Scribd Logo
Upload

Welcome to Scribd!

  • SED02

    Uploaded by

    George Alexiadis
    20 views70 pages

    Copyright

    © © All Rights Reserved

    Available Formats

    PDF or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    Report this Document

    Copyright:

    © All Rights Reserved
    Download as PDF or read online from Scribd
    Flag for inappropriate content
    20 views70 pages

    SED02

    Uploaded by

    George Alexiadis

    Copyright:

    © All Rights Reserved
    Download as PDF or read online from Scribd
    Flag for inappropriate content
    of 70
    Structural Engineering Documents 2 Hajime UMEMURA Haruo TAKIZAWA UU std ky 3 Utah te) td) CONCRETE BUILDINGS International Association for Bridge and Structural Engineering |ABSE ‘Association Internationale des Ponts et Charpentes IPC Internationale Vereinigung fir Brickenbau und Hochbau IVBH About the Authors: Recipant of his doctors! dro in 1949 ftom she Univesity of Toxvo, {aif onthe mornanct of baling structure, He ws ade the eer. interasted inthe arthavakerosstont ‘sgn of reintoread eonrete Bus ‘word of precicing engineers anda ‘hored vrius tecnica! pobliestions. His aversfied seve include the thosnitation of US-dapen coopers Feiparch, and hes curenty the View President ee interntiona! Assoc ‘on for Earthquake Engnening orn in Tokyo in 1987, Horse ‘TAKIZAWA studied atthe anaineer- Ingsehool of the University of Tokyo lind obtained hi doctor! degre in 41074 under the guidance of Proteror ‘member of Hoke Un I Sappor, hare he hole Profesor of Stratural Engineering ‘Technology n Paradens, USA Ins teaching and aseren me a of reinforced conerets budge Structural Engineering Documents 2 Hajime UMEMURA Haruo TAKIZAWA DYNAMIC RESPONSE OF REINFORCED CONCRETE BUILDINGS International Association for Bridge and Structural Engineering IABSE Association Internationale des Ponts et Charpentes IPC. Internationale Vereinigung fiir Briickenbau und Hochbau_IVBH ‘copyint © 1962 by Inmatona Astocition for ridge and Struct! Ensnerng ‘Al is reed No part ofthis Book may be reproduce in any frm o by any mats, ace or mechani, Ineusng photocopying recog by any infomation soap andrea syst, without erin in wen tom he publisher. Primedia Sulzer Publier: 1ABSE- Airc 1VBH ETH. Hingaete (04.8008 Zien, Sitar a:ovar726e7 ‘elon: B22 188 ABS CH “eleg1ABSE, C8003 Zieh CONTENTS Seprxegesen Introduction Principles of member-by-member analysis ‘Stifnes, strength and detocmebility of elements {Load deformation relation of elements [Mechanics of fully nonlinear beams and columns Additional aspects in member by:member modeling: Examples of member-by-member analysis of planar frames ‘Simple modelings of overall behavior Basen Introduction 1. INTRODUCTION As the necesity of « reliable evoluation of the structural sofety ogoinst desteu ‘earthquakes is increasingly felt, more ottention has been given to the dynomi performance of building frames deformed into their ultimate states. The primery ‘concern therein is @ better understonding of the process of failure by relating the mechanics of structure to the characteristics of intense ground shaking. This type of question is being osked repeatedly in the formulation of earthquoke-resistant building standards a: well os in the design of specific important structures. In terms of the dynamic structural analysis, an important step forward to mee current need will be token by reflecting the nonlinear structural behavior in @ sufficient agreement Perticularly strong-motion response studies along this Tine can be of a practical Importance for lowrise reinforced concrete (R/C) buildings, becavse this closs of structure is comparatively vulnerable to eorthquoke excitations. The vulnerability has been demonstrated by their critical damage sustained in recent earthquakes. The Present report, which briefly outlines the important ospects in modeling the dynomte failure of R/C frames from the particular point of view, is intended fo provide 2 summary of the progress having been made during there twenty years, In the 60's ond 70's, the wide oveilobility of high speed electronic computers and the axtociated development of numerical techniques have had o significant contribution to the ease of ccorrying such studies. Also, the 1968 Tokachi-okt earthquake in Japon ond the 1971 Son Ferondo earthquake in ‘California were epoch-making events in this Feld of structural engineering. References [1, 2]* are among the review popers related to the current subject In formulating the strong-motion response behavior of @ structural system, the # involved are well separated into two main categories. One is the so-called "material" nonlinearity that refers to the Inelastic and hysteretic properties of restoring force arising from the hereditary strest-strain relations of constituent elements. The other is the geometrically nonlinear effects axtoclated with the deformed configuration of system. For the R/C framed structures of practical interest, leration to the former factors alone suffices fer attaining most of the present fe, while the latter may become fully responsible for the system destobilizati in the limited region near ultimate collapse. Reasonably close identification of the inelastic ond hysteretic performance of R/C buildings ond its odequate mathematical modeling can be indeed a difficult practice. Throughout the current report, an empirical approcch to @ set of generalized nonlinear formulations is emphasized on the basis of the direct use of the member properties derived from experiments, With differing levels of complication, this leads too straightforward, but often cumbersome, menber-by-nember modeling of total system, or te 0 considerably simplified modeling that incorporates the restoring force characte of members. To use stress-strain formulations for the entire structure Ts judged less desirable, since such opproaches will be for more unwieldy without ony compensating improvement in reliability. It is in fact probable that when accounting for the inelastic ond hysteretic action of the stress-strain models with current concepts ond methods the laborious analysis will not describe actuel behavior os well os the * Numerals in brackets indicate the reference nunber in the 2 Dynamic response of reinforced coneete buildings member-by-menber models can. ‘A synthetic view of the whole problems related to the member-by-member formulations js included in this report, which Indicates necessary fundamentals for the in-depth study of dynamic behavior, An elaborate analysis of the current type can positively reflect all minor details of structure, and there is one-to-one correspondence between the components of on actual frome and Its idealized system. By use of real ‘onelyticel models for the constituent elements, this will greatly contribute to predicting fr reproducing, with @ desired accuracy, the behavior of « specific building under revere earthquakes. (On the other hand, the interest of onalysis is frequently token in the overall Performance of structure disregarding the microscopic local effects. A necessity alto frises to examine the erude but general nature in the strong-motion response of ‘certain group of buildings. In such instances, simpler and gress models of total system re preferred 10 long at the reduction does not impolr the essential features in the nonlinear dynomte response. Some of the work in the current area have proceeded ‘along the direction of developing @ macroscopic model that results, with relatively mall efforts, in a satisfactory estimate of the significant trends. In edition to the ‘complex modeling, enother emphasis of the report 1s put upon this aspect of simple formulations. 2, PRINCIPLES OF MEMBER-BY-MEMBER ANALYSIS. ‘The evident advantage of @ menber-by-member enalysis lies in the quar ‘assessment of overall structural safety with a direct reference to the process of the ‘damage wstained by each contituent element. Provided that complications in the procedure of computation pose no problem, this Is best fitted to interpret the dynomi failure of a porticulor due to 0 destructive ground motion. Also the onalysis nt capacity of existing buildings by Since the advent of high speed electronic computers, rapid progress hos been made in the general orea of nonlinear stress analyses. Most of the researches ore however found within the traditional context of structural or solid mechanics. Even though steel frome: may be suficiently analyzed by meons of such approaches, the member~ by-member analysis of R/C frames it under considerably different For instance, the conventional anclysis of material nonlinearity often fells to evaluate, with 2 proctical sufficiency, the inelastic deflection of R/C structures, Inelastic ond hysteretic performance of R/C frames would be clarified on the boris of the fundamental ler of mechanics, if the material properties of concrete and Ieinforcement were realistically modeled under arbitrory triaxial stress history and then ‘combined with such displacement discontinuities ot flexural and/or shear cracks, truthing and spalling, bond slippage ond dowel cction, Not a few studies along this Tine con be found in'the litercture, and the enalyses have been partly applied to cxamining the results of the leading test of structural components. However, the Current state of the ert is far from fctory; od hoe rationalization: have been frequently used for toking account of complicated