11 edition features a ti
'~;
a new addition to the chaptet on Random Vibration' ~ "~i~nse of strtu:tuni~.c.... . ,', modeled as 11 multi degreaof.freedom 5},st.em,5ubje<ted to sevenl random frees to a random modon at the base of the Stn!cture , ~ coverage of the concept of damping. indudi!'lg the evaluation of equivalent v!scovs
dim?ing ,
Method
Structural Dynamics
Theory and Computation Fourth Edition
Structural Dynamics
Theory and Computation Fourth Edition
Visit Chapman & Han's Internet Resource Center for ;nfonnarion on our new publications,
links (0 useful siles on the World Wide Web and an opporrunity to join our email mailing list. Point your browser to: http://www.chaphalI.com or
Mario Paz
Speed Scientific School University of Louisville Louisville, KY
A service of
IcDP
Chapman & Hall 2..{i Boundary Row London SEI 8HN England
Tnomas Nelson Australla 102 Dodds Street South Melbourne, 3205 Victoria, Australia International Thomson Edltores Campos Eliseos 385, Plso 7 CoL Polanco n560 Mexico D,P Mexico International Thomson Publishlng Asia 221 Henderson Road #0510 Henderson Building Singapore 0315
Honor your father ana your mother, as the ,"ord your God has commanded you, that you may long endure and that you may fare welL .. Exodus 20:12
TO THE MEMORY OF MY PARENTS
Benjamin Marr:an paz 8a1ma Misrl Paz
All rights reserved. No part of this boole covered by th~ copyright hereon may be reproduced or used ill any form Or by any meansgraphk, electronic, or mechanical, including photocopying, recording" taping" or information storage and retrieval systemswithout the written permission of the publisher.
2 3 4 5 6 7 B 9 10 XXX 01 00 99 98 97
Library of Congress
Paz.. Mario.
Struct~
Ca!aloging~in.l?'ublication
Data
~
4th <:<:1.
p. em. Indudes bibliograpt<jca! references and index. ISBN 0412~74613 L StructutaL dynamics. L Title. TA6S4.P39 1997 62U7<lc11
British Uhul")' Cataloguing in PubHcation Data availabJe
%3741$ elP
"$tructu.ral Dynam.ics: Tlu:my and Compulation" is intended to present terhl.ically accurate and authoritative information from highly regarded sou.~es. Thepublishcr; editors,. advisots, and :.:ontributors have made every reasonable effort to ensu.re the accuracy of the information. bi.(t cannot assume responsibility [or the accuracy of all information, ot for the :.:otJ..Sequences o[ its USi'.
To order this or any other Chapman & Hall book,. please contact International Thomson
'
n.~
n.
;fn<'c:...,c:.;;t.:(I(\""r13Uif\.l=iA''':\":~f.
CONTENTS
xxi
U U L3 1.4
LS 1.6
1.7 1.8 1.9
1.!O
U I
Newton's Law of Motion i 8 Free Body Diagram I 9 D' Alembert'S Plinciple I 10 Solution of the Differential Equation of Motion I II Frequency and Period I 13 Amplitude of Motion I 15 Undamped SingleDegreeofFreedom Systems Using COSMOS I 20 Summary I 22 Problems i 23
ix
3J
2.1 2.2
2.3
2.4 2.5 2.6 2.7
Viscous Damping i 31 Equation of Motion i 32 Critically Damped System I 33 Overdamped System / 34 Underdamped System / 35 Logarithmic Decrement I 37 Summary / 43 Problems i 44
5.5
5.6
5.7
5.8
Fourier Analys" I 139 Response to a Loading Represented by Fourier Series I 140 Fourier Coefficients for Piecewise Linear Functions / 143 Exponential Form of Fourier Series I 144 Discrete Fourier Analysis I ]45 Fast Fourier Transform I 148 Program 4Response in the Frequency Domain i 150 Summary I 156 Problems I 156
162
Undamped System: Harmonic Excitation / 47 Damped System: Hannonic Excitation / 50 3.2 Evaluation of Damping at Resonance I 58 3.3 Bandwidth Method (HalfPower) to Evaluate Damping, / 59 3.4 Energy Dissipated by Viscous Damping I 61 3.5 Equivalent Viscous Damping I 63 3.6 Response to Suppon Motion / 66 3.7 Force Transmitted to the Foundation I 76 3.8 Seismic InstrUments I 79 3.9 :3.10 Response of OneDegreeofFreedom System to Harmonic Loading Using COSMOS i 81 3.1l Summary I 88 Problems I 92 3.1
7
4
6.7
6.8 6.9 6.10
Principle of Virtonl Work I 162 GeneraLlzed Slngle~DegreeofFreedom SystemRigid Body I 164 Generalized SingleDegreeofFreedom SystemDistlibu~ed Elasticity I ,67 Shear Fo,ces and Bending Moments I 172 Generalized Equation of Motion for a Multistory Building / 177 Shape Function I 180 Rayleigh's Method i 185 Improved Rayleigh's Method! 192 Shear Walls I 195 Summary I 199 Problems I 200
205
96
Impulsive Loading and Duhamel's Integral I 96 Numerical Evaluation of Duhamel's IntegralUndamped System i 105 Numerical Evaluation of Duhamel's IntegralDamped System I 109 Response by Direct Integration I 11 0 Program 2Response by Direct Integration I 116 Program 3Response to Impulsive Excitation f 119 Response to General DynamiC Loading Using COSMOS I 124 Summary I 131
7.1 7.2
73
7,4
7.5
7.6 7.7
7.8
7.9
7.10
Nonlinear Single DegreeofFreedom Model I 206 Integrmion of [he Nonlinear Equation of Motion I 208 Constant Acceleration MeLood I 208 Linear Acceleration StepbyStep Method I 2l! The Newmark Beta Method I 2 14 Elastoplasric Behavior I 215 Algorithm for the StepbyStep Solution for Elastoplastic SingleDegreeofFreedom System I 217 Program SResponse for ElastopJastic Behavioa, System I 221 Nonlinear Stn.lClUra; Response Using COSMOS I 224 Summary I 228 Problems I 229
Contents
Contents
Xl
RESPONSE SPECTRA
233
8.1 8.2 8.3 8.4 8.5 8.6 8.7 8.8 8.9 8,10
Response Spectrum for Support Excitation! 237 Tripartite Response Spectra I 238 Response Spectra for Elastic Design I 241 Influence of Local Soil Conditions! 245 Response Spectra for Inelastic Systems I 247 Response Spectra for Inelastic Design ! 250 Program 6Seismic Response Spectra .I 257 Response Spectra Using COSMOS I 260 Summary I 265 Problems I 266
1 1.6 Combining Maximum Values of Modal Response I 334 11.7 Forced Motion of a Shear Building Using COSMOS .I 335 11.8 Summary J 346 Prob;ems ! 348 12
DAMPED MOTION OF SHEAR BUILDIt;GS
352
Equations for Damped Shear Building I 353 Uncoupled Damped Equations I 354 123 Conditions for Damping Uncoupling J 355 12.4 Program ll~Absolute Damping From Damping Ratios I 362 12,5 Summary I 364 Problems I 364
13
12.1 12,2
PART II
271
13.2
Stiffness Equations for the Shear Building J 272 PLl Effect on a Plane Shear Building! 275 Flexibility Equations for the Shear Buildip.g J 278 Relationship Between Stiffness and Flexibility Mame""s J 280 Program 7Modeling Structures as Shear Buildings I 281 Summa,), I 283 Problems ! 283 287
14
133
]3.4 13.5
13.6
Static Condensation Applied to Dynamic Problems J 370 Dynamic Condensation J 380 Modified Dynamic Condensation! 387 Program 12Reduction of the Dynamic Problem! 391 Summary I 393 Problems I 393
PART III
10
397
Natural Frequencies and Noonal Modes! 287 Orthogonality Property of Ute Noonal Modes I 294 Rayleigh's Quotient J 298 Program 8Natural Frequencies and Normal Modes I 300 Free Vibration of a Shear Building Using COSMOS! 301 Summary I 304 Problems J 305
310
14.1
14.2
14.3
14.4
II
ILl 11.2
Modal Superposition Ylethod i 310 Response of a Shear Building to Base Ylotlon I 317
Static Properties fnr a Beam Segment J 400 System Stiffness Matrix J 405 Inertial Properties.Lumped Mass ! 408 Inel1iai PropertiesConsistent Mass I 41O Damping Properties ! 414 External Loads ! 414 Geometric Stiffness ! 416 Equations of Motion I 420 Element Forces at Nodal Coordinates I 427 Program 13Modeling Structures as Beams J 430 Dynamic Analysis of Beams Using COSYlOS I 433
XII
Contents
Conten:s
xlii
14.12
15
15.1
15,2
15.3
15.4
15.5
15,6
Eleme"' Sliffness Matrix for Axial Effects I 443 Element Mass Motrlx for Axial Effects I 444 Coordinate Transformation I 449 Prooram 14Modeiinc:r Structures as Plane Frames! 458 ~ ~ Dynamic Analysis of Frumes Using COSMOS I 460 Summary / 465 Problems I 466
469
Program 17Modeling Structures as Plane Trusses f 520 Stiffness and Mass Matrices for Space Trusses I 522 Equation of Motion for Space Trusses I 525 Program l8Modeling Str'Jctures as Space Trusses I 526 Dynamic Analysis of Tmsses Using COSMOS f 528 Summary I 536 Problems I 536
19
19, I
16
16.1
16,2
16.3 16,4
16.5 16.6
16.7
16,8
16.9
Local and Global Coordinate Systems f 470 Torsional Effects I 471 Stiffness :vIatrix for a Grid Element I 472 Consistent Mass Matrix for a Grid Element I 473 Lumped Mass Matrix for a Grid Element f 473 Transformation of Coordinates I 474 Program ISModeling Structures as Grid Frames I 480 Dynamic Analysis of Grids Using COSMOS f 483 Summary I 487 Problems f 488
19.2
19.3
Plane Elasticity Problems f 539 19, J.l Triangular Plate Element for Plane EJastlcitv Problems I 540 19, L2 Library of Plane ElosticilY Elements (2D Eleoents) I 552 Plute Bending I 555 j9.2.1 Rectangular Finite EleT:1ent for Plate Bending I 556 19.2.2 COSMOS Library of Plate and Shell Elements f 565 Summary I 573 Problems I 575
20
17
17.6
17,7 17 ,8
17.9
Element Stiffness Matrix f 49l Element Mass :vIatrix f 493 Element Damping :vIatrix I 494 Transformation of Coordinates f 494 Diffetenrial Equation of Motion f 503 Dynamic Response I 504 Ptogtam 16Modeling Structures as Space Frames I 504 Dynamic Response of ThreeDimensional Frames Using COSMOS f 507 Summary I 510 Problems I 5\0
511
202 20.3
18
18,1
.:;'11
Incremental Equations of Motion I 578 The WIlson8 Method I 579 Algorithm for StepbyStep Solution of a Linear System Using the Wilson8 Method f 582 20.31 Initialization f 582 20.3.2 For Each Time Step f 582 20A Program 19Response by Step Integration f 587 205 Newmark Beta Method f 588 20,6 Elastoplastic Behavior of Framed Structures f 589 20.7 :vIembet Stiffness Matrix I 590 20.8 Membet :vIass Matrix I 593 20.9 Rotation of Plastic Hinges I 595 20.10 Calculation of Member Ductility Ralio ! 596 20.11 TimeHistory Response of Mul(idegreeofPreedom Syster.rs Using COSMOS f 597 20.12 Summary I 602 Probler.rs f 604
20,1
xiv
Contents
Contents
xv
PART IV
2!
607
609
23.7 23.8
2LI 21.2 2 L3
21,4 2 L5
21.6 21.7
Flexural Vibration of Uniform Beams J 610 Solution of the Equation of MociOf: in Free Vibration J 6 i ! Nftrura~ Frequencies and Mode Shapes for Uniform Beams / 613 21.3.1 Both Ends Simply Supported I 613 21.3.2 Both Ends Free (Free Beam) ! 617 21.3.3 Both Ends Fixed I 6 i 8 2L3.4 One End Fixed and [he other End Free (Cantilever Beam) I 620 21.3.5 One End Fixed and the other End Simply S"pponed I 622 Orthogonallty Condition Be[weer. Normal Modes I 622 Forced Vibration of Beams J 624 Dynamic Stresses in Beams I 630 Summary I 632 Problems I 633
635
13.9 23.10
23.11 23.12 23 13
Correlation I 662 The Fourier Transform I 666 Spectral AnalysIS I 668 Spectral Density FU .. otlon I 672 NarrowBand and WideBand Random Processes I 675 Response::o Random ExcI(3[ion: SingleDegreeof~Freedom System I 679 Response to R2.ndom Excitation: Mulr!pleDegree~ofFreedom SyStem! 685 Random Vibration Using COSMOS! 696 S ummay ! 700
PART VI
24
EAHTHQUAKE ENG:NEERING
705
t:NIFORM BUILDIC'lC CODE 1994; EQUIVALENT STATIC LATERAL FORCE METHOD 707
22
22.4
22.5 22.6 22.7 22.8
Dynamic Matrix for Flexural Effects I 636 Dynamic Matrix for Axial Effects I 638 Dynamic Matrix for Torsional Effec[S I 641 Beam Flexure Including AxialForce Effect J 642 Power Series Expansion of ::,e Dynamic Matrix for Flexural Effects I 646 Power Series Expans~on of the Dynamic MatrIX for Axial and for Torsional Effects j 648 Power Series Expansion of the Dynamic Matrix Including the Effect of Axial Forces I 649 Summary I 650
PAHT V
23
RANDOM VIBRATION
653
651
RA,'fDOM VIBRATION
23.1
23.2
23.3
Statlsrical Description of Random Functions I 654 Probabiliry Dens;ry Function I 657 The Normal Distribution I 659
f
r:.r.;n
Earlhquake Ground Motion! 708 Ec;:..tivalenr Seismic Lateral Force J 7:2 EanhquakeResiSlam 8esign Methods I 712 Static Lateral Force Mcrhod I 713 145 Distribution of Lateral Forces j 718 24.6 Story Shear Force I 718 24.7 Horizontal Torsional :Vloment I 719 24.8 Oveflureing Moment I 720 24.9 Srory 8rift Limitm'on I 720 24.10 PDelta Effect (PLl) I 721 24.11 Diaphragm Design Force I 723 24.12 Program 23 UBC94 Equivalent Static Lateral Force Method I 732 24.13 Simplified Three Dimensional Earthquake Resistant Design of B uilcings I 739 24.l3.l Ylodeiing the Building I 739 24.13.2 Transfomla[jon of Stiffness Coefficients j 740 2413.3 Center of Rigidity I 742 24.13.4 Story Eccentricity I 743 24.13.5 Rotational Stiffness! 744 24.13.6 Fundamental Period I 745 24.13.7 Seismic Factors I 745 24. I 3.8 Base Shear Force! 746 24.13.9 Equivalent Lateral Seismic Forces! 746 24.13.10 Overturning Moments I 747
xvi
Contents
24.13.10 Story Shear Force / 747 24.13.12 Torsional Moments / 747 24.13.13 Story Drift and Lateral Displacements / 748 24.13.14 Forces and Moments on Structural Elements I 749 24.13.15 Computer Program / 750 24.14 Equivalent Static Lateral Froce Method Using COSMOS / 756 24.15 Summary I 761 25 UN1FORIvl BUlLDl.t'!G CODE 1994: DYNAMIC :vlETHOD
25.1
766
25.2
Modal Seismic Response of Buildings / 766 25. Ll Modal Equation and Participation Factor / 767 25.1.2 Modal Shear Force I 768 25. L3 Effective Modal Weight / 770 25.1.4 Modal Lateral Forces / 771 25.1.5 Modal Displacements / 771 25.1.6 Modal Drift I 772 25.1.7 Modal Overturning Moment / 772 25.1.8 Modal Torsional Moment / 772 Total Design Values I 773 Provisions of UBC94: Dynamic Method I 774 Scaling of Results I 776 Program 24UBC 1994 Dynamic Lateral Force Method i 783 Summary i 787 Problems I 788
APPEI"1DlCES I 789 Appendix I: Answers to Problems in Part I I 79 I Appendix II: Computer Programs I 80 I Appendix Ill: Organization and their Acronyms / 804 Glossary i 807 Selected Bibliograpby I 815 Index / 819 Diskette Order Fonn I 825
The basic Structure of the three previous editions is maintained in this fourth edition, although numerous revisions and additions have been introduced. A new chapter to serve as an introduction for the dynamic analysis of structures using the Finite Element Method has been incorporated in Part 1lI, Structures Modeled as Discrete Multidegree,ofFreedom Systems. The chapter on Random Vibration has been extended to include the response of structures modeled as multidegreeoffreedom systems, subjected to several random forces or to a random motion at the base of ~he structure. The concept of damping ~nc1uding the evaluation of equivalent viscous damping is thoroughly discussed. The constant acceleratlon method to determine the response of nonlinear dynamic systems is presented in addition to the Hnear acceleration method presented in past editions, Chapter 8, Response Spectra now includes the development of seismic response spectra with consideration of local soil conditions at the site of the structure. The secondary effect resulting from the lateral displacements of the building, commonly known as the P!l effect. is explicitly considered through the calculation of the geometric stiffness matrix. Finally, a greater number of iHustrative examples have been incorporated in the various chapters of the Dook using the educational computer programs developed by the author or tbe professional program COSMOS.
xvm
Preface
xix
The use of COSMOS for the analysis and solution of structural dynamics problems is introduced in this new edition. The COSMOS program was selected from among the various professional programs available because It has the capabiiity of solving complex problems in structures, as well as in other engineering fields such as Heat Transfer, Fluid F:ow, and Electromagnetic Phenomena. COSMOS includes routines for Structural Analysis. Static, or Dynamks with linear or nonlinear behavior (materlu! nonUnearlty or large displacements), and can be used most efficiently in the microcomputer. The larger version of COSMOS has the capacity for the analysis of structures modeled up to 64,000 nodes. This fourth edition uses an introductory version thot has a capabiiity limited to 50 nodes or 50 elements. This version is induded in the supplement, STRUCTURAL DYNAMICS USING COSMOS '. The sets of educational programs in Structural Dynamics and Earthquake Engineering that accompanied the third edition have now been ex.tended and updated. These sets include programs to determine the response in the time or frequency domain using the foB (Fast Fourier Transform) of structures modeled as a single oscillator. Also included is a program to deler:nine the response of an inelastic system with e!astopiastic behavior and a program for the development of seismic response spectral charts. A set of seven computer programs is included for modeling structures as twodimensional and threedimensional frames and trusses. Other programs. incorporating modal superposition or a stepby~step timehistory solution, are provided for calculation of the responses to forces or motions exciting the structure. In addition. in this fourth edition, a new program is provided to detennine the response of singleor muItidegreeoffreedom systems subjected to random excitations. The computer programs for earthquakeresistant deSign have been updated using the latest published seismic codes. The book IS organized into six parts. Part I deals with st:1lctures modeled as singledegreeoffreedom systems. It introduces basic concepts and presents methods for the solution of such dynamic systems, Part II introduces conccpts and methodology for solving multidegreeoffreedom systems through the use of structures modeled as shear buildings. Part III describes methods for the dynamic analysis of skeletal structures (beams, frames. and trusses) and of continuous structures such as plates and shells modeled as discrete systems with many degrees of freedom, Part IV presents the mathematical solution for some simple structures modeled as systems with distributed properties. thus baving an infinite number of degrees of freedom. Part V introduces tbe reader to the fascinating topic of random vibrations, which is now extended to multidegree~offreedom systems, Finally, Part VI presents the current topic of earthquake engineering with applkations for the design of earthquakeresistant
; A cQnVen:erH form lo order this supplemelH is provided in lhe back of the book.
buildings following the provisions of the Cnifonn Building Code In use in the United States. There is a detailed presentation of the seismic analysis of buildings modeled as threedimensional structures with two independent hori~ zontal motions and one rotational motion about a vertical axis for each story of the building. A computer program for the implementation of this simplified method for seismic analys~s of buildings is jncluded in the set of educational programs. Scientific knowledge may be presented from a general allencompassing theory from which particular or simple situations are obtained by introducing restricting conditions. Alternativeiy, the presentation may begin by considering particular or simple situations that are progressively extended. The author has adopted this latter approach in which the presentation begins with particular or simple cases that are extended to more general and complex situations. Funhennore, the author believes that a combination of knowledge of applied mathematics, theory of structures. and the use of computer programs is needed today for the succe$.').fu! profes;;;iollal pfflctice of engineering. To provide the reader with such a combination of knowledge has been the primary objective of this book. The reader is encouraged to inform the author on the extent to which this objective has been fulfined. Many of my students. colieagues, and practidng professionals have suggested improvements, identified typographical errorS. and recommended additional topics for inclusion. Ali these suggestions were carefully considered and have been included in this fourth edition whenever possible. I was fortunate to have received valuable assistance and insight from many individuals to whom I wish to express my appreciation. I am grateful to Jeffrey S. Janover, a consulting engineer from New Jersey, who shared his expertise in the implementation of professional computer programs for the solution of complex engineering probtems. I appreciate the discussions and comments offered by my colleagues Drs. Michael A. Cassaro and Julius Wong who helped me in refining my exposition. I am also grateful to my friend Dr, Farzad Naeim who has coIlabornted with me on Seismic Response Spe<:rra in the International Handbook of Earthquake Engineering: Codes, Programs and Examples (Paz. 1994) of which I am the editor. I have incorporated some of the material from the Handbook in updating the chapter on Response Spectra, I also wish to acknowledge Dr. Luis E. Suarez from the University of Puerto Rico in Mayaguez, who provided me wiLi copies of his work in random vibrations and of his class notes on the Finite Element Method. It is with great satisfaction that I acknowledge the help re<:eived from four of my fonner students: Christopher Biles, who carefully studied and commented on Chapter 23, Random Vibrations, as he worked on his Masters' theSIS on that subject; Mahomet Sharif for providing me with actual cases of random vibration problems selected from his professional practice; Zair HiJlal, who made skillful use of the computer in preparing some of the new figures
xx
in the book; and Cleryl Hoskins who most carefully checked the solution of the problems for some chapters of the book A special acknowledgement of gratitude is extended to Dr. Edwin A. Tuttle, emeritus professor of educatlon, who provided many suggestions that helped to improve the clarity of my presentation. I aiso wish to express my sincere gratitude to my friend Jack Bension for his professional help in editing the revised sections of the book. My thanks also go to Ms. Debbie Jones for her competent typing sieills in the revisions. To those people whom I recognized in the prefaces to the previous editions for their help, I again express my wholehearted appreciation. To my wife Jean a special thanks for carefully checking the structure of the book and for most graciously aUowing me time to prepare this new edition, particularlY during sev.eral "working vacations." As with the third edition, this volume is dedicated to the everlasting memory of my parents.
March, 1997
Natural phenomena and human activities impose forces of time~dependent variability on structures as simple as a concrete beam or a steel pile, or as complex as a multistory building or a nuclear power plant constructed from different materiais. Analysis and design of such Stn.lctures subjected to dynawic loads involve consideration of timedependent inertial forces. The res islance to displacemenl exhibited by a struclure may include forces which are functions of the displacement and the velocity. As a consequence, the governing equations of motion of the dynamic system are generaUy nonlinear partial differential equations which are extremely difficult to solve in mathematical terms. Nevertheless. recent developments in the field of structural dynamks enable such analYSls and design to be accomplished in a practical and efficient manner. This work is facilitated through the use of simplifying assumptions and mathematical models, and of matrix methods and modem computarionai techniques, In the process of teaching courses on the SUbject of structural dynamics, the author came to the realization thai there was a definite need for a text which would be suitable for the advanced undergr'J.duar:e or the beginning graduate engineering stJdent being introduced to this subject. The author is familiar with tJ}e existence of several excellent lexts of an advanced nature but genxxi
xxli
eraUy these texts are. in his view, beyond the expected comprehension of the student Consequently, it was his principal aim in writir.g this book to incorporate modem methods of analysis and lechniques adaptable to computer programming in a manner as clear and easy as the subject permits. He felt that computer programs should be induded in the book in order to assist the student in the application of modem methods associated with computer usage. In addition, the author hopes that thIs text will serve the practicing engineer for purposes of selfstudy and as a reference source. In writing this text. the author also had in mind the use of the book as a possible source for research topics in structurdl dynamics for students working toward an advanced degree in engineering who are required [0 write a thesis, At Speed Scientific School, University of Louisville, most engineering students complete a fifth year of study with n thesis requirement leading to n Master in Engineering degree. The author'S experience as a thesis ndvisor Jeads him to believe that this book may weB serve the students in their senrch and selection of topics in subjects cunently under investigation in structural dynamics. Should the text fulfill the expectations of the nuthor in some measure, par~ ticuiariy the elucidation of this subject, he will [hen feel rewarded for his efforts in the preparation and development of the material in this book.
MARIO PAZ
PART I
Structures Modeled as a SingleDegreeofFreedom System
December, 1979
1
Undamped Single DegreeofFreedom System
It is not always possible to obtain rigorous mathematical solutions for engineering problems, In bct, analYlical solutions can be obtaine<;i only fo: certain simplified situations. For problems involving complex malerial properties, loading, and boundary conditions, the engineer introduces assumptions and idealizalions deemed necessary to make [he problem mathematically manageable but still capable of providing sufficiently approximate solutions and sat
4
lsfactory results from the potot of view of safety and economy. The link
between the real physical system and the mathemaricaHy feasible solutloil js provided by the mathematical model which is the symbolic designation for the substitute ideaHzed system including a.1I the assumptions imposed on the physi~ cal prublem.
1,1
DEGREES OF FREEDOM
In structural dynamiCs the number of independent coordinates necessary to specify lhe configuration or position of 2. system at any time is referred to as the number of degrees of freedom. If:, ge:1eral, a continuous structure has an
3
Undamped
Sjng!e~Oegree"of~Freedom
System
~F~t=~ ,
,.,
llwnlllHl 1ill1lj
~
pm
lfJ' Ft._==~;
'"
n
onedegree~ofMfreedom
on the behavior of the real physical system. Nevertheless, from a practical point of view, the information acquired from the analysis of the mathematicai model may very well be sufficient for an adequate understanding of the dynamic behavior of the physical system, including design and safety requirements.
systems,
infinite number of degrees offreedom, Nevertheless. the process of idealization or selection of an appropriate mathematical model permits the reduction in the number of degrees of freedom to a discrete number and in some cases to just a single degree of freedom. Figure I.! shows some examples of structures that may be represented for dynamic analysis as onedegreeoffreedom systems, that is. structures modeled as systems with a single displacement coordlnate. These onedegree~of~freedom systems may be described conveniently by the mathematical mode} shown in Fig. 1,2 which has the foHowing elements: (1) a mass element m representing the mass and inertial characteristic of the structure; (2) a spring element k representing the eiastic restOing force and potential energy storage of the stnlcture~ (3) a damping element c representing the frictional characteristics and energy losses of the structure; and (4) an excitation force F(t) representing the external forces acting on the structural system. The force F(t) is written this way to indicate that it is a function of time. In adopting the mathematical model shown in Fig. 1.2, it is assumed that each element in the system represents a Single property; that is, the mass m represents only the property of inertia and not elasticity or energy diSSIpation, whereas [he spring k represents exclusively elasticity and not inertia or energy dissipation. FinaIly, the damper c only dissipates energy. The reader certainly realizes thar such "pure" elements do not exist 1n our physical world and that mathematical models are only conceptual idealizations of real structures. As such, mathematical models may provide complete and accurate knowledge of the behavior of the model itself, but only limited or approximate information
>v
tj~J/
fa)
Ib)
one~degreeMof~freedom
,
1
'.
~~,
Fig, 1.4 Force displacement relation. (a) Hard spring, (b) Linear spring. (c) Soft spring.
Structures Modeled as a
Sing;eDegree~ofFfeedom
System'
The curve labeled (a) in Fjg, 1.4 represents the behavior of a "hard spring," in which the force required to produce a given displacement becomes increasingly greater as the spring is defonned. The second spring (b) is designated a linear spring because the deformation is directly proportional to the force and the graphicaJ representation of its characteristic is a straight 1ine. The constant of proportionality between the force and displacement (slope of line (b)] of a linear spring is referred to as the spn'ng constalll, usually designated by the lelter k" Consequently, we may write the foUowing relatlon between force and displacement for a linear spring.
F,=ky
(l.l)
For two springs in parallel the total force requJred to produce a relative displacement of their ends of one unit is equal to the sum of their spring constants. This total (orce is by definition the equivalent spring constant k( and (s given by
(12)
2: k,
;"'1
(1.3)
A spring with characteristics shown by curve (c) in Fjg. 1.4 is known as a "soft spring," For such a spring the incremental force required to produce additional deformation decreases as the spring deformation increases. Undoubtedly, the reader is aware from his previous exposure to mathematical modeling of physical systems that the linear spring is tne simplest type to manage analytically. It should not come as a surprise [0 learn that most of !be technical literature on structural dynamics deals wHh mode's using linear springs_ In other words, either because the elastic characterisrics of the structural system are, in fact, essentially linear, or simply because of analytical expediency, it is usually assumed that the forcedeformation pmperties of the system are linear. In support of [his practice, it should be noted that in many cases the displacements produced in the. structure by the action of external forces or disturbances are small in magnitude (Zone E in Fig. 1.4), thus rendering lbe linear approximation close to tbe actual structural behavior.
For two springs assembled in series as s.hown in Fig_ 1.5(b), the force P produces the relative displacements in (he springs
LlYI
and
Lj.,
1.
=~ k2
Then. (he total displacement y of the free end of the spring assembly is equal to y = LiYI ;. Ll)'2. or substituting LiYI and LlY2,
(1.4)
Consequently, the force necessary 1.0 produce one unit dis.placement {equivalent spring constant) is given by
1.3
Sometimes it 1s necessary to determine the equivalent spring constant for a system in which two or more springs are arranged in parallel as shown in Fig. 1.5(.) or in series as in Fig. L5(b).
k=P , y
Substituting y from this last relation lnto eq, (1.4), we may conveniently express the reciprocal value of :.he equivalent spring constant as
1 +_.
k,
k,
(1.5)
In general for n springs in series the equivalent spring constant m.ay be oblained from
( 1.6)
Fig. 1.5 Combination of springs. (3) Springs in parallel. (b) Springs in series.
k,
k,
Struc\ures Modeled as a
Slng\e~OegreeofFreedom
System
displaceme~t
or
velocity of the mass In at any time f. for a given set of initial conditions at time f = O. The analytical relation between ~he displacement. y, and time, (, is given by Newton's Second Law of Motion, which in mod~rn notation may be , expressed as
in which the mass moment of inertia 10 and the moment of the forces Mo are determined with respeet to the fixed axis of rotation. The general motioo of a rigid body is described by two vector equations, one expressing the relation between tbe forces and the acceleration of the mass center and anorher relatioothe moments of the forces and the angular motion of' the body. This Ias~ equation expressed in its scalar components is rather complicated, but seldom needed in structural dynamics.
F=ma
(1.7)
where F is the resultant force acting on a particle of mass m and a is its resultant tlcce1eration_ The reader should recognize rhat eq. (1.7) is a vector relation and as such it can be written in equivalent fonn in terms of irs components along the coordinate axes .x, y, and z, namely
(1.8a)
(I,8b) (L8c)
The acceleration 1S defined as the second derivative of the position vector with respect to time; it foHows that eqs. (1.8) are indeed differentiar equations. The reader should also be reminded that these equations as stated by Newton are directly applicable only to bodies idealized as panides, that is, bodies that possess mass but no volume, However. as is proved in elementary mechanics, Newton's Law of Motion is also directly applicable to bodies of finite dimensions undergoing translatory motion. For plane motion of a rigid body that is symmetric with respect to the reference piane of motion (xy plane), Newton's Law of Motion yields the following equations:
(LlO)
L F, = m(ad,
(1.93) (L9b)
where the spring force acting in the negative direction has a minus sign, and where the acceleration has been indicated by y. Ir. this notation, double overdots denote the second derivative with respect to time and obviously a single overdot denotes the first derivative with respect to time. that is, the velocity.
(J ,9c)
In the above equations (aG);~ and (aG)" are the acceleration components, along the x and v axes, of the center of mass G of the body; Ct is the angular accelerati{}J;; lr; is the mass moment of inertia of the body with respect to an axis through G, the center of mass; and 'kMc; is the sum of the moments of all the forces acting on the body with respect to an axis through G, perpendicular to the xy plane, Equations (1.9) are certainly a}so applicable to the motion of a rigid body in pure rotation about a fixed axis. For this particular type of plane motion, altematively, eq. (L9c) may be replaced by
( 1.9d)
,.)
(0)
'c)
Fig. 1.6 Alternate free body diagrams: (a) Singie degreeoffreedom system. {b) Show~ jng only external forces. (c) Showing external and inertial forces,
Undamped
Single~Deg(eeotFreedom
System
11
1.6
O'ALEMBERT'S PRINCIPLE
An alternative approach to obtain eq. (LlO) is to make use of D'Alembert's Principle which states that a system may be sei in a state of dynamic equilibrium by adding to the external forces a fictitious force thai is commonly known as the inertial force. Figure L6(c) shows the FED with inclusion of the inert:al force my. This force is equal (0 the mass multiplied by the acceleration, and should always be directed negatively with respect to the corresponding coordinate. The application of D'Alembert's Prineiple allows us to use equations of equilibrium in obtainJng the equation of motion. For example, in Fig. 1.6(c), the summation of forces in the y direclion gives directly
mji + J . :y = 0
(0)
w
'<I
(LlI)
N
which obviously is equivalent to eg. (1.10). The use of D' Alembert's Principle in this case appears to be triviaL This will not be the case for a more complex problem, in which the application of D'Alembert's Principle, in conjunction with the Principle of Virtual Work, constitutes a powerful tool of analysis. As wUi be explained later, the Principle of Virtual Work is directly applicable [0 any system in equHibriuffi. ft follows then that this principle may also be applied to the solution of dynamic problems, provided that D' Alembert' s Principle is used to establish lhe dynamic equilibrium of the system.
fd
Fig. 1.7 Two representations of the simple oscillator and corresponding free body
diagrams. or
my+ky=O
which is identical to eq. (a).
Exam ple 1.1. Show that the same differential equat:on is obtained for a springsupported body moving verlically as for the same body vibrating along a horizontal axis, as shown in Figs. 1.7(0) and 1.7(b).
SOlution: The FEDs for these two representations of the simple oscillator 1.7(c) and 1.7(e), where the inertial forces are inchIded. are shown in Equating to zero the sum of the forces in Fig. L7(c), we obtain
m'j+ i..:y;;; 0
(a)
1.7
When the body in Fig. 1.7(d) is in the static equilibrium posit;on, lhe spring is stretched Yo units and exerts a force kyo = W upward on the body, where W is the weight of the body. 'When {he body is displaced a distance y downward from this position of equilibrium the magnitude of the spring force is given W + ky, since kyo = W. Using this result and applying by F. =. k(yo + y) or it to the body in Fig. L7(e), we obtain from Newton's Second Law of Motion
The next step toward our objective is to Dnd the solution of the differenlial equalion {Lll). \Ve should again adopt a systematic approach and proceed to first classify this differential equation. Since the dependeot variable y and its second derivative y appear in the first degree in eq. (1.11), this equation is classified as linear and of second orGeL The facl that the coefficients of y and y (k and m, respectively) are conslants and the second member (rigbt~hand side) of the equalion is zero further classifies the equation as homogeneous with conslant coemcien~s. We shodd recall, probably with a certain degree of satisfaction, that a general procedure exists for the solution of linear differential equations (homogeneous or nonhomogeneous) of any ordeL For this simple, second~order djfferentja~ equation we may proceed direclly by assuming a trial solution given by
y=A cos WI
 (W+Ay)
+ W=my
(b)
( 1.12)
12
13
or
y=B sin
W!
(Ll3)
where A and B are constantS depending on the initia~ion of the motion whlle w is a quantity denoting a physical characteristic of the systert: as it wiE be shown next. The substitution of eq. (1.12) into eq. (1.1 j) gives
( In,,}
+ k) A cos
w!;;;:;
(1.14)
Next, we should determine the constants of integration ,4, and B. These constants are determined from known values for the motion of the system which almost invariably are the displacement Yo and the velocity Va at the jojtiation of the motion. that is, at time t = O. Tbese two conditions are referred to as initial conditionJ, and the problem of solving the differential equation for the lnltial conditions is caUed an initial value problem. After substituting. for l = 0, y ~ Yo, and }' = Uo into eqs. (Ll?) and (!.IS) we find that Yo=A
(LJ9a)
be equal to z.ero or
(1.15)
Vo=
Bw
(l.!9b)
Finally, the substitution of A and B from eqs. (1.l9) into eq. (Ll?) gives The reader should verify that eq. (1. 13) is also a solution of the differential equation (LlI), with OJ also satisfying eg. (US). The positive roO! of eq. (LlS),
(1.160)
y=yocos wt+sm wt
w
Vo .
( 1.20)
is known as the natural frequency of the system for reasons that will soon be apparent. Equation (1.16a) may be expressed in tem.s of the static displacement resulting from the weight W = mg. The substitution into eq. (1. i6) of In = Wlg results in
which is the expression of the displacement y of the simple oscillator as a function of the time variable 1; thus we have accomplished our objective of describing the motion of the simple undamped oscillator mOdeling structures with a single degree of freedom.
w=
Hence
(kg
yW
i
w=;V YSl
'8
(LJ6b)
wT= 211' or
(1.21 )
where Y$! = W Ik is the static displacement due to the weight W. Since eilher eq, (Ll2) or eq. (Ll3) is a solution of eq, (l.ll), and since this differentiaI equation 1$ linear, the superposition of these two solutions, indicated by eg, (1.l7) below, is also a solution. FurthemlOre, eq. (Ll?), having two constants 0: integration, A and B, is, in fact, the general solution for this second~order differential equation,
y=A cos wtB SIn wI (1.17)
T= 21T
W
The expression for velocity, ):, is found simply by differentiating eq. (1, 17) with respect to time; that is,
The period is usually expressed in seconds per cycle or simply in seconds, whh the tacit understanding,that it is '<per cycle." The value reciprocal to the period is the "aluralfrequency f. From eg. (l.21)
y=
Aw sin wi
+ Bw cos
wt
(US)
f==
21T
( 1.22)
:"".'
",:i;
14
Undamped
Single~Oegree,ofFreedom
System
15
l~l;nll
M~T
10.691~lit\,
114 in.
. 4 (JIl)
3 X 30
10'
and
I 1 =+60 ;0.69
k, = 9.07 lb/in
Tne natural frequency f is usually expressed in hertz or cycles per second (cps). Because the quantity w d!ffers from the natural frequency f only by the constant factor, 2'lT, w also is sometimes referred to as the natural frequency. To distinguish between these two expressjons for natural frequency, w may be called the circular or an8ular natural frequency. Most often, the distinction j5 understood from the context or from the units. The natural frequency f is measured in cps (IS indicated. while the circular frequency w should be given in radians per second (radfsec). Example 1.2. Detennine the natural frequency of the system shown in Fig. 1.8 consisting of a weight of W = 50,7 lb attached to a horizontal camilever beam through the coil spring ka, The cantilever beam has a thickness t ~ in, a width b = 1 in modulus of eiasticity E= 30 X 106 psi, and a length I. = ! 2.5 in. The coil spring has a stiffness, k, = 10.69 (Ib lin).
Solution: The deflection L1 at the free end of a uniform cartilever beam acted UpOrl by a static force P at the free end is given by
The natural frequency for this system is then given by eq. (1.16a) as
w=Jkt:!m (m
Vligandg=386 in/sec 2)
w = /"9ih X 386/50,7
w
= 831 rad/sec
1.32 cps
PI.' 3EI
(123)
or
Tne corresponding spring constant k, is then
P
j
J
(1.24)
k!=;:;;o)
3EI
I.
c = j y~ + (uo}w)J.
ran
(1.25) (1.26)
where I ?7.bt (for rectangular section). Now, the cantilever and the coil spring of this system are connected as springs in series. Consequently, the equivalent spring constant as given from eq. (1.5) is
y,
and
tan fl=y
uo}w
( 1.27)
16
Undamped
Sing!lDegreeof~Freedom
System
17
Ck'.
I
I
~
(1.28)
The simplest way to obtain eq. (1.23) or eq. (1.24) is: to :nultip;y and divide eg, (1.20) by the factor C defined in 0'1. (L25) and to defme a (or (3) by eq, (1.26) [or 0'1, (1.27)J, Thus
Example 1.3. Consider the frame shown in Fig. 1. 11 (a). ThIS is a rigid steel frame to which a horizontal dynamic force is applied at the upper level. As part of the overall structurai design it is required to determine the natural frequency of the frame. Two assumptions are made: (1) the masses of the columns and walls are negJigible; and (2) the horizontal members are sufficiently rigid to prevent rotation at the tops of the columns. These assumptions are not mandatory for the solution of the problem, but they serve to simplify the analysis. Under lhese conditions, the frame may be modeled by the springmass system shown in Fig. 1. 11 (b).
Solution:
Y=
I Yo wi + C\CCOS
I
1o
W=200 X 25
1 82.5 in
4
= 5000 lb
E = 30 X 10'psi
k
k 12 X 30 X 10' X 165 (15x 12)' 1O,185Ibli"
sin ex = Yo . C
and
(L29)
cos
0: =
uo!w
(1.30)
Note: A unit displflcement oftlte top of afixed column requires aforce equal 12EUI}, Therefore, the Mruralfrequency from eqs, (U6) and (1.22) is
The substitution of eqs, (1,29) and (1.30) into eg, (1.28) gives
y :;;;: C (sin a cos wI
+ cos
0'
sin wt)
(1.31)
_1_1 10,185
21T~
386
(Ans_)
tv
4,46 cps
5000
The expression within the parentheses of eq. (1.3 J) is identical to sin (wt + a), which yields eq. (1.23), Similarly, the reader should verify, without difficulty, the fo:m of solution given by eq, (1.24). The value of C in eq_ (Ll3) [or eq, (L24)] is referred to as the ampliLOde of motion and the angle 0' (or f3) as the phase The solution for the motioo of the simple oscillator is shown graphically in Fig, 1. [0.
y
L '" 15'
we x 24
~
I
.L

lj t::
m
r'
tbl
FII)
11///////5J~
Fig. 1.11 Onedegree~of~freedom frar:le and corresponding mathematical model for Example 1.3.
18
U,'ldamped
Sing~eDegTeeof~Freedom
System
19
F
k
...........
",ml,}"""
~a)
,,'
~b)
Fig. 1.12 (a) Water lower tank of Example 1.4; (b) Mathemutical modeL
'Ir
Example 1.4. The elevated water tower tank with capacity for 5000 gallons of water shown in Fig. 1.12(a) has a natural period in laterai vibration of 1.0 sec when empty_ When the tank is fuiI of water. its period lengthens to 2.2 sec. Determine the lateral stiffness k of the tower and [he weight \.11 of the tank. Neglect the mass of the s~lpporting columns (one gallon of water weighs approximately 8.34 Ib) Solution: In its iatem! motion, the water tower is modeled by [he simple oscillator shown in Fig. L12(b) in which k is the lateral stiffness of tile tower and m is the vibrating mass of the tank. fa) Natural frequency We (tank empty):
1'1= 10,860 ib
Subs~ituling
(Ans.)
:11
tl
:~
and
k= lila Ibiin
(Ans.)
!: r ,
(a)
(b) Natural frequency Wy: (tank full of water) Weight of water W",:
Example 1.5. The steel frame shown in Fig l.13(a) is fixed at the base and has a rigid top that weighs 1000 lb. Experimentally) it has been found that its naturai period in iateral vibration, is equal to 1/10 of a second. It IS required to shorten or lengthen its period by 20% by adding weight or strengthening the columns. Determine needed additional weight or additional stiffness (neglect the weight of the COlumns).
Solurion: The frame is modeled by the springmass system shown in Fig, 1.13(b). Its stiffness is calculated from
w, ~ 5000 x
T,
8.34 = 41,700 Ib
Wt    .   )  _.. _  
Ikg
\ 1'1 + 41.700
(b)
YV In
217 0.:
as
2r. _
[T
Squaring eqs. (a) and (b) and dividing correspondingly the left and right sides of these equations, results in
~L
1000
(2.2)'
(1.0)' =.
W + 41,700 W
k = 10,228 In lin
20
21
lO
T,
2 Ti"
10,228 x 386
Solution: The analysis is performed using a single spring elemem with one concentrated mass element. The foBowing com<nands are imp]emented in COSMOS:
(I) Set view to the XY pia"":
DISPLAY VIEW, 0,
>
0,
vn:w_PAR 1,
> VJ EW
(b) Shorten the peliod to T, = O.S X 0 1 = O.oS sec by strengthening columns in Llk:
>
GRID
>
PLANE
PL>.NE, Z,
0, 1
co =2..".
0.08
Solve for M:
i ,(_10~.2_2_8,.~.,."M,....:..)(",3",8",,6)
Y 1000
(3) Establish a grid with two diVIsions in the X and Y directions, then use the scale command;
GEOHE'l'RY
>
C,
GRIDON,
0, 1,
GlUDON 1.,
2, 2, 2
(Ans.)
DISPLAY SCALE, 0
>
DISP_PAR
>
SCALE
1.10
>
CRPCORD,
The foi;owing example is presented to illustrate the use of the program COSMOS in the analysis of structures modeled as single degreeoffreedom systems. Detailed explanations for the llse of COSMOS including numerous examples with data preparation and results are presented in the supplement STRUCTURAL DYNAMICS USING COSMOS . Example 1.6. An instrument of mass m = 0.026 (lb . secL/in) is mOl;.nted on isolation springs of total spring constant k = 29.30 (lb/in). Model the system as an undamped singledegreeoffreedom system (Fig. 1.14) and de;tennine its natural frequer:cy.
29.3 (ib/in):
>
RCONST
ReOM'ST,
1, I, }, 1,
29.3
>
1f
PAR..b.Y"...J,!:ESH
>
1
}CCR
2,
1,.
>
EGROUP
EGHOUP, 2. MASS, 0,
0, 0, 0, 0,
C,
0, 0, Q
(9) Define real constant for mass element: m = 0.026 (lb . sec2 /in):
PROPSETS A tonVenlelll foffi'l ror or>Jering ~his supplement is provided :in the lacs! page of this volume,
~CON$'l',
>
RCONS1'
2, 2, 1, 7,
0,
0.026,
0,
0,
22
23
(5)
>
PARhlCMESH
~
>
!'CPT
l"CFT .. 2, 2,
>
NMEHGE, 1, 3. 1, O.aOOL O. 0, 0
)' = A cos
fJJJ
+ B sin
mt
(12) App!y constraints in an degrees of freedom at node 1. and an degrees of freedom exce.pt UY at node 2:
LOADSBe > DP1', 1, l\L, STRUCTURAL 0, 1, 1
>
;)ISPLHKT'S
>
DPT
A; Yo
B =v(jiw
w=Jkim is the r.atural frequency in rad/sec w
(13) Set the options for the frequency analysis to extract one frequency using the Subspace Iteration Method with a maximum of 16 iterations, and run the frequency analysis:
A,NAI.''iSIS
>
FREQ!BlJCK
>
A_FREQUENCY
~
>
FREQ/BUCK
>
R_FREQUENCY
~FREQUENCY
(6)
>
= C sin (UJI + a)
LIST
>
FREQLIST
or
FREQUK'CY (2.AD / SEC)
3,35597E+Ol
FREQUENCY (CYCLES/SEC)
FREQUENCY#
1
PERIOD
{SECONDS)
5.34278:+:):)
:.87:68:)1
1.11
SUMMARY
and
c
tan a
llnfJ
The mathematical model of a structure is an idealized representation for its analysis. The number of degrees of freedom of a system is equal to the number of independent coordinates necessary to describe its ,Position. The free body diagram (FBD) for dynamic equilibrium (to allow application of D' A~embert's Principle) is a diagram of the system isolated from all orher bodies. showing the external forces on the system. including the inertial force.
(3)
an
PROBLEMS
L 1 Determine the natural period [or the system in Fig. P U. Assume Lhllt the, bellm and springs supporting the weight Ware massless. 1.2 The foHowing numerical values are given in Problem 1.1; L = 100 in, E! lOlI(lb_inl), W=3000 Ib, and k 2000!biin. If the weight W has an initial
(4)
The stiffness or spring constant of a iinear system is the force necessary to produce a unir displacement
24
Undampeij
Sjngle~DegreeQjFreedom
System
25
1.S
Detennine the natural frequency of the fixed beam in Fig. PLS carrying a concentrated weight W at its center. Neglect the mass of the beam.
EI
,
Fig. Pl.1.
T
Y
~.
..
fl
1.12
ml../2
I
Y
=I
~
Fig. P1.5.
OJ'=' 20 in/sec, cetermlne the displacement and the veloctry 1 sec later. 1.3 Determjne the natural frequency for norlzontaf motion of the sreel frame in Fjg.
1.6
P1.3. Assume the horizontal girder to be Infinitely rigid and neglect the mass of
the columns.
The numerical values for Problem 1,5 are given as; L = 120 in, El = 10 9(Ib . in2), and W 5000 lb. If [he initial displacement and (he initial velociry of the weight are, respectiveiy, Yo = 0.5 in and Vo = 15 in/sec, determine the displacement, vel~ odty, and acceleration of W when time ( = 2 sec, Consider the simple pendulum of weight W illustrated in Fig. Pl.?, A simple pendulum is a particle or concentrated weight that oscillates in a vertical arc and is supported by a weightless cord. The only forces acting are those of gravity and cord tension (I.e., frictional resistance is neglected). If the cord length is L, determine the motion ir the maximum 9sdllation angle B is small and the lnllial displacement and' velocity are 80 and 80 respectively.
1.1
Fig, Pl.3.
1.4 Calculate the natural frequency in the horizontal mode of the steel frame in Fig.
PL4 for the following cases: (a) me horiz,ontal member is assumed:o be infmitely rigid; (v) the horizontal member is flexible and made of steel~WJ:O X 33.
~ I\~
I I \
:'
;
W"" 25 Kips
;
Fig. Pl.7.
il
'I
11
II
;:
15'
W10
x 33
1.8 A diver standing at the end of a. diving board that cantilevers 2 ft osciIlates at a frequency 2 cps. Determine the flexural rigidity EI of the diving board. The
I
_L
"
1'5'
Fig. PlA.
wejght of the dIver is 180 lb. (Neglect the mass of the diving board).
1.9
A bullet weigh1ng 0.2 Ib is fired at a. speed of 100 ft/sec into a wooden block weighing 50!b and supported by a spring of stiffness 300 IbJin (Fig, PL9). Determine the displacement y{t) and velocity v(t) of the block after r sec,
,,j; ..
26
27
1.12
Sbow thilt lhc nalural frequency for the system of Problem L 11 may be expressed as
Fig. P1.9.
f= ie
where w mg, .It, is the critical budding weigbt, and 10 is the natlJral frequency neglecting the effect of gravity. 1.13 A vertical pole of length L and DexuaJ rigidity EI carries a mass m at ~ts top. as shown in Fig. P LJ 3. Neglecting the weight of the pole, derive the differential equation for small horizontal vibrations of the maSs, and find the natural frequency, Assume tba( Lhe effect of grav!:y is small and nonlinear effects may be neglected,
1.10
An c;evator weighing 500 Ib is suspended from" spring having a stiffness of 600 IbJin. A weight of 300 Ib is suspended through a cable to the elevator as shown schematically in Fig. PLIO, Determine the equation of motion of the elevator if the ca:::'Je of the suspended weight suddenly breaks.
=p ,
Fig. F1.l0.
tr
CPT~
I
'I . ddri
1.11
Write the differential equation of motion fo~ lhe inverted pendulum shown in Fig, P Lll and determine its natur<ll frequency. Assume small oscllJations, and neglect the mass of the rod.
Fig. P1.l3.
Equihbril1m
POS:tiOf!
1.14
Show the natural frequency for the system in Probler:> 1,13 may be expressed as
i~
/ = fn
y1 
W~,
where 10 is the natura! frequency calculated neglecting the effect of gravity and W<;Y is the critical buckling wejght.
1.15 Determine an el<p:ession for the n~tu:a! frequency of the weigbt W in each of the cases shown in Fig. P;' !5. 'fhe beams are uniform of cross"sectiona! moment of mertia J and modulus of ela5!Jci[y . Neglect the mass of the beams
Fig. Fl.l1.
28
29
w
k
L_.l
2
I
I
~k
,,,
(bi
JA,;.
~
I
4
~Hinge
0
C
{ 1:
I
a
Fig. P1.17,
!'t"r
[WJi
bl
1.18
Ie)
lei
Determine the natural freqt:e::cy of v;bration in t,1.e vertical direction for t.ie rigid foundation (Fig. PLlS) transmitting a uniformly distributed pressure on Ihe soil having a resuliant force Q = 2COO kN, The area of the foot of the foundation is 3 A = 10 mJ The coefficient of elastic compression of the soil is k "'" 25,000 kN 1m (Problem contributed by Professors Vladimir N, Alekhin and Aleksey A Antipin of the Urals Siate Technicai University, Russia.)
Fig. P1.15.
1.16
A system (see Fig. Pl.16) 15 modeled by two freely v;brating masses m; and!n2 interconnected by a spring having a constant k. Detennine for this system the differential equation of mo!ion for the relative displacement u = Y2  Yl between the two masses. Also determine the corresponding natural frequency of the system.
Fig. PL18.
1.19 Calculate the natural frequency of free vibration of the chimney on elastic
foundation (Fig. PL 19), pennitting the rotation of {he structure as a rigid body about the horizontal axis xx. The total weight of the structure is W with its center of gravity at a height h from the base of the foundation. The mass moment of inertia of the structure about the axis xx is 1 and the rotational stIffness of soil is k (resisting moment of the soil per unit rotation). (Problem contributed by Professors Vladimir N. Alekhin and AIeksey A. Antipin of the Urals State Technical University, Russia.)
~.nl,
k ~.
I
'1J'l
/=#/d')"'W/A/d/,Pflil.awV~$pp1bh'fliIP
Fig. P1.16. 1.17 Calculate ~he natura! frequency for the vibration of the mass In shown in Fig. PL17. Member AE 15 rigid with a hinge at C and a supporting spring of stiffness k at D~ (Problem contributed by Professors Vladimir N. Alekhio and A!eksey A Antipin of t::,e Urals State Technical University, Russia.)
30
~T./Jw rI :
b
~ . +j
2
Damped Single DegreeofFreedom System
Fig. Pl.19.
We have seen in the preceding chapter that the simple oscillator under idealized conditions of no damping, once excited, will oscillate indefinitely with a constant amplitude at its natural frequency_ Experience indicates, how~ ever, that it is not possihie to have a device that vibrates under lhese ideal conditions. Forces designatec as frictional or dampIng forces are always pres~ ent in any physical system undergoing motion. These forces dissipate energy; more precisely, the unavoidable presence of these frictional forces constitutes a mechanism through wh:ch the mechanical energy of the system, kinetic or potential energy> is transformed to other forms of energy such as heaL The, mechanism of this energy transformation or CissJpation is qujle compiex and is not completely :mderstood a! this time. In order to account for these dissipative forces in the una:ysis of dynamic systems, it is necessary to make some assumptions about these forces, on the basis of experience.
r,
"
2.1
VISCOUS DAMPING
iJ .. "
In considering damping forces ;n the dynamic analysis of strJctures. it is usually assumed (hal these forces are proportionaJ to the magnitude of the
.~ "
32
33
velocity> and opposite to the direction of motion. This type of damping is known as viscous damping; it is the type of damping force that could be developed in a body restramed in its motion by a surrounding viscous fluid. There are situations in which the assumption of viscous damping is realistic and in which the dissipative mechanism is approximately viscous. Nevertheless, the assumption of viscous damping is often made regardless of the actual dissipative characteristics of the system, The primary reason for such wide use of this method is that it leads to a relatively simple mathematical analysis.
which, after cance1!aiion of the common factors, reduces to an equation called the characteristic equation: for the system, namely
mp2 + cp + k == 0
The roots of this ;:;uadratic equation are
(2,2)
P,=
P2
~~... :t It~r~
2m'~
l2m)
(2,3)
2.2
EQUATION OF MOTION
Thus the general solution of eq. (2.1) 1s given by the superposition of the two possible solutions, namely
y(l)
Let us assume that we have modeled a str"JclUral system as a simple oscillator with viscous damping, as shown in Fig. 2.1(a). In this figure, In and k are, respectively. the mass and spring constant of the osci!!ator and c is the viscous damping coefficIent We proceed. as in the case of the undamped oscillator, to draw the free body diagram (FED) and apply Newton's Law to obtain the differential equation of motion, Figure 2.1(b) shows the FED of the camped oscillator in which the inertial force my is also shown, so that we can use D' Alernbert's Principle. The summation of forces in the y direction gives the differential equation of motion,
= C,e'"
+ C,e"
(2,4)
where C\ and C1 are constants of integration to be determined from the initial conditions, The final fonn of eq. (2.4} depends on the sign of the expression under the radical in eq. (2.3). Three distinct cases may occur: the quantity under the racical may either be zero, positive, or negative. The limiting case in which the quantity under the radical is zero is treated first The damping present in this case is called crilical damping,
my+cy+ky
(2,1)
The reader may verify that a trial solution y ;;;; A sin wt or y = B cos t:J! will not satisfy eg. (2.l). However, the exponential function yo:: Cef'i coes satisfy this equation. Substitution of this function into eg. (2.1) results in the equation
(~) __ k
2m
or
err:::::
(2,5)
mCp',,' + cCpe"
+ kCe"
=0
2 J1m
(2.6)
(,j
where Ccr designates the critical damping value. Since the natural frequency of the undamped system is designated by w ,{ki';;, the critical damping coefficient given by eq. (2.6) may also be expressed in alternative notation as
CCf
= 2mw = w
2k
(2,7)
(0)
A,,', :
"'Ic~j
I ~~"
In a critically damped system the roots of the characteristic equation are equal. ane, from eq, (2.3), they are
PI =P2= (2,8)
Fig. 2.1 (a) Viscous damped OSCj1Jl'HOf. (b) Free body diagram.
