been completed. in Chile, the People’s Republic of China, Indonesia, and Armenia, It is not surprising that most applications are for important buildings that house sensitive intemal equipment. The basic dilemma facing a structural engi- neer charged with providing superior seismic resistance of a building is how to ‘minimize interstory drift and floor accelerations. Large interstory drifts cause damage to nonstructural components and to equipment that interconnects sto ies. Interstory drifts can be minimized by stiffening the structure, but this: 44 INTRODUCTION 3 leads to amplification of the ground motion, which leads to high floor acceler- ations, which can damage sensitive internal equipment. Floor accelerations can be reduced by making the system more flexible, but this leads to large inter- story deifts. The only practical way of reducing simultaneously interstory drift and floor accelerations is to use base isolation; the isolation system provides the necessary flexibility, with the displacements concentrated at the isolation level "The concept of base isolation is quite simple. The system decouples the building or structure from the horizontal components of the ground motion by interposing structural elements with low horizontal stiffness between the struc ture and the foundation. This gives the structure a fundamental frequency that is much lower than both its fixed-base frequency and the predominant frequen- cies of the ground motion. The first dynamic mode of the isolated structure involves deformation only in the isolation system, the structure above being to all intents and purposes rigid. The higher modes that produce deformation in the structure are orthogonal to the first mode and, consequently, to the ground motion. These higher modes do not participate in the motion, so that the high energy in the ground motion at these higher frequencies cannot be transmitted into the structure. The isolation system does not absorb the earthquake energy, but rather deflects it through the dynamics of the system; this effect does not depend on damping, but a certain level of damping is beneficial to suppress possible resonance at the isolation frequency. “The fist use of a rubber isolation system to protect a structure from earth- quakes was in 1969 for an clementary school in Skopje, Yugoslavia. The Pestalozzi School, a three-story concrete structure designed and built by Swiss engineers (Fig. 1.1), is isolated by a system known as the Swiss Full Base Isolation-3D (FBI-3D) System [116]. Unlike more recently developed rubber bearings, the rubber blocks used here (Fig. 1.2) are completely unceinforced so that the weight of the building causes them to bulge sideways. Glass blocks Fig. 1.3) acting as seismic fuzes are intended to break when the seismic load- ing exceeds a certain threshold. Because the vertical and horizontal stiffness of the system is about the same, the building will bounce and rock backward and forward in an earthquake. These bearings were designed when the technology for reinforcing rubber blocks with stecl plates—as in bridge bearings—was not highly developed or widely known, and it is unlikely that this approach will be used again “Most recent examples of isolated buildings use multilayered laminated rub- ber bearings with steel reinforcing layers as the load-carrying component of the system. Because of the reinforcing steel plates, these bearings are very stiff in the vertical direction but are very soft in the horizontal direction, thereby pro- aie! 2.35) ‘The natural frequencies ? are given by Mé‘a? = Ko! and we assume that $'C/ = Oif ij ‘The matrix equations of motion reduce to the N + 1 equations D> MOG + (n+ meds + cobs + kev» = (m+ ma)ilg (2.36a) Git 2oiBiai+ Fa Lily tig) 021, 2 WN (2.360) where L, are the participation factors of the fixed-base modes, that is, The fixed-base modal masses are given by M,= "Mo! We can write these equations in the form (2378)34 THEORETICAL BASIS OF SEISMIC ISOLATION and Lady + 9, + Debian + 9? nN @smy qi = Lilly The complete modal analysis of these N + 1 equations is given in Kelly [65], where the procedure for the frequencies and mode shapes of the composite: system is described. In most cases the modes higher than the first mode play no role in the design of the structure or the isolation system; therefore only the: first mode needs to be included. ‘Comparing the equations of motion [Eqs. (2.37a) and (2.37b)] with the previ- ous set for the isolated 1-DOF structure (Eqs. (2.7a) and (2.7b)], the equations can be made to correspond if we replace v, in the elementary analysis with Livp, lig with Liidg, and 1 em wid _ iM, 1 nm siving My s, remy Ht Late) + ere a(Lity) + 03 Lay) = Lally (Lis) + Gy + 2e01Bidh = —Lalig In the solution of these equations, the result for q1 follows that for v, in the simple 1-DOF system. ‘The basic results for the 1-DOF structure, namely, that leolom = Sy SuCen Bs) @38 Ce = [Sa(wod, BE) +e7C1 ~ yPSZ0?, BY? 2.39) 22 EXTENSION OF THEORY TOBULOINGS 35, are replaced as follows. The maximum relative base displacement is given by Ulm = ty Ls Sar) 240) and because L; appears on both sides, the result is the same as before. To obtain the base shear, we have PLSHwE, 6), ALS ob ed (41) with «%, 6F calculated as before and ¢ replaced by ¢; = w3/a}. The relative displacement vector vis given by vege! 2.42) and if we neglect the damping, contributions, the inertial force on each element F= Ky=9:K6! = gMé'o? (2.43) ‘The total horizontal force on the superstructure is FR = giw2LiM, (2.44) and this is expressed in terms of the base shear coefficient Cs through Com = rR 2.45) Thus BM [3S2(es, Bs) + (1 ~ 71)e7L382(7, 8)? 2 FM (3005.64 —)PePSHoh ad? 2.46) with ¢ = w/a? as established previously.36 THEORETICAL BASIS OF SEISMIC ISOLATION 2.3 ANALYSIS OF COUPLED DYNAMIC EQUATIONS In most structural applications it is assumed that the damping is small enough that the effect of the off-diagonal components is negligible and the required solution can be obtained from the uncoupled modal equations of motions. The previous analysis neglects the off-diagonal components, leading to very simple results for base displacement, base shear, and interstory drift, In many isolated structures designed according to the most recent California, design codes, the code requirements are so conservative that the designers are using additional viscous dampers in an attempt to control the large design dis- placements, and damping factors for the isclation system of the order of 0.50 are obtained. Clearly, at this level of damping the equations cannot remain uncou- pled and a complex modal analysis should be used. In complex modal analysis, however, we lose the physical insight that led to the simple results of the uncou- pled solution (65). For this reason a similar approximation to that employed there will be used in this section to demonstrate the effect of high levels of damping in the isolation system on the response of the structure, It is interesting to note that to zero order in ¢, the four damping terms are 2a) = 2p — 2wafa = i= 2onBe 2 = 2s so that the off-diagonal components are ofthe same order as the diagonal terms, Recalling that Ly = O(1) and Lz ~ O(e), we assume that the influence of M42 01 the result for gis negligible but the influence of Xxg} on ga could be significant. ‘Thus, we assume that Eqs. (2.14) and (2.15) are modified, so that qi(#) is given by the solution of and qa(®) by GO) + 2wrBria + wi ge = Lally — rd ‘To aid in simplifying the solution, it is useful to take the Laplace transform of these equations using LIY@! fj 14-70 29 ANALYSIS OF COUPLED DYNAMIC EQUATIONS 37 In terms of the Laplace transform, we have Las) S24 2w stop Laas) + Dalasas) S?+2unBas+a} | (6? +2w2hys + wi\s? + 2a,B\s+ ay) = -L2A,()A(6) + XoL1A2(s)a(8) where G(s) = LT [ig]. The term A2(s) can be reduced by partial fractions to atbs e+ds AO= FT Raybis toy * 4 oabs FoF where after considerable manipulation we find 7 (e282 ~ 20181) wt - oF D D }(2u,62 ~ 20181) D 247 D= (3 - 03)? + QaaB2 ~ 2018 w} 20262 — 0320161) (2.48) ‘The inversion of the two terms of A2(s) follows from ]= eres a mle and 1 gat — esin ye88 THEORETICAL BASIS OF SEIGMIC ISOLATION so that the inversion of (a+ bs)/(s? + 2uiBis +0?) is x80 sin be" cos Bit + (a~ bwrBs) and that of (c+ 3)/s? + 2uaBps-+ 0} is eo sin Be deco Sy 4(¢~ das oy EN Bt where Bi=0i1~6)'* B= 020-6)? ‘The final result of g,(t) and ga(®) is obtained by convolution and substitution of a, b,c, and d from Eq. (2.47) as ao=-2 f. iig(t — Ne“ sin Bir de (2.49) a= —b S f e220 sin Bat r)iiglt) dr D +hals { j. e-POD cos Ty( ~ ahi) dr 7 Qu2B2) - (2 +36) 1 Hev810-" sin y(t — rilg(t) dr + SiGesbp a oiteiee fie u(t ita) a PEROT" p-erB20-9 c05 Tylt — Digs) de == i 0-9 es Galt — rigs) _ (at + wP)w282- w}208) 1 sw) sin Galt ~ r)iig(r) dr = fie (c= itn ar} (2.50) ‘These convolution integrals can be computed for any choice of iig(t), but for ‘the purpose of this demonstration, it is necessary only to have a sense of the ‘order of magnitude of the results. 23 ANALYSIS OF COUPLED DYNAMIC EQUATIONS 39 ‘Terms in 1, 2 can be expressed in terms of the nominal frequencies w», , by use of Eq. (2.9), from which we have —Ayepaj yt? ‘The denominator D of each term in Eq, (2.50) can be written as D=(w}— of) — dufahBh + BD) + doreor(o} + 0)6i62 which reduces to 1 2-2)? + dye? 2026762 Tage Me? eb? + tyuded 40 - yu2ufo36e +40 — 7) Peaseoe wo? + 03)B 82) 51) A further reduction of each term is possible if we assume the following ‘orders of magnitude: y= 00) e«1 To the first order in ¢, we have aoe B- 20-241 and the multipliers of each integral become40 THEORETICAL BASIS OF SEISMIC ISOLATION 1 $ Tey Ud 2nd or U+d- 2d wFQw262 eDurb Day (of + eDwrbe ~ 0320181 Dez sgiving for the four terms in parentheses in Eq. (2.50) {tsa -ane[ fl ete? cos aie nigin dr ae EAPL-D cos Balt — Diig(7) « -6 i eP=D sin By(t — Dig) dr Pa fe? sin e-em ae} ‘The results for g1 and q2 to first order in ¢ are thus Li al Lf eainnsn send 8 eoeabll-” sin Galt — Dil) dr 1d =2yen)| f° em 008 Tye — Nill) dr { If =f esc woe] = fi enme-M in Bite nin dr +62 fame sin mae -migin arb 2.53) 123. ANALYSIS OF COUPLED DYNAMIC EQUATIONS 41 It is convenient to denote the convolution integrals in Eqs. (2.52) and (2.53) by Ii, Tos Is, and 4, where Iy= ff rete sin Bide igh dr he I er ePO—9) sin Dat ~ r)iig(t) dr Iy= J) ee c06 Bilt — nih) dr Ie =f eh cos Gz(t — Diig(s) dr In this analysis the quantities of interest are the interstory drift and the floor accelerations, which are represented in this simple model by v, and d,. In this simple model they are related by Ks|Uslevax. so that the evaluation of v, will also provide the floor acceleration. The inter- story drift v, is given by ibs + 28 leading to Ly, 1 la u +tu-a-ya2n 7 l-d-na & 1 -S0-0-YePalr 7 Hd neni {U1 + (= 2ye)](l3 ~ La) ~ Bila + Bala) 2.54) Is useful here to separate the three contributions to the drift as follows: @ that produced by the base shear generated by the isolation system Sy @s5) yo Gi) that from the uncoupled modal equations42 THEORETICAL BASIS OF SEISMIC ISOLATION 1 w=tu-a-nea 2.56) 7 and Gi) that from the coupling terms, which is generally neglected in most anal- yses, 2 1S (01+ ~ 270s ~ fe) ~ ils + Bal} 1 2 -a-nePete 7 dane esr ‘The convolution integrals /1, [2, Is, and 1g can be estimated for the purpose of this demonstrative analysis by response spectrum methods. We recognize that FE Mila = S061, 80) Hallas = So2sB2) ‘where Sp is the displacement response spectrum. ‘The expression f. 8D 008 ealt ~ iit) de is the relative velocity response spectrum Sgy(w, 8) for a single-degree-of-free- dom oscillator of frequency « and damping actor 6. This we approxima by the pseudo-velocity response spectrum Sy(w, 8) given by wSp(w, B). The pe values of the four convolution integrals in parentheses in Eq. (2.53) will occur at different times and should be added by the SRSS method, leading estimates of the maxima of the three contributions to We have [AY eas = €L4 Sp(wo1, Br) U1. ~ yellaSolo2, B2) [OP ome = G= erat, 1% 7 3 l 7 1 Lo? Irae = — {+ = 27 lw ?S3(1, Bd + 0355 (2, Ba) +607 SH (01, B:) + 30285 (w2, Bad}? (2.58) 23, ANALYSIS OF COUPLED DYNAMICEQUATIONS 43 Al design codes for seismically isolated structures are based or constant-veloc- ity spectra, so that the various terms in the above can be related through Sv(w,6) = SvH@) where Sy is constant and H(8) is a suitable damping modification function that decreases with increasing 8 and is unity at 8 = 0.05. Many such functions have bbeen used, either as tables in code documents or as continuous functions. A particularly simple form is the Kawashima-Aizawa function [60] 1s HO= 108 +05 (259) where H(0) = 2, H(0.05) = 1, and H—0.5 as +e. Using a constant-veloc- ity spectrum of this form and the results for the modal quantities Ly, L>, 1, i -.. from the earlier section (after considerable manipulation), we obtain the following results: [Pima = ¢ SO 16) [ePna = = 99" St. 85) [WP Imex = 2eBp {[1 + 2(1 ~ 2y)e + BNH7(6,) 41420 ~2ye + sherGay” Se (2.60) Clearly, for small values of 8, say 8 = 0.10, the first term, | mus is the dominant term. For all values of 2p the second term, Ula, is always much less than the first term and is neglected, The significance of the third term, luP'Imax» depends on the value of 8. For the usual values of 6,, the value of Fis small compared to unity, so the ratio between them becomes ra 14 HG) + HB) oak 26o{C BREED Hf (B2)) 2.61) Now 6; = 6s and 1 1 B= Gaye B+ 1B)44 THEORETICAL BASIS OF SEISMIC ISOLATION Suppose we edopt the Kawashima-Aizawa formula for H(8) and take ¢ sk. 7 = }, and 6, = 0.02; then the ratio of the two terms is 033 when 6, = 0.10 nd increases 10 1.80 when By = 0.50. To put this into numbers appropriate for an example project, suppose that the code-mandated displacement at the MCE is 76 em (30 in.) for 59% damping and a period of 2.5 sec. To reduce this: to a more accepsable level, suppose that linear viscous dampers are added to bring the damping to around 50%, at which point the reduction factor is 0.57. ‘The displacement is now acceptable, and in code notation the elastic base shear becomes F, = KD, which before was and is now reduced to 0.285W, which, again, seems quite reasonable; however, the viscous force Fy=206MD which is out of phase with F,, is, for 8 = 0.50 and D = wD, exactly the same as F,, and the maximum base shear is V2F, =0.40W ‘The corresponding maximum floor accelerations given by [ils = 2a [iets = € and lite an = ©? » 2eB8s {{1 + 2C1 - 2ye + 8776) in Sy +1420 —2ye +BnAMB)P” = Zep BolL1 + 2U ~ Zy)e + BFJH7(1) + [1 +201 ~ 2y)e + B3H7(6.)