and unclarfied factors. It may be ‘lio affirmed thot the overly detatled characteristics obtained from the nonlinear ‘analyses of stress ond strain fields in the interior of an individual component ore Principles of member-by-member analysis 9 unnecessary for the purposes of describing structural behavior. On the other hand, serious difficulty is felt in the more elementary method following cross-sectional Properties of mombers, The influences of crack pening ond bond slippage ar well as the action of shear cannot be properly accounted for, which are closely related to the Important Features such as the stiffness reduction after cracking and the degradation of hysteresis loop [e.9., 3, 41. In the last three decades, much research effort in the various fields of structural ‘engineering hos been made for investigating the inelostic ond hysteretic performance of structural elements by meant of laboratory experiments. Notably in Jopan, great ‘emphasis has been placed on the empirical derivation of the menber character Under thelr states of stress in a frame acted upon by lateral forces, One way to developing @ better method of member-by-member analyses can be obtained by Incorporating directly the properties of various elements derived from the test results. This Is the standpoint having been taken by Aoyana [5] in his general formulation for the inelastic deflection analysis of R/C structures. In the practical method of on empirical nature, the structure under consideration was treated at an axemblage of columns, girders, shear walls, beam-to-column connections ond foundations, and the Inelastic characteristics of these elements were to be independently estimated by wse of cvailable experimental dato. He performed static enalyses of the Inelastic deflections of o multistory frame subjected to reverzal of lateral load, the results of which were found in @ sotisfactory ogreement with the cbservation in its field test, Noteworthy ‘lso is an advontage of this approach to permit to examine the domoge level of each member modeled in the numerical onalysis with o direct relation to the actual one observed in the corresponding laboratory test. The bosic principle of the stove opproce! However, solution of certain problems Ts required for its completeness. The member Properties obtained from experiments are inevitably restricted to thore under simple conditions of loading; thus they do not straightforwardly coincide with the states of stress of the constituent members in o loaded frome. For example, the behavior of columns is mostly examined under the state of antisymmetric distribution of uniaxial bending moment along the axis of menber and under constant axial compression. Although this type of single-degree-of-freedom loeding, in which the loading verlable | single component of applied end-moment only, can be most preferable in view of the current interest, they are subjected to the action of forces more or less perturbed from the reference state. Another component of end-moment and the axial force tend to vary in the cose of planar frames; the point of inflection shifts instantaneously while their restoring force properties are cffected by the instantoneous magnitude of axial compression. In addition to these factors, the influences of initial strex due to dead ‘and live loads are not negligibly small. ‘On the other hand, the action of biaxial bending moment and twisting torque mokes the sitvation in the case of space frames still more complicated, The loading variables culminate in tix degrees of freedom, In order to have © mathematically tractable mode! of each structural element along the above empirical approach, the whole proces: of modeling may be conveniently resolved into the following three steps. The first is concemed with the modeling of inelasti cond hysteretic characteristics obtained directly frem experiments under simple conditions of looding (usually, with @ single loading variable). This can be then separated into the skeleton curve properties for monotonic leading ond the hysteresis properties for eyelic looding. The latter must allow to specify the load-deformation relation 4 Dynamic response of renforead concrete buildings under arbitrary loading history. The second step deals with the extension of the to those under more general conditions of loading. AAs explained above, this is intended to take account of the actual states in a loaded frome. In the third step, the effect of the deformation rate of member is treated with «© specific emphasis upon the damping forces in non-elastic region. 3, STIFENESS, STRENGTH AND DEFORMABILITY OF ELEMENTS With the aforementioned principles in mind, a basically empirical epproach is token to study the properties of R/C framing members. From this standpoint, it becomes desirable that the important porameters describing the inelastic and’ hysteretic behavior of constituent elements ore given in the form of empirical equations, These should include relations between end-moment and end-retation for flexural or flexural-thear columns and girders including the flexure of shear walls, thove between shear force ‘and sheor deformation for sheor-fotlure-type columns and shear walls, and those between panel distortional moment end shear distortion for beam-to-column conne: Some of the lood-defermation relations are simply illustrated in Fig. 1 Hoving been derived by reviewing the data of monotonic loading terte, «number of ‘empirical equations for evaluating the skeleton curve porometers are available for vorious structural elements and for different modes of failure. Such studies are porticulorly found in Japanese literature; the reference [6] which examined the plastic state of R/C beams in flexure is among the pioneering works, OF those equations, a self-contained group developed by Unemura, Aoyama and their coworkers [7 to 9] has been widely used for practical purposes. Following the guiding principle referred to in the previous chapter, they made o series of investigations for correlating the ‘empirical characteristics of typi Certain set of significant factors. Although being of nature of pilot study, this includes the skeleton curve parameters in flexural (Flexural-sheor) failure of girders and columns, and sheor Failure of beam tercolumn connections and shear walls. Even the reliable characterization of the Failure process of girders ond columns cannot be simple. Port (a) of Fig. 1 shows the members subjected to antisymmetric loading with the single varying factor of bending moment and shear, ond under invariant axial force. Depending upon the specimen details, their ultimate states are differently characterized by the phases of flexural yielding, thear failure, bond failure, rushing and spalling of concrete, steel buckling, ond so on. A total review on this subject is beyond the scope of the current report ond should be best left for more specialized literoture. The following discussions begin with the relatively simple evaluation of their flexural (flexucel-shear) parameters. In most coses of practical interest, the load-deformation diagram is featured by several breakpoints, the skeleton curve tending to become 0 polylinear type rather than tingle rounded one. These points can be interpreted to reflect the flexural cracking ‘of concrete in tension, the development of flexural or flexural-shear crack of concrete, the yielding of tensile longitudinal reinforcements and the compression failure of concrete, all of which occur at ond near thelr critical sections. Astoclated with the cracking, moderately reinforced members exhibit o sharp break away from the elastic bbranch at @ very low level of bending moment as compared to the yielding of reinforcements, Also, thelr moment-bearing copacity does not increase appreciably Suifnes, strength and deformability of elements 5 (2) cotums and Girders (©) shear watts I Rs g ap => s (c) tean-to-ctumn Conactions (6) SubeElmants for Catums and Girders Fig. 1 Framing elenents of R/C structure and their Toad-defornation relations examined in experiments 6 Dynamic response of reinforced concrete buildings beyond this yield up to the ultimate failure related to the crushing of concrete, The post-yield branch con be fully plastic; the ductility then deteriorates according to the progress of crushing and spalling. Thus @ tllinear expression of the skeleton curve characterized by the two points of crack and yield becomes a prerequisite for any more sophisticated ideclization. Four parameters are necessary therein, for which conven tionally used are the initial stiffness at uncracked stage, the erack ond yield moments, tnd the reduction factor of secant stiffness defined ot yield point and relative to the initial stiffness. Given the geometry and the reinforcing specification including the mechanical properties of materials, the emplcteal equations prepored by Sugano et el [8,9] perm fo evaluate all these parameters. While other stondordized procedures in the plas theory of BC sections may be vsed for the crack and yield capacities, the most important contribution of their study x found in the for the stffess reduction factor aP yield point, Corteltion factor aed ‘equation ore a set of dimensionless structural parameters related to the te forcements, end the oction of sheor and axial compression. This ean provide o great ht into the nature of momenteresisting R/C frames. On the other hond, much fs not known quontitatively on