34
35
Since the two roots are equal, the generai solution given by eq. (2.4) would provide only one independent constant of integration, hence, one independent solution, namely
(2.9)
2.5
UNDERDAMPED SYSTEM
When the value of the damping coefficient is iess [han the critical value (c < c.: r), which occurs when the expression under the radical is negal~ve, the fOots of the characteristic eq. (2,3) are complex conjugates, so that
Another independent solution may be found by using the function (2.10) This equation, as the reader may verify, also satisfies. the differential equalion (2.1). The general solution for a critically damped system is. then given by the superposition of these two solurions,
(2.11 )
p,
p,
Ik    c +i 1
2m 
~ m
('2m, C'lT
(2.13)
where i = J~ is the jmaginary unit For this case, i[ is convenient to :nake use of Eu!er's equations which relate exponential and t.:igonometric functions. namely
ei~ = cos
+i
sin x
i s;n x
(2.14)
e  H = cos
T1H~ substitution of the roots
X 
2.4
OVERDAMPED SYSTEM
Pi and pz from eq. (2.13) into eq, (2.4) together with the use of eq. (2.14) gives lhe foilowing convenient fonn for the general Solullon of the underdamped system;
Y (l)
In an overdamped system, the damping coefficient is greater than the value for critical damping, namely
(2.12) Therefore, the expression under the radical of eq. (2.3) is positive; thus the two iools of the characteristic equation are real and distinct, and consequently the solution is given direcdy by eq. (2.4). It should be nOled that, for the overdamped or the critjcally damped system, the resulting motion is not oscil~ latory; the magnitude of the osciHations decays exponentially with time to ~ero. Fjgure 2.2 depicts grapbknJJy the response for the simple oscjHator with c'rilicat damping. The response of the overdamped system is similar to the motion of the critically damped system of Fig. 2.2, but the return toward the neutral position requires more time as the damping is increased,
=e
{,.;.'2>:.jl
(A cos woe
+ n sin
(/Jot)
WD,
(2.15)
where A and B are redefined conStants of integration and frequency of the system, is given by
the damped
(2.16)
or
(uv=
W
(2.17)
This lost result is obtained after substituting, in eq, (2.16), the expression for tbe undamped natural frequency
(2.18)
(2.19)
36
structures Modeled as a
Singje~DegreeolFreedom
System
Damped
Sjngle~DegreeofFreedomSystem
37
Finally, when the initial conditions of dispLacement and veiodty, Yo and VOl are introduced, the constants of integration can be evaluated and substituted ilito eq. (2,15), giving
y(t):;:; e
The value of the damping coefficient for rea; stmctures is much Jess than the critical damping coefficieJ1t and usuaUy ranges between 2 and 10% of the critical damping value. Substituting for the extreme value g= 0.10 into eq.
(2.17),
<Vol
' "0
+ yogw, _sm
) wpt,
(2.20)
WI))
Wo= Q,995w
(225)
It can be seen that the frequency of vibration for a system with as much as a
10% damping ratio is essentially equai to the undamped naturai frequency, Thus, in practice, me natural frequency for a damped system may be taken to be equal to the undamped natural frequency_
(2.22)
2,6
LOGARITHMIC DECREMENT
and
Uma="'UJDYO
(VD
+ Yu~w)
(223)
A graphical record of the response of an underdamped system with inilial displacement Yo but start:ng with zero velocity (uo:;;; 0) is shown in Fig, 2.3. It may be seen in this figure that the motion is oscillatory, but not periodic. The amplitude of vibration is not constant dcring the motion but decreases for succeSSIve cycles; nevertheless. the oscillations occur at equal intervals of time. This time !ntervul is designated as the damped period of vibration and is given from eq. (2.17) by
(2.24)
A practical method for determining experimentally the damping coefficienl of a system is to initiate free vibration, obtain a record of the osclIJatory motion, such as the one shown in Fig, 2.4, and measure the rate of decay of the amplitude of motion, The decay may be conveniently expressed by the logarithmic decrement (] which is defined as the natural logarithm of the ratio of any two successive peak amplitudes, y: and Y2, jn free vibration, that is,
y, 8=ln
(2.26)
The evaluation of damping from the logarithmic decrement foHows. Consider the damped vibration motion represented graphically in Fig, 2.4 and given analytically oy eq, (2.21) as
y(t) = Ce ~w.. cos (wn1 a)
y\tl
Yo
Peak
Fig. 2.4 Curve showing j::eak displacements and displacements at the points of genc),.
tan~
.... "',,
,"
38
39
We note from this equation that, when the cosine factor is unity, the displacement is on points of the exponential curve y (I) = Ce ~f"" as shown in Fig. 2A However, these points are near but not equal to [he positions of maximum displacement. The points on the exponential curve api?car slightly to the right of the points of maximum amplitude. For most practical problems, the discrepancy is negligible and the displacement curve may be assumed to coincide at the peak amplitude. with the curve y (1) = Ce .. wi so that we may write, for two consecutive peaks, Yl at time '1 and h at To seconds later,
Yl
and at time 1'2 ::;::c 1J + Tv, corresponding to a period later, whcn again the cosine function is equal to one and the sine function is equal to zero
(I
and
[1
is then
Ce f "",
and laking nalural logarithmic results in the logarithmic decrement in term of accelerations as
and
Dividing these two peak amplitudes and taking the natural logarithm, we obtain (2.27) or by substituting TD , the damped period, from eq. (2.24),
8=
which is identical to the expression for the logarithmic decrement given by eq. (2.27) in terms of displacements.
(2.28)
Example 2.1. A vibrating syslem consisting of a weight of W = 10 jb and a spring with stiffness k;;::. 20 lb/in is viscously damped so that the ratio of two consecutive amplitudes is LOO to 0.85. Delennine: (a) the natural frequency of the undamped system, (b) the logarithmic decrement, (c) the damp~ lng ratio, (d) the damping coefficient, and (e) the damped natural frequency.
Solution: second is
As we can see, the damping ratio {can be calculated from eg. (2,28) after detennlning experimentally the amplitudes of two successive peaks of the system in free vibration. For small values of the damping ratio, eq. (2.28) can be approximated by (2.29) Alternatively, the logarithmic decrement may be cak:l1ated as the ratio of two consecutive peak accelerations, wbich are easier to measure experimentaty than dispJaceme:1ls. In this case, raking the first and the second derivatives in eq. (2.20, we obtain
y(l)
= 27.78 radlsec
J=
4.42 cps
Ce 1~[WCOs(wDIa)Wosin(wda)]
Y(I)
= Ce ,~ [[ 
((J)vt
~ a)J)
y,
= In
100
0.85
= 0.163
0.163 ==0.026
2,,
40
Damped
Single~DegreeofFreedom
System
41
(d) The damping coefficient is obrained from eqs. (2,6) and (2.19) as
0.037 Jb . sec
277(0.05) = 0.314
and the ratio of two consecutive amplitudes 1.37
(e) The natural frequency of the damped system is given by 0<;. (2. J7),
so that
WD=w.:~tj
tub
y,
(0.026)' 27.77 rad/sec
= 27.78; I
Example 2.2. A platfonn of weight W 4000 Ib is being supported by four equal columns. that are clamped to the foundation as weH as to the platform, Experimentally it has been determined that a static force of F i 000 lb applied horizontally to the platform produces a displacement of Li = 010 in. It is estimated that damping in the structure::. is of the order of 5% of the critical damping. Determine for this strocture the following: (a) undamped natural frequency, (b) absolute damping coefficient, (c) logarithmic decrement, and (d) the number of cycles and the time requlre.d for the <1mplitude of motion to be reduced from an initial value of 0.1 in to O.Ol in.
Solution: (a) The stiffness coefficient (force per unit displacement) is computed as
k=~= ........=
(d) The ratio between the first amplitude Yo and the amplitude YI: after k cycles may be expressed as
Y"w, Y:.:
YI<
Yt
Yl
In
Y;
In 0,1
In 10 0.314
Ll
1000 0.1
Wa
is given by
3 1.06J I  (0.05)'
and the damped period T[) by
110,000 X 386
1  =..  
4000
31.06rad/sec
""'~
211'
31.02
= 0.2025 sec
64 3.8.In
lb'sec
= 8TD
1.62 sec
c=
32.19 lb sec
in
Example 2.3, A machine weighing 1000 Ib is mounted through springs having a total stiffness k 2000 lb lin to a simple supported beam as shown in Fig. 2.5(a). Detennine using the mathematical model shown in Fig. 2.5(b), the equivalent mass mE. the equivaient spring constant kE' and equivalent damping
!:
il Ii
,_ f
42
43
~.~.=.~=
f
(3)
c"
(lb'sec\ 12792 [~~~~!
\ 10
(Ans)
(\
Fig, 2,5 (a) System for Ex:affiplc 2.3; (b) Mathematica! modeL
Cs
Ccr=O.10X 12792
Ilb.seC) 12.79,.\ 10
(Ans)
coefficient c for the system assumed to have 10% of the criticai damping. Neglect the mass of the beam.
2.7
SUMMARY
So/mion: The spring constant k.b for a uniform simple supported beam is obtained from the deflection 3 resulting from a force P applied at the center of the beam:
Hence,
kb~5
48 X 10'
IT
481
=    = 7500 Jb/in
Real stmcturcs dissipate energy while undergoing vibratory motion. The most common and practical method for considering this dissipation of energy is to assume that it is due [0 viscous camping forces, These forces are assumed to be proportional to the magnitude of the veiocilY but acting in the direction opposite to the motion. The ftletar of proportionality is called the viscous dampiflg coefficient. It is expedient to express tbis coefficient as a fraction of the crifical damping in the system (the damping ratio {= c ICe,)' The critical damping may be defined as the least value of the damping coefficient for which the system will ;'lot oscll1ate whe!': disturbed initially, but it simply will return to the equilibrium position. The differential equation ef motion fer the free vibration of a damped single degreeorfreedom system is given by
40'
The equivalent spring constant is then calculated using eq. (1.5) for two springs in series:
my + C)' + k)l =
=.+2000 7500
kF= 15791blin
The analytical expression for the solution of this equation depends on the magnitude of the damping ratio. Three cases are possible: (l) eriticaUy damped system <tf:::: 1), (2) underdan1ped system ([< :), and (3) overdamped system (i> 1), For the underdamped system (1;< J) the solution of the differential equation of motion may be written as
(Ans)
44
Dar:;ped 2.5
SingleDegree~ofFreedom
System
45
in which
(jJ;;;
Show that the displacement for critical and overcritical damped systems with init:al displacement Yo and velocity Vo may be written as
y=e<dlyo(:l
{ki;;;
{=1
Y"" e where
EM. r
;Yo cosh L
sinh wi;!
1 for
and Yo and tlo are, respectively, the jnitial displacement and velocity. A common method of detennining the damping present in a sys~em is to
2.6 A structure is modeled as a damped osciIlntor with spring constant k "" 30 Kips! in and undamped natura! frequency w= 25 rad/sec. Experimentally il was found that a force 1 Kip p;>oduced a relative velocity of 1.0 in/sec in the damping
, (e) the logarelement Find: (a) rhe damping ratio fj, (b) [he damped period T D and (d) the ratio between two consecutive amplitudes. ithmic decrement
evaluate experimentally the logartthmic decrement, which is defined as the natural logarithm of the ratio of two consecutive ueaks for dis!)iacement or for acceleration. in free vibration, that is, .
a,
2.7
2,8
In 2.4 it is indicated that the tangent points to the displacement curve correspond to cos (wot a)"" L Therefore the difference in We( between any twO consecutive tangent points is 211". Show lhat the difference in Wpf between any two consecutive peaks of the curve is also 211", Show that for an underdamped system in free vibration the logarithmic decrement may be written as
(5: = .. ~
The damping ratio in structural systems is usually less than 10% of the critical damping (f< 0,]). For such systems, 6e damped frequency is approximately equal to the undamped frequency_
2.9
In ''YiH
v'
where k IS the number of cycles separating two measured peak amplitudes Yi and
It has been estimated that damping in the system of Problem 1.11 is IO% of the
I.
PROBLEMS
2.1 Repeat Problem L2 assuming that the system has 15% of crit:cal damping.
2.10
2.2 Repeat Problem 1.6 assuming that the system has 10% of critical damping, 23 The ampiitude of vibration of the system shown jn Fig. P2.3 is observed to
decrease 5% on each consecucive cycle of motion. Determine the damping coefficient c of the system~ k = 200 Ib lin and m = 10 lb sec1 l!n. 2.11
critical va~ue. Determine the damped frequency In of the system and the absolute value of the damping coefficient c. A single degreeoffreedom system consj.5ts of a mass with a weight of 386 lb and a spring of stiffness k 3000 !b/in. By testing the system it was found that a force of JOO Ib produces a relative velocity i2 in/sec. Find (a) the damping ratio , (b) the damped frequency of vibration If), (c) logarithmic decrement 8, and (d) the ratio of two consecutive amplitudes. Solve Problem 2.10 when the damping coefficient is c=2Ib, sec fin. For each the systems considered in Problem L15, determine the equivalent spring COnstant k! and (he equivalent damping coefficient CE in the mathematical model shown in Fig. P2.12. Assume that the damping in these systems is equal to lO% of the critical damping.
IT , "'T
_C k
2.12
0:
:
,
i ,
I
Fig. 1'2.3.
2.4 It is observed experimentally that the ampIitude of free vibration of a certain structure, modeled as a single degreeof freedom system, decreases from 1 to 0.4 in 10 cycles. What is the percentage of critical damping?
Fig, P2.12.
46
Structures Modeled as a
Sing!eDegreeof~Freadom
System
2.13
A vibration generator with two weights each of 30 lb with an eccentricity of lOin rotating about vertical axes in apposiLe direclions is mouo!ed on the <roof of a onestory building with a roof thai weighs 300 Kips. It is observed that the maximum latera! acceleration of 0.05 g occurs when the vibrator generaior is rotating at 400 rpm. Determine the equivalent viscous damping in the struCture.
2.14
A syslem is modeled by two freely vibrating masses ntj aud nt2 interconnected by
a spring and a damper element as shown In Fig. P2, 14. Determine for this system the differential equation of motion in terms of tbe reln.rlve mOtion of '.he masses U=YlYl
Fig. P2.14. 2.15 Determine the relative motion t/ = y~ ~Yl for the syslem shown in Rg. P2.14 it~ terms of the natural frequency fl), damped frequency W.o, and relative damping. Hint: Define equivalenr mass as M = mj1Ji21{ml + 111,).
In this chapter. we will study the motion of slructures idealized as single~ degree~offreedoIr. systems excited harmonkaHy, lhat is, structures subjected to forces or displacemenl<; whose magnitudes may be represented by a sine or cosine funclion of time. This type of excitation results in one of the most important molions in the study of mechanical vibrations as weB as in applications to structural dynamics. Structures are very often subjected to the dynamic action of rotating machinery which produces harmonic excitations due to the unavoidable presence of mass eccentricities in the rotating parts of such machinery. Furthermore, even in those cases when the excitation is not a harmonic function, tbe response of the structure may be obtained using the Fourier Method, as the superposition of individual responses to the harmonjc components of the external excitation. TIltS approach will De dealt wjth in Chapter 5.
3.1
The impressed force F(t) aCling on the simple oscilhHor in Fig. 3.1 is assumed to be harmonic u:1d equal to Fo sin w/, where Fo js the peak amplitude and iiJ
47
48
49
or
y=
Fo km&/
(3.5)
f.y...j
)b)
in which r represents the ratio (frequency ratio) of the applied forced frequency to the natural frequency of vibration of the system, that is,
:t'h
fo 5ir.;:;r
iii r=
(3.6)
'"
Combining eqs. (3.3) through (3..5) with eq. (3.2) yields
y(l) ==.1 cos
Fig. 3~1 (a) Undamped osclllator harmonically excited. (b) Free body diagrf'.!:1.
is ~"e frequency o~ the force in radi,ans per second. The differential equation ob.amed by summrng all the forces m the free body diagram of Fig. 3, I (b) is
wt+ B Sin
wt+~
1 r
(3.7)
my + ky = Fo sin
?Ix
(3.1 )
If the initlal conditions at time t:=: 0 are taken as zero (Yo";;:; 0, constants of integration determined from eq. (3.7) are
A=O,
B=~17
Of)
0), the
rFolk
which, upon substitution in eq. (33). gives ,,:here YC(,I) is the complementary solution satisfying the homogeneous equatlOn~ that IS, eq: (3.1) wIth the lefthand side set equal to zero; and Yp(l) is the partlc~lar solu~Jon based on the solution satisfying the nonhomogeneous differentJal equatIOn (3,1). The complementary solution, v (!) is 2:iven by eg (l.I7) as < ,
y (I)
~
   , (sm iilt  r sm WI 1  r~
Folk.
.)
(3.8)
y .. (1) =A cos
where w
Wl:
B sin
(ut
(3.3)
[kim"
The nature of the forcing function in eq. (3.1) suggests that the partIcular solution be taken as
yp(t) = Y sin iiJt (3.4)
As we can see from eg. (3.8), the response is given by the superposition of two hannonic tenns of different frequencies. The reSUlting motion is not hannonic; however, b the practlcal case, damping forces wiH always be present in the system ar:d will cause the last tenn, i.e., the free frequency term in eq. (3.8), to vanish eventually. For this reason, this term is said to represent the transient response. The fordng frequency term in eq. (3.8). namely
y{t) =
F,lk
1
Si;1
wt
(3.9)
wher~ Y 15 the peak value of the particular solution. The subsritut10n of eq. (3.4) Into eq. (3.1) followed by canceHation of common factors gives
?naif"'" kY= Fo
is referred to as the steadystate response. It 1s clear from eq. (3,8) that in the case of no damp1ng in the system, the transient will not vanish and the response is then given by eg. (3.8). It can also be seen from eg. (3.8) or eg. (3.9) that when the forcing frequency ~s equal to the natural frequency (r == LO), the amplitude of the motion becomes infinitely large. A system acted
50
51
upon by an external excitation of frequency cOinciding with the natural frequency is said to be at resonance, In this circumstance, the ampljtude will' increase gradualiy to infinite. However, materials that are commonly used jn practice are subjected to strength limitations and in actua; structures failures occur tong before extremely large amplitudes can be auained.
For this purpose, t:le reader should realize that we can write eq. (3,10) as
(3.13)
with the understanding that only the imaginary component of Fo etOJ<, i.e., the force component of Fo sin WI, lS acting and, consequently, the response will then consist only of the imaginary pan of Ihe total solution of eg, 13.13). In other words, we obtain the solution of eq. (3.13) which has real and imaginary components, and disregard the real component. It is reasonable to expect that t)le parlic"Jlar solution of eq. 13.13) wllJ be in the fOfm of
(3.1~)
3.2
Now consider the case of the onedegreeoffreedom system in Fig. 3.2{a) vibrating under the jnfluence of viscous damping. The differentia! equation of motion is obtained by equating to zerO the sum of the forces in the free body diagram of Fig. 3.2(b). Hence
my + cy + ky = Fo sin
[ol
(3.10)
Substitution of eq. (3.14) into eq. (3.13) and cancellation of the factor e'1:.Jt: gives
The complete solution of this equation ag;::jn consists of the complementary solution Ye(r) and the panicular solution yp(t). The complementary solution is given for (he underdamped case (c < C,.) by eqs. (2.15) and (2.19) as
Y,,{t) = e~(A cos
(,tJo(+
 m{j,}C + iciiJC + kC = Fa
or
B sin wni)
(3.11 )
c
ano
lIl&l""t" fe&!
(3.15)
into eq. (3.10) and equating the coefficients of the sine and cosine functions, Here we follow a more elegant approach using Eu}er's reJat:on, namely
eg~
(3.15) may
,'f
if
e;:"
cos wI + i sin WI
)'p
or
(3,16)
I I
Ii
I
where
(bl
J
(3.17)
I~
Fig. 3.2 (a) Damped oscitlator harmonically exciled. (b) Free body diagram.
I
,
.i
52
Structures Modeled as a
Sir,gle~Degfee..j)f~f(eedom
System
53
The response to the force in Fo sin WI (the imaginary component of Foe i'''') is then the imaginary component of eq. (3.;6), na:nely,
0.125
I I
t, 0
(3.18)
,
o.
!
i ,
,
or
i i
rii
,
(3.19)
,
is the amplitude of the steady~s"'te motion. Equations (3.18) and (3.17) may conveniently be wnuen in terms of dimensionless ratios as
(3.20)
1
!
i
I;;'
<125
i i i 1
i
II
,
1\
~
i
i
II.
!
and
, 'e
tan B=
(3.21)
,/
tI~~
II I
i\,,\
where Ys, = Folk is seen to be the static deflection of the spring acted upon by the force F0; g= c lew the damping ratio; and r iJJl lV, the frequency ratio. The total response is then obtained by summing the complementary solution (tran
1,\ "o
"" I'\.
'\,
I I
, r<~'t ~
I.
i1: 1 t,
'\. ,\ 1\.\
r:0
~ ;::,.
2. 3.
sient response) from eq. (3.11) and the particular solution sponse) from eq. (3.20), that is,
(steady~state
re
(3.22) The reader should be warned that the constants of integration A and B must be evaluated from initial conditions using the total response given by eg. (3_22) and not from just the transient component of the response given in eq, (3.11). By examining the transie<lt component of response, it may be seen that the presence of the exponential factor e {"'" will cause this component to vanish, leaving only the steadystate motton which is given by ego (3.20). The ratio of the steady~state amplitude of Yp(l) to the static deflection )'5: defined above is known as the dynamic magnification factor D, and is given from eqs. (3.19) and (3.20) by
Y
:F'ig. 3.3 Dynamic magnificatiou factor as a function of the frequency ratio for various amounts of damping.
It may be seen from eq. (3.23) that the dynamic magnification factor varies with the frequency ratio r and the damping ratio f Parametric plots of the dynamic magnification factor are shown in Fig, 3.3. The phase angle 8, given in eq. (3.21), also varies with the same quantities as it is shown in the plots of Fig. 3.4. We note in Fig. 3.3 lhat for a lightly damped system, the peak amplitude occurs at a frequency ratio very close to 1~ that is, the dynamic maQnification factor has its maximum value virtually at resonance (r = 1). It can'" aiso be seen from eq. (3.23) that at resonance the dynamic magnification factor is inversely proportionai to the damping ratio, that is,
D(r=I)=21;
D= 
y"
= """E==;c;==.c:r
(3.23)
(3.24)
54
Structures Modeled as a
180
Single~Degree~ofFreedom
Syslern
55
~I
l
0.1
~O
I.
,
i
..
f
I
!
/:02
/
...r
,
1
b iiz
0
/,
i..,.. fj ,
/ 0,'1
I
, V
I V
/,
'
::::
4 
El
'
L~{I
~. r' !
V r. l = ~f:::IIV

/'
ri
' ,2::
~ !
large mass of the machine. Figures 3.5 and 3.6 show, respectively, the schematic diagram of a beam~machine system and the adapled model. The force at the center of a slmply supported beam necessary to deflect this point one unit (i.e., the stiffness coefflclent) is given by the formuia
I I I
if
0
!
i
..
,
k=~=
L'
l'
04 / 1
.~
,
I
W I
I
,
+_
i
3
The natural frequency of the system (neglecting the mass of the beam) lS
I
2
Y III
1_
rT
161~92() V 16,000/386
I~ =
38.65 rad/sec
FreQuency ratio
r '" w/w
i
i
60
fj
Although the dynamic magnification factor evaluated al resonance is close to its maximum value, it ~s not exactly lhe max.imum response for a damped system. However, for moderate amounts of damping, the difference between the approximate value of eq. (3,24) and the exact maximum is negligible.
Example
3~1.
1in i:;,
weight W 16,000 lb. The beam is made of two standard S8 X 23 sections with a clear span L = 12 ft and total cross~sectiona! moment of inertia 1= 2 X 64,2 = 128A in4. The motor runs at 300 rpm, and its rotor is out of balance: to the extent of W' = 40 lb at a radius of eo = 10 in. What will be the amplitude of the steadystate response if the equivalent viscous damping for the system is assumed 10% of the critical? This dynamic system may be modeled by the damped osciHato/ The distributed mass of the beam will be neglected in comparison with the
SoIUlion:
, m
~,
~~Td.
1,1
1 ...
/##,$#/////.&7/.,w///)..w~
TT
!mm'}~;
'y
'"
Fig. 3.6 (a) Mathematical model for Example 3.L (b) Free body diagram.
'I
56
57
Ftr<:.:;=;s':;S
tK;P~$=iS1 T
I
W8X 20
Referring to Fig, 3,6, let m be the total mass of the motor and In the unbalanced rotating mass. Tnen, if y is the vertical displacement of the nonrotating mass (m  m ') from the equilibrium position, the displacement YI of m' as shown in Fig. 3.6 is
J
f'5'
(b)
Yt = Y
+ eo sin
Wi
(a)
Fig. 3.7 (a) Diagram of frame for Example 3.2. (b) Mathematical model.
The equation of motion is then obtained by summing forces along the vertical direction in the free body diagram of Fig. 3.6(b), where the inertial forces of both the nonrotating mass and the unbalanced mass are also showll. This summation yields
(m  m')y + m),
+ cj + ky = 0
(b)
Example 3.2. The steel frame shown in Fig. 3.7 supports a rotating machine that exerts a horizontal force at the girder level, F(t) = 200 sin 53! lb. Assuming 5% of critical dampir.g, determine: (a; the steadystate amplitude of vibration and (b) the maximum dynamic stress in the columns, Assume that the girder is rigid.
Solution: Thi~ structure may be modeled for dynamic analysis as the damped QsciHator shown in Fig" 3,7(b), The parameters in this model are computed 'as follows:
(a)
Wl
(c)
This last equation 1S of the same form as the equation of motion (3. I 0) for the damped oscillator excited hannonically by a force of amptjrude
(d)
200
= 0.0936 m
w
Substjtuting in this equation the numerical values for this example. we obtain
Ym
15,000
F,=(40)(1O)(31.41)'!386= 1022lb
The amplitude of the steadystate resulting motion from eqs< (3< 19) and (320) :5 then
r=
iii
53 ==0<715 w 7AI
1022!6L920
(Ans<)
'. A coit dls.ptacement at (he top of a pir.ned supported column requires a force equal to 3El JL).
58
59
3EIY
= 2018 Ib
;; "
l.1 \l\lU
and the maximum stress
=V
mIlA
L = 36,324 Ib . in
" , , " =
0
~ c
crm~~ =  
Mm."
lie
36,324
2136
E ., E
:r
~ ,.
psi
(Ans.)
"
3.3
We have seen in Chapter 2 that the free~vibration deeay curve oermirs the ev~iuation .of d~mping of a singledegree~offreedom system by si~pjy calcu~ latmg ~h~ loga:l{hr:n d~crernent as shown in eq. (2.28). Another fechnique for detennlOlOg dampmg 1$ based on observations of steadv~st3,te harmonic re~ sponse, :hi~h requir~s barmor.ic excitations of the st~cture in a range of ,.freq~encles In lb~ neighborhood of resonance. With the application of a harmOnIC force Fo S10 iVt at cJoseJy spaced values of frequencies, the response curve f?r the structure can be plotted, re,sulting in displacement amplitudes as
a functiOn of the applied frequencies. A typical response curve for such a moderately damped structure is shown in Fig. 3.8. h is seen from eq, (3.24) that, at resonance, the damping ratio is given by
and Y", is the maximum amplitude. The error involvec in evalualing the damp~ ing ratio using the approximate eq. (3.26} is not significant in ordinary Structures, This me.thod of determining the dampir.g ratio requires only some simple equipment to vibrate the sln;c~ure in a range of frequencies that span the resonance fre,quency and a transducer for measuring umpHtudes; nevertheless, the evaluation of the static displace:r.ent YH = Folk may present a probJem since, frequently, it is difficult to apply a statk lateral load to the structure.
t=
2D(r 1)
(3.25)
whe:~ D(r= 1) is the dynamic magnification factor evaluated at resonance, pracuce. the damping ratio is determined from the dynamic magnification factor evaluated at the maximum amplitude, namely,
in
~=
2Dm
(3.26)
where
D=!::.
}"
60
.12
61
I
,
kp'~k ~ 0.11,34 io
, I
.10
!
i
or
i
g=~
1 iiJ,'W,
I
i
,
I
I
"
I
,,';'Pt'ilk
0,0802
,
since
'"
j,+/,
(3.27)
,
i
. 02
f H
I
~ ~_ fZ  ( l ~ i.87
I
,
I:
,
i
oB
12
/'
V'
V' ' A
I
f. ""
f,
1'7.05: 18.97
'\
r=
Example 3.3.
degreeof~freedom
]
/
and
/=~2
 /,+1,
:1
r;:f7;"' ~= 5.2%
I, fr"'~""'1"
"
Experimental data for the frequency response of a single system are plotted in 3.9. Detennine the damping ratio
10
14
16
18
20
22
I' ,.
i
i
of this system.
26
2.
Solution: From Fig. 3.9 the peak amplitude is 0.1134 in; hence the amplitude at halfpower is equal to
0.1134/[2
0.0802 in
response amplitude in eg. (3.20) equal to 1 fj2 times the resonant ampUtude given by eq. (3.24), that is,
Squaring both sides and solving for the frequency rat;o results in
f=l,  11
or by neglecting
j,+/1
18.92  17.05 Ij= i8.92';' 17.05 5.2%
(Ans.)
2g' 
r~= J
22 + 2~
r,=IIje
r2= 1 + ~~
3.5
Finally, (he damping ratio is given approximately by half the diffe:ence between these halfpower frequency ratios, namely
The energy, Eo, dissipated by viscous damping during one cycle of harmonic vibration of frequency (i; is equal to the work done by the damping force cy during a differential displacement dy integrated over one ;;>eriod of vibration T = 2.,,lw. Heoce,
Eo
'"I" (cy)dy
_0
r "1"1f/';'
Jo
dy (ey)  , dt = d. ..0
[2 . . /,;
ej'dt
(3.28)
'
62
63
The velocity, y;;;;; y(t), for damped oscillator acted upon by the harmonic force, F= FQ sin wr, is given by the derivative of eq, (3.19) as
y(l)= Ywcos ((I;r (J)
2(rY
(3.29)
Eo=cy2
where
a?L
('2"./w
= 21T!;rky1
(3,30)
2(r
Folk
!
" i
"
1:
21T(rkY'
(=,
c~
r=.
w
Tv
It =1 V JJJ
and
Cee
= 2jk;;:;
which is equal to the energy, ED, dissipated per cycle by (he viscous dampjng force as given by eq, (3.30).
Equation (3.30) shows that the energy dissipated by visco:Js dnmping is proportional to the square of the amplit"Jde of the motion Y. The energy dissipated by viscous damping is provided by the work Wp performed by the external force. During one cycle. the work of the external force F:::::: Fo sin WI is
WF =
3.6
", .  1"'" . 1
()
Fo
3m lIJt dy
1,
()
Po
SUI
dy wI dr dr
r1,,j::'
J,
r2f.'r:.
= 1TFOY sin e
(331)
To demonstrate that work, W p , of the exciting force in eq. (3.3 1) is equal to the energy dissipated, ED> by the viscous force in eq. (3JO), we need to substitute the sine of the phase angle 8 intO eq. (3.31). Thus, from eq. (3.2i), we have sin 8 2fT cos8 =1=7
freque:lcies 11 and A at amplitude equal to Thus, the eqUivalent viscous damping (3.26) as
Y.
2Y."
(3.32)
si~rfT+ cos 2 0 = (1 
sin'
(2~r)'
rZ)l
+ (2r)z
+/'
64
Structures Modeled as a
S;ng!eDegreeo~Freedom
S:lsteI":"J
65
Alternatively, the equivalent viscous damping ~eq may also be evaluated experimentally using the expression for the logarithrn1c decrement fj from eg. (2.28) 0' app<oximately from eq. (2.29) as
R~ ~ ____ _
E: "" Maximllm st.'"ai:!
(3,34)
energy sto;ec
The most common definition of equivalent viscous damping is based On equating the energy dissipated, in a vibration cycle of the actual structure, to the energy dissipated in an equivalent viscous system. Hence, equating the energy, EO, dissipated in a cycle of harmonic vibration detennined from experiment to the energy, ED, dissipated by an equivalent viscous system given by eg. (3.3D), we have
Fig. 3.10 Restoring ferce vs. displacement during a cycle of vibration showing the energy dissipated e" (area within the loop) and the maximum energy stored (triangular
and
Example 3.4. Laboratory tests on u structure modeled by the damped springmass system (Fig, 3.1 1a;, are conducted to evaluate equivalent viscous damping usjng (3) peak amplitude and (b) energy dissipated. The experimental restoring forcedisplacement plot at resonance is shown in Fig, 3,11 (b).
100 ~ , _ _ _,
W"'" 38.6lh
or
k
(3,35)
c
F(l)
LS.sinwr(Jb)
in which Es, the strain energy stored at maximum displacement if the system were eIastic, is given by
1 , E , ~kY 2
(,)
(3.36)
~
i1845(lb,in)
In the determination of the energy. EO, dissipated per cycle and the elastic energy, E" stored at the maximum displacement, an experimer:t is conduced by vibrating the structure at the resonant frequency for which r= wlw= L At th:s frequency, damping in the system has a maximum effect. With appropriate test equipment and measuring instrumentation. the restoring force and displacement during a cyde of vibration are measured to obtain a plot of the type shown in Fig. 3.10. The area endosed in the loop during one cycle of vibration is equal to energy dissipated. E', and the triangular area corresponding the amplitude Y is equaJ to the maxImum strain energy E., Consequently, the equivalent viscous damping is evaluated by eq. (3.35) from the experimental results E~ and En with r= 1, obtained from resisting forcedisplacement plot
Oisp!aem(1!\l
(b)
Fig. 3.11 (a} ~iatherr.atical mode! for Ex.ample 3.4. (b) Ex.perimental restoring forCeamplitude plor.
66
Response 01 OneDegreeofFreedom
Sys~em
10 Harmonic Loading
67
Solution:
where the support of the simple oscillator :nodding the structure is subjected to a hannonic motion given by the expression
y~ (r) =
Yo sin
wt
(3.37)
and
w= 1
Ik =,1f\OO =3L62radlsec
OJ
where Yv is the maximum amplltude and liJ is the frequency of the support motion. The differential equation of motion is obtained by setting equal to zero lhe sum of the forces (includIng the inertia! force) in the corresponding free body diagram shown jn Fjg. 3.12(b}. The summation of the forces in the horizontal direction gives
(3,38)
by eg. (3,32)
The substitution of eq. (3.37} into eq. (3.38) and the rearrangement of terms result in
(b) F,..quivalent viscous damping calculated from energy dissipated E" = 0.66
(lb' in) and E, = O.845(lb, in) shown in Fig, 3.11:
my + cy + ky
The
kyo sin
w( + cWyo cos
wI
(3.39)
0.66
4'11'(1.0) 0,845
=0,0621 = 6,21%
by eg. (3.35)
lWO harmonic terms of frequency w in the righthand side of this equation may be combined and eq. (3.39) rewritten as [similarly 10 eqs, (1.20) and (L23)]
my + cj + ky ~ Fo 51n
3,7 RESPONSE TO SUPPORT MOTION
where
(WI
+ f3)
(3.40)
There are many actual cases where the foundation or support of a structure is SUbjected to time varying motion. Structures subjected to ground rnot~on by earthquakes or other excitations such as explosions or dynamic action of machinery are examples in which support mottOl)S may have to be considered in the analysis of dynamic response. Let us consider in Fig, 3.12 [he case
(3.41) and
tat, f3 = cwlk = 2Ft,
(3.42)
h is apparent that eq. (3.40) is l.~e differential equation for the oscillalor excited by the harmonic force Fo sin (fur + 13) and is of the same fo:m as' eq. (3.10). Consequently, the steadystate solution of eq. (3.40) is given as before by egs, (3.19) and (3.20), except for ,he addition of the angle {3 in tile argument of the sjne function. that is,
Folk
y(Il
(3,43)
y,
Fig. 3.12 (a) Damped simple oscillator harmonically excited through its support. (b) Free body diagram including inertial force.
/1+ (2r[,)' .. . ,
sin (WI + {3
I!)
(344)
Equalio::t (3.44) JS the exp:ession for the relative transmission of the support motion to the oscillator. This is an important problem in vibration isolation in
68
69
which equipment roust be protected from harmful vibrations of the supporting structure. The degree of relative isolation is known as transmissibility and is defined as the ratio of the amplitude of motion Y of the oscillator to (he amplitude Yo, the motion of the support From eq. (3.44), transmissibility Tr is then given by
2~7~~k~~~~~
(3.45)
Analogously to the motion transmitted, we may find the acceleration transmitted from the foundation to the mass. The acceleration transmilted to the mass is given by the second derivative of y(l) in eq. (3.44) as
ji (I)
fr~uer.cy
(3A6)
while the acceleration )Ii(1) of the foundation is obtained from eq. (3.37)
(3.47)
(l/sec) for the type of t:lbber pad material used in the isolation. 'What is the stiffness of the isolation requ:red to reduce the acceleration transmitted to the instrument to 0.0 I g? .
Solution:
The acceleration transirnissibillty, Tn is then given by the ratio of the amplitudes of the acceleration in eqs. (3.46) and (3.47). Hence,
(3.48)
. to 0.1 and beginning with an assumed value ~;;;; 0.10 for the damping ratio, we have
T,=~
J1 + (2r~J' 2 +(2rl)
J(1
r)2gSS
=0.1
(a)
It may be seen that the transmissibility of acceleration given by eq. (3.48) is identical to eq. (3.45), the transmissibility of displacements, Hence, the same
A pJot of transmissibility as a function of the frequency ratio and damping ratio is shown in Fig. 3.13. The curves in this figure are simHar to curves in Fig. 3.3. representing the frequency response of the damped oscillator, the major difference being that all of the curves in Fig. 3.13 pass through the same It can be Seen in Fig. 3,13 that damping point at a frequency ratio r = tends to reduce the effectiveness of vibratlon isola:ion for frequencies greater than this ratio, that is. for r greater than
/2.
= 13.346
wlw;;;' 3.653
62.83radlsec
r;;;;
w=27if=27T1O
w
Exampie 3.5. A delicate instrument weighing 100 lb is to be mounted on a rubber pad to the floor of a test laboratory where the vertical acceleration is 0.1 g at a frequency of f= 10 cps. [t has been determined experimentally that the ratio of the stiffness, k, to the damping coefficient, c, IS equal to 100
wlr = 17.20rad/sec
0.259 x 17.20' = 76.641blin
70
71
we obtain
klc= 100
u
Yo
where 0 is given in eq. (3.2:).
j (l
r')' + (2rEj'
(353)
Example 3,6, If the frame of Example 3.2 (Fig. 3.7) is subjected to a sinusoidal ground motion y .. (t) = 0.2 sin 5.3[, determine: (a) the transmissibility of motion to the girder, (b) [he maximum shearing force in the supporting columns, and (c) maximum stresses in the columns.
whIch is somewhat less than the assumed value , 10. If desired, an itera!ive cycle could be. performed introducing = 0.086 in eq, (a) and repeating the calculations. Equation (3.43) provides the absolute response of the damped oscillator to a harmonic morion of its base. Alternatively, we can solve the differential equation (3.38) in terms of the relative motion between the mass m and the support given by
(3.49)
Sotwion: 3.2 as
k~21361blin
0.05
Yo = 0.2jn
Y. = 0.0936 in
tv
= 7A1 rad/sec
tv = 5.3 rad/sec
r
0.715
mu
+ eLl + ku =
F~rr(t)
(3.50)
where Fef'r(t) =  my~ may be interpreted as lhe effecti~e force acting on lhe mass of the oscillator, and its displacement is indicated by coordinate 11. Using eg. (3.37) to obtain y, and substituting ;n eq. (3.50) results in
(Ans.)
mil + cu + kH
(35 J)
Again, eg. (3.51) is of the same form as eg. (3.10) with Fo myod. Then, from egs. (3.19) and (3.20), the steadystate response in tenns of relative motion is given by (3.52)
j (!
 r')'
+ (2rtl'
0.206in
Then the maximum shear force in each column is (Ans.) or substituting (c) The ,maximum bendIng moment
iiJ'
kim
72
l1es';)or.se of
Of:e~DegreeofFreedom
73
T,
and
0<6869
y~
0<6869 X 25
L72in
(Ans<)
(b) Speed for maximum vertical amplitude: At relatively small values of damping. the maximum ampiitude occurs close to the resonance condition of r= L However, in this problem. the value of the damping, 40%. is large zoe it is necessary to determine the value of r that wiH result in maximum amplitude. Sqnaring eq, 0.45) and substituting the damping ratio 0.4 gives
1 + (0<8r)'
1 + 0<64r'
1; = U?j2 + (0<8r)'
The vertical displacement of the tires is y{t) = Y{J sin W!, where Yo = 2.5 in. The forcing frequency is iiJ = 211ft, where the forcing period T=' Llv is the time taken by the automobile with velocity u to traver the distance L over one cycle of curved roadway. Hence,
then
d(T;) ~ 0
dr
and
gives r = OJ = 0.893 w
W
(a) 'Ylaximum displacement Y:
v
~
27TU IL
(a)
w= O<893w= 0<893 X
1050 ~ 937rad/sec<
50 mph
7334 ft/sec
18A3rad/sec
Maximum vertical amplitude occurs at this value of the forcing frequency, which by eq. (a) corresponds to the velocity
w=
2"X 7334
fiJL
(937) (25)
2".
2"
(Ans)
w= / \ m
r;;
I
I,
{""iooo X 386
=.
= 1O.50rad/sec
At this velocity, the transmissibility and ampIitude of vibration of the automobile would be calculated by eq. (3A5) as
T,
and
y
w 18A3 r~~w
1050
5 in.
I ,
I !
1.755
L6S
Yo
 ...;
Fig. 3.14 Idealized bedroad for Example 3.7.
L65 X 25 ~ 4J3in
(Ans)
Example 3.8~ A machine having a total weight of 1800 lb, including its foundation. is to be isoLated from the vibration of the ground, which is 900 cps
. i
; 5:
74
75
owing to other machines operating near. Determine the stiffness of rubber isolation spring (0 limit the transmitted vibration to lJ 10: (a) neglect damplng and (b) consider damping given by the exp:ession c = k 1170 obtain experimentaHy [units of c (!b sec lin) and of k Oblin)]_
SolutioJt:
and
143.24
\ 4.663
r;;
(a) = 0
o
T,
k=7!8L381b/in
The damping is then determined from
Y
Yo
~..,.,,
:!:o/)
= 0.1
as
c=k/170=718L31l70
(1._/)= !O
1+,.'=10 1"=11
1'=3.3166=w
c = 42.241b sec/in
and the damping ratio as
w
rad
w =  = 143.24217 sec
900
c"
2~
42.24
348.34
12%
m= w=
~I'
1800
386
.  = 4.663 Jb sec l 1m
;= 0.11
gives
=    = 43 lS8rad/sec
143_24 3.3166
r'  6.7916  99 =0 .
r'
(Ans.)
13.909 14324
1'= 
W
w
= 3.73
~
Then
k = w 2m
(43.188)'(4.663) = 8698Jb/JO
0.10
(b) Assume
w==38.40= / 3.73 \ m
= =
T' _
= 2/ 6876X4.663
\1O)
and
~=0.113=
40.45 358.14
[1%
w 1'==3.65
w
This calculated value, 11 % for lhe damping ratio, agrees with the last value tried. Therefore, the requked sprjng conswnt for the isolation is k 6876 tb/in as calculated above,
7$
77
3.8
and
(359)
In the preceding section, we determined the response of the structure to a ham10nic motion of its foundation. In this section we shall consider a sjmilar problem of vibration isolation; the problem now, however, is ~o find the force transmitted to the foundation. Consider again the damped osci;Iator with a hannonic force F(t)= Fo sin wt acting on its mass as shown in Fig, 3.2. The differential equation of motion is
mji
Then from eqs. (3.54) and (3.57), 'he maximum force AT transmitted to 'he foundation is
+ (.'); + ky = Fc sin wi
1 + (2fr)'
(3.60)
with the
stead)'~srate
In this case, the transmissibHity T~ is cefined as the ratio between the amplitude of the force transmitted to the foundation and the amplitude of the applied force. Hence from eq. (3.60) AT I 1 + (2fr)'Fo = Y (1  r')' + (2rfJ'
where Y and
e,
T,
(3.61 )
(3.54)
and
tan
()=
2fr
1=7
It is interesting to note that both the transmissibility of motion from the foundation to the structure, eq. (3.45), and the transmissibility of force from the structure to tbe foundation, eq, (3.61). are given by exactiy the same function. Hence the curves of transmissibiHty in Fig. 3.13 represent either type of transmissibility, An expression for the total phase angle 4J in eq, (3,57) may be determined by taking the tangent function to both members of eq. (3.59), so that
tan
The force transmitted to tbe support through the spr:ng is ky and tbrough the damping element is c~v. Hence the total force transmitted F T is
Fr=k:y+C)'
Differentlatjng eq. (3.19) and substituting 1n eq. (3.55) yield Fr= Y[k sin (iiJr 8)+cw cos (WI 8)] or
". 0 '" ' T= Y ~'k' F +cw" s;n(i5JlIJ+,..,J
(3.55)
Then substituting tan 0 and tanp, respectively, from eqs. (3.21) and (3.58), we obtain
(3.62)
(3.56)
(3.57)
in which
(3.58)
Example 3.9. A machine of weight W.= 3860 Ib is mounted on a sImple supported steel beam as shown in Fig, 3.15(a). A piston that moves up and down in the machine produces a harmonic force of magnitude Fo = 7000 lb and frequency liJ:=: 60 radlsec. Neglecting the weight of the beam and assuming 10% of the critical damping, determine: (a) the amplitude of the motion of the machine, (b) the force transmitted to the beam supports. and (c) the corresponding phase angie.
78
Response o!
OneOeg;ee~ofFreedom
79
the foundation is
(Ans.)
w
y
30,000 ksl
Ar=F,T,= lO,827lb (c) The corresponding phase angle fror:1 eq. (3.62) is
"1"'12010,"
(Ans,)
(,I
Ib)
Fig. 3.15 (;1) Beammachine system for Example 3.5. (b) !V1alhematlcal model.
Solution: The damped oscil1ator in Fig~ 3.1S(b) is used to model the system. The following parameters are cakula:ted:
3,9
SEISMIC INSTRUMENTS
rT
When a system of the type shown in Fig, 3J6 is used for the purpose of vibration measurement, the relative displacement between the mass and the base is ordinarily recorded. Such ar. instrument is called a seismograph and it can be designed to measure either the displacement or the acceleration of the base, The peak relative response U/Yo of lhe seismograph depicted in Fig, 3.16, for hannonlc mOlion of the base, ;$ given from eq. (3.53) by
{=OJ
r=
u
=0,6
y,
/'
.f (l :'~')' + (2r,fj'
(3.63)
Po 00' T= . lin
(al From eq, (3.20), Ihe amplitude of motion ;s
y= (7;==if~"""",'F= 0.,075 in
with a phase angle from eq. (3,21)
(Ans,)
A plot of this equalion as a funcllon of the frequency ratio and damping rat10 is shown in Fig, 3,17. It may be seen from this figure that the response is essentially constant fo. frequency rattos r> 1 and damping raLio ~;;;:: O.S. Consequently, the response of a properly damped instrument of this type is essentially proportional to the basedisplacement amplitude for high frequencies of motion of the base. The instrument will thus serve as a displacement :neter for measuring such motions, TIle range of applicability of the instrumenl is 1n
B= tan .. !
= 10,6'
1547
Fig, 3,16 Model of seismograph,
80
81
3.10
t
~
0.15
Example 3.10. Determine the steadystate response of the system in Example 1.6 subjected to a hannonic force, F 1000 sin wI (lb) applied at tbe mass In in the y direction, The force frequency w varies between 2 and 10 cps, Neglect damping.
t '" 0.4
Salurian: The analysiS ~s a continuation of Example 1.5, A harmonic analysis is performed over the frequency runge 2 to 10 Hz.
(15) Denne analysis type as hannonic using one natural frequency, units of exciting frequency tn Hz, frequency range from 2 to 10 Hz. 1(0) frequency points in the frequency range, linear interp0lation. and request printout of relative displacements and reiative velocities:
ANALYSIS
>'
POST.DYN
>'
?D_A'fY?E
lO
PD_ATYPE,
5, 1, 1, 2, 10, 1000, 1,
PD_G.iRTYP
(16) Define dynamic forcing function us frequency depended force over full frequency range: creased by. redudng the natural frequency, i.e., by reducing the spdno stiffness . a or mcreasmg the mass. New consider the response of the same instrument to a harmonic acceleration of the base Y$;;;; Yo sin Wt. The equation of motion of this system is obtained from eq. (3.50) as
ANA::"YSIS POS"l'_DYN;. PD_CURVES PD_elJR'!'YP!' 1, 1, 0
AKA:"YSIS
~ POST~DYN
:>
> PDCURVES
;. PDCURDEF
PD_CURDEF, 1, 1, 2,
{i;t
The steadystate response of this system expressed as the dynamic magnification factor is then given from eg. (3.23) by
STRUCTUR.L.~
;. FORCE ;. ?ND
FN'b, 2, FY, 1, 2, I, 2, 1, 0
(3.65)
(18) Request output for the amplitude of the displacement and of the acceleration in function of the frequency:
ANALYS:S
PD~PR::;:N7,
)<
This equation is represented graphically in Fig. 3.3. In this case, it can be seen 0.7, the value of the response is from this figure that for a damping ratio nearly constant in the frequency range 0 < r< OA Thus it is clear from eq. (~.65) that the response indicated by this instrument wi!: be directly proportlOnal to the baseacceleration ampIitude for frequencies up to about six .tenths of the natural frequency. Its range of appHcabiHty wUl be :ncreased by increas~ ing the natural frequency, that is, by increasing the stiffness of the spring or by decreasing the mass of the oscillator. Such an instrument 1S an accelerometer.
>'
PD_PRINT 1000, 1
(19) Request plots for displacement and for accelerarion at node 2 in function of the frequency:
P.NA!,YSIS
:>
POS_DYN
:>
PD.. _OUTPUT
:>
P::LNRESP
PD_NRESP,
fu~ALY$!$
1, 2, 0
:> P05_DYN > PD 1, 1000, 10, 0
OC~PU~
) ?D_PL07
PD_PLO~,
.' ..""
82
Structures Modeled as
a.
SingleDegreeofFreedom SYSlem
83
POS_DYN
:>
R.....!)YNAMIC
R_DYNAMIC
(21) Activate XY plot infonnarion for Y displacement at node 2 as a func~ tion of frequency, and pJot the displacement YS, frequency for node 2:
orSPI,AY
:>
XV_PLOTS
>
ACTXYPLOT
v
y
,I
, , ,
'i'\
,111
,i ,
,'II
H
1
"
f! 
to
'"
I,
i
(
L
7 : \
'.l.l
I ,
lJ,
'~,
:1
LI1
J.'6
L~
\i,\!:
""I.~
8.4
9.2
j$
F!l(C Ifll)
(a)
,of frequenc)', and generate plot of the acceleration vs. frequency for node 2, Then replot with acceleration range from 6000 to lE6:
lE'St,,~~~
::>
1,
0,
2N
s.!'i&S<<l>S!
  '_ _
1_ _
1
!:I {l12(+$$ ,
O:i:SPLAY > XY_PLOTS > XYPLOT XYPLOT, 1 DISPLAY XYRANGE. DISPLAY REPAINT
::>
~r
]
,
iT :
'
I
i
,:':::::::1=;~~
t
I
I ,'
\
~l
.
,
!I
y s
{l3~'{lSt+1'~~ I,
"~"
C{ll5t'!i~
4~
,!.
1
I, I I'
'
(24) Obtain a hard copy of the frequency response for displacement and acceleration:
CONTROL j LASERJET, DEVICE > LASERJET 150,0
t t2 ~,
!
:
! \ j  I L., : +'..
, _ I __
~'
. , I, . ,"=,.~,
'~
Ii
i' u
FRSI)!'U)
Figure 3.18 shows the frequency response in terms of the displacement and of the acceleration,
Example 3.11. Solve Example 3,10 assuming that the damping in the system is equal to 10% of the critical damping.
,
\
I
(b)
t.
';',
Fig. 3,18 Frequency response for Exnnlp1e 3.10. (il) Amplllude of the (b) Amplitude of tbe acceleratiol:.
d~splacemem;
" !;
~'
84
85
Solution: The analysis is a continuation of Examples 1.6 and 3.10: A harmonic analysis is performed over the frequency range of 2 to to Hz. 10% of critical damping is assumed. The following commands in COSMOS follow those of Examples 1.6 and 3.10: (25) Define modal damping for mode 1 as 10% of critical damping (0.10):
_~ALYSIS
'"
'"
'" '"
i
i
'\
) POST_DYN
1,
>
PD_DN1P!GAP
>
PD_MDAMP
,"
'" , '"
PD_MDAMP I L L
.1
I \ I \
!I
II
\
1\
I
(27) Activate XY plot information for X displacement at node 2 as a function of frequency, and plot the displacement vs. frequency for node 2. Then replot with displacement range set to 200 maximum:
DISPLAY ) XY_PLOTS ) ACTXYPLOT ACTXYPOST, 1, FREQ, UY, 2, 0, 12, 1, 0, 2N DISPLAY XYPLOT XYPLOT, 1 DISPLAY ) XV_PLOT XYRANGE XYRk~GE, I, 1, 2, 10, 0, 200 DISPLAY ) VIEvCPAR ) REPAINT REPAINT
, ,
IL
.\
'6.6 6.4
,
,
FRE01H,1
9.2
1$
7.6
(a)
2E('l~
(28) Activate XY plot infonnation for X acceleration at node 2 as a function of frequency, and plot the acceleration vs. frequency for node 2. Then replot with acceleration range set to 2E5 maximum:
DISPLAY ) XV_PLOTS ) ACTXYPLOT ACTXYPLOT, 1, FREQ, AY, 2, 0, 12, I, DISPLAY XYPLOT XYPLOT, 1 DISPLAY ) XY PLOT XYRANGE XYRANGE, 1, 1, 2, 10, 0, 2E5 DISPLAY REPAINT 0, 2N
I
I
(\
, ,
lE+~S
1
8G.:Il.:ll!l
I I II
/
I \
1\
."
6.8
6SGGS 4GGGS
I'
=9.2
lG
2GSn
Figure 3.19 shows the frequency response plot for Example 3.11 in terms of the amplitudes of the displacement and of the acceleration.
,~
2 2.8 3.6
~
.4
5.2
FREOIH,)
7.6
8.4
Example 3.12. Detennine the steadystate response of the system of Example 1.6 when a vertical harmonic acceleration of magnitude 0.3 g and frequency UJ in the 2 to 10 cps is applied at the support. Neglect damping. Solution: The analysis is a continuation of Example 1.6. A harmonic analysis is perfonned over the frequency of 2 to 10Hz. The following commands in COSMOS follow those of Example 1.6:
(b)
Fig. 3.19 Frequency response for Example 3.11. (a) Amplitude of the displacement; (b) Amplitude of the acceleration.
86
87
(15) Define analysis type as hannonic using one natural frequency, units of ex.citing frequency as Hz, frequency range from 2 to [0 Hz, 1000 frequency points in frequency range, linear inrerpolalion, and reque..')t relative displacement and relative velocity response printout:
ANALYS:::S
J
. ,
:
"
POST_DYN )
?D_ATYPE
, ,
(16) Define dynamic forcing function as frequency dependent base acceleration in the Y direction:
PD_CURTYP, M1"ALYSIS
,
, ,
J.
i
ANALYSIS
?D_CURTY? PD_CURDEF
"
,
;
fJ~1
!
I
=t=
ANALYSIS ) POST_JYN
PD_BEXC'!) PD_BASE
0
PD_BASE, 1, 0, 386.4, 0,
" , ,
, ,.
./
,.,
'
J.
..
"
).
I !
PD._OUTPUT PD_PRINT
0, 10, 1, 1000, 1
>
1, 0, L
0,
(a)
ANALYSIS ) POS_DYM > PD OUTPUT PD_PI,OT, 1, 1COe, 10. 0 (18) Execute harmonic analysis; AJ,"VALYSIS
FLJYNAMIC
:>
PO_PLOT
POST_OYN' ) R_DYNAHIC
(19) Activate XY plot information for Y displacement at node 2 as a fune tion of frequency, and plot the displacement vs. frequency for node 2. Then replot with displacement range set to 10 maximum:
DISPLAY ) XY~PLOTS ) ACTXYPOST ACTXYPOST, 1, FREQ, UY, 2, 0, 12, 1, 0, 2N
DISPLAY ) XY_PLOTS ) XY~~GE XYRANGE, 1, 1, 2, 10, 0, 10
, , ,
REPAINT
(20) Activate XY plot information for Yacceleration at node 2 as a function of frequency. and piot the acceleration vs. frequency for node 2. Then replot with acceleration range set LO 10,000 maximum:
DISPLAY ) XY_?LOTS ) ACTXYPOST ACTXYPOST, 1, FREQ, AY, 2, 0, 12. 1, 0, 2N DISP~AY , XV_PLOTS > XY~~GE XYRANGE, 1, 1, 2, 10, 0, 10000 REPAINT
(b)
Fig:lre 3,20 shows the frequency response for Example 3.12 in terms of the amplitudes of the cisplacement and of the acceleration,
Fig. 3.20 Frequency response. fer Example 3,12, (a) Amplilude of the displacement;
88
Response of
One~DegreeofFreedom
89
Example 3.13. Solve Example 3.12 assuming that the damping in the system is equal to 15% of the critical damping.