}"? Sy 23. ANALYSIS OF COUPLED DYNAMIC EQUATIONS 45 Floor acosteration, g By Fig. 2.2. Infiuence of damping on floor accelerations. For the same example the these two components and their sum as @ function of By are shown in Fig. 2.2. We see that liis|max becomes equal t0 liis|max When Bp is around 0.26, but their sum has a minimum value at about 0.12. ‘This result implies that the addition of dampers (leading to large values of 8»), while controlling the isolator displacement by reducing vp, has the coun- tereffect of increasing the interstory drift and the floor accelerations. For a constant-velocity design spectrum the accelerations generated by the coupling terms become the dominant term. It is not widely appreciated that in base-iso- lated structures the higher modes, which carry both the floor accelerations and the interstory drift, are almost orthogonal to the base shear, so that a low base shear is not a guarantee of an effective isolation system. In this respect the effort to improve the performance of the system by adding excessive damping is a misplaced effort and inevitably self-defeating,CHAPTER 3 ISOLATION SYSTEM COMPONENTS 3.1 INTRODUCTION Base ‘solation is now a mature technology and is used in many countries, and there are a number of acceptable isolation systems, the construction of which is well understood. Nevertheless, the concept appears to have an irresistible attraction to inventors, and many new and different systems of isolators are proposed and patented each year. Many of these new systems will prove 10 ‘be impractical and some might actually be lethal, but the number continues to increase year by year. ‘Most systems used today incorporate either elastomeric bearings, with the elastomer being either natural rubber or neoprene, or sliding bearings, with the sliding surface being Teflon and stainless steel (although other sliding surfaces ‘have been used). Systems that combine elastomeric bearings and sliding bear. ings have also been proposed and implemented. ‘This chapter will discuss as many of the available systems as possible within the constraints of space and time. With the increasing number of new systems and modifications of existing systems, it is possible that some may be over- lookec. We apologize in advance to inventors or vendors of isolation systems that have been inadvertently excluded. 3.2 ELASTOMERIC-BASED SYSTEMS ‘Natural rubber bearings were first used for the earthquake protection of build- ings in 1969 for the Pestalozzi School in Skopje, Macedonia (see Chapter 1), 4748 ISOLATION SYSTEM COMPONENTS ‘The bearings are large rubber blocks without the steel reinforcing plates used today and compress by about 25% under the weight of the building. The bear- ings have a vertical stiffness that is only a few times the hozizontal stiffness and the rubber is relatively undamped. This system was tested on the shake table at the ERC in 1982 [117]. Characteristic of isolation systems of this kind, the horizontal motion is strongly coupled to a rocking motion, so that purely horizontal ground motion induces vertical accelerations in the rocking mode. ‘The system also has foam~glass blocks on either side of a rubber bearing that are intended to act as fuses to prevent movement in the building under wind, internal foot traffic, or low seismic input. The system is still in place and is ‘monitored from time to time. Since this building was completed, many other buildings have been built on natural rubber bearings but with internal steel reinforcing plates that reduce the lateral bulging of the bearings and increase the vertical stiffness. The internal steel plates, referred to as shims, provide a vertical stiffness that is several hun- dred times the horizontal stiffness. These multilayered elastomer bearings pro- vide vibration isolation for apartment blocks, hospitals, and concert halls built ‘over subway lines or mainline railroads. In 1975 Decham 2t al. [38] suggested that this approach could be used to protect buildings from earthquake ground ‘motion, and an intensive experimental and theoretical research program was begun at the ERC to develop this concept. Laminated elastomeric bearings ccan be differentiated into low-damping or high-damping types. 3.2.1 Low-Damping Natural and Synthetic Rubber Bearings Low-damping natural rubber bearings and synthetic rubber bearings have been widely used in Japan in conjunction with supplementary damping devices, such as viscous dampers, stee! bars, lead bars, frictional devices, and so on. The elastomer used in Japan comprises natural rubber, while in France neoprene has been used in several projects, The isolators have two thick steel endplates ‘and many thin steel! shims, as shown in Fig. 3.1. The rubber is vulcanized and bonded to the steel in a single operation under heat and pressure in a mold. The steel shims prevent bulging of the rubber and provide a high vertical stiffness but have no effect on the horizontal stiffness, which is controlled by the low shear modulus of the elastomer. The material behavior in shear is quite linear up to shear strains above 100%, with the damping in the range of 2-3% of critical. The material is not subject to creep, and the long-term stability of the ‘modulus is good. It is possible to manufacture an isolator with no damping and exactly lin- ‘ear shear behavior. Such bearings were made for an isolation system proposed for nuclear power plant application by the Central Electricity Generating Board in the United Kingdom. These bearings were intended to be used in conjunc~ tion with a viscous damper developed by the GERB Corp. in Germany [68]. ‘The combined system was tested at the EERC, and the tes's confirmed that the bearings were completely linear to 150% shear strain and completely without 82 ELASTOMENC-bASED sySTEMS 48 Fig. 3.1 Low-damping natural rubber bearing, damping. The intent of the design was to provide an isclation system that could exactly correspond to the linear viscous dynamic model. While the rubber fitted the model, the tests showed that the response of the damper was not exactly that of a linear viscous element. The advantages of the low-damping elastomeric laminated bearings are ‘many: They are simple to manufacture (the compourding and bonding. pro. cess to steel is well understood), easy to model, and their mechanical response is unaffected by rate, temperature, history, or aging. The single disadvantage is ‘that a supplementary damping system is generally needed. These supplemen- tary systems require elaborate connections and, inthe case of metallic dampers, are prone to low-cycle fatigue. ‘Many applications of this type of system have been used in Japan. The energy dissipation elements consist of a variety of steel-yielding devices, including tapered rods, coiled yielding springs, lead bars, and fricional elements. A vari- ant of this approach is the lead-plug bearing. Developec in New Zealand in the 1970s, the lead-plug bearing is now the most frequently used isolation system. 3.2.2 Lead-Plug Bearings ‘The lead-plug bearing was invented in New Zealand in 2975 (110, 111] and has been used extensively in New Zealand, Japan, and the United States, Lead-plug bearings are laminated rubber bearings similar to low-dimping rubber bearings bbut contain one or more lead plugs that are inserted iato holes, as shown in Fig. 3.2. The stec! plates in the bearing force the lead plug to deform in shear. The lead in the bearing deforms physically at a flow stress of around 10 MPa (1500 psi), providing the bearing with a bilinear response [129]. The lead must ‘it tightly in the elastomeric bearing, and this is achieved by making the lead plug slightly larger than the hole and forcing it in. Because the effective stff- ness and effective damping of the lead-plug bearing is dependent on the dis- placement, it is important to state the displacement at which a specific damp- ing value is required. Lead-plug bearings have been extensively tested in New Zealand [27], and there are very complete guidelines on their design and model-50 ISOLATON SYSTEM COMPONENTS. Superstructure anchor Pier anchor Elastonerc bearing Lead eS Fig. 3.2 Lead-plug isolator. ing (23, 80). Buildings isolated with these bearings performed well during the 1994 Northridge and 1995 Kobe earthquakes. 3.2.3 High-Damping Natural Rubber Systems (HDNR) ‘The development of a natural rubber compound with enough inberent damp- ing to eliminate the need for supplementary damping elements was achieved in 1982 by the Malaysian Rubber Producers’ Research Association (MRPRA) of the United Kingdom [39]. The damping is increased by adding extrafine carbon block, oils or resins, and other proprietary fillers. The damping is increased to levels between 10 and 20% at 100% shear strains, with the lower levels cor responding to low hardness (50-55 durometer) and a shear modulus around 0.34 MPa (50 psi) and the high levels to high hardness (70-75 durometer) and a high shear modulus [1.40 MPa (200 psi)]. The methods of vulcanization, bond- ing, and construction of the isolators are unchanged, “The material is nonlinear at shear strains less than 20% (Fig. 3.3) and is char- acterized by higher stiffness and damping, which tends to minimize response under wind load and low-level seismic load. Over the range of 20-120% shear strain, the modulus is low and constant. At large strains the modulus increases due 10 a strain crystallization process in the rubber that is accompanied by an increase in the energy dissipation. This increase in stiffness and damping at large strains can be exploited to produce a system that is stiff for small input, is fairly linear and flexible at design level input, and can limit displacements ‘under unanticipated input levels that exceed design levels. ‘The damping in the isolators is neither viscous nor hysteretic, but somewhat. in between. In a purely linear viscous element the energy dissipation is quadratic in the displacement; in a hysteretic system it tends to be linear in displacement. Tests on a large number of different rubber isolators at the ERC (Fig. 3.4) 32 ELASTOMERICHASED SYSTEMS 51 2607 Fo |-©-P-8.48 Mpa (50 pad a 0 hp 1000 st 200 {-o=p=10.34 Mpa (1800p anh 5 | 8 10.00 + i oo Le reaipenee | Sree ine eo Esai 0.00 | Z a ee ‘Shear strain (4) Fig. 3.3. Stress-strain characteristics of high-damping natural rubber isolators. demonstrate that the energy dissipated per cycle is proportional to the displace. ‘ment around the value of power 1.5, This characteristic can be exploited so that it is possible to model the bearing response, which combines linear viscous and elastic-plastic elements. Another serendipitous advantage of the high-damping rubber system is that it Provides a degree of ambient vibration reduction, The isolators will act to filter52 ISOLATION SYSTEM COMPONENTS Bearing ID: HR030-2 Vertical pressure: 1000 PSI Strain ranges: 5% 10% 25% 50% 75% 100% 150% 200% 250% 300% 350% Horizontal load/deflection 3. ISOLATION GYSTEMS BASED ON SLIONG 5 28 Cnamannenrs Ner000 20 BlUL In SALCTURE Se PUR TENACES TB ALT THERTONOTNTRGIRES a B68 Deflection (in.) Fig. 344 Hysteretic behavior of high-damping natural rubber isolators. out high-frequency vertical vibrations caused by traflic or adjacent underground railways. This effect was demonstrated in a shake table test program carried out at ERC in 1985 [39a 3.3 ISOLATION SYSTEMS BASED ON SLIDING A purely sliding system is the earliest and simplest isolation system to be pro- posed. A system using pure sliding was proposed in 1909 by Johannes Avetican Calantarients, a medical doctor in England. He suggested separating the struc- ture from the foundation by a layer of tale (shaded portion in Fig. 3.5). As is evident in his diagrams, Calantarients clearly understood that the isolation sys- tem reduced accelerations in the isolated building at the expense of large rela- tive displacements between the building and the foundation, for he designed a set of ingenious connections for utiities—in those days restricted to gas lines and sowage pipes—to accommodate these displacements (see Fig. 3.6). In fact, Calantarients’s system incorporated all the elements now considered necessary in a base isolation system: a method of decoupling the building and the founda- tion, a method whereby utility lines can withstand large relative displacements, and a wind restraint system. Fig. 3.5 Calantarients’s base isolation system using a layer of tale as the isolating medium, Isolation was first considered as a seismic-resistant design strategy by the Italian government after the great Messimo-Reggio earthquake of 1908, which killed 160,000 people in unreinforced masonry buildings—the typical build- ing type for the area. Almost all buildings of this type collapsed [19]. After the earthquake a commission was appointed to make recommendations for rebuild- ing the area with earthquake-resistant structures that were both economical and safe. The commission considered two approaches to earthqueke-resistant design: The first approach isolated the building from the ground by either inter- posing a sand layer in its foundation or using rollers under columns to allow the building to move horizontally; the second approach involved a fixed-base $) ‘Time history analysis may be used in lieu of response spectrum analysis, but if either the isolation system or the superstructure is highly nonlinear, a time history analysis is required, 4.4 STATIC ANALYSIS ‘The static analysis formulas provide displacements and forces and are based ‘on constant-velocity spectra over the period range 1.0-3.0 sec. In UBC.94 [54] and OSHPD-96, the value of the constant-velocity spectrum is derived from66 CODE PROVISIONS FOR SEISMIC ISOLATION the Applied Technology Council provision ATC-3-06 [13] and for Z = 0.40, a soil factor $ = 1, and 5% damping is 0.60 m/s (23.6 in./sec), leading to a displacement spectrum Sp given by Sy TZ om 4 Sp (0.60) = 0.2527 m (10ZT in.) ‘The spectrum is then modified by a soil factor and a damping factor and adjusted. for other seismic zones, leading to the required design displacement D. The three levels of displacement to be calculated are as follows: + D, the design displacement, being the displacement at the center of rigidity of the isolation system at the DBE; + Dr, the total design displacement, being the displacement of a bearing at 1 comer of the building and including the component of the torsional dis- placement in the direction of D; and + Dray, the total maximum displacement, being the total design displacement covaluated at the MCE. ‘The design displacement D in UBC-94 and OSHPD-96 is the starting point for the entire design process and must always be calculated whether or not dynamic analysis is used. It is based on the assumption that the superstructure deformations are negligible and is given by D a) O25ZNSITr (ease in) 2 a where Z = seismic zone coefficient (e.g., 0.4 in zone 4) N = near-field coefficient 51 = soil coefficient T; = effective period in seconds B = damping coefficient ‘While the concept is still the same, the UBC-97 formulation is more com- plex. A large number of new terms have been added to the code. For example, there are now six different displacements that have to be computed, The num- ber of soil profile types has been increased to six, of which three are hard rock, rock, and soft rock. There are four seismic coefficients to be calculated, but in zone 4, where most isolated buildings in the United States are located, it is necessary to calculate the following factors: NV, and Ny, which depend on seismic source type and seismic source distance; My, which depends on ZN; and Cay and Cys, which depend cn My, Na, and Ny. The result is that the simple static analysis computation of the earlier versions of the code has been replaced by a sequence of table definitions and formulas. 