the deformation developed eround the ultimate fatlure, The ductility hos been recognized to be highly dependent upon the magnitude of acting ‘axial force and the amount and specification of web reinforcement, which reflects the sealing of concrete cove onthe oucited buckling of lrgiucnalrenfercenents 10}. The empirical equations for evaluating the necessary porameters in the shear failure of girders and columns, which are not included in the above group of equations, can be portly found in the literature. Lee [11] examined the ultimate deformation of shear failure-type beams ot well of their crack and ultimate capacities, while Arokawa [12, 13] derived @ set of equations for the stret asiociated with shear cracking ond shear failure of beams, A study of Ikeda [14] extended the applicability of these equations to shear-failure-type columns by including the influence of exial compression upon the levels of stress, The reference [15] is @ summary report for the studies mode in Japon. ‘The capacity of shear walls to resist lateral forces hat been of o long time research ‘concern. Their deformation is usvolly resolved into the t shear, only the contribution of shear being indicated in Fig. 1(b). An empl ‘examination of their crack and ultimate strength is found in the above-noted work of Sugano [8, 9]. In @ series of investigations concerning the earthquake performance of shear walls, Tomii (16, 17] identified the deformation at ultimate failure. On the other hand, much effort has been mode to prevent the brittle sheor failure of R/C buildings since the 1968 Tokachi-oki earthquake in Japan. One direction in such studies was taken to derive @ gross criterion on the shearing opacities of total structure [18 to 23]. The practical concept of "wall index Introduced therein is worthy of o special notice in relation to the shear strength of columns and walls, Instead of the ossumption of rigid and pointwise joints, idealization of the beam-to~ column connection by @ panel subjected to sheer distortion is @ modern opproach. This allows to take account of the shear erackings sometimes observed within the joint area. The behavior of the connection element is specified in terms of the panel distortional moment (average shear stress multiplied by @ geometrical factor of its volume) and the Suitfnes, strength and deformability of elements 7 shear distortion as shown in Fig. I(c). Use of enpirical mentioned study of Sugano and Koreithi [9] leads to o tri skeleton curve, An additional illustration in Fig. 1(é) is the in-plane behovior of foundations. Their three~degree-of-freedom motion consists of the components of heaving, swaying and rocking. The next discussion is done for the sub-elements combined with columns «2 proctical building frame. Some typical examples havi alven in Fig. Ile). The stiffness, strength and deforma affected by their combination with R/C slabs. An approximate procedure to account for the influences applies corespondingly to the elastic modeling ct T-shape beam by means of the standardized evaluation of effective width. However, the actual Influences on the important parameters such os the yield moment and the stiffness reduction factor ot yield point are not yet cleorly Understood. References [24 to 26] ‘re among the experimental contributions to this problem. Siab reinforcements become fully responsible for the increase of yield moment when the upper side is in tension, The effects of spandrel or wing walls can be more complicated, In some cases, these sub-elements dominate over the behavior of girders or columns, turning their failure from flexurel type to sheor type. The severe damage to RVC buildings in the 1968 Tokachi-oki earthquake in Japon, which wot caured by shear failure of the columns under constraint of spondrel walls, is © good lesion of the significance of the present effects, Several experimental studies on the behavior of structural elements spandrel or wing walls can be found in the literature (27 to 32]. In closing the gross survey in this chapter, flexurel properties of R/C sections are supplemented by means of o more fundamental approach. A strotghtorword Integration of the moment-curvatre relation of sections doesnot necessarily lead to the load ‘deformation relation of @ R/C column or gitder, a mentioned previously. In spite of this severe initaton, the monent-curvature relation ean play on impertent role ‘examining the inelastic ond hysteretic characteristics of R/C members [33 to 41]. Evaluation of the above relation is bosed on the filament formulation; the uniaxial stress-atain relations of concrete and longitudinal reinforcements are specified in e certain way, and then assembled over the entire section by use of the Bernoulli-Euler hypothesis thet the section always remains @ plone. With no computational difficulty, this permits to occount for the irregular configuration of section and the leregular ‘rangement of reinforcements. Moreover, the influences of the varying mognitude of ‘axial force and the action of biaxial bending monent can be readily incorporated into the formulation. The reliability of onalysis depends upon how realistically the stress- strain relations of concrete and reinforcements ore modeled, For instance, the descending branch of concrete beyond the point of peck compression strength wos separately specified in a manner to reflect the confining effects, Also the stress= bearing capacity of cover concrete beyond the ultimate compressive strain was different, in some of the literature, from that of core concrete. In formulating the stress-strain ‘elation of reinforcements, it is @ recent trend to toke account of their monotonically fond lsotropically expanding yield criteria as well as the strain hardening ond the Bouschinger effect. Along with these studies, 9 strictly analytical approach has been token which opplies the finite element formulation of continuum to the member [e.9., 42]. This is best 8 Dynamic response of reinforced concrete buildings fitted for examining the mechanism of sheor resistence; however, a totally deleted from the present report particular expect 4, LOAD-DEFORMATION RELATION OF ELEMENTS. Besides the properties of skeleton curve exhibited under @ monotonic loading, the ‘effects of hereditary response can be fully significant on the dynamic performance of « structural system, To date many restoring force modeling: have been propored for providing « better opproximation of the actual behavior of constituent menbers, each having its own individual character and historical raison d'etre. In this chapter, @ brief survey on these models is presented for o general class of hysteretic characteristics (Vide alto the excellent reviews on this subject by Aoyama (43, 44], Bresler [45], Pork [46], Sozen [471, ete.) with the primary emphatis on R/C structural elements. The simplest model for the study of inelastic and hysteretic mechanical systems te the bilinear model, which includes the elostoplstic system as 1 special case. In the pat ‘9 number of studies [e.g., 48 to 84] were done with o view 10 clarifying its dynomie behavior from not only theoretical but alto practical points of view, Nature of the response to both steady ond transient excttotions wos well examined by vse of deter ie and stochastic approaches, the implication: in their results having been reflected Upon the eorthquoke-resstant structural design of many buildings. Actual R/C members, however, cannot be sufficiently characterized by the elementary load-deformation telations; tether, the skeleton curve has a very marked tamation branch of signficent softening that then leods to a plastic branch. Upon reversed loading, the odaiional fects of sifness degradation become important particularly in the Righly inelostic region, ond yield crticolly low capacity of hysteretic energy loss, Such features cbserved commonly in W/C system ore far from thore represented by the ideal elesto~ plastic of bilinecr models, which mokes i totally umrealistic to give even @ rough fstimote of the dynamie response of R/C structures on the basis of the response propertoe derived trom them. (One direction of the attempts to generalizing the restoring force models can be traced long o Tine of the some properties of hysteresis as in the case of elastoplastic or ls general concep cond boc fo very ery work by Mesing [85] and is often referred to as "Masing type", oF synonymously called "normal “nondegroding". The hypothesis used therein postulates that the curves defining i branches in a restoring force diagram are geometrically similar to the skeleton curve, oF more specifically, the restoring force curve related to 0 closed symmetrical loop is of the same form ot that of the skeleton curve enlarging it by @ factor of two, A well-known family of single rounded skeleton curver, which war originally formu~ loted by Romberg and Osgood [86] and is featured by differing degrees of gradual break away from the elastic branch, was associated with the hysteresis concept by Jennings [87 to 90]. In eddition to’ the work of Jennings, references [91 to 93] ore ‘among the studies examining the opplicablity ond dynamics of the Ramberg-Osgood systems. Then Iwan [94] submitted © more generalized formulation. The latter per to specify the skeleton properties of a particular system with much versatility. Tl bosed on the assumption that a general hysteretic system can be thought of as an cxemblage of ideal eloitoplastic