Solution: The analysis IS a continuation of ExampJe 3.12. Damping is increased to 15% of critical damping.
(21) Define modal damping for mode I as 15% of critical damping (0.15):
]\l\AL'fSIS > POST_DYN > P;:}_DAMP iGAP ?D_HDA.,'\J?, 1, 1, 1, 0.15
O.J:O;
>
POST_DYN
>
R_D~NAMIC
y " , ,
,.,
1 .. 25
/
II
:
'\
,
I
I
'.
, .,
9.1S'
I
L
/1
!
.\
(23) Activate XY P!ot information for Y displacement at node 2 as a function of frequency, and plot the displacement vs. frequency for node 2. Then replot with displacement range set to 0.5 maximum:
DISPLAY> XY_?[,OTS ) AC'rXYP['OT ACTXYPIJOT, _, FREQ, UY, 2, 0, 12, DISPLAY > XV_PLOT ) XYRA~GE XY?.ANGE, 1., 1, 2, 10, 0, 0.:) DISPLAY > VIEW~PAR > REPp_IN1' REPAINT
.!"
,,1III.SS
'" II
!
I
,
0, 2N
., , ,.,
i
FfIQHhl
'.'
'iL7
"
(a) .......~.~..:~,
(24) Activate XY plot information for Yacceleration at node 2 as a function of frequency, and plot the acceleration vs. frequency for node 2. Then repiot with acceleration range set to 500 maximum:
DISPLAY:> XY_PLOTS > ACTXY?LO'f ACTXYPLOL l, FREQ, AY, 2, 0, 12, 1, 0, 2N DISPLAY :> XY_P:01' :> XY?JL~GE XYRANGS, L L 2, 10, 0, 500 DIS?LAY :. VIE~CPAR > REPA:)lT REPAINT
y "
, ,
Figure 3.21 shows the frequency response for Example 3.13 in terms of (he amplitudes of the displacement and of the acceleration.
3.11
SUMMARY
In this chapter, we have determined the response of a single degreeof~freedom system SUbjected to harmonic loading. This type of loading is expressed as a sine, cosme, or exponential function and can be handled mathematically with minimum difficulty for the undamped Or damped structure. The dIfferential equation of motion fo: a linear single degree'offreedom system is the secoodo:der diffe:e:1tJai equation
(b)
my + cy + ky = Fo sin w!
eq. (3.10)
Fig. 3.21 Frequency response for Example 3,13. (a) Amplitude of the displacement; (b) Amplitude of the acceleration.
90
91
or
a y + 2~wY + w Y = ~ Slfi
2
with respect to the support, In this latter case, tbe equation assumes a much simpler and more convenient form, namely
F m
._
wt
mil
+ cu + ku.
Fef! (l)
eq. (3.50)
in which
w to
in whkh
F.;.:r(t)
m)'J(t)
Ii
[i,
r;;W
= /~ It!
The ueneral solution of eq, (3.10) is obtained as the summalion of the com
plem~n\ary
(transient) and
y
~he
particular
(steady~slate)
J (1
soh1l10ns. namely
= e ,~ (A
,~
cos
WDI
~
+B
.
Slfl
+ (lrt;)
_____
____
~J
~~'
transient solution
in which A and B are constaf'lt.5 of integration,
steadystate solulion
For harmonic excitation of tbe foundation, the solution of eq, (3.s0) in terms of the relative motion is of the same form as the solution of eq, (3.10) in which the force lS acting On the mass, In this chapter, we have also ShOWli that the equivalent damping in the system may be evaluated experimentally either from the peak amplitude or from the bandwidth obtained from a plot of the ,1mpJitudefrequency curve when the system is forced to harmonk vibnuion. Most commonly, equivaient viscous damping is evaluated by equating the experimentally measured energy dissipated in the system during a vibratory cycle at the resonant fret:;uency to the theoretically calculated energy that the system, assumed viscously damped, would dissipate in a cycle_ This approach leads to the following expression for the equivalent viscous damping:
E~
=
w"
and
wJ 1 
in which
It = energy dissipated in the system during a cycle of harmonic vibration at reSOnance E~ strain energy stored at max;mum displacement jf the system were elastic r = ralio of forced vibration frequency to the natural frequency of the system.
Two related problems of vibrating isolation were discussed in this chapter:
(1) the motion transmissibility, that is, the relative motion transmitted from the foundation to the structure; and (2) the force transmissibility which is the relative magnitude of the force transmitted from the structure to the foundation" For both of these prob!ems, tbe transmissibility js given by
'i
The transient part of the solution vanishes rapIdly to zer~ because ~f lhe negati ve exponential factor leaving only the steadystate SOIU1l0D: Of partlcui~r significance is the condition of resonance (r = iiJlw =:. 1) for whlch the ampltttldes of motion become very large for the damped system and lend 10 become
.. infinity for the undamped system. The response of the structure 10 support or foundaHo.n motl~n can, be obtained in terms of the absolute motion of the mass or of tis relatlve motJon
I I
T< \
1 (1
~1+(2rt;)'
r'l' + (:irt;)'
,I
III
92
93
PROBLEMS
3.1 An elec(;ic motor of total weight W = 1000 Ib is mounted at (he cemer of a slm;!ly supported beam as shown in rig. P3.L The unbalance in the rolOr is W' e "" I lb in, Determine the ste<1dystate amplitude of vertical motion of rhe motor for a speed of 900 rpm, Assume that the damping in the system lS 10% of the critical damping. Neglect the mass of the supporting beam.
3.7
the stiffness of the isolation springs required to reduce the vertical motion amplitude of the instrument to 0,01 in. Neglect damping, Consider the water tower: shown in Fig. P3.7 which is subjected to ground motion produced by a passing train in the vicinity of the tower. The ground motion is idealized as a harmonic acceleration of the foundation of the tower with an amplitude of 0.1 g af a frequency of 10 cps. Determine the motion of the tower relative to {he motion of its foundation, Assume an effective damping coefficient of 10% of the cri~icai damping in the system.
Fig. P3.I.
3.2 3,3
Delenn~:)e
Prob~
lem 3.1.
Detennine the steadystate amplitude fOf the horizontal motion of the steel frame in Fig. P3.3. Assume the horizontal girder to be infinitely rlg:d and neglec: both the mass of the columns and damping. critical damping,
3A Solve for Problem 33 assuming that the damping in the system is 8% of the
Fig, P3,7,
3.8 3.9
15'
W10X 33
~
Fig. P3.3.
Determine the transm;ssibiUty in Problem 3.7, An electric moror of total weight W"'" 3330 lb is mounted on a simple supported beam with overhang as shown jn Fig. P3,9, The unbalance of the rotor is W' e 50 lb ' in. (a) Find the amplitudes of forced vertical vibration of the motor for speeds 800, 1000, and 1200 rpm, (b) Draw a rough plot of the amplitude versus rpm. Assume damping equal to 10% of the critical damping_
3,5
3.6
For Probiem 3.4 determine; (a) the maximum force transmitted to (he founda!iotl and (b) the transmissibility,
A delicate instrument is to be spring mounted to the noor of a test laboratory where it hes been determined that the floor vibrates vertically with harmonic motion of amplitude 0, I in at 10 cps. [f the instrument weighs 100 lb, determine
;:g;;:
= 10'",
~'L
,c2.5'~2.5'~
Fig, 1'3,9,
I
,1
94
95
system,
ratio, (e) the amplitude of the exciting force when the peak amp:itude of the vibrating mass :s measured to be 037 In, and (d) fhe amplitude of the exciting force when the ampJimtle measured is at lhe peak frequency assumed to be the resonant frequency.
3.11
Determine the damplng in a system in which during a vibration lest under a harmonic force it was observed that at a frequency 10% higher than the resonant frequency, the displacement amplitude was exactly onehil.if of the resonant amplitude.
r4>flt)
Fig. P3_15.
Fig. P3.12.
3.12 Detennine Ihe natural f,equency, amplitude of vibration, and maxilnum normal stress in [he simple supported beam carrying an engine of weight W = 30 xN (Fig. P3.! 2). The engine rotates at 400 rpm and induces a vertical force
3.16
F(:)=Ssinwr(&>""c;400/60(21t) 41.9rad/sec). (E=210X10 9 (N/m\ /= 8950 x W s (m4) Sect. f!lOdulus. S = 597 X 1O6m J.} (Problem contributed by Professors Vladimir N. Alekhin and Aleksey A. Amipin of the Urols State Technical Universfty, Russia,) 3.13 A machine of mass m rests on an elastic floor as shown in Fjg. P3, l3. In order to find the natural frequency of the vertlcal motion, a mechanical shaker of mass nt: is boIled to the machine and run at various speeds until the resonant frequency II is found. Detennine the natura! frequency I~ of the floormachine system in tenns of fr ar:d the given daHL
A s;:ruclmal systeCl modeied as 11 damped oscillatOr is )';ubjecled to the harmonic excit..tion produced by an eccentric rOLOr. The spring eon;;!ant kane the mass m are known but nO( the damping and :he amount of unbalance in the rotor. From measured amplitudes Y, al resonance and Yl at a frequency ralio rl "'" 1, determine expressions to calculate the damping ratio and the amplitude of the e:r.:citb& force F, at resonance.
3,17
A system is modeJed by two vibrating masses nil and III; interconnec(cd by it spring k and damper element c (Fig. P3.!f). FOr nannonic force F= Fa sin w( acting on mass /Il2 determine: (a) equation of motion in [crms of the reJative motion of the (wo masses, U )'l  y:; (b) the slcacYwstate solution of the relative motion,
~r' y,
Fig. P3.17.
Fig. P3.13.
3.~4
Detenn:::e the frequency at which the peak amplitude of a dumped oscilblor will occur. Also. determine the peak llmplitude and corresponding phase angle. A sLructure modeled as a damped springmass sys,em (Fig, P3.IS} with mg=2520Ib, k~'89,OOOlblin> and c=lJ2Ibsec/in is subjected to a harmonic exclting force. Detennine: (a) the natural frequency, (b) the damping
3.15
97
4
Response to General Dynamic Loading
Fir)
Fig. 4.1
=F(r)
Rearrangement yieJds
du
dr = .:...cc":,:,:,:,,
(4.1)
In the preceding chapter we studied the response of a single degree~of~freedom system with harmonic loading, Though this tYpe of loading is important, real structures are often subjected to loads that are not harmonic. In the present chapter we shall study the response of the single degreeoffreedom system to a general type of force. We shall see that the response can be obtained in tenns of an integral that for some simple load functions can be eva1uated analyticaUy. For the general case, however, it will be necessa.'1' to resort to a numerical integration procedure.
where F(r) d1' is the impulse and du is the incremental velocity. This incremental velocity may oe considered to be an initial velocity of the mass at lime r. Now let us consider this impulse F(r) dr acting on the structure represented by the undamped oscillator. At the time r the oscHlalor will expedence a change of velocity given by eq. (4.1), Thi_$ change in veiocity is then introduced in eq. (1,20) as the initial velocity DiJ together with the initial displacement Yo;::;: 0 at time r producing a displacement at a later time t given by
dyer) _
F(r) dr .
mw
....  sm "'(I  r)
(4.2)
4.1
The loading fl;nction may then be regarded as a serres of short impulses at successive incremental times dr, each producing its own differential response at time I of the form given by eq. (4.2). Therefore, we conclude that the total displacement at time t due to the continuous action of the force F (r) is given by the summation or integral of the differential displacements dy (t) from time t = 0 to time t, that is.
yCt) =1 
mw ,
I'
(4.3)
98
StruClures Modeled as a
Sjngle~De9feeofFreedorr.
System
99
The integral in this equation is known as DHhamei's inregrai. Equation (4,3) represents the total displacement prod'Jced by the exciting force F (1') acting on the undamped oscmatOr~ it includes both the steadystate and the transient
= O. If the function F(T) cannot be expressed analytically, the integral of eq. (4.3) can always be evaluated approximately by suitable numerical methods.
2 I
  


To include the effect of initial displacement )'0 and initial velocity Dc at ~ime t 0, it is only necessary to add to eq, (4.3) the solution given by e.q, 020)
for the effects due to the initial conditions, Thus the total displacement of an undamped sing!e degreeoffreedom system with an arbitrary load is given by y(t).,,;yocos
Fig. 4.3 Response of an undamped Single applied constant force.
degreeof~freedom
system
(0
a suddenly
wt+~sin (JJ{+~J
w
"
mw ()
(4.4)
Applications of eq, (4.4) for some simple forcing functions for wbich it is possible to obla;n the explicit integration of eg. (4.4) are presented below.
4.1.1
Constant Force
Consider the case of a constant force of magnitude Fo applied sudc!enly to the undamped oscillator at time f:::':" 0 as shown in Fig. 4.2. For both initial displacement and initial velocity equal to zero, the application of eg. (4.4) to this case gives
y(/)=~ I
where y~! = Folk. The response for such a suddenly appUed constant load is shown In Fig. 4.3. Il will be observed that this solution is very similar to the solution for the free vibration of the undamped oscillator. The major difference is that the coordinate axis! has been shifted by an amount equal to YSI Folk. Also, it should be noted that the maximum displacement 2Y~l is exactly twice the displacement that the force Fe wou:d produce If it were applied statically. We have found an elementary but important TesUIt: the maximum displacement of a linear elastic system for a constant force applied soddenly is twice the displacement caused by the same force applied statically (slowly). This result for displacement is a;so true for the internal forces and stresses in the structure.
mw
f'
Fes!n W(lT)dT
4.1.2
Rectangular Load
Let us consider a second case, that of a constant force Fo suddenly applied but
and integratlon yields
y(l) ~
only during a lim~ted time duration Iff as shown in Fig. 4.4. Up to the time l", eg. (4.5) applies and at that time the displacement and velocity are
F, . cos ,!
InW
w(t 
7) 0
I'
Y (/)
(1k
Fe
cos
Wf)= y,,(l
cos
Wi)
(4.5)
and
F(ll
j
~c
For the response after time ld we apply eq. (1,20) for free vibration, taking as the initial conditions the displacement and velOCity at td. After replacing I by t  Id, and Yo and Uo by Yd and u", respectively, we obtain
1'1
y(t) =
100
~01
2.0:,nn"'I'Iri,7'TTlT,ril,
in I i i'''ITn I
L6H+t't+:
.J ~
HH~~r.~+i4+H+~L~~itH
~!/+IHH+li+;I'ili,Hi"m '
f3= 2 Sin    sm   2 2
a+{3 ,
a{3
s o'! .JHftttHiHtlI't'ttI' :/ 1
'I
i ,i I ' ~H+~~+r'i~++H
J4+J,f+,+'!+
Li
I .
"'4 \ . [ (,
(4.8b)
Fig. 4A
Maximum dynamic load factor for the undamped oscillator aCled upon by a
rectangular force.
tJ ) 
cos wt }
(4.6)
If the dynamic load factor (DLF) is defined as the displacement at any time t divided by the static displacement y" ~ Folk, we may write eqs. (4.5) and (46) as DLF=lcos(.o[, and DLF:::: cos w(t [d) [$(d
The use of dimensionless parameters in eq. (4.8) serves to emphasize the fact that the ratio of duration of the time. that the constant force is applied, to the natural period rather than the actual value of either quantity' is the important parameter. The maximum dynamic load factor (DI.F)m"H obtained by maxizlng eq, (4.8), is plotted in Fig, 4.4. It is observed from this figure that the max.imum dynamic load factor for loads of duration (.1 IT'?::.05 is the same as if the load duration had been infinite, In general, the maximum response occurs during the application of the load, except for loadings of very shor~ duration (t.1IT<O.4). In such cases, the maxinum response may occur during the free vibration after the cessation of the load; it is then necessary to extend the loadbg time for a durati<.m of about one period, in which the magnitude of the load is set equal to .zero. Charts. as shown in Fig, 4.4, which give the maximum response of a singledegreeof~freedom system for .a given loading function, are called response spectral charts. These charts are extremely useful for design purposes, as win be discussed in Chapter 8. Response spectral charts for impulsive loads of short duration are often presented for the undamped system. For short duration of the load, damping does not have a significant effect on the response of the system, The maximum dynamic load factor usually corresponds to the first peak of response and the amount of damping normally found in structures is not sufficient to appreciably decrease this value.
cos
Wl,
(4.7)
4.1.3
Triangular Load
It is often convenient lo express time as a dimensionless parameter by simply using the natural period instead of the natural frequency (w:::: 2r./T). Hence eq, (4.7) may be written as
We consider now a system represented by the undamped oscillator, jnitially at rest and subjected to a force F(t) that has an initial value Fo acd that decreases linearly to zero at time td (Jolg. 4.5). The response may be computed by eq. (4.4) two intervals. For the first interval, 7'~ la, the force lS given by
sn
DLF= I
and
DLF ~ cos 2
'\r
II
Ii \
Tf
cos 27T,
T
(4.8a)
Yo 0:;:;0,
uo=O
102
Structures Modeled as a
2.0
Sjng!eDegree~ofFreed()m
System
'!O3
I III I
I II i
' ! TTl
I
!!
0:
~
1. 2
U4 I
I
I
I
I
Fir!
9. o. 8
o.
lL Xl
J.+
I I I
I I
UL!
These values may be considered as the initial conditions at time r = Id for this second interval. Replacing in eq. (L20) t by t td and Yo and Vc. respectively, by Yet and ()d ,md no:ing that FC,) = 0 in this interval we obtain the response as
. WI  SIn
we;
 f J )' J
Teos
Fo
uJt
1_ .
IJ F'~
I
on
II i
I
0.2
.ill
,
I,
U' : i L1 ,o 5
II
~ '~ r
= Folk
I.
gives
DLF = ~[S1O
WI,
tvl ~ S10
(jJ
(I 
t d)
cos wt
(4.12)
Fig. 4.5 Mllxiffium dynamic loud fnctor for {he u;tdamped oscillntor acted upon by triangular force,
In terms of the dimensionless time pantmeter, this last equation may be written as
DLF =
The substitution of these values in eq. (4.4) and integration gives
"
'10
2,, SIO
T
' .
'
2,,1\T
}iT.
cos
2,,T
(4.13)
(4.9)
DLF
2m,dr
(4.10)
The plot of the maximutn dynamic load factor as a function of the relative time duratiot'. (dlT for the undamped oscillator is given in Fig. 4.5. As would be expected, the maximum va!ue of the dynamic load factor approaches 2 as lalT becomes large; that is, the effect of the decay of the force is negligible for the rime required for the system to reach the maximum peak. We have slUdied the response of the undamped oscillator for two simple impUlse loadings: the rectangular pulse and the triangular pulse. Extensive charts have beer, prepared by the U.S. Army Corps of Engineers I, and are available for a variery of other loading ;Jll]ses.
ld_
I d),
we
Eo (Sin
k
wlt!
Wid
'I
!
Example 4.1. A oneslUry building, shown in Fig. 4.6. is idealized as a ISft high frame with twO steel columns fixed at the base and a rigid beam supporting a weight of W 5000 lb. Each COlumn has a :nOment of inertia Ix;::; 69.2 inA and a section modulus S = IJ:1c = 17 in) (E = 30 X 106 psi). Determine the m.2.ximum response of the frame to a rectangular impulse of ampli~ tude 3000 Ib and dUHltior: id = 0.1 sec appIJed horizontally on the top member of the frame. The response of interest is displacement at the top of the frame at'.d the bending stress in the COlumns.
and
Vrl=
Fo I'.
k \
wsm
wl,'+I
Cd
COS
wt,:
1\
(4.11)
1,1 1
j U.S. Army Corps of Engitleers, Dcsig" oj SUilcfw,:s fO ResiSf the EffeCf:! Of ATomic Weapons, Manunls 415, 41$, 3nd 416. Maret: !5, 1957: Mnnuals 417 lind 419, Janury IS, 1958; Manuals 41$,420.421, January 15, 1960.
104
105
F"fZ'Z===Zj
4.2
In many pract:caj cases the appUed loading function is known on]y from
experimental data as in the case of seismic motion and the response must be evaluated by a numer:cal method. For this purpose we use the trigonometric identity sin <(/ T) = sin wt cos WT cos wt sin (;)7, in Duhamel's integraL Then, assuming zero initial conditions, we obtain Duhamel's integral, eq. (4.4), in the form
1 y(t) =sin wr
W Sx20
..;;
Fig. 4.6
Idealized frame
fOf
ExampJe 4.1.
mco
f'
Solwio":
Natural period:
12X30X IO"X2X69.2
(IS X 12)'
or
wi 
(4.14)
m= 5000 = 12.95341b.sec'lin
386
~
=
J:
T= 211' '
J\zfaximum displacement:
~
= 0.2446 sec
(4.15)
:i. T
(DLF)~,
=~ .. = OA08
0.2446
= 1.9
(from
4.4)
The calculation of Duhamel's integral thus requires the evaluation of the integrals A (0 and B (I) numerically. Several numerical integration techniques have been used for this evaluation. In these techniques the integrals are replaced by a suitable summation of the function under the integral and evaluated for convenience at n equal time increments, liT. The most popular of these methods are the trapezoidal rule and the SImpson's rule. Consider the integation of a general function J(T)
A(I)=
y" =
Y."
Fa
3000
(Ans.)
'.
f'
I(T) dT
(4.16)
and for Simpson's rule
A (I) = L!rj(l, + 4/,
.
l>f = 6El
:r Ym"'
= 6 X 30 X
0m~~
(4.17)
(1'=
M s
15,083 pSI
(Ans.)
where n ~ l/1J'T must be an even number for Simpson's rule. The implementation of these rules is straightforward. The response obtajned will be approxi
'. ':.;
106
Response 10
107
I
F(r'>1 1
where
t1F, = F(I,)  F(lI_ ,)
"~~~
and
;
F(t,l
IF;
I I I
L1I'=(;[;_1
The substitution of eg. (4.20) jnlo eq_ {4.18) and integration yield
."'~
;
I I
1
I
"
F!r)
_:'~_._L~_"~ _~
I I I
________
f,_ 1
r,
t"
i
(4.2])
:l(,~l
II
I
!
i
mate since these rules arc based on the substitution of the function 1 (r) for a piecewise linear function for the trapezoidal rule, or piecew~se parabolic function for Simpson's rule. An alternative approach to the evalualion of Duhamel's integral is based on obtaining the exact analytical solution of the integral for the loading function assumed to be given by a succession of linear segments. This method does n?t introduce numerical approximations for the integration olher than lhose inherent in the round off error, so in this sense it is an exact method. In using this method. it is assumed that F (r), the forcing function. may be approx1mated by a segmentaHy linear function as shown in Fig. 4,7. To provide a complete response history, i~ is more convenient to express the integrations in eg. (4.15) in incremental form, namely
A(r,)=A(I,_,)+
L1F;
w(!, cos Wi j
(i_I
cos
Wi_I)}
(4.22)
Equations (4.21) and (4.22) are recurrent formulas for the evaluatlOO of the integrals in eq, (4,15) at any lime!:.:.;; I;.
Exnmple 4,2. Determine the dynamic response of a tower subjected to a blaslloading. The idealization of lhe stucture and (he blaSlloadJog are shown
(4.18)
F(r)
li.1
B(t,)=B(IH)+
f.,
(4.19)
K'" 100 kim.
where A (l,J and B(t,) represent the values of the integals in eq. (4,15) at time tl_ Assuming that the forcing function F(T) is approxin;ated by a piecewise linear function as shown in Fig. 4.6, we may write
Y,
{"
(4.W)
Fig. 4.8
'"
108
Structures Modeled as a SingleDegreeciFreedom SYSlem For this system, the natural frequency is
109
Solurion:
4.3
Since the loading is given as a segmented linear function, the response obtained using Duhamel's integral, eq, (i!.14), with the coefficients A (/) and B (t) detennined from eqs. (4.2]) and (.:!.22), will be exact The necessary calculations are presented in a convenient tabular fonnar in Table 4.f for a few time steps. The integrals in eqs. (4. J 8) and (4.19) are labeled L1A (t) and LlB (I) in this table, since
L1A(ti)=A(ti)
The response of a damped system expressed by the Duhamel's integral is obtained in a manner entirely equivalent to the undamped analysis except that the impulse F( r) dr producing an intral velocity dv = F(r) drlm is substituted into the corresponding damped freevibration equation. Setting Yo = 0, Vo = F(T) dTlm, and substituting I for t  r in eq. (2.20), we obtain the differential displacement at time t as
dy (/) = e '
,f",(IT)
F(r) dr
(4.23)
A(ti~:)=)
~
,.[;
1,_,
Surruning these differential response terms over the entire roading interval results in
y(i) =_1_
and
LlB(li) B(I,)
B(li~l)
"
{'., F(r) sin wrdr
mwo
Jo
F(T)e!<4(Ii)sinwD{l
r)d7
(4.24)
Since the blast terminates at t = 0,060 seCt the values of A and B remain constant after this time. Consequently) the free vibration that foHoVots is obtained by substituting these values of A and B evaluated at t = 0.060 sec into eq. (4.14), that is,
y (I)
which is the response for a damped system in terms of the Duhamel's integraL For numerical evaluation, we proceed as in the undamped case and obtain from eq. (4.24) e' "" (4.25) y(t) ~ (AD(/) sin WD' BD(t) cos WDI}'mwo where
AD(ti)=AD(t;~I)~'
r.,
(4.26) (4.27)
or
Y(I)
BD(t;)=BD(I;~I)+
0.8130 sin 31.621 Ll338 cos 31.621
r;.,
For a linear piecewise loading functIon, F(r) given by eq. (4.20) is substituted into eqs. (4.26) and (4.27) which requires the evaluation of the following integrals: (4.28)
TABLE 4.1
I (sec)
Fer)
0 120,000 ]20,000 0 0 0
WI
L1A (I)
A (I)
LlB (t)
B(I)
Y (I)(in)
0 0.078 0.512
1.134 1.395
(4.29)
omo
0.040 0.060 0.080 0.100
0.000
a
0.6324 1.2649 :.8974 2.5298 3,1623
0 1082
1376
113 0 0
0 486
2404 3585
(430)
3585 3585
1.117
(4.31)
~
'i
ii;
110
Struclures Modeled as
a.
SingleDegreeofFreedom System
111
where It and I~ are the integrals iIldicated jn eq,s.. (4.28) and (4.29) before their evaluation at the limits. In terms of these integrals, Ao(fj) and Bn(ll) may be evaluated after substituting eq. (4.20) into eqs. (4.26) and (4.27) as
J ~i' _ .. yl')
tlF, \
tlF,
tl', [,
(4.32)
(3)
(4.33)
Finally, the substitution of eqs, (432) and (4.33) into eq. (4.25) gives the displacement at time Ii as
e (~i Y Ct i ) =   {Ad!;) sin
m(liD
Waf, ~
(4.34)
(bl
4.4
The differential equation of motion for a one.degree~offreedom system repre~ scored by the damped simple oscillator, as shown in Fig. 4.9(a), is obtained by establishing l:he dynamic equiHbrium of the forces iIi the free body diagram, Flg. 4.9(b):
my + cj + ky = F (I)
(4.38)
I
m
rill
in which the function Fet) represenlS the force applied to the mass of the oscillator. When the structure, modeled by the simple oscilhHor, is excited by a motion at its support, as is shown in Fig. 4. t O(a), the equation of motion obtained using the free body diagram in Fig. 4.9(b) is
(4.36)
:;;;;;J)~ffi/ffi/
bl
In this case, i: is convenient lo express the displacement of the mass relative Lo rhe displacement y.. of toe support, namely
u=yy,
The substitution
Ii
(4.31)
and its derivatives frorn eq. (4.37) into eq. (4.36) results in
(4.38)
Comparison of eqs, (4.35) and (4.38) reveals that both equations are mathematically equivalent if (he r:ghthand side of eq. (4,38) is interpreted as the effective force (4.39)
Fig. 4.9
diagram.
(a) Damped simpJe oscillator excited by Ihc force F(1). (b) Free body
112
Structures Modeled as a
ShgleDeg~eeofF::"eedom
System
113
On the other hand, the panicular solution of eq. (4.42) takes the form
(4.45)
Consequently, the solution of the second order differential equation (4.35) or eq. (4.40) gives the response in terms of the absolute motion y, for the case in which the mass is excited by a force> or in terms of the relative motion u =)' )'1> for the strucrure excited at its support.
II)
111 '\
4.4.1.
The method of solution for the differential equation of motion implemented in the computer program presented in this chapter is exact for an excitation function described by linear segments, The process of solution requires for convenience that the excitation function be calculated at equal time intervals LIt. This result is accomplished by a lineer interpolation between points defining the excitation, Thus, the time duration of the excitation, including a suitable extension of time after cessation of the excitation, is divided into N equal time intervals of duration For each interval .1t, the response is caiculated by considering the initial conditions at the beginning of that time interval and the linear excita~ion during the intervaL The initial conditions are, in thjs case, the displacement and velocity at the end of the preceding time interval. Assuming that the excitation function F(r) is approximated by a piecewise linear function as shown in Fig, 4.7, we may express this function by
(4.46)
FicA;
~C
.at.
The substitution into eq. (4.43) of li}e complementary solution Yc from eq. (4.44) and of the particular solution YP from eq. (4.45) gives the total solution as
(4.47)
The velocity is then given by the derivative of eq, (4.47) as (4.41) in which ti = i t3t for equal intervals of duration .11 and i 1,2,3" .. , N. The differential equation of motion, eq. (4.35), is then given by
e ......" .. 'i'[(WpOi,;'WC;)
 (WDC i
COs
WD(t'i)
+ ,;'wD i )
sin wD(tli)] + A,
(4.48)
t " ! 1<,
, Ii \ F+.~IF , \ Lli !
. ft
d,
(4.42)
The solut:on of eq. (4A2) may be expressed as the sum of complementary solution YC' for which the second member of eq. (4.42) is set equal to zero, and the particular solution )'Pl that is,
}' =}'~ +}'p
The constar.(S of integration Ci and Di are obtained from eqs. (4.47) and (4.48) introducing the initial conditions for the displacement Yf and for the velocity :If at the beginning of the interval .dt, that is, at time ti. Thus, introducing into eqs. (4.47) and (4.48) the initial conditions and solving the resulting relations yields
c,.= y; Bi
(4.49)
(4.43) The evaluation of eqs. (4.47) and (4.48) at time Ii + ilt results in the displacement Yi+1 and the velocity Yi+l at time !j,..!. Namely,
(4.50)
The complementary solution is given in general by eq. (2.15), which for the interval liSt:S:C;+ Lit is
Yc = e  {",(Ir,lle, cos
UJ[)(t 
t,)
+ D; sin
WD(t  (,)]
(4.44)
114
Structures Modeled as a
Single~DegreeoIMFreedom
System
115
and
Yi+1 efwd'(D,(wocos wD..1tt'wsin
waLll)
(4.51)
120
Fitl
Finally, L'1e acceleration at time 1; '1'1 t/ + Lit is obtained directly after subs.tituting and J',., from eqs. (4.50) and (4.51) into the differential eq. (4.35) and letting t""'" t/ + ..1r. Specifically.
y,.,
I
I
! ,
,
I I
. ,
I I
, ,
I
(4.52)
The substitution of the coefficient Ai, B" C i and D, from eqs (4.46) and (4.49), together with c ~ 2gklw into eqs. (4.50) and (4.5]), results in the following formulas to calculate the displacement, velocity and accelaration al the time step l{+ I = ti + Lk
0,02
0.04
Ibi
....;;,;,, thee)
0.06
Fig. 4,11
I
Ii
I
(4.53)
(4.54)
(4.55)
Equations (453), (4.54), and (4.55) are recurrence formulas to calculate, respectively. the displacement, velocity, and acceleration at the next lime step li+ I = Ii + Lli from the previously calculaled values fOf the these quantities at the proceeding time step f i Because these recurrence formulas are exact, the only restriction in selecting the length of (he lime step, Ll.r, is that it allows a close approximation to the excitation function and that equally spaced time intervals do not miss the peaks of this function. This numericai procedure is highly efficient because the coefficients in eqs. (4.53), (4.54), and (4.55) need to be calculated only once. Example 4.3. Detennine the dynamic response of a tower subjected 10 a blast loading. The idealization of !he structure and dhe blast loading are shown in Fig.4.1 J. Assume damping equal to 20% of the critical damping.
Solu.tion: Since the. loadir:g is given as a segmental linear function, the response obtained using the direct metbod will be exnct The necessary calculations are presented in a convenient tabular format in TabJe 4.2. For this system, tbe natural frequency is
UJ
!;
Sin wDClf
2f +   cos
wAr
Wodl
1+ 11 /~
2/; \ 1 J
wilt}
I
Jk/m
~ 30.984 lad/sec
31.623
,1,
116
Structu~es
Modeled as a
Si~gleOegree:)H~reedcm
System
117
TABLE 4.2
t, sec
Yi in
Yt in/sec
0 1O.692 25.155
Yi in Isec 2
0
990.754 430.768  1142.511 982.581  522.555
F,lb
0
0.074 0.451 0.926 1.044 0.778
0
120000 120000 0
problem, stores the data in a file, and then calculates and prints the response at equal increments of time. Maximum values for the response are also calculated and printed. A description of the programs used in this book and a fonn to order them is provided in Appendix II. Example 4.4. Solution: Mass:
In
17.096
. 4.821
From Fig. 4.11, we have the following data: 100 X 1000 = 100,000 (lb . in)
0
0
. 20.191
Spring constant: k
0.16517 C' = 1.16755 X 10' D' = 6.1439 X 10' 0.61286 C"=7.60730X 10. 5 D'=8.9097X 10 5
Uu =
Natural period: T = 2"1T ~ :;:; 0.20 sec Select hme step for integration: Lit = T 11 0:;:; 0,02 sec
= 6.1439 X lO , X
1.2 X 10'
0.074 in
lU?:JT DA71"
~n.st:R
y, 8.9J24x IO'x 1.2 X 10'= lO.692 in/sec y, =  31.623'XO.0742x0.2X31.623x 10.692+ l.2X 10' 1100 = 990.754 in /sec'
Thus, completing the first cycle of calculations in the direct method of solution, Introducing the calculated values Yh Yb and y! into the recurrer::ce fommlas eqs. (4.53), (4.54), and (4.55), we obtain the response at time = 0.04 sec. The continuation of this process results in the response of this system as showr. in Table 4.2 up to 0.10 sec.
e\CJ':A~'lON
MASS
S?IHNG CQNSTAM'l'
c"
1'II1E STEP cO' :NTEGRA1'!ON
l.~A?
Ii "
.1)2
C " (;
'tItE
Excr'!',",?lON
'2
TIME
() OISC
;,;xcrT,~':'XC"
O.l)1C
120000.0:)
O.OC
0.150
:).00
VC".JX:
Ace.
(L
4.5
o.c,jO
0.0:'0
(1.000
0.074
0.000
aco
to.692
2'),,155 17.096
9!n.:;:n
43(;. 7~()
t1~:z.SU
~1Jl2.581
The computer program described in this section calculates the response of the simple oscillator excited by a time~dependent external force acting on the mass or by an acceleration applied to the support. The excitation is assumed to be piecewise linear between defining points. The response consislS of a table giving at equal increments of time the displacement. the velocity, and the acceleration of the mass for the case of the oscillator excited by a force applied to the mass. For the case of the osciHator excited at its support, the response is given in terms of the displacement and the velocity of the mass relative respectively to the djsp]ac,ement and velocity of the support ar:.d of the absolute acceieratJon of the mass. The program implements the direct integration method to calculate the response. The program has been written in an interactive mode with the user. It starts by requesting information about the data needed in the solution of the
0.451 O. 92~
L044
J.ns
4.821
<:O.Bl
25.215
'On.55S 13.248
42L'S~
c .l40
C.H;:)
O 1)0 O.HlQ
, m
;).l06
21.511
;J.!HO
O.PS
"0.S53 0.424
59!1.09S
1.()DS
5:<9. MIG
i;:<4.160
to. 2~"
11.260
:;'1.10$
1:.5'>'2
l1LJU
5.6n
"3D.4?7
w.x.
D=SP"~ACl':.'{N"l'
)'AX. VE;t.OCl'J'Y
;rs .22
~
><AX. Att!;;"e.:AAT!O!'l
!H2.S1
118
119
Example 4.5. Consider the tower shown in Fig. 4.11, but now subjected to a constant impulsive acceleration of magnitude y~ = 0.5 g during 0.5 sec applied at the foundation of the tower. Determine the response of the tower in terms of the displacement and velocity of the mass relative to the motion of the foundation. Also, determine the maximum acceleration of the mass,
Solution:
Mass: m
4.6
Spring constant: k
Program 3, the same as Program 2, gives the response of the simple osdliator to an excitation specified as linear segments between pOJn;s defining the loading function. Program 2 may be used for any excilation function. while Program 3 is restricted to certain specific excitation functions predefined in the program, A total of eight possible excitation functions are implemented 1n the program. There is also a provision for a ninth fUllction labeled as "new function" which may be imp:emented by the user The excitation functions in the program are s!~own graphically :n Fig. 4.l2 and given analytically by the following expressions:
E, (T); P(l)' sin fP(2)*T P (3))
E,(T)~P(l) E,(T)~P(l)*T
Damping coefficient: c
Te:O
(4.56) (4.57)
(4.58)
0.02 sec
;0
Time (se:~c,)_ _+..:S"pport Acceleration (g)
a
0.5
0.5
0.5
p(1If~"'~T
nc "
f 4 !Tl
Esln
E6
(n
J
Ii!
i I
T
.G2
1
P(1)
i
p\ll~
T ,
"'lex
\ACccU:AAn~
or \7AAVI'tY OR ?E:RO)
366
. , ~~1'(21'"'
P"l~
T
E/cn;>,:!OO
Z:XCI1'ATXOfl
1'(2)
pml2 P(21
(lj Gr~ph ilq. (>i.51)
0.500
0.50
:1.'>00
".AX.
2 For excitation at the supper:: Displacement :tnd velocity ue n.~~ative 10 the sup]Xlrt. whereas
the acceleration
)$
absoiute vall;.e.
Fig.4_12
func~ions
II
120
121
E,(T)
= P(l}*T
=P(l)*P(2) TIP (2)
Os T,;; P (2)
T>P(2)
(4.59)
= 0.20 sec
E,(T)=P(1)*(I
Os T,;; P(2)
T> P (2) OST5P(2)12 P(2)12<1'5P(2) T> 1'(2) 051'';;1'(2) T> 1'(2)
(460)
=0
E,(T) = P (I) * [2 *T IP(2)J
=2*P(I),[l
TIP (2)],
o
E,(T)=P(l)*sin [P(3) *71
Di?l)"T CATll;
(4.62)
:v.ss
SPRING CONS'!.' ..... ""!' PAJ;fING COi"f""lC!NT Tn!e S"!'P Of W'l'EGRA'l""l:)M RS?OMS K1.X. T!M ISOX (GRAVIn OR ZeRO)
PhRN:':'ER$ OF "7,;e XCITATICII,
<:(T)
~
o
E, (T) = P(l)IEXP [I' (2) * 71
" "" ,
m
100
l,OJONI 0
1'",,0
(4.63)
THAX
""
C
.0:\
.:l is
E, (T)
= !'lEW FU!'ICTION
C'll)SW!P,2}TPO)l
T">C
In eqs. (4.56) through (4.63) T is the time variob!e and P(I), P(2) and P(3) a;e cons~ants or parameters to be input as data i:l order to complete!y define the specific excitation function selected.
1\l001l0
)1)
Exampie 4.6.
O;nPJT RElhIL1'S:
in Fig. 4.13 subjected to the sinusoidal force F(l) Fa sin WI applied at its top for 030 sec, (b) Check results using the exact solution which in this case is <lvui:able in closed form. Neglect damping.
T;"'
DIS7:
VH.rA:.
Ace
0,(::)0
'~().H9
0.000
0.010
0.0:10 C .00':
0.016
o.oue
L1SS 5_ ,,$5
11. $70
o .O?t
52S.lSa
';S(L
~8J 6~0.10.j
Solwion:
0.030
(Lt:n
O.iHO
c,zn
O.
4S~
]~,
Mass:
III
:l.()SO
J.060
0.0;0
o.
Spring constant: k
1.c:n
l.J96
1.199 1. SS5
o.oan
0.090
" "
211.229HZ
16$
$1:1:.513
~16.Z&~
).67.611
~S20. 916 len] .9$0 1451 .He
:896 S~
O,Hl::>
O.H{> 0.120
::. !}IE
1.5.4;
13.C1<
10.~H
17(14
'50
.HO
0.9J7 O. JB5
1112
~:l56
o .DC
0.1'"
O.J.!)O 0.10
(1.1,70
17.;;.'5
61.205
~2
1:1S ~H
O.
l~O
::L 1\;0
(I.NC
0_210
, , nc
,
~3.U' 2.7a~
0.29: l m 171
.(0)
,
n.2?9 10.S.n
~SjI.H5
6B6. Hi )1.0112
I~H.l!),):
:lS~L~O,
'"
BH_732.
4l}
~)i!.0:;'$
:it}
lL60,
18.}g.j
2SU.269
].115_74:4
0.220 0.23:)
C . ./4:J
SC .2""
19.5B
)()94 .181
~JC9
2GO~
2. :31
m
.1Si;
L2:t
0.09>
'O}.2$t5
HS.EOl
Fig.4.13
0.2"0
lOS: .496
122
Stn;clures Modeled as a
O.21S0
0,,;10
SinglaDegreeof~F(aadom
l21,lS1l
System
123
LL?S
:L ~:>o 1. )98
~126.4116
US.574
(). ?:SC
o.n::;
0,300 C.310
},.AX . or SPLACEMElIlT'
4.213 4.674
4.711
Results shown in the above table are sufficiently close to corresponding values given by the computer in part (a) of this problem. .
0.02,
.. n
1;<12
">X.
"
'5$$.)2
Example 4.7. Solve Example 4.6 selecting the time step 111 equal to and 0.005. Then compare the displacements at time 1= 0.1, 0.2, and 0.3 sec with the response obtained in Example 4.5 using the exact solution of the differential equation,
om,
(b) The exact solution for the response of <1 simple oscmator to the sinusoidal force Fo = sin wt, with zero initial displacement and velocity, from eg. (3.8) is
Fo /. = "':::T ISin "
Y (t)
kmar \
wI
sm wll
W.
SolutiOfl: In solving the probiem using Program 3, it is only necessary to modIfy existing file by selecting option 2 (Modify Existing Data File) at the computer request for file information, The only modification necessary i:1 the dala is to change the time step to the values prescribed by this problem, The foHowing table, records the results provided by the computer and their comparison with the exact solution calculated in ExampJe 4.6:
where w is the natural frequency in rad/sec, W the forced frequency also in rad/sec, and Fo the amplitude of the sinusojdal force. Substituting corresponding numerical values for this example yields
100,000 I . Y (I) ~ ]00,000 100 (30)' \ SIn 30t
~
     "..
4t = 0.02 sec
Dispi. (in) 1.560 3.092 4.570
% error
4t
(in)
0.005 sec
% error
DlspL
% error
0.1
30 31. 6231 . \ sin 31.6231, J 02 0.3
10 (sin 30,
Example 4.8. A structural system modeled in Fig. 4.14(a) by the simple oscilJato( with 10% (~= 0.10) of critical damping is subjected to the impulsive load as shown in Fig. 4.14(b). Determine the response.
Solution:
Mass:
ttl
Spring
9000 sin 30t + 948.7 sin 31.623t
The evaluation of the response at specific values of time results in the ing table:
t
follow~
500011;
" 
 
Y (t) (in)
yet)
(in Isec) 2.9379  11.6917 26.0822
HI)
(in Isec 1)
 1466.51 2907.53 4298.04
o
(a)
0.1
Cb)
Fig, 4.14
fOT
Example 4,8
124
125
= 0.20 sec
0,20 sec
50,000; Ib, 0 S r :s; O. I sec, o 1> 0, I sec,
,1A;
0.2 sec.
PLAN::, Z, 0, 1
Y.ASS
SPRING CO:<S?A..'IT
Dh.'l?!NG C02F'F!ClN'f
l.CCCC
53.~" .C~
(4) Define element group I using the SPRING element fonnul"tion with 2
nodes:
PROPSETS ) EGROUP
EGROU?, L SPRING, 0, 2, L 0,
c
ii
=
~
'tun::
?:Sl'o.>.J$E O'_\X.
o.
0,
10('1')
PO) '1'
(H':'cP (2)
e
?tI~l<M2'l'O:R
0
~ao:Jo
1'>;>(2')
Ptl)
PARAlis:78R P (2;
,"
1000000
Att,
::0.1>)(
f:"O(;()
o. oct! O. no
0.040 O. J6C
!).:lee
O.CJO
C,OO"
(l,015
J.n"
~. 22~
0,
0,
0,
0,
0,
m;;;:
C,1)S
.691
O,27S
{j, 4~6
7.922
o .lDO
0.120
O,57{i
8.646 L39"l
_8,O!
0.140
0.51'
0,26$
lOa, 0,
0, 0, 0, '0, 0
c .1&0
(l.2:(0
c.on
.iL t&Z
;; .4~()
us
10
0.:'10
(8) Activate element group I (spring) and real constant 1 (k ~ 29,30 Ib/in), and generate one element along curve 1:
CONTRa!... > ,'1CTIVE ) ACTSET
l"..'<X. otS;>:,;>.CEY;;NT ,.
P.AX. VLOC:Ti '"
1,,';>10
1S .le
S9l_2C
4,7
(9) Activate e~ement group 2 (mass) and real constant 2 (m = 100 lb' sec!./
in), and generate one element at point 2:
CONTROL ~ ACTIVE ACTSE'J', EG, 2
ACTSET, He, 2
>
ACTSET
Solution: The analysis is pre:onned using a single spring element wIth one concentrated mass element The following commands are implemented:
Ii
126
127
07) Define modal damping for mode 1 <:s 20% of critical d4mping (0.20);
NODES ) NMERGE
NYERG?:, 1, 3,
1, 0,0001, 0,
0, 0
(11) Apply constraints in all degrees of freedom <'I.t node I, and <'!oil degrees of freedom except UX at node 2:
LOAD_Be STRUCTURAL DrSPLMKTS DPT, L AL, 0, L 1 D?':', ", ey, 0, 2, UZ, RX, RY,
, ,
,
RZ
DPT
i ,
i
(12) Set the options for analysis to exlract one frequency using the Subspace Iteration Method with a maximum of 16 iterations, and run the frequency analysis:
ANALYS1S
:>
POS':'_DY~
>
R_DYNAHIC
FREQ /BUCK
A_FREQUEKCY
(20) Activate XY piot information for X displacement at node 2 as a function of time, and plot lhe displacement VS. lime for node 2:
A_FREQUENCY,
1,
0,
0,
0,
lEOS,
0,
1806,
0,
G,
D,
0
>
ANALYS:S
PREQ/8UCK ) R_FREQUENCY
1, TIME, OX,
2,
12, 1, 0,
2N
R_?REQUENCY
DISPLAY
XYPLOT,
:>
XY_?LOTS ) XYPLCT
HESUL':'S
:R<;;:,157
1
LIST
:>
FREQr..IST ?requency
(cycles/sec} 5.03292e+OO Period (seconds; 1,98692e01
(21) Activate XY plot informalion for X acceleration at node 2 as a functio:1 of time, and plot the acceleration VS. lime for node 2:
CISPLAY ~ AC?XYPOST,
XY~PLOTS
> ACTXYPOST
1. TIME, AX,
2,
12, 1, 0, 2N
i.
!
D!SPLAY
XYPLO'!',
:>
XV_PLOTS ) XYPLOT
I'! ,.
(14) Define analysis type as modal time history using one natural fre~ quency, 1000 time steps starting at t = 0 with a time increment of 0,002 sec; use default values for all integration parameters and reques~ printout of relzrive displacement and absolute acceleration:
A."'ll,LYSIS
:>
Figure 4.15 shows the displacement and acceleralion response for the first 1 sec <Jfter the excitation is applied. Example 4.10. A water tower modeled as shown in Fig. 4.16{a) is subjected to the impclsive ground acceleration depicted in Fig, 4.16(b). Deter~ mine: (a) the maximum displacemcnt at t!:e top of the tower, ar.d (b) the maximum shear force at lhe base of (he tower. Neglect damping and take iii = 100 (rad/sec).
POST__ DYN
2, L
:>
FD_ATYPE
?D._ATYPE,
1000,
0,
0.002.
C,
0 S,
0.25,
(15) Define dynamic forcing function as timedependent force: ANALYS:S ~ POST_DYN PD_.CUR';'YI?, 1, 0, :) ANALYSIS :. POST~9YN
PD_CURCEF, 1, 1, 0,
II
P~_CURVBS
PD_CUR~BF
Solwion; The analysis is performed using a Single sprbg element with One concentrated mass e1emenL Tbe foHowing commands are implemented:
120000,
(1) Set view to the XY pla.ne:
0,
.02,
120000,
.O~,
.06, 0, 1, 0
DISPr,AY
VIE\,v, 0,
>
0,
AC?IVE }
1
ACTSE~
'":'C,
LOADS_&: :.
STRUCTURAl,
1, 2, 1
>
FORCES, FP'I'
FP'!',
2, f'X,
GRIJ
0, 1
:>
PLANE
128
Structures Modeled as a
SjngleDegree~of~Freedom
System
129
1.1)2
& .61
&.65
J
tI
)
I
I
, ...
 ....
I
i
2M1
, a(t)"'300 ain/3t (in/aec')
0.48
3
t:l,! :2
i,
R E
L
,
i
! ,
i , ....
S.96
tl,24
&
.~2
=1\\
\.
/" ..........
'+
0,
0.0
'i"!\";f;;1
(t2
"~'l
s .S
.,' '"
s.<
G.5
''
aId
(al
Fig.4.16
(a)
'50
I
i
"
no
A
I
I"'~
, , s
"0
/\
EGROUP
0, 2, L 0, 0, 0, 0
'"
2~tl ~8&'
I
/'""'
.,
1,. 'SPRING,
\
i
'...
''
nIJ IH>'
SSi'il
12SG .
V
3.1
"
.:,
0.'
TIME
.:,
, .1
3.a
. g
0, 0
(bl
Fig. 4.15 Computer plot for rhe response of Exarr:p!e 4.9. (a) Displaceme':'lt. (b) AcceleratioL
(8) Activate element group I (spring) and real constant I (k ~ 100,000 Ibl
in)~
CONTROL :> ACTIVE > ACTSET AC'ISE'L EG, 1 ACTSE'!, RC, 1 MESHING " PAR!>'LMESH ) M_CR M_CR, L L 1, 2, .3., 1
130
131
(9) Activate element group 2 (mass) and real constant 2 (n:::::: 155 Jb . sec 1 f in), and generate one element at point 2:
CONTROL AC7SET. ACTSE'!', MESHING
M~PT,
2, 2,
PD_OU'l'PU'T ,
PD_NRESP
NMERGE,
1, ), 1, 0,0001,
(11) Apply constraints in aH degrees of freedom at node 1, and all degrees of freedom except UX at node 2:
LOADSBe ;.. STRUCTURAL DPT. 1, ALL, 0, L 1
~
DISPLMNTS ) CP'r
(19) Activate Xl' plot infomuHion for X displacement at node 2 as a function of lime, and plot the displacement VS. time for node 2:
DISPLAY ;. XY_PLOTS ) ACTXYPQS,=, AC'l'XYPOST, 1, ':'J:ME, UX, 2, 12, 1, DJ:SPLAY ) XY_PLOTS > XYPLO'l' X,{PLOT, 1 0, 2N
RY, RZ
(12) Set the options for the frequency analysis to exJract one frequency using the S'Jbspace Iteration Method witb a maximum of 16 iterations, and n.l!1 the frequency analysis:
PREQ !BUCK > A .. FREQUENCY A_FREQUENCY, ~, S, 15, 0, 0, O. 0,
>
A..~ALYSIS
(Figure 4.17 shows the plot of the displacement a: node 2 as a function of time.)
(20) Request SCJn for maximum displacement in the X direction at node 2 from t = 0 to 0.2 sec.
A.:,"JALYS:r:S > POST_DYN ) PD_OUTPUT :> P:U1AXMIN PD_M..Ji.X:Hl';, L 1. 10, 2, 2. 0, 2 ANALYSIS } POS'T'~ .. DYN > PD_PREP.lJ..RB
PD_PREPARE, 1
lEC5,
0,
1E06,
Or 0,
0, C
>
R_FREQUENCY
R.]REQUENCY
(rad/sec} 8.032:ge+C1
Period (seconds)
7.82250e02
MaxilTlIJm displacement:
(14) De5.ne analysis type as modal time history llsing one natural frequency, 1000 time steps starting at t = 0 with a time increment of 0,002 sec; use default values for all integration paramelers: and request pri:1tom of relative displacements and relative velocities:
ANALYSIS ) POST_:)YI", :. PD_ATYl?E PD_ATYPE, 2, 1, 1000, 0, .002, 0, A,,), 0,25, {)
Type
Slep. no. 70 87
Value (in~
Positive Negative
0.18323
~0.18413
4.8
SUMMARY
PD_CURTYP,
ANA:'YSIS ;,
1, 2,
1,0,
1
>
POST~DYN
PD_CURVES
PD~Ct:RDEF
PD__CURDF,
.2,:3CO, 100, 0, 0, 0, 0, C
In this chapter, we have shown that the differential equation of motioc for a linear system can be solved for any forcing function in terms of Duhamel's integral. The numerical evaluation of this integra! can be accomplished by any standard method sHch as the trapezoidal or the Simpson's rule. We have
'32
S:ructu:es Modeled as a
Sir,gleDeg(ee~or~Freedom
System
,.,
~
.IS
,N'
Q .l2
u
R E l
a.ea
0 04
/\
i ,
i I
I
\
\
!
I
:
r
f;
Ffti1'1==============i I
W8X 24
20Kip~
~Y
'f.lJI4
e.l.iB
[sV r
i i
m
1\
I\
j
,
I II
I
3.12
+V mJ ;
\
Ii I Ii !I! i ; ,
!/ J!m~
~.18
$.12
oS .IS
\,
iil.2
Q
I
0.$2
rL~4
,
I
I
1!l.1i8
\/
.\
I ! .. , ,_
11
@.l4
,v,
:lLlG
~.~&
G,!
..,
'J
TIME
Fig. P4.l maximum horizontal deflectioIl. Assume the columns massless and the girder rigid, Neglect damping.
i<1g. 4.17
Displacement
re~ponse
preferred the use of a numerical integration by simply assumIng that the forcing function is a linear fU:1ctio:1 between defining points, 3:1d, on this basis, we have obtained the exact response for each time increment. The computer programs described in this chapter use the direct integration method. In this method, the differential equation of motIon is solved for each time increment for the conditions existent at the end of the preceding interval (initial conditions for the new interval) and for the action of the excita~ion applied during the interval, which is assumed to be Ii:1eaL Two programs were presented in this chapter: (1) Program 2 to calculate the response of a single degreeof~freedom system excited by a force applied to its mass (or by an acceieration at the support). (2) Program 3 to determine the response of a single degreeoffreedom system excited by one of the impulsive functions specified in the program, These programs anow us to obtain the response in tenns of dIsplacement, velocity, and acceleration as a fU:Jction of rime for any single degreeoffreedom system of linearelastic behavior when subjected to a general force function of time applied to the mass or to an acceleration applied to the support.
Fig. 1'4.3
4.4 The frame shown in Fig. P4.1 is subjected to a sudden acceleration of 05 g applied to its foundation. Determine the maxImum shear force in the columns.
Neglect damping.
4.5
Repeat Problem 4.4 for 10% of critical clamping. Use Duhamel's integra! to obtain the response of damped simple oscliiator of stiffness k, mass tn, damping ratio {, subjected to a suddenly appJied force of magnitude Fa Assume ini~ial displacement and initial velocity equal zerO. Establish ~he equation of motion for the system in Problem 4.6 and solve it by superposition of complementary and particular solutions with initial conditions for dispiacemellt and velocity equal to zero,
4.6
PROBLEMS
4.1 The steel frame shown in Fig. PA 1 is subjected. to a horizontal force applied at the girder leveL The force decreases linearly from 5 Kip at time t = 0 to zero at ! = 0.6 sec. Determine: (a) the horiZOp.lat deflection at 1=0.5 sec and (b) the 4.1
134
4,8
135
A trailer being puUed by .. truck moving a: constn.nt speed u is idealized as a mass In connf'.cted to the {ruck by a spring of stiffness k" Determine the governing differential equation and its solution if the truck starts fOrr. rest.
4.9
Determine the response of an undamped systerr. to a ramp force (Fig. P4.9) of maximum magnitude FiJ Ilnd duration Id starting with 7.ero initial conditions of
displacement and velocily.
8 (/)/g
Birl/g'" 0.1.1.
o
+fJ(t)
U1
Ib'
ia)
Fig. P4,14
Fig. P4.9
4.10 Determine the maximum displacement at the top pf the columns and maximum bending stress in the frame of Example 4.1 ass:.;ming that the columns afC pinned at the base. DIscuss the effect of base fixity. Delermine the maximum response (displacement and bending stress) for the frame of Example 4.1 subjected to a trinaguJar food of initia! force ['0 == 6000 !b linearly decreasing to zero at (ime IJ == 0.1 sec. For the dynamic sySlem shown in Fig. P4.!2, determine and plot the displacemenl as a funCtion of time for the inlerval 0 S 1:S 0.5 sec. Neglect damping.