44 STATIC ANALYSIS 67 Although all isolated projects are currently designed using dynamic analysis (based on time histories, as there are many computer programs now available {or this purpose), stauc analysis is stll required 10 ensure that the design quanti- ties do not fall below certain minimal levels determined from the static analysis. ‘According to UBC-97, the two basic displacements to be calculated are Dp and Dy, or DBE and MCE displacement at the center of rigidity of the isolation system. They are calculated using the formulas Dp= (etre vere, ‘mm or in. 42) Du = eu mm or in, 43) where g is the gravitational acceleration, Cy and Cy are seismic coefficients, Tp and Ty are isolated periods, and Bp and By are damping coefficients cor- responding to the DBE and MCE level responses, respectively. The terms Cvp, and Cyjq are functions of the seismic zone factor Z, the site soil profile type, and one of the two near-source factors, Ny. Definitions of these and other factors needed for calculating Dp and Dy are presented in the following sections. 4.4.1 Seismic Zone Factor Z Seismic zone factors vary from 0.075 for zone 1 to 0.40 for zone 4 as shown in. Fig. 4.1 (UBC-97, Table 16-). Seismic zone factors are the same as in previous editions of the UBC. 1 2A 28 a 4 Fig. 4.1 UBC 97 ceiemic zone factors.88 CODE PROVISIONS FOR SEISMIC ISOLATION 4.4.2 Site Soil Profile Type Site soil profiles of Sq to Sz are based on the average shear wave velocity in the top 30.5 m (100 ft) of soil (see Fig. 4.2 or UBC-97, Table 16-J). This velocity varies from under 180 m/s (590 ft/sec) for the soft soils (Sz) to over 1500 m/s (4921 ft/sec) for the hard rock profile (S,). It should be noted that S, rock in continental United States is found only in eastern United States. Hard rock found in California is classified as Ss. for example. Another class of soil profile type (Sp) requires site-specific evaluation and is not classified based on average shear wave velocity and constitutes very poor site soil conditions prone to liquefaction problems. 4.4.3 Seismic Source Types: A, B, and © Seismic faults are grouped into three categories based on the seriousness of the hazard they represent. Faults capable of producing large-magnitude earthquakes (0 2 7.0) and have a high rate of seismic activity [annual average seismic slip rate, SR, of 5 mm (0.2 in.) or more] are classified as type A sources. Faults capable of producing moderate-magnitude earthquakes (M < 6.5) with a rela- tively low rate of seismic activity [SR <2 mm (0.08 in.}] are classified as type C sources. All faults other than types A and C are classified as type B sources. 444 Ne ‘Source Factors: N, and N, ‘Two factors are used to model the ground motion amplification due to near- source effects. The first, Ng, is intended for the short-period range correspond- ing to a constant-acceleration segment of response specira. The second factor, Ny, which corresponds to the midperiod range ot constant-velucily segaient of the response spectra, is the primary near-source factor used in seismic isola- tion applications. Near-source factors are functions of closest distance to the seismic source and the seismic source type (see Figs. 4.3 and 4.4). UBC-97 So Se Se Sa Fig. 4.2 UBC-97 soil-type categories (shear wave velocity range is shown to the right of site profile type), Solsmic Source Type Closest Distance to Known Seismic Sources Fig. 43 Nearsource factor Ne as a function of source type and distance. defines the site-source distance as the closest distance between the site and the vertical projection of the fault on the surface (see rg in Fig. 7.7). The surface projection need not include portions of the source at depths 10 km (6.2 miles) oo greater; therefore, a site sitting directly on the top of a fault deeper than 10 kan (6.2 miles) is not considered a near-source site. 82 i sotmte source = ‘Type Closeet Distance to Known Seismic Source Fig. 44 Near-souree factor Ny as a function of source type and distance,70 CODE PROVISIONS FOR SEISMIC ISOLATION 4.45 MCE Response Coetticient My ‘The MCE response coefficient My is intended to estimate the MCE response based on the DBE shaking characteristics. As such, Mjy is defined as a function of ZN, and varies from 2.67 for ZN, = 0.075 to 1.20 for ZNy 2 0.50 (see Fig. 4.5). The logic for assigning larger My values to smaller DBE events stems from the fact that in regions with low seismicity the gap between the DBE and MCE events is generally much larger than that in the zones of high seismicity. ‘Values of Myy are listed in UBC-97, Table A-16-D. 4A6 Spectral Seismic Coefficient ‘These coefficients are intended to define the minimum spectral ordinates to be used in design. The terms Cyp and Cap correspond to constant-velocity and corstant-acceleration regions of the DBE spectrum, respectively; Cvay and Cau perform the same function for the MCE spectrum. For seismic isolated struc- tures, Cyp and Cap are the same as Cy and Ca defined for conventional struc- tures by UBC-97, Tables 16-Q and 16-R. The values of Cyy and Cay, however, are given in the Seismic Isolation Appendix of UBC-97 (Tables A-16-G and AICP), ‘As shown in Figs, 4.6 and 4.7, Cyp and Cap are functions of seismic zone factor and site soil profile type. Also notice that for zone 4 the values shown in these figures must be multiplied by the appropriate near-source factor Ny or Na Cyps Sym and Cap: Cay Cootficiont, My MCE Response 19 Intensity, 2M, Fig. 45. MCE response coeficient Mu. 44 STAC ANALYSIS. 71 Sa Se Se So Se Soil Profile Type Fig. 4.6 Seismic coefficient Cyp. ‘Sesmic Coefficient, Cy ‘Similar information for Cyyy and Cay is shown in Figs. 4.8 and 4.9. Here, the MCE response coefficient comes into play. The values shown in these figures for MyZN, or MyZN, values larger than 0.40 should be multiplied by the factor indicated in the figure. 44.7 Damping Coetticients: Bp and By The effective damping in the system, 6, at the DBE and MCE response levels, (referred to as Bp and By) are computed from 1. ( total area of hysteresis loop 80 = 3 ( Ss ) (4a total area of hysteresis loop ( Kyu,maxDyy os Ss Se Se SS Soil Profile Type Fig. 4.7. Seismic coefficient Cap.72 CODE PROVISIONS FOR SEISMIC ISOLATION [aMGaN ore lemyzny -0.18 JaMZN, =0.20 JaMZNy =0.90 yyZNe0.40(Mutibiy SS Se SCS. Soil Profile Type Fig. 48 Seismic coefficient Cv. where Kp,max and Ky,max are effective stiffness terms defined in Section 4.4.8. ‘The damping reduction factor B (Bp for the DBE and By for the MCE) is given in terms of 6 in tabular form (UBC-97, Table A-16-C), with linear interpolation to be used for intermediate values. A very close approximation to the table values is given by 1 a - 46 25(1 - In 8) 46) where f is given as the fraction of critical damping (not as a percentage). Values. of B from the code and the formula are shown in Fig. 4.10. [BM yZN, 0.075 lm MyZNg =0.15 lO MyZN, =0.20 lea MyZN, =0.90 (ZN,20,40 (Mult FE RzNe30 40 (Matty by Soil Profile Type Fig. 49° Seismic coefficient Cam. 44 staTiC ANALYSIS. 73, 8 from UBS [28 from Formula Critical Damping Ratio Fig. 4.10 Damping coefficient B from UBC and formula, 4.48 Effective System Vibration Periods: Ty and Ty ‘The periods Tp and Ty that correspond to the DBE and MCE response are computed from an 68) ‘weight of building gravity (Fi ~ F5)/(@5 - D5) (Fy ~ Fia/ Dia ~ Pid minimum value of Kp,. &t Dp as determined by testing maximum value of Kp,en at Dp as determined by testing Kunin = minimum Value of Ky, at Dy as determined by testing Kuycnae = maximum value of Ky.er at Dy a8 determined by testing The values of Kymin Kp,maxy Ka,mins 20d Kiy,mex are not known to the designer during the preliminary design phase. The design procedure will begin with an assumed value of Ker, which is obtained from previous tests on simi- lar components or by using the material characteristics and a schematic of the proposed isolator. After the preliminary design is satisfactorily completed, peo-74 CODE PROVISIONS FOR SEISMIC ISOLATION totype isolators will be ordered and tested, and the values of Kp, mint Kp.moxe Krnin» 2nd Ki4,nax Will be obtained from the results of the prescribed program of tests on the prototypes. The terms Fs, Fp, Fig, Fig and Dp, Dp, Diy, Dy are the maximum and minimum forces and displacements on the prototype bear- ings, corresponding to DBE and MCE response levels, used to determine the ‘mechanical characteristics of the system. The results of the prototype tests are then used to refine the preliminary design, and when dynamic analysis is used, they establish bouncs on the various design quantities. Because the effective stiffness and the effective damping are usually dependent on the displacement, the process of compating the effective system periods and damping is an iter” ative one. 9 Total Design Displacements: Dyp and Dry ‘The total design displacements Drp and Dry (which include torsion) are given ine Dro=Do(1+y a>) a9) be nue bu(t49 pp) ao ‘where e is the actual sccentricity plus 5% accidental eccentricity and y is the dis- tance to a corner perpendicular to the direction of seismic loading. This formula assumes that the seismic load KeD is applied through the center of mass, which is located at a distance e from the center of stiffness (as shown in Fig. 4.11) Assuming a rectangular plan, with dimensions 6 x d and a uniform distribution of isolators, the torsional stiffness of the isolation system is Kea(b? + d?)/12, and the rotation @ is thus 6 KyDe_____12De Kal +a Fat ‘The additional displacement due to rotation is 12De wai” leading to Egs. (4.9) and (4.10) above. If the actual torsional stiffness of the system is computed and the additional displacement due to KeD through € ‘turns out to be less than the value given by Eqs. (4.9) or (4.10), then this value ccan be used, but it mast be at least 1.1 times Dp and 1.1 times Dy, respectively. 44 STATIC ANALYSIS 75 Fig. 411 Plan dimensions for celoulation uf Drp and Dra ‘The total maximum displacement Dry is required for verification of the stability of the isolation system. 4.440 Design Forces ‘The strength design forces that the superstructure and the elements below the: isolation interface are to be designed for are based on the design displaceme D. Elements below the isolation system (working stress design) are calculated using the formula Vo = Ko,naDo ay ‘The strength level for design of elements above the isolation system in terms of the minimum lateral seismic shear force is given by the formula Ko,muDo v, Ri (412) where R; is a design force reduction factor (ductility factor) ranging from 1.4 to 2.0. Table 4.1 shows a few examples of fixed-base structural systems with corresponding reduction factor R. In all cases the value of V, should not be less than + the seismic force requited by the UBC provisions for a fixed-base structure, + the base shear corresponding to design wind load, and + one and a half times the lateral force required to fully activate the isolation78 CODE PROVISIONS FOR SEISMIC ISOLATION TABLE 4.1 Reduction Factors for Fixed-Base and Isolated Construction Construction Rr R Special moment-resisting frame 20 as Shear wall 20 33 Ordinary braced frame 16 36 Eccentric braced frame 20 70 system, that is, the yield load of a lead-plug rubber bearing or slip threshold of a sliding bearing system. ‘The reduction factors for fixed-base design are very much higher than those fer isolated design for a number of reasons; one major element is period shift. “As the structure yields, the period lengthens and the force demand is reduced. ‘Simultaneously, the damping in the structure is increased because of hysteretic action due to yielding in the structural system. In addition, overstrength and redundancy tend to spread the yielding to other elements. In the case of an isolated structure, only overstrength and redundancy are applicable. For example, if a moment-resisting steel frame is used for the superstructure, ‘the reduction factor allowed by the code is 2.0 to allow for overstrength and ‘redundancy, implying that (a) the structural system will yield at a force level that is twice the nominal yield level and (b) the system will be just on the verge of actual yield at the forces specified by the code formula. Period shift in the structure counters the effectiveness of the isolation system because it decreases the separation between the fixed-base period and the isolation period and could bring larger forces to the structure and more participation from the higher modes. In addition, the damping in the isolated structure will not be as great as in the fixed-base structure. For all these reasons, the Ry factors are substantially smaller than the R factors for the same struczural system. Note, larger ductility demands can mean damage to the structural and nonstructural ‘components; thus, the requirement for lower Ry values is equivalent to damage control for isolated structures. 44.11 Vertical Distribution of Force Ir earlier versions of the code, the vertical distribution of te inertial forces on the structural system was based on the assumption that the participation of the higher modes was negligible and that the accelerations wer roughly the same at all levels of the structure, There was some concern, however, that this might not be sufficiently conservative, and the vertical distribution was changed in subsequent editions of the UBC code to one where the lateral force at level x, denoted by F,, is computed from the base shear V, by 44 STANCANALYS'S. 77 Tien where w, and w; are the weights at level i or x and h, and hy are the respective. heights of structure above isolation level. ‘This formula leads to a triangular distribution of force, While the basic theory ‘would indicate that the distribution should be close to a uniform one, a triangu- lar distribution is specified to account for higher mode contributions generated by nonlinearities in the isolation system, for example, because of lead plugs in clastomeric bearings or the effects of friction in sliding bearings. ‘What are the implications of this code requirement in terms of a seismic base shear coefficient Cs. If Eq. (4.7) is substituted into Eq. (4.1) and divided by the structure weight W to give Cs, we find FeaVs ap Ve NZS AL For example, if N = 1, Z = 0.40, and § = 1.40 (UBC-94 $; soil type, which is roughly equivalent to UBC-97 S¢ soil profile type) and the system has 10% equivalent viscous damping, then so that for a frame structure with Ry = 2 and a period of 2.0 sec the code prescribed Cs is 0.1175. The comparable Cs for a fixed-base structure in the same zone and soil type is given by 060 1 OS TERR and because R= 8.5 when R; = 2, the design shear coefficient is equal o 0.0706 for #1.0-sec period fixed-base building. Clearly, the reductions in the seismic force that are expected from isolation (around ¢'/*) are not available to the designer 4.4.12. Drift Limits ‘The maximum interstory drift limits for isolated buildings are also more severe than the limits for fixed-based buildings and should not exceed 0.01/R, (Le, the clastic deformation due to Ko, mmDo applied in a triangular pattern to the structure7B CODE PROVISIONS FOR SE:SMIC ISOLATION should not exceed 1%), This limitis less than half of 2.5% permitted for fixed- ‘based buildings with fundamental periods of less than 0.70 sec and half of the 2% limit permitted for fixed-based buildings with longer fundamental periods, 4.5 DYNAMIC ANALYSIS Design spectra: Site-specific specira are required if + Tw 23.0 sec oF «+ the soil type is Se, Sp or + the structure is located withia 10 km (6.2 miles) of an active fault Dynamic analysis is also required ifthe effective period of the isolated structure, Tp, is greater than three times the elastic fixed-base period of the structure ‘above the isolation system. If a sive-specific spectrum is used, it may be higher, but if smaller, it cannot be less than 80% of the codified response spectrum. 