sub-elements having different yield levels, or collection of perfectly elastic ond rigidplastic (Coulomb-tlip) elements. Even though the wide class of restoring force models may provide @ realistic description of the Losd deformation relation of elements 9 behavior of steel structures with welded joints, it connot be directly opplied to R/C structures from the reasons already noted. Dynomie properties of the Iwan system have been investigated in a series of reports [95 to 10]. ‘Another direction to improving the elastoplostic model was token by Clough [102, 103]. His modeling was motivated by accounting for the prominent effects of stiffness degrads~ fion ebserved during the eyelte loading of R/C strictural components. Viewing from the historical developments, the degrading model having elestoplestic (or bilinear) skeleton may be conrisered at the proto-type of o class of R/C hysteresis models proposed thereafter. detailed examination of 1h dynamics can be found in the Titerature [104 to 106], which clarified the implications of the stiffness degradation in relation to the original nondegrading system. With the bilinear skeleton curve and Fdealizing the degradation of reversal stiffness in different ways, several versions of the Clough model hove been defined in more recent studies of the earthquake response of VC structures [107 t0 113]. When including the breakpoint of eracking by meons of trilinear representation of the skeleton curve, a better approximation can be obtained such as the degrading trilinear systems by Sugano and Fukada [114 to 116], and by Muto et al [117]. In particular, the Fukada 8 @ simple expretion of R/C character hysteretic energy loss, which accompanies this stifiness degradation, was also formulated into another generalized model of Iwan [118 to 120]. Again the model comprises distribution of fundamental mechanical elements; in the current instance, a new compo nent of ireverible or directional slip behavior is vied to account for the irecoverable modification oF deterioration of system during the loading process. His attempt to Including @ larger class of hereditary systems in a physically motivoted way 1s admittedly of much theoretical interest. However, certain difficulties are felt in modeling actual YC performance by means of the Iwan formulation. This sill fails to reflect even the bore-minimum feotures of R/C systems. In constructing an elaborate model that is as consitent as desired with the experimen tally observed characteristics, one is necessarily ferced to resort to formulations of @ more empirical nature. Although it is impractical to test every conceivable care of leading history ond consequently certain assumptiors must be introduced to perm! ‘extensions of the empirical hereditary law, such approach can be the most efficient in formulating the complicated behavior in question. For instance, one of the notable Features known from an inspection of R/C load-deformation diagrams is the trend of peck orientation [121]; for the suecesive reversal 2f loading the restoring force curve tends to the abrolute maximum point ro far experienced in the opposite direction of leeding. Its postive consideration hot motivated the invention of the stiffness degreding models of Clough and others. Of the R/C models that fall within the current category by Takeda (122, 123). spection of the cyclic behavior of flexural-failure-type R/C. components. The model hot been reported to be in an ettential agreement with the ‘observations in laboratory experiments and to succesfully the results of dynamic failure testing of R/C frames on a shoking table [e.g., 1220 125]. With a specific empharis on the empirical choracterstics of complete ond symmetrical hnypteresis loops, a different group of restoring force models has been used by Shiga et al [126 to 131], and by Toni et al [132 to 140], in which the hysteresis loop is modeled separately from the skeleton curve, This approach is quite reasonable and will offer a 10 Dynamic response of reinforced concrete buildings Wide possibility for future vie since R/C hysteresis loops are not so uniquely deter~ mined by the employed skeleton properties. Even if confining the problem to the simpler coses of flexural failure, thelr geometry is noticeably affected by the amount of longitudinal reinforcement, magnitude of axial compression, additional action of sher, specification of reinforcement (round or deformed) and details of web reinforcement. Moreover this can vary with the amplitude of deformation; some of the factors ore closely related to the significant influences of bond slippage. Correlation of the shope of hysteresis loop with the factors is, however, not yet clearly understood. According to 2 qualitative examination by Kokisho [141], a more stable loop con be obtained, in the flexural-fallure cases, under smaller axial ond shear forces and with more web reinforcement, and the smaller the axial force is ond the larger the shear force and the ‘amount of reinforcement are the more appreciably observed is a trend of hard-spring (lnverse-S) loops. Recent progress on this subject can be found in a study by Nakata ¢t al [142], where several empirical equations for practical wie were developed relation to a set of column parameters. Associated with the hard-spring property, studies of the conventional slip and double bilinear models from the interest of nonlinear vibration theories [143 to 146] are also worthy of a special notice. In oddition, a previous reference [64] includes the earthquake response study of slip-type elastoplastic systems. ‘Among several important aspects concerning the stabilized hysteresis loop, most commonly ‘exomined are the loop area corresponding to the hysteretic energy loss during one cycle ‘of reversal loading and the degree of loop pinching toward an inverse-S form os well os the effective secont stifness. In porticuler, the loop area hat been of a very long time concern since the pioneering work of Jacobsen {147, 148]. Following his intro- ‘duction, the concept of equivalent viscous damping factor is frequently used to. descr the amount of energy loss relative to the potential energy measured ot peak deformation. ts application to R/C systens was reported in an early study of Osawa [149], the previously noted work by Shiga et el [126 0 131] lying along the experimental extensions. Also the references [150 to 152] are related to the subject of the equivalent damping in its use for approximating the dynamic response of simple nonlinear systems. Combination of the varying loop area with the reduction of effective stiffness leads to «@ considerably simplified description of the Inelostic and hysteretic performance of R/C structures; by using the dota of accelerogroph records, lemura et al {153 to 158] Identified and modeled the deteriorating force-ceflection relation of nine-story R/C tly damaged during the 1971 San Fernando earthqueke in California @ similar line of modeling, © design-oriented procedure for taking account of yielding response of multstory K/C buildings was. developed by Shibato et ol [159 0 168]. On the other hand, a straightforward reflection of the loop geometry was the main concern in the previously referred studies of Tont et ol (1320 140]. They reported results of ‘examination of earthquake response; the effects of loop pinching cannat nt of the porameter representing the hysteretic eneray loss, which consistent with on intuitive understanding. The pinching behavior of loops has been ‘also incorporated into the modified Takeda models (164, 165, 111]. A remaining feature relevant to the R/C restoring force characteristics is the gradual decline of load-bearing capacity observed during a successively cyclic looding under specified anplitudes of deformation. The trend may be insignificant and relatively stable loope con be obtained in most cores of flexural failure. On the contrary, this instability of hysteresis loops becomes important in the cases of shecr failure. With the clostoplastic skeleton curve, Matsushima [166] examined the effects of the successive Loaddetormetion relation of elements 11 deterioration of yield strength in evaluating the earthquake response. An important contribution to modeling this class of sheat-failure-type restoring force properties wos ‘made by Aoyama [167 t0 173, 1], in which e serie: combination of bilinear ond negative slip models wos used for understanding the severe shear fatlure damage to R/C columns sustained during the 1968 Tokachi-cki earthquake in Japan. The oforementioned works of Tekayanagi and Schnobrich [164], ond of Nakata et ol [142], include o ‘mathematical modeling and an examination of empirical data for the current feature, respectively, ‘As outlined in the foregoing gonerol remarks, sufficiently realistic formulation of the restoring force relations of R/C structural components is indeed on unwieldy problem, ‘and the current stote of the ort still leaves much r2om for more advanced investigations. In particular, laborstory test data should be obtained ond reviewed, under a definite i le, for all the vorieties of structural specifications of practical impor 31 formulations must follow with different levele of compl the most sophisticated model ir rendered versatile enough to the full details in a specific case, while the mode! for practical design purposes being reduced to a bare simplicity by retaining the extential features therein. In the