Fit)
4.15
Repeal Problem 4.14 for 20% of crj(icai darr.ping. Determine for the tower of Problem 4,15, the maximum dispiacement at (he top of the tower relative to !hc ground displacement.
4.16
4.17 The frame of Fig. 4,17(3) is subjected to horizontal support motion shown in Pjg. 4J 7(0). Delermine a:e maxireum absolute deflection at the lOp of the frame.
Assume no damping,
4.11
4.12
,owt""
I"
,~~
~"'liwc)
~W8XZO~
"
~~T
10
0.2
0.4
1
Ill)
I , I
1.0
(b)
~~
Fig. P4.17
11
{ !~ecJ
10J
Fig. P4.12
4.13
4.14 Repeat Problem 4.12 for 10% of eThical camping. The lower of Fig. 1'4. (4(a) is subjected (0 horizOnlal grOlmd acce!ef3tion a (0 shown in Fig, P4.14(b). ;)etermine the relative displacement at the top of the tower at lime I 1.0 sec. Neglect damping.
4.18 4.19
RepeaL Problem 4.J7 for 10% of cHkal damping. A strucfural systeo modeled by the simple oscillator with 10% {s:= 0, 10) of criticar damping is SUbjected to [he impulsive load as shown in Fig, P4,19. Determine fhe response.
136
137
r;!)
(Ibl
FW""
1QS\(\Wf
(hI
0.0
Fig. P4.!9
IL\,
Fig, N,22
0.2
4,20
A water tower modeled rtf> shown in Fig, P4.20(a) is subjected to g:rouod shock given by the function depicted in Fig, P4.20(b) Determine: ~a) the maximum displacement at the top of [he tower and (b) the ma;omum shear force a! tbe base of the tower. Neglect damping. U,\C time step for integration.:11 0.005 sec,
4.23
The steel frame in Fig. P4.23 JS subjected Lo the ground motion produced by " passing traill in its vicinity, The ground motion is idealized <!s a harmok acceleration of the foundation of the frame with amplitude 0.1 g at frequency 10 cps. De{ermine the maxlmum response in terms of the relative displacement of the girder of the frame and ~he motion of the foundatiOn, Assume 10% of the critical damping,
Jolt)
(,)
0.23
rhec)
Fig. P4,23
(bl
Fig. P4.20
4.24
Detenninc the maximum stresses in the columns of the frame in P(oblem 4.23 using the maximum response in terms cf relative motion. Also check that the same results may be obtained using the response in terms of the maximum absolute acceieration. A machine having a weig.it W= 3000 ib is mounted through coil springs to a steel beam of rectangular cross sect jon as shown in Fig. P4.25(a). Due to malfunctioning, the machine produces a shock force represented in Fig. P4.25(b), Neglecting the mass of the beam and damping in the system, determine the ma:;;imum displaCement of tr.e machine.
4.21 4.22
Repeat Problem 4,20 for 20% of the crhical damping. Dete~ine the maxImum response of rhe tower cf Problem 4.20 when subjected to the iC1pulsive g:uuod acceleration depicted in Fig. P4.22.
4.25
138
S~ruclures
Modeled as a
Sjngje~Degree~of~F(eedom
System
y
k~ =
18000 Ib/io.
~==============::=======,=e=3=4~8=,="~,==i>
Il0'~'I0'1
101
F(r)
I
f(d
5
Fourier Analysis and Response in the Frequency Domain
6500"'~ 
~~
F(n"'F~sio3!.41Sl
I
0.0 0.1
;!~cj
Fig. P4.2S
4.26
For Problem 4.25 determine (a) the maximum [ensUe and compre3Sive stresses in the beam ar.d (b) the maximum force experienced by the coil sprin.gs during the sno:;k.
This chapter presents the applica~lon of Fourier series to determine: the response of a system to periodic forces, and (2) the response of a system to nonperiodic forces in the frequency domain as an alrernate approach to the usual analysis in [he time domain. In either case, the calculations :equire the evaluation or integrals that, except for some relatively simple loading rU:1Ctions, employ numerical meti:ods for their compu:ation. Thus, in genera!, to make practical use of the Fou:ier method, it is necessary to repiace the integrations with finite sums.
5.1
FOURIER ANALYSIS
The s"...lbject of Fourier series and FOt~.rier analysis has extensive ramificatJons in its application to many fields of science and mathematics. We begin by considering a single~degreeoffreedom system under the action of a periodic loading, that is, a forcing function [hat repeats itself at equai intervals of rime T (the period of (he function), Foune, has shown that a periodic function may be expressed as the summation of an infinite number of sine and cosine tenns. S:Jch a sum is known as a Fourier series..
139
140
Structures Modeled as a
F(r)
Single~Degree~of~Freedom
System
141
each component of the series. When the transient is omitted, the response of an undamped system to any sine term of the series is given by eq. (3.9) as
y" (I)
n = ,SIn
h Ik.
r~
I 
nWl
(5.4)
= nwlw and
w=
\l
I 
r~
For a periodic function, such as the one shown in Fig. 5.1. the Fourier series may be written as
F(t) = ao
+ al cos wi + a2 cos
2Wl
+ hi
or
The [otal response of an undamped, single~degree~offreedom system may [hen be expressed as the superposition of the responses to aU the force terms of the series, including the response C10lk (steadystate response) to the constant force Qo. Hence we ha ve
y (l) =
~+
k
,,=11
(5.6)
F(!) = Qo
(5.2)
where w= 27TIT is the frequency and T the period of the function. The evaluation of the coefficients Qo, a,,, and h" for a given function FCt) is determined from the following expressions:
aD =  1 T "
When the damping in the system is considered, the steady~state response for the general sine term of the series is given from eq. (3.20) as
b"lk sin (nwr  iJ) y" (r) = ; ? ? ,(I  r;t + (2r",;t
?
(5.7)
J""
or
F(I)dt
a" = 
T "
b" = T :,
2J"" 2J""
(lr~)sinnwt2rl1;cosnwt
where [i in the limits of the integrals may be any value of time, but is usually equal to either  TI2 or zero. The constant Qo represents the average of the periodic function F(t).
y,,(I)=T'
(lr;)'+ (2r",;)'
5.2
y" (t) =
T'
(5.9)
The response of a singledegreeoffreedom system to a periodic force represented by its Fourier series is found as the superposition of the response to
Finally, the total response is then given by rhe superposition of the terms expressed by eqs. (5.8) and (5.9) in addition [Q the response to the constant
142
143
lj t:
m
(.)
FIr)
J)?&///};)~7
T/:;::=J:~ FO~;?To
T 2T
(b)
3T
Fie;.!)

__ 1 _________ _
oF,
t.erm of the series. Therefore, [he total response of a damped singJedegreeof freedom system may be expressed as
_ ao y () r 
F{r,_ I)
,,
+ (2r,,)
1 cos nwt
J
1,_ I
r,
c"
IClc,1
(5 [0)
Fig. 5.3 Piecewise linear forcing function.
Example 5.1. As an application of the use of Fourier series in determininothe .response of a system to a periodic loading, consider the undamped simpl~ oscillator in Fig. 5.2(a) which is acted upon by the periodic force shown in Fig. 5.2(b)
Solution: The first step is to determine the Fourier series expansion of F (t). The corresponding coefficients are determined from eqs. (5.3) as follows:
5.3
ao=~rTFordr= Fo
T)o T
T
Proceeding as before in the evaluation of Duhamel's integral, we can represent the forcing function by piecewise linear function as shown in Fig. 5.3. The calculation of Fourier coefficients, eq. (5.3), is then obtained as a summation of the integrals evaluated for each linear segment of the forcing function, that is, as
N J" = [ )'
2
ao
T7 , I
2
N
F(r) dr
(5. [ [)
li_'
~
Fo n7T
a"=T~
b" =
N
f"
1;_: IiI
(5.! 2)
, I
The response of the undamped system is then given from eq. (5.6) as
yt~?~L
2 f" T I~I
(5.13)
()
Fo
_k
or in expanded form as
y(r) =
~
where N is the number of segments of the piecewise forcing function. The forcing function in any interval [,_I :5. t:5. t, is expressed by eq. (4.20) as
Fo
Fo sin wt
1Tk (l  r;)
Fo sin 2w{
21Tk (l  4r;)
2k
F(r) = F(r,.,)
+ ~(r .1r;
flF,
r,.,)
(514)
where
TI
in which .1Fi = F(l,)  F(ti I) and ..Jt, = {i  {i_I' The integrals required in the expressions of a" and b" have been evaluated in eqs. (4.21) and (4.22) and
144
145
designated as A (r,) and B (I,) in the recurrent expressions (4.18) and (4.19). The use of eqs. (4.18) through (4.22) to evaluate the coefficients a" and bl< yields
elF \ 1') (sjn nwt ill,
(i
i
sin nWf, 1)
elF,
n~w .:1/;
'> ')
sin n(~tJ)}
!
(5.15)
?""11'
,; ~ elF
The interval of integration in eq. (5.20) has been selec(ed from zero to T for the periodic function. It should be noted that the exponential form for the Fourier series i:1 eg. (5,19) has the advantage of simpliciry when compared to [he equivalent trigonometric series. eq. (5.2). The exponential form of the Fourier series can be used as before (0 deter:uine the dynamic response of structural systems, However, a more effective method is available for the determination of the coefficients as well as for the cakulatIon of the response for a single degree of freedom excited by the force expanded as: in eq. (5.; 9). This method, which is based on Fourier nnalysis for the discrete case, is presented in the next sections.
en
cos nWI;
5.5
(5.16)
cos
nZ;)/;_I))~
When [he periodic function FCr) is supplied only at N equally spaced time intervals (dl TrN) 10, l" [2, ' ' ' , [/1 h where t} = jLlr, the integrals in eqs_ (5.3) may be replaced approxImately by the summations
a,. ~  ) F (ti) cos nMj 4t T i"'O
The integral appearing in the coefficient ac of eg. (5.3) is readily evaluated after substituti;1g F{r) from eq. (5.14) into eg. (5,11), This evaluation yields
1 Nl
(5.17)
b_=~ .. T
0, 1,2, ...
(5.21)
5.4
The Fourier series expression (5.2) may also be written i:1 exponential fonn by substituring the trigonometric functions using Euler's relationships:
where w= 27T11. The above definitions for the Fourier coefficients have been slightly altered by omitting the factor 2 in the expressions for an and h". In this case eq. (5.2) is then wriaen as
F(r,)
(5.22)
cos nwt =
:;2
e;,Iid, T e
,""Wi
(5.18)
If we use complex notation, egs, (5.21) can be combined into a single form by defining
The result of this substitution may be written as and using Euler's relationship
(5.19)
(5.23)
(5.24)
to
where
C=~Ii
I) into
eq. (5.23)
(5.25)
(5.20)
C"=y
147
Substituting
'i =
1,2,."
(5.26)
Equation (5.26) may be considered as an approximate formula for calculating the complex Fourier coefficients in eq, CS.20), The discrete coefficients given by eq, (5.26) do nOt provide sufficier.t infonnation to obtain a continuous function for F(r); however, it is a most importanr fact that it does allow us. to obtain all the discrete vaiues of the series [F(IJ)} exactly [Newland, D.E., 1984). This fact leads to the fonnal definition of the discrete Fourier transform of [he series {F(!j)}.j=O, 1,2, N 1, given by
"q
As a matter of interest, Example 5A is presented later in this chapter to illuslrate the importance of choosing the nl.:mber of sampling points N for Lhe excitation function sufficier.tly large Lo avoid spurious resufts due to aliasing. Having represented an arbitrary discrete function by a fwire sum, we may then r..lso obtain as a discrete fl:nction the respor.se of a simple oscillator excited by the ham:onic components of the loading function. Again, only the steadystate respor.se will be considered. The introduction of the unit exponential forcing function E" = e''''nt into the equation of morior:, eq. (3.13), leads ro
(531)
which has a
steady~stale
(5.32)
" = O. L 2, .. " (N  !)
(5.27)
When eq. (5.32) is i:1troduced i:lto eq. (5.3 t), it is found thai :he function H(w,), which will be desigr.mec as the complex frequency response function, takes the form
H(w,,)
F( [i) 
L.,
(5.28)
,,=0
(5.33)
l) in order to maintain the sym:netr)' of transfonn pair eqs. (5.27) and {5.28). It is important to realize that in the calculation of the summation indIcated in eg. (5.28), the frequencies increase with increasing jndex n up to n N n. It will be shown very shortly that, for n> N 12, the corresponding frequencies are equal to the negative of frequencies of order N n. This fact restricts the harmonic components that may be represe:1ted in the series to a maximum of N 12. The frequency correspondIng to this maximum order WIIn. = (N fl)liJ is known as the Nyquisl frequency or sometimes as [he joldlng frequency. :\1oreover, if [here are harmonjc components above W,v/2 in the original functior.. these higher compor.ents wilt introduce dis.tortions in the lower harmonic corr.ponents of the series, This phenomenon is called aliasing [Newland., D.E., 1984, p. 118). In view of this fact, it is recommended that the number of intervals or sampJed pOints N should be at least twice the highest harmonic compone:1t present in the function, The Nyquist :requency &J y is given in radians per second by (5.29) and in cycles per second by
The range of the sumation it: eq. (528) has been limited froal 0 to (N
H(w,,)="
k(l
+ ro + 2/f,/;)
1.
Therefore, the response)'4 (;j) al tirt.e {j = jiJr to a harmoniC force cOr:1ponent of arr.pliwde en indicated :n eq. (5.28) is given by (5.34) ar.d the total response due to (he N hannonic force components by
f ; 2." =
.y
(cps)
(5.30)
(5.35)
148
Structures Modeled as a
Single~Cegree~ctFreedom
System
149
where ell is expressed in discrete form by eg. (5.27). In the determination of the response y (t) using eg. (5.35), it is necessary CO bear in mind that in eg. (5.28) the force component of [he frequency of order n is equal co [he cegative of [he component of (he frequency of order N  n. ThIs fact may be verified by substitcting  (N n) for n i!1 the exponemial factor of eq. (5.28), In this case we obtain
(5.36)
The evaluation of the sum wHl be most efficient if the number of time ir.crements N into which [he period T is divided is a power of 2, that is,
(539)
where lY! is an imeger. In this case, :he integers j and n can be expressed in binary form. For the pUf'I,?ose of iHustrution, we wi!: consider a very simple case where :he load period is divided ~nto only eight time increments, that 1s. N = 8, M = 3. In this case, rhe icdices in eqs, (5,27) and (5.35) will have the bicary ::eprese:1tatlon
since e 2""/:= COS i sin = l for all integer values of j. Equation (536) together wiili eq. (5.28) shows that harmonic components of the force corres~ ponding to frequencies of orders Ii and  (N  n) have the same value. As a consequence of this fact, r'l w,,/w, where (seJecting N as an even number) as
bJ,,=niiJ
jo + 2j, + 4[,
~5.40}
w=.i KIm
for
n::S;Nt2
and
(SAl)
w"
(Nn)w
for
n>NI2
where the frequency corresponding 1:0 fi = tV /2, as already mentioned. is the highest frequency [bat can be considered in the discrete Fourier series. The evaluation of [he sums necessary to determine the respo<1se using the discrete Fourier tninSfOffil is greatly simplified by the fact that the ex.ponential functions involVed are harmonic and extend over a range of Hl, as demonstrated in the next secrion,
We note that tbe first factor on the righthand side is unity since from eq. (5.38)
W~l
5,6
A numerical technique is available that 1S efficiem for computer detenr.ination of the response in the frequency domain. This mewod is known as the fast Fourier transfonn (FFT) [Cooley. P. M., eL aL (1965)1. The corresponding compurer program is reproduced as a subroutine of computer Program 4. The response it; frequency domain of a singledegreeoffreedom system to a general force is given by eq. (5.35) and the coefficients required are compmed from eq. (5.27). It can be seen (hat either eq. (5.35) or eq. (5.27) may be represented, excep( for sign in the exponent by the exponential functron as
where 1= j !nZ + 2j2n: + hnl is an integer. Therefore, only the remaining three factors need to be considered in the summations. These summations may be performed conveniently in sequence by introducing a new noration to indicate the successive steps in the summation process, Thus the first step can be indicated by
."'1
A ()) =
2: A'" (n)W::;
(5.37)
",=0
where
W,v=
(5.38)
150
151
An
F(t)
w ~ 38.61<
0.16 0.64.
t(sec)
100 K/in.
1201< 
Fan method
(b)
The final result Am UO,jl>j2) is equal to A U) in eg. (5.37) or (5.41). This process, indicated for N = 8, can readily be extended to any integer N = 2M. The me [hod is particularly efficient because the results of one step are immediately used in the next step, thus reducing storage requirements and also because the exponential takes the value of unity in the first factor of the summation. The reduction in computational time that results from this formulation is significant when the time interval is divided into a large number of increments. The comparative times required for computing the Fourier series by a conventional program and by the fast Fourier transform algorithm are illustrated in Fig. 5.4. It is seen here how, for large values of tV, one can rapidly consume so much computer time as to make the conventional method unfeasible.
Example 5.2. Derermine the response of the tower shown in Fig. S.S(a) subjected to the impulsive load of duration 0.64 sec as shown in Fig. S.5(b). Assume damping equal to 10% of [he critical damping.
Solution:
Problem Data:
=3
~T~im:::::e~(c:s::e::c,)+_~F=o::r::.ce=,(,,1 b:,)_
5.7
The computer program presemed in this chapter calculates the response in the frequency domain for a damped singledegreeoffreedom system. The excitation is input as a discrete function of time. The computer output consists of two tables: (1) a table giving the first N complex Fourier coefficienls of [he series expansions of the excitation and of the response and (2) a table giving the displacement history of the steadystate motion of the response. This second table also gives [he excitation function as calculated by eq. (5.28), thus providing a check of the computations. The main body of this program performs the tasks of calculating, using the FIT algorithm, the coefficients ell and the function F(l j ) in eqs. (5.27), (5.28), and the response Y (I) in eq. (5.35).
o
120,000  120,000
o
2H.. E:D4
!NPUT D':'7.:'
NiJMEE~
OF
POZ~TS
DEF":::l:KG THE
E;~,::r;.!'7To!l
~"
S?R!NG CONSTnNT
DA.!1?!NG COEFt'"ICI2!IT
A.V; "
100 100000
;'K ;
c "
G ;
632
0
152
1'~,~t:
Single~DegreeotFreedom
System
'!'!11"
~XC!TA'!'!ON
:} ,ec'}
J.e?
lZ:'O:,,) ,co
O. ;;:~D
o. co
Exponent of N = 2M: /vl = 4 Gravitational index: G = 0 (force on the mass) Excitation function:
Time (sec)
0.00 O.lO 0.20 OAO 0.45 0.60 :.00
Force (Kip)
D!AC
0.e;:;:,>
O.C<>JC
o
[0
"co
c. ';l~~C
OvlS
1':,'5 '1'10
O.tEn
j010 0 DOJ l O.
o
6
0.znr:
<),
;;~:. j
evoo
COl5
o. ~~~I)
_0.0337
,1:0:)
o o
:. ",41
.J Ccc:;.
,) OeD;)
l1C;;;;C
C(>~}
5JC:;C 01%
:;<';>co
5J)('0 O:CO
j.onc
:u:os~
l1JXC.
joe:
>;;)},5::::;: OT
?O~:,1"'$
:);';t1ytxC :"S
::XC1:;;,:O<!
0.:010
",_'d"iS
C,"..>:7!~ :::;:'HT:C:!~:'"
Determine the response of the simple osciIlator shown in 5,6(a) when subjecled to (he forcing function depicted in Fig. 5,6(b). Use lkf = 4 for the exponent i:1 N 2M. Assl.!me 15% of the critical damping,
Example 5.3.
'!:Xl?OC!f':."':'
m"
::msx
',rHZ
iGRAV:'t:.' C? ZZROl
:;:XC!1"'A"t1Ol:
!::<:;r:;?<r!CIl
J.G;!D
0.00
lao
O. ,150
11, CO
20J 0,600
a,OJ
G,,;0
o.,)C
cess
~';!
000"
lieS!
<J
~O
0021 0004
DOC;'
CC"" 0000
Fit)
0.115$
OS'5
ClS11
_0. )007
0 (lOC7
O. DOOl
LO
'J DJO
0 0$75
Cl :'314
oooa
"J.,DS)
0 :j('C7
10
0.115$
e
0
000"
:.:
tb)
o .:a4
0,SH3
C lOS::
en,
0.0040;.
0.e<);)4
c o
~s;.;;
C043
_l.:$C~
.5,,52
154
155
x.v.c.
FORCE RZ,\L
0.0(100
O.JGS2 0.00%
a.ooon
C.\)OGO
fCRCE UV,O.
v.ODOO
o .~ooo
wn.2S00
5. SC'OC
. 0 ,0S;1 wO.CG}S
,,).O;;S{
coooo
~O.
0. aDeD
 _....
Time (sec) 0 0.125 0.250 0.375 0.500 0.625 0.750 0.875 1.000

0000
(1.0(100
~O.OJOO
B.?SOO i$.0000
~l.
Number of Sampling Poims for the N 8 N 16 N=32  0.0416 . 0.0153 0.0052 0.Oi78 0.0221 0.0178 00052  0.0153  0.0416
....
HZs
0.)750
G.~n5
0.0091 0.C1<4
G.OO$6 O. (4)0
0.l'000
LSO\lO GOOl'
~. ~OO() ~.COOO
;:: .0.000.
N= 64
(l.oOOC
O.OOGO
(I.oooc
0.,0000
O. G:)OO
O.5COO
0.%25 0.62S>J
O.6S7S 0.7SCO O.$U:' C. S7S0
(l.H~:;
O. ~21D :).0167
o.OOGe 0,000:)
J,OC:i"J
:. SOC'O C.OOOO
a.M'"
0.0116 0.0018
c.oooo O.cooC'
0,0:)00.
J .COO:)
o.eOOO
o.oeoo
O.OCOO
", GOCl'
~.OCOO
C.oooc
O.OOOD 0,000;) 0,0(0)
o.oon
0.00':
0.00;';0
O.Ol'OO
0.0000
Example 5.4. Consider a smgledegreeoffreedom und<::!mped sysrem in which k 0:::: 200 lblin, m 100 lb sec 1 /in subjected to a force expressed as
....
_

0.4201 0.4698 0.5 i07 0.5358 0.5443 0.5358 05107 0.4698 OA201
_
0.0416
...
.~
P(I)
" =L
""
(a)
lvJ = 3, 4. 5, and 6 corresponding to N = 8, 16, 32, and 64 sampled points. Then discuss (he results in relation to the limitations imposed by the NyqUIst
frequency.
The fundamental frequency of the excitation function. eq. (a), is Wj f2rr= t sec. Since the ~i!ghest component in eq. (a) is of order Wlf) = 16wJ, to avoid aliaSing, the number of sampled poims should be at least twice that order, that is, :he minimum number of sampled points should be N 32.
WI
Sotwion:
the Nyquist freqnency. The response obtained for N g or N = 16 gives spuriolls numericai results, . Results in Table 5.2. which "vere obtained using N = 8 sampled pomts, also verify t:tat when the exciting force ContamS.hiHTI:onic compone~ts higher ~~atn the Nyquist frequency which corresponds, In Lhls case, to N/  4, the re",ul s are again spurious. . . A fir:al comment is in order. The example ptesen[ed, havlflg equal.amplltude for all the components of the exciting force, serves to e:mphasne the importance of choosing tr.e number of snmpling point~ N s~ff:clent,ly large to avoid aliasing. In pn!c~ical siwa~ions r.ormally the higher :larmOnlCS hav~ a much smalier amplitude thar: that of the fundamental Or lower frequencIes.
TABLE 5.2
Time
With a simple moodification of Program 4, the applied force is calculated in the program, instead of being supplied through a numeriC2.1 [abLe as nor~ many required by the program. The results given by the computer for this example are conveniently arranged in two tables: Table 5.1, giving the dis~ placement response to the excitation having aU 16 harmonic components as prescribed for this problem; and Table 52, Showing the dispJacement response to a reduced number of harmonic terms in (he excitation function. For this example, in which the exciting force is supplied in 16 harmonic components, the response given in Table 5.1 corresponding to N = 32 or N =:;:; 64 may be considered the exact SOlution. A comparison of the response l6 with (he exact solution (N = 32) shown for sample points N = 8 or N dramarically demonstrates the risk of not choosing N sufficient!y large enough so that none of the frequencies of the components in the exciting force exceed
Displacement Response for Example 5.4 (Excitation Force Sampled at N~ 8 pointS) _____ . _____________ N'Jmber of Harmon~.~ Comp0n.~nts in the Exciration Force
(sec)
0 0.125 0250 0.375 0.500 0.625 0.750 0.875 1.000
IV
N=8
N=16 0.8531 0.9357 1.022 1.071 1.089 1.071 1.022 0.9357 0.8531
156
157
~ "" C.10
~r'
5.8 SUMMARY
Fig. PS.2.
In general, any periodic function may be expanded into a Fourier ser:es, eq. (5.1), whose terms are sine and cosine functions of successive multiples of the fundamental frequency. The coeffjc;en~s of these functions may be calculated
5.3 The spnngrn2$5 system of Fig, P5,2 is acted upon by the t:mevaryL1g force shown in Fig:. P53. Assume that (he force :5 pedodic of period T:= 1 sec and determme steadystate Ies?or:se of the system by applying Fourie; series expansion of F(t).
the
forms, The discrete form of these !mnsfonns, eqs. (5.27) and (5.28), penni!s
their use in numerical appiications. An eXlremely efficient algOrithm known as the fast Fourier transform (FFT) can save as much as 99% of the computer time otherwise consumed in the evaluation of Fourier complex coefficients for the eXCHattOn function and for the response of a dynamlc system
,clfi
PROBLEMS
S.l
Determine the first three terms of [he Fourier series expansion for the lime varying force shown in Fig. PS.I.
5".4 The cantilever beam shown in Fig. PS.4(a) carries a concentrated weight at its free end and it is subjected to a periodic acceieration :rt :its support which is the rectified sine f:.mcrion of period 7=OA sec and amplitude )te"'" 180 as shown 1fl Fig. P5.4(b}. Determine: (a) the Fourier series expansl0n of rhe forcing function. and (b) the steadystate response considering oni:; three terms of the series, damping in the system and assume L1e beam massless.
05
1.0
30xL.
U
Fig. P5.1.
20
.(sec)
'0
~(rl
T
Y,
T ,
I.)
(hi
5.2
De!ermine the steadystate response fo: the damped springmass system shown in Fig. P5,2 rnl,H is acted upon by (he forcing func~ion of Problem 5. L
FIg. P5.4.
~he
Frequency Domain
159
Solve Problem 5.4 using PlOgram 4. Take 16 Fourier terms. Inpu[ [he values of the excitation functions Ul intervals of 0,025 sec. Solve Proble:n 5,4 in the frequency domain using Program 4. Take the exponent of N 2 M , M=4, Input [he effective force, F df = mYt(!) cZllculated every 0.025 sec. Repear Problem 5.6 assuming 20% of critica: damping. The forcing fur:c~ion shown in Fig. P5.&(0) is assumed w be periodic in the extended interval T= lA sec. Use Prog:rom 4 to determine the fir:i.t eighl: Fourier coefficienrs and the steadystate response of a .mucture modeled by the undar:J~ ped osciHawr shown in Eg. P5.8(b),
5.12
Cor.sider the sy:item shown in Fig. P5.12 and its loading with assu:ned period T= 2 sec. Determine: (a) Ihe firs! four rerrns of the Fourier series expansion for [he forcing functior: in terms of Po; (b) Lhe firs! four terms 'Of the Fourier series expansion fer the resp'Onse.
5.7
5.8
F(t)
2.0
{ill (b)
Fig. P5.12.
0.' 0.4 0.6
0.8
1.0
I 1.2
1.4
r(secl
1.4 sec
0: Fourier series;
input
I,)
'h,
Fig, PS,g,
,~ec.
Repeat Probiem 5. J 3 as:;uming: that (he systel7'l has 20% of [he critical darr.ping.
Ob:ro;n the close solution for the system in Proble:n 5.12 by considering (he halfcircle sinusoidal excitation as the s"Jperposition of two sinusoidal functions: PI = Pa sin Tt1 smrring ,n'" 0 and P! and Pc sin 17(1  I) stanir:g at r = 1 sec as shown in Fig. PS.iS.
P 1
5.15
5.9
5.10
Solve Problem 5.8 in the frequency domain using Program 4, Take M=4 for the expOl:ent in N = 2M. Cse Prograrr. 4 to determine: (1) the Fourier senes expansion of the rorcir:g func(:on shown in Fig. P5.10(a), and (2) the steadys(ate :esponse caiculmed in rhe freql.:ency domain for the springmass system in Fig. P5.1O(b). Assurr:e ! 5% of the critical damping. Take />1 = 3 for the exponent in N 214 compare results with those in the solution of Example 5.3.
p~
1r\;stnw:
L:::, _ ,
Fitl
t
ft; ""
200 K/in.
0.1 0.2
Fig, PS.15.
I
O.S 1.0
r sec
r~'''''' 10"'~
i,
(,j
5.16
(b)
Fig. PS.IO.
5.11
Solve Problem 5.10 in the frequer:cy domain l'sing Program 4. Take M 5 for the expOJ1eIU in N 2 M.ColT.pare resules with those in the solutIon of Example 5.3.
A single~degree.off:eedom system having a natural period of ;).8 sec ar.d stiff" ness of 5000 Iblin subjected to an impulse of dUr<Hion 0.5 sec which varies as shown in Fig. F5 16. Compute (he response with an exte(l<;ion of I> 1 sec for which the value of force is zero. Use F'rograrr: 4 to Obtain: (a) the discrete transform of the forcing function and of [he response. (b) the displacerr:e!lt response, and (c) the applied force calculated using the ir:verse discrete trans:orm. Neglect damping and discreljze (he forcing fl'ncrior: uSing a rime slep Dr = eu SeC.
160
161
5.19 5.20
tOOClb
Repeat ProbleCl 5.18 assuming thal: the damping in the system is 5% of the criticaL Solve Problem 5. ('8 for an accele:adon at lhe base of tower (hat varies as a symmenical triangu!ar load as shown i:1 Fig, P5,20,
Fig. P5.16.
5.17 5.18
Solve Problem 5.16 assuming a 10% of the critical damping in tle system,
The water rower shown in Fig. P5.J:'8{a) is subjec(ec to impu:s:ve acceleradon of its ba$e lilac varies as half the 3ir.e function shown 1n Fig. P5.1'8(b). {)$e Program 4 to determine: (u) :he discrete Fourier coefficients for the excitation and for tile respo:"Jse. (b) (he re:1I!(ive displacement of rhe tower with respect to the ground displacement, and (c) ~he excitation obtained by the inverse discrete transform. Use an extended ex;ciration of [Otal duration 1.6 sec and time step .::::il 0.1 sec. Negrect damping.
W"'" 3S.6K
K""
11
lOO~
'"
O.3g
O~';;O';g'"1.76[g~
(a)
(b)
Fig. PS.1S.
A{r)
1.2
I.'
seo
Fig. PS.20.
163
6
Generalized Coordinates and Rayleigh's Method
Principle, which establishes dynamic equiHbrh:m by the inclusion of the inertial forces in the system. The principle of virtual work may be stated as foHows: For a system that is in C<tuilibrium, the work done by all the forces during an assumed displacement (virtual displacement) that is compatible wjth the system constraints is equal to zero. In general the equations of mOlion are obtained by introducing virtual displacements corresponding to each degree of freedom and equating the resulting work done to zero. To illustrate the application of the principle of virtual work 1O ob(ain the equation of motion for a singledegreeof~freedom system, let us consider the damped oscillator shown in Fig. 6J{a) and its corresponding free body diagram in Fig. 6.1 (b). Since the Inertial force has been included among the ex.ternal forces, the system is in "equilibrium" (dynamic equilibrium). Consequently, the principJe of virlual work ~s applicable. If a virtual displacement by is assumed to have taken place, the total wo:k done by the forces shown in Fig. 6.1(b) is equal to zero, lbat is,
my8y
or
{my + cy + ky 
(6.1)
Since is arbitrarily selected as not equal to zero, the other factOr in eg. (6.1) must equal zero. Hence,
oy
In the preceding chapters we concentrated our efforts in obtaining the response to dynamic loads of structures modeled by the simple oscillator, that is, structures that may be analyzed as a damped or undamped springmass system, Our plan ;n the present chapter is 10 discuss the conditions under which ! structural system consisting of multi pie interconnected rigid bodies or having distributed mass and elasticity can stiH be moceled as a onedegree offreedom system. We begin by presenting an alternative method to the direct .application of Newton's Law of Motion, the principle of virlual work.
my + cS' + ky 
F (I) ~ 0
(6.2)
Thus we obtained in eq. (6.2) the differential equation for the motion of the damped oscillator.
6.i
An alternative approach to the direct method employed thus far. for the formulation of the equations of motion is the use of the principle of vinuai work. This principle is particularly usefuL for relatively complex structural systems which contab. many lnterconnected parts. The prir.dple of virtual work was originally stated for a system 1:1 equilibrium. Nevertheless, the principle can readily be appJied to dynamic systems by the simple recourse to D'Alernbert's
162
cy ,
,!
T:
mg
N
"<I
164
Structures Modeled as a
Sing!e~Degree~ofrreedom
System
165
6.2
Most frequently the configuration of a dynamic sysrem is specified by coordinates indicating the linear or angular positions of elements of t3t system. However. coordinates do not necessarily have to correspond directly to displacements; they may in general be any independent quantities that are sufficient In number to specify the pOSition of all parts of u1e system. These coordinates are usually caIled generalized coordinates and their number is equal to the number of degrees of freedom of the system. The example of the rigidbody system shown in Fig. 6.2 consists of a rigid bar WIth distributed mass supporting a circular plate at one end. The bar is supported by springs and dampers in addition to a singie frictionless support. Dynamic excitation is provided by a transverse load F(x,t) varying linearly on the portion AS of the bar. Our purpose is to obtain the differential equation of motion and to identify the corresponding expressions for the parameters of the simple osci~lator representing this system. Since the bar is rigid, the system In Fig. 6.2 has only One degree of freedom, and, therefore, its dynamic response can be expressed \vith one equation of motion. The generalized coordinate could be selected as the vertical displacement of any point such as A, B, or C along the bar, or may be taken as the angular position of the bar. This last coordinate des.ignated &(t) is selected as the generalized coordinate of the system. The corresponding free body diagram showing all the forces including the inertial forces and the inertial moments is shown in 6,3, In evaluating the displacements of the different forces, it is assumed that the displacements of the system are small and, therefore, vertical displacements are simply equal to the product of the distance to support D multiplied by the angular displacement 8= 8(1). The disptacements resuhing at tbe points of application of [he forces in Fig. 6,3 due to a virtuai displacement 58 are indicated in this figure, By the principle of virtual work~ the total work done by the farces during this virtual displacement is equal to z.ero. Hence
 !
I,}
'"'
cLB
4LiriLii
miiL
D
2LM
~L
2LKfi
Fig. 6.3 Displacements and resultant forces for system in Fig. 6.2. or, since 88 is arbitrarily set not equal to zero, it follows that
(6.3)
I,
lm(~r
The differential equation (6.3) governing the motion of this system may conveniently be written as
(6.4)
where J", Coo, K, and F (t) are, respectively, the generalized inertia, generali7.ed damping, generalized stiffness, and generalized force for this system. These quantities are given in eq. (6.3) by the factors corresponding to the acceleration. velocity, displacement, and force tenns, namely,
r =/0+ 11 + 4ifle +
c~
me
= cL
K' = 4kL'
r
lr:'as~!uni!
(t) = i L'!(I)
AC===~~~====~======~=====i~=i~
Fig, 6.2 Example of
slngledegree.of~fteedom
01 length}
Rigid OOf
Example 6.1. For the system shown in Fig, 6.4, detennine the generalized physical properties M', C', K' and generalized loading r (f). Let Y(t) at the point A:: in 6.4 be the generaiized coordinate of the system.
Solution: The free body diagram for the system is depicted in Fig. 6.5 which shows all the forces on the two bars of the sytem including the inertial
rigid system.
166
Stfuc~tJre$
167
M'
C
I,
+4 
3m
c
3
K' = 3k,
and
Fig. 6.4 System for Exnmple 6.1
P" (I) =
force and the )nertial moment. The generaHzed coordinate is Y (1) and the displacemenr of my point in the system should be expressed in terms of this coordinate; neveriheJess. fOf convenience, we select also the auxiliary coordinate Y j (I) as indicated jn Fig. 6.5. The summation of the moments about point A I of all the forces acting on bar Ai  Bh and the summation of moments about Bl of the forces on bar Ai. B 1,) give the roUowing equations:
6.3
(6.5)
30
Subs~itljting
h 3 .. a 1'Y 2 Y+maY+cY+k.
4 3
'3
\J Y,2a+3ak,Y=Q
(6.6)
YJ from eq. (6.5) into eq. (6.6), we obtain the differential equation for the rnotion of the system in termS of the generalized coordinate Y(t), namely
The examp1e presented in the preceding section had only one degree of freedom in spite of the compiexicy of the various parts of the system because the two bars were interconnected through a spring and one of rhe bars was massless so that only one coordinate sufficed to completely specify the motion, If the bars were nor rigid, but could defom) \11 flexure, the system would have an infinite number of degrees of freedom However, a singledegree~offree dom analysis could still be made, provided Ihai only a single shape could be developed during motion, that is, provided that the knowledge of the disp)ace~ ment of a single point in t;)e system determines the displacement of the entire system. As an illustration of this method for approximaling the analysis of a system with an infinite number of degrees of freedom with a single degree of freedom, consider lhe cantilever beam shown in Fig. 6.6. In this iHustration, the physi<;aJ properties of the beam are the Dexural stiffness EI(x) and its mass per unit of iength m (x). It is assumed that the beam is subjected to an arbitrary distributed forcing function p (x, t) and to an axial compH'..ssive force N_
168
Generalized Coordinates
169
In order to approximate the motion of this system with a single coordinate, it is necessary to assume that the beam deflects durjng its motion in a prescribed shape. Let if>(x) be the function describing this shape and, as a generalized coordinate, yet) the function describing the displacement of the motion corresponding to the free end of the beam. Therefore, the displacement at any point x along the beam is
y(x, 1)
~
The factor f is required for the correct evaluation of (he work done by tbe flexural moment increasing from zerO to its final value M(x) [average value M(x)!2J. Now, utilizing eqs. (6.10) and (6.11) in eq. (6.12), we obtain
v~
)0
(6.13)
4> (x)y(t)
(6.7)
Finally, equating the potenrial energy, eq. (6.13). for the continuous system to the potential energy of the equivalent system and using eg. (6.7) results in
where 4>(L) I. The equivalent onedegreeoHreedom system [Fig. 6.6(b): ",ay be def:ned simply as the system for which the kinetic energy, potential energy (straIn energy), and work done by the external forces have at all times the same values in the two systems. The kinetic energy T of the beam in Fig. 6.6 vibrating in the pattern indicated by eq. (6.7) is
! K'Y (t)'
or
K= fEJ(X){4>" (x)}'dx
where
T~ r."~m(X){4>(X)Y(t)}'dx
0'
(6.8)
Equating this expression for the kjne~ic energy of the continuous system to the kinetic energy of the equivalent single~deg:eeof~freedom system t.M"Y(r)l and solving the resulting equation for the generalized mass, we obtain M
The generali7.ed force F"{t) may be found from the virtual displacement
OI(t) of the generalized cnordinate Y(l) upon equating the work performed by
(6.9)
the external forces in the structure to the work done by the generalized force in the equivalent singledegree~of~freedom system. The work of the distributed external force p (x, t) due to this virtual displacement is given by
The fiexu:al strain energy V of a prismatic beam may be determined as the work dOiH:; by tbe bending moment fof (x) undergoing an angular displacement dO. This angu]ar displacement is obtained from the wellknown formula for the flexural curva!ure of a beam. namely (6.l0) or
W=
Substituting
LLp (X,t)8ydx
p (x, 1)4> (x)OI dx
oy
(6.15)
The work of tlle generaUzed force F(t) in tbe equivalent system corresponding to the virtual displacement OY of tbe generalized coordinate is
W
d8=dx
M(x)
= ret) IlY
(6.16)
E1
(6.1l)
Equating eq. (6.15) with eq. (6.16) and canceling the factor oY, which is taken to be different from zero; we obtain tbe generalized force as
r(t) ~
since dy /dx e, where B, being assumed smaH, is :aken as the slope of the elastic curve, Consequently, the strain energy is given by
1
rL
p(x,t)4>(x)dx
(6.17)
v~ f1M(X)d8
(6.11)
Similarly, to detennine the generalized damping coefficient) assume a virtual d~spiacement and equate the work of the damping forces in the physical
170
sysrem with the work of the dampmg force in the eqUivalent offreedom system. Hence
C'Y6Y =
. J'
0
c (x)y8y dx
where c (x) is the distributed damping coefficient per unit length along the beam. Substituting 6y = (x)oY and ]I: (x)'i' from eq. (6.7) and canceling the COITLt'J:lon factors. we obtain
expression waS conveniently set equa! to the initiaL length of the beam L instead of to its borizontal component L I. Now we define a new stiffness coeftlcient to be called the generalized geometric stiffness K~ as the stiffness of the equivalent system requlfe.d to store the same potentia! energy as lhe potentia! energy stored by the normal force N, that is,
IK~Y(!)' = No(!)
C': fC(X)[(X))'dx
(6.18)
SUbstituting 8(1) fro:n eg. (6.21) and [he derivative dyldx from eq. (6.7),
we have
which is the expression for the generalized damping coefficient To calculate tne porentlal energy of the axial force N which is unchanged during the vibration of the beam and consequenlly is a conservative force., it is necessary to evaluate the horizontal component of the mOlion 8(f) of the free end of the beam as indicated in FIg. 6.6. For tnis purpose, we consider a differential element of length dL along the beam as Sh?wn in Fig. 6.6(.a} ,\he length of this element may be expressed as
I, 2 I KeY(,) : N)
2'
2,[
r'l,Y(t)d<l>]' dx
dx
(6.22)
or
or
(6,l9) Now, integrating over the horizontal projection of the lenglh of beam (L f) and expanding in series the binomIal expre.sslon, we obtain
Equations (6.9), (6,14), (6.17), (6.18), and (6.22) give, respectively, the generalized expression for the mass, stiffness, force, damping,_ and geometric stiffness for 3 beam with distribuled properties and load, modellng it 3S a , simple oscillator. For the case of an axial compressive force, the potentiai energy in the beam decreases with a loss of stiffness in tbe beam. The opposite is true for a tensile axial force, which resuhs in an increase of the flexural stiffness of the beam. Customarily, the geometric stiffness IS determined for a compressive axial force. Consequently, the combined generalized stiffness K; is then given by (6.23)
Finally> the differential equation for the equivalent system may be written as
Retaining only the first two terms of the series results in
(6,24)
(6.20)
The critical buckling load N n is defined as the axial compressive load that reduces the combined stiffness to zero, that is,
L=L +
or
0(1)
K; = g' 
K~ ~O
=L
(6.21)
The substitutian of K' and Ko from eqs. (6.14) and (6.22) gives
The reader should realize that eqs. (6.20) and (6.21) ll\voive approximations
since lhe series was truncated and the upper limit of the integral in the final
172
173
(6.25)
(dldx)' dx
less accurate than the displacemeflls, because the derivatives of approximate shape functions tends to increase errors in the shape function. A better approach to estimate the equivalent forces is to compute the inertial forces that resulted in the calculated displacements. These inertial forces are
given by
p (x, /)
111 (x)ji(x,
Once the generalized stiffness K~ and generalized mass M'" have been determined, the system can be analyzed by any of the methods presented in the preceding chapters for singledeg~eeoffreedom systems. In particular, the square of .he natural frequency, 0', is given from eqs. (6.9) and (6.14) by
I)
(6.30)
(6.31)
E1 (x)" (x)' dx
At any time t the equivalent load p(X,I) given by eq. (6.31) is applied to the
m(x)'(x)dx
The displacement y(t) of the generalized singledegreeoffreedom system is found as the solution of the differential equation
M'Y(I)
+ C'Y(t) + K'Y(t)
(!)
and the displacement y (x. r) at location x and time t is then calculated by eq.
(6.7).
structure as a sta~ic load and the internal forces (shear forces and bending moments), as wen as the stresses resulting fron these internal forces, are determined. To provide an example of the determination of the equivalent one degree of freedom for a system with dIstributed mass and stiffness, consider the water tower in Fig. 6,7 to have uniformly distributed mass in and stiffness EI along its length with a concentrated mass lVi mL at the top. The tower is subjected to an earthquake ground motion excitation of acceleration ag(t) and to an axiai compressive load due to the weight of its distributed mass and concentrated mass at the top. Neglect damping in the system. Assume that during the motion the shape of the tower is given by (x)
6.4
a [E1(x)a;r a'Y]
(6.28)
hi,FI
i I
(6.29)
The equivalent force p (x, I), calculated using eq. (6.29), depending On derivatives of the shape function, wi[] in general. give internal forces that are
Fig. 6.7 Water towe!;' with distributed properties for Example 6.2.
174
175
Seiccting the lateral displacement Y (/) at the top of the tower as the generalized coordinate as shown in Fig, 6,7, we obtain for the displacement at
any point
By setting
K; =
O. we obtain
trig .. ' .  (3".  4\
Q
16
'
(5.33) The generalized mass and the generalized stiffness of the tower arc compuled, respectively, from eqs. (6.9) and (6.14) as
(!hg)~r
E[
2(317'  <iTl.'
(6.39)
M'
fi:L_ M ;()1T8)
271
The equation of motion in terms of the relative motion u given by eq. (350) for the undamped syslem as
Y (I)  y, (I) is
(6.34)
(6.40)
and
K"
1 1 : cos'dx o ,)L) 2L
(6.35)
<
(17 \'
11X
where M' is given by eq. (6.34), K; by eq. (6.38), and the effecrive force by eq. (6.17) for the effective distributed force and by tMAg (I) for the effective concentrated force at the ~op of the lOwer. Hence
".41 K ;321!
F;" (I) =
where Pdr(X,t) = 
L p,,,(x, 1)</1 (x) dx  I"La, (1) Jo ina;;(t) is the effective distributed force. Then
The axial force is due to the weight of tbe lower above a particular section, including Lhe concentrated weight at the top, and may be expressed as
(6.36)
F;,,;
iiw,(r)1>(x)dxliILa,(r)
Substitution of ;:P (x.) from eq. (6.32) :nto the last equalion yields upon in~ {egration
F;,,;  .....~~('"'I)~L~ ('TT 1)
71
where 8 is the gravitaliona: acceleration. Since the normal force in this case is a function of x, it is necessary in using eg. (6,22} to include N(::::.) under the integral sign, The geometric stiffness coefficient K~ is then given by
K~=
(6.41)
dO
Example 6.2.' As a numerical example of c~lculating the response of 2. system wIlh distributed properlies, consider the water tower shown in ~ig, 6.7 excited by a sinusoidal ground acceleration i.l g (t) = 20 sin 6,361 (in Isec~), Model the slIucwre by assuming the shape given by eq, (632) and determine the response,
Solution:
The numerical values for this example are:
It. = 0.1 KJp' sec 1in per unit of length
1
EJ
L~
32L'
16
4)
(6.38)
w= 636 rad/sec
176
177
From eq, (6.34) the generaHzed mass is 0.1 X 1200 21T (51T 8) Kip'sec:!
').. F4'::
m
.".
!47.21~';.111
1
Fl~ 1..
w4 L2 1013
32 X
(i200)"
'::~(31T'  4)
0.1 X 386
1
= 2[,077 Kiplin
F,(')
m.
,,"
('j
w=/K;/lvr
and the frequency ratio
11.96rad/sec
=
6.36
'"
11.96
=0.532
(0)
(<l
Fig. 6.8 (a) Multistory building sllbjected to lateral forces F,Ct). (b) Column mode:nng the building s'bowing assumed lateral displacement function y{x,t)"" tP(x)Y(t), {c) Equivalent stngle~degreeof~freedom system, (d) Free body diagram.
6.5
or
F;" =
Hence the steadysta:e response (neglecting damping) in terms of relative motion is given from eq. (3.9) as
Consider a multistory building such as the model for the fourstory building shown in Fig. 6.8(a) subjected to lateral dynamic forces Fi(t) at the different levels of the building. The mass of the building 15 assumed to be concentrated at the various levels (floors and root) which are assumed to be rig;d in their own planes~ thus only horizontal displacements are possible in such a building" To model this structure as a single degree of freedom [Fig. 6.8(c), the lateral dIsplacement shape y (x, t) is defined jn terms of a single generalized coordinate Y(I) as
y(x,l) = w(x)Y(t)
(6.42)
(Ans.)
The generalized coordinate ye'l in eq. (6.42) is selected as the lateral displacement at the top level of the building which requires that shape function be assigned a unit value at that level; that is. (H) = 1.0 where H is the height of the building.
I
I
178
179
The equation of motion for the generalized singledegreeoffreedom system in Fig. 6.8(c) is obtained by equating to zero the sum of forces in the corresponding freebody diagram [Fig. 6.8(d)], that is (6.43) where MO, C', KO, and FO (t) are, respectively, the generalized mass, generalized damping, generalized stiffness, and generalized force, and are given for a discrete system, modeled in Fig. 6.8(b), by
M ~
(6.47) The substitution of the generized damping coefficient, (, from eq. (6.46) and of K = w 2Mo from eq. (6.47) into the differential equation of motion, eq. (6.43), results in
O
I I I
m;q,;
(6.44a)
;"'1
N
C ~
K ~
c,Llq,;
k;Llq,;
(644b)
i= I N
conveniently measured relative to the motion of the base Yo (I), that is, (6.44c)
U;
;""1
(I)
y; (I)  Yo (I)
F~Jl,i
(6.49)
(t) ~
F;Ct)<P;
(644d)
i= I
Also, in this case, the effective forces are given from eq. (4.39) as
F~ff,i
where the upper index N in the summations is equal to the number of stories or levels in the building,
=  m,yo (I)
(6.50)
The various expressions for" the equivalent parameters in eq. (6.44) are obtained by equating the kinetic energy, potential energy, and the virtual work done by the damping forces and by external forces in the actual structure with the corresponding expressions for the generalized singledegreeoffreedom system. In eq. (6.44), the relative displacement d, between two consecutive levels of the building is given by . (6.45) with o = 0 at the ground level. As shown in Fig. 6.8(b), Tn, and F;U) are, respectively, the mass and the external force at level i of the building, while k i and Ci are the stiffness and damping coefficients corresponding to the ith story. It is convenient to express the generalized damping coefficient C of eq. (6.44b) in terrns of generalized damping ratio thils by eg. (2.7)
The differential equation for the generalized singledegreeoffreedom system excited at its base is then written as (6.51) where the generalized coordinate for the relative displacement, V(r) is
U(I)
~
YCt)  )'0(1)
(6.52)
The generalized effective force F;JJ(t) is calculated by eq. (6.44d) and (6.50) as
F;ff~ yo(1)
m;<P;
I
(6.53)
r;
j=
c" ~
(6.46)
i= I
(6.54)
180
181
r'
r'=
(6.55)
6.6
SHAPE FUNCTION
The use of generalized coordinates transforms a multidegreeoffreedom sysu tem into an equivalent singledegreeoffreedom system. The shape function describing the deformed structure could be any arbitrary function that satisfies the boundary conditions, However, in practical applications, the success of this approach win depend on how close the assumed shape function approximates the actual displacements of the dynamic system. For structural buitdings, selection of the shape function is most appropriate by considering the aspect ratio of the structure, which is defined as the ratio of the building height to the dimension of the base. The recommended shape functions for highrise. midrise, and tow~rise buildings are summarized in Fig. 6,9. Most sejsmic building codes use the straightline shape which is shown for the midrise building. The displacements in the structure are calculated using eq. (6.7) after the dynamic response is obtained in terms of the generalized coordinate.
... I '" I I I !
'"
J'
ii OJ

8@12ft96ft
(a)
,~
(
~ +
AW
A'
I __ J...
!!
N
iii
'"
~
A fourstory reinforced concrete framed building has the dimensions shown in Fig. 6.1 O. The sizes of the exterior columns (nine each
on lines A and C) are 12 in X 20 in. and the interior columns (nine on line B)
Example
6.3~
24ft
1
Af
(0)
24ft
1~
J
are 12 in X 24 in for the bottom two stories, and, respectively, 12 in x 16 in and 12 in X 20 in for the highest two stories, The height between floors is 12 ft. The dead load per unit area of the floor (floor slab, beam, half the weight
Fig. 6,10 Plan and elevation for a fourstory buHding for Example 6.3; (a) Plan. (b)
Elevation.
Ix
I I 82t8 hi orx)
~Dj
pfl.
q,(x)=xfH
~!~'"
(b) MIDR1SE
(c) HiGHRISE
of columns above and below the floor, partition walls, etc.) is estimated to be 140 pst The design llve load is taken as 25% of an assumed live load of 125 psf. Determine the generalized ma~s, generalized stiffness, generalized =0.1), and the fundamental period for lateral damping (for damping ratio vibration perpendicular to the long axis of the building. Assume the foHowing shape funetio"s: Ca) (x) =xiH and (b) ",(x) sin ('11Xi2H) where H is the height of the building.
(a) .l,.oW~RISE
I.5<HID<3
HJD>3
{:rJ >., 1 cosnx
12U
Fig. 6.9 Possible shape functions based on aspect ratio (Naeim !989. p. 100).
I
1
Solution: 1. Effective weight at various floors: No live load needs to be considered on the roof. Hence. the effective weigbt at all floors, except at the roof, will be 140 + 0.25 X 125 = 171.25 pst and the
182
183
effective weight for the roof will be 140 pst. The pbn area is 48 ft x 96 ft 4608 ft~. Hence, the weights of various leve~s are:
and for the cobmns 12 in X 20 in, I 1 12x20' , 8000 in , 12 X 3 X 10' x 8000 k = 144'
Kips K,
~
= 965 '
Kiplin
789.1 X3
2. StOI)' lateral sliffness: It wjll be assumed t:'1at horizontal beamandfioor diaphragms are rigid compared to the cobmns of the building in order to simplify the hand calculation. In this case, the stiffness belween two consecutive Ievels is given by
k=
12El
where
and the generalized critical damping C;, and [he absolute generalized damping C', respectivaly, by eqs, (2,6) and (2J9) as
C;, = 2[j('M'
I:;: U 12 X 2D 3 = 8000 in4 (moment of inertia for the concrete section for
columns 12 in x 20 in) Therefore, for these co:umns,
1 and
c=(C'u
96.450 Kip!in
k=
TABLE 6.1
"':
1.000
L"p,
0,250
and
The total stiffness for lhe first and second stories is [hen K, = K,
0,750 0.250
= 18 X 96.45 + 9 x
1758
166,67 = 3236 Kiplin
2
0500
0250 0,250 0,250
Similarly, for the columns 12 in X 16 in. of the lOp and third stories. I ' ,2 X 16' = 4096 in', 12
k=
~
12x3X lO'X4096
....~~:=
49A Kiplin
625250
184
Slrl!ctures Modeled as a
SjngJeOegree~ofFreedom
System
185
TABLE 6.2
6.7
RAYLEIGH'S METHOD
Level
4
,p,
1.000
1,p;
1n;7
1.671
k;tJ.l/Ji
(Kip sec' fin) (Kip lin) 0.076 10.154 1.745 0.217 82.782 1.022 0.324 339.702 0.300 0.383 476.686
1.671
1758
3
1758
2.040 2.040
3236
2.040 3236
M*
= 4.738 K* =
909.324
In the preceding sections of this chapter the differential equation for a vibrating system was obtained by application of the principle of virtual work as an alternative method of considering the dynamic equilibrium of the system. However, the differential equation of motion for an undamped system in free vibration may also be obtained with the application of the Principle of Corr~ servation of Energy. This principle may be stated as follows: 1f no external forces are acting on the system and there is no dissipation of energy due to damping, then the total energy of the system must remain constant during motion and consequently its derivative with respect to time must be equal to zero. To illustrate the application of the Principle of Conservation of Energy in obtaining the differentia] equation of motion, consider the springmass system shown in Fig. 6,11" The total energy in this case consists of the sum of the kinetic enorgy of the mass and the potential energy of the spring_ In this case the kinetic energy T is given by (6.56) where y is the instantaneous velocity of the mass. The force in the spring. when displaced y units from the equilibrium position, is ky and the work done by this force on the mass for an additional displacement dy IS  ky dy. This work is negative because the force ky acting on the mass is opposite to the incremental displacement dy given In the positive direction of coordinate y. However, by definition. the potentia] energy is the value of this work but with opposite sign. It follows then that the total potential energy V in the spring for a final displacement y will be
Table 6.2 shows the necessary calculations to obtain using eqs. (6.44a) and (6.44c) the generalized mass M" and the generalized stiffness K for this example assuming ,ptA) = sin (ml2H). The natural frequency is then calculated from eq, (6.47) as
IM'
'i.'
and the generaljzed critical damping C~.;t and the absolute generalized damping C, respectlvaly, by eqs. (6.48) and (6.49) as
c,,=2[j?M' =2j(909.324)(4.738)
and C =
= 13L276(1bseclin)
v=
J:
kydy
jky'
(6.57)
Adding eq. (6.56) and (6.57), and setting this sum equal to a constant, will give
For this example, either of the, two assumed shape functions results in essentially the same value for the fundamental period, However, (x) = x III is a slightly better approximation I to the true deflected shape than is </> (x) sin ('l7X 12ff) because To> T,.