4.5.1. Time History Analysis Pairs of horizontal components from at least three recorded events are neces- sary for a time history analysis. The events must be representative of the site, soil, and source characteristics and have durations consistent with the DBE and MCE. Time histories developed for a site within 15 km (9.3 miles) of a major active fault are required to incorporate near-fault phenomena, although near- fault phenomena are not defined. More detailed information on time history analysis requirements are given in Chapter 7. 4.5.2 Scaling For each ground motion pair, the SRSSs of the 5% damped spectra are com- puted. The motions are then scaled (multiplied by a factor) so that the average Of the SRSS spectra does not fall below 1.3 times the target spectrum for the DBE or MCE by more than 10% over 0.57 seconds to 1.257 seconds. When dynamic analysis is used, the design values are calculated in the following way: + If three time histories are used, the design must be based on maximum. response quantities. «= If seven times histories are used, the design can be based on average response quantities ‘When dynamic analysis is done, it is possible to have design displacements ‘and design forces that are less than those given by the equivalent static formulas. 45 DYNAMICANALYSIS. 79 Code-specified limits, which limit to what extent the design values can fall below the static values, are summarized in Table 4.2 The total design displacement Dro for the isolation system can be reduced to not less than 90% of that given by the static formula, and the total maximum displacement Dryy can be reduced to not less than 80% of the static formula result. The Drp and Drye are computed from Dp and Dy by the use of multi- pliers, and the code permits a further reduction by replacing Dp and Dy in the statie formulas by Dp, and Dfy, where Dp= —— 22 = aor (4.14) Du Di = ——2*__ Vis C/tw (4.15) with 7 being the elastic fixed-base period of the superstricture com the empirical formula ofthe code, eae This further reduction is to allow for the flexibility ofthe superstructure. The static formulas [Eqs. (4.2) and (4.3)] assume that the supersructure is rigid and that if some deformation takes place in the superstructure, the displacement in the isolation system is reduced. But by using the two-degree-of-freedom (2-DOF) model developed earlier and ignoring the approximations for small 6, it is possible to show that this formula is not quite correst. This straightfor- ward analysis and the result, which depends on mass ratio and frequency ratio «24/, and is shown in Fig. 4.12, demonstrate that the formula corresponds to the result for 7 = 1, corresponding to my = 0 (jc. a single-degree-of-freedom system). For all values of 7 < 1 and for «p/w, < 1, the analysis shows that the correction [Eqs. (4.14) and (4.15)] overestimate the reduction in Dp and Diy due to flexibility in the superstructure, TABLE 4.2 Code Minimum Values When Dynamic Analysis fs Used Parameter ‘Static Response Spectra Time History Dro Dro = 1.10Dp 090075 080Drp, Drm Drm = 1.104 80D ry 0.80074 Vs kp,saaxDD 20.90V, 20.90V5 Ve regular kp,maxDo/Ry 20.800, 20.607, VY, irregular kp, maxD0/R} 2L0V, 20.8Vs. 0.010/Rr o.015/Rp 0.020/Ry80 CODE PROVISIONS FOH SEISMIC ISOLATION yD or - 0 or oa oe ™ Fig. 4.12 Reduction of design displacement dve to flexibility inthe superstructure 46 OTHER REQUIREMENTS FOR NONSTRUCTURAL, COMPONENTS. For strength design of nonstructural components, either fixed-base requirements. may be used or the peak response seismic force computed. Components that cross the interface should be designed for Dray, while fixed-base design proce dares are used below the isolation system. In addition, there are other detailed. system requirements that must be considered; these include environmental con- ditions, wind forces, fire resistance, and lateral restoring force. 47 PEER REVIEW ‘The code mandates that there be a peer review of the design of the isolation system. The review team should include persons licensed in the appropriate dis- ciplines and experienced in the theory and application of isolation. The review should evaluate the seismic criteria, the preliminary design, and the final design, In addition the panel should review the prototype test program and the pro- posed quality control test program. The peer review panel normally includes ‘threa people: a structural engineer, an expert in ground motions, and an expert in seismic isolation. 48 DESIGN AND TESTING REQUIREMENTS FOR ISOLATORS 81 4.8 DESIGN AND TESTING REQUIREMENTS FOR ISOLATORS ‘There are a number of design requirements for the isolator units that are derived from loads produced on the isolators due to overturning of the building as a result of horizontal accelerations. The building has to be checked for global overturning at the MCE with the full dead load being used in the calculation, Uplift of individual isolators is permitted when this check is made, When some isolators uplift, others will experience increased downward load, and the isol tors are required by the code to be designed and tested for this increased down- ‘ward load as well as net upward load, if any. The precise design requirement states that an isolator should be stable when displaced to the total maximum displacement Dry under 1.2DL + L.OLL + Emax and 0.8DL~ Emin at the MCE, where Emax is the maximum downward and Enig is the minimum downward (or meximum upward) vertical load on an isolator caused by overturning of the superstructure. The increased or decreased dead load is intended to provide an allowance for vertical accelerations in the ground motion. This requirement appears again in the testing requirements for the isolators. In many isolation projects the information needed for the preliminary design of the isolation system and the superstructure is obtained by qualification or pre- prototype tests that are not covered by the regulations. After the preliminary design is completed, isolators are manufactured, and a very extensive program Of prototype tests are performed. The code requires that at least two full-sized specimens of each type of isolator be tested. The tests required are a specified sequence of horizontal cycles under DL + 0.SLL for small horizontal displace- ‘ments up to the total maximum displacement, with these primarily establishing the mechanical characteristics of the bearings for use in verifying the design. A sequence of extreme load tests are required where horizontal displacement cycles are combined with maximum and minimum downward loads. The maxi- mum vertical load for these tests is defined as 15DL + 0.5LL + Emm and the ‘minimum is 0.8DL ~ Emig. In some cases it is possible that the minimum load ‘may actually be a tension load. Although rubber bearings can take a certain amount of tension, it ie very dif ficult to carry out such a test because most test machines are unable to generate tension forces and apply shear displacements simultaneously. Upward forces on isolators tend to occur in designs where the horizontal forces in the superstruc- ture are carried by a few lines of lateral resistance. This leads to high over- turning forces on a few isolators because the dead load on these isolators is ‘not enough to overcome the uplift forces. If numerous columns participate effi- ciently in resisting the horizontal load in the superstructure, the dead load in each column will eliminate the possibility of uplift on all bearings, avoiding the problem of designing the isolators for tension forces, thereby simplifying the testing requirements. Note, if tension is produced in the bearings, not only will be bearings need to be tested in tension, but the connections of the bearings to the foundation and to the building will also need to be tested,(CODE PROVISIONS FOR SEISMIC ISOLATION ‘There are other prototype test requirements in the code that are not generally strictly adhered to because of the lack of suitable testing facilities. The code requires that isolators with a behavior that is dependent on loading rate be tested at real loading rates. Rate dependence is defined as follows: The hysteresis loop fat the design displacement at the frequency of the isolated structure does not differ by plus or minus 10% if the tet is carried out at any frequency in the range from one-tenth to twice the design frequency. Although natural rubber is relatively insensitive to rate of loading and is well within this range, other polymeric materials and elastomers may not. The power requirements for a test of this Kind are quite surprising. If two isolators carry, say, 500 tons (L111 kips) {for a system with a 2.0-sec period and 30 cm (11.8 in.) displacement) fone on top of the other, as is the usual test method, and are loaded to their ‘design displacement at 2.0-sec period in a sinusoidal manner, the peak power requirement is about 1800 hp or 1.4 MW. ‘Similar unattainable test requirements are specified for units dependent on bilateral loads where it is stated that these units must be tested in two directions simultaneously. The eriterion for bilateral dependency is similar to that for rate dependency, and the requirement is never carried out because of the lack of suitable test equipment. ‘Although there are no specific requirements for quality control tests for pro- auction isolators, the code demands that such a program be established and reviewed by the peer review panel. The design engineer usually specifies a quality control program that includes a series of standard tests on the material, Compression tests, and combined compression and shear tests on all production isolators. In addition, a small percentage of isolators may be selected at random, land tested to the MCE displacement. Extra isolators are often kept under load at the site for periodic checks of the long-term stability of the material properties. Although this is not required by the code, the code does specify that access for inspection and removal of all isolators be part of the design. 49 OSHPD-96 REQUIREMENTS. ‘The design of hospitals in California is governed by the State Code Title 24, Part 2, which is administezed by OSHPD [15]. The current regulations published in May 1996 for base-isolated hospitals are very similar to the 1994 UBC require- ments, except that both site-specific ground motions and dynamic analysis are always required. The site-specific ground motions are evaluated for OSHPD by the California Department of Mines and Geology (DMG), and their approval of the design spectra for the DBE and MCE is necessary. ‘The civil code and the UBC differ in the following respect: Static analysis cannot be used for the design, only for minimum requirements; therefore, even, though dynamic analysis is mandatory, a static analysis still must be carried ‘out. The dynamic analysis must include @ time history analysis, except under 49 OSHPDS6 REQUIREMENTS 8S. very restrictive circumstances when a response spectrum analysis is accepiable. Site-specific design spectra have to be prepared for DBE and MCE events, From these site-specific design spectra, time history pairs of recorded events. are determined by using the same scaling procedure outlined in the UBC. ‘The duration of the selected and scale time histories has to be consistent with, the magnitude and source characteristics of the site-specific DBE and MCE cevents, and if the site is near field, then near-field effects should be incerpor- ated, “The requirement for design review has become considerably more restrictive. “The enforcement agency (e.g., OSHPD) retains the right to require the owner of the facility to retain a review panel and have the review panel serve in an advisory capacity to the agency. The testing requirement for the protorypes are unchange “The civil code has additional requirements for hospital buildings that are not mandatory in the UBC code. The isolation system must be monitored for the life of tie building and access for monitoring and replacement of the system must be provided. The building is to be instrumented with accelerometers that must be approved by the enforcement agency prior to installation. After every signi‘icant earthquake, the structure, the isolators, and the strong motion recorders must be inspected by a structural engineer retained by the owner, and an inspection report must be provided to the enforcement agency. In addition, a proportion of the isolators must be removed every 10 years to be tested and compared with the tests done prior to installation. ‘Another problem facing the designer is the difficulty of obtaining the approval from the DMG for the design ground motions. In the recent past the DMG has taken a very aggressive role in the selection of design spectra and cearthquakes for time history analysis, often resulting in considerable delays in the approval process. AS a result, the designer may adopt a very conservative ‘ground motion in the hope that a rapid approval results, thereby significantly impacting the cost and performance of the isolation system. ‘As the codes for seismic isolation design have evolved in the United States, the trend has been from the simple, straightforward, and rational version of the Yelow Book to the more complicated and less rational versions in the more recent codes. The Yellow Book was put together by a small group of activists ‘who t00k the position that base isolation was a simplifying approach to se:smic design and formulated a design process that was informed by the elementary theory. The requirements were consistent with the theoretical basis of isolation and included incentives for designers to use designs that were consistent with the simple theory. AAs the codes have evolved, however, they have been influ- enced by professional engineers who are not comfortable with the technology and see it as too complicated—even dangerous. Their influence on the codes has made the codes correspondingly inconsistent with the theory and has fur- ther complicated the process by insisting on excessive testing requirements at ‘every level of the procurement process.84 CODE PROVISIONS FOR SEISMIC ISOLATION 4.10 OTHER CONSIDERATIONS ‘One feature that has persisted through all versions of the UBC isolation regula- tions is the scaling of the time histories. In essence, the code requires an increase of 30% in the target spectra to account for bilateral ground motion. Isolation systems are always isotropic, however, and the maximum isolator displacement ‘ean be in any direction. Since the basic static formula for maximum displace- ‘ment is intended to be applied in any direction, it is not clear why the dynamic analysis should include bilateral displacements. Furthermore, this requirement is in direct conflict with the detailed systems design requirements for conven- tional structures, which states that the “requirements of orthogonal effects may be satisfied by designing such elements for 100% of the prescribed design seis ‘mie forces in one direction plus 30% of the prescribed design seismic forces in the perpendicular direction” (UBC-97, Sec. 1633.1) [55]. Obviously, the result- ing vector summations are very cifferent (sce Fig. 4.13). "The extensive testing requirements for prototype isolators remain from the earlier code versions. New requirements for inspection and replacement have bboen added, including requiremeats for periodic monitoring, requirements on repair or retrofit of the isolation system, and a requirement for a horizontal displacement monitoring device. In total, the 1997 version of the UBC regulations for seismic-isolated struc~ tures has completed the process of turning the simple straightforward and ratio nal code developed in the 1986 Yellow Book into a complicated and conser. vative set of requirements that will seriously undermine the use of isolation technology by the general engineering community. The whole impetus for de- 180% rule Y-Force 100%-80% rule 100 X-Force Fig. 4.13 Inconsistency between code requirements on orthogonal effects. 