Femainder of this chapter, additional comments are given without becoming involved in those diseusions, ing « schematic expression, Fig. 2 summarizes the most important aspects in IVC restoring force relations. The Figure implies, in parts (a) and (6), that the lefi~side idealizations should be replaced by the right-side ones. This leads to {2 degrading trilinear or quadrilinear formulation thet satisfies the basic requirements for RVC systems. On the other hand, necessity of the flowings fram left to right in parts (c) ond (4) is influenced by the individual sitvatiors, Pinching action of the degraded loop into a regulor-S form occurs in prestressed concrete components accompanying still Poorer copacity of hysteretic energy loss. It should be again emphasized thet the egradation, pinching and instobility ore closely dependent upon the amplitudes of eformation. ‘A conventional formulation of the degrading trilineer (or quadrilinear) system is ‘explained in Fig. 3. This is based on the stiffness degrading concept originally ‘proposed by Sugono and Fukada as previously referred to. The model wos then mod for cllowing to specify the different skeleton characteristics in positive and negotive directions of looding (Fig. 3(0)|, the necessity of which is always encountered in the stance of girders. The post-yield skeleton curve in each loading direction may be replaced by ancther polylinear or curvilinear expeesion; the quadrilinear form ean eccount for the deterioati ty ossociated with the erushing and spalling of the hysteresis rules, the geometry of @ double-parallelogram on formation plane [the shaded area’ In Fig. 3(o)] plays a key role os shown the right helf of Fig. 3(b). Once on absolute max'mum deformation Ts newly atta ‘long one of the post-yield branches, the double-perallelogram deforms on the plone in ‘2 manner which is determined by the’ absolute moxinum points in the two directions [Oma Fax) and (Orin » Frin) in thet figure]. Itt deformed configuration geometrically similar to the original shape olong the separate axes of load and defor= ‘motion, With the ordinary bilinear hysteresis low keing applied within the current ouble-porallelogram, this realizes @ model whore reversal stiffness degrades monotoni= cally after yielding. For a direct comparison, the corresponding Mosing type interpre tation is indicated in the left half of Fig. 3(6}. The double-porallelogram translates on 12. Dynamic response of reinforced concrete buildings (0) Hysteresis ; ‘f_. ee inte tm) (s Loop Properties ~ Degradation j7.. 6 Loop Properties - Pnchina J (c) tysteresi Seto mare a Pe pans tm) (4) nysterets Loop Propertfes = Instabitity relation of R/C systens Fig. 2 Schematic display of the important aspects in modeling restoring force Mechanics of fully nonlinear beams and columns 13 serch point i poston station ¥ (e690 + aie pont In pottie 1245) © ie point In neat (Geax Bi (=) ‘ten F conorent © (Petal (2) Qquadritinear Skeleton Curve Having Different Properties in Two Opposite Directions of Loading, and the Associated Double-ParalTelogran (&) Tho Difterent Laws of Hysteresis Assoctated with the GuadriTinear Sketeton curve Fig. 3 Formulation of the degrading quadrilinear system, including the different hysteresis: Taw of nondegradation the plane according te the loading beyond the yield points. ‘When applying the degroding low, the so-called equivalent viscous damping factor becomes constont for any post-yield closed and syrmetvical loop. This is given by (2/2)(I-ay)ac__ where ac and ay stond for the ratio of crack versus yield copocities tnd the reduction foctor of secant stiffness defined ot yield point and relative to tlostic stiffness, respectively. An advantage of the R/C system, in oddition to its simplicity, is roughly the same eveluction of peak inelastic displacements of earthquake Fesponse es does the more elaborate model of Takeda. However, the model is seriously flowed by its several unrealistic features in the lood-deformation relations. For instance, the actual degradation of stiffness occurs alzo prior to yielding and con be generally more significant. Thus the hysteretic energy lose may be still overestimated. 5, MECHANICS OF FULLY NONLINEAR BEAMS AND COLUMNS Several fundamental aspects in the menber-by-member formulation are noted in the following two chopters, which are relevent to the modeling techniques for constituent ments. With the empirical load-deformation relations completely specified under ple conditions of loading, the first concer is given to the experiment-to-analyeis lge" problems for girders ond columns os illustrated previously, These modelings resort to some reasonable simplifying essumptions since the practice involve cer 14 Dynamic esponse of reinforced conerete buildings substentiol difficulties due to portial lack of empirical knowledge. An attempt to have an overall view of the whole problems was made by Takizawa [174]. For both planar and space frame analyses, the flow of modeling procedures from the current point of view may be summarized os shown in Fig. 4. ‘Among the diversified aspects indicated in the Figure, development of the so-called beam model" plays the mest fundamental role. This is used to formulate the hereditary response of the cl ment. Stated more precisely, successive evaluation of 2 2 instantaneous stiffness matrix is the interest therein for a simply supported member subjected to Hi ‘uniaxial bending moment with fluctuating location of inflection point ond to the odditional ection of constant axial force. The matrix need be specified in terns of the hereditary histories of two end-moments, under a premise that restoring force properties cre known for @ single-degree-of-freedom loading. Usvally antisymmetric distribution (of bending moment olong the axis of menber is token for the reference frame. It should be noted that the cross-sectional properties generally differ between the two ical sections as well os the different properties during positive and negative flexures. For gieders and columns foiling in shear, thelr shear deformation should be treated Separately from the flexural deformation. This is olso the case when idealizing {shear wall by means of @ deep column. The stifnest relation for there ‘obtained by summing the flexural flexibility matrix, the evaluation of which needs o certain beam model, to the contribution of shear deformation, In the early days of member-by-member analysis, the ideal elastoplastic properties were assumed for all the mombers [175 to 180]. ‘Also the location of inflection points was self-evident for the Remberg-Osgood beams in single-boy symmetrical frames {181, 182]. Limited to such exceptional situations, the beam model can be of no vie. In the subeequent efforts to obtain o closer idealization, various beam models have been eveloped ond were implemented into the computer programs of member-by-member onalysis. Two-component formulation of Clough [183 to 185] stonds out as the first proposal of the beam model. This assumes the conventional law of bilinear hysteresis, ond is one of the most widely ured modeling [e.g., 186 to 190]. The dual component member ‘contists of @ combined elostie member plus on ideal elastoplastic member in parll Its extension to the nondegrading trilinear hysteresis was independently made by Aoyama [5], which is referred to at “divided beam" model of “comperite beam" model. The latter is a four-component R/C member consisting of on elastic member combined With theee ideal elastoplastic members in porallel, and can account for the cracking ‘ond ylelding characteristics different at the two critical sections. Giberton [191, 192] developed @ somewhat i one-component model. His system is on elastic member with i fpringe concentrated at two menber ends. The equivalent springs are ‘embody the inelastic ond hysteretic action that occurs within the actual member. This Cllow: enough versatility to match ony lood~deformation relation, which is the meta rearon for ite wide uie [e.g., 19810 195, 165, 111]. He maintained that his model ‘war more advantageous thon the two-component model because of this efficiency. The rrulti-component formulations of Clough and Aoyama, however, may be modified into ‘on entirely general form that allow similar versatility os does’ the Giberson's formu- lation. Actually this generalization wos reported by Takizawa [174]. The fundamental [Mechanics of fully nonlinesr beams and columns 15 uy i g a ®) i 0 : | Hd | E me i 3 hy TEH | 5 Hl) alle : bi ET Lae ia BD ag — (HIS i Hie ; ; il ! 