~he
structure
will vibrate elmer to the free condition with less imposed constraints, thus with the stiffness
reduced and longer period.
186
Slructures Modeled as a
Sir;g!e~Degree~oi~Freedo:n
System
187
myy + lyy =0
Since
y cannot
~hat
(659)
Fig. 6,12
Spring~mass
simple oscillator of Pig, 6.1l, and assurr.e that the motion is hannonk. This assumption leads to the equation of motion of the form
which is the natural frequency for the simple oscillator obtained previously from the differential equation of motion. This method, in which the naturai frequency is obtained by equating maximum kinetic energy with maximum potential energy, is known as Rayleigh's ,"1etltod.
(6.60)
y= wC cos
(WI + a)
(6.61)
where C is the maximum displacement and we the maximum velocity. Then, at the neutral position (y = 0). there will be no force in the spring and the potential energy is zerO. Consequently, the entire energy is then kinetic energy and (6.62)
At the maximum displacement the velocity of the mass is zero and ail the
Example 6.4. In the previous calculations on the springmass system, the mass of the spring was assumed to be so smaH that its effect on the natural frequency could be neglected. A better approximation to the true value of the naturaJ frequency may be obtained using Rayleigh's Method. The distributed mass of the spring could easily be considered in the calculation by simply assuming that the deflection of the spring along its length is iinear. In this case, consider in Fig. 6.12 the springmass system for which the spring has a length L and a total mass In r Use Rayleigh's Method to detemline the fraction of the spring mass that should be added to the vibrating mass.
Solution: The dispiacemen~ of an arbitrary section of the spring at a distance s from the support will now be assumed to be u = sylL. Assuming iliat the motion of the rr.ass 111 is harmonic and given by eq. (6.60), we obtain
energy is then potential energy, thus (6.63) The energy in the system changes gradually over onequarter of the cycle from purely kinetic energy, as given by eq. (6.62), 10 purely potential energy. as given by eq. (6.63). If no energy has been added or lost during the quarter cycle, the two expressions for this energy must be equaL Thus (6.64) Canceling common factors and solving eq, (6.64) wi!] give
C Sin (wt+ al
D
(6.66)
The potential energy of the uniformly slretched spring is given by eq. (6.57) and its maximum value is (6.67) A differential element of the spring of length ds has mass equal to nlflslL and maximum velocity um~. = &tim~, wsCIL. Consequently the lotal kinetic energy in the system at its maximum value is (6.68)
.1 .
~
~
l
(6.65)
,j;'i
;, '1 i , ,
"
,i .i
:,i :.."
;~I
:l
188
~89
After integrating eq. (6.68) and equating it with eg. (6.67}, we obtain
kC ~UJC 2 2
2 2( m+I m,,\
3 !
(6.69)
E=:=~L~~r
y
(a}
(OJ
Fig. 6.13 (a) Cantilever beam of uniform maSS with a mass concentrated at its tip. (b)
(6.70)
Assumed deflection curve.
(i)
end as shown in Fig. 6.13(b). For this static load the deflection at a distance x from the support is
(6.72)
Rayleigh's Method may also be used to determine the natural frequency of a continuous system provided that the deformed shape of the structure is described as a generalized coordinate. The deformed shape of continuous structures and also of discrete structures of multiple degrees of freedom could in general be assumed arbitrarily. However, in practical applications. the success of the method depends on how close the assumed deformed shape will come to match the actual shape of the stmcture during vibration. Once the deformed shape has been specified, the maximum kinetic energy and the maximum potential energy may be determined by application of pertinent equations such as eqs, (6.8) and (6.13). However, if the deformed shape has been defined as the shape resulting from statically applied forces, it would be simpler to calculate the work done by the extema] forces, instead of directly determining the potential energy. Consequently, io this case, the maximum kinetic energy is equated to the work of the forces applied staticaHy. The fonowing examples illustrate the application of Rayleigh's Method to systems
with distributed properties.
where y = deflectio71 at the free end of the beam. Upon substitution into eq, (6,72) of y ~ C sin (WI+ a). which is the harmonic deflection of the free end, we obtain
u
(6.73)
Tbe potential energy is equated to the work done by the force F as it gradually increases from zero to the final value F. This work is equal to i Fy, and its maximum value which is equal to the maximum potential energy is then
Ym~'" = iFC= 2L3 L.
3El '"
(6,74)
since the force F is related to the maximum deflection by the formula from elementary strength of materials,
Yma~=C= 3EI
The kinetic energy due to the distributed mass of the beam is given by
FL'
(6,75)
Example 6.5. Determine the natural frequency of VIbration of a cantilever beam with a concentrated mass at its end when the distributed mass of the beam is taken into account. The beam has a total mass mb and length L. The flexural rigidity of the beam is EI and the concentrated mass at its end is m, as shown in Fig. 6.13.
Solution: It will be assumed that the shape of deflection curve of the beam is that of the beam acted upon by a concentrated force F applied at the free
T=
lL .!..L~~~)U2
o 2
I.,
dx
(676)
L !
and using eq. (6,73) the maximum value for total kinetic energy will then be
(6,77)
190
191
After integrating eg. (6.77) and equating it with eq. {6.74), we obtain
31 C' 2L'
140,
(6.78)
have durjng vJbra ..ion, A choice of a shape that gives consistently good results is the curve produced by forces proportional to the magnitude of lhe masses acting on lhe structure. For the simple beam. these forces could be assumed to be the weights W! = mig, W2 = tn18,,, . W N = mNg due to gravitational action on the concentrated masses. The static deflections under these weights may then be designated by y" y, .. _, YN The potential energy j5 tben eq~al to work done during toe loading of the beam, :hU5,
(6.81)
f~~
277
211 1 ,I
\11 L
1
'\
31
(6.79)
,m+40
33.\
II 'b'
!1lb at the end of the beam, a more accurate value for Lhe natural frequency of the cantilever beam is obtained compared to the resull ob~'iined by simply neglecting its dist.ributed mass. In practice the fraction 331140 is rounded to 114, thus approximating the natural frequency of a cantilever beam by
For harmonic motion in free vibration, the maximum velocities under the weights would be WYh W),?, ... , WYN, and therefore the maximum kinellc energy would be
(6.82)
J=
(6.80)
When the maximum potentia! energy, eq. (6.81), is equated with the maximum kinetic energy, eq. (6,82), the natural frequency is found to be where
The approximation given by either eq, (6.79) or eq, (6,80) is a good one even for the case in which m = O. For this case the error given by these formulas is about 1.5% compared 10 the exact solution which will be presented in Chapter 21.
(6.83)
or
Example 6.6. Consider in Fig. 6.14 the case of a simple beam carrying several concentrated masses. Neglect the mass of the beam aod determine an expression for the natural frequency by application of Rayleigh's Method.
Solution: In the application of Rayleigh's Method, it lS necessary to choose a suitable cu~e to represent the defonned shape that tbe beam will
)'i
it must be remembered that these are not gruvity forces at all but substituted
forces for the inertial forces. For example, in the case of a simple be.am with overhang (Fig. 6.15) the force at the free end should be proportionai to m)(F1 m3g) but directed upward in order to obtain the proper shape for the defomled heam. In toe application of Rayleigh's Method, the forces producing the deflected shape do not necessarUy have to be produced by gravitational forces. The only requirement is that lhese forces produce the expected deflection shape for the
y,
v,
y,
192
193
m1
""
_Y,
fundarr:ental mode, For example, if the deflected shape for the beam shown in Fig. 6.14 is produced by forces designated by f,. j" .... f" instead of the gravitation forces WI! W~, . " W,\', we wiU oblain, as in eq. (6,81), the maxjmum potential energy (6.84) which equated to the maximum kinetic energy. eq, (6.82), will result in the following formula for the fundamental frequency:
Then an improved value for the natural :requency may be obtained by loading the strocture with the inertial loads associated with the assumed deflection. This load results in a new deformed shape which is used in calculating the maximum potential energy. 'nle method is better explained with the aid of numerical examples. Example 6.7. By Rayleigh's Method, determine the natural frequency (iower or fundamental frequency) of the two~story fr.!rne shown in Fig. 6.16. Assume that the horiz.ontal members are very rigid compared to the columns of the frame. This assumption reduces the system to only two degrees of freedom, indicated by coordinates YI and Y2 in the figure. The mass of the structure, which is lumped at the floor levels, has values mj = 136 lb sec 1 lin and!tl2 = 66 Jb sec1lin. The total stiffness of the first Story is k j = 30,700 Iblin and of the second story k2 = 44,300 ib lin, as indicated in Fig. 6,16,
(6.85)
Solution: This structure may be modeled by the two mass systems shown in Fig. 6.17. In applying Rayleigh's Method, let us assume a deformed shape for which Y1 1 and Y1 = 2. The maximum potentia! energy is then
(6.86)
~ !(30,700)(1)' + 1(44,300)(1)'
6.8
37,500 lb in
(a)
The concept of applying inertial forces as static loads in determining the deformed shape for Rayleigh's Method may be used in developing an improved scheme for the method. In the application of the improved Rayleigh's Method, one would start from an assumed deformation curve followed by the calculation of the maximum values for the kinetic energy and for the potential energy of the system. An approximate va!ue for natural frequency is calculated by equating maximum kinetic energy with the maximum potential energy.
~ I (136)",'
+ 1(66) (2w)'
(b)
=200w'
194
195
or in the ratio
y, = 1.00
y,
= 1.34
(c)
Introducing these improved values for the displacements YI and h into eqs. (a) and (b) to recalculate the maximum potential energy and maximum klnetic energy results in
(d) (e)
= 12,57 rad/sec
Equating maximum pOlential energy with max.imum kinetic energy and solving for the naturai frequency gives
w~
Of
~. ~
f= 2.00 cps
This last calculated value for the natural frequency f= 2.00 cps could be further improved by applying a new inertial load in the system based on the last value of the natural frequency and repealing a new cycle of calculations. Table 6,3 shows results obi.ained for four cycles. The exact natural frequency and deformed shape, which are calculated for this system in Chapter 10, Example 10.1, as a twowdegrees~offreedorn system, checks with the values obtained in tl1e last cycle of the calculations shown in Tab!e 6.3.
13.69 fadlsec
w 2".
2.1& cps
The natural frequency calculated as f;;;:; 2,18 cps is only an approximation to the exact value, since the deformed shape was assumed for the purpose of apply:ng Rayleigh's Method. To improve this calculated value the ~atu~al frequency. :et us load the mathematical model in Fig. 6. 17(a) with the mertlal load calculated as
.of
6.9
SHEAR WALLS
w'y,
Horizontal forces in buildings, such as those produced by eanhquake molion or wind action, are often resisted by structural elements called shear walls.
TABLE 6,3 Improved Rayleigh's Method Applied to Example 6.6
. j
The equilibrium equations obtained by equatjng to zero tbe sum of the forces in the free body diagram shown in Fig. 6,l7(b) gives
30,700y,
= 25,489
24,739
Deformed
Cycle
Inertial Load
Natura]
Fl
25,489 21,489 19,091
F2
24,739 ;8,725 12,230
Frequency
2.18 cps 2.00 1.88 1.88
:1
!
44,300(y,  y,)
and solving
YI
'~'~~~~~
"~'
= 1.64
=
J
4
I: 1.32
1.1.27
y,
2.19
196
Structures MOdeied as a
x
Slng\e~DegreeofFreedom
Sysler,
197
+. x
Floor mass
L y
L
ioj
Fig. 6.18 Mathematical model for shear wall and rigid floors
The de.formed shape equation is assumed as the deflection curve produced on a cantdever beam supporting three concentrated weights lV, as shown in Fig. 6.19. The static deflections Yit Y:h and Y3 cafculated hy using basic knowledge of strength of materials are
These structural elements are generally designed as reinforced concrete walls fixed at the foundation. A single cantilever shea! wall can be expected to behave as an ord;nary flexural member if its lengthtodepth ratio (LlD) is greater than about 2. For short shear walls (LID < 2), the shear strength assumes preeminence and both flexural and shear deformations should be considered in the analysis. When the floor syscem of a multistory buildIng is rigid, the structure's weights or masses at each floor may be treated as concentrated loads, as shown in Fig. 6.18 for a threestory building. The response of the structure is then a function of these masses and of the stiffness of the shear walL In practice a mathematical model is developed in which the mass as weI! as the stiffness of the structure are combined at each fioor ieveL The fU.I1damental frequency (lowest natural freqoency) for such a structure can then be obtained using Rayleigh's Method. as shown in the following illustrative example. Example 6.8. Determine> using Rayleigh's Method, the natural period of the threestory building shown in Fig. 6.18. All the floors have equal weight W. Assume the mass of the wall negligible compared to the floor masses and consider only flexural deformations (1... fD > 2).
Solution: The naturai frequency can be calculated using eq. (6,83), which is repeated here fOf convenience:
y,
y, ~ 162 y, ~ W
49 V(L J
El
WL 3
0.3025 EI
92 WL'
El ~ 0.5679
WL'
EI
(b)
El
or
lEI
i
~
,,,iIg"Wv. L.. UI
Solution: The lateral deflection Lly,,> conSidering only shear deformation for a beam segment of length L1x, is given by
(a)
w= 1,,
i= 1
Y ~WDI
(a)
'",."". ,
198
where
Rayleigh's Method
199
TABLE 6.4
v=
A
shear force
cross~sectional
Calculation of the Natural Frequency for the Shear Wall Modeled as Shown in Fig. 6.18
area
DIL
O~OO
y, (in)
0,09259 0,)3600
O~26620
Yl" (in)
y, " (in)
0~56790
w" (radisec)
roo (cps)
~~~~~
~~~~~~~
O~50
0.30247 0.37483
O~59193 O~95376
0.65465
O~91489
29,66 27.67
first story is
3W(Ll3) WL y" = ~~ aAG   aAG
(b)
23.30
19~O5
0.48322
O~78704
Ll77GS 1.65509
1.46032 2.11161
2~90764
\.34862 1.95585
2~736S8
15.71
13~21
3.69079
11.33
~~
~~~~~~~~~~~~~
At the second floor the shear deflection is equal to the first floor deflection plus the relative deflection between floors, that is
YJl
Y~I
2W(U3)
aAG
5WL 3aAO
Factor of
(c)
fE/iwI.',
since the shear force of the second story is V= 2W, and at the third floor
6WL 3aAG
The next iHUSlrative example presents a table showing the relative import~ ance that shear deformation has in calculating the natural frequency for a series of values of the ratio D i L
(d)
The total deflection is then obtained by adding the flexural deflection determined in Example 6.8 to the above shear deflections. Hence,
Example 6.10. For lhe structure modeled as shown in Fig. 6.18, study the relative imporlance of shear deformation in calculatin.g the natural frequency.
SolWton: In {his study we will consider, for the wall, a range of values from 0 to 3.0 for the ratio D IL (deplhto~length ratio). The deflections Yh Yl, y} al the floor levels are given by eqs. (f) of Example 6.9 and the natural frequency by eq. (6.83). The necessary calculations are conveniently shown in. Table 6.4. It may be seen from the last column of Table 6.4, that for this example the natural frequency neglecting shear deformation (D /L = 0) . is
Y ,162
Y,
IS WL)
i '
WL aAG 5WL
49 WL'
We can see better the relative importance of the shear contribution to the total deflection by factoring the first terms in eqs. (e). Considering a rectangular waH for which A = D X 1, E 10 2~S, 1= ID'1i2, a = 1.2 (I = thickness of the wall), we obtain
IS WL ' [ 1+1.8751~1 ID\'] y,=~
4~72[EIjWL' cps, For short walls (DII. mation becomes increasingly important
> O~S)
6.7
SUMMARY
162 EI
;L j
y,
~I+O~9S7l 162 El L i
162 El ,
49
WL) [
'D)"
y,~~
92 WL' r1 +O~611\{D)'1,
L
j
(f)
The concept of generalized coordJnate presented in (his Chapter permits the analysis of multiple interconnected rigid or elastic bodies with distributed ProPM erties as singledegrceoffreedom systems. The analysis as one.,degree~of freedom systems can be made provided that by the specification of a single coordi~ nate (the generalized coordinate) the configuration of the whole system is determined. Such a system may then be modeled as the simple osciilntor with its various parameters of mass, stiffness, damping, and load, calculaLed to be
200
Structures MOdeled as a
SingleDegree~ofFreedom
System
201
dynamica!ly equivalent to the actual system to be analyzed. The sOlution of thIs model provides the response in terms of the generalized coordinate. The principle of virtual work which is applicable to systems in static or dynamic equHibdum is a powerful method for obtaining the equations of motIon as an alternative to the direct appiication of Newton's law. The principle of virtual work states that for a system In equil!brium the summation of the work done by ail its forces during any displacement compatible with the constraints of the system is equal to zero. Rayleigh's Method for detennining the natural frequency of a vibrating system is based on the principle of conservation of energy_ In practice, it is applied by equating the maximum potential energy with the maximum kinetic energy of the system. To use Ray[eigh's Method for the detennining of the natural frequency of a discrete or a continuous system, it is necessary to assume a defonned shape. Often, this shape is selected as the one produced by gravitational loads acting in the direction of the expected dispJaceme,lts. This approach leads to the following formula for calculating the nmuml freqc:ency:
1
6.2 6.3
, . Y
Un!fOf/l'l disk
TOldl mass'" m
!~e:(tMsibleCilble
Rigid beaM Total m<l~ '" 200
:
!
Fig. P6.2.
Determine the generalized quantities M", C', K", and F (t) for the structure shown i:~ Fig. P6.2. Select yet) as the generalized coordinate. Determine the generalized quantities M, Co., KA, and F" (I) for the structure shown in Fig. P6.3. Select OCt) as the generalized coordinate.
I i 82.: WiY,
w=
I~'~
eq. (6.83)
2.: W,y;
where }; is the deflection at coordinate i and W, concentrated weight at this coordinate. Shear walls are structural walJs designed to resist lateral forces in buildings. For short walls (LID:::; 2) shear defonnations are important and should be considered in the analysis in addition to the flexural deformations.
PROBLEMS
6.1 For the system shown in Fig. P6.1 determine the generalized mass M', damping C, stiffness K', and the generalized load F* (r). Select Y(r) as the generalized coordinate.
Y(I)
Fig. P6.3.
FOI (he elastic cantilever beam shown in Fig. P6A, determine the generaiized quantitIes M", K, and F~(r). Neglect damping, Assume that the deflected shape is givea by (x) = 1 cos('rrxI2L) and select Y(l) as the generalized coordinate as shown in Fig, P6A. The beam is excited by it concentrated force F(f) Frf(r) at midspan.
1
6.5
~'ig.
P6.J.
Defermine the generalized geometric stiffness KG for the system in Fig. P6A if an axia! tensile force N is applied at the free end of the beam aIong the x direction. What is the combined generalized stiffness K;?
'202
203
FltJ
ConCI!n\r;;1ted
Flexurz:1
rigidity As.ru'fled
=:0
man ""m
EI
~hape
Fig. po.g.
Fig. PM.
6.6 A concrete conical post of djameter d at the base and height L is shown in Pig. P6.e;, It is assumed lhar the wind produces a dynamic pressu:e poet) per unit of projected area along a vertical plane. Determine the generalized quantities M', K', and F" (1). (Take modulus of elasticity E{ "'" 3 X 10 6 psi; specific weight y= ISO lb/ft) for concrete.)
6.9
Determine the natural frequency of the simply supported beam shown in Fig. P6.9 using Rayleigh's Method. Assume the deflection curve given by 2 (x) = Y sin 1'fX IL The total mass of the beam is IItb = 10 th sec /in, flexural rigidity 1 lOt lb inl, and length L = 100 in, The beam. carries a concentnncd mass at the center m = 5 lb ' sec" lin.
m,
~Y(/}
T
~Lll
,.dj
Fig. P6.6.
6,7 A simply supported beam of total unifomlly distributed mass mb, flexural rigidity Ef, and length L, carries a concentrated mass m at its center. Assume the dellection curve to be the defleclion curve due to a concentrated force at the cerner of [he beam and determine the natural frequency using Rayleigh's Method. Determine the natural frequency of a simply supported beam which has a [otaluniformly distributed mass ma. flexural rigidity sions shown in Fig, P6.8. Assume that during vibration the beam is of the shape produced by a concentrated force applied at the beam. with overhang 1, and dimen deflected curve free end of [he
Fig. PG.9.
6.10 A twostory building ls modeled as the frame shown in Fig, P6,lO. Use
Rayleigh's Method to determine the natural frequency of vibration for the case in which only flexural deformation needs to be considered. ~eglect the maSS of the columns and assume rigid floors. (Hint: Use cq. {6,83)l.
~W===1IY'
!\Total)
{ (Toul)
+,
L
~L...
Fig. P6.10.
6.8
6.11
Solve Problem 6.lD for the case ir:. which the columns are short and only shear deformation needs to be considered. (The lateral force V for a fw~.d column of
204
Structures Modeled as a
Singl$Degree~o~~Freedom
System
length L. cross~sec(!onal area A, is approximately given by V = AG,d IL, where the lateral deflection,) G is the shear modulus of eiasticiry and
6.12
Calculate the natuml frequency of rhe shear waH carrying concentrated masses at the floor levels of a [crees.tory belJding as shown in Fig, P6.12. AssurTle that the deflec~ion shape of the shear wall is rhat resulting from a conceO[ra~ed larera! force applied a~ its rip_ Take :rexural rigidity, 1 3,0 X 101< lb  in 2; length L 36 ft; concentrated masses, m = 100 Ib  sec! li n and mass pe: unit of length along the wull, m= 10 lb sec 2 /inz,
y
7
Nonlinear Structural Response
Fig. P6.12.
6.13 Solve Problem 5.12 on the assumption that the deflection shape of the shear waH IS (hat resulting from a lateral unifonn load applied along irs lengrh.
In discussing the dynamic behavior of singledegree~offreedom systems, we assumed that in the model representing the structure, the restoring force was proportional to the d:spiacemeor. We also assumed the diSSipation of energy ~hrough a viscous damping mechanism in which the damping force was proportional IO the velocity_ In addition, the mass in the model was always considered [0 be unchanging with time. As a consequence of these assump~ tions, the equation of motion for such a system resulted in a linear, second~
(7. I)
In the previous Chapters it was iHustrated that for particular forcing functions such as harmonic functions, jt was relatively simple to solve this equation (7.1) and that a general solution always existed in tenus of Duhamel's integraL Equation OJ) thus represents Ihe dynamic behavior of many structures modeled as a singledegreeoffreedom system, There are, however, physical situations for wh:ch this linear model does not adequately represent the dynamic characteristics of the structure, The analysis in such cases requires the
20:5
2GB
Structures Modeled as a
SlngleOegreeof~Freedom
Syslern
207
introduction of a model in which the spring force or the damping force may not remain proportional, respectively, to the displacement or to the velocity. Consequentiy, the resulting equation of motion will no longer be linear and its mathematical so]mjon. in genemJ, wi1l have a much greater complexity. often requiring a numerical procedure for its integration.
F,(yl
slope: secant
,
siope tangen, ~
: 1:,.:;, .....
7.1
Figure 7. J (a) sho'w's the model for a sing\edegreeoffreedom system and in Fig. 7,1 (b) the corresponding free body diagram. The dynamic equilibrium in the syslem is esmbllshed by equating (0 zero the sum of the inertial force F t (l) tbe damping force F D (1), the spring fo:ce F, (t), and the extema! force F (f). Hence, at time t, the equilibrium of these forces is exp:essed as
(7.2)
Subtracting eq. (7.4) from eq. (7.5) results in the difference equmion of motion in terms of increments, na,T',ely
(7.6)
Considering the case that in (his equation, the mass is constant, F, (r) = my; that the damping force is proportional to [he velocity with the damping coefficient also constant, F b (r) = cj; and the resistance force or spring force is a fUnction of the dispiacernem, Fj(f) = F;(y), we may then express eq. (7.2) as
(7.3)
Hence, at time t,. using the notation Yi = Y (().
Ft:rthermore, we ass:Jme that t:le incremental resisting or spring force is proportional to the incremental displacement, thar is i1F J (yj) = k,Lly"
'ii =
Fit,)
y(t,) und
Yi = Y (t
l ),
(7,7)
+ .c.1;
(7.5)
~Y
,4
.~~
Spdng
~~~~
~!
Damper
~lE
f+ F(c}
The coefficient k i is cefined as the c"JITem evaluation for the resisting force per unit displacement (st1ffness coefficient) whlch may be taken as the slope of the tangent of the force~disp!acemem function, at the initiation of ~he time step .::1; for ~hc imcrval L1t or as the slope of the secant Jir:e as s.hown in 7.'2 fOl the plot of the resisting force F$lV). The value of (he coeft1cjenl Xi is calculated at a displacement corresponding to time l; ar:d assumed to remain constant during t~e increment of time Lk Since, in general, this coefficient does not rerr.ain constant during thar time increment, eq. (7.7) lS an approximale equation. The incremental displacement .1)", incremental velocity Lly, and incremental acceleration Jy are given by
F,. (tl
F(}
(7.8)
=
Fig. 7.1 (a) Model for a singJe~degree~offreedom system. (b) Free body ding:am showing {he inertial force, the damping force, the spring force, and the eXtern::>l force.
(7.9)
J,v, = 'Y(I,
(7.10)
208
209
7<2
Among [he many methods available for (he solution of the nonlinear equation of motion. probably one of the most effective is the stepbystep inregrntion method. 1:1 this method, the response is evaluated at successive increments ,dt of time, usually taken of equa~ lengths of time for computational convenience. At the beginniilg of each interval, the condition of dynamic equilibrium is established. Then, the response for a rime increment .1I is eva:uated approximarely 00 the basis that (he coefficients k (y) and c CY) remain constant during the intervai ,jt. The nonlinear characterisrics of rhese coefficiems are considered in rhe analysis by reevaluating these coefficients at (be beginning of each tIme increment. The res.ponse is then obtained using the displacement and velocity calculated at the end of the time interval as the initial conditions for the next time step. As we have said for each time interval, the stiffness coefficient k(y) and the damping coefficient c (j) are evaluated at the initiation of the interval but are assumed to remain constant until the ne.~t seep; thus (he nonEnear behavior of the system ~s ap?roximated by a sequence of successively changing linear systems. It should also be obv:ous that the nssumptlon of constant mass is unnecessary; it could just as weil also be represented by a var:able coefficient There are many procedures available for performing the stepbystep imegrarion of eq. (7.12). Two of the most popular methods are the constant acceleration method and the linear acceleration method. As the names of these methods imply, ;n the first method the acceleration is assumed to remain constant during the (ime interval .1r, while in the second merhod, the acceleration is assumed to vary linearly during the ir::tervaL As may be expected, the constant acceleration method is simpler but less accurme when compared with the linear acceleration method for the same value of rhe time incremem, We shall present here in detail both methods.
ti .......... t I f f ' :
A,
l
Fig. 73
Integrating this equation twice with respect to the time between the limits and! ::esults in
y(r)=:i<~
j
<
I,'
20';+),,+J(II,)
(7.J 2)
and
y(t) =y, +yII,)
+
l, +
I
(7J3)
= I,
+ ill gives
(n4)
and
(7.15)
7.3
In the constarnt acceleration method, it is assumes that acceleration remains t; and t, ~ 1 = [, + .1l as shown in 73.
The value of L'1e constant acceieration during the interval is taken as the average of ~he values of the acceleration )I, at the inii:iation of [he time step and y,.;. :, the acceleration at the end of the time step. Thus, the acceleration y (r) at any time I during rhe time interval Lit is given by (7< II)
at
where ay; and Lly; are respectively the incremental displacement and incremental velocity defined by egs (7.8) and (7<9). To use the incremental displacement in the analysis, eq. (7.15) is solved for y;.; and substituted 'nto eq< (7.14) to obtain: (H6)
and
2
ar
2y,
(7<17)
210
SUuctures Modeled as a
Sjng!e~Degree~ofF(eedom
System
211
Nnw subtracting
'ii
7.4
(7.13)
The substitution into eq. (7.7) of fly, and flY" respectively. from eqs, (7.17) and (7.18) gives
y'. Oy"
ill
2y'
Ij'
+ k fly =
f
LlF
,
(7.19)
Equation (7.19) is then solved for the incremer.cal displacement ~Yi to ob~ain
Lly,
Ir. the Unear accelera(ion method. it is assumed that the acceleration may be expressed by a linear function of rime during the time interval LlI. Let to' ar.d (,;. I = II' + ill be, respectively, the desigr.ation for the time a~ the beginning and at the end of the time interval ,cll. [0 this type of analysis. the material properties of the system C; and k; may include any form of nonlinearity. Thus it is not necessary for the spring force to be only a function of displacement or for the damping force (0 be specified on~y as a function of veiocity. The only restr1ction in the analysis is that we evaluate these coefficients at an instant of :ime t, and then assume [hat they rerr.ain constant during the ir.crement of time ill. When the acceleration is assllmec to be a linear funcllon of time for the interval of time Ii or f, {! = f; + .1r as depicted in Fig. 7.4, we may express the acceieratjon as
(7.20)
.. ri ) \'
"
=)'
,
.. + ~ Yi (t [)
I
,}
ilt
'
(7.25)
ki
is
4m _ k = r , Llr
+ Llr + k'
2c
(7.21)
where LlYi is given by eq. (7.10). integrating eq. (7.25) twice with respect to time berwee!1 the ;imits I, anc I yields
y(t) = Y,
and
+ Yilt
iiYi
,
(7.26)
LlF,
4m
(7.22)
y{t):;;;; Yi
+ y;(l
 ri) :
{r fi)
(7.27)
The displacementy;+) y(ti + Ll/) at time ti't ( I; +..1{ is obtained fron eq. (7.8) after solving for incremental displacement LlYi in eq. (7.20). The incrementnl velocity is calculated by eq. (7.17) and the velocity atI time Ii + i = t; + ill from eq. (7.9; as
(7.23)
!i
.vi+l
ti+!
[j+,dt, is
obtained directly from the differencial equation of motion, eq. 0.3), rather than using eq. (7.16). Hence, from eq. (7.3j:
(7.24)
Fig, 7.4
212
213
and
(7.29)
To obtain t,1e displacement y;...; = yeti + .de) at time is substituted into eq. (7.8) yielding
Yi.' = y,
T
ti", !
!i
where Lly, and .6Yi are defined in eqs. (7,8) and (7.9), respectively. Now to use the incremental disp!flCement .:1y as the basic variable in !he analysis, eq. (7.29) is solved for the incremental acceleration Lly, and then substi~uted inro eq. (7.28) to obmin
Llj,
Lly,
(7.37)
Then the incremental velocity LlYi is obtained from eq. (7.31) and the velocity at time [;+; = t; +.cit from eq. (7,9) as
Lly,;;y,3j,
61
(7.30)
. '.' . Yi+!=Y;.!J.Yi
(7.38)
and
(7.31)
Finally, the acceleration y,.,. 1 at the end of the lime step is obtained directly from (he differential equation of motion, eq. (72). where the equation is written for tlIT.e t,"' i ;;;;; Lit. Hence, after setting F; s: my,' + 1 in eq. (7.2), it follows [lim
'i .:
The substitution of eqs. (7.30) and (7.31) imo eq. (7.7) leads to the following form of the equation of motion: 6
L1r v 2 .. '
(739)
(7.32)
FinaHy, transferring in eq. (7,32) all the terms containing the unknown incremental displacement LlYi to the lefthand side gives
k,L1y, (7.33)
in which
ki
k,
='
(7.34)
It ShOUld be noted thar eq. is equivalent to the static incrementaIequilibrium equation, and may be solved for the in~remental displacement by simply d~viding th.e effective incremental force tJF, by the effective spring Constant k" that is,
Lly
=~'
fJF
k,
{7J6)
where that damping force F DCt,.;. I) and the spring force Fs(t l + I) are now evaluated := rj + Llt. After the displacement, velocity, and acceleration have been detennined at time t; .. 1:= t,  L1, the outlined procedure is repeated to calculate L1ese quantities at the following time step ti + '2 = t, ... I + .dt, and the process is continued to any desired finai value of time. The reader should, however, realize that this numerical procedure invotves two significant approximations: (l) the acceleration is 3>sumed to vary linearly during (he time increment ilt; and (2) the damping and stiffness properties of the system are evaluated at the initiation of each time increment and assumed to remain constant during the rime interval. In general, these two assumptions introduce errors that are small if the time step is short. However, these errors generally might tend to aCcumuiate from step to Step. This accumulation of errors should be avoided by imposing a total dynamic equilibrium condition at each step in the analysis. This is accomplisned by expressing the acceleration at each step using the differential equation of motion in \vhich the displacement and velocity as well as the stiffness and damping forces are evaluated at that time step. There still remains rhe problem of [he selection of rhe proper time increment ,dt. As in any numerical method. the accuracy of the stepbystep integra::ion method depends upon the magnitude of the time increr:1eot selected. The following factors should be considered in the selection of Llt: (1) the natural period of the structure; (2) the rate of variation of the loading function; and (3) the complexity of the stiffness and damping functions. In general, it has been found that sufficiently aCCurate resulrs can be obtained if the time interval is taken 10 be no longer than onetenth of the
at time !,,,.!
214
215
natura! period of the structure. The second consideration is that the interval should be s:nal! enough to represent properly the variation of [he load with respect to time. The third point that shouid be considered is any abrupt variation it: the rate of change of tbe stiffness or damping fUJ1CliofL For ex.ample. in the usual assumption of elastoplastic materials, the stiffr.ess sud
Now, the substitution of eqs, (7.42) and (7.43) into the incremental equatlon of motion, eq erl), results in an eqUaliOD to calculate the incremental displacement .dy;, namelY,
(7,44)
denly changes from linear elastic 10 a yielding plastic phase. In this case, tc obtain the best accuracy. it would be desirable to select smaller time steps in the neighborhood of such drastic changes.
where ~he effec:ive stiffness k" und the effective incremental force !JF; ure given respective;y by
_ m (3!Jr'
k,=kl+"
+1
7.5
2{3!JJ
(7,45)
Tlle Newmark Beta Method includes, in jtS formulation, several timestep methods used ~or the Solulion of iinear or nonlinear equations. It uses a numerical parameter designated as {1 The method, as originally proposed by Newmark (1959), contained in addition to /3, a second parameter y. Particular numerical values for these parameters leads (0 weBknown methods for the solution of the differential equation of motion. the constant acceleration method, and the linear acceleration methcd. The Newmark equation can be written in incrementa] quantities for a COnstant time step Llt. as
(7.40)
!JF,. = .dF, +
In
mIl \
4/3 J '
(7.46)
10 these equations, C; and k, are respec;:ively tbe damping and sriffness coefficients evaluated at the :nitia: time I, cf the time step Lit Ii":" I  I;. 1n the implemen;ation of t~e :'\ewm2.rk Beta Me6od, a numerical value for the parameter {3 is selected. Newmark suggested a vabe in the range 1/ 6.::f f3:S: 112. For f3 1 14. the method corresponds (0 the constan~ accelera~ion method and for f3 i 16 tc {he linear acceleration method.
and
(7.41)
7.6
ELASTOPLASTIC BEHAVIOR
in which the incremental displacement ny: and incremental velocity LlYi 3re defined, respectively, by eq (7,8) and (7,9), It has been found that for values of y different than in, the method introduced a superfluous damping in this system. For this reason, this parameter is generally set as '1= 1/2, The solution of eq, (7.41) for .1.9; and its subsequent substitution j nto eq. (7.40) after setting y =. ! 12 yield
(7.42)
and
(1.43)
If any stmcture modeled as a 5ingledegreeof~freedom system (sp:ingmass system) is aUowed to yield plustically, then the restoring force exerted is likely to be of the fonn shown in 7.5(a). There IS a porrion of the curve in which linear ei.astic behavior occurs, Whereupon, for any fu::ther defonnation, plastic yielding takes place. When the strucUJ.re is unloaded. the behavior is .again elastic untiJ further reverse loading produces compressive plasric yleldir.g. The stmcture may be subjected to cyclic loading and unloading in this manner. Energy is dissipated during each cycle by an amount thnt is proportional to the area under ~be curve (hysteresis ;oop) as indicated in Fig_ 7.5(a). This behavior is after'. simplified by assul'TJing a definire yield point beyond which additional displacemem takes place ar a constant value for the reswring force without any further increase in the load. Such behavior is known as elastoplasIic behavior: the corresponding forcedisplacement curve is shown in Fig.
7.S{b)
For the structure moceJec as a springmass system, expressions of the restoring force for a system with elastoplastic behavior are eaSily w:inen,
216
217
R/!!$,Or;P9
loree
R
y=O
in which
y~~x
Conversely. if y decreases to y~, the system begins a plastic behaVIor tn compression along curve C and it remains on this curve as long as y < O. The sysrem returns to an elastic behavior when the velocity again changes direction and y > O. In this case, the new yielding limits are given by
y,
Yrr,in
y, = Yxi'
+ (R,  RJlk
(7.50)
Fig, 7.5 Elasric~pl,l$(ic :structural models. {a) GenerJi plustic behavior. (b) E!asmpiastic behuvoiL
in which Ymin is the minimum displacement along curve C, which occurs when .y = 0. The Same condition given by eq. (7.48) is valid for the system [0 remain operaung along any elastic segment such as Eo, E" ];.".as shown 1n Hg. 7.5(b). We are now interesled in calcuiatlng the restoring force at each of the possible segments of the elasrop[astrc cycle. The restoring force on an elastic phase of the cycle (E" E,. E, .. ) may be calculated as
These expressions depend on the magnitude of the restoring force as well as upon whether the morion is such that the displacement is increasing l": > 0) or decreasing (y <0). Referring to Fig. 75(b) in which a general elastoplastic cyc!e is represented, we assume that the initial conditions are zero (yo ~ 0, Yo 0) for the unloaded structure. Hence, initially, as the load !s appJied, rhe system behaves elastically along curve The displacement y" at which plastic behavior in tension may be initiated, and the displacement Yn at which plastic behavior in compression may be initiated. are calculated, respectively, from
R=R,
on a plastic phase in tension as
(y,y)k
(7.51)
R=R,
and On the plastic compressive phase as
(752)
(7.S3)
y, = R,Ik
and
(7.47)
The algOrithm for the stepbystep linear acceleration method of a single degreeoffreedom system assuming an elasroplastic behavior is outlined in the fonowing section.
where Rr and Rc are the respective values of the forces that produce yielding in tension and compression and k is the elastic stiffness of the structure. The system will remain on curve Eo as long as the disp:acemem y satisfies
7.7
<y,
(7.48)
If the displacement y increases to y" the system begins to behave plastically in tension along curve T on Fig. 7.5(b); it remains on curve T as long as the velocity y > O. When y < 0, the system reverses to elastic behavior on a curve such as EI with new yielding points given by
(1)
(2) (3)
Input values for k, m, c, R" R,., and a table giving the time magnitude of [he excimtion Fl'.
Set Yo and Yo;= O. Calculate initial acceleration:
Ii
and
Yr =
)'ma1
F(r = 0)
(7.49)
(7.54)
218
219
(4)
(4)
= 31Llt, a, ~ 61Ll1,
YI
OJ
dli2, a,
6/d:'
(5)
(7.59)
Solve for the incrementa! displacement:
Lly;
(5)
Rilk
(7,55)
y, = R,lk
For each lime step:
(l)
(760)
(6)
Use the following code to establish (he elastic or plastic state of the
system: KEY
= 0 (elastic
behavior)
(7)
KEY =
(7.56)
+ Lly,
(7.62)
(7,63)
Calcui:lle Ihe displacement y and 'eJocity j at the end of the time step and set the value of KEY according to the following conditions;
(8)
Yi )= y, T i.lYi
{a)
When the system is behaving elastically at the beginning of the iime step and
Calculate acceie:ation Yi~ I at the end of time interval using the dynamic equmion of equilib:i.urr.: (7,64)
y,<y<y, KEY=O
KEY = I KEY =  [
(b) When (he system is behaving plastically in (cosion a( the begin, fling of the time step and at which
j>O KEY = 1
R=R,
or
if KEY = I
if KEY =  1
(7,65)
y<O KEY=O
(c) When the system is behaving plastknlly in compression at the beginning of the time step and
R=R"
y <0
(3)
KEY=j 0
;;>0 KEY
CalcUlate the effective stiffness:
Example 7.1 To illustrate the hand calculations in applying the stepbystep integration method described above, consider the single~degreeoffreedom system in Fig. 7.6 w:th elasroplastic behavior subjected to the loading histOry ;1S shown. For this example, we assume that the damping coefEcient remains constant (~= 0.087). Hence the only noniinearities in the system arise from the char,ges in stiffness as yielding occurs.
where
Solution:
kp k for elastic behavior (KEY
=
0)
= 1 Or
 1)
k=~~=~.
12E(2!)
Ll
12 x 30 X 10'
x 2 X 100
(7.58)
(15 X 12)'
12,35 Kiplin
220
221
20'
or
k = k, +
where
ibl
128,22
k" = k
kp
;.:!;;!
0 (plasric behavior)
f 6
L.V
l<l
lit, =
(0
dF,
+ 12.822Y; + 0,6137y,
Fig. 7.6 Frame with elaStop!l!stic behavoir subjected (b) Loading. (c) Elastoplastic behavoir.
damping coefficient
sec lin
0, Initinl acceleration is
yo=
Yield. displacements are
F(O) =0
The necessary cakulz.tions may be conveniently arranged as illusrrared in Table 7.1. In this example with elasroplastic behavior, the response changes abruptly as the yielding StartS and stops. To obtain better accuracy, it would be desirable to subdivide the time step in the ceighborhood of the change of state; however, an iterative procedure would be required to establish the length of the subintervals. This refinement has not been used in the present analysis or in the computer program describec in [he next section. The stiffness computee at the initiation of the time step has been assumed to remain constant during the entire time increment. The reader is again cautIoned that a significant error may arise during phase transitions uniass [he time step is selected
relatively small,
R, v =~ I k
7.8
and
y~. =
1.215 in
The natural period is T = 21T r;;ik = 0,8 sec (for the elastic system). For
The same as for ~he oilier programs presented tn this book, Program 5 initially requests information about the data flle: r.ew, modify, or use existent data file. After the user has selecIed one of these options. the program requests the name of the file and the necessary input dam. The program continues by reading and
\1")0;
dC'4N"":o6
a:N~g~g~$~?1j~;g~~~~
I I
N
OI'X>OONa
.........
00 <r)r'1QC'l 0 0 l.fl0\ 0\
00 NVi
223
'NMMN:N~0N
~v
>'2
r r 0> r0
""
'" "
N
~("o,)")("")
""
ooddoood
I
roooo
'"
..:
~
~
c. E
"l
k. c:.
:;;?
r r r r.,. '" rr r r r rV '" .,. " "" " .,. " 0 0 0 0 "': " ci ~ '" ~ '" '" '" '" '"
~
r
tCO 0
OJ
"
'" '"
""
'" '"
'"
.2
::E c.
printing the dara and by setting the mitial values to the varioLls constants and variables in the equarions. Then by linear interpolation, values of the forcing function are computed at time increments equal to the selected time step Lll for the integration process, In the main body of the program, the displacement. velocity, and acceleratIon are compLted <:::t each time step. The nor.lir:ear behavior of the restoring force is nppropr.meJy cons:Gered in the calculation by the variable KEY which is tesred through a series of condllional statements in order to determine the correct expressions for the yield poinrs nod the magnitude of the restor:ng force in the system. The output consists of II table givlng the displacement, ve]ocity, aod accel~ eration at time increments Lit, The last caluma of the tD.ble shows the value of the tndex KEY which provIdes Information about the stale of rhe elastop. lastic system. As indicated before, KEY = 0 for elastic behavior and KEY = 1 or KEY = : for plastic behavior, respectiveiy, in [eosior: or in compression" Example 7.2 Using the computer Program 5, find the response of the structure io Example 7, J" Then repea( the calculation assuming elastic behtvoir. P:o~ and compare results for the elaswpiasric behnvoir with the elas[ic response. Solution: Problem Data (from Example 7.1).' Spring constant: k = 12.35 Kiplin Damping coefficient: c = 0,274 KIp' sec lin Mass: m = 0.2 (Kip sec' /in)
"
" iii
iii
"0 0
,~
.0
>.
6. <1)
N
.;
u
;; il
c ,,,
:>..~ ,S
~"
e N <': .,.., 0 od N
0
,..,
'"
0 oc 0 ,... "" N = '" '" r'" '" '"  '" oZ '" 0 ~ "" N r"" I '" '" '" ' " i I " I '" I
<n
&:. r
.,..,
.,..,
00
0 0>
~,
'0
~~ggg88
~1T'l>no.riV)"';""';
<::O~~
... "
'" ::J
c:
:.
J,
c:
~
~~~~::::=:~~ti6~~~~~
1 I
0
I,I')vOOOOQ'\...<O:TOOV;>.QrO:
Max. restoring force (tension): Rt = 15 Kip Max. restorir.g force (compresslon): R" = ~ 15 Kip
oci~""';.Dr..:r..:.nNciir..:<Atrl~oC
O\)MOOI,I')OOO\OOOOOOMN
a.
~
<1)
Selecl time step: Lit = 0, I Sec Gravitational index: G = 0 (force or. the nass) The same computer program is used [0 obtain the response for a completely elastic be!iavioL J[ is only necessary to assign;) large value to the maximum restoring forces in tension ~nd in compression, These assigned values should be large e:1011gl: in order for tl1e structure to remain in the elastic range. For the present example. R/ 100 Kip 3:lG R," = ~ 100 Kip were deemed adequ&te in this case. In order to visualize and faciljtate a comparison be{ween the elastic and inelasric responses for this example, the displacements are plotted in Fig, 7.7 w1[h [he response during yielding shown as dashed lines.
" c 0
z
'" "
~
'"
;;; "" co '" '" "" 0: "i '" "i '" ci 0 "" <:5 ;:;;
N
M
'Voo{"0t010000
~oonr\Q>nO{"0Vir ~<::O)t~{"0{"0NO
,.., '" ;:;; " "" '" N '" '" f:! "" ..., "" '" "" "': r'" '"
0N
00
N
N
""
..:
OJ ..J Ql
oi<A~r..:cO.n~o\.,oMOO'nOO

I
... ....
.::t:
V',
222
224
Structures Modeled as a
Single~OegTeeofFreedom
System
Displacement !In.!
225
4.D~
tiS
AX
AK
OA.:~?I":G
:'::~::;:
0 0
,
35
~";~
CCS?F!cr::"
0
H
::;.'!,
S':'!i:?
o.r
nr::'SCftN!:'!CN
?;)!l.CS IS TS:iS:O"
?8?.cZ :::m??:::ss~ml
. '",
"
0
orr.;\
:11';
RSS~"=;''G
RSS?C~!~:G
"
ZG, CO
0
:,
1':;;
~3
,t
t 20D
30~
~;//.
1/
E1utic (~cnse
S;;'C:C:;:,:C;i
,
::YlS
0;;
.0~
, ,
'!'!~'S
2X::::T.;T:(',d
4SJ
T::',>;
Z:KCt'!'.::;,:ox
1.
:')\'
no
.>::0'1
(N
,
Yielding/""
CU':?'..:'..' ,,;;:5(,.':"75.
1.CGD
::J~s"'t.
"/S:':>:.
;..ce.
C.OCO'i :a. 9714 25 .7n.\
300C
::.DOI;O
?: ?J
:n:6
0.2325
c.i4$S
,.
~), 2'~C
:>. ~83',
::"::3:' ;")49
0: 7 3
+
,
c
JOel
6,,;;:8
',2679 9154
65CC
:3 lH:
Fig. 7.7 Comparison of elastoplastic behavoir with elasTic response for Example 7.2.
o .~o::
SO()
s. dSJ
C. CDC C.7C0
0:15
S"7/S
.,
~9';;()
630'<
no'!
.72Y1
on. ;'(l~:
31 . ?S5e
.{1 )1$90
soc so:;
oe::;
l.SS?;> 0.,,1;64
"as
lCO
,.
$:;10
~Ja,
.,
) GRID
G. 1
PLk~B
BC52
_:4 936:
n258 1122
120Z
on]
oS ]77';
?:"?..ol\lE,
Z,
:leo
400 l. ,c:;
50C
0'
..
;
~
9.4756 5051
B:.2?
56.7'~S
PH
6~ZS
:!ss:.
n57
. :C2S
). C0RVE$
:>
capCORD
5:1,12$5
6911
\,. 2lsa
7:;0
st:i
59/1
~;56
i.3aM
s~. 5~6~
~7?_
CRPCORD.
1. C, 0, 0, 1, G, 0
.'1471
il"3e
~;;J.S
,sec: ooc
$753
Ol';' .,
,
liiO 3
59 441{
(4) Define element group 1 using the TRL'SS2D element formulation with etastoplastic behavior
PROPS 2'rS ). EGROH? GROUP, 1, 'fRUSS2C, 0, 0, 0, C, 1, 0, 0
't ga97
s . .ns~
7.9
Example 7.3
Solution: The analysis is perfonned using a single truss element with one cO:1cemrated mass element. The following commands are implemented:
(I) Set view
VI~\\!,
(6) Define elastic modulus for truss elernem base OIl 1 in length and area I in' (k = AEIL = 12,346 Iblin). Also define yield 'tress as 15.000 Ibl mZ and tangem modulus as tE~7 Ib/tn 2:
?ROPSETS ) MPROP
to
the XY plane;
:>
DISPL.?,';(
V:::E.:'i1 J?AR )
\rIE'd
0,
Dt
1,
226
227
= 0 and
of 0.05 sec:
EGROUP
K~SS,
2,
0, 0, O. O. 0, 0, 0
(9) Define real constant for mass element: m ;:;;; 200 !b sec:? lin:
PROP$ETS
ReOl'1ST,
2.
~
I
RCONST
2,
1,
7, 200, 0,
0, 0,
0,
0I
(7) Set the options for [he nonlinear dynamic analysis using default values for toierance parameters and run analysls:
t>.NALYSIS ~ NOKL::NE:A::\ > A~K0NLINS:l_)=<. A NONL1N2AR, :J, l , 1, 20, 0.801, 0,
N,
Y,
0,
PARA."CMSSH .. 11 ?T
0.001, 0.01. 0
,;';"lALYS!S > NOl'lLINEAR
R_NOtt,;L'::~SAR
2, 2,. 1
NM2RGE 0.000;", 0,
(18) Activate XY plot info,wation for X displacement at node 2 as a function of time, and plot the displacement vs. time for node 2:
0, 0
C:S?LAY ~ XY_?LOTS ~ ACTXYPOST ACTXYPOST, 1, TH!E, UX. 2, 12 .\., DISPLAY XY~?=O'T'S:> XYPLO'!' XYPLO?, 1
v,
(12) Apply constraints in all degrees of freedom at node 1, and all degrees of freedom except UX at node 2:
LOADSBe ) $'l'RUC'TURAL .. DISPLMN'T'S
;>
2N
CPT
HZ
(19) Activate XY plot information for X velocity at node 2 as a functio;i of lirr:e, and plot the veiociry vs. time for ;iode 2:
DISPwAY > XY P~OTS } lCTXiPOST ACTXYPOST, 1, TIME, VX, 2, 12, .\., DIS!?LAY > XY ... PLor{~S } xY2LOT XY?LCYr, 1
([3) De:'ine cyr.amic fOfcIng function and apply as force to node 2 in the
X
d~rection:
LO_!;'DS~8C
0,
2N
> FUNe_cuRvES
>
CUBDEF
CuRC2F,TIME.
1,
1,
0,
>
0,
.45,
)
20000,
FPT
1.1,
0,
L2,
:'0000, 2.0,0
LQ.lWS Be .)
FPT, 2,
STRCC':'URAL FX, 1, 2, 1
~O:.qCES
(20) Activare XY plo~ informatio:l for X acceleration at node 2 as a function of time, and plOl the acceleration vs. ~~me for :lode 2:
DISPL.!;'Y ) XV_PLOTS } ACTXYf'O$T ,l:!,C'rXYPOS'L L 'rIME, AX, 2, 12, 1, DISP.!:.AY XY ...,?LCTS > XYP:"'OI' x YPL01' , 1
(14) Define Rayleigb damping coefficients [C = ak + pm]. C" = 2(km),n = 3143 10' sec lin; C 0.087 (C,,) = 273.4 lb sec lin: Using a= 0.01, fJ [273.4(0.01)(12346)11200 0.75
ANALiSISNONSINEA:R NLROf..Hi? NL_RDAMP, 1, 0.01, 0.75
0,
2N
[elmS
of (:1)
djspJaceJ:len~,
228
229
i
[
!
I ,
i
J.5
rI
x
2
1.5
I
I
I
lit
!
i ,
I
[
, ,
i 't
I
I '\ I I \!
~
I
:r
....r ,
, / I
II
I
I
!
I
I
'.5
,,
, .4
I I~' , .8
!
/ f~
=~
,
!
1.2
1.4
I/bl
II '~ ___ ~ '\
I
I 64 I
I
I
, ,
, , ,
I ,
48
12
A X
!
/
16
,
1
i , ~ I , ~ i
i I
, ,
I
I
I
! i
,7
!
r
? '\
\
\
I ,
2N
! 5
~r
i
,
31
~"K I ~
.
I
I
i
I~
I
II
1\
i ,
I .2
qs
i~t
,
I
~
:?
5,
go
I .8
EJ
I I
[
.4
!;!
I
.5
i
!
I
I
I ,
I.,
I ,
!
I .6
\ II ,
1.8
II I , ,
11,
1.8
OJ.2
T!HE
,,'
j
iol
I
/
I
,
I
Fig. 7 ..Be
I
,
L
./
/
i
" '\
i\
I ! \ I \ I/ \
I '"I
4
I \ I \
I
,
8
co
,
I
"...
, .2
IT
I
II
Y
I
!
l
\
I
I
i
l.6
I
I
1.8
, .4
'.6
, .8
1.\
T!HE
Ib)
F;g.7.8
Response for Example 7.3. (a) Displacement; (b) Velocity; (c) Accelenuion.
motion of an eanhquake or the effects of nuclear explosion. In these cases, it is not realistic [0 assume that the structure will remain linearly elastic and it is then necessary [0 design the Structure to withstand deformation beyond the elastic limiL The simplest and most accepted assumption for the design beyond the elastic limit is [0 assume an elaswpi<lsric behavior. In this type of behavior, the structure is elastic until the restoring force reaches a maximum value (tension or compression) at which it remnins consmnt umil [he motion reverses its direction and returns to an elastic behavior. There are many methods [Q solve numerically the differemial equation of (his type of mOlion. The stepbystep linear acceleration presented in this chapter provides satisfactory results with relatively simple calculations. However, these calculations are tedious and lime consuming when performed by hand. The use of a computer and [he availability of a computer program, such as [he one described in this chapter, reduce the effor[ to a simple routine of data preparation.
7.10
SUMMARY PROBLEMS
7.1 The singledegreeoffreedom of Fig. P7.1 (a) is subjected to the foundmiol1 acceleration history in Fig. P7.I(b) Delermine [he maximum relative displacement of the columns. Assume eiastoplas[ic behavior of Fig. P7.I(c).
Structures are usually designed on [he assumption that the structure is linearly elastic and [hat it remains linearly elastic when subjected [0 any expected dynamic excitation. However, [here are siwations in which the structure has [Q be designed for an evemual excitation of large magnitude such as the Slrong
230
231
m"'O,5~W
.,I()
7.4
Repeat Problem 7.2 for the acceleration history show:. in Fig. P7.4 applied horizontally to the fO:Jndation.
50
Ibl
50
Fig. P7,4
7.5
7.6
Ie)
7.7
Fig. P7.1
7.2
Solve Prcb!em 7.1 as;;.uCiing elastic behavior of t~e strJctu:e. (Hint Use :::omputer Program 5 with R,o= 200 Kip and l?", = _. 200 Kip.) Solve Problem 72 :or elastic behavior of the struL'ture. Plot the rimecisplacement rcsponsc and compare with results :rom Problem 7.2. Determine the ductility rado from the results of Probler:l 7.2 {DuctHity ralio is defined as the ;ario cof the maximum displacement to the displacement at {he yield point}. A stru~ture mOdeled as springmass shown in Fig. P7.8(b) is subjected to t:'e loading fcree depicted in Fig. P7.8(a). Assume elastoplastic behavior of Fii!. ~. P7.8(c). Determine the response.
Determine the displacement history for the structure In Fig. P7.1 when it is SUbjected to the impulse loading of Fig. P7.2 applied hOIlzomaHy at the mass"
Fit}
7.8
o
7.3
0.1
0.2
Fig. P7.2 Repeat Problem 7.2 for the impulse loading s~own in Fig. P7.3 applied hClr1Z0ntally at the mass.
,cUI
200K~
o
r"
w . 38$1<:
1l77
~F({)
~l
__ 
~
L21=
:c)
~".;',
Fig. P7.3
Fig. P7.8
232
7_9
1.10
10
damping. So~ve Problem 7,& assuming elastic be\)uviOl of Ihe $)'$ll~m. (Hint: Use Program 5 with ff,= 1000 Kip and Rc~ , 1000 Kip.)