4.11 STEPBY.STEP PROCEDURE FOR UBC.97 COMPLIANT DESIGN BS veloping isolation systems by creating cost-effective, simple, strategies to create: earthquake-resistant structures has been lost. 4.11. STEP-BY-STEP PROCEDURE FOR UBC-97 COMPLIANT DESIGN 4.11.1 Preliminary Design Steps ‘Step 1: Establish Seismic Zone Factor Z. Determine the seismic zone factor by ‘establishing the seismic zone of the project’s site from UBC-97, Fig. 16-2, and the corresponding seismic zone factor Z érom Fig. 4.1 (UBC-97, Table 16-D. Step 2: Establish Site Soil Profile Category. Determine the site soil profile type from Fig. 4.2 (UBC-97, Table 16-1). Step 3: Establish Seismic Source Types. For each controlling seismic hazard ‘source (Le., seismic faults), determine the corresponding seismic source type from Section 4.4.3 or UBC-97, Table 16-U. Step 4: Establish Near-Source Factors N, and N,. For each seismic source type established above, determine corresponding near-source factors N and My from Figs. 4.3 and 4.4. (UBC-97, Tables 16-S and 16-1). Step 5: Calculate Maximum Capable Earthquake Response Coefficient My. Multiply Z and V, as calculated above to obtain ZN,. Use Fig. 4.5 (UBC-97, Table A-16-D) to read the corresponding value of My. Step 6: Determine Seismic Coefficients Cyp and Cap. With seismic zone factor and site soil profile as established in the above steps, obtain appropriate seis- mic coefficients Cy and C4 from Figs. 4.6 and 4.7 (UBC-97, Tables 16-R and 16-Q). Call them Cyp and Cap, respectively. ‘Step 7: Determine Seismic Coefficients Cy and Cay. Using soil profile type determined in step 2, multiply My, Z, and N, to obtain MyZNy. Use Fig, 4.8 (UBC-97, Table A-16-G) to obtain Cy. Similarly, calculate MyZN, and use Fig. 4.9 (UBC-97, Table A-16-F) t0 obtain the value of coefficient Came ‘Step 8: Determine Structural System Reduction Factor Ry. Obtain the reduction factor Ry corresponding to the structural system used above the isolation interface from Table 4.1 or UBC-97, Table A-16-E. The value of Ry for virtually all cases is either 2.0 or 1. Step 9: Select the Type of Isolation Bearings and the Damping Coefficients Bp and By. Select the bearing type (or combinations) to be used. Using the information presented in Chapter 5, select an appropriate conservative esti- mate of the damping level provided. For example, for high-damping rubber systems, use a damping level of 10-12%. With this information use Fig. 4.10 (UBC.97, Table A-16-C) and read the corresponding value of the damping86 CODE PROVISIONS FOR SEISMIC ISOLATION coefficient. Assign the same value to Bp and By at this time, Use linear interpolation if necessary. ‘Step 10: Select a Desired Isolated Period of Vibration Tp. Decide on an initial estimate for the isolated system fundamental period of vibration at the design basis displacement level. Generally speaking, an isolated period between 2.0 and 3.0 sec is desirable. Step 11: Estimate the Effective Stiffness of the Isolation System. Use a simple equation [such as Eq. (4.7)] to estimate the stiffness of the isolation system for the isolated period established in step 9. Assign this value to both Kp, min and Kp,nsx for the moment, ‘Step 12: Estimate the Minimum Design Lateral Displacement Dp. Use Eq. (4:2) ‘with the values calculated in the previous steps and calculate the initial esti~ mate of the minimum design displacement Dp. If this value is larger than what is acceptable for your project, you need a stiffer system. Go back 10 step 10 and start with a smailer estimate of the vibration period. Otherwise, proceed 10 the next step. Step 13: Establish the Minimum Design Lateral Forces V» and V,. Use Eqs. (4.11) and (4.12) to estimate the minimum design lateral forces for the iso- lation systems and structural system at or below the isolation interface (Vs) and structural elements above the isolation interface (V,), respectively. Also notice the limitations on V, as described in Section 4.5 of this chapter (see UBC-97, Sec. 1658.4.3). If the values of either Vy or V, are larger than what is acceptable for your project, you need a softer system. Go back t0 step 10 and start with a larger estimate of the vibration period. Otherwise proceed to the next step. Step 14: Perform a Preliminary Design of the Superstructure Elements. With V, estimated in step 13, calculate static lateral forces at each level of the building using Eq. (4.13). Use these lateral forces for preliminary stress siz~ ing of superstructure elements. For drift design use the maximum interstory drift ratio above the isolation system of 0.010/R; for the static force proce- dure (UBC-97, Sec. 1658.6). For response spectrum analysis and time history analysis the maximum interstory drift indices are 0.015/Ry and 0.020/R;, respectively (UBC-97, Sec. 1659.8). If the period of the fixed-base super- structure as designed is significantly different from that assumed in calcu- lating the limitations on V, in step 13, go to step 13 and verify the adequacy of V; as assumed. Otherwise, proceed to the next step. ‘Step 15: Perform a Preliminary Design of Isolator Units and their Distribution. ‘Using the preliminary displacement, stiffness, force, and damping properties established in the previous steps and design procedures presented in Chap- ter 3, design the isolator units to resist the gravity load, lateral load, and displacement requirements. Also establish a preliminary force-displacement hysteretic diagram for each type of isolator unit used. These diagrams may be based on the test results obtained from tests conducted oa bearings of similar type and size or on properties provided by isolator manufacturers. 4.1. STEP-BY-STEP PROCEDURE FOR URC-27 COMPLIANT DESIGN 87 ‘The hysteretic properties assumed at this stage are subject to verification and possible revision during the final design stage when the results of the prototype and production tests become available. 4.11.2. Final Design Steps ‘Step 16: Construct Mathematical Model(s) of the Isolated Structure. Con- struct mathematical (je., computer) model(s) of the fixed-base and isolated structure subject to requirements of UBC-97, Sec. 1659.5. Incorporate the force-displacement characteristics of the isolation bearings obtained from ‘step 15 in your models. There are basically two methods for construction of, ‘such models: ‘a. A model of base isolation system with stick model representation of super- structure. This approach is useful when multiple execution of models sub- jected to various time histories or minimum eccentricities are required. "Typically this may be achieved by constructing a 3D-BASIS model of the base isolation system incorporating the mass properties, eigenvalues, and eigenvectors of the fixed-base superstructure. The limitation of this ‘method is the inherent assumption of rigid diaphragm floors that must be validated and lack of net-uplift modeling capabilities. b. Am integrated model of the isolation system and superstructure con- structed with ETABS, SAP-2000, or similar computer codes capable of ‘modeling nonlinear characteristics of the isolators, if necessary. The dis- advantage of this approach is the sheer volume of data to be processed for numerous computer runs that are associated with any realistic seismic isolation project. ‘The recommended approach is to use the first method during the design iter- ations and the second approach during the final stages to confirm the accuracy Of the results obtained and to design for elements inadequately represented by the approximations inherent in the first method. ‘Step 17: Select an Appropriate Lateral Response Procedure. Use the informa- tion presented in Sections 4.4 and 4.5 or UBC-97 criteria selection provisions (UBC-97, Sec. 1657) to select an appropriate design procedure (i.e, static Iateral response procedure, response spectrum analysis, or time history anal- ysis). Regardless of the procedure required by the Code, we recommend that ‘he static lateral and response spectrum procedures be performed before any time history analysis is attempted. Idemify seismic demand definitions to be utilized (Le,, lateral forces, design spectra, time history pairs). Step 18: Finalize the Target Values of Design Displacements and Isolated Peri- ‘ods. Use the computer model(s) constructed in step 16 and seismic demands established in step 17 in a series of analysis and design iterations to arrive at a design of the isolation system and a superstructure that satisfies the basic project displacement and force response objectives. Finalize the values of83 CODE PROVISIONS FOR SEISKEC ISOLATION design displacement Dp and maximum displacement Dyy for the project. Notice that these values are not the same as the minimum value calculated in step 12 of preliminary design. As selected here, Dy should be larger than. Dp, and the Dp assigned here should be larger than the value calculated in step 12 of the preliminary design. Establish the isolated period at design dis- placement and maximum displacement levels (T'p and T'y). If possible, take: advantage of reduced displacements and base shears permitted by the code: for dynamic lateral response procedures introduced in Eqs. (4.14) and (4.15) (UBC-97, Sec. 1659.2). Step 19: Finalize the Target Values of Effective Stiffness. Except for the very unlikely case of an isolation system with markedly different positive and neg~ ative response characteristics, use the computer model and results obtained: in steps 16-18 to establish the target values of effective stiffness terms as follows: DBE base shear Ko,nax= Ko.nia De MCE base shear Kegoue = Kunin Da ‘Step 20: Verify the Effective Period Suggested by the Mathematical Model. Ver- ify the effective periods Tp and Ty as determined by the mathematical model ‘against those calculated by minimum values represented by Eqs. (4.2) end 3). ‘Step 21: Verify the Damping Level Suggested by the Mathematical Model, Use ‘Eqs. (4.4) and (4.5) to verify the damping level assumed in or reported by the mathematical model and, if necessary, recalculate the damping coefficients Bp and By (Fig. 4.10 or UBC-97, Table A-16-C), ‘Step 22: Verify Design Displacements and Forces against Code Minimum Val ues. Compare the design displacements reported by the mathematical model against the corresponding UBC-97 code minimum values as represented by Eqs. (4.2), (4.3), (4.9), and (4.10). If possible, take advantage of reductions permitted by Eqs. (4.14) and (4.15). If calculated design displacement values are less than code minimums, then adjust all response parameters (including ‘member forces and deformations) proportionally to code suggested values. Also verify reported base shears against code minimum values represenced by Eqs, (4.11) and (4.12), Step 23: Verification of Performance as Suggested by the Prototype Bearing Test Results. Upon the availability of prototype bearing test results, revise the mathematical model constructed in step 16 to reflect the lower bound and "upper bound bearing properties suggested by the prototype test results. Recal- culate effective stiffness values Kp,maxs Kv,miny Ky,max and Kymin Using 411 STEP-BY-STEP PROCEDURE FOR USC-97 COMPLIANT DESIGN BB Bgs. (4.16)~(4.19) (UBC-97, Sec. 1665.5.1). Repeat steps 20-22 using the revised mathematical model(s). Step 24: Verification of Performance as Suggested by the Production Bearing “Test Results. Upon the availability of production bearing test results, revise the mathematical model constructed in step 16 to reflect the lower bound and upper bound bearing properties suggested by the production test resulls and actual distribution of individual isolators. Recalculate effective stiffness values K,mes: Ko,mins Kimasy and Kunin using Eqs. (.16)-(4.19) (UBC- 97, Sec. 1665.5.1). Repeat steps 20-22 using the revised mathematical mo- dels): — DF bsmax— FB.mee Km = Seo be (4.16) DF ima ~Fovmin Koma = 417) Kua,max = (4.18) DY Finn — Fitna Ku, in = Din (4.19) Step 25: Verification of Performance as Suggested by the Production Bearing Test Results. Upon the availability of production bearing test results, revise the mathematical model constructed in step 16 to reflect the lower bound and upper bound bearing properties suggested by the preduction test results and actual distribution of individual isolators. Recalculate effective stiffness values Kp, mass Ko,mins Ki,maxs 810 Kunin Using Eqs. (.16)-(4.19) (UBC- 97, Sec. 1665.5.1). Repeat steps 20-22 using the revised mathematical mo- deics). 4.11.3 Design Example Consider a small building with a plan dimension of 40 x 20 m (131 x 65.5 £0). The total weight of the structure is estimated at 1600 tons (3550 kips). The lateral-load resisting system consists of reinforced-concrete shear walls, and the building is regular in both the plan and the elevation. The actual distance between the center of mass and the center of rigidity of each floor is 1.0 m 9.4 in).90 CODE PROVISIONS FOR SEISMIC ISSLATION ‘The project site is located as seismic zone 4 with an Sg soil profile and it is located about 1.5 km (0.93 miles) away from an active Fault capable of pro- ducing 7.9 magnitude events with an average seismic slip rate, SR, larger than 5 mm (0.2 in,). The fixed-base period of the building is 0.70 sec. The isola- tion system should provide effective isolated periods of Tp = 2.4 and Ty = 2.7 sec, respectively, and a damping of 15% critical. A margin of +10% vari- ation in stiffness from the mean stiffness values of the isolators is considered acceptable. Use the step-by-step procedure detailed in Section (4.11.1) to determine the minimum design displacements, minimum lateral forces, and maximum permit- ted interstory drift ratios according to the UBC-97 requirements, SOLUTION Step 1. From Fig. 4.1 (UBC-97, Table 16-1), Z = 0.40. ‘Step 2. Site soil profile category is given as Se. Step 3. From Section 4.4.3 (UBC-97, Table 16-U) for M > 7.0 and SR > 5 mun (0.2 in.), the seismic source type is A. Step 4. Step 4: From Fig. 4.3 (UBC-$7, Tables 16-5) for seismic source type ‘A and closest distance of less than 2'km (1.25 miles), Ng = 1.5. Similarly, from Fig. 4.4 (UBC-97, Table 16-T), Ny = 2.0. Step 5. ZN, = (0.4)2.0) = 0.80. From Fig. 4.5 (UBC-97, Table A-16-D) for ZN, 2 0:50, My = 1.20. Step 6. For Z = 0.4 and soil type Se from Fig. 4.6 (UBC-97, Table 16-R), Cy = Cy = 0.96N, = 0.96(2.0) = 1.92. Similarly, from Fig, 4.7 (UBC-97, Table 16-Q), and 16-Q), Ca = Can = 0.36N, = 0.36(1.5) = 0.54. Step 7. MyZNy = (1.20)(0.40)(2.0) = 0.96 > 0.40 and MyZNq = (1.20)(0.40)(1.5) = 0.72 > 0.40, From Fig. 4.8 (UBC-97, Table A-16-G), Cy = 24MyZNy 2.4(0.96) = 2.30. Similarly, from Fig. 4.9 (UBC-97, Table A-16-F), Caw = 0.9MyZN, = 0.9(0.72) = 0.65. Step 8. For reinforced-concrete shear wall system from Table 4.1 (UBC-97, Table A-16-E), Ry = 2.0. Step 9. For preliminary design purposes, 15% damping is assumed. Therefore, from Fig. 4.10 (UBC-97, Table A-16-C), Bp = By = 1.35. Step 10. Tp and Ty are given as 2.4 and 2.7 sec, respectively, ‘Step 11. From Eqs. (4.7) and (4.8) SOLUTION 91 Ww Tp = 28 Rovnink 5g, [ 280001000) . i 2.40 = 2 9 egy > Kovain = 1117 tons/m (63 kips/in) . 