2 | Ge a 3 lg fie e | i : im : z Kf es et il 2 ‘deo of modeling wos extensionally interpreted there; the pattern of beam partition ts rot prescribed but varying according to the curent state of menber. Three-component Partition suffices for the modified formulation, The beam model generalized in such ‘2 manner and the one-component model can be complementary one in the meaning thot the roles of force ond deformation are converse in representing the reduction of stiffness, Both the generalized composite beam model ond the one-component model ore flawed 16 Dynamic response of reinforced concrete buildings by certain serious disadvantages, even though their uniphisticatedness may be very attractive. Several cluney and unreotonble points feature the former modeling [174]. ‘On the other hond, the equivalent spring idealization in the latter model con be "when the inelastic zones develop far from the critical sections, the # action between the two ends depending on the elasticity of member alone. In relation to this cipect, some versions of the one-component model have appeared In the'Iiterotuwe (197, 198, 141], The concentrated springs were correspondingly replaced by two prescribed regions that were subjected to inelestcity of uniform distribution, The reglors were located, for example, In the length of ten percent of clear span, of rember depth, from the menber ends, Also, the two models of Clough ond Giberson ‘were implemented into menber-by-menber frame analyses to moke a close comparison Under the some low of bilinear hysteresis (199, 200). No significant discrepancies in ‘exponie were reported in the study With @ view to constructing © more occeptable model of R/C beams, two different ‘opproachet were independently taken by Takizawa [201 to 203, 174] ‘and by Ota [204, 205, 109]. One is based on the assumption of the instantaneous distribution mode fof cross-sectional flexural ond shear rigidities clong the axis of member, and the intro= ‘duction of the notion of preudo-ontisymmettie rotation, This can toke account of any hereditary relation. In addition, en important case con be modeled where the inflec~ tion point is located outside the member. The other consists of « combination of two cantilevers with time-varying span length, the properties of which are specified by ‘means of @ reference cantilever. The elementary theory of beam deflection is used in the loter formulation, A medification of its modeling procedures has led to preserving the symmetry of instantaneous stifnest matrix [174], ond reliability of the model wos portly examined in the reference [206]. The beam madels summarized above con be well arranged Into a unified representation form. This permits to make @ comparison of their validity. The question may not be onswered in the affirmative, ince the available date of experiments are limited, and ‘moreover, all the modelings involve fotal drawbacks accompanying the drastic reduction of the related complexities, However, @ tentative conclusion can be derived on the felatively poor opproximations by the compotite beam and one-component models, judging from clearly different trends of stiffness reduction under particular conditions of loading [174] Necestty of including additional two factors Into the beam model arises in some cases. ‘tress coured by dead and live loads. Its consideration for gleders cannot be streightforword because of the violation of the linear distribution ‘moment olong their longitudinal axis. The other is the time-dependence of Inelastic ond hysteretic characteristics; the skeleton curve specified at the beginning may vary noticeably during the process of dynamic response. With columns, for ‘example, the skeleton parameters such as the crack and yield moments tend to fluctuate according to the Instantaneous response of axial force. It is a recent trend nto the latter effect (207, 164, 165, 208]. For taking eccount nal Factors, some approximate formulations ore ovailable (174). However, @ simple treatment of the initial stess os the initial contribution to tend-moments wffices for mast of practical purposes. Also the time-dependent effect ‘due fo the fluctuation of column axial force will be often negligibly small for @ given geometric form of building elevetion.. Modeling of « building structure as @ space frome mekes the matter still more compli= Mechanics of fully nontinear beams and columns 17 cated. This i primarily due to biaxial bending ard torsion. Not much is known ‘bout the important expects for member-by-menber analy of R/C structures, A two dimensional extension of one-dimensional restoring force systems for formulating the biaxial bending behavior of R/C columns (which will be discussed in a later chopter) may be regarded to fall within the current category of bridging models in the space frame problem, Two techniques on significantly different modeling boses ore noted for supplementing the preceding discussions. These are "multiple spring beam model” and "layered beam model." The multiple spring model corresponds to the moment-curvature formulation that can be found in the early work of elestoplastic or bilinear system [20?, 210]. Hsu [211] developed this type of beam modeling in which the member war lumped into several sub-elements along its longitudinal axis anc thelr restoring force properties were separately specified. Takayanagi and Schnobrich [164], and Emori and Schnobrich [65], recognized its edvantage in describing the Lehavior of shear walle whore inflection point lies outside the elements, and usec the modeling for analyzing R/C systems of frame-wall coupling, In the reference [164], inelastic and hysteretic interaction between bending moment and axial force wot alzo formulated into the onelysis, In the remaining modeling of layered beam, the crass section wat discretized into © number of layers to which stress-strain relations of material were assigned, Thus this coincides with the previously mentioned filament fermulation of R/C sections [33 to 4l],, its direct application to frame analysis being ‘ound in the reference [212]. The torm of layered model was sed by Emorl and Schnabrich [165]. In their analysis, an elestic beam with the critical zone of layered section modeled the exterior column members ot base story. They emphasized its odvonrage 10 toke account of the time dependent variation of flexural properties according to « significant change in the axial force. Then the discussion proceeds to o different subject of geometrically nonlinear columns. The earthquake response of R/C buildings is vivally examined by disregarding th destabilizing action of gravity, which implicitly sppores limited deflections. This he geometry of structural elements remeins basically unchanged throughout process ond that infinitesimal strain approximation moy be ved in der the governing equation of motion. In fact, the axtumption will be satisfactory for ‘most practical purposes, since even seriously damaging response ts restricted to rela~ tively small loterol drifts. However, the influence of geometrical nonlinearity grows dually as the deformations of « sufficiently ductile system progress for into the ylelded region. What then oceurs is the marked reduction of its effective lood~ carrying copacity. This results from the overturning effect of vertical compressive loods acting through sidesway drifts, ond is commonly referred to as gtavity or Pod. ‘effect. With increasingly large deformation, the dettabilization eventually dominates ‘over the restoring opacity. Thus, in case the ultimate failure is examined for ductile YC structures, the above assumption fails and the effects of nonlinesr geometry in the deformed configuration must be adequately token inva account. In the investigations of the lateral ultimate strength of steel beam-column, the PA term hos been a matter of intense concern. Also this has been formulated into simple systems modeling the gross performance of total stru:ture. The related literature on dynamic response dates back to early work of Ruge [213] and Jacobien [214], 18 Dynamic response of reinforced concrete buildings beginning in the 1990's. Recent studies [215 to 217, 182, 218 to 220, 198, 221 10 291, 111, 232] have included the effects of gravity or nonlinear geometry into both member~ by-member and simplified analyses. The trend was toward more reliable exominations of the completely nonlinear behavior necr collapse, One of their important aspects, which is the inclusion of the gravity term into the equation of overall structural perfor mance, is left for the discussions made in a later chapter. The geometrically nonlinear column has been mathematically modeled in different levels Considerable ease of formulation ts provided under a simplifying developed clong the member. It is @ usual respect to lateral drift [e.g., 182], while its rigonometric expression con be found in the study of Jennings and Husid [216]. This term is specified by means of the so-called geometric stiffness matrix in the formu~ lation of frame analysis. With different modes of deflection aszumed along the longi ves of linear ond Hermitian interpolations for determining the matrix have frequently appeared in the literature. Consistency with the deflection eurve evaluated from the concurrent analyzes of material nonlinearity is not necessarily reflected upon this type of modeling. In addition to the above matrix symonymourly led inital stress matrix, contribution of another additional term represented Initial or large displacement matrix becomes relevant to the Incremental formulation of completely nonlinear columns. A more fundamental treatment of the finite deformat follows the we of Green's etroin and Kirchof's ress inthe general crea of solid . ADDITIONAL ASPECTS IN MEMBER-BY-MEMBER MODELINGS In succession to the previous chapter, discussions proceed to summarize some problems Important for menber-by-member analyses. These are panel element formulations of bbeam-to-column connection and shear wall, slipping interaction between structural components, and uplift of foundation. Treatment of viscous damping forces in the evaluation of inelastic ond hysteretic response is clio examined with relation to the properties of dynamically loaded elements. During the lost two decades, the local behavior of beam-to-calunn connection zones hot oftracted considerable