8
Response Spectra
1.11 7.12
Solve Problem ?,9 assuming c!Jsiic behavior of lhe sys:em. Fig, P?I2(a) is A suucwle modeled as [hI:': damped splingm:ls.s 5yslem shown subjecteu to the !im~acctleriH;on ex.citation (luing at ils suppOrt. rne: excitation (unCI!On is expreised:)5 a{l}::::: (Ju!(J). Whl!feJ(f) is depicted in Fig. P7.12(b). Determine the maxlmUln v,llne (hUI lhe f:lclOr (In nJly have for the S(fllCIlIH! 10 remain e!J:.;{ic. AS~ume !h;)( the SlruclUre has.: nil elns[op!a$!ic behavior of FIg. PI. !1(c}.
il,
A
fir'" ;!)OJ _ ,_ _,
O.~
0.3
(6
101
id
fig. P?t2
In Ihis Chapter, we inirodllce (he concept of response SpeClrl.tffl, v{hich in recenr years has gained wtde: nccepwnce in SlfUCI'Jfl11 dynamic ptuctlce, particularly in eanhqu"Ke englliteril1, design. Slated brieny, the response spec:nlm is a plot of lhe Glilximur:! respof'.sc (maximum displacement, veloci,,!, acceleration, or any olher qu<ulli!y of in:erest) to Zl specified toad fU!lcLion for all possible singledegreeoff!eedom syslems, The abscissa of the SpeC{fUm is the natural frequency (or reriod) of lhe system, and Ihe ordin;][e (he maximum response. A plot of ihls lype is shown in Fig. 8.1, in which a onestocy building is subjected (0 a ground displacemen( ;nuicated by the function Y,(f). The response speclraJ curve shown in Fig. 8.!(a) gives, fOf any single"degreeof~
or
[he mass
til
relative to the
displacement at the support, TI!\JS, to dClcfmine Ihe response from an nvailab!e speclra! chan, fOf (. specified excitJlion, We need only iO know the :\<lIUf;)! frequency of the Sy$ltliL
8,1
To
;I!u$u~[e (he construction of J rCspof\:;e SpecI(;.}1 cllan, consider in Fig. 8.2(a) the unda~Ylped osci!lam( subjcC:;:d :0 onehalf period of lhe sinusoidal
2)3
8
Response Spectra
In this chapler, we
j~t1Dduce Ilk. t:lJllI.':CPl or re.f!1oWiC .'ipeClmm, which in recent yents has :lintu wide ,H..'CCP\;lllC(: iI) ~((Ut:lllf;1l uyn:\m!c prl.lcljc~. p::lniculady iii eanhqui1kc tngi:1Ccfillg th!:i!t;11 St;l\ed hriefly. lilt: response spectrum 1$ il plot or tht.O l11ilXll1HlIn 1!:;;ij)0!lSC (1l"IXllillilli ci:;placcH\clH, vdacilY, pccdef<J;lon, or aoy other qUi:iuity of jllu.:rcsl) !O a speciCI(:J !O~HJ [ullction [or ,111 possible s.ifiglc~cle.grecofr(ce(;O!ll sy::il!:ms. !"!l<! i\U\CISS:I ::'If the ::;:pc.cl(Um is (he na[urn! fJ(;.qutncy (or period) of the SY:>icll;, :;;)(~ tIll! oH.!;n;He tht 111aximum response, A plot of this type is showll In (:ig. 8.!, j;, wll!cli ;t enestory building is subjected (0 a grcunu Ji:;p:acCIl1l.::di i:I(:ic,llCJ oy the [uHc\lon Y,(t)_ The n> spcn~e spectral curve sliewn ill i:,g. l:L i(:\) gi"'I~S, [or ;:,\ly singkdegrceo[freedom system, (he flHlXinHlnl 0J!>p!:'ce.:nerH cr the:. muss III relaflve to lhe displi1cemen( O[ ill:! suppert Tho::., (8 dc(cnni,ie (hi,! ;e.spomc (;om ao available spectra; chan, [or a srecjfi~d l:xcit:too:\, wt: need only to kllow the oalUral fre.qucilcy of lhc s'yslem
8.1
To il!u:ar;Hc lh< cO:1s(n.:ctioJ: of it (C:'PCIl$C SjltCHJI ch:irt, consider in pig. 8.2(a) dIe Und~!nlpcJ 8sci!LLIO( 5uuj';Cld 10 cn(.:ll:!i[ pcriDd o{ (he. :;inusoid;)i
234
Response Spectra
235
I
_ _ _ _ _ _ _ _ _ _=~_!NatlJr~! frequent;:y. f
'
~...
{b)
/////,r),ddddN&&dL/d
,(tl
The Solulion of eg. (8.1) may be found by any of the methods studied in the preceding chapters such as the use of Duhamel's integral (Chapter 4) or the stepbystep linear acceleration method (Chapter 7). However. in this example, owing to the simplicity of the exciting force, we can obtain the solution of eq. (8.1) by the direct method of integration of a linear differential equation, that is, the superposition of the complementary solution Yc and the particular solution yp
Y=Yc+Yr
(8.41
Fig. 8.1
[0
The complementary solution of eq. (8.1) (righthand side equals zero) is given byeq. 0.17) as
ground excitation.
UJt
(85)
exciting force shown in Fig. 82(b). The system is assumed to be inrtially at rest. The duration of the sinusoidal impulse is denoted by la. The differential equation of motion is obtained by equating to zero the sum of the forces in the corresponding free body diagram shown in Fig. 8,2(c), that is,
my+ky=F(t)
(8.1 )
in which w=jklm' is the natural frequency. The particular solution fo( the time interval O:i l:i I, is suggested by the righthand side of eg. (8.1) to be of the form
)'p=Csin
wt
(8.6)
The substitution of eq. (8.6) into eq. (8.1) and solution of the resulting identity gives
C=_~o~
in which
r Fo sin F(t) = 1 ,0
and
wt
(8.2)
kmni
(8.7)
(8.8)
Introducing the initial conditions y (0) = 0 and y (0) = into eq. (8.8) and calculating the constants of integration A and B. we obtain
Fit)
k
m
~F(I)
(e)
+~~
(b)
(8.9)
w=
2rr
t/
I )
Fig.8.2
F(t)
(a) Undamped simple osclllatcr subjected to load F(r). (b) Loading function Fe sin (i)(OSo('5;ld). (c) Free body diagram.
','" ':,."
236
Response Spectra
237
After a time lti, the external force becomes zero and the system is then in free vibration. Therefore, the response for {> (d is of the fo~ given by eG:. (8,5) wIth the constants of integration determined from known values of displacement and velocity calculated from eg. (8.lOa) at time 1 ld_ The final expression for the response is then given by
)'
)'.s:
I
I'
2 7T
{i (
\1'
~ 
1.1
.for
{:.?:: t.1
(S.IOb)
It may be seen from eq, (S.lO) that the response in terms of yly" is a function of the ratio of the pulse duration 10 the natural perIod of the system (talT) and of time expressed as liT. Hence for any fixed value of the parameter t,Jr, we can obtain the maximum response from eq. (8.1 0). The plot in Fig. 8.3 of these maximum values as a function of I"IT is the response spectrum for the half~siflusoidal force duration considered in thjs case. It can be seen from the response s;>ectrum in Fig. 8.3 that lhe maximum vaiue of the response (amplification factor) }'IY~I = 1.76 occurs for this particular pulse when (uIT~ 0.8. Owing to the simplicity of the ip.put force. it was possible in this case to obtain a closed soiution and to plot the response spectrum in teons of dimen~ ;:;ion!ess ratios, thus making this plot valid for any impulsive force described by onehalf of the sine cycle. However, in general, for an arbitrary input load, we cannot expect to obtain such a general plot of the response spectrum and we normally have to be satisfied with the response spectrum plotted for a completely specified input excitation,
2.

8.2
An importanl problem in structural dynamics is the analysis of a syStem subjected' to excitation applied to (he base or support of the structure. An example of such input excitation of the base acting on a d,lInped oscillator which serves to model certain structures is shown in Fig, 8.4, The excitation jn this case is given as an acceleration function which is represented in Fig. 8.S. Tile equation of motion lhat IS obtai!1ed by equaling to zero the sum of the forces in ~he corresponding free, body diagram in Fig. 8.4(b) is
m), + cU
(S.l! )
~"
1' 1. 5
ki
i
2.,.;p
11'
r",
I
Of)
(tJ;;:
r:.:
:/
0'
~
Y"
Y,ll)
o /
o
F(tJ
'!
~F(r)
15
FoIl<.
F,~
"
0. 0
1.0
I ,
~V M/\~
i
V~'
4.0
I
Fig. 8.5
Acce;e,a~ion
Fig. 8,3
ha!f~sinusoidal
238
Response Spectra
239
Equation (8J2) is the differential equation of motion for the damped oscillator in terms of its absolute motion. A Hlort useful formulation of this o[ob1em is to express eg. (8.12) in terms of the relative motion of the mass w'ith respect to the motion of the support, that is, in terms of the spring defonnation. The relative displacement u js defined as
my+ku=O
(8.17)
u= y Ys
Substitution into eg. (8. (2) yields
it
(8.13)
We observe from eq. (8.17) tbat the absolute acceleration is at aU times proportional to the relative displacement. In particular, at maximum values, the spectral acceleration is proportional to the spectral displacement, that is, from eq. (8.17)
(8.18)
+ 2gwu + uiu =
Ys (I)
(8.14)
The formulation of the equation of motion in eq, (8, 14) as a fur.ction of the relative motion between the mass and the support is particularly important since in design It is the deformation or stress in the "sprIng element" that is required. Besides, the input motion at the base is usually specified by means of an acceieration function (e.g . earthquake accelerograph record); thus eq. (8.14) containing in the righthand side the acceleration of the excitation is a more convenient fonn than eq, (S.12) which in the righthand side has the support displacement and the velocity, The solution of the differential equation, eq. (8.14). may be obtained by any of the methods presented in previous chapters for the solution of onedegreeoffreedom systems. In particular, the solution is readily expressed using Duhamel's integral as (8.15)
where UJ = ik)m is the naturaJ frequency of the system, So. }'mau and SD =: Um,n' When damping is considered jn the system. it may be rationalized that the maximum relative displacement occurs when the relative velocity is zero eu = 0). Hence we again obtain eq. (S.18) relating spectral acceieration and spectra! dispracement. However, for a damped system, the spectral acceleration Sa, is not exact[y equal to the maximum acceleration, although in general, it provides a good approximation. Equation (8.18) is by mere coincidence ~he same as the relationship between acceleration and displacement for a simple harmonic motion, The fictitious velocity associated with the apparent harmonk motion is the pseudovelocity and, for convenience, its maximum value S~ is defined as the spectral velocity, that is
(8.19)
mY+Cli+ku=O
(8.16)
Dynamic response spectra for sjngledegreeoffreedom e]astjc systems have been computed for a number of input motions. A typical example of displacement response spectrum for a sing!edegreeof~freedom system subjected to support motion is shown in Fig, S,6. This plot is the response for the input motion given by the recorded ground acceleration of the 1940 E1 Centro earthquake. The acceleration record of this earthquake has been used extensively in earthquake engineering investigations. A plot of the acceleration record for this eartbquake is shown in Fig. 8.7. Until the time of the San Fernando, California earthquake of 1971, the EI Centro record was One of the few records available for tong and strong earthquake mot10ns. In Fig. 8.8, the same type of data that were used to obtain the displacement response spectrum in Fig. 8,6 are plotted in tenus of the spectral velocity, for several values of the damping coefficient, with the difference that the abscissa as well as the ordinate are in these cases plotted on a logarithmic scale. In this type of plot, because of eqs. (8.18) and (8.1 9), it is pOSSible to draw diagonal scales for the displacement sloping 1350 with the abscissa, and for the acceleration 45"" so that we can read from a Single plot values: of spectral acceleration, spectral velocity, and spectral displacements.
240
Response Spectra
241
] LI
0
/"
\~\,
AI
U
I
I
i
I
i
~'
''yfv.,
i i
i
i
.~
2. 0
1.0
I
,
! '~I i '
~
~
E
~
<
\
!
I
V
,
1 .S
5
2O
tal
I
>
'" ~
~
o. 1
0.05
!
1
0,02
'u~
O.OS 0.1
0.2
Natural frequency, I:pS
\8 \J]
(8.20)
Fig.8.8 Response spectra for elastic system for :he t940 EI Centro earthquake {f:oom Blume et al. 1961.)
Fig. 8.6 Displacement response spectrum for elastk system, subjected to the ground motion of 1940 El Cemro earthquake, (from Design of Multisfory Reinforced Building for Earthquake MOlions by J. A. Blum, N. M. Newmark, and L R Corning, Portland
Cement Association 1961.)
For constant values of So, eq, (8.20) is the equation of a straight !jne of iog versus log/with a slope of 45", Analogously, from eq. (8,19)
S~
2r.f
To demonstrate the construction of a tripartite diagram such as the one of Fig, 8.8, we write eq. (8.19) in terms of the natural frequency / jn cycles per second (cps) and take the logarithm of the tenus, so that
Jog S"
 log! ~ Jog27r
So
(8.21)
S,
log S,
Y/9
For a constant value of SI>' eq. (S.2i) is the equation of a straight Hne of log 5<1 versus log f with a slope of 135''.
8A
" i
(
_:~h~I\~httl~tJ.i~i~/v.,~I/Wii~~f,\JlrM~'N*/~"~
~q
o
11
"I,
10
'~'~!'~'~~"';;:'~'~'.LJ.....i.~~'~'Tj(m:.~i:
IS M 25
Fig. 8.7 Ground acceleration record for El Centro, California earthquake of May 1g, 1940 northsouth compOnent.
In general, response spectra are prepared by calculating the response to a specified excitation of singledegreeoffreedom systems with various amounts of damping. Numerical inlegration with short lime intervals are applied to caJcuiate the re~ sponse of the system. The step~by~step process is continued until the total earthquake record has been completed. Tne largest vatue of the fcnclion of jnteresl is recorded and becomes the response of the system to that excitation. Changing the parameters of the system to change the natural frequency, we repeat the: process and record a new r:laxirnum response. This process is repeated until all frequencies of interest have been covered and the results ploued. Since nO two earthquakes are alike, this process must be repeated for all earthquakes of interest
242
Response Spectra
2.<t3
Until the San Fernando, California earthquake of 1971, rJlerc were few recorded strong earthquake motioT.;s because there were few accelerometers emplaced to measure them; the El Centro, California earthquake of 1940 was the most severe earthquake recorded and was used as the basis for much analytical work. Recently, however, other strong earthquakes have been recorded. Maximum values of ground acceleration of about 032 g for the El Centro earthquake to values of more than 0.5 g for other earthquakes have been recorded. It can be expected that even larger vaJues will be recorded as more instruments are placed doser to the epicenters of earthquakes, Earthquakes consist of a series of essentially random ground :notions, esually the northsouth, eastwest. and vertical components of the ground acceleration are measured. Currently, no accurate method is available to pre~ diet the particular motion that a site can be expected to experience in future earthquakes. Thus it is reasonable to use a design response spectrum which incorporates the spectra for several earthquakes and which represents a kind of "average'; response spectrum for design. Such il design response spectrum is shown in Fig. 8.9 normalized for a maximum ground acceleration of 1.0 g.
freque!'cy, cps
Fig. 8.10 Elastic design spectrum normalized to peak ground acc~leration of LO g fer 5% damping. (Kewrnark and Han 1973,)
Fig. 8.9 Basle design spectra normalized to 1.0 g. (From Newmark and Hal! 1973.)
This figure shows the design maximum ground motion and a series of response spectral plots corresponding to various values of the damping ratio in the system. Details for the construction of the basic spectrum for design purposes are given by Newmark and Hall (973), who have shown that smooth response spectra of idealIzed ground motion may be obtained by amplifying the ground motion by factors depending on the damping in the system. 1n general, for any given site, estimates might be made of the maximum ground acceleration. maximum ground velocity, and :naxjmum ground displacement. The lines representing these maximum values are drawn on a tripartite logarithmic paper of which Fig. 8.10 is an example. The lines in this figure are shown for a maxJmum ground acceleration of 1.0 g. a velocity of 48 in/sec. and a dlsplace ment of 36 in. These values correspond to motions that are more intense than
N
244
Response Spectra
245
TABLE 8.1
~

Percent Damping
Displacement
o
0.5
2.5
2.2 2.0 1.8 L4 1.2 Ll

AcceleratJon
6.4
5.8 5.2
I
2
of 10 strong motion earthquakes. The tab:e gives recommended amplification factors as well as the corresponding standard deviation val'Jes obtained in the study. Tbe relatively large values shown in Table 8.2 for the standard deviation of the amplification factors ;::rov:des f'Jither evidence on the uncertainties surrounding earthquake prediction and analysis. The response spectra for designs presented in Fig. 8.9 bas been constructed using the amplification factors shown in Table 8,1, Example 8.1 Example 8.1. A Slructure rr:ode~ed as a singledeg;:eeof~ freedom system has a natural period, T = 1 sec Use l~e response spectral method to determine the maximum absolute acceleration, the maximum relative displacement, and the maximum relative pseudoveJocity for: (3) iJ. foundation tT'.Otion equal to the EJ Centro earthquake of 1940, nne (b) the design earthquake with a maximum ground acceleration equal to 0.32 g. Assume 10% of the critical damping,
43
2.6 1.9
5
7
10 20
LO
1.5 L3 Ll
1.5
those gene~al!y expected in seismic design, They are, however, of proportional magnitudes which are generally correct for most practical applicarions. These
maximum values norrnnlized for a ground acceleration of LO g are simply scaled down fOf olher than 1.0 g acceleration of the grouild, Recommended amp! ificalion factors to obtain the response spectra from maximum values of the ground motion are given in Table 8. L For each value of the damping coefficient, Lhe amplified displacement lines are drawn a1 the left, the ampli~ fled velocities at the top, nnd the amplified acceleration at lhe right of the chart. At a frequency of approximately 6 cps (Fig. 8.9), the amplified accel~ eration regron line intersects a line sloping down toward the maximum ground acceleration value at a frequency of abot:.l 30 cps for a system with 2% damping. The lines corresponding to other values of damping are drawn para;lel to the 2% danlping line as shown in Fig. 8.9. The amplification factors in Table 8.1 were developed 00 the basis of earthquake records available at the time. As new records of more recent earthquakes become available, these amplification factors have been reca[~ culated, Table 8.2 shows the results of a statistical study based on a selection
TABLE 8.2
 "..
Solution: (a) Fro:r. the response spectra 1n Fig. 8.8 with 1 IT == 1.0 cps, corresponding to the Cilrve labelec ~ = O. J 0, we read on the three scales the following values:
SD
Su
3.3 in
18.5 in Isec
(b) From the basic design spectra in Fig. 8.9 with frequency f= 1 cps and 10% critical damping. we obtain after correcting for 0.32 g maximum ground acceleration in lhe following resu;is: S" = 9.5 X 0.32 So = 60 X 0.32 So 3.04 in
= 19.2 in Isec
.. "~""
.. "
..
~~..~~.
"~
..
~."~
..
Displacement
~
Acceleration
~~
.. ""~ .. "
..
~~
Standard Deviatjon
Factor
Standard Deviation
8.5
5 10
0.828 0.630
..
2.032 1.552
1.201
3.075
2.281 1.784
0.738
0.502 0.321
 ..
0481
~ Newmark, N. M., and Riddell, R., Inelastic Spectra for Seismic Design: Seventh World
""""
Before the San Fernando earthquake of 1971, earthquake accelerograms were llrnited in number, and the majonty had been recorded on BIluvium, Therefore, it is only natural that the design spectra based on those data, such as those suggested by Housner (1959) and Newmark~Hali (1973), mainly represent alluvial sites. Since 1973. the wenith of information obtained from earthquakes worldwide and from scbseqt.:ent studies have shown the very significant effect that the Iocnl site conditions have on spec~ral shapes.
246
Response Spectra
247
10TAL
104
, 1"" I
sAl.O  !5 RECOReS
"
.........,./OEEP COHSION~SS SOILS /~ (>250 FT) ~ 30 il.HOiH!S
~ECOROS:
Fig. 8.11 Average acce),i:f<ltion spectra for dif~eren[ ;:;oi! conditions (After Seed et al. 1976; from Seed and klfiss :982.)
An example of a conscrvat:ve design spectrum is shown in Fig. 8. j J. 'fhis figure shows four spectral acceleralion curves representing the average of normalized spectral values corresponding to several sets of earthquake records registered on four types of soils. The dashed line through the poims A, B, C, and D defines a possible conservalive design spectrum for rock end stiff soil sites. Normalized design spectral shapes. as those includes in the recent edition of the Unifonn BUI:ding Code (ICBO, 1994) (Fig. 8.12) are based on such sImplifications. The UBC spectral shapes become trmnear. when drawn on 0. tripanite logarithmic chart, similar in shape to 3, NewmarkHall spectrum. UBe spectra! shapes (leBO, 1994) can be expressed I by the following rather simple formulas:
Fig, 8.12 Normaltz.ed design $pectHI shapes contained in Uniform Building Code. (ICBO 1994.)
SA
=i
+75 T
for
SA =2.5
for
for
(8.24)
5, = 2.28SIT
T>O,915 sec
S, = I + 10 T
SA =2.5
for
O<TSO.IS sec
0,15 <TsO.39 sec T> 0.39 sec
(8.22)
for
for
S.
= O.9751T
where SA in the spectral acceleration for 5% damping nonnalized to a peak ground acceleration of one g, and T is the fundamental period of the building. It should be noted that values obtained from the UBC speclraj chart of Fig. 8,12, Or alLematively, calculated with eqs. 8.22 to 8.24. are too conservative. In actual design practice, these values are scaled down by lhe structufil1 fac!or Rw (Chapter 25) with values between 4 and 12, depending on the type of building. However the UBC establishes limitations for tbee resultant base shear force obtained by rhe dynam~c method reiative to the base shear given by the static method of a:Jalysis, as explained in Chapter 25.
j
0<T;;;O.15 sec
(8.23)
8.6
~t1d
36C
For cenain types of extreme events such as nuclear bla.st explosions or strong motion earthquakes, it is sometimes necessa.ry to design structures to withstand strains beyond the elastic limit. For example, in seismic design for an ear1hquake of moderate intensity, it is reasonable to assume elastic behavior for a weUdesigned and cons~ructed structure. However, for very strong motions.
248
Response Spectra
249
lQrce
?l,m,e
Y'lrl~Y
"rk'.,
Y,
p.~ ?l~~tic
k
Y",.,
Y
Elastie
fbi
this is not a realistic assumption even for a welldesigned strJcture, Although stnlctures can be designed to resist severe earthquakes, it is not feasible economically to design buildings to elastically withstand earthquakes of the greatest foreseeable intensity. In order to design structures for strain levels beyond the hnear range, the response spectrum has been extended to include the inelastic range (Kewmark and Hall 1973), Generally, 'he elastoplastic relation between force and displacement, which was discussed in detail in Chapter 7, is used In structural dynamics. Such a forcedisplacement relationship is shown in Fig. 8.13. Because of the assumption of elasloplastk behavior, jf the force lS removed prior to the occurrence of yielding. the materia! wE] return along its loading Hne to the origin. However, when yielding occurs at a displacement y/. the restoring force remains constant at a magnitude R" If the displacement is not reversed, the displacement may reach a maximurr: value Ym;!:; if, however. the displacement is reversed, the elastic recovery foHows along a Hne parallel to the initi2J line and the recovery proceeds elastically until a negative yje!d value R<; is reached in the opposite direction. The preparation of response spectra for such an inelastic system is more difficujt than that for elastic systems, However, response spectra have been prepared for several kinds of input disturbances. These spectra are usually plotted as a series of curves corresponding to definite values of the ductility ratio }Jr. The ductility ratio fL is defined as the ratio of the maximum displacement of the structure in the ihelastic range to the displacement corresponding to the yield point yy, that is,
(K25)
N~tV(~llJef;od
T, Sl:e
Fig. 8.14 Response spectra for undamped eiastoplutic system for the 1940 EI Cc.ntro earthquake. (From mume et aL t 961.)
The response spectra for an undamped singiedegreeoffreedom system subw jected to a support motion equal to the EI Centro 1940 earthquake IS shown
.f
.~
in Fig. 8.14 for several values of the ductility ratio. The tripartite logarithmic scales used to plot these spectra give simultaneously for any singkdegreeoffreedom system of natural period T and specified ductility ratio fL, the spectral values of displacement~ velocity, and acceleration. Similarly, in Fig, 8.~5 are shown the response spectra for an elastoplastic system with 10% of critical damping. The spectral velocity and the spectral acceleration are read directly from [he plots in Figs. 8.14 and 8.15, whereas the values obtained for the spectral displacement must be multiplied by the ductility ratio in order to obtain the correct va!ue for the spectral dispJacement. The concept of ductility ratio has been associated mainly with steel struc~ tures. These structures have a loaddeflection curve that is often approximated 3S the elastoplastic curve shown in Fig. 8.I3(b). For otber types of structures, such as reinforced concrete structures or masonry shear walls, still. conveniently. the load<ieflection curve is modeled as the elastoplastic cur.re. Aithough for steel structures, duc(ility factors as high as 6 are often used in co;lapselevel earthquake design, lower values for the ductility ratio are applicable to
t, riC,
i (.
"
250
Response Spectra
251
!
.!1 <0'
J';
i
0.1
0.2
3.
Fig. 8.15
Response spectra for elastop:atic system with 10% critical damping for tbe
Frequency, cps
masonry shear walk The selection of ductility values for seismic design must also be based on the design objectives and the loading criteria as well as the risk level acceptable for the structure as it relates to its use. For reinforced concrete structures or masonry walls, a ductility factor of to to 1.5 seems appropriate for earthquake design where the objective is to limit damage, In other words, the objective of Emit damage requires that structural members should be designed to undergo little if any yielding. When the design objective is to prevent cOllapse of the structure, ductilities of 2 to :} are appropriate in this case.
Fig. 8.16 Inelastic desjgn spectrum nonnaHzed to peak ground acceleration of 1.0 g for 5% damping and ductility factor j.J.. = 2,0, (NOTE: Chart gives directly Su and Sit; however, displacement va[nes obrainec from the chart should be multiplied by ).J. to obtain
SD')
8.7
In the preceding section of this chapter. we discussed the procedure for calculating the seismic response spectra for elastic design. Figure 8,9 shows the elastic design spectra for severa] values of the damping ratio. The same procedure of constructing a basic response spectrum that consoHdates the "average" effect of
several earthquake records may also be applied to design in the inelastic range, The spectra for elastoplastic systems have the same appearance as the spectra for elastic systems, but the curves are displaced downward by an amount that is related to the ductility factor JL. Figure 8.16 shows the construction of a typical design spectrum currently recommended (Newmark and Hall 1973) for use when inelastk action is anticipated. The elastic spectrum for design from Fig. 8.9 corresponding to the desired damping ratio is copied in a tripartite loga.rithmic paper as shown in Fig. 8.16 for the spectra corresponding to 5% damping, Then lines reduced by the specified ductility factor are drawn parallel to elastic spectral lines in the displacement region (the left region) and in the velocity region (the central region), However, in the acceleration region (the right region) the recommen
252
Response Spectra
253
ded reducing factor is . This last line is extended up to a frequency of about 6 cps (point P' in Fig. 8,16). Then the inelastic design spectrum is compieted by drawing a line from this last point P' to point Q, where lhe descending Hne from point P of the elastic spectrum intersects the line of constant acceleration as shown in Fig. 8.16. The development of the reduction factors in the displaceme,nt and velocity regions is explained with the aid of Fig. 8.17(a). This figure shows the forcedisplacement curves for elastic and for elastoplastic behavior. At equal maximum displacement Ymn for the two curves, we obtain from Fig. 8.l7(a) the following relationship:
However, in the acceleration region of the (esponse speclrum, a ductility reduction factor f.L does not result in dose agreement with experimental data a.~ does the recommended reduction factor !2J.L  1 . This factor may be rationaily obtained by establishing the equivalence of the energy' between the elastic and the jnelastic system, In reference :0 Fig. 8.17(b) this equivalence is eSlab.lished by equating the area under elastie curve "Oab" with the area under the jnelastic curve '<Ocde," Namely,
F,.yy
(8.28)
or
F
.'
= FE
fJ.
(8.26)
where FE and Fy tire loe elastic .md inelastic forces corresponding respectively to the maximum elastic displacement)lF. tinct 10 the maximum inelastic displace~ ment Yp. and wbere y., is {he yield displacement. The substitution of }' = Ff)k, )'J = FJk, and YP = p,.y; into eq. (8.2&) gives
where FE is the force corresponding to the maximum displacement J'm~. in the e!astic CUrve and F" is the force at the yield condition. Equation (8.26) then shows that the for~e and consequently the acceieration in the dastoplastic system is equa! to the corresponding value in the elastic system reduced by the ductility factor. Therefore, the spectral acceleration S" for elasloplastic behavior is related to the elastic spectral acceleration SaC as
(8.27)
Or
Thus,
Fe Fy = .."'('= ... ~=
\' 2Jt .and, consequenlly,
;1
;1
Iflelulic respmlSe
/
!nela~lk ~sponll
.L1k
/'
I
F, 
.L1k
I
I
a
I
I
b
Yc
c / 1 d 7,f+_.!.\,'
,e
,.,
Fig. 8.17
Thus, the inelastic spectral acceleration io the acceleration region is obtained by reducing the elastic spectral by I;,e factor fij;.~_ The inelastic response spectrum thus constructed and shown in Fig. 8.16 gives directly (he 'values for the spectral acceleration S(I and spectral velocity SUo However, the values read from this chan in a displacement scale must be multipiied by the ductility factor Jt to obtain the spectral displacement Sf)' Inelastic response spectral charls for design developed by the procedure eX.plained are plesented in Figs. 8.18, 8.19, and 8.20 correspondingly to damping factors ~;. 0,5%, and 10% and for ductility ratios JI = I, 2, 5, and 10.
.,
i l
::
254
Structures Modeled as a
SjngleDegfee~ol~Freedorn
Syslem
Response Spect:.3
255
.'
FI4;quency. cp!j
Frequ(!ncy. Cj}:1
Fig. 8.18 Undamped inelastic oe:;.ign spectra normalized to peak grounJ acce;eralion of 1.0 g. (Spectra! values 5<1 [or acceleration and S~ (or velocity are obtained direcUY frolD the graph. P.owever, values of S9 for spectral displacement should be amplified by the ductility ~atio jL.)
f;
Fig. K 19 Inelastic design spectra normalized to peak ground acceieraJion of 1.0 g for 5% damping. (Spectral y"lues Sa for acceleration and S" for velocity art direcdy obtained from Ihe graph. However, values of SD for spectra! displacement should be amplified by the ductility rutio ,u.)
I ,
sjngledegree~offreedom sys~
f ,
.
tern of Examp]e 8.: > assuming that the struchlre 1s designed to withstand seismic motions with ar; eiastopimaic behavior having 11 ductility ratio jJ = 4.0. Assume damping equal to 10% of the cri[ical damping. (a) Use the :"esponse spectra of lhe EI Centro earthquake. (b) Use the design response spectra. Solution: (a) From the response spectrum corresponding to 10% of the critical damping (Fig. 8.15), we read for T = l sec and [he curve labeled
t ,
The factor 4.0 IS required in the calculation of So since as previously noted the spectra plotted in Flg. 8.15 me correct for acceleration and for pseudovelocilY, but for displacements it is necessary to amplify tile values read from the chart by the ductiH[y ratio.
(b) Using the inelastic design spectra with 10% damping for design in Fig, 8,20, corresponding to 1 cps, we obtain the following maximum values fo: the
response:
,I
:1
S[J
= 3.84 in
'" = 4.0
Slj = 6.2 in/sec and
'I
I I
= 0.096 g
I'
[I I
255
Response Spectra
257
8,8
The computer program described in this chapler serves LO calculate elastic response spectra in ierms or spectral displacement (maximum reiafive displacement). spect:al velocity (maximum relative pseudo velocity), and spectral acceleral~on (maximum absolute acceleration) for any prescribed timeaccelerationselsrr.ic excitation. The response is calculated in the specified r<l;<ge of frequer:cies using the direct integration method presented in Cntlpter 4, Example B.3 lise Program 6 to develop the response speClra fo: elastic systems SUbjected to the first ten seconds of the 1940 EJ Cen!ro earthqu .. ke. Assume :0% dampiflg. The. digitized values corresponding to the acccJernlions recorded for the. firsl 10 sec of the E1 Centro earthquake are given in Tble 8.3.
Solution.:
\C.I'.".)
r. S_)
ON1PlNC RA1'lC
'nNe STLI" IN'tt(;i!I\T10,",
,'CCf:LER1.Ti0l1 GAAV:;1"(
," "
"
,HI>
~'II\;O,
(CI".S
Sf>E:(:T
P1SI'"
$.ttcT
VEtX
(IN)
24.65
( INfSEC]
Fig. 8.20 Inelastic: design spectra norm;Jlh:cd to pcnk ground acc:e!cnlliou of 1.0 g' for 10% damping. (Spectral values S" fo: acceieralion and Sv for velQcity are directly obtained from the graph. Howcycr, vah.;es or Sf) for spr.ctral displacement should be ali'.plified by lhc ductility Hillo p..)
lJ,05
1)'1!)
,
tJ
'>0
:2:\ .86
O. I') 0.20 O.
2~
""
" n
u
c. }O
G JS
As can be seen, these spectfJl values based On the design spectrl.m are
somewhat different than those obtair:ed from the respo:lse spectrum of the EI Centro earthouake of J940, Also, if we compare these rest.;lls ror the elastoplaSlic benavi~r Wilh the results in Example 8.1 for the elastic slructure, we observe that the maximum relative. disp!aceme:lt has essentially the same magdtude whereas the acceleration and the reli'ltive pseuGovelocity are appreciably Jess. This obSe;vtllion is in general true for roy structure when inelas~k response is compared with the response based On elastic behavior.
O. '0 O.
~
11.t)2
"
$:
H,
20 .31
n n.n
"
"
';;!i .02
1S.17
6'1.25
1>. 1~
(/.50
LaO 1. 50
]. CO
,
2
".H
2.41
.t~
"
""
H
~:).
&1.
$~
1S
127.4()
n.n
25.H
H, U
lU_S7 )16.111
H9.~()
L.50
1.17
O. 7~
) .oc
J. 'jU
1 J. 96
261.1.\
~ ...
O.
$)
1l.H
\. )!i
coo
4. ~o
S _flO
\lA, 0.31
".JO I). OS
lo,a
9.1>&
$.0
15'1.
lSS.
294.
to.oo
15.00
20.CO
1."0
0.02
1).l)l
L,'
". SOl
" "
III
liS .0.4
120.01'
"
~.
TABLE 8.3
Digitized Values of the Accel~ration Recorded 10r the First Ten Seconds for the EI Centro Earthquake 011940
.~~~~ ~~"
 
Time (sec)
Ace,
(ace. g ')
Time (sec)
Ace.
Time
(se.c)
(ace. g ')
Time (sec)
Ace.
(ace. g ';
0.0000
0.2210 0.3740 06230 0.7890 0.9410 10760 1.3840 1.5090 1.8550 2.2150 2.4500 2.7080 3.0680 3.3860 3.6680 4.0140 4.3140 4.6650 5.0390 5.3020 5.5100 5.8000
O.OlOS
0.0189 0.0200 0.0094  0.0387  0.0402 . 0.0381  0.0828  0.1080 0,1428 0.2952 0.2865 0.1087 0.0520 0.1927 0.0365 0.0227 0.1762  0.2045 0.0301 0.1290 0.1021  0.0050
0.0420 0.2630 0.4290 0.6650 0.8290 0.9610 10940 1.4120 1.5370 1.8800 2.2700 2.5190 2.7690 3.1290 34190 3.7380 4.0560 44160 4.7560 5.1080 5.3300 5.6060 5.8090
0.0020 0.0001 .. 0.0237 0.0138  0.0568 0.0603 ~. 0.0429  0.0828  0.1280 0.l777 0.2634  0.0469  0.0325 0.1547  0.0937 0.0736  0.0435 0.1460 0.060S 0.2183 0.1089 0.0141  0.0275
0.0970 0.2910 0.4710 0.7200 0.8720 09970 1.1680 14400 1.6280 1.9240 2.3200 2.5750 2.R930 3.2120 3.5300 3.8350 4.1060 4.4710 4.S31O 5.1990 5.3430 5.6900 5.S690
0.0159 0.0059 00076 0.0088 0.0232  0.0789 0.OR97  0.0945 0.1144 0.2610  0.2984 0.1516 0.1033 0.0065 0.1708 0.031 I il.0216  0.0047  0.2733 0.0267 0.0239 0.)949  0.0573
]0660
1.3150 1.4810 1.7030 2.0070 2.3950 2.6520 2.9760 3.2530 3.5990 3.904il 42:>20 4.6180 4.9700 5.2330 5.4540 5.7730 5.8830
00001 0.0012 0.0425  0.0256  0.0343  0.0666 0.1696 0.0885 0.2355  0.3194 0.0054 0.2077  0.0803 0.2060 0.0359 0.1833  0,1972 0.2572 0.1779 0.1252 0.1723 0.2420 . 0.0327
TABLE 8.3
(continuation}
~~~.
Time
(sec)
Ace.
(ace. g ')
Time
(sec)
Time (sec)
Time (sec)
Ace.
(ace. g ')
0.0665  0 0200  0.0603 0.0033 0.0017 0.0385 0.0113 0.0360 0.0272 00109 0.0036 0.0068 0.0716 00468 0.066)  0.0369 0.1534  0.0022 0.0955 0.)301 0.0816 0.0586 0.0093
~ ~
5.9250 61320 62290 6.3820 6.5200 6.6030 67280 0.8520 71210 72260 'i4250 76000 7.7520 7.9600 8.1260 8.2780 85330 8.8180 8.9560 9.1500 9.4410 9.8150 )0.0200 )0.1500
0.0216 0.0014  0.0381  0.0162 0.0043  0.0170 0.0009 0.0022 0.0078 0.0576 0.0186  0.0628  0.0054  0.0140 0.0260 0.Q305  0.0344 .. 0.0028 OJ 849 0.1246  0.1657  0.0881  00713 0.0024
5.9800 6.1740 6.2790 6.4090 65340 6.6450 6.7490 6.9080 7.1430 7,2950 74610 76;10 7.7940 7.98'10 8.1660 8.3340 8.5960 8.8600 9.0530
92530
9.5100 9.8980 10.0500 10.1900
0.0108 0.0493 0.0207 0.0200 00040 00373 0.0288 0.0092  0.0277  0.0492  0.2530 0.0280 0.0603  0.0056  0.0335 0.D2.46 0.0104 0.0233 0.1260  00328 0.0419 0.0064 .. 0.0448 0.0510
6.0130 6.1880
6.31.60
6.4590 6.5620 6.6860 6.7690 6.9910 7.1490 7.3700 7.5250 7.6690 7.8350 8.0010 8.1950 8.4030 8.6380 8.8820
9,0950
9.2890 9.6350 9.9390 10.0800
0.0235 0.0149  0.0058  0.1760 .~ 0.0099 0.0457 0.0016  00996 0'{1026 0.0297 0.0347 0.0196  0.0357 0.0222  0.0128 0,0347  0.0260 0.0261 0.0320  0.0451  0.0936  0.0006 .. 0.0221
6.0850 6.1980 6.3680 6.4780 6.5750 6.7140 6.8110 7.0740 7.1710 7.4060 7.5720 7.6910 7.8770 8JI700 8.2230 8.4580 8.7350 8.9150 9.1230 9.4270 9,7040 9.9950 10.1000
~
_...
..,.=.~~.

260
Response Speclra
26i
8.9
JIS
Example 8.4 Use the Program COSMOS to genenHe response spectra for :.elastic system subjected to the 1940 EI Centro earhqunke. Assume damping equal (Q 10% of the crillcsl dampir.g.
Solafion: In generating response spectra, the program, COSMOS bas tbe capability to calculate the spectra at any specified coordinate of [he structure, Because of this capability. it is necessary to model the structure and calculate the natUfal frequencies before requesting the generMior; of ~he response spectra at a specified coordinate. Example 8.~ is implemented in COSMOS as a spring. mass system with spring constant I and mass value 1. Then the natural frequency is calculaled and the response spectra at the base of this s),slem (coordin:He i) generated. Timeacceleration values for EI Centro earhquake (Table 8.3) are inpot in COSMOS through a previously prepared file that 'contains 186 digiH7ed point of acceleration curve recorded for El Centro earhquake of 1940 for the first 10 seconds. Figure 8.2i shows the plot of this record obtained using COSMOS. The acceleration response spec:ra ge.nerl!;ed by COSMOS for frequencies :n the range 0.314 rad/sec (OJ cps) to 125.7 rad/sec (20 cps) in shown in Fig. 8,12.
"'f:' r
I
lSG ,
,,1 \ I (\ , i \/j
....
..
I
!
IN
'\
".. :
'=IH=:.
I
r1
..
:
I
i
~
1'1
J
i i
,
......
~
,
12. ::;
"
I I
SL:i
I
I 12 ~
15
87.:;
1?S
FR[O lR.d/S'!)cl
Fig. S.22 Ac::elel:likm response spcctwrn ;){ Ihe base of an e!;istic syste:n subjected 10 ground nlOllOO of 1940 E! Cenlro carthquilke,
Solution: The analysis is performed ~!iing n single spri:Jg element wHh one concentrated mass element The following commands are implemented in COSMOS:
(l) Set view
[0
the XY plane:
>
I I
!
i
,
':
'
:,VY

~rv~ I
VIEW
.$ .IS
,1
,
&.tB
>
PLANE
,
&
,A J .Ia
9.15
E! ' vr~'._",
,
I ,
<
0, 0, 0
,..
I ,
I I
(4) Define element group J using the SPRING element formula::on with
............... .......
two nodes:
S.S
"
com
1,
0,
lillE
Fig. 8".21 Ground acceleration for the nest lO scc recorded for the po!.ent of {he 1940 El Centro earthquake.
262
Response Speclra
263
use default values for all integra~jon parameters; and request printout of relative disp:acemem and relative velocities:
ANALYSIS )
PD~ATYPE,
POST~DYN
l'CCi{,
?D_A?YPE
2.
1,
2000,
Ot
.01,
0,
0.5,
C.25,
EGROUP,
2,
MASS,
0,
0,
0,
C,
0,
J,
=l
(15) Prepare file ELCENTRO,GEO in a formal suilable for input in COSMOS, cont<'iining the values of the ground acceleration for 1940 EL CE~TRO EARTHQUAKE shown in Table 8.3, aod lisl the firsl five lines of lhis file
P!)_Cl:RDF,
1,
RCONST RC:ONST', 2, 2, 1. 7, 1, 0, 0, 0, 0, 0, 0
G,
.J10S,
.H1.
.0001
.O~12
.221,
. }?4,
"P .
"
S, 13,
.0425 .tlJ'D
.6:n,
.71, .0256
769, ,0331,
:1_?T, 2, 2,
(16) Define dynamic forcing function as timedependent hase acceleration in the X direction, Read forCing function curve from file "ELCENTRO.GEO":
ANALYSIS> PD_CURTYP,
POST_D'iN ,. ::URVF::S
>
PD~C()R?YP
o.
C,
l, 0, 1
U'l'ILI'1'Y ) "!?ILS
CONTRCL
PO_BASE,
l',
386.4, 0,
0,
(17) Activate XY plot information for acceleration input at node 1 as a function of time and generate plot (see Fig. 8.20:
DISPLAY ) XY._PL07S :'> ACTXYPRE l, ;., TIt!, L 12, L DISPLAY X':CPLO?S) XYPLOT
ACTXYPRE, XYP:"OT,
(l2) Set the options for the frequency analysis to extract ~ frcquency using the Subspace Iteration Method with .t: maximum of 16 jterations. and [un tte freqJency analysis:
p~ALysrs
1E05,
0,
106,
C,
0,
0,
0
R_FREQUENCY
(18) Define modal damping for mode 1 as. 10% of critical damping (0.10):
ANA!. 'ISIS ~ ?OST. __ DYN > Pi:LDhliP;'GAP ) PD_MDAN?, 1, 1, 1, .1 PD_HDA.."1P
FREQUENCY
imMBER
PERIOD
(SECONDS)
ANA~YSIS
>
(RJ:i.D/SEC)
PD...:.Nf:sP,
C 1000000E+01
0.1591549E+00
0.6283185+01
(14) Define analysis type as modal time history using one natural frequency, 2000 time steps starting at 1=0 with a time increment of 0.01 sec;
R_DYNM{IC
264
Response Spectra
265
(21) Define analysis type as response spectrum generation using one natural frequency, starting at 0.314 fad/sec (0.05 Hz). ending at :25.7 TJdJsec (20 Hz), linear scale, with 200 points in frequency range, 10% of critical damping, for node 1 in the X direction:
ANALYSIS , ?OST_DYN )
PD~ATYPE
Use an editor to display the output file (EX8w4.0UT) and print the firsl nine lines of the response spt"~[ra:
RISSfOl'lS~
SIT.c7I\lJIf
'OR
NOD:;
~N
!)~l\gCT;OI
,
(S!tCl
,
TRUE ktt. ('S""'UX I\Et. PSEU:xl
AU$
PD.J,':''::.'PE.
Q, Q,
3,
1,
.314,
125.7,
1,
200 ..
1,
0.1,
I,
0,
PElno!)
'tRUE Rtt.
'tRUE
0, 0
Dr$p
\itt. C.';6!HE.Oj!
Q_::tlln;,~:
;c
.Q .2H2~JE.'.t
VEL
Ace
".:n10E.n~
jj.'H~t"~;;
Q.H11t,01 1.20HE.C)
O.1"f)f..:j1
~.,>~;.;:.c'
'.1J7S)SE.ry,
b.lsn48E4': .n.1)a%tE,ol
jj_il~J1n:.;n
POST~CYN
) R_DYNAM:C
J. ,tf"lUE.t2
:.lun~E.;JZ
C.~2SH(f.'n
C.~~6?Ht.Ol
~_:17HSE'01 O.2S~SHE.(,:
O_l)~S?lE.Ol
(23) Activate XY plot information for X acceleration at ~?de 1 as a fUfictlor: of frequency (response spectrum), and generate plo~ (see Fig. 8.20):
DISPLAY > XY_PLOTS > ,t..CTXYPOST
C.no{,>ol
O.nH~.Oi
1),}HH~(t\
O.HnB'Cl
O.HU!:.O:
o.a~5L9.Cl
O.5~Ht.CI}
". ,'64,1,.0:
o.n~9'n>~,
O.'n~9J!t.h
().S;;$MS!:.Ol
O.n~678E.n
q.lO~lH!t.Q:;
dJ.6i1.06;;:.02
o.$P~Ht.u<
O.H'1)~E.OJ
OAOn)e,H
o.
;:;nS.E'~l
O.20a)"E'~1
].,.:I$S6'H.:>OZ
I
:~
ACTXYPOST,
D::;:SPLAY
1, FREQ, AX,
1, 12, 1,
0,
IN
XY_PLOTS) XYPLOT
B,10
SUMMARY
pOINT
FR8Qt:ENCY (BAD/SEC)
SPECTRAL
ACe.
,
2
,
3
6
7
9 10
3 .140eOl 9 441e01 1. aOOeOl 1.S74eOl 2.204e01 2.834eOl 3.464e01 4.09Se01 4.72SeOl S_3SSeOl
Response spectra are pl0iS tha~ give the maximum response for a single~ degreeoffreedom system subjecied LO a specified excitation, The construction of these ploes requires tbe solution of singledegreeof~f:eedom systems for a sequence of values of the natural frequency and of the damping ratio in the range of interest. Every so;ut:on provides only one point (the maximum value) of the response spectrum. In solving the single~degreeoffreedom systems, use is made of Duhamel's integrak or of t!:e direct method (Chapter 4) for elastic systems and of the step~by~step linear accelerJ.ion method for ir,elastk behavior (Chapter 7). Since a la:ge number of systems must be analyzed to fully plot each response spectmm, the task is Jengthy and time consuming eveu with the use of a computer. However, once these curves 3re constructed and are availabJe for Lhe excitation' of interest, the analysis for the design of sl.rlJctures subjected to dynamic loading is reduced to a simpJe calculation of the natural frequency of the system and the use of the response spectrum. In the following chapters dealing (hat structures which are modeled as systems with many degrees of freedom, it will be shown that the dynamic analysis of 3 system with Jt degrees of freedom can be transformed to the problem of solving n systems in which ead1 one 1S a singledegree~offreedom system. ConsequentlY, this transformation ex.tends the usefulness of response spectra for singledegreeoffreedom systems to the solution of systems of any number of degrees of freedom. The reader should thus reaiize the full importance of a thorough understanding and mastery of the cor.cepts and methods of solutions for single~degreeof~
Response SpeClr3
8.4
267
freecom SVSlems, since tbese S'lme. methods "fE also llpplicabk to systems of many deg;ees of freedom afler {he prob~eflj has been transformed to if\depcn~ dent s;nglt>degreeofrreedom systems,
8.5
Derennine the lll~iiCirnum S((CSS in (!.e COhOllllS or the frame of Proble:':) g.] The frame sbowll ;;~ Fig. Pil.t i;; sub;ee\O!d to Ihe excll:!.liol\ prowLced by the El Centro eanhqu~ke of i940. A~$ume 10% camping ano from. lhe appropnllle chnrt de:ermine lilt specllal v;)ttJe~ fOf di~;):aCtmeIH, velocity. and :,cceleftIlJon . ..,\s
$.urne e)..Slic b.;ll,,"VlOr.
8.G
DROBLElVS
8.1
The steel frame show!' in Fig. P8.1 is. subjeclcd ~o horizontal fo.:ce al the girder level of (1000 sin 101 lb) for a lime clllf~lion of half a cycle or lhefofciilg sine funcliol"l. Use !!Je appropriate r..::spons.e spectral chtlf{ to obtaIn {\) rnr.ximum Jisp!acemem. Neg.lec{ d:l.rnping.
I 11 K,p.
Repeal Pmbkm 8 .5 using thl: b;\sic Gcsign spC<.:lra g,iven In Fig. 8.9 to delermine tile ;;pec!fal vahres for acceli:w;jon, ve:OCily, ;u1d displacement. (Scale down speciml Y<llucs by r<it:lOI' 032.;
8,7
A slructure morJded as t:~e s:>ring'll:ass systcm sJ,own in fig. ?8.7 is ;:,ssumd 10 be. subjcc{c\J to ;:, suppon mOllon :1fouuced by (hI! El Cenlfo eanhquake of
1940. Assuming ei~lstic bch:\.vior and using IliC ;,prropria!c response spect{;l; ch::.tn find thi!: !\)3Xj;Y;lHll fcia:ive displacement between lhe mass and Ihe suPPOrt. Also co:nplle :hc m,\f:m:um foro..:..: .:c!ing on Ihe spring.. Negkcl dan\ping.
l\OCC WI
'" 0
'0, tl
\,>~ilCl
'l~
W{;
l
S.2
S.]
~.'~ 1'
x 20
~1
E,g
Fig. P8.1
Ro::pe(l.l Problem 8"7
JS~U!ll;llg :lm
Determine the maximum Stfcsses in [be COI...,lnlIS or the (Hune of P(():.lle:n 8. L Cons',der the rralne shown in F:g. \)8.3(;;) suojeCied to (l. foundillion excitation
8.9
8.10
(0
l produced by a h:!!f cycle or Lhe runction til =::; '200 sin WI inise:: as ~bown in ]~ig, pg,3(b). DeLermi.:'!c tht m<1x~murn hOri:r.Olltlll disl,lacemem or the gU'tkf relative to the moCon of Ihe founu:>riofl. Neglect daruping.
C(}ns~der Ihe '1jJringm>lss .'YSIClll of Problem 8.7 ,\I\J tiS$UI11 lila! the spring dement follOws, :;,;1.1 tlun c!aslOp!:I~!ic b;:hnvior with a l\l.J.:;.iwutll valu~ fOr (he f~SIOfL!g fo(c~ ill tc.cSIOIl <oj( In C(":lp:I;;.5~;OH i!i equal to !Ialf the value of the elastic maxinn!!1\ rcree ;;) the !ipri!l& calculated til Problem 8.7. Determine ;he sp<!cu;L! value fOJ (he displacClHcrll, Neglect 01UUpIOg. (lfint: S!ar( by (l.ssuming J.1. "" 2, DnJ spectral vabc So, ca!cd:ill! p., ;"!nu niH.! new spectral vulues, etc.)
8.11
itU
l f,
A Mfucture modtk~ as <.l sl:lg!::ucgn:.eoffn:.:eJclll system has a natural period r"" 0.5 sec, Use (he fCSponsi! spcclc:ll me~hod tQ dCicnnine ill the clastic range lhc m:t,:im..;:n :"!bsohne 'Hxderatl<);). ~ilc ;:,::ximuli\ rdauvc displacemem. and lhe n:l.aximurn p:'l;.'uJcve\oci:y for: (aJ j\ found;:.rion JllOtion ec;ual LO :he EI Centro earlilqu:.tKe of ;940 and (b) Ihe Gi:!lign spectrum wilh a maxinlllffi ground :!.cceler<iltOtl equal to 0.3 g. Neg!ect d(l.lIlping. SOlve Problem 8.12 (l.ssUl.\ing elaslop!a$I;C .alio jJ. 4.
bdlJ'<'\O{
ILl3 8.14
Use Program 6 10 de.,.dop a lutile :l:.tving: the $p<!cuaJ v:;.Ll:es [Of oi5planmcl:'ls. velocities, [lJlJ acce!er:..t:Ons in tb"~ r:Hlge of [requCl,cy from O.JO cps 10 1.0 cps.
268
The excitation is a constant acceleration of rr.('lg:1ilutie O.O! g applled for ;0 sec. Neglect domping. 8.15 Use Program 6 to deve!op the response spectra for elastic syStems subjected to the first 10 sec of {he 1940 El Cen{fO em1hquake. Neglecl damping. The digitized values corresponding [0 (he accelernrions :ecorded ror the firS( to sec of the EI Centro earthquake ue given in Table 8.3. 8.16 Repeal Problerr. 8.! 5 corresponding to a system with JO% of the critical dam?ing.
PART II
9
The Multistory Shear Building
In Part I we analyzed and obtained the dynamic response for structures modeled as a sing;edegreeoffreedom system. Only if the structure can assume a unique shape during its motion will the singledegree model provide the exact dynamic response. 0!herwise, when the structure takes more than one possible shape during motion, the solution obtained from a singledegree model will be an approximation to the true dynamic behavior. Structures cannot always be described by a singledegree model and. in ge:1erai, have to be represented by multipledegree models. In fact. structures are continuous systems and as such possess an jnfinite number of degrees of freedom. There are analytical methods to describe tie dynamic behavior of continuous structures thal have uniform material properties and regular ge~ ometry, These methods of aT1a1ysis, though interesting in revealing information for the disc:ete modeling of structures, are rather complex a:ld are apphcable only ro relatively simpie actual s.tructures. They require considerable mathematical analysis .including [he solution of partial differential equations. which will be presenred in Chapter 20. For ~he present. we shall consider one of the most instructive and practical types of stmcture which involve many degrees of freedom, the multistory shear building.
,~
27l
272
273
9.1
A shear buildng may be defined as a structure in which there is no rotation of a horizontal section ac the level of 6e floors, In (his respect, the deflec(ed building will have many of [he features of a cantilever beam tha[ 1s c.eflec!tc by shear forces only, hence the name shear building. To accomplish such deflection in a building, we muSt assume that: (1) the towl mass of the StrJCt:.ue is concentrated at the levels of the f1oors~ (2) the girders on the floors are lnfmilely rigid as compared to the columl1$; and (3) the defofF.lation of the structure is independent of the axial forces present in (he coiumns. These assumptions transform the problem from a structure with an infinite number of degrees of freedom (due to the distributed mass) to a structure that has only as many :::egrees as it has lumped masses at the floor levels, A threestory struCf:;re modeled as a shear building [Fig. 9. ita)} will have three degrees of freedom, that is, (he three horizontal displacements at the floor :evels. Th.e second assumption introduces the requirement that the joints between girders and columns are fixed against rotation. The third assumption leads to the condition that [he rigid girders will remain horizontal dunng motion. lc should be noted that the building may l:ave any number of bays and that it is only as a matter of convenience that we represem the shear building solely in terms of a single bay. Actually, we can further idealize tne shear building as a single column [Fig. 9.2(a)], having concentrated masses at the floor levels with the understanding that only horizontal displacements of these masses are possible. Another alternative is to adopt a multimass spring system shown in
ta.)
Ib)
Fig. 9.3(aj to represent the shear building. [n any of the three representations depicted in these figures, the s:~ffness coefficiect or spring conStant Ki showr. between any two consecutive masses is (he force required to produce a re:2t!Ve l.:nit displacement of the two adjacent floor leveis, For a uniform column wilh tne tWO ends fixed against rotation, the spring constar.t is given by
12El
k=
(9.1 a)
m, m,
k, ro,
k,
Ir,..j
,
1
I
F'll)~
K1(Yl
_m'i4
Yi t
A,
';
AI
I Y'I
I
.,
I
r+ '"'J{YJy?l...!
F~it)~ +m~
F*){Y:.tYl)
~k~ (Yly , l
_!
F,t!)+!
I')
J
II
4m~
X I >'1
" ,
,~
.
Fl
rYz~
(b)
:()+!
*!Y;
.J
I")
repreSenla~fOn
of a sh<!zu building.
274
275
and for a column with one end fixed and the other pinned by
3EI k=L'
(9.1b)
where E is the material modulus of elasticity, I [he crosssectional moment of inertia. and L the length of the column. It should be clear that all three representations shown in Figs. 9.1 to 9.3 for the shear building are equivalent. Consequently, the following equations of mmion for the threestory shear building are obtained from any of [he corresponding free body diagrams shown in these figures by equating to zero the
It should be noted that the mass matrix, eq. (9.4), corresponding to the shear building is a diagonal matrix (the nonzero elements are only in the main diagonal). The elements of the stiffness mmrix, eq. (9.5), are designated stiffness coefficients. In general, the stiffness coefficient kij is defined as the force at coordinate i when a unit displacement is given at j, all other coordinates being fixed. For example, the coefficient in the second row and second column of eq. (9.5) k n = k2 + k3 is the force required at the second floor when a unit displacement is given to this floor.