160001000) _, . a 2.70 = 2 4 [ pO = Kivi = 882 tons/m (0 kips/in) Since we assumed a +10% variation about the mean stiffness values, . aun _ ‘i Koyoux = (1.10) G2 = 1365 tons/m (77 kips/in.) . (882) Kizsnm= (1-10) S22 = 1078 tons/m (61 kips/in.) Step 12. From Eqs. (4.2) and (4.3) Dp = G/EDCIDT _ .81/4x7K1,92)(24) a aa 85 m (3.5 in.) _ B/4x)C ym Te _ (981/447 V2.3002.7) _ Dy = GAEDE Tn, CSV IVE «1.14 m 45.010) ‘Now, let us calculate total displacements including torsion. The additional acci- dental eccentricity required by the code is 5% of the plan dimension perpen- dicular to loading. Therefore, (0.05)(40.0) + 1.0 = 3.0 m (118 in.) In Eqs. (4.9) and (4.10), the value of the multiplier in both formulas is 12 4, 126.0) Ley peggy = 1+.) GEO = Lowe Hence Dro = 1.018(0.85) = 0.87 m (34 in.) Dry = 1.018(1.14) = 1.16 m (46 in.) Now Jet us examine the minimum design displacements permitted for ‘dynamic analysis (Eqs. (4.14) and (4.15)]:92 CODE PROVISIONS FOR SEISMIC ISOLATION Do oas Mp = SS = = 0.82 m (32.1 in V1I+(T/ToP V1 + (0.70/2.40P c ? Du Lis = 1.10 m (43: Ph SG TTae ~ VTSOTORTOF Step 13. The minimum base shear strength at or below the isolation interface from Eq. (4.11) is ‘p,maxV = 1365(0.85) = 1160 tons (2580 kips) = 73% of total weight ve and for design of the system and elements above the isolation interface from Eq. (4.12), Vg = Keumeo. - E5089) «580 tons (1290 kips) = 36% of total weight COMMENTS The large minimum displacements imposed by the UBC-97 requirements make the designing and implementation of a seismic isolation for this building extremely difficult. Fortunately, UBC-97 near-field factors, which are the pri- rary reasons for these conservative estimates of displacements, are expected to bbe phased out in the upcoming edition of the code. The year 2000 code, which isto be called the International Building Code (IBC-2000). is based primarily on the NEHRP-97 guidelines. These guidelines use hazard maps developed by US. Geological Survey (USGS) for the continental United States. These maps do not impose a near-fault factor on the top of the hazard estimate for the site as the near-field effect is already considered in establishment of seismic hazard. TBC-2000 is expected to lead to much more reasonable, lower bound values for esign displacements This design example highlights the UBC-97 inherent bias in the performance requirements for fixed-based versus isolated structures. While fixed-base build- ings are routinely designed to a life safety objective, isolated buildings are expected to perform with uninterrupted operation during and immediate oceu- pancy after the design earthquake. The code does not allow seismically isolated structures the same limited life safety objectives accorded to fixed-base build- ings. Given the fact that seismic-isolated structures, even when they fail, pro- vide a much more enhanced protection of life and limb for their occupants com- pared to their fixed-base counterparts, restriction of isolated designs to imme- diate occupancy objectives is not reasonable and must be changed. CHAPTER 5: MECHANICAL CHARACTERISTICS AND MODELING OF ISOLATORS 5.1 INTRODUCTION ‘The design process for an isolation system will generally begin with a pre- liminary design using parameters from a previous project or from data from ‘a manufacturer to estimate the possible maximum displacement of the system and maximum values of various controlling quantities (such as shear strain) and also for estimating the structural base shear, stability of the isolators, and pos- sibility of uplift. After this preliminary design process is completed, examples Of the final design of the isolators will be ordered and subjected to the code- ‘mandated prototype test program. Depending on the results of the prototype tests, the preliminary design may or may not need to be modified. In order to minimize the number of iterations in the design, it is essential to have accurate data and good design procedures in the preliminary design phase. This chapter will provide the isolation designer with the information needed for making the preliminary design, 5.2. MECHANICAL CHARACTERISTICS OF ELASTOMERIC BEARINGS ‘The mechanical characteristics of multilayered elastomeric bearings have been studied for many decades, and while exact analyses using nonlinear techniques are still quite difficult, simple predictions based on elastic theory have been developed by many researchers and verified by laboratory testing and more recently by finite element analysis. The most important mechanical property of94 MECHANICAL CHARACTERISTICS AND MODELING OF ISOLATORS. the isolation bearing is, of course, its horizontal stiffness; this is given by GA Ku= ‘where G is the shear modulus of the elastomer, A is the full cross-sectional area (which may differ from the area of the reinforcing shims), and 1, is the total thickness of the rubber. The maximum horizontal displacement D is related to the maximum shear strain by ey ‘The vertical stiffness Ky and the bending stiffness—expressed as ET by analogy with beam theory—are also given by a simple linear elastic theory and are needed for designing a bearing. ‘The vertical frequency of an isolated structure, often an important design criterion, is controlled by the vertical stiffness of the bearings that comprise the system. In order to predict this vertical frequency, the designer need only compute the vertical stiffness of the bearings under a specified dead load, and for this a linear analysis is adequate. The initial response of a bearing under vertical load is very nonlinear and depends on several factors. Normally, bear- ings have a substantial run-in before the full vertical stiffness is developed. This run-in, which is strongly influenced by the alignment of the reinforcing shims and other aspects of the workmanship in the molding process, cannot be pre- dicted by analysis but is generally of litte importance in predicting the vertical response of a bearing. ‘Another important bearing property that must be analyzed for design is the buckling behavior of the isolator. In order to conduct this analysis, the response of tae compressed bearing to bending moment is necessary. Referred to as the “bending stiffness,” this can be ascertained by an extension of the same analysis that is done to determine the vertical stiffness. ‘The vertical stiffness of a rubber bearing is given by the formula _ BA Kv=- where A is the cross-sectional area of the bearing (in this case it usually is taken as the area of the shim plates), ¢, is the total thickness of rubber in the bearing, and £; is the instantaneous compression modulus of the rubber-steel composite under the specified level of vertical load. The value of E- for a single rubber layer is controlled by the shape factor S, defined as loaded area force-free area 52 MECHANICAL CHARACTERISTICS OF ELASTOMERIC BEARINGS 9B, which is a dimensionless measure of the aspect ratio of the single layer of the elastomer. For example, in an infinite strip of width 2b and with a single-layer thickness 1, b see : 7 on for a circular pad of diameter ® or radius R and thickness £, * ad Zz 2) and for a square pad of side dimension a and thickness 1, set 6.3) For a single pad in the form of a complete circle, the compression modulus E. is given by E, = 6GS* and for a square pad, the result is In some cases bearings are designed with unfilled central holes. The result for a bearing with an inside radius a and outside radius is Eq = OXGS? where yo Ata? =e? = a2V/Un b/a)h Ora? If a/b, then d+ 1; hence, E, = 6GS%, which is the result for the full circular pad. If a/b 1, by writing a/b = 1'~ ¢ and letting e-0, we find that \ + # and E,—+ 4G, which is the result for the infinite strip. It is inter- esting to evaluate how rapidly the result for \ approaches 2, To illustrate this point, the solution for ) is plotted versus the ratio a/b and from 0 < a/b <1 in Fig. 5.1. Clearly for the case when a/b > 0.10, the value of is almost t¥0-96 MECHANICAL CHARACTERISTICS AND MODELING OF ISOLATORS 0 01 02 08 04 05 08 07 08 09 4 ab Fig. 5.1 Reduction of compression modulus E- for an annular pad, thirds, indicating thatthe presence of even a small hole has a larg effect on E; therefore, in most cases for bearings With central holes, the value of E. should be taken as 4GS? rather than 6GS*, Under direct compression the constraint provided by the bonded steel shims produces a shear stain in the rubber, which is denoted by +. If the nominal compression strain e. is given by ‘where A is the vertical displacement, then Ye= 6Se. ‘This is the maximum shear strain developed at the edges of the pad and often used in design. ‘The maximum shear strain due to compression is not the only shear strain quantity of interest to the designer. It is also useful to estimate the average strain in the following manner: Because rubber is somewhat strain sensitive, Gis often modified according to the strain level, particularly in highly filled rubbers. In compression the shear strain varies widely over the volume of the pad; therefore, the appropriate value of the modulus used to estimate the average strain is based on a calculation of the elastic stored energy in the pad From this calculation the average shear strain ‘Yave is given by 52 MECHANICAL CHARACTERISTICS OF ELASTOMERIC BEARINGS 87. Yee = V6Se6 Although a certain degree of trial and error is needed, computations of this kind allow the designer to estimate the appropriate value of G, which can then be used to estimate the vertical stiffness. First we must assume a value of G in ‘order to calculate ¢, and from that calculate “jaye; we then modify G and iterate as necessary. Because the modulus is not very sensitive to strain above about 20%, few iterations are needed. The bending stiffness of a pad is computed using a similar approach with the same type of displacement assumptions. The pad is assumed loaded by a pure moment M, ard the deformation is assumed to be a rotation of the top and. bottom bonded plates, as shown in Fig. 5.2. The relative angle between the top and bottom plates is denoted by a, and the radius of curvature p generated by the deformation is related to a by Fig. 5.2 Rubber pad between rigid constraint layers in pure bending [30]98 MECHANICAL CHARACTERISTICS AND MODELING OF ISOLATORS M= Dea = For a circular pad of radius R, the result is 3aG_ Rt Mo SB 1D and for a beam of circular section aR i [But in this case, taking the E in EI as E. = 6GS?, we have the result +Rt/12 for I, which is one-third the beam moment of inertia. The difference is caused by the fact that the pressure distribution varies as a cubic parabola across the pad, whereas in a beam, the bending stress distribution is linea. In the case of 4 square pad, the effective EY is very close to one-third the beam EI, being (EDe = £03291) For the circular pad with a central hole, we have (ede = 2031 B40 ‘The shear stresses induced by the bending are given by yo = 6S¢s where ¢s = R(a/t) is the edge compression strain produced by bending. The average shear strain (in the sense of overall shear strain energy) is given by yt Vista =visk & 2s 4a va 7 Sep ‘When the shape factor of the pad becomes large, the effect of compressibility of the rubber begins to be important. Compressibility can be approximately 52 MECHANICAL CHARACTERISTICS OF ELASTOMERIC BEARINGS 98 incorporated in the previous formulas through the ad hoc modification [73] a z ‘where 1/E; is the compression modulus assuming incompressibility behavior and K is the bulk modulus of the material. More exact fanulas have been developed for this effect, but the bulk modulus is such a difficult quantity to ‘measure that use of the simple ad hoc formula is good enough for design pur- poses. The value of K varies quite widely in reference material, ranging from. 1 low of 1000 MPa (145,000 psi) [74] to 2500 MPa (363,000 psi) (78]. The value most commonly used, and which appears to fit laboratory test results on bearings, is 2000 MPa (290,000 psi). When rewritten for Ee, the ad hoc equation becomes EK EO EE and for the circular pad, we have EZ = 6G; thus 6os?K 6GS+K aft = 868° [ 136657 | When 5 is small, we have B.=665:(1- S¢") K so that for, say, better than 10% accuracy, $ is such that xK 657 < with K = 2000 MPa (290,000 psi) and G = 0.7 MPa (101.5 psi), S<7. ‘When S$ is large and 6GS?/K is much greater than unity, x neer[i- gh]102 MECHANICAL CHARACTERISTICS AND MODELING OF ISOLATORS where R is the radius of curvature of the dish. The first term is the restoring force due to rise of the mass, providing a horizontal stiffness Ku which produces an isolated structure period T given by T= 2nV/Rig ‘which is independent of the carried mass. The second term is the friction force between the slider and the concave surface. The coefficient of friction p depends ‘on pressure p and sliding velocity D. The coefficient decreases with increas- ing pressure and becomes independent of velocity for speeds above 51 mm/s (2 in,/sec) at pressures greater than about 14 MPa (20 ksi). A typical hysteresis loop for a FPS system for a shake table experimental program is shown in Fig. 3 [5]. The very linear nature of the restoring force, the high stiffness before iding occurs, and the energy dissipation due to the sliding friction are clear from this figure. The equivalent (peak-to-peak) stifiness is given by Www Ker tp ‘The damping produced by friction at the sliding surfaces can be estimated os 0.0 Base Shear / Weight -0.3 25 0.0 25 Isolation System Displacement (1N.) Fig. 5.3 Hysteresis loop from shake table tests for FPS. 54 MECHANICAL CHARACTERISTICS OF FRICTION PENDULUM SYSTEM 103 by the code formula oq = 2#08 of hysteresis loop a “aKa? ‘The area of the hysteresis loop is 4uWD; thus 4yWa 8° SetW/R)D + uWid which ranges from 2/x for small D to 2u/(rR/D) as D increases. For example, if D = 254 mm (10 in.) and R = 1 m (39.4 in.) with 4 = 0.06, we have 8 12%. To understand the geometry of the FPS, it is useful to invert the equation above relating T to R. For fixed period , we have gt? 2 = £7. = 103 R= Fy > tor so that We need a radius of around 1 m (39.4 in.) for a 2-sec period. If the horizontal displacement D of the system is 254 mm (10 in.), then the upward displacement by given by woa[t-or nt) will be around 32 mm (1.26 in.). Thus, a horizontal motion of #254 mm (£10 in.) at around 2-sec period also generates a vertical motion with a range of 32 mm (1.26 in.) at a period of 1 sec. The approximate form for the vertical displacement indicates that the vertical displacement is roughly quadratic in the horizontal displacement Another aspect of the FPS is that if the displacement is less than a certain factor of the radius, the restoring force can be less than the frictional force and the system will not be recentering. This factor is obtained by equating the two terms in the force equation. Thus, the system will not recenter if D/R y. This ccan be a problem in long-period systems; for example, if T = 5 sec, R = 6.35 1m (20.8 ft), with 4 = 0.06, the system wall not recenter if D < 381 mm (15 in)104 MECHANICAL CHARACTERISTICS AND MODELING OF ISOLATORS ‘The extremely simple modeling of the FPS makes it very attractive; however, its very simplicity is the main disadvantage of the system. It is essentially a one-parameter system, and that parameter is controlled by the radius of the ‘concave surface. In orde: for the various articulated surfaces to slide together, all surfaces have to be spherical. Thus, the response is linear over the entire ange of displacement. 