atention in the area of steel and R/C structural engineering. In ts a generally eccepted understanding that their resistance to shear can have on cppreciable Influence upon the overall performance of structure, For R/C fromes of proctical interest, utimote failure at the connection zones will be improbeble and this he contribution of thelr shear dlstotion to the overall deflection may not become #0 ing of Framing menbers. The [ola distortion however acts 8 an importent factor t0 increate the flexibility of elostic system, Moreover the zone tends to soften markedly dye To shear crocking. In practical terms of frame onelysis, the joint area has been modeled by © panel of Certain geometrical volume deformed by the uniform action of sheor [233 to 239, 193, 240 t0 242, 174]. The classical notion of the rigid zones that suppres flexural and shear deformetions at member ends is being accounted into this type of idealization, Its four-degree-of-freedom displacement field consists of combined rigid-body displace ments (horizontal end vertical tronslations, and rotation) and shear distortion. The contribution of this shear deformation to the deflection of R/C frames can be closely Additional spects in memberbymember modelings 19 the reference [237]. examined by 2 proctical method suggested According to the 4-DOF specification of displacenent field, it turns out that the ‘concept of ponel distortional moment becomes relevant to the sheer panel modeling [174]. An illustration of Fig. 5(a) is intended to give @ simple interpretation of the dlistortionel moment, where the forces acting on the panel are Finally composed into the equivalent four'sesultonts of F, P, Mond M. As indicated in the figure, the lost resultant is derived os the deviatoric component of the acting moments related to the ‘coupling forces of Qy and Gy. The self-equilibrating moment obvioutly corresponds to the shearing sess of H/Vp transmitted in the interior, ond distorts the panel without Contributing to its tronslatory and rotatory motions, Accordingly, the modulus of igiity mol 5 the shear deforma to the distortional moment and is called "panel stifness." In addition to the deformed configuration of the connection zone, Fig. 5(b) includes an illustration of the displacement compatibili ‘When using the ordinary idealization of shear wall by means of @ deep column, its treotment opplies correspondingly to the instance cf column members and poses no specific problems. Its pone! element formulation [243, 244] con provide better ‘pproximation, even though the current state of the art is limited to elastic analysis. In the lotter modeling, four points at salient comers of the wall are chosen to transmit the internel forces insteod of the two member ends. Each nodal point hos four degrees ‘of freedom for the in-plane motion of horizontal end vertical translations, rotation, ond shearing distortion, as shown in Fig, 5(e). The 4x4 porameters are wed to determine necessarily ond suficiently the interior displacement field. Using @ reference frame of x and y coordinate axes, the x-component of particle displacement, for exemple, is ‘ossumed to be represented by a two-variable polyromial of x and y which ts linear and ‘cubic with respect to x and y, respectively, On the bosis of this specification of displacement field and the associated expression of two-dimensional strain state, appli iple of virtual work leads to the complete. characterization of The nodel point forces introduced during this process are horizontal fend vertical forces, and rotational and distortional moments. The pair of shear defor- ‘motion and distortional moment at nodal points war not included in the original formu lation by Wogewa (243). ‘The bond deterioration between concrete ond reinforcing steel that occurs within a joint core can influence significantly the restoring force characteristics of R/C systems. Therefore, @ special attention hat been paid in the literature [109, 164, 165, 111] to the rotation caused by bond slippage of tensile reinforcement along its embedded length, In such studies, joint rotational spring locoted ot the member ende wor used to ‘account for this factor of additional flexibility. The pullout action of longitudinal reinforcement from its onchored element may be interpreted to cleim an intercomponent formulation. Together with the cracking developed over plural components, this roize: 0 rerious cbjection to the ordinary Principle of member-by-member modeling in which inelastic ond hysteretic properties ‘re separately modeled at individual components avd then combined simply by means of the requirements of displacement compatibility. It appears that the literature on this particular subject is quite limited; the effect of intercomponent continuity of longitu= inal reinforcements was partly examined in the reference [245]. 20 Dynamic response of reinforced concrete buildings Thedue et ten it see cs ttt sr URNS ses seco 1S: eer Seer et Petecagea cog let rte tore N Bateman Smart ete reo as Feats Oregetf = tt rn foe J Siaeancnnimetes ae 2, eesti sendy SB 4 MRS Bae 8AM MA IIE, nity cm (a) Four-conponent conposition of Forces Acting on Bean-to-Colunn Connection (©) before conttguration of Bean-Cotumn (c) before configuration of MATT Element Sitassenbioge Fig. 5 Panel elenent formulations of bean-to-column connection and shear wall Additional aspects in mamberby:member madelings 21 Instead of assuming the simply supported footings tied by foundation-beams, the necestity often aries to make o close formulation af the local behavior at the bose portion of structure. Its practice resorts mostly to the work in @ separate field of Imechanies and foundation engineering. Also this poses @ well-studied problem of tnalytical solid mechanics: derivation of Impedance relations for the dynamically leaded rigid body resting upon or embedded into o half space of stress wave propagation medium. With such prominent factor: at the mechanical properties of underlying soil tnd the geometrical specifications of footing including the embedment depth, the Impedance for vertical, horizontal, rocking end twisting motfone tends to. depend significantly ypon the frequency of sinusoidal steady-state loading. In partieulor, the equivalent stiffness ond damping for the rocking motion have been found to ext rnonelementary potterns of frequency dependence. For structures of slender elevation, lateral loading will frequently produce bose overturning moment thot exceeds the resisting capacity provided by self weight. What then occurs is the uplift motion with diminishingly small area of contact between the Structure end the underlying system. The tronsient uplift condition har been @ matter ‘of intense concern in the practical design of structures. Beyond this condition the rocking spring of foundation becomes geometrically nonlinear, which hes the potential of causing analytical complications. Recent efforh [e.g., 246 to 248] were mode to evaluate currently available techniques for predicting the uplift response. ‘Then follows the formulation of damping properties. The restoring force characteristics of structural components will be more or less affected by the rate of deformation. Most of the ovailable data were, however, obtained under quasi-static conditions, although much research effort hos been recently mede to examine their dynamic effects by various experimental means, In the conventional fromework of structural analysis, only the additional term of viscous damping is responsible for the effects of dynamic leading. As is well known, the damping capacity of o total structural system can be mostly citibuted to the “geometrical (or radiation) dampirg" that results from the dynamic building-foundotion-soil interaction... The inherently low damping of structural compo= nents oF their subassomblages (say, 1% of critical damping) har been evidenced through experiments and observations. Hence the modeling of damped structural systems should bbe on a separate besis without applying the simpler notion of “proportional damping.” The damping capacity assigned 10 the base portion of structure differs olzo among the different components of its rigid-body mot Synthesis of the individual damping properties will offen lead to 0 somewhat large value (soy, 5% of critical damping) evaluated for the apparent damping factor of total system. Certain systematic collections for its empirical date can be found in the Titercture, which were ebtained through cbservatiors of the dynamic response of existing buildings. Such studies consistently indiccte that the damping factor hos @ {91088 tendency fo vary in inverse proportion to the fundamental period of structure. ‘suggests an approximately constant value of the opparent retardation time, which is seemingly influenced by local subsoil conditions. mind, reviewed is the expression of viscous damping term in the inelastic and hysteretic response. Let the dynomic restoring force F of © 1-DOF system be represented by 22. Dynamic response of reinforced concrete buildings FAD, 6) = FC) +c(0)0 in which F is the static restoring force end stands for the hereditary history of the deformation 0. In the elastic stoge, the two terms are explicitly giv -e With the spring and dashpot constants of Ke ond Cq- Two auxiliary express of damping are Te=Ce/Ke ond te Ca/(2/Rke), where Tey Ge and M represent the fetardation time, ‘action of eritfeal damping and osociated