9.2
+ klYI
m,y, + k, (y, 
m,l' + k, (y, 
This system of equations constitutes the stiffness formulation of the equations of motion for a threeswry shear building. It may conveniently be written in matrix notation as
[M] {Y}
+ [K] {y}
[F]
(9.3)
where [M] and [K] are, respectively, the mass and stiffness matrices given, respectively, by 0
[M]
= ~
[ml
m2
(9.4)
[
[K]
= 
kl k,
0
+ k,  k2
k2 + k3
 k,
~j]
k, FI
(9.5)
and {y}, {)i}, and {F} are, respectively, the displacement, accelermion, and force vectors given by
(r)}
(9.6)
Buildings are generally designed to resist two types of forces: (l) vertical (gravitational) forces originating from the dead weight, occupancy, and equivalent loads; and (2) lateral forces caused by earthquake or wind loading. These two types of forces induce at each story of the building axial vertical forces, horizontal shear forces, overturning moments, and torsional moments. The overturning moments and also The torsional moments of the columns or other lareral resisting elements (structural walls) of the buildings have two contributing components: (1) the primary moments resulting from the vertical and lareral loads applied with their respeC[ive moment anns measured from the undefonned building configurmion; and (2) the secondorder moments resulting from the vertical forces acting over the incremental moment anns caused by the lateral deflection of the building. This latter secondorder comribution [Q the overturning moment is commonly known as the PL1 effect; P designating the vertical force and L1 the lateral displacement of this force. Generally, the PL1 effects in lowrise buildings are negligible because the total lateral displacements of low buildings are kept relatively small by the story drift limitations specified in the design criteria. However, in taller midrise and highrise buildings, where lareral displacements may be much larger, the maximum interstory drift limitations does not ensure that PiJ. effects will be negligibly smalL Since these secondorder effects are manifested after the building has experienced lateral displacements, an iterative process may be used to determine P/J effects. However. the PL1 effect can be introduced directly in the stiffness matrix when (he axial vertical forces are consmnts (weight), thus making it unnecessary to conduct an iterative process. In Fig. 9.4(a), the PJ effect is illustrated for a threestory shear building with lateral displacements y, subjected to constant vertical forces Wi at the various levels of the building. As shown in Fig. 9.4(b), due to these lateral displacements, secondorder overturning moments Mi given by eq. (9.7), are developed at the various stories of the building.
(9.7)
276
277
E,
h, F, F,
The application of eq. (9. i 1) to the threestory shear building in Fig. 9.4 results in
M,
I'<
(9.12)
I'<
Ail
h,
~:
h,
..... M 1
E,
,,/ q ,
(a)
\ hi
(9.13)
(1))
Fig. 9.4 Threestory shear buHding showing (a) lateral displacement )'" (b) story oyer~ tumiog momentS Ml and equivalent couples ,.0; X hi.
P;=
" IW
j""i
{q} = [KG]{Y}
(9.14)
with n = 3 for the threestory building in Fig. 9.4. The second*order moments i'VIi of eg. (9.7) can be substitured by equivalent lateral force couples of magnitude equal to the overturning moments M t has shown tn Fig, 9A(b), that is
where [KG] can be referred to as the lateral geometric stiffness matrix which for n~story shear building is given by
r
KG
T
P , P,
h:
P,
P;
,
hl
0 0
0 0
0 0
(9.9)
Solving eq. (9.9) for F,' and substituting MI from eq. (9.7) yields
P,
h,
Pj
h,
h)
(9.15)
(9.10)
0
with Yo= O. Therefore, the force qi at level i of the building is given by (9.11)
0
0
h,,_
"
1\ 
i I
+ P"
P,,I h,,_1
p"
hf;
P~
The consideration of the PtJ. effect in [he shea:: bl:ilding will rhen be equival e m
hr.
278
279
= [Ke]
[M] {Y}
or
[lvf] (y) + [[K]  [Kc] (Yl = {F}
(9,16)
{f)lY',
From eg. (9.16), it is seen that the P~!J. effect may be intmduced in the dynamic as well as in static analysis by modifying the stiffness ma:rix [K] which is decressed by the geometric matrix lKG], In the sQlution of eq, (9.16), the latera! d!splacement will be larger man those calculated with eq. (9.3) in which pLJ effects are not conside:ed. These increased lateral displacements will then result in greater values for the shear forces at [he varioL<s stories of buHding.
F,
It'H1=:::'::==I
+
9.3
An aHernative approach in developmg the equation of motion of a S[rUClUre is the flexibility forrnularion. In this approach, the elns[ic p!:'operties of the structure are described by flexibility coefficients, which are defined as deflections produced by a unit load applied at one of the coordinates. Specifically, the flexibility coefficient ~~j is defined as the displacement at coordinate i when it unit static force IS applied at coordinate j. Figure 9.5 depicts the flexibility coefflciems corresponding to unit force applied at one of the srory levels of a shear building. Using these coefficients and applying superposition, we may
stfHe that the displacement aC any coordinate is equal to the sum of the producrs ot" fiexibHiry coefficients ar that coordinare mUltiplied by the corresponding forces. The forces acting On [he threestory snear building (including [he inertial forces) are shown in Fig, 9,6. Therefore, the dispJncemenrs for the threestory building may be expressed in tenns of the flexibility coefficiems as
y,
miN/"
Rearranging the tenns in these equations and using maulx notation, we obtain
c'l"+
;
IJ.J.
(yl
(9,17)
t+IU....l I
I
where [M] is the mnss ",.trix, eq. (9.4), [tJ is the flexibility matrix given by
r~!J:~l ,
(9,18)
I.)
Ib)
tel
and {y}, {y}, and {F) nre, respectively, the dlsplacement, acceleration, and force vectors given by eq. (9.6).
',.
280
281
9.4
Inserting these expressior.s fo, (he flexibllity coefficients into mlHrix, eq. (9.18). resuits in
[~e
flexibility
The definitions given for either stiffness Of flexibility coefficients are based on static considerations in which the displacements are produced by static forces. The relationship between stat~c forces and displacements may be obrained by equating to zero the acceleration vectOr O'} 10 either eq. (9.3) or eg.
(9, (7), Hence
k;
k
k, l
[j]=
k,
+k,
t
+k,
kJ
(9.22)
I
k,
[K]{y)~{F)
+k,
k,
++k,
k!
k) ,
~
(9" 19)
(920)
[fJ {F}
{y}
The extension of the tlexibility matrix for a rhree~s[Ory shear :,uilding to any number of stories is obviolls from the pattern of eq. (9.22).
From these :elatlons it follows that the stiffness matrix (K} and [he flexibility matrix [t1 arc inverse manices, that is
9.5
[KJ = [fJ
or
[fJ = [K]
(921}
Consequently, the flexibility matrix Ul for a shear building ;nay be o~tained either by calculating the inverse of the stiffness matrix or directly f:om lhe definition of the flexibility coefficients. Taking this last approach, we obtain, for :he t~reestory shear building shown in Fig. 9.5(n),
kdu = 1
and
The computer program presented in this chaprer serves to model a S(t"..!cture as a shear building. Such modeling requires the development of the s{iffness and mass matrices of the sys~em" In Program 7. the elements of these matrices are stored in a file in preparation for dynamic analysis such as the calculation of natural frequencies and the determination of the reSponse of the structure when subjected w dynamic forces or to seismic exciratio"n. Program 7, stiffness and mass matrices of a shear building. contains the option of iocludino the p,jj effect, thus of calcu;ating the combined stiffness marrix (KcJ = [K} :;, [Kd. Smce the stiffr.ess of the mass matrices are symmetric, only the elements in the upper triangular portIon of these malrices need to be stored in files. The notation used, in this case, consis(s of numbe,ing consecutively those elements of the matrix located in the upper triangular part. For a matrix of order 5, the numbering of elements is as follows:
and
and
l
,
1(1)
(2)
symmeu.c
since [he flexibility coefficients for springs in series are given by the summation of the reciprocal values of the spring constants.
Using this notation, we need (0 store only about half of [he total number of coefficients in the stiffness matrix. As a matter of fact, we could save more memory in {he computer :;y no[ slo.ing the zero coefficients at the wp of each columr. of the m2trix. The implementation of such saving in computer memory
282
283
requires special coding to locate each entry in the diagonal of the ~t~x and, therefore, each coefficlen! of the matrix. Such coding has not been tmp{emenred in the computer programs presented in this volume. Example 9.1. Detennine the stiffness and mass matrices for the fourstory 6 shear building shown in Fig. 9.7. The modulus of elasticity is E = 2 X 10 psi. Neglect the PLl effect.
Solution:
5.S4';Z.. :U 5.2";:n;::+cz
;;.SA7n~C2
3.2'?F~CZ
c.oOD:n::.. JD c .OOJ"E+CO
G.ooon_()Q
I),::;{)O();::.OJ
:.OOG),,:'00
o. COOCf>CO
.
00')0:;:.00
M'j~SCO
0,OGDO;;:~0()
1)
.cooo;;.oo
OOCO;;;~OG
(l . DC M)1,;' C0
: _CONSGO
, ,
~)"GOS~;;O
O.0C:JO?:+;;iC
OJY:2~0Q
1.0"J0('c;;.co
Example 9.2.
P~d
effect,
.Sq59<:02
c) .~:H::>02
l.Z73SS.CZ
000(;::;.0<:
c,
,.
0. :::::OOS_co
,.)
o.coon_cc
~}
,,~"n~Cl ~;H!>C2
27J6I'+Cl
:s~nS>G2
0<:;00;;; .. 0,]
(). 0;0C2~OC
.)
nl",:.n
l~C
,C0
c.:sn~>o~
1.C~O
,
0
. OC.O;;S~:;'C
J _JC%E""CO
l, OOCC;:;~,jC
~'. jOCCS_~C
O.OOJC,,;:)(l
0, COvO __J{;
~.
o.ocon~Oj
~eQ,
0;;)
o .1"':lH:~C$
(].:S9~r>O$ O.l':l~:~j~
o,;:oc::t".oc
QOOt,;);>(;O
O.JOO:;'C.oo
())C1~t)D
!SCl.GO
0 QOO')_(;O C. m')oc",c;;
OOD'):;:"CC
o .CO~::O
". CO()CE~')C
:$:). all
9.6
SUMMARY
The shear building idealization of structures provides a simple and useful mathematical model for the analysis of dynamic systems. This model permits the representation of the srrucmre by lumped rigid musses interconnected by elastic springs, In obtaining the equations of motion, two different fonnulal:tons are possible: (1) [he s{iffness method in which the equations of equilibrium are expressed in terms of stIffness coefficients and (2) the fleXibility method in which the equations of compatibiEty are written in terms of flexibility coeffi cients, The stirn,ess matrix and the tlexibillty matrix of a system, in reference to tne same coordinates, are inverse matrices.
J""79,55in:~"
totl)!
In'" 1
t
I
180 10
PROBLEMS
I
~/ r
....
10079 total
55!~ 4~
1
9.1
'/JW
iZ7T
For the threestory shear building verify that the stiffness matrix, eq, (95), and the flexibility matrix. eq. (9.22), are iJ1Verse matrices. 9.2 For r.~e twoStory shear buiJding shown in Fig. P9.2 derennine the stiffness and flexibility matrices ar.d then verify that these matrices are inverse matrices,
284
285
;:'1
([)1========~========:jt
WlO
lBOO !b/ft
...
Y,
x 21
W10 X 21
1.'
....v
<25'+3IJ'i
Fig, P9.1.
9.3
For the rhret>.srory sheaf building shown in Fig, N.3 obtain the $[lffness and flexibility matrices and show thaI rhese matrices are inverse mmrices. AU the columns are steel members W 10 X 2 L
tOOO lb/ft
+
tt)
Fig, P9.6.
FJW.
r
)
9.8
De[ermine the stiffness and moss matrices for the shear building shown in Fig. P9.8. The ::nodulus of elnslicllY is "" 30 X 10 Pi>l.
15CX) Jb/h
F1irl
)2'
2000 IbJh
Y
30'
"""w
.a &
.a;,&
Fig. P9.3.
J
multimass~spring
9.4
9.5 9.6
Write the differential equation of motion using the stiffness formulation for (he
shear building in Fig. P9.2. Model the structure by a system. 9.9 Write the differential equation for the motion of the shear building In Fig. P9,],
Model \he stmclure as a shear column with lumped masses as the floor levels.
The threestory shear building in Fig. P9.5 IS subjected t;::; a foundation motion which is given as an accelennion function Y.(r). Obtain the stiffness differemjllJ: equation of motion. Ex?ress the disj:llacement of the floors relative to the foun~ dllticn displacement (i.e . u, "" )" Yf) 9.10 9.11
Use Program 7 Ie de,ermine the stiffness and mass matrices for (he she;:;r buliding shown in Fig. P9.9. The modulus of elasticity is f. "" 30 X !OI> pst. Neglec( the p. Ll dfecL
Solve Problem 9.9 inclUding lhe P::l
9.7
Generalize the results of Problem 9.6 and obtain the equations of motion for a
shear building of n. slories.
Use Program 7 to de!er:nlne the stiffness i:.nd mass matrices for a uni form (ensrory shear bui!ding: m which the mass :u each level it: 20 {lb sec'llin}; ~he imerstory heigh, is 12 ft; rhe mcdulus of elasticity is 3.0 X 106 psi; the ~()tol 9 flexur<:.l slifrncss is: :fEf 24 x 10 (lb in;') for e.!! the stories of {he buildinC" Neglecl the P.J effect. e'
286
10 lib ~ec~!;n,l
r rim==""""'1
l12"
10'
10
Free Vibration of a Shear Building
fl'<
~if2mrr'3"3'5~~
'""," ,. 400 in. ~
Fig. P9.9.
9,12
\Vnen free vibration :5 u:lder considerarion, the structure is not subjected to any external exciUlllon (force or suppor:: motion) and irs motion is governed only by the initi~1 conditions. There are occasionally circumstances for which it is necessary to detennine (he mmion of the structure under conditions of free v:bra!ion, but ~his is seldom [he case. Nevertheless, (he analysis of the Structure in free morion provides the most impor::um dynamic properties of the structure which are [he natural frequencies and tbe corresponding modal sbapes. \Ve begin by considering both formulations for the equations of motion, namely, tne stiffness and the tlexlbiliry equarlons.
10.1
The prOblem of f'ree vibration req~ires that the force vector {F} be equal to zero in either lhe stiffness, eg. (9.3), or fiexibiiilY, eq. (9.17), formulations of the equations of motion, For the stiffness equation with {F} to}, we have
( 10.1)
287
288
289
For free vibrations of the undamped structure, we seek solutions of eq. (10.1) in the fonn
[DJ
= [f] [,M]
(10,8)
i= I, 2, "" n
(yj
{a} sin
eM  a)
(102)
(10,9)
where at' is the amplitude of motion of the ith coordinate tnd n is the number of degrees of freedom. The substitution of eq. (to.2) into eq. (10.1) gives
 w'[M; {a} sin (we  a) ~ IK] {a) 'in (we  a)
or rez.rranging
~erms
where [f] is the unit matrix with ones in the ma~r. diagonal and zeros every* where else. For a r.ontrivial solution of eq, (iO.9), it 15 required that the detennioant of the coefficient matrix of {a} be equal to zero, thz.t 15, [DJ  I/w' [I] I 0
(10,10)
=0
(10.3)
lIK]
{OJ
Equat~o;1 (to.to) is a poiynornial of degree n in I Ivi. This polynomial is the characteristic equation of tne system for the fleXibility formulation. For each
whkh, for the general case, is set for n homogeneo~s (right~hand side equal to zero) algebraic system of linear equations with n unknown displacements a, ar.d an unknown parameter (,1/", The fonnulation of eq. (lO.?) is an impof:.ar.t mathematical problem known as an eigenproblem. Its nontrivial solution, that is, the solution for which not all at = 0, requires that the determinan! of the matrix factor of {a} be equal to zero; in this case,
one of the n solutions for (I IW') of eq, 00,10), we can obtain from eg, (lO,9) corresponding solutions for the amplitudes a, in tenus of an arbitrary constant. The necessary caJcuiatior,s are better explained with the use of a numer~cal example, ExampJe 10.1. The building to be analyzed is tr.e sjmple steel rigid frame lOJ, The weights of the floors and walls are indicated in the shown in f;gure and 3re assumed to include the structural weight as well. The building consists of a series of frames spaced 15 ft apart. It is further assumed that the stmctural properties are aniform along the length of the building and, therefore, the analysis :0 be made of an inte.:ior frame yields the response of the entire building.
1[K]  w'[Mj
In general, eq. (lOA) results in a polynomial equation of degree n In u} which 2 should be satisfied for n values of w . This polynomia: is known as the characteristic equation of the system. For each of these values of ul' satisfying the characteristic (lOA), we car. solve eg, (10.3) for a" a2. ,,,,0" in tetros of an arbitrary constant. Analogously, for tr,e flexibillty formulation. we have for free vibra::on from eq, (9,17) with {F} 0,
"''!
~:::W~,;=:50=='b=l=ft::::~.. ~ Y,
20 psf W10 X 21
10'
{y}
+ Ifl [M]
{OJ
(l05)
r=======H....
W",f
Y,
Vole again assume hanuonic motion as given by eq. (10.2) and substitu!c eq, (10.2) into eq, (10.5) to obtain
{al
Or
WIO X 45
(1M)
>
~~
I.)
(107)
"t
290
291
In the usual manner, these equatjons of motion are solved for free vibration by substituting
a)
a)
sin (M 
(l0.1l)
h
Fig. 10,2 Mult!mass~spring moce! for a twostory sheor building. (u) Model (b) frcc
sin (<:Jt  a)
body diagram.
Solution: The building is modeled as a shear building and, under the assumptions stated, the entire building may be represented by the springmass system shown in Fig. 10.2. The concentrated weights, which are each taken as the total floor weight plus that of the tributary wi1i1s, are computed as follows:
IVI~ IOOX30x 15+20X 12.5X ISx2=52,5001b
fk:+k 2  m l wZ
 k,
lo
<
co
(lO.I2)
For a nontrivial SO,'utl'on, we requl're thot ~ ... the d e t ennmant 0 f i he COeftlclent matrix be equai to zero, thar is,
(l O. 13)
1361b . sec' lin The expansion of this detenninant gives a quadratic equarion in
IV, ~ 50 X 30 X 15 + 20 X 5 X 15 X 2 = 25.500 lb
(t,/", name!y,
(lO.l4)
m, = 661b sec'lin
Since the girders are assumed to be rigid. the stiffness {spring constant) of each story is gi ven by or by introducing the numerical values for this example, we obtain
8976&>'  lO,974,800w' + 1.36 X 10' =
(10 IS)
;;;
12(21)
and the individual values for the steel column sections indicated are thus k;
=_.~_..C'::.",,=,,=30,7001b/in
u/ wi =
140
lO82
12 X 30 X 10' X 248.6 X 2
X
12 x 30 X lOb X 106.3 X 3 k,. = ..:::....:.c:.......__._. '  ~ 44 300 Ib /in (lOX 12)" , The eqUlltions of mm!on for [he system~ which are obtained by considering In Fig. 1O,2{b) the dynamic equlHbrium of each mass in free vibration, are
m,YI
T
w: = 11.83 radJsec
Wl
= 32,89 radisec
k'YI  k?(y~
y,)
=0
0
292
293
T,
= 1 = 0.532 sec
t,
tl
= 0.191 sec
To solve eq. (1012) for the ampHtudes at and Q1. we note that by eqcating [he determinant (0 zero in eq. (l0.l3), the number of indeper:denr equations is one less. Thus in the present case, the sY3tem of two equations is reduced to one independent equation. Considering the nrst equation in eq. (10.12) and
(.j
lb,
Wj
Fig. 10.3 Norm;,l modes for Example 10.1 ~a} Firs. mode. (b) Second
44,300a" = 0
We have introduced a second subindex in o! and Q2 [0 indicate (hat the va;ue Wi has been used in this equation. Since in the present case there are two unknowns and onty one equarion, we can solve eq. (10.16) only for Lhe relarive value 0:' all to all' This relative value is known as the nonnal mode or modal shape corresponding to the first frequency. For this example eq. (to. i6) gives
Q2'
= L263
a"
It is customary to describe the normal modes ~y assigning a unil value ro one G:: equal to unity so that
is called a norma! or natura! made of vibrafiofL The shapes (for this example fall and O:fliad <lre caned normal mode shapes or simply modal shapes for the corresponding natural frequencies (UI and W:t. The two modes thar have been obtained for lhis example are deprcred in Fig. 10.3. \Ve often use the phrase first mode or fundamental mode to refer to the mode assocIated With the lowest frequency. The other modes are sometimes called harmonics or higher Iltmnol1ics. It is evident that (he modes of vibration, each having !ts own frequency, behave essentially as single~degree.of freedom systems. The total motion of [he system, that is, [he tola! SOlution of the equations of motion, eq. (10,1), is given by the superposition of [he modal hannonic vibrations which in tenus of arbitrary constants of integration may be wriuer. as
a:H
LOOO
(\0.17)
SimilarlY, substitlltir.g the second natural frequency, (10.12), \ve obtain the second nonnal mode as
UlZ
(10.19)
an = 1.000
a" =  1.629
(10.18)
Here C( ar:d as well as Ct\ and ((1 are four COnstants of integration [0 be detennined from fO'Jr initia: conditions which are the initial disptacemer.[ and velocity for each mass in the system, For a [wodegreeo~freedom system, these initial conditions are
y, (0;
H should be noted that aJrhough we obu:.ined only ratios, the amplitudes of motion could, of course, be found fro:n initial conditions. We have now arrived at two possible simple hannonic motions of the slructure which can take place in such a way that all [he maSses move in phase at the same frequency, either w, Of W:J.. Such g motion 0: an undt.rnped system
is convenient
[Q
294
(Ct!
Buifding
295
and U'2 in eq, (lO.l9) in faver of other cor:stanrs, Expanding the trigonometric functions and renaming the constants, we obtain
YI (/) Clall
sln
Wit
ell/ il
cos Wit
Wit
'T
C~au
sin
w}}
(r)i
+ C4Gn
+ C 4G!Z
cos ~l
COS WJI
h(/)
+ C;.a~"2 sin
(10.21)
in which C 1, C 2, C)., and C", are new consrunrs of integration, From [he first twO in:tiul eoocirion;:; in eq. (l0.20), we have (he followIng two equations:
00.22)
Since the modes are independent, these equmions can always be soived for C z and Similarly. by expressing in eq. (10.2 I) the velocities at time equal to
These equations are exactly rhe same as eq. (lO.i2) but written in Lilis form they may be given l.l stade incerpre(arion as [he equilibrium equations for the system acted on by forces of magnitude ffltw"J,a, and mzviaz appJied to masses ml and m;b respectively. The modal shapes may then be conSidered as the static defleclions resulring frorn rhe fo:ces on [he righthand side of eq, (025) for any of the twO modes. Tnts interpretation, as a static problem, alIows liS to use the resu:(s of the general smtic ~heory of linear s{ruc(Ures. In partiCUlar, we may make use of Betti's (hecrem, which states: For a structure acted upon by two systems of loads and correspond:r.g displacements, {he wor.i< done by the first system of loads rr:oving [hrough the displacements of the seccnd system is equal (Q [he work done by this second system of loads undergoing the displacements produced by the first load system. The two systems of loading and corresponding displacements which we shall consider are as fot~ lows: System I: fo:ces
zero, we find
(0.23)
twO
UJ~anm?
the system in tenns of the two modal vibrations, each proceeciog at its own frequency, completely independem of the other, the amplitudes and phases being detennined by the initial conditions.
displacemegts
ar:.
an
afiJIlJG\lG['L
10.2
or
(10.26) If the nat1Jral frequencies are differem (WI
orthogonality propeny. This property constitutes the basis of one of the rr:.OS[ a.ttractive methods for solving dynamjc problems of mulridegreeoffreedom systems, We begin by rewriting rhe equations of motion in free vibration, eq. (103), as
r= UJ2),
[.'c {a}
w'
,M] {a}
( 10.24)
which is the so~cal!ed orthogonality rela:ionshlp between modal shapes of a two degree~of~freedom system. For an ndegreeof~freedom sysre::u in which the mass marrix is diagonal, the orthogonality condition between any tWO modes i and j may be expressed as
(10.25)
'"
0,
for i ')Ai j
(10.27)
296
297
for iY<j
(10.28)
Solution: The substitution of eqs, (IO,17) and (10,18) together with the values of the masses from Example 10.1 ieto the normalization factor required in eq. (1O.30) gives
,j(136) (100),
in which {a,} and {aJ are any two modal vectors and [tvt] is the mass mat:ix of the system. As mentioned before, the amplitudes of vibration in a normal mode are only rdative values which may be scaled or nonnalized to some extent as a :natter of choice. The foilowlng is an especially convenient nonnalization for a general system: (10.29)
+ (66)(1.263)'
1.629)'
=
= "
,31 311.08
/(136)(joO)' + (66)( 
4>" =
"
which, for a system having a diagonal mass matrix. may be written as
=
1.00
0,06437,
4;11=
v
LOa
311
=0.0567
1.263
1.31
= 00813,
cf>n
 1.6287
;31L08~=
 0.0924
(10,30)
The normal modes may be conveniently arranged in the columns of a matrix :':r.own as the IT'..odal matrix of the system, For lbe general case of n degrees of freedom. the modal matrix is written as
in which ij is the nonnaIized i component of [he j modal vector. For normalIzed eigenvectors, the orthogonality condition is given by
l.p"
12"
4>'"1
?" ,
(10.34)
(.p};~M]{.p); =
L n~
dJa1 "
rp'm:
fori
for i = j
c: 0.31)
(10,35)
Ar.other orthogonality condition is obtained by writ~ng eq, (10.24) for the nonnaHzed j mode as where [4>] T is the matrix transpose of [4 and [.1'11] the mass matrix of the system. For this example of two degrees of freedom, the modal matrix is
<p
[K] {.p}; =
w; [MJ {.p};
Then premultiplying eq, (10.32) by {cf;}[ we obtain, in view of the orthogonality condition of eq. (10.31), the followIng orthogonality condition between eigenvectors:
[ J = l00813
 0.0924j
(10.36)
{4>}TrK] {.p}; =
wJ
To check the orthogonality condition, we simply substitute the normal modes from eq. (10.36) into eq. (10.35) and obtain
fOf i
'" j
fori
(10.33)
01
005671
I 0j'
00924(lo :
Example 10.2. For the tWOdegree shear building of illc.strative Example 10.1, determine the normalized modal shapes and verify the orthogonality condition between modes,
We have seen (hat to determine the natural frequencies and nonnal modes of vibration of a structural system. we have to solve an eigenvalc.e problem.
29E
299
The direct method of solution based on :he expansion of the determinant and the solution of [he characteristic equm::on is limited in practIce to systems having only a few degrees of freedom. For a system of many degrees of freedom, the algebraic and numerical work required for the solution of an eigenproblem becomes so immense as to make the direcr method impossible. However, there are many numerical methods available for the calcu!ation of eigenvalues and eigenvectors of an eigenprobtem. The discussion of these methods belongs in a mathematical text on numerical methods rather than in a tex! such as this On srructural dynamics. For our purpose we have selected, among :he various methods available for a m;merical solution of an eigenproblern, the Jacobi Method, which is an iterative method to calculate [he elgen~ values and eigenvectors of the system. The basic Jacobi solution merhod has been developed for the solution of standard eigenproblerns (i.e., [IV( being [he identity matrix), The method was proposed over a century ago and 3as been used extensively. This method can be applied to aU symmetric matrices (K] with no reslricdon on the eigenvalues. It is possible to transfom1 the generalized eigenproblem, [[K]  w' [M]] = {<P} = {o} into the standard fonn and still rr.ainta!n the symmetry required for the Jacobi Method. However, this transformation can be dispensed with by using a generalized Jacobi solution method (Bathe, K J. 1982) which operates directly on [K] and [M].
The ratio given by eq. (10.37) is known as the Rayleigh's qumiem. This quotient has the foHowiing properties: 0) It provides the eigenvalue Wj when the eigenvector {<Ph is introduced in eq. (10.37). When a vector {} different from a eigenvector is used, then eq, (10.37) provides a value (:)1 {hat lies between the smallest eigenvalue, cuf and the largest eigenvalue w~, (3) Final1y. if a vector {} that is an approximation to eigenvector {<Ph. correct to d decimais, is used. then the value of ai obtained from eq. (10.37) is accurate to 2d number of decimals as an approximation to w;'
Example 10..3. Use Rayleigh's quotiem to calculate an approximate value for the first eigenvalue of the structure in Example 10.1 beglning with t!le approximate eigenvector for the first mode {}T = {LOO L50}, then iterating using eqs. (1025) and (10.37) to converge to the eigenvalue and eigenvectOr for the first mode.
Solution: The substitution of the given vector {4>i' {LOO 1.50} and the matrices [K] and [M] into :he nume",:or of eq. (l0.37) results in
I'
42.025
10.3
RAYLEIGH'S QUOTIENT
Several iterative methods for the solution of an eigenprob!em make use of Rayleigh's quotienL Rayleigh's quotient may be obtained by premultiplying eq. (10.32) by the transpose of the modal vector {4>}J. Hence,
{1.00 LSO}
The property of the mass matrix [AI] of being positive definitely; renders the product {4>};[M] {4>h'" 0, thus, it is permissible to solve for
w;:
uF = .............
= 147.9
= {4>}'[,o/f}{P
(10.37)
= 1.00
and
Q:;.
= 1,24
{4>} T = {LOO
1.24} yields
o.
300
301
This value of fJ} is virtuaily equal to the solution w~ = i40 calculated in. Example 10.1. Anothe: popular iterative method to solve an eigenproblern, that is, for structural dynamics, to calculate nmura! frequencies and modal shapes, is. [he 'subspace ireration method. Computer programs for the solution of an eigenproblem using the Jacobi method or the s:.lbspace iteration method are included as alternative methods in [he set of educat.:onal programs accompanYlng thi.s textbook. Commercial computer program for the solurion of an eigenproblem problem such as COSMOS usually provide severa! other methods in addition to the Jacobi and subspace iteration method. The computer program COSMOS gives the user the option of solving for natural frequencies and modal shapes by any of the following methods, Jacobi, subspace iteratio:l, Lanczos, inverse iteration power, and ::omptex eigenvalue analysis.
;},' S;)'J'Jt>C;)
C.
O~OC');:.'N;
O. GOCoo,:_co
C'. CODCOf:.O:'
:'.10000a.01
:;,Mt<lS.(';O
e:;~s"'.';'.tvC5
?~.;
:
:n);:,,;
.;s;;,:
l.
10.4
The program presented in this section uses the generalized Jacobi method to .determine the natural frequencies and corresponding modal shapes for a struc~ ture modeled as a discrete system. Example 10.4. Use Program 8 to solve the eigenproblem cOtTesponding
n.1Hn
i),)2'lH
O.sno!.
fL[3$SS
10.5
[KJ =
3000  150~
 1500
01
Example 10.5, DeLennine the nan:ral frequencies and modal shapes for the threeslory shear building modeled as shown in Fig, tQ,4.
So!wion: The analysis is perfonned using three spring elements and three concentrr..ted mass elements.
(M] =
[~
0
I
~l
V:S\\
C, L
P'~R
V.:EH
VIEW,
Cl,
0,
Solution: The execution of Program 8 to calcuiate natural frequencies and modal shapes requires the previous preparation of a file containing the stiffness a;)d :nass matrices of t:Je system. This me is crez.ted during the execution of one of [he programs to model the structure or by execution of the auxiliary Program Xl. Thls program accepts as input the stiffness a:ld mass matrices of the s~rucrure and creates the flIe required to execute Program 8 In the solmion of Example 10.4, the required nIe was created by executing Program XL
PLANE, Z,
0, 1
CURV2S
0, 0,
>
CR?COR~
CRPCOR;),
:i,
0,
C,
;),
2,
G,
0,
:3,
0,
',j,
3,
0,
3C2
Free Vibrat"lon of a Shear Building CON7ROL ;,. ACTIVS AC':'SET ACTSEl', EX:;, 2 AC'TST, ?C, 4 MESHING > PARAlC1H::SH: f'CPT, 2, 2, 1
303
50 KJiO.j
10'
40 Kilo.'
L)
l
,
6'
:1() Ktin.
Fig. lOA Mu(he;nr.nical model for a threestory shear building of Example W5.
:CPT,
3,
::;,
(4) Define element group I using the SPRING element formulmion with two nodes and element group 2 using (he MASS etemem fomlUlation:
CONTR.OL
> ACTLVE: ACTSE:7 Po.C'!'SE'I', Re. 5 !1ESHING > PAR.A'!'LMSH M_I?'T, 4., 4, 1
1,
SPR=~G,
D,
0,
2, 1, 0, 0, 0, 0
0, 0, 0. 0, 0
EGROUf:'. 2, HASS,
0,
(5) Defi:lt real conSi.ftnts for spring eiemems: k = 50,000 Ib lin. 40,000 Iblil1, and 30,000 Ib/in:
?RO?S'1'S
N1";ER.GE 0,0001, 0,
0,
(9) Apply conStrtmts in all degrees of freedom at node degrees of freedom except UX at nodes 2, 3, and 4:
LCADS~BC
t.
and in all
(6) Defi:le real constnnlS for mass eJements: m ~ 25,88 lb sec:: lin, 15.53 lb sec'lin, and 10.35 lb sec' /in:
?RO?SE'1'S
DP7, 1, AL,
D.?'l' , 2, t;y,
0, 1, 1
C, ;;, 1, UZ, 8.X, RY, RZ
RCONS'1'
!?fHN1'~O?S,
0,
{7) Generate spring and mass elements with appropriate e!ement gro~p and real constants actlv<J.red:
(11) Se[ :he optIons for {he frequency analysis to extract rhree frequencies
il
CON'I'ROL ~ ACTIVE ;,. ACTSET AC':'SE':' , EG, 1 AC'!'SET, RC, 1 XESHING > ?AR3M.MSSH fo M_CR
M_CR, 1, l, l, 2, 1, 1
j
A_FREQU E:KY
A_FREQUENCY,
3,
S,
16.
lE{)S,
c,
lE06,
R_FRQUENCY
PARA."CHESH
AC'!'SET, RC, 2
tCC~,
2,
?
2,
1, 2,
1, 1
!
M_CR
CONTROL
.f..CTSE?,
ACTIVE ) ]>.CTSE:T
3
?Af<J'_'1_MSSH
Re,
:>
MESHING
::CCFL 3,
3,. L
2, L
Frequency
l 2 J
Frequen.cy (Raciisec)
2.S1B60e+Ol
S.9871e .... Ol 8.J6760e,..Cl
Ft'eC;:".lency
(cycle/sec)
Period
(seconds)
::.49471eOl
4.00848efOO
9:.Q697ge+OO 1.J3174e+Ol
1.10256e01 7,50595e02
304
305
DIS1,:31', D:;::S;:'IS'l',
LIST
>
DISLIS'r
constit~tes the ~odal matrix [<PJ of the system. It is particulady convenient to nomahze the eIgenvectors to satisfy the following condjtion:
{}TrM](};
;=1,2, .. ,n
~h",?'"
',)X
:}~"'~oc
'.:0:
~.(j0e~()t'
',>?
?;':
,f:[
,()C'?';::; . ::Je~C;; .0;:e00
G.00e C:;
M
,,1
OOeCC
C .C(j"TCO
:;.OO~<C O.~Q":) G.CJe~;'J0
S.5s",~i)~
1 . SBe:>"
0. JGe~";;
,;;;;",~:)C
;) .::::;e~v7
1
.OCe~a0 ,).OOe~C::;
0. ::H;e~00 . O(\e.~0
D.OO.:~{)D
where the normalized modal vec~ors {<Ph are obtained by dividing the components of the vector {ah by /"{a}T[iv!J {a}~. The normalized modal vectors satisfy the following important conditions of orthogontlity:
{};(M]{h=O
{h;M]{},=!
:>;e'(;'
FK
:).00>[<00
>Iode sn??'"
";y
().
(loe~C(,
:;:,:
'L(;~~O:;
;IX
llY
:LME"~O
?2 ;;
.~()"'~S'c:
for i fori=J
0DecO
~".;;~e(',:
O.0)",JC
;;,00>1>00
G.OQM00
2.';'0",02 :,91;,.n
0%
"'.C::;o;)<:
J,
OCe~C,;
,.
(; .:C""C(j
OOe~DC
D ~:e(() () .:;:;..,.::;0
J.CCe+CC
t"::>;),,, Co:.
;;":'
0.00'1100
,'!o:le sic"p'"
:';1.
3D"!>OO
ar.d
{d>};[X]{};=O
rtx
(; . (;C",,,,C;)
,w
j.GJe w ;)0
0.30900
3.66.;\:)2
C _0Cq~CS
O. CDeJC
?0]f~C() :.OiJe~OC
l\Z
C~",Oj
C. G()e .. Ct 0.:;':200
c.rteC]
G. aCe'CU
;;. DOe.O:::
C.CC?Co
C .C~",CO
C,;)0e~:!)
.,
10.6
L}i?e~:::
}%,n
{};[X] {}j
wi
Ooe~DO
G.00".]C
C. ;;C",.JJ
c. :)O~~~~
.). :;Q!!_()J
SUMMARY
and
[1]
The motion of an undamped dynamic system 1n free VIbration is governed by a homogeneous system of differential equations which in matrix notation is
[M) {y}
+ [X] (y)
= {OJ
The process of solving this system of equations leads to a homogeneous system of linear algebraic equations of the fonn
(lK]  ,} [,11]) (o}
{OJ
which mathematically ~s known as an eigenproblem. For a nontrivial solution of this problem, it is required that the determinant of the coefficients of the unknown {a} be equal to zero, that. is,
. [KjI
in which [<P) is the modal matrix of the system and rfl] is a diagonal :natrix containing the eigenvalues ul in the main diagonal. For a dynamic system with only a few degrees of freedom, the natural frequencies and modal shapes may be determined by expanding the determinant and cakulating the roots of the resulting characteristic equation. However, for a system with a large number of degrees of freedom. this direct method of so!ution becomes impracticaL It is then necessary to resort to other numerical methods wbich usually require an iteration process. Among the various methods available for the solution of an eigenproblem, we have selected the Jacobi and the Subspace Iteration Methods.
w1 (MJ
I= 0
PROBLEMS
1O~1
The roots of t,1)is equation provide the natural frequencies Wi It is then possible to solve for the unknowns {aJ, in terms of relative values. The vectors {all corresponding to the roots w~ are the modal shapes (eigenvectors) of the dynamic system, The arrangement in matrix format of the modal shapes
wr
Determine the natura: frequencies and nOf":llai modes for the twostor}' shear buiIding sho\."itn i:l Fig. PI O. L
306
307
Ib
k<;.l
lin..
FI,,,,,., "" 5
.:
i
\5"
~
t 1 '" 5GO lb/in.
Y,
/).7j>y)/,//
Fig. P]O.!.
10.2 A certain SlruC!UrC has been mou<!ied as % lhreedegreeof~rreedcm system having the numerical values i::,dicnted in Pig. ?IO,2. Delen'nine the :,.atural freque:;cies of Ihe slrUCllJ:e and the corresponding norma: modes. Check your JnSwer using computer Progrnm 8.
k I ""
500 Iblll'L
Fig. PIOA.
10.5
Consider (he uniform shear building in which rhe mass of each Door is m and
~he
stiffness of each slOry is k. Determine :he general fonn of the system of differentia! equations for a uniform shear building of N stories, Wrile a main computer program which will use Subrouiine JACOBi. of Program 8 to caku;me rhe natural frcque:,.cie!> <.tnd r:iodnl shapes for a uniform shear blli!dir.g as described in Prob:em 10.5. Express (he natural frequencies in fu:,xlion of lhe quantity kIll).
Modify the computer program required in Problem 10.6 to incorporate lhe
fil
10
~._.
II
:.:0 KIln.
!D.G
I..
10.7
10.3
Assume <l shear building model for the Frame shown in Fig. P~O.3 and deterfmne the r.a:ural frequendes and oonn3J modes.
option of seHing: Ihe ratio kIm to a value that will result in the uniform shear bu:lding having a prescr:bed fundar:,en!al natural period T sec/cycle.
10.8
Write a main compute, program thnl will use Subroutine JACOSr of Prog~am 8 to calcu!ate lhe natural frequencies and modal shapes for a shear buildir;g of
N stories in which the stiffness of each story changes linearly from k at the first !;!ory to k,'l rH [he Nih slOry. Assume equal mass at ench floor.
10.9
Modify the computer prograr:i requested in Problem to.a for the case rhat the mass in each floor vnries linearly from m at the first floor to m,. at the Nth floor.
Modify the compu(er program req:;ested in ?roblem !D.S to inco:porate the
10.10
j
Fig. PIO.3.
lQA A movab~e strucmrtt! frame is supported on rollees as shown in Fig, PIDA.
option of seuing: :he m.:lgnirude of the Ooor mass to a value that will result in
the shear buijdi~g having a prescribed fundamental natun!: period T sec/cycle.
10.11
~ i
Find the naturaf frequencies and modal shapes :or the threedegreeoffreedom shear building tn Fig. P! 0, 1 I. Use the results of Problem 10.11 to write Ihe expressions for the free vibration displacements)'\o Y2, and)'1 of ,he shear building in Fig. PIO.II :n terms of constants of integra(ion. The stiffness matrix (Xl nne the milSS matrix (MJ of u s{ruc~ure modeled .lIS a threedegreewoffreedom SYStem are given respectively by
10.12
10.13
30B
309
10.15
Solve the eigenproblem corresponding to a structure for which [he stif!'ness and ma:;s matrices are:
1263,4
0
i,052,800
~
631.68
15,792 263.4
15.192 263.200
0 1.052.800
~
0
0 631.68
0
0
[K]
=r
15.192
ISymmetric
L
and
;
15.792 256.300
263.4
1.05~.800J
r
'
3.7143
0.6429
~
7.7381
238.10 7.7381
89.286
0
0
0.6429 7.7381 3.7143
Fig. PIO.Il.
[M]=
[K]=
lot~
" .'
;210
o
210
5
5
I ISymme;ric
37143
138.1
~7~3J
8~2861
238.1
~
J' (Ib!in)
0.333
and
23.070
[M]
0
1~86
1886:
1886\ (lb, sec1 /in)
o 23.070
1886
187.200,
Oeterr::tioe the natural frequencies and corresponding modal shapes of the smJc
ture.
10.14 Solve the eigenvalue problem co;responding ness and mass matrices are
LO
:ne stjff
r
[K]
12.0000
15.0000
12.0000
! :~~~~~j
5.1430
and
[.0\1) =
0.8169 0.1286
0.1286 OJ1571
~ 0.0740]
0.1286 0.8169
,.
~.
0.0740 0.1286
311
11
Forced Motion of Shear Building
normal mo~es of vibration. We retun to the equations or marion, eq, (9.3), which for the partIcular case of a twodegreeofrreedom shear buiiding are
mlYj + (k ,
+ kl)y,
k'JJl
k:tYl = FI (r)
fl/l,V'l 
+ klYl
F2 (I)
(11.1 )
. We seek to transform this coupled system of equation into a system of mdependenl or. uncoupled equations in which each equation contains only One unknow.1 f<Jnctlon of tln)e. It is firSt necessary to express the solution in terms of the normal modes m<Jltiplied by some fnctors determining (he contribution of e~ch mode. In the case of free motIon, these factors were SInusoidal fuacl:Ons of time; i~ the prese:H case, for forced modon, they are genera! ~unct:oas of tIme which we desigai.ue as l,(f). Hence the so:ution of eg. (1 L!) IS assumed fO be of the form
(1 1.2)
In the preceding chapter, we have shown lhat the Free motion of a dynamic
system may be expressed in terms of free modal vibrallons. OUf prese;1t interest is to demonstrate that the forced motion of such a system may also be
expressed in terms of the normal modes of vibration and thal (he total response may be obtained as the superposition of the solution of independent modal
equations, In other words, our aim in this chapter is to show that [he normal r:1odes may be used to transfo:Tn the system of coupled differential equations into a set of uncoupled differential equations in which each equation contains only one dependent variable" Thus the modal superposition method reduces the
To ~et~rr:ltne the appropriate functions 2, (t) and z'J (t) that wiH uncouple eg. Lt 15 advantageous to make use of the orthogonality relations to separate the modes. The orthogonnJity relations are used by multiplying the first of eq. (,11.3) by ~II a~~ rh~ second by ali. Addition of these equations after multiplica
OJ ..J),
(IlAa)
problem of finding the response of n multidegree~off[eedom system to the determination of the response of slngledegree<offreedom systems.
aIL
we
(I lAb)
1 L1
We have shown that any free motion of a rr.ultidegreeoffreedom system may be expressed in terms of normal modes of vibration. h witi now be demonstrated that the forced motion of such a system may also be expressed in i,erms of the
310
The results. obtained in eqs. (11A) pennir a simple physical interpretation. The force (hat IS effective in exciting a mode is equal [0 the work done b t" t 1 f " y "e ~x erna. or~e ?Isplaced by the modal shape in question. From the mathemm~ leal pOint or vle~, what we h~v~ accomplished is to separate or uncouple, by a change of vnnables, (he onglnal system of differential equations. Conse
312
3,3
root of the sum of the squares of tte modal contributions (SRSS method). Thus the maximum dis.placements may be approximated by
(! LS)
and
(11.9)
m'"o' ............,..
rr"~c.s ~,
KI :::; w MI
7 I
and Kz = w~M 2> the modal spring constan[s~ and Pi (r) :::; a nF I (I) + (/2fF:. (t) and P? (I) = tl12F! (f) + <122F2 {t), the modal forces. Alternarivejy, if we r.ad u~ the previous normalizatior., eqs. ([0.29) or (1030), these equ[trions may be wntten
simply as
21 +
l2
Wlzl : ;: ; PI (r)
=
+ W;Z1
Pl
(r)
(I ;6)
P,
(11.7)
The resutts obmined by appiication of the SRSS merhod (square root of t::e sum of [he squares of modal contributions) may substantially underestimate or overestimare the rowl response when two or more modes are closety spaced, In (his case, anot::er method known as the Complete Quadratic Combination for combining modal responses to obtain the total response is recommended. The disct,;ssion of such a method is presented in Section 11.6. The rransformation ftom a system of two coupled cifferential equations, eq. (Il.l), to a set of two uncoupled dif:eremial equations, eq. {l1.6), may be extended to a system of N degrees: of freedom. For such a system, it is particularly convenient to use matrix notation. Wi~h such notation, rhe equation of motion for a h'ne,Ur system of N degrees of freedom is given by eq. (9.3; as
The solution for the uncoupled differential equations, eqs. (1 LS) or eqs. (! 1,6), may now be found by any of the methods presented in the previous chapters, In particular, Duhamel's integral provides a general solution for these equations regardless of the functions describing the forces acting on t~e struc~ ture. Also, maximum values of the response for each modai equation may readdy be obtained using available response spectra, However, the superposi~ tion of modal maximum responses presents a problem. The fact is that these modal maximum values will in general not occur Simultaneously as the transformation of coordinates, eq. (I 1.2), requires. To obviate the difficulty, it is necessary to use an approximate method. An upper limit for the max~mum response may be obtained by adding the absolute values of (he maXlmum modal conl;ibutions, that is, by SUbstituting ZI and Zl in eqs. (I L2) for the maximum modal responses [:1:\"", and Zlm,) and adding the absolute vaiuts of the terms in these equations, so lha(
YlmlV'
(I U 1)
(ll.l2)
The premultip!icalion of eq. (11.12) by the transpose of the ith modal vector,. {<ttl;, results in
(lLl3)
= ; <ttllZlm~~ i
+ !<PnZzmaJ
I 2ZZ2mu I
(11 8)
The results oblained by this method will overestimate the maximum response, Another method, which is widely accepted ar.d which gives 2. reasonable estima~e of the maximum response from these spectral vaiues, is the square
'TIle onhogonality conditions between normalized modes, eqs. (103 nand (10,33), imply that
(l ];4)
StrlJ~lJres
315
and
(! 1.15)
20'
1.2,3, " N
where the modal force P; (I) is giver. by
(I U6)
10"
(11.17)
Equation (11.16) constitutes a set of N uncoupled or independent equations of motion in terms of the modal coordinates Zi. These uncoupled equations, as may be observed, may readily be written afrer the narural frequencies to and the modal vector {}; have been determined in the solution of the correspond~ Lng eigenpmblem as presented :0 Chapter 10. Exampie 11.1. The twostory fram~ of Example 10.1 is acted upon at the floor levels by triangular impulsive forces shown in Fig. 1l.1. For this frame. detemline tht! maximum floor displacements and the max!mum shear forces in the columns.
(,)
,OJ
Fig. 11.1 Shear building wich impulsive loadings. (a} TWO!jWfY shear buUding. (b) Impulsive load:J)gs.
The 5ubstitu(ton of :hese values: into the uncoupled equations of motion, eqs. (i 1.6),
Solution: The results obtained in Examples iD.I and 10.2 for the free vibration of this frame gave [he following values for the natural frequencies and normalized modes:
0.1
T,
and
Id
0.188
0.06437, 0.08130,
1>"
 =   .. =0524
TJ O.l91 in which the modal natural periods are calculated as 211' T\ =  .. = 0.532 sec w,
From
0.1
The forces acting on the frame which are shown in Fig, 1 LI(b) may be expressed by
and
w,
F, (I)
= lO,OOO (l
 Ill d ) Ib
4.5, we obtain
in which
fd
= 0. 1 sec and
F,(t) = F,(I) =0,
fOfi>O,1 sec
122
316
317
Fo..
z!q
?'770
_ . _ = 1.18
 1281
1082.41
where Z:;,.,..u is tl':e m~mum modal response, (</>,;  <Pi _ L/) f...'1e relative modal displacement of story, (WIth <hoJ = 0), and k, the stiffness of the story, For this examp!e we have for the tirst story
" = 121, _ 12 X 30 X 10' X 248.6  "(15 X IZ)j
""I
As indicated above these maximum modal values do not occur simultaneously and therefore cannot simply be superimposed to obtain the maximum response of the system. However, an upper limit for the absolute maximum displacement may be calculated with eqs. (11.8) as
Y,~,= 10.06437 X 9.621
15,3451b/in
v" =
Vim"
9502 Ib
= ,9502' ;.
1253'  95841b
A second acceptable estimate of the maxir.lUm response ~s obtained by taking the square root of the sum of the squared moda! contributions as indicated by eqs. (U.9). For this example, we have Y'm,., = /(0.06437 X 962)' + (0.0567 X 1.441'
22,146Ib/in
= 3607
Ib
V" 0.62 in
Y'm" = .) (O.o8i3O XT62)':;' (
(al
112
(IUS)
in which k is the stiffness of the story and .1y the difference between the displacements at the two ends of the column. Since the maximum displacements calculated as jn eq. (a) may have positive or negative values, the relative
dis~
The response of a shear building to the base or foundation mot1on is conveniently obtained in terms of floor displacements relative to the base motion. For the twostory shear building of Fig. 1L2(a), which is modeled as shown in 1 L2(b). the equations of motlor: obtained by equati~g: to zero ri:e sum of forces in the free body diagrams of Fig. 11.2(c) are the folJowing:
placements of the two ends of the column. The maximum possible value for Jy could be estimated as tr.e sum of the absolute maximum dj5placements at
the ends of the columns. However, this procedure will in most cases greatly
overestimate the actual forces in the columns. The recommended procedure is first to calculate the shear forces in the columns for each mode separately and then combine these modal forces by a SUItable metbod, such as the square root of the sum of the squares of modal contributions. This procedure is based on the fact that ::nodal displacements are known with their correct reLative sign and not as absolme values.
(11.20) where Yfi = ys(t) is the displacement imposed at the foundatiot. of the structure. Expressing ~he floor displacements relative to the base motion, we have
liZ =)1, 
)1..
(1121)
318
319
where F, and
+m 1Q l1
mIG;I"+ mla il
m:G,!! r m}a~l
(l126)
S!1e.1l
Tr:e relation between rhe modal displacements ment U I, !f; is given from eqs, (l J .2) DS
" z~
(1127)
(I L22)
In practice I( is convenient to introduce a change of variables in eqs. (l1.25) such thnt the second members of these equiHions equal )11 (t), The required change of variables to accomplish this simplification is
In
(1128)
which when introduced into egs, (I! .25) gives
(11.23)
We no:'e that the righthand sides of eqs. [11,23) are proportional to the snme function of time, y:(!). This fnct leads [0 a somewhat sirn;?ler solution Com~ pared to the solution of eqs. (1\.6) which rr;r.y contain different functions of time if! each equation. For the base motion of the shear building, eqs. (11.4)
81 + w~g:! = Y,{t)
(! 1.29)
may be written as
Finally, solving for g( (/) <l:ld 81 (r) ip. [he uncoupled ec.s, (1 L29) and substituting the solution :nto eqs. (11 ,27) and (I ! ,28) give the response as
(l L30)
( 1124)
or
When the maximum modal re.sPO:1se glrnJ< and &:m;u are obtained from s,pectral charts, we may estimate the maxi;r:'Jm va!ues II rm~~ and !fJm~~ by the SRSS combir:atlon method as
(1125)
(lUI)
320
321
.~.~
~i
where y~ y, (r) ;$ the acceierarion function exciting the base of the struCture. Equations (11.34) may conveciemly be written in matrix notation as [NfJ {ii} + [K] {u} ~  (M] (I) LeI)
,.
in whtch It,,n, the rr:ass matrix:, is a diagonal matrix (for a shear building), [KJ, the stiffness n1:2..lrix, is a symmetric matrix, { I} is ;l Vector with all its elements equa! to 1, Y. = j.\ () is the applied acceleration at the foundation of the bui:d~ ing, and {u} and {it} are, respectively, the displacement and acceleration vectors relative to the mOtion of (he foundatico. As has been demonstrated, the system of differential equaiol1s (i 1.35) can be uncoupled [hrcugh the rransformmlon given by eq. (11.1 J) as
(11.36)
ilt
where [cP~ is the modal matrix obtained in the solution of cor.esponding eigenproblem l[K] . w' [I,'f}] {"'} = :O}. The substitution of eq. (11.36) into eq. (I 1.35) followed by premultip!icalion by the transpose of {he hh eigenvector, {};' (the ith modal shnpe), results in
(11.37)
The equations of motion for a;) Nsrory s~enr building .(Fig. 11.3(a)) sub~ lected to excitation motion at its base are obtawed by equatmg to zerO the sum
of fcrces shown in the free body diagrams of Fig. 11.3(b), namely
m,y, "k,(y,y,)k,(y,y,)
mJ, + k,(y,  y;)  k)(y)  Yo)
which upon introduction of onhogonalily property of the nonnalized eigenyec~ tors feqs. (11.14) and (1r.:5)1 resu![s in the modal equations
(11.38)
=0

" mNl'iN1
,. ') .rN2
kN ,(y , v  y .vI ) = 0
(11.32)
=0
(i
(11.39)
L 2, "', ,"I)
(1l.33)
results in
mlii;+kILl!kl(tlzlIl) m?ii2
mlys
=  mij,
+ k:.u'1
K.) Cu) 
f(1)
m,,,,,,
(i = 1, 2, 3, .. , N
(11.40)
kN(U",
U,y_:)
mNIY5
m,"/Y$
(l134)
322
323
then
8, + 140g, = 108,47
(l1.4 i)
ii, + lO8ig,
108,47
(bl
and
and their solution, assuming ZerO initial conditions for velocity and displace~ ment, 1s given by eqs. {4.S) as
( 1142)
where So/and SA, are, respectively, the spectral displacement and spectral accelera[ion for the jth mode. The participation facto;s Jj indicated in eqs. (11.39) or (11.40) are the factors of the excitation function y, (!) in eo... {ll.38). As presented in Chapter 8, response spectral charts are prepared as the so!ution eq. (11.38) (wlrh for vfllues of ine natuml frequency w, in the range of interest Thecefore, the spectral values obtained from these charts Sm or SA; should be multiplied as indicated in eqs. (11.41) and (I L42) by the participation factor 1~, which was omitted in the calculation of spectral vaiues.
140
cos 1 1.831)
cos 32.891)
(c)
g,(r) = 1()il2(1
i08.47
r, ')
The response !n terms of the relative motion of the stones .<:.t ~he floor levels with respect to the displacement of {he base is giver. as a f1JTIction of tlme by eq. (i 1.27), and rhe appropriare consideration of (he participation factors, as
U,(I)~
Example 11.2. Determine the response of the frame of Example 1 L 1 shown in Fig. 11.2 when it is subjected to a suddenly applied constant accel~ eration ]1 = 0.28 g at its base.
+ J.61"'XO.0924XO.lOO(lcos 32.891)
or, upon simplification, as
i<,
Solution: The nat'Jral frequencies and cOITesponding r:oIT.:J<11 modes from cc.lc'Jiations in Examples 10. J a~d 10.2 are
WI
f.!J?
= 32.89

rad Isec
.p" = 0.06437.
<{;" = 0.08130.