5.5 MODELING OF ISOLATION BEARINGS BY BILINEAR MODELING In practice all isolation bearings are modeled by a bilinear model based on the three parameters Ky, Kz, and Q, as shown in Fig. 5.4, The elastic stiffness K, is either estimated from available hysteresis loops from elastomeric bearing tests or as a multiple of Kz for lead-plug bearings and friction pendulum bear- ings. The characteristic strength Q is estimated from the hysteresis loops for the elastomeric bearings. For lead-plug bearings Q is given by the yield stress, in the lead and the area of the lead, while in the friction pendulum bearings it is given by the friction coefficient of the sliding surface and the load carried by the bearing. The postyield stiffness can be accurately estimated or predicted for all three types of bearings. ‘The effective stiffness, defined as the secant slope of the peak-to-peak values in a hysteresis loop, is given by 2 Ker= Kit D=d, Foree Fig. 5.4 Parameters of basic hysteresis loop for example problem. 155 MODELING OF ISOLATION BEARINGS BY BILINEAR MODELING 105 where D, is the yield displacement. In terms of the primary parameters, @ OO Re and the area of the hysteresis loop (the energy dissipated per cycle), Wo, is given as Wo = 4Q(D- D,) ‘The effective damping Serr is defined by bay = 2OD- Dy eR gD? This can be expressed in nondimensional quantities by defining a nondimen- sional displacement a2 "3 and a nondimensional characteristic strength Q ED, whereby the effective damping becomes 2a _y-1 * Otay yet on For fixed a, 8 = 0 at y= 1 and tends to zero as y+ =», The maximum value of B given by d8/dy = 0 occurs at with 2a 1 Wt C8 Bons Fara) Now a= Q/KsD,; this from Fq. (5.6), we have106 MECHANICAL CHARACTERISTICS AND MODELING OF ISOLATORS Ki-k % 69) This leads to the curious result that the maximum value of the effective damping depends only on the ratio of XK; to Kz. The second slope, Kz, is easy to determine for any type of isolation system, but the first slope, Ky, is usually done by eye and can vary over a wide range. Since the characteristic strength Q can also be accurately determined, the value of K, has no influence on the effective stiffness but obviously has a strong influence on the damping and, in particular, on the maximum value, which can be an important promotional value for a proprietary system. To illustrate the effect of the selection of K; on the damping, consider a system with the same Q and Kz values (thus the same effective period at all values of D and the same hysteresis loop) but modeled by different values of Ki Fig. 5.4). ‘We take Q = 44.5 KN (10 kips) and K2 150 KN/m (2 kips/in, Kj = 51K; Comesponding to a friction pendulum system K} = 21K Corresponding to a lead-plug bearing K} = 6Kz Corresponding to a high-damping rubber bearing K{=3K2 Another example of high-damping rubber bearing ‘The values of Dy, a for these are as Dy = Spry = 2-5 mm (0.10 in) 50.0 pe 45 635 mm Di= MS 635mm@2siny a2=200 p. 45 in op= 45 rs4mmilooiny a =50 f= 4S Lsmmasoiny — ahn20 ‘The peak values of f and the displacement at which it occurs are as follows: Bax Bax Bax = 480 at D = 20.8 mm (0.82 in.) 410 at D = 35.6 mm (1.40 in.) 268 at D= 87.6 mm (3.45 in) 1.170 at D= 173.5 mm (6.83 in) 56 IMPLICATIONS OF BILNEAR MODELING 107 fal’ oa TK oa: 02 Kniky oa os 04 05 08 07 08 09 1 Fig. 55 Effective damping as a function for displacement for different choices of Ki. ‘The variation of f with D for the three cases is shown in Fig. 5.5, illustrating how the damping at small displacements can be very high if K; is selected to be large. It also shows that as D becomes large, all B-versus-D curves tend to ‘become the same. The point is that the same hysteresis curve when modeled by a bilinear model can have a very different pattem of damping values, depending. only on how the initial stiffness is selected. To put the values used in Fig. 5.5 in perspective, suppose that the second slope, Kz = 350 kN/m (2 kips/in.), corresponds to an isolator in a system with 1 period of 3 sec. The carried weight on an isolator is then 2x9 29 396 = 786 KN (176 kips On Cre Kips) 44.5 KN (10 kips) then means o w= Z=56% ‘The characteristic strength which represents the friction factor for a friction pendulum system or the strength ratio for a lead-plug bearing, 5.6 IMPLICATIONS OF BILINEAR MODELING ‘The problem facing the isolation designer is that having obtained the value of Cv for the site and assumed possible values for period and damping, there108 MECHANICAL CHARACTERISTICS AND MODELING OF ISOLATORS is still the constraint of maximum acceptable value for the MCE displacement Dy. Suppose then that T,, 6, and D are specified and itis necessary to calculate the parameters of the model, namely, K2 and Q, to provide these values. With the period specified, the effective stiffness Kem is given by (3) then Wo = 2nKenD*B can be calculated, leading to W)\(2n\?_, saxo) ~20(™) (28)' os At this point we neglect D, and estimate Q from Wo O= Gp y KendB leading to an initial value of Q/W. Itis then possible to estimate Kz = Kew-Q/D and D,, assuming, for example, for a lead-plug bearing that Ky = 10K? or for a friction pendulum system that Ky = 100K2. With the approximate value of Dy, the estimate of Q and Kz can be improved, and with these values the design of the system can proceed. For example, suppose the code formula for Dy leads to a displacement of 76 cm (30 in.) for a period of 2.5 sec and 5% damping. To reduce this to a feasible level of displacement, say, 50 cm (20 in.), requires that B= 1.5 and 6 = 0.30, The value of Ker for T= 2.5 sec is Keg =0.644W KN/m (Kee = 0.01636W kips/in.) and the value of Wy is given as Wo = (0:3)2n(0.644)(0.25)W = (Wo = (0.3)2x(0.01636)(400)W 308W kN/m. 12.3W kips/in.] from which we get (neglecting D, from the first estimate) 158 IMPLICATIONS OF BILINEAR MODELNG 108 223 (aan; 91%) Ka (064s S84) w (“eee =0366w = 0.00865W and 0.154 0.34 sess 05 (= swesy = 198) A recalculation of Q, including this estimate of Dy, is Q/W = second estimate of K3 as 171, with a Kz =0.306W KN/m (Kz = 0.00768W kips/in.) ‘Thus we need a lead core that produces a characteristic force of 17% of W to get this level of damping. Suppose now we postulate a service level earth- quake (SLE), for example, that earthquake that might have a 50% chance of being exceeded in 50 years, and assume that for this level of input the seis- mic coefficient Cy is half that at the DBE, that is, Cys = $C)D, then Cys = (1/2.4)Cyyy. Therefore, we can anticipate a displacement Ds that is not more than (1/2.4)Dy. Assuming that in this case D is approximately 125 mm (S in), at this level of displacement we have om oa Keg = 0.306W + O55 W (“me z ") = 1674 N/m = 0.04201" kips/in and Wo = 40 ~ D,) Wo = 4Q(D - D,) 4(0.171)(0-125 ~ 0.51) 4(0.171X5 1 sow) 10506 kN/mm 2.066 kips/in. iving110 MECHANICAL CHARACTERISTICS AND MODELING OF ISOLATORS. p-031 and B=154020% f= 152 ‘The period T = 1.56 see. Assuming that Cys = (1/2.4)Cyy and that Cy leads to 76 cm (30 in.) at T = 2.5 sec and 6 5%, we have 30.156 1 Ps 7a 75 15D = 130 mm (5.13 in.) ‘The unreduced base shear at this displacement is KesD: .674W x 13 KenDs 0.22W kN 0.0420W x 5.13 0.22W kips which is only just above the characteristic strength Q = 0.171W: therefore, with this design, very little isolation effect is possible. For a friction pendulum device the results are similar. Since D, for the frie~ tion pendulum system is generally negligible, the required Q is only 0.154W, corresponding to a friction factor of 15.4%, The required K2 in this case is, ousa ousa kyo. 2154 (" ores 2254 = 03007 eN/m 008659 tps If we assume that for a SLE we have a displacement Ds of around 125 mm. (S in), we find 0.125 z~oaeiws 22580 (‘Scomeon ss = 1.573WkN/m 0.0395 kips/in. Wo = 4(0.154W)0.125 Wo = 4(0.154W)S 0.077W kN/m_ = 3.08W kips/in, giving 56 IMPLICATIONS OF BIINEAR MODELNG 111 B=0496 T=161e0 R=20 reading to 1 161 30 Sa F359 = 10 mm Gin) land the unreduced base shear is 4 x Kyr = 0.189W, which exceeds the slip force, 0.154W; therefore, for this level of input, the system will not move, ‘These results mean that the code formulas and the modeling procedure lead to isolation systems that will not work at what might be considered occasional rather than rare levels of earthquake input. In case the impression is that a displacement of 76 cm (30 in.) at 2.5-sec period and 5% damping represents the extreme case of the code formula, it should be noted that buildings in downtown Berkeley and all of the University of California at Berkeley are within 2 km (1.24 miles) of an active fault that could produce a moment magnitude greater than or equal to 7.0 and has a slip rate exceeding 5 mm (0.2 in.) per year, thus qualifying for seismic source type A. With these site parameters and Sp soil type, the MCE displacement at 2.5 sec and 5% damping can exceed 1m (40 in), 5.6.1. Energy Dissipation in High-Damping Natural Rubber Bearings The mechanical characteristics determined from cyclic testing of seismic iso- lation devices at a constant frequency are typically expressed as a function of displacement D using two parameters: the effective stiffness Key and the equivalent viscous damping eq. (Note that in the following discussion, global bearing properties such as displacement, force, and viscous damping coefficient axe used. Altematively, the model can be formulated in terms of rubber mate- rial properties such as strain, stress, and material damping coefficient, thereby allowing analyses to be performed based on only material properties and not requiring test results from actual bearings.) High-damping natural rubber isolators are characterized by stiff behavior (high Keg) at shear strains up to about 25%, a substantial decrease in Ker up (0 a strain of 150-200%, and then an increase in Keir above this strain due to strain crystallization in the material. (The strains at which these stiffness changes are observed depend on the elastomer compound.) Examples of the variation of Ket and Bq 28 a function of displacement are shown in Fig. 5.6 for the bearings evaluated in this study. For high-damping natural rubber bearings, Gag typically decreases with increasing strain but Wp does not. Several recent tests of elastomeric isola- tors [64, 68] have shown that over a wide range of strains Wo is proportional to the shear strain raised to an exponent of approximately 1.5. In practice, the actual value vaties from about 1.45 to 1.75, depending on the compound112 MECHANICAL CHARACTERISTICS AND MODELING OF ISOLATORS 18 2 10 Eajvalort Viecous Damping (ercont) tt 3 co 70 Bo Stain (erent Eo. gs ial * 10 0 Displacement (mm) Fig. 5.6 Stiffness and damping of high-damping natural rubber isolators. under consideration.) Similar results, expressed in terme of displacement, can be seen for the bearings in this study in a log-log plot of displacement versus Wo (Fig. 5.7). In this case the exponent is about 1.61. Considered on a molee- ular level, the nature of the high-damping natural rubber material suggests that the internal energy dissipation mechanisms may be a combination of frictional and viscous components; however, no fundamental physical process that leads to a constant near 1.5 has been identified, Definition of Energy-Based Mode! The empirical observations outlined above can be used to model an elastomeric isolation system as a combination of a linear elastic spring element, a pure hysteretic element (Wp proportional 56 IMPLICATIONS OF BILNEAR MODELING 113. — Measured Data ~~ Lesst-Squares Fit oo os to v5 20 10 (Displacement) Fig. 5.7 Energy dissipation characteristics of high-damping rubber isolators to D!®), and a pure viscous element (Wp proportional to D*®). The goal is to determine the properties of each of these elements such that the Wo of the com bined system is proportional to approximately D' over a given strain range. The four parameters necessary for the basic form of the model are Ky (the clastic stiffness of the linear spring element), Kz [the elastic stiffness of the hysteretic (elastic-plastic) element), F, (the yield force of the hysteretic cle. ment, alternately D,, the yield displacement of the hysteretic element), and C (the viscous coefficient of the linear damping element). Once these parameters are determined, the model can be implemented directly in any nonlinear strue- tural analysis program that includes this basic set of elements (for example, SD-BASIS [96], DRAIN-2DX [6], or LPM [59)). ‘To derive the equations for the nonlinear element properties, the energy dis sipated by the model in one cycle is first expressed as, Wo, = 4F,(D ~ Dy) +xCuD? (6.10) which, neglecting the terms in D, for large bearing displacements under a design level earthquake, becomes (sl) where a and b are unknown constants Based on the observations in the previous section, the energy dissipated in a bearing during a cyclic test at a constant frequency can be approximated by the relationship14 MECHANICAL CHARACTERISTICS AND MODELING OF ISOLATORS Wo, = eD™ 6.12) where a 5 1.5 for filled elastomers and c is an experimentally determined constant, The coefficients a and b, which define the energy-based model, are then determined by minimizing the difference between the energy dissipated in the model (Eq. (5.11)] and the energy dissipated in the bearing (Eq. (5.12)] over some displacement limits Dy and Dz as follows: Dy (Wy ~ Wo)? dD ©.13) 6.14) ‘The best choices of a and b can be shown to be (2552) (mig -( ss) ( 5 G2)-(2) (et) (get (25?) (552!) -( 752! 6.16) If the model parameters F, and C are given by Fy=a/4 and C= b/tma) jin over which displacement range should the energy difference be minimized? If Dy and Dy are selected as zero and the maximum anticipated displacement, respec- tively, itis unlikely that good agreement will be obtained over the entire displace- ment range. Therefore, D; and Da should be chosen to bracket the expected maxi- ‘mum displacement by perhaps +20-30%, ensuring that the energy dissipated dur- ing the peak displacement cycles will be accurately represented by the model, Repeated analyses with some iteration may be required to converge on the appro priate displacement range, depending on the intensity of the input. 