inertia, respectively. The choice of idealizing assumptions to account for the viscous damping in the non clastic stage con become a crucial one in terms of the reliobility of dynamic onal Here are exomined three different idealizotions. Specifying the instontaneous ond. Incremental force-deformation relation by dF = Kd0 ond introducing the ouxiliary expressions of T= C/K and [= C/(2VAK), these ore Twel C= (Kio Ce, Type Hl ond Type tt = (Keg), The fundamental stondpoint taken in each type of modeling is self-explanatory. Under euietcra Fig. 6 Influence of different assumptions of viscous damping in inelastic and hysteretic response Examples of member-by:member analysis of planar frames 23 the softening condition of K/Ke <1, their relative mognitudes of instantaneous damping Tie obviously in the order of (Type I) < (Type Ill) < (Type I). which of the three types provides @ rela~ ‘eppeats that the damping in the non-elostic le larger than modeled by Type | ond is clove to Type Ill. Most commonly used in the literature is the "tangent stiffness proportional damping” of Type IL This probably leads to the underestimation of the effect of damping after non- elastic action has token place. Application of the damping of Types | and Il to o }OF formulation can be straightforward. Also the Type Ill’ damping for o multi DOF system was exomined in the reference [164] 2y use of @ certain approximation of oscillation mode. Apart from the reliability of the various idealizations of viscous damping, it will be informative to illustrate the resulting dynamic response. One example calculation for this purpose is shown in Fig. 6, where © I-DOF degroding trilinear system is subjected to an earthquoke motion. In the case of Ce=1%, the discrepancies that result from the different exsumptions are seen to be relatively small. This is porticularly the case between Type | and Type Ill. On the other hand, serious differences feature the case of ce= 5% according 10 the higher intensity of excitation. The damping of a= 1 % may be used to approximate the inherent damping copacity of structural components or their subessemblages os noted previously. Thus the above lustration indicates not +0 crucial © choice for the separate formulation of damping in member-by-monber analysis. On the contrary, the results highlight the ue of considerable caution in the simpler modeling of the proportional damping. This is becouse the apparent damping capacity of total system may amount to Ce= 5% or more by the contribution of the rodiation damping, 7. EXAMPLES OF MEMBER-BY-MEMBER ANALYSIS OF PLANAR FRAMES, Along the lines so far reviewed, the member-by-member formulation of @ 8/C frame proceeds toward an elaborate microscopic description of its performance with full ‘ttention to local effects. The choice of the levels of complication to represent structural elements is eiuetal one in terms of the practice of lengthy comput well as the ease of programming. For both research and practical purposes, vs computer programs have been developed to implement the analytical procedures choven A few examples of such anclyses are shown in this chapter to demonstrate the advan- tages of member-by-member modeling. is token up, The purely moment-ressting frame fs one of the design exompies that are ineluded in the oppendicer of the ALJ (Architectural Institute of Jopon) Stendords for Structural Caleuletiont of Reinforced Concrete Struc= tures with Commentary = 1971, This may be regorded ot o typical R/C structure in Jopan, designed in occordance with the Jopanese Building Codes ond the Al) Ston~ dards;'© wotking stress design of sections wes performed by specifying @ slxmic coefficient of 0.2 at each story and by applying the stress anol lostcity of materials [normal concrete with design strength 6,>8,. This is entirely inconsistent With the findings from member-by-member analysis: 6, =8,>5, for the example A and 5,* 5,6, for the example 8. The latter patterns of story drift response con be Interpreted’ to reflect their modes of failure; the Is and 2nd stories in the example A. ‘and all the stories in the example B are coupled each other, thus forming 2-DOF and 1-DOF mechonisms of yield-hinge kinematics, respectively. On the contrary, the results of the two analyses appear to coincide well in the examples C ond D. Even though their modes of feilure are of a hybrid type, performance of the weak column portion dominates over the weak girder portion in bath stiffness and strength. The Findings in the example E showing an ideal 1-DOF failure mechanism and in the ‘exomple F having @ dominant weak girder portion opply correspondingly to the comments siven to the example B. ‘As demonstrated in the li ited illustrations, serious differences con generally ar between the response predicted by the member-by-nember model and the story-by-story ‘model. The two can be in reosonable agreement cnly when the structure is of @ weak column type. It hos been found thot such diserepencies are not attributable to an improper determination of sheor-beam parameters but to the common nature inherent in the series combination of yielding elements. In o shear beam system, yielded drift can separately develop at relatively weak element accompanying @ significant action to suppress the progress of yield at its upper elements, The trend of failure concent= ration at o single element becomes particularly notable when the base element is 32 Dynamic response of reinforced concrete buildings h oe f cm Fig. 9 Comparison of structural response resulting from the three different beans of menber-by-nenber, story-by-story and “equivalent” analyses ‘Simple modelings of overall behavior 33, vulnerable, Beoring the fatal drowbock of story-by-story formulation in mind and with a view to ‘obtaining a better prediction by use of other simple models, two different epprocches have been taken by Tokizewa [203] and by Aoyoma [254]. "Both the two studies intended to eccount for the development of inelastic drifts strongly coupled among Stories. In the former opprocch, menber-level properties were condensed by pres the mode of drift os suggested by yield-hinge kinemetics, Depending indi the failure mechanism, application of the modeling principle results in de "equivolent" 1-DOF or multi-DOF system of total structur beam model is included os its special case. A formulation similar to the equivalent 1-DOF system wos also developed by Pique [255] and by Sciidi ond Sozen [111], where structures were successfully assumed to deform according to their elastic fist mode shopes. On the other hand, Aoyama Introdiced @ modified shear beam model of YC frames. Consideration of the failure coupling among stories har led to vse of nonlinear bending springs along the axis of the story-by-story system in addition to the ordinary shear springs. Then the previous examples are again token up for illustrating the advantage of the equivalent formulation. According to the present principle, thelr degrees of freedom are one, two ond three for the examples 8, £ end F, the example A, nd the examples C ond D, respectively. The last 3-DOF cotes coincide with the conventional shear beam idealization and ore not reexamined. Associated with the bore-minimum reduc , the Ist ond 2nd stories in the example A anc all the stories in the examples B, E ond F constitute a single story-ossemblage element, Separation of story drifts within the story-ossemblage element follows the distribution of interstory heights under o constant translational angle. On the basis of the ipecified modes of displacement, generalized internal forces are introduced in the irdividuol elements, end their equa~ tons of motion ore derived os usual by means of the Lagrangean opprouch. The foregoing Fig. 9 has included the resulting response of story drifts for the examples A, B, E ond F. From this comparison, it con be immediately seen that the equivalent 1-DOF ond 2-DOF formulations provide more reasonable ogreenent with the results from the menber-by-member analysis than does the story-by-story formulation. The example ‘onalyses strongly suggest the capability of predicting the esienticl feotures of story drift response by vie of © very simple dynamic model, when the mechanism to produce inelastic deformations is considered carefully ond reflected foithfully upon the modeling. |v should be again emphosized that the ordinary stay-by-story formulation con become totally improper for frames with strong columns and weak girders, In the equivalent formulation, the method of incorporating inelastic ond hysteretic properties of structure is of simple os in the shear beam idealization, ond allows @ more satisfectory desctip~ tion of overall performance. Applicability of the equivalent formulation, of couse, requires extensive examinations for various frames of procticol interest. However, ity line of thinking is certainly more reasonable than the shear beam modeling, ane can be regarded as @ natural ‘extension of the conventionel concept. The current simplification may be justified only for short-period structures in which higher-mode components of displacement are not oppreciably excited by eorthquake motions. Moreover the failure mechanism under 0 dymomic excitotion may differ from the one exhibited under a static loading. This may 2A. Dynamie response of reinforced concrete buildings also vary according to @ miner change in the distribution mode of lateral forces or to

    You might also like