U, =
(dl
In this example, owing to [he simple excitation function (a constant acceler~ a[ion), it was possible to obtain a dosed SOfU(lon of the problem as a functIon
of lime. For <l complex excitation function such as the one produced by an actual ennhquake, l[ would be necessary either to resort to numerical integration to obtain the response or to use response spectra if available. The maximum modal response is obtained for [he present example when the cosine functions to eqs. (c) are equal iO 1. in this case the rr.aximum modal response is then
glrnJ< ;;;;:
y, = 0.28 X
The participation facto:s (Ire calc':Jiated from eqs. (11.39) with the denornina~ ~ors set equal to unity since (he modes are normalized. These factors are then
F,
L5S 0.20
(e)
F, =
 1.613
(n)
glrnJ<
324
Forced MOlion 01
Sneer 8uild:t1s
325
and the maximum response, calculated from the approximate formulas: (I 1.31), is
1M]
(f)
01 = [136 0 66
The possible maximum valves for 'he response calculaled from eqs. (d) by setting the cosine functions to their maximum value resuit in
U;i1\~"
ti;m4>
1.426in
1.789 in
(g)
Previous Lo the execution of Program 9, it is necessary (0 execute one of [he programs to model the S;fuCWre (for example, Program 7 [Q model the structure as <l she2.r building) followed by the execution of Program 8 to solve rhe eigenproblem. The execution of these programs will prepare the file required w execute Progr2.m 9. Alte!"p.a~e!)', when the eigensoturion is known, we would execute the auxiliary Prog:am X1 to prepare the :equired fEe. Since the ejgen~ solution in the prese:1t examp.e is known, we p:oceed to execute Program X2:
Input nata and Output Results
which for this particular exampie certainly compares very well with the approximate results obtained in egs. ([) above.
113
Program 9 calcvlates the response of a linear system by SUperposiE~on of the Solulions of the moda! equations. Before one Can use this program. it is necessary to so! ve an eigenproblem to determine the nat:Jral frequencies tnd modal shapes of the structure. The prog:am determines the response of the structure excited either by timedependent forces .appJied at nodal coordin.ates or a timedependent acceleration at the support of the structure.
C.OiEJG
.O~2~:)
0,0%70
n
!'I.iMlJ.SR
..r.
OAT",
2XT!;;:?;)",;,,,, 701<C2$
Example n.3. Use computer Program 9 to find the response of the frame analyzed i:1 Example 11,2.
Solmion," The natural freqaencies and the modal matrix for this struC[Ure as calculated in Example 10.1 are
(VIIi .83 rad Jsec
Wz
or
FORCt #. COORD
,., ;"<'IR<: FOr!':;:;;: 1$ A?JiI\.:EP. (HJ;t. \:;1" (>(',:H'tt CB7H1::;,,c 'the fORe"
G.OO l.IJO
O.2Si) O.2!lO
and
[ (ill
326
327
COORD
~,.
elSE.
1"";'1;.
'laoc,
19~
i"..AoA.
Ace.
1.405 1. na
a.
,0. oS"
m,
Example 11.4.
Soiwion:
Input Data and Output Results
1
rm~~ NE'~2
>,
U . OJ,
C~C
GRAV!1'h1'IOtl';L HiQ;{
force F F'J sla Wr which is i.:pplied at ~he level of the second floor. In this = Fo s:n Wf become case eqs. (l L I) With FI (I) 0 and
r;
.OOM
OQOO"v(lO~
I). tO~o
c.oooo
{l.jCCOl
J.4000
Y::
= y.~ sin WI
(11.44)
!;:XC!1'>.rtON
3.0000
J.!OOO
C WOO
2.000C.:HiCO
A:rer substitution of eqs. (l1.44) inro eqs_ (11,43) and cancellation of [he commOn factor si;1 wr. we then obttin
(k, + k, 
v.coco
:l.onno
m,w') Y,
k,Y,
=0
(11.45)
:'!AX.
!HS?t...
tv.X. Vi.OC.
?64.33
~J. ~851
H.:'>.x. Ace,
l1>'7.{10C
c.snn
o. i5t\$~
11.4
which is a sysrem of two equations in two unknowns, Y I > and Yl  This system always has a un:que SOlutiOn except in [he case when the determinant formed by the coe:ficiems of [he unknowns is equal :0 zerO, The reader should remember [hat in this case the forced frequency w would equal one of the natu!'al frequencfes, since this de(erminam when equated co zero is precisely [he condition used for determining t:te natural frequencies. In other words, unless the structure is forced to vibra(e at one of the resonam frequencies, the algebraic system 0: eqs. (l L43) has ~ unique sol:Jtion for YJ, and Y::. Example 11.5. De(ermine the steadystate respor:se of the twoStory shear LO,OOO sin 201 is appHed to [:te buildir:g of Example 10.1 when a force F?(r) second story as shown in Fig_ [lA,
When :he excitation, that is, (he external forces or base motion, is ha:r.uook (sine or cosine function), the analysis is quite simple and [he response can readHy be found without the use of modd analysis. Let us consider [he two'sOry shear building as shown in 11.4 subjected 0 a single harmonic
328
329
Solution: The natural frequencies for this frame were dete:mined in Example 10, I to be
fiJi
_____
m,'~_;>+
V,
= II,83
rad/sec
F;{tl
fJ)2
= 32.89 rad/sec
Since the forcing frequency is 20 rad/sec, the system is not at reSonance. The steadystate response is then given by solving eqs. (11 A5) for Y! and Y::. Substituting numerical values in this system of equations, we have (75,000  136 X 20') Y,  44,300 Y, = 0  44,300 Y
+ (44,300  66 X 20') Y,
10,000
44), the
steady~state
response is
(b)
Damping may be considered in the analysis by simply including dampjr,g elements in the ~nodel as is shO\~ln in Fig, 11.5 for a twostory shear building. The equations of motion which a:e obtained by equatir.g to zero the sum of the forces in the free body diagram sbown in Fig. j L.5(c) are
:t
Ie)
Fig, 11.5 (a) Damped shear Ol!ilding with harmonic load. (b) Multidegree model, eel Free body diagram,
mass~$pring
m,Y, + (0,
+ (k, + k,)y,
(ll ,46)
= cos W[
+ i sin
rut,
Now, considering [be general case of applied forces of the fo!T.l given by (11.47) we, conveniently, express such force in complex fonn as
(l1.48)
Real {(Fe
iF;)e'Wt}
Real {(F"
cos
!F;)(cos
[;J{ 7
i sin Wi)}
= F"
wt + Fs sin
lu(
(11.49)
wbet, is the expression in eq. (l1.47). Assuming that the forces FI (I) and F~(t) in eg. (J 1.46) are in the fonn giver. by eg, (11.47), we substitute eg, (l L48) into eq, (l 1.46) :0 obtain
mlil
with the tacit understanding tbat oniy the real part of the eq. (11.48) is the applied force. Vile show that [he real part of the complex force in eq. (11.48)
~ cJ:  k;:)';
(11.50)
330
331
The solution o~ the complex system of eqs. (11.50) will, in general, be of the form Y{i) = (Yo: + iY,;)e''&':. Since on!y the re.al parts cf::he fcrces in eqs. (1 LSO) are applied, the sOlution is also only the real part of y (i). Then, analogously to ego (1 1.49).
(1151) For the stea.dy~s~ate response of eqs. (11.50), we seek so;utlor!s 1!1 the form of
Solwiofl:
3071bseclin
(a)
The substitution of numerical values for this example into eqs. (i J .53) gives the following system of equations: (20,600 + 15.000i) Y,  (44,300 + 8860,) Y, = 0
Y\ = Yle'~
Y2::::
yl/,j.;
(11.52)
 (44,300
8860i) Y,
(b)
y, = 00006814 + 0.26865;
Y,
=
0.06309
+ O. 137775i
into eqs. (l: .50) and leteing Yl Yd + i Y,I and Yz ;;;;: Yc:; + i Yib reselts 1:1 the foEowing syS(em of complex algeb:aic equfI[ions:
Therefo::e, by (he re:2.tio:1 established in eq. (j [.54), [he steadystare response is given by y, (r)
(l1.53)
AS aiready stated, the response is (hen ~ound 'JY solving !he complex system of equations (11,53) and considering only the real part of the solution. Hence, from eq. (11.51), we have
y,
+ 3.144) in
(Ans.)
YI (t)
wt
(11.54)
Y: (fJ
wr 
Ys1 sin Wi
in wpjch YI = Y + iYs !, Y2 Yel + iYs:; is the solution of the complex equations. ' t l (11.53), The necessary calculations are better explained lhrough the use of a !1ur:1ericai example.
Wher: these rest.:lrs a;e compa;ed with those obmined fo~ the undamped Struc~ lure in Exampie 11.5, we nOie only a smail change in the amplitL.de of motion. This is at ways the case for systems ligh(1y damped and subjected to harmonic excitation of a frequency [hal is not close to one of the natt.:ftll frequencies of the system. For (his example. the forced frequency Wj = 20 rod/sec is relativelY far from the naU!;al frequenctes WI ! l "83 rad lsec Or UJz = 32.89 rad Isec which were calculated in Example : 0.1.
Example 11.6. Determine [he steadyscale response for the tWQstory shear building of Example 11.5 in which damping is considered in me analysis (Fig. 11.5). Assume for this example that the damping constants Cl and C2 are, respectively, proportional to the magnitude of spring constants k; and k'l in which the f.acmr of proportionality, ao = 0.01.
11.5
Prog:am 10 calculates the ;esponse 10 hannonic excitations of a strucrurru sys:em for wh:ch the s:if::1ess and mass matrices have been determined by one
332
333
of the programs modeling the structure (Program 7 to model the structure as a shear building). DampIng in the system is assumed to be proportional to the stiffness and lor mass coefficients, that is, the damping matrix is calculated as
y, (I) y, (I)
or
0.00068 cos 20,  0.2686 sin 20t 0.0631 eos 20r  0.1378 sin 20,
[C]=ao[M]+a,[K]
(11.55)
b which ao and at are constants specified in the input data. The program calculates the steadystate response for structures subjected ro harmonic forces applied at the nodal coordinates or a harmonic acceleration applied at the base of the structure. Example 11.7, Obtain tbe response of tbe damped twodegreeoffreedom shear building of Example 1 L6 using computer Program 10. Solution:
y:(t) =0.2686
s;n (20,+3.'44)
y,(t)
The results given by the computer, as expected. are the same as the values calcdated in Example I 1.6.
Example 11.8. For the structure modeled as a four~story shear bullding shown in Fig. 1L6, determine the steadystate response when subjected to the force FCt) = 10,000 sin 20r (1b) appl:cd at the top floor of the building. Assume dampIng proportional to stiffness (factor of proportionality al = 0.01. Modulus of elasticity E 30 X 10'" pSI.
m?""
K?AC '"
.c:.
~ .4300S~G{ ~ .4~CO.>o~
Fltl
V
V
1
I '" 7955 in.4 .......... 180 in.
m"'1
4
..........
180 in.
,
r:.
<;"S:~<;~
,
lDC:;:
4 ........
180 in.
t t
w5,
:;;:93~2
As given by the output results. of com?uter Progrz,w 10, the response for this twodegreeoffreedom system is
W
fOf
Ex<.:.mple 1:.8
334
335
D';'''i~t:'lC
S1';rrOiESS F;>CTOR
FOI\Ct:O fitCUEi'K'Y
jR.~o!sec)
G.S47H:.O~
_).27)7.01
i:.'H))(,O'l
o.OOOO1l;,QC
O.I)()002,CQ
"1.2717.02
O.OOQOR.:)()
).:n31r;:.0;:
5. S1n:::.J~ .;:.:037<:,01
D.OOOC;:'O()
<, ,:n}7~02
o. OCOCi;~OC
<L?737S.0<:
C. )(!(;05,00
Applic,:rion of the SRSS method for combining modal respo:1se generally provides a;1 acceptable eSlimation of the tOlat maximum response. However. when some of the modes are closely s~O!ced. the use of tne SRSS method :nay result in grossly underestimating or overescim::nlng [he maximum response. In 9ar::c:.!lar, large e:Tors have been founG ir: (he analysis of rhree~dimetl$ional struct'Jres in which tors:onal effects are signifko.oL The tenn "closely spaced" refe:ring to modes, intiy arbi:rarily define [he case when the difference between two no.cGral frequencieS is within lC% of the smnlles[ of [he twO frequencies. A formGlmiun known as the Complete Quadratic Combinurlofl (CQC). which is based on the lheory of random vibrations, has been proposed by KiHreghian {l98D) anc by Wilson e{ nl. (l9St). The CQC method, which may be conSidered as an eXlension of dIe SRSS method, \S give.n by the foHowing
equation:
1.C()ON:~:'!O
J.O(ji)"~CG
o.cuooe.oo
0.0'0,,02.00
'.0000.i)(; 0.0000.30
J .(lOOC';,OO
(;.000).00
0.0000:::.00
!..tooo.>oo
0.3000:;:.00 0.0000<:.00
iLoCCOE.QO
D.CI}(HH:'CO
o.ocoo.oo
1. ()OOOi.:.OQ
(l157)
COORD.
FC COVol'Otl:4T
,
ljOCll
(1 1.58)
Fe COKf'CNZNT
where r;;: r.vi!w, is lhe rntio of the narural frequencies or order i and j and ~i and ; the corresponding damping 1::0:; for modes j and j. For constao[ modal d4tmping [', eq. (: 1.58) reduces to
2.6qn:'OI 1.1"70,,.01.
.i.7ll211;.Ol 3.215H:.Ol
:'LO)5".(11
2 .lO{t:,c'l
~.4qlE.Oj
"S'Jse.Ol
,')
S'(; +r)r,n
~..,,~~
...
(l1.59)
11.6
The sqJare rOOt of the sum of squared comribctioflS (SRSS), to estimate rhe totnl response from calculated maximum modal values, may be expressed in general, from eq. (J j.41) or eq. (l : .42), as
It is important to note (ba(, for i eq. 01.58) or eq. (1 i.59) yields Pij= 1 for any value of [he damping ratio, including ~::::: O. Thus, for an undamped strJcture, the CQC melhod (eq, (l 1.57)) ~s idemicaJ to {he SRSS method (eq. (1;.56)]
(: 156)
where R is the estima{ed response (force, displaceme:lt, etc.) 3r n specified coordinate and R; is the conesponding maximum response of the i[h mode at that coordtnate
11.7
I
i
Exnmple 11.9. De,ermtne ~he In<lximum displacement at the noor levels of (he [hree~srory shear bui1dlng (Fig. ! J.7a) :)ubjcc[ed [0 impulsive triangular loads shown b fig. 11.7b, The wt;ll ~Liffness of the columns ar e<::.ch Story is
337
(7) Activate set 1 and generate three spring elements along curves I, 2, and 3:
CON'f?OL
AC~T'SET, ~
ACTIVE ;. ACTSET
1
ACI'SET, EG, 1
ReF
(8) Activate set 2 and generate three mass elements at points 2, 3, and 4:
Fig. 11,7 (a) functions"
Three~story
shear building for Example 11.9, (b) Impulsive loading i\C1'SE':', EG, 2. AC'T'SE'Y', Re, 2 MESHING > PARitl:_MESH > r:: ?T f'LPT, 2, 4, 1
(9) Merge and compress nodes:
k = 1500 lb lin and the mass at each fioor level is m Neglect damping in the structure,
= 0.3886
lb sec'lin).
Solution: The analysis is performed using three spring elements and three concentrated :nass elements.
(1) Set view to the XY plane:
VIEVJ, 0, 8, L 0
N>:3RGE,
MESHIN~
L
>
9, :,
:;:,GOO1,
>
;),
:),
NOJES
CCMPRESS
NCO~!PRESS,
(10) Apply constraints in all degrees of freedom at node L and in all degrees of freedom except [IX at nodes 2, 3, and 4:
GEOMETRY
GRID
>
?L?$E
DPT, 2,
'J':{,
PLANS, Z, :J, 1
0,
4, 1, UZ,
RX,
RY,
!<..Z
CR?CC2..:J
C?J?CORD, L
0, :::, 0, L
0, 0, 2, 0, C, 3, C, 0, 3, C, 0
0,
(4) Define element group using the SPRl:.~G element fOnT,alation whh two nodes and elemem group 2 using the ~1ASS element formulation:
PROPSETS ;.
EGROD?
c,
0
(12) Set the options for the frequency analysis to extract three frequencies using the Subspace Iteration Method with a maximum of 16 iter2.tions, and run the frequency analysis:
A..."1ALY$rs ~ FE:..E<:::; /BUC3': :> Zl"_r~rtEQU2NCY A_FREQ:.J3NCY, 3, s, :6, 0, 0, 0, ;),
(5) Define the real COnstant for spring elements: k = 1500 lb 1m:
?E\O?SETS ;. RCONS'fRCONST, L L 1, 1, 1500
1S05,
0,
1206,
0,
C.
(}, 0
= 0.3886 Ib
0, 0, 0
R_FREQUSNCY
RCONST
RCONST,
2 , 2 , 1 , 7,
3886,
(L
0,
0,
:>
LIST
:>
?:KEQLIS':'
338
f
fr.squency {cycles /.secl
<1.10C63e+OC
339
Frequency#
1 2 3
I J .33")
]." 11953e.;.02
1.23303e+Cl L78178e1'Ol
S.11010e:J2
5.512J6e02
5.54$$
(4) Define analysis lype as rnodnl time history using three flt!.turai frequencies, 500 time steps s[aning at 1=0 with a time increment of 0.00 I sec; iJSe
default values for a:l integration parameters; and reques: printout of relative disp1acemen!s and re!ative velociries:
ANALYSIS
>
R R R E E t. l l
?OST~~)yN
"
?O_ATY?S
ti[l)iII_nt~l
+_+_
t+'
PD ..},T'::'PE,
2.
3.
500,
0,
.001,
0,
0.5,
0.25,
0
3 14<1 G
time~dependent
>
force:
5 568"2"
~1,?9~,+_~
PO_CURVS PD_CUROEF
PD_CUF{'l'YP
PO_CURTY?,
AJ.'1A.LYSrS ' PD_CURDEF,
1, L
C, L
0 0, lCOO, .2, 0, ,0
PD_CURVSS
I eo .4\ S
C.:\lW!
~I+ .l~.es.s'.;!
"
JGG:'
O,JSO)
iIl.q01J;2"
&
~.5 ~5GI
TiME {sec]
Fig. 11.8 Latera! disploce.ment funcdons at the threestory building of Example J 1.9.
fPT
FPT, .:L
FPT, 4,
1, 2, 2, ), :L 4,
1 1 1
(17)
Reques~
(20) Activate XY plot infoIT:1mion for X displacement at nodes 2, 3, and 4 as a func[;on of time, and plot the displacement vs. time for these modes (see
Fig. 1:.8);
DISPLAY XY_PLO';'S) AC':'iCYPOST ACTXY?OST, 1, 'i11'1E, UX, 2, 12, L
ANALYSIS )
?D ..NRESP,
1,
2, ), 4,
PO_OUTPUT
AC':'XYPOST.
ACTXYPOS':',
2, TINE,
), '":'XME,
UX,
UX,
>
),
4,
0, 10. 1, 0,
L 0,
2N 3N
4~
8,
DISPLAY
XY PLOTS
XYPLO?S
XYPLOT, 1, 1, 1 (2!) Activule XY plot information for X velocity at nodes 2, 3, and 4 as a function of time, and viOl the velocity vs. time fOf these modes (see Fig. 11.9);
DISPLAY
~ XY_?~CTS ~
(19) Request Scan for m<l.:\imum displacement in the X direction at nodes 2, 3, and 4 from tOto 0.5 sec;
ANALySIS )
PI)~MAXMIN,
PD_HAXMIN
ACTXYPCST
PD~PREPARE
PO_PREPARE,
ANALYS~S
1 ;. POST_DYNAMIC
L TI11E, VX, 2, 12, L 0, 2N ACTXYPOS?, 2, ':'!ME, VX, 3. 10, 1, 0, IN ACTXYPOST, ), l'It1E, VX. 4, 8, I, 0, 4N
ACTXYPOST,
PD_MAXLlS1'
XYPLO~
34Q
t
829.22
667'1.3
SaSE.
3Q36.S
," ,,
L L L
X X
162i'1.Ei
E"
RRR
A AA
Bas
2S2.S1
14 ! S 2 3@33.'
65l
(i'1isec)
" s (inJsecl )
.2"
."
35,31
$1 llGl!
62$9
(\) , (il,J5QJ
4~G,
<l50!
TIME (sec)
T1ME (sec)
Fig. 11.10 La{ual acceleration fur;doos at the threestory building of Example 11.9 Fig. 11.9 Lateral veloci[y functions at the three5tory builc.:ng of Example i 1.9.
(I) Set view to the XY plane: (22) Activate XY plot information for X acceleration at nodes 2, 3, rond 4 ns a function of time, and plot the acceleration YS. time for these modes (see Fig. 11.1 0):
DISPLAY > ACTXY?OST, ACTXYi?OST, ACTXYPOST, XY_PLOTS > ACTXYPCST 1, TIME, AX, 2, 12, 1, Of 2N 2, TIME, AX, 3, 10, L 0, 3N 3, TIME, AX, 4, 8, L 0, 4N
DISPLAY ,. VIE_PAR ) VIE VIE..), 0, 0, 1, 0
, ,
GECMETRY
PL,b_I\lE, Z,
> G~ID
) PLANE
0,
GEOMETRY ) CURVES
CRPCORD
Example 11.10. For the structure modeled as a fourstory shear building shown in Fig. 11.11 a. determine the response when it is subjected to the force shown in Fig. ll.llb applied at the top level of the building. The modulus of elnSlicity is E::;.; 2.0 X 10 6 psi. Assume dampi!1g in the system proportionaJ ro the stiffness coeffiCIent (Co 0.01). The crosssectional area of the columns is 0.5 in X 5.0 in.
Solution: The analysis is perfonned using four beam eiemenrs and four concer.trated mass elemems.
CRPCORD, 1, 360, 0, 0, 360, lac, 0, 360, 180, 0 CRPCORD, 2, 360, 18C, 0, 360, 360, 0, 360, 360, C, CRPCORD, 3, 360, 36G, 0, 360, 540. 0, 360, 540, 0, CRPCORD. 4, 360, 540, 0, 360, 720, 0, 360, 728,
(4) Define element group i using the BEAM2D elemerHs, tT.2.;:enal properties, and cross~sectiDnal consrants and genera[e mesh with two nodes and one element along each Cl!rve:
?ROPSE?S
EGROIJP,
11?ROP,
>
EGROUP
BEAM2D,
EX,
L
1,
0, 0, 0,
0, 0,
0,
0,
342
t !
,
(8) Request printout of mode shapes:
F(t)(:b}
343
Piti
180'm,
, t
i
! .. 79551f'!'
ANALYSIS
P::\l:NT _OPS
PRINT_0PS,
0,
0,
0,
::.,
0,
0,
G,
0,
(9) Set (he options for [he frequency analysis to extract four natural frequencies using the Subspace Iteration Method with a maximum of 16 iterations, ar.d 11m the frequency analysis:
ANALYSIS)
'l~~' SU
Illl
m" 1
I'" 79.55 in.4
+
y,
Ib]
FREQ ISUCK
A~fREQUENCY
A_FREQ:.JENCY,
4,
5,
16,
}
0,
0,
0,
0,
iEOS,
:),
1EC6,
0,
0,
0, 0
>
f\u\lA:VSIS
FRQ /SUCl'(
R... FREQUENCY
R~FRQUENCY
FREQLIST
shea~
P;:;i?quency number'
1 2
3
Frequency {radl5<?cl
Period ( seconds}
PROPS E'I'S
> :KeONS'}'
RCONS'C,
1,
1,
8,
5,
lS9,~,
M~CR
5,
0,
0,
G,
0,
0,
a .707046CE+OO
O.245S5~5 .. OO 0.:602743:+00 0.:'306568+00
MESHING '
H_CR, L
PARA~_MSSH
}
1
<1.
1. 2,
(5) Define group 2 using mass eiemems, real constants, and assign a mass
.1 poinls 2, 3, 4, and 5:
PROPSE':'S
EGROUP,
>
(IJ) Define analYSIS type as modal time history using four natural frequen~ cies, 300 time steps st2.ning at l 0 wirh a rime increment of 0.01 sec; use default values for all inl:egration parameters; and request printout of relative displacements and reliHive velocities;
ANALYSIS
~
EGROUP
2, HASS,
e,
POST_DYN )
PD_ATYPE
0,
v,
0, 8,
0,
Q,
0,
0,
0,
0,
PD_Al'YP:,
0,
2,
4,
JOG,
O. . Cl,
0,
,05,
0.25,
0,
M_P7
:n
the X direction:
~
ANA:..rYSIS ANA;"YSrS
POST_CVN ".
PD_CURVES PO COROE?
PD~CURTYP,
0, 0
L 0, 1000, .25, 1000, .5, 0, 5, 0
NMERG,
L
>
12,
1,
>
0,0:)01.
0,
0,
PO_CURDEF, AC':'SET,
1,
1
MESHING
NODES
NCOV.PRESS
NCOHPRESS, 1. 8
(7) Apply constraints in all degrees of freedom at node 1, and in a!i degrees of freedom except UX at nodes, 2, 3, 4, and S:
:"OADSBC ) STRUCTUR..A.L
~
CPT,
i3~OOl):
5,
ex,!,
5,
= 0,
DISPL}l)l'TS
D?T
DP':',
1,
ALL,
0,
OPT, 2, Uy, 0,
L 1 5, 1, UZ,
RX,
RY,
RZ
344
345
lIllUt)
X; XI
R R R R: E I: E E l L ' :..
'\AAA S ;, Ril
5 SSS
(inJsec~)
(in)
". ~iI++t:..C_ .. 1':+ " .~.:+ !" @.S 1.2 \.I! 1
. 3
,.,
! .5
2.1
, .1
TIME (sec)
four~story
Flg, 11.13 Luteral accelenltlon functions at the fourstory building of Example I UO,
'
PD_OUT?D'r
1,
>
PD_NRS?
5, \)
PD_NRESP,
2,
3,
4,
Example 11.11. Determine using COSMOS the steadystale respOf1se for the building of Example I L 10 wheT'. it is subjected ro a harmonic force F(;) ~ 1000 sin 20t (lb) applied horizontally a the top leveL
!LDYNAlJ;IC
Solution: The following commands are implemented in COSMOS after the comma:1ds listed in Example 11.1O up to poir:t lO:
w
~
(16) Activate XY plot information for X displacement at nodes 2, 3, 4, and 5 as a functior: of time. and plot the response curves (see Fig. II, 12):
(1) Define ana:ysis tYpe as frequer:cy respor.se for an excited freque:1cy, 20 (radlsec):
.!<l!ALYSIS ) POS'r_CYN ) ?D_A?YPE ?D_,i:t.TYPE, 5<" 4. 0, 2C, 20, :.., 0,
D:::SPLAY
>
XV_PLOTS
>
ACTXYPO$'I
2, 12, 1, C;, 2N 3, 10, ::., 0, 3N 4, 8, 1, 0, 4N
ACTXYPOS?
DISP::"AY )
4, TIME, UX. 5, 6. 1, 0, 5N
XL_PLO'TS ;. XYPL07
1, ::., :.
{l2) Define (he dyr:amic force F (1) ;;;;; 1000 sin &t using the following commands:
l\1.'\l)'.LYS:S ,.
PC_CURTY?,
) PD... CURTYP
XYPLOT,
_,
(17) Activate XY plot information for X accelero.don at nodes 2, 3, 4. and 5 as a funCtlon of time. and piot [he response c;;rves (see Fig. l L 13):
CrSPLAY :> XY_?LCTS :> ACTXYPOS'T' ACTXYPOST,. TIME, AX, 2, 12, 1, O.
ACTXYPOST, ACTXYPOS'I\ ACTXY?CST, DISPLAY > XYPI.,CT, 1, 2, TIME, 3, TIME, 4, TIME,
XY~I?LOTS
.k.NALYSIS } PD_CJS\DEF,
2N
"~X,
3,
'2,
AX,
~<x,
5, } XY?::'OT
:0, 1, C, 3N 8, 1, v, 4N .0, 1, C, 5N
Fl'i'D.
5,
FX,
10000,
5,
1,
(13) Request values for [he ar:lplitudes of the response in terms of displacemenr, velocity, all::! accelen:tion at the four levels of the buildmg:
:::..,
1,
346
347
1,
1,
0,
0,
L
j
1
PD_NRSP
TABLE 11.1
(Continued)
ANALYSIS ).
POS'r~DYN
> PD_OUTPUT
PD_NRESP, L
2. 3, 4.
5
Nod" ,,t:":I$1. :.,I>i'SI<' ;) OOOCOC>(Hl iO.01l01lE.()O) y"e::i'nsi. \pl>i'se1 C,{jO(lOOE.OO ,0.ooOOE.OCI o.nCOocE.OC (O,OOOCEOO)
!q'~'s~,
y~oti'! .. on
(ph<ls~l
~r"ti'tio"
{"M~e:l
fi
= 0,
(ph"liel
O.D01l()Ot~OO
O.OO:)O(}>:.O(l
to.OOOOt~OOI
o.~&a~g.:.o) .';~56
O.NlOCCE .. OC
i!LCOOD<:~CO)
ANALYSIS
fCDYl\AMIC
(16) Use an ediror to !1St !.he results from the OUTPUT FILE. (see Table Il,"i for the steadys:ate response in lerms of dis;::iacement. velocl!y. and
.31~00t,(n
O. JOOCOS.CC 10.COOCE.COI
O. t)OOOo;.OC (J.OOOOE.i'lC!
(1.00000,>00
~O.OOOO.oC)
0.00000<:.00 IC.OCOOODI
O,OGOOO"'OO (D.OO()CE.CO)
acceleration):
1" . t ~ S2
ACCt.RA,tO"S
Hc4'i!
,,t.cdnsL
I"h"s~j
/(C"",,1. (ph"",,) 0
D?0"COd)~
z
~Ot b~
,on
11,8
SUMMARY
i.,has~l
For the solution of linear equations of matior:, we may employ either the modal superposition method of dynamic analysis or a stepoystep numerical
TABLE 11,1 Steady~State Response (Displacement, Velocity and Acceler~ alian) ror Examp~e 11.11
~'~''~'~''
1'.)'1%
:'. )Sl$
0.720&)E.O{ , . ])4H:Oll D,.:HOE.C1
1~.i'lOOOE'()OJ
[0
.OOOOf:~O())
n,ooecoo:.oo
Node
x~(
.. n$l.
"':<'In,,l.
lph"5e! 3 (lOOOO";'OO ,0,00CC[,:10) O.JOOO:lE<OC
fO.~OOOE.DC)
y~<:ot,,~,
'In
(ph~~<)
(p'rM''')
o OOOOOE~{lJ
:Q,O~OOE.OO)
OOoeCE.Oe 10.0030E.(0)
C.OJOOOE~OO
O.CCOOO(~OO
O.'CCOOEC~JO
c.n~24E+Ol
OC(lOE.CO)
O,OOOOJ~O:)
!O.COOOt.OOl
integration procedure" The modal superposition method is restricced to the anatysis of structures governed by linear systems of equations whereas the s(ep~by~step methods of nc:rr:erical integration are equaily applicable to systems With linear or nonlinear behavior. We have deferred the presentation of the ste;:;~bys~ep integration method to Chapter 20 on time history response of multidegreeoffreedom systems, In the present chapter, we have introduced fhe modal superposition method in ob~aining theresponse of 0 sheor building subjected to either force excita~ [ion or to base motton <:!nd have demonstrated that the use of the nonnal modes of free vibration for Iransforming the coordinates leads to a Set of uncoupled
348
349
differential equations. The solution of these equations may then be obtained by any of the methods presented in Part I for the singJedegreeoffreedom system. When use lS made of response spectra to determine maximum values for modal response, these valUes are usually combi ned by the square roOl of the
500010
sum of squares <SRSS) method. However, the SRSS method could seriously overestimate or underestimate the total response when some frequencies are
closely spaced, A more precise method of combining maximum valJes of the
Th~s
n:.erhod
has been strongly recommended in lieu of the SRSS method. In the particular case of harmonic excitation. the response may be obtained in closed fann by
simply SOlving a system of algebraic equations in wh~ch the unknowns are the amplitudes of the response at the various coordinates.
F;g. PILI.
PROBLEMS
11.1
Determine the response as a function of UC)e for the twostory shear building of Problem 10.1 when a cons!.o'lot fo:ce of 5000 lb is suddenly applied at the level of the second floor as shown in Fig, PI L L Bays are 15 f: apan.
m
,
F
11.2 Repeat Problem 11" 1 if the excitatlon is applied to the base of the structJre jn the form of u suddenty applied acceleration of mugnitude 05 g,
11.3 Delermine [he maximum displacement at the floor levels of the (hree~s{ory shear building [Fig, Pl1.3(a)) subjecLed to impulsive triangular loads as shown in Fig. Pi L3(b). The lOlal stiffness of the columns of each story is k = 1500 Ib/in and 2 the muss at each floor level is m "'" 0.386 [b . sec lin. Solve Problem 11.3 using Program 9. Set time increment Lir=O.Ol sec and toeal time of integration T:'en "'" 0.3 sec, Detern:.lne (he maximum shear force in the cobmns of the second story of Problem I 1.3. (Hint: Calculflte modal shear fon::es and combine concribmions using method of square root SUC1 of squares.)
F,
Y,
I"
11.4
Fig. PIU.
11.5
1l.9
11.6 Use computer Program 9 to obtain {he time history response of the (hreestory building in Pl1.6(<.I) subjected lO (he suppOn acceleratior:. ploned in Fig.
PIL6(b), Determine the response for a toeal rime of LO sec using time slep Lit"'" 0.05 sec, and modnl damping coefficient of 10% for alt [he modes ( = 30 x 10' psi).
For the Structure (shear building) shown tn Fig. P 11.8 determine (he stead>'.stare motion for the following load systems (loads in pounds):
(a)
f,
(b)
= 2000 cos
r,
I,
1L7
Find the steadYStaie response of the shear building shOwn in Fig. PI! 7 subjected to the harmonic forces indicated In the figure. Neglecr damping,
11.10
A:so load the :;;[rucn.:re simultaneously with load sy!>Lems (n) and (b) and verify :he $upCr90si(ion of reSC;1:s. For {he str,Jc[ure modeled as a focrstory shear building shown in P1UO. delermine the s:eacYSLUte respor;se when i[ is subjected to the force F 10000 sir: 20: Cb) applied at lhe tOP floor of the building. The modc!us of
11.8 Solve Problem 11,7 assuming damping coefficients proponional to the slory stiff ness, C; 0.05 K;
350
351
"'1 ".
~:I
jO'
Ac:cele:ration
~
9 C.5
CA
elasticity is 2.0:x tO~ psi. AS;iume domping in the syste:n propo;:ionat to the stiffness coef(1cienl (c)) co O.O!}.
0.2 0_2
fitl
m~
0.2
lacjil.
Ttm~
f(d(lhj
0.1
0,2 0.3
OA:
O. J
0.9
1.0
(!s~C)
~O_!
 0.2
lS0.ln.
,b,
Fig. Pl1.6.
_v,
o
C .25
Ib,
0.5
r[rec)
+\
12'
I
Fig. Pll.lO,
F, '"
2000
5;n
lrJI!!b_~z~:;Z:z~~a~z~a*f""" Y,
WI .. 3860 In
F:g. PIL7.
353
12
Damped Motion of Shear Buildings
12.1
~or ~ viscously damped shear building, such as the three.. s[ory building shown lfl Fig. 12: t, [he equations of motion obtained by suer.ming forces in the correspondlflg free body di&grarr.s are
m;'y;
+ CtY! + kIll
FI
(I)
m?"'h + c: Cy~
Y2)
= F2 (f)
(i21l
F~ It)
In the previous chapters, we de[ermlned the natural frequencies and modal shapes for undamped structures when modeled as shear buildings. We also
,.)
Fig. 12.1 (u) Damped shear bui:dir._!l, f,~l _ . Marhem",;c'; .. ~ .., medeJ' _ (c ) F ree b 0 d y diagram.
354
Damped Motion of Shear Buildings then (he coefficient of the damping term in eq. ([ 2.6) will :educe to
( 12"2)
355
{F(rll
where the matrices and vec(Qrs are as previously defined, except for [he damping matr.x (C], which is given ::;y
or alternatively as
(12"3)
z" + 2"w"z'1
(12.9)
[n the next seedon, we shall esmblish ~he conditions under which the
damped equations of monon may be t:;:ms:omH!d to an uncoup:ed set of independent equations.
in which case
M" ~ ();'[M]{4)"
K" = {};[K] {"oL
(12"lOa)
= w;M"
2(w",M",
(12.too)
C"
= {,p};[C] ("oj" =
(};{F(I)}
(l2.lOc)
(12.lOd)
122
F.,(r)
To solve ~he d:fferential equations of motion, eq. {l2.2), we seek to uncouple these equatlor.s. We, therefore. inlroduce the trnnSfOli:lation of coordinates
( 12.4)
wi1! resu:t in
where [~] is the modal matrIx obtai!1ed in the solution of lhe undamped free~vibration syste:n. The substitu~ion of eq, (12.4) and ~ts derivatives ir.ro eq,
(12"2) lends to
(12.5)
which PremultiplYlng eg. (l25) by the transpose of the nth modal vector {};, yields
:5
12.3
It ~s noted that the orthogonality property of the mouai shapes,
In the derlvat;on of ~he uncoupled damped equation (12.12), it has been assumed that the nonnal coordinate transformation, eq. 02.4), that serves !o
~nco~ple the inertial and elastic :orces also uncouples the damping forces. It
m"n (IV)
:s of mterest to consider the conditions under whiCh this uncoupling wi:! occur, ,hat is, the fO!"T:l of the damping matrix [C] to which eq" (12"8) applies" When the damping IT',atrix is of the form
causes all components except the nth mode in the first and lhird terms of eq. (12.6) to vanish. A similar reduction is asslIlned to apply to the dampi!1g te!m in eq. (i2.6), thal is, if it is assumed :hat ( 12.8)
[C]
00[,\4] + a, [K]
(12" 13)
i~ which Q" and at are arbirrary proportionaliry factors. the Orthogo!1<tlity cor,dltion will be satisfted. This m~y be der.lonStrared by applying the onhogonal
356
357
ity condition to eq. (12.13). that is, ?remultiplying both sides of this equation by the transpose of the nth mode {'}~ ar.d postmultlplying: by the modal matrix
[<P], We oblain
(};[C][4'J
ao[};[M][tP]
fro;n which
(12.18)
+ a,{J;[1Q[4'l
(12,14)
The orthogonality cenditions, eqs. 02.7), then redt.:ce eq. (l2.14) to (1>l; [C] [<Pl ~ 001 ); [M] (l, +", (4)); [K] (4>l"
Equation (12,i8) may be used to determine the C0nstams 0i for any desired values of modal damping ratios corresponding to any specified numbers of modes. Fo; example. to eva!uate these constants specifying the first fcur modal damping ratios 5j, [,' .." we may choose i J, 2. 3, 4. In tn,s case eq. (12.18) gives the following system of equations:
t
;,
~,
f ,",
W:;
w,
which shows that, when the damping matrix is of the form of eq. (12.:3), the damping forces are also uncoupled with the transfonnation eq. (12.4). How~ ever, it can be shown that there are other matrices formed frorn the m.ass and stiffness m;;.!rices which also satisfy the orrhogonality condition. In general, the damping matrix may be of the fonn
(12,15)
21
w,
OJ,
(;)~ w_~
w~ ~ w, ,
l w)
"', "',
5
,
,
7 '
7 ,
(iJ~
"'; w, J
a, " a, )
a,
(12.19)
~HQ]{a}
( 1220)
where fQ] is a squflre matrix: having different powers of the naturaL freque:l~ cies. The solution of eq. (12.20) gives the constants {a} as
(12,21)
where i can be anywhere in the range :x: < i < 00 and the summation may include as many terms as desired. The damping matrix, eq. (12.13), can obviously be obtained as a speclaJ case of eq. (l2.l5). By raking twO terms correspondbg to i 0 and i = 1 in eq. (12, i5). we obwin the damping matrix. expressed by eq. (12.13), With this fOnT! of rhe damping matrix it is possible to compute the damping coefficients necessary to provide uncoupEng of a system having any desired da;nping rarios in any specified number of modes. For any mode n, the modal camping is given by eq, (12.lOcJ, that is
Finally the damping matrix: 1s obtained after the substitution of eq, (12.2l) into
eq.
(I2, 15),
If [C] as given by eq. (12.15) is subs[ituted in the expression for C" we obtain
(12.16)
It is interesting to oeserve from eq. (12, !8) that in the special CaSe when the damping ;nar::1X is proportional to [~e mass {C} = Go [tl1](i = O}, the damping ratios are :nversely proportional to the natural frequencies; rhus the hjgher modes of the structures win be given very little dampir.g. Anaiogously, when the damping is proportionai t:c rhe stiffness matrix ([C) = {II {K}), the damping ratios are directly proponion<lJ to the corresponding natural frequencies> as can be seen from eq. (12.18) evaJumed for i 1; and in this case the higher modes of the strJcture will be very heavily damped.
Example 12.1. Determine (he absolute d<lrnping coefficients for the S!ruc~ ture presented in Example 1O.~, Assume 10% cf the crir~ca! dampi:1g for each
mode. SOlution:
From Example 10.1, we have [he fo1!owir.g infcrmarion.
Now, using eg, (10.24) (K [ <pl" ~ w~\1 {4>U ond perfonning several algebraic operations, we can show (Clough and Penzien 1975, p, 195) that the damping coefficienr associated with any mode n may be written as
(12,17)
Nawml frequencies:
w,
W~ =
358
359
Modal matrix:
Flna:!y, substi:ut:ng this :natrix into eq. (12.15) yields the damping matrix as
[100 I: .26
Mass malr]x;
(b)
I ,C]
l0
1136 0]1
42172
 1.4654]
573.5
66,' _. 3.0193
4.7556
199.3
 19931 31391
136
[M)=.
Sliffness matrix:
There is yet a second met.1od for evaleatlog the damping matrix. corres. por:ding to any set of specified medal damping ratiO. The method may be explained stnning with the relationship
(12.22)
_I
[A]=
r
L
2f,~,ld'
C
o
2f,w,M ,
0
0
26w,u,
02.23)
= 0.0:85J
O.C{)OOI146
"1=
We also calculate
in which ,he modal masses Mj, M" M} .... are equal to one if the modal matrix {<P} h(lS been normalized. H is evident that the damping matrix: (C) m.y be evaluated by postmu!tip!ying and premuillpiying eq< (12.22) by the inverse of the modal malrix [tP]l and its inverse transpose (<p]T, such that
[C]
= [4>]T[A1 [IP] ,
02.24)
'iY ,
,f)
Therefore, for any specified set of modal damping ;atlos {fl. matrix. [A] can be evaluated from eq. (12.23) and the damping mwix [C] from eq. (12.24). However. in practice, lhe inversion of the modal matrix is a large computational effort, Ir.stend, lakmg advantage of onl,ogonality properties of the. mode shapes, we cun deduce the following expression for the slstem damping matrix:
(12.25)
Then
2: a, ([Ur' [Kj)'
{1" !
0.01851 [
551.475
l67U45
 325.7381 . I
671.145]
EqUlltion (12.25) muy be obtained from the condition of orthogonality of the norma! modes given by eq. 00.35) ;!s
[I]
(12.26)
l
we obtuin [ <P}T[lvf]
(12.27) .
360
361
(12.28)
in which [A1]
:=: [M']T since the mUss rr:atrix [A1J IS a symmetric matrix. Finally, the substitution into eq. (l1.24} of eqs. (12.27) and (i2.2S) gives
0,07
0.06
~
0
Cl STEEL {
2nd mod~
[e] ~ [M1
which results 1n eg. 02.25) aher substttuting matrix fA) frorr. eg. (12.23). The damping matrix (C] obtained from eq. (l2.25) will satisfy the orthogona!ity propeny and, therefore, the damping term in the differen(ial equallon (12.2) will be uncoupled with the s[.me transformation. eq. (12.4), which serves to uncouple the inertial and etasric forces. h is of interest to note in eq. (l2.25} thal the contribution to the damping m,ltrix of each mode is proportional to the modal damping rario: thus any undamped mode wiH contribute nothing to the G.::mping ma,rix. We should mention at this point the circuf!lstances under which it will be desirable to evaiuate me elements of the camping matrix, &s eq, (12,15) or eq. (12.25). It has been stated that absolute structural damping i.s a rather difficult quantity to determine or even to estimate. However, modal damping ratios may be estim2.ted on the basis of past experience. This past experience includes laboratory determination of damping in different matenals, as well as damping values obtained from vibration tests in exisri:1g buildings and other srructures. Numerical va;ues for damping <'2.110S in structures are generz.Hy in the range of 1% to 10%. These values depend on the type of structure, marerials utiliz.ed, nonsm.lctu[.[,l e:ements, elc. They also depend on the period and [he mz.gnimde of vibratiO!1. It has also been observed that damping ratios corresponding to higher modes have increasing values. Figure l2,2 shows the values of camping ratios measured in existing buildings as reported by H. Aoyama (1980) [in Wakabayashi (1986)). It may be observed from this llgufe that experimental values for damping ScD:trer over a wide range, and that it is difficult !O give definite recommendations. The scatter observed in Fig. 11.2 is typical for experiments conducted to determine damping. On this basis, the obvious Gon~ elusion should be [hat the assumption of viscous darr.ping to represent damping does not de$cribe the real mech<.:nism 0: energy dissipation in structural dy namics. However, for analytical expediency and also because of the uncertain~ ties involved in atterr.pting other formulation of damping, we still accept the assumption of viscous damping. At this time, the best recommenda(10n that can be given in regard to d2.mpil1g is [Q use conservative values. 1% to 2% for steel buildings and 3% to 5% for reinforced concrete buildings for the fundamental frequency. and co r.ssume damping ratios for the higher modes to increase in proPOft!O!l to the natural frequencies. Thus, on the basis of giving
0,05
{o
STEEL
REINF. CONCR.
O,04~..
!
3ld
mod~
t:. o ._
0.03
0.02 0.01
.,
h. STEEL
REiNF. CONCR.
"
Cl
~:{o
Co
+01:'.....
1
1
0.1)0' ... 0
Fig. 12.2 Dam~;ng ratios measured in existing buildings. [H. Aoyama in Wakabayashi (1986).J
some coosiderarion to the type of s~ructure, we assig:i numerical values to the damping ratios in aU the modes of interest. These values are then used directly in the modal equations or [hey are used to determine the dampi:1g matrix that is needed when the dynamic response is obtained by SOme analytical method other than modal superposition, e.g., the time history response of a linear or nonlinear s.ystem. Example 12.2. Determine the dump;r.g matrix of Example ILl using the method based on eq. (12.22).
Solution:
= 11.83 rad/sec
[M; = 1 . '" .0
1"'6
66.
362
Oamped Motion of Shear Buildi.;gs The modal and mass matrices are
363
To determine [CJ. we could uSe eitr.er eq. {l2.24) or eq. (12.25). From Example 10.2, the nOnTIalized mocal matrix 1s
[>P) =
CO.064J7 00567]
r04330
(>P]
~
 0.7421
0.7228 "!
: 0.7967
lO.4330
l 0.0813
0.0924
and S.370 C 6.097
[M] =
'l
0.8169 C.1286
 0.0740
0.1286
0.8571 0.1286
 0.0740 I
C.1286] 0.8169
2f,W1M,
Then by eq. (12.24)
(Cj
1[2.366
0
Solwion: The execution of Program [l requires a file that IS prepared during t;,e execution of Program 8 to solve the eige;1problem of a structure previously modeled, Altematively, the required file is prepared by the auxiliary', Program Xl. In the solullon of Sxamp!e 12.3, the auxiliary Progrnm X2 is executed, followed by the execution of Program t 1,
Input Data and Output Results
i'ROORAH U
ONH'tfl!G HATiHX
PitTA 1'!l.
Oil
[C]=l198
NATlJRAi.. l"i<Q0FW';:ES
tC.?5.l:
0.221
0.514
L241
l~rG:;N'\ftcroRS:,
which in this case of eql:al damping ratios in all the models checks with the damping matrix obtained in Example 12,1 for the same St::UCture using eq.
0.4)10
~0.1H~
(12.15).
o.'n~
Hil'lJ'l' OATA,
12.4
Program 1 I, "DAMPING," serves to calculate the absolUlc dumping coefficients, that is, the elements of the damping matrix from known values of modal damping r2:tios. Example 12.3. Use Program : 1 to calculale the damping matrix for a structure with three degrees of freedom for which t;,e squares of the natural frequencies (eigenvall..es) are
7 AOOCl>CZ
1.
?lJ"no,
,
I
.' ,
.,
OtlH'Vf I<I;$UI.'$;
wi
L96!8,
~ = IS 3927,
wj = 60.7968
I
;
4.9211;:>01 ':<.)0)<>01
2.l03J"Dl
5.29SB~,J;
.PH'!>O ,':;310::<
f.pJSZC}
L)Ol'E~'):
364
365
12.5
SUMMARY
5 Ksec 1lin.
The most common method of taking into account the dissipation of energy in structural dynamics is to assume in the mathematical model the presence of damping forces of magnitudes that are proportional ro the relative velocity and of directions opposite to the motioo. This type of damping is known as viscous damping because it is the kind of damping that will be developed by motion in an ideal viscous fluid. The inclusion of this type of damping in the equations does not al~er the linearity of the differential equations of motion. Since the amount of darnp:ng commonly present in structural systems is relatively smali, iiS effect IS neglected in the calculation of mllurnl frequencies and mode shapes. However. [0 uncouple the damped differential equations of metlon, it is necessary to impose some restrictions on [he vaJues of damping coefficients in the sys(em, These resrrictions are of no consequence owing to the fac~ thm in practice it is easier w determine or to estimate modal damping ratios rather than absolute damping coefficients. In addition, when solving (he e~u(Hions of motion by the modal superpositIon method, only damping ratios are required. When the solution is sought by orher methods, the absolute value of the damping coefficients may be calculated from modal damping ratios by one of the twO me:hcds presenred ln this chapter.
~ K1 = Kiio.
3OC()
K 1 = 2000 KJin.
~ Y,
y,
Fig_ PI2.3.
12.6
Use Pro_gmm 7 (lnd S (Q model a.:l.d de[erf:")ine rhe natural frequencies and nomal modes tor th~ (ive~s[ory shear building shown in Fig. PI2.6: then use Program ! 1 to de:erf:")me the darr,ping ma(rix c:orresponding to an 8% damping ruti!) In ail {he modeS.
Repeat Problem 12.6 fo:: rhe follewing values of Ihe GWdal dampiGg ratios:
(,*0.20,
.;,~O.15, 6~O.IO, ;,,~O.05,
<,=0
PROBLEMS
12.1 The stiffness and mass matrices for a certain [wodegree~of~freedom Sll'.lcmre are
[K] ~ [ _ 200
12.2
12..3
Determine the dampir.g matrix for this sysrem corresponding to 20% of the critical damping fer the first mode and [0% for the second mode. Use the method :,ased on eqs. (J2.i6) and (l2.17). Repeat Problem 12.! using the merhod bosed on eqs. (12.22) and 02.25). The natural frequencies and corresponding nonnal modes (arranged in the modal matrix) for the threeslory shear buUdlng show;) ~n Fig. PI1.3 are WI = 9.3! radl sec, W2 = 20.94 rad/sec, w) = 29.00 rnd/sec. and
O.l 114
 0.1968 0.12451
[<PJ~!02117
 0.0277
0.2868
0.2333
lO.2703
0.2114]
Y,
De:.ermme the dampl:>g r.Hmix for :he system correspor.ding to damping railos
12,4 12.5
of 10% for all the modes. Repeat Problem 12.3 for dacnping ratios 0: 20% :or aU the modes. Repeat Proble[J1 12.3 for [he foUowing value of modal dampi:"l:g rar:os:
(, ~ 0.2.
K,
10.000 Kiln.
(,
0.0
Fig. P12.6.
387
13
Reduction of Dynamic Matrices
13.1
STATIC CONDENSATION
A pracrical merhoc of accomplishing [he reduction of [he stiffness matrix is to identify chose degrees of freedom to be concensed as dependent Of second~ ary degrees of feedom, and express them in terms of the remaining indepenM den! Or primary degrees of freedom. The relationsh!p between [he secondary and primary degrees of freecom is found by establishing the static relmion between the:n, hence [~e name Sialic Condensation lWelhoa (Guy an, RJ., 1965). Tr.is relationship provides the :f1eans to reduce the stiffness matrix. This method is also used in sta(~c problems to eli:ninate unwanted degrees of freedom such as the imernal degrees of freedo:n of an element used with the Finite Element Method. In order to describe lhe Swtic CondenSation Method, let us assume that thos.e (secondary) degrees of freedom to be reduced Or condensed ue arranged as the first s coord!nates, and rhe remaining (primary) degrees of freedoD are rhe hl.';;( p coordinates. With [his arrangement, the sriffness equation for (he struCture may be written using partition matrices as
(13,1)
In the discretization process, it is sometimes necessary to divide a structure into a large number of elements because of changes in geometry. loading, or material properties. When the elements are assembled for the entire structure: the number of unknown displacements, that is, the number of degrees or freedom, may be quite large. As a copsequence, the stiffness, mass, and damplog matrices will be of large dimensions, The solmlon of the corresponding eigenproblem ro determine narural frequencies and modal shapes will be difficnit and, in addition, expensive, In such cases it is desirable to reduce the size of rhese matrices in order to make the solution of tr.e eiger.problem more manageable and economical. Such reduction is referred to as condensation .. A papular method of reductiOn is the SUllie Condensation Merilod" ThIS method, t!:lough s.imple to apply, is oniy approximate and may produce relative!y large errors in the res.ults when appiied to dynamic problems. An improved method for condensatior. of dynamic problerns, which gives virtually exacl results, has recently been proposed. This method, caned the Dynamic Condensarion Merhod, will be presemed in this chapter afier the introduction of me Static Condensation Method.
366
where {YI} is the displacement vector corresponding to the s degrees of :reedom to be reduced and {Yp} ~s lhe vector corresponding to the remaining p incependem degrees of freedom. In eq, (13.1), it was assumed that the external forces were zero at lhe dependent (i.e., secondary) degrees of freedom; this assumption is no( mandatory (Gallagher, RH . 1975), but serves to simpHfy explanations withom affecting the final results. A simple multiplication of the matrices on the left side of eq. {l3.!) expands this equation into twO matrix equations, namely,
[Ku] {yJ + [K,,] {y,} = {OJ
[K,'] {y,} + (K,,] {y,} = {F,}
(13.2)
( 13.3)
where
en is the
(13.5)
Substitming eq. (13.4) and using eq. (13,5) i:i eq. (13.3) results in the reduced
(]3,6)
368
369
Equation (13.4), which expresses the static relationship between the secondary coordinates {tTl and primary coordinates {yp}. may also be written using the identity {ypl = [I] {yp} as
rfJ and [K] are precisely the transformarion matrix and the reduced stiffness matrix defined by eqs. (13.4) and (13.6). respectively. In this way, the GaussJordan and the reduced elimination process yields both the transformation matrix s;Hfncs$ rr:atrix [~. Tnere is thus no need to calculate [Ks~]  t in order to reduce L~e secondary coordinates of the system,
en
Example 13.1. Consider [he twodeg:eeof.freedom system represented by the rr.odel shown in Fg. :3.1 and use static condensation to reduce the first coordinate.
Or
{y}
= [TJ b
Solution:
p}
(13.8)
ns
where
(13.9)
Substituting eqs. (13.8) and (13.9) inlo eq. (13.1) end premultiplying by the transpose of tIl results jn
[TJT [K]
II
[TJ {y,}
[[7)'[1]1 I{O}
"l{Fp };
[0
kl2
(13.13)
or
[TJT[K]
(13.14)
(l3.1O)
thus showing that the reduced stiffness matrix [KJ can be expressed as a t~nsformation of the system stiffness matrix [K]" It may appear that the calculation of the reduced stiffness matrix (K] given by eq. (13.7) requires the inconvenient calculation of the inverse matrix [KssJ I. Flowever, tb.e practical application of the Static Condensation Method does not :equire a matrix inversion, Instead the standard GaussJ Qrdan elimination process is applied systematically on the system's stiffness matrix [K] up to the elimination of the secondary coo:dinates {y,}, At this stage of the elimination process the stiffness equation (13.1) has been reduced to
~x,
~x,
!{~}
\ {F,,}
(13.1\)
t:l:,:i2l::
Fig. 13.1 Mathematical model for a tWOdegreeoffreedom system.
370
,
,
(13.16)
V
371
We can now check eq. (13 iO) by simply performing the indicated multiplica~ (Ions, namely
This method of reducing the mass and damping matrices may be justified as follows: The potential elastic energy V and the kinetic energy KE of the s.tructure may be written, respectively, as j(yjT[K] [y) (13.19)
(13.20)
KE=1/Y}r[M](j}
which agrees with the result given in egs. (13.!4).
Ana~ogous)y, the virtual work EWo! done by the damping forces Fd ;;;;. (C] {j} cOITesponding to tl1e virtual dlsplacemenr {8y} may be expressed as
oW"
{/iy/T[C](y}
(13.21)
13.2
Introduction of
~he
In order [0 reduce the mass and the damping matrices, it is assumed that ;:he same static relationship berween the secondary and primary degrees of freedom remains valid tn the dynamic probiem. Hence the S2me rransfonnation based on s(Uric condensation for the reduction of the stiffness matrix is also used in reducing the mass and damping matrices. 1n general this method of reducing the dynamic problem is not exact and imwduces errors in (he results. The magnitude of these e!Tors depends on the relative number of degrees of freedom reduced as well as on [he specific selection of these degrees of freedom for a given structure. We consider first [he case in which the discretization of the mass results in a number of massless degrees of f;eedom_ In :his case it !$ oniy necessary to car.; out :he static condensation of the stiffness matrix and to delete from rhe mass rr.arrix the rows and COlumns corresponding to the maSSless degrees of freedo:