56 IMPLICATIONS OF BILNEAR MODELING 115: ‘The selection ofthe final model parameters, Ky and Ko, is best made using the shape of the measured hysteresis loop at the target displacement. This is shown schematically in Fig. 5.8. The logical choice for K2 is the tangent stiff- jess at 220 displacement; in the figure, Kz was determined by performing & least-squares ft on the data to 470% ofthe maximum displacement (soli lines) ‘Tne choice of Ky is best done by eye to match approximately the intial unload- ing sifiness after the peak excursion of the loop. Note that while K may remain approximately constant over a wide displacement range, the best choice for Ky (the unloading slope) will likely vary. 5.6.2 Adjustments to the Model to Account for High-Strain Stiffening ‘An additional refinement to the basic model is required ifthe bearing is expected. to displace into the stiffening portion of its force-displacement curve, and this is best achieved by incorporating 2 gap element that is activated ebove a cer- tain displacement. The most straightforward approach is to use a near elastic gap element that des not dissipate energy; this type of element is available in several widely used analysis programs, including DRAIN-2DX and SAP2000. From actual bearing tests, however, it is apparent that the stiffness does not change suddenly but increases rather smoothly as a function of strain until near failure. The stiffening model incorporated in the program 3D-BASIS-ME [125] captures this behavior relatively well by combining a hysteretic element with an elastic element having a stiffness that can be defined to increase linearly over a defined displacement range. This approach exhibits hystereic behavior that is closer to that recorded in real bearings, but it still does not guarantee that. the energy dissipation characteristics will be captured accurately. In particular, 6 21] Strain Leve: 100 percont ze eg ° - ‘slope =Ke to 20 ° 2 40 Displacement (mm) Fig. 58 Selection of model parameters—K} and Ka116 MECHANICAL CHARACTERISTICS AND MODELING OF ISOLATORS because the stiffening element is elastic, it may not reproduce the substantial dissipation that high-damping natural rubber bearings exhibit upon unloading from the stiffening portion of the force-displacement curve. ‘The gap/stiffening element proposed here alleviates this shortcoming by defining separate loading and unloading curves that enhance the energy dis- sipation in the stiffening portion of the loop. The element force in both curves is a power function of the displacement in the element, and the exponent in the unloading branch is greater than that in the loading branch to provide dissipa- tion. The general expressions for the element force in terms of the displacement Fg = {Ku 98n(D\(\D| ~ Dew)" for loading 6.18) 'S= 1 Ki sgnD\(|D| ~ Dy)" for unloading G9) where Fs is the force in the element, Dyap is the displacement where stiffening initiates, n and m are the hardening exponents, and K; and Ky are the stiff- ness coefficients (loading and unloading). Becanse Kz and Ky must be defined such that the loading and unloading branches intersect at the top of the loop, Ky is, therefore, dependent on the previously calculated maximum force and displacement. The exponents n and m are best determined by eye to achieve a close approximation to the measured loading and unloading curves. Finally, to ensure that the composite model (the combination of linear, vis- cous, hysteretic, and stiffening elements) dissipates the desired quantity of energy at the target displacement, we need to define the parameters for the viscous and hysteretic elements at the displacement at which the gap element is activated, then determine the required Wp at the target displacement using the logarithmic relationship, and finally calculate the energy dissipated by the ‘composite clement at the target displacement. ‘The energy dissipated by the gap element in a fully reversed hysteresis cycle ccan be shown to be (5.20) Om=tar| Note that this calculation assumes that the frequency of loading at the target displacement is the same as at the displacement where the gap element is acti- vated. If this is true, the energy dissipated by the viscous element will have een increased inadvertently. To correct this shortcoming, the coefficient of the viscous element is reduced so that the energy dissipated by the model equals that dissipated in the bearing (based on the observed logarithmic relationship). 56 IMPLICATIONS OF BILINEAR MODELING 117 6.6.3 Comparisons with Experimental Data The validity of the model and its sensitivity to frequency variations are evalu- ated by comparing its performance to recorded bearing hysteresis loops. While the goal of the formulation is to dissipate equal energies in the model and the tearing. the fit with the observed force-displacement (neglecting stiffening) depends in large parton the selection of parameters Ky and K2. As mentioned above, the selection of K2 is straightforward; in the examples shown here it is taken as the least-squares fit on the data to 70% of the maximum displacement (Gefore stiffening). The choice of K; is not as easy, however, because it defines the stiffness under two separate modes of behavior—the initial stiffness when the bearing is first loaded and the unloading stiffness after a displacement peak has been reached. To achieve the close fit in the fully reversed loop shown in Fig. 5.9, X; has been chosen to be much lower than the initial tangent stiffness to the unloading curve. If K; is too soft, however, the assumption that D, is negligible compared to Dnax that was made in the derivation of the model may rot be correct, and the energy dissipation of the hysteretic element will be less than required. In most cases the majority of the energy is dissipated by the vis- cous element, so this should not be a significant problem. But if reproducing the behavior of the bearing at small displacements is essential, a large initial stiffness will be required for the hysteretic element, with the trade-off that the unloading behavior near the peak displacements will not be matched closely ‘An additional consideration in the basic model relates to the frequency of loading, since the coeflicient of the viscous element is defined to be inversely related to the loading frequency [Fq. (5.17)}. In the examples provided here, the bearings were tested at a rate of 1.0 Hz; however, a typical design fre- — Recorded Data — Analytical Model +0 Strain Level: 100 percent 48 40 20 20 ey 0 Displacement (mm) Fig 5.9 Comparison of analytical model with recorded data.148 MECHANICAL CHARACTER'STICS AND MODELING OF ISOLATORS: quency might be 0.5 Hz. If the model derived from 1.0 Hz testing is subjected to cyclic loading at 0.5 Hz, the viscous element will dissipate substantially less energy than anticipated. Most high-damping natural rubber compounds do not exhibit significant rate dependence within the range of frequencies anticipated in seismic applications, so there will be relatively lite change in the energy dis- sipated by the real bearing, as illustrated in Fig. 5.10. It is important to note that the oversensitivity to frequency in the model will lead to conservative results in terms of displacement because the energy dissipated reduces as the loading fre- ‘quency reduced. For the low-frequency ground motions that are most critical to the behavior of isolated structures, the energy dissipated will be underpredicted, leading to conservative displacement estimates. This problem can be salved by modifying the frequency used to define the coefficient of the viscous element over a series of analyses until the computed response frequency in the cycle or cycles at the maximum displacement equals that assumed in the analysis. This assires the proper energy balance between the model and the bearing. When the model is extended to larger displacements, three parameters must be defined: the displacement at which the gap/stiffening element is activated, the exponent on loading, and the exponent on unloading. (Both exronents may vary somewhat with displacement, depending on the shape of the loop and the accuracy with which the force~displacement relation ship must be matched.) Figure 5.11 compares observed hysteresis data to 350% shear strain, with the response of the model defined using the pro- cedure outlined above. In this case stiffening begins at a displacement of 4 in. (225% shear strain), and the parameters of the viscous and hys- teretic elements are also defined at this displacement. The stiffening ele- ‘ment has an exponent upon loading of 1.5 and an exponent upon unload- ing of 1.7. The parameters Kz and Ky are defined based on the maxi- 21 Recorded data 0.5 He “Analytical Model defined at Force (kN) ° ‘Strain Love 100 percent 45 to 20 ° 20 % Displacement (nm) Fig. 5:10 Sensitivity of model to loading frequency. 56 IMPLICATIONS OF BILNEAR MODEUNG 118 — Recorded Data — Analytical Mode! Strain Level: 350 percent ee ae er ee ee ee) Diplacement (mm) Mig. 5.11 Large displacement behavior of the model ‘mum recorded force in the bearing and their respective exponents. To ensure appropriate enexgy dissipation in the analytical loop (including the contribution from the stiffening element), the coefficient of the viscous element is reduced by 34%, and the final analytical loop area is only 1.4% different than the mea- sured aca. The analytical relationship docs not precisely uack Wie measured force-displacement (particularly near-zero displacement where the measured loop is slightly pinched) but should be satisfactory for dynamic analyses.CHAPTER 6 BUCKLING AND STABILITY OF ELASTOMERIC ISOLATORS 61 INTRODUCTION ‘A multilayered elastomeric bearing can be susceptible to a buckling type of insta~ bility similar to that of an ordinary column but dominated by the low-shesr stiff- ness of a bearing. The previous analysis of the overall deformation of a single pad ccan be used in a buckling analysis that treats the bearing as a continuous compos- ite system. This analysis considers the bearing to be a beam, and the deformation is assumed to be such that plane sections normal to the undeformed central axis remain plane but not necessarily normal to the deformed axis. ‘The theory for the buckling of isolation bearings is an outgrowth of work by Haringx in 1947 on the mechanical characteristics of helical steel springs and rubber rods used for vibration mountings. This work was published in a series of reports, the third of which [49] covers the stability of solid rubber ro¢s. The Haringx theory was later applied by Gent [44] to the problem of the stability of ‘multilayered rubber compression springs, and it is this application that forms the basis for the theory given here. ‘The method parallels the linear elastic analysis of the Euler buckling of a column, To model the rubber isolator as a continuous beam, it is necessary to introduce certain modifications to the quantities defined in the previous section, Consider the bearing to be a column of length h with a cross-sectional area A and define the shear stiffness per unit length as Py = Gas, where Ay is an effective shear area given by 124122 BUCKLING AND STABILITY OF ELASTOMERIC ISOLATORS where h is the total height of the bearing (rubber plus steel) and i, is, as defined carlier, the total heigat of rubber. The increase in A is needed to account for the fact that steel does not deform in the composite system. The bending stiffness is similarly modified, so that (ENeq for a single pad of thickress ¢ becomes Els, where Els= Bey * In terms of these quantities, the overall horizontal stiffness Ky (which was GA/t,) becomes ‘The usual situation for a bearing in an isolation system is shown in Fig. 6.1. ‘The bearing is constrained against rotation at both ends and is free to move sideways at the top. The result for the critical buckling load Pe: is the solution of the equation P+ PPs~PsPe=0 1) from which the critical load Pui: is given by —P s+ PE + 4PsPe on Paw 2 If we now assume that Ps = GA and Peas 1 6GS71x? 2xtst 3° 8 ~oa( a ) then, for most types of bearings where $2 5, Pg >> Ps, the critical load can be approximated by Pes = (PsPe)'? (63) 61 INTRODUCTION 12% i Fig. 6.1 Boundary conditions for an isolation bearing under a vertical load P (the bearing buckles with no lateral-force constraint but is prevented from rotating at each end). Using this expression and recalling that rece y conan 64) we have AV (x? 1 aga h oa (ca #)"*( Leostar 2) 2xGASr where the radius of gyration is denoted by r = V/I/A = a/2V33 for a square bearing with side dimension a and /4 for a circular bearing with diameter # ‘The critical pressure Pare = Pe/A can be expressed in terms of $ and the124 BUCKLING AND STABLITY OF ELASTOMERIC ISOLATORS ‘quantity S2, referred to as the aspect ratio or the second shape factor, defined by * a B= ot or SF os =z SS for a circular bearing 22 G ® ss. ke veo” * In actual design the load carried by a bearing (say, W) will be less than the critical load, and neglecting the effect of the vertical load on the horizontal stiffness Ky of the bearing, that is given by Ky = GA/t,, which in turn is related to the horizontal frequency «through 5) 3. Kn ohe are ‘Thus, the safety factor SF against buckling, which is defined by SF = Pax/W, becomes VixSwpr e SF= 69) All other things being equal, the safety factor increases with shape factor S, frequency «7, or bearing size (either a or ®). ‘The bearing size will, of course, depend on the carried load. If the pressure p= W/Ais specified, then wn WE (4) for a circular bearing ©” “| pg (2) tore seve tein 6» 2V3 \P If the pressure is fixed, the safety factor will diminish as WY, leading to the ‘unexpected result that buckling can become a problem for bearings that are lightly loaded 1 INTRODUCTION 125. To get a feeling for the magnitude of the quantities involved, suppose that the safety factor must be at least 3, the shape factor S is 10, and the frequency is x radians per second (2.0-sec period), all of which are typical values. In this case, r must be at least 3x 9810 V2 x 10% x? = 67.0 mm (2.64 in.) If the bearing is circular, then # = 268 mm (10.6 in.). This minimum dimension is independent of the carried load or the pressure, but if the pressure is specified, say, for example, at 6.90 MPa (1000 psi), it translates to a minimum load of 390 KN (88 kips). For most buildings the bearings will be much larger than this minimum size end the carried loads will be in the hundreds of tons, so that buckling is not likely to be a problem for the design. ‘There have been cases, however, when it has been necessary to design iso- lators for lighter loads, for example, circuit breakers in electric power plant switch yards. These components weigh as little as a few tons and the bearing design should be steble under loads that range from 10 to pethaps 50 KN (2-10 ips). Bearings developed for such applications are referred to as “enhanced stability bearings.” In these bearings the tilting stiffness is increased either by making the bearing an assembly of small bearings connected by steel plates {69] oF by individual rubber discs connected by steel plates [128]. The first application of a bearing of this kind was carried out for circuit breakers at the Edmunson Power Plant near Bakersfield, California (69]. The stability analysis given here for the single bearing is applicable to these enhanced stability bear- ings. Influence of Vertical Load on Horizontal Stiffness When the load carried ‘by the bearing is ccmparable to the buckling load, the horizontal stiffness Ky is reduced. The reduction is obtained by using the same linear elastic analysis, and is given by a If the load is less than 0.32 times the buckling load, the accuracy of the usual formula for Ky is better than 10%, and in most designs that requirement will ‘ensure that this is the case. ‘The downward displacement dy of the top of a bearing carrying a vertical ‘oad P and displaced through a sideways movement at the top of D is also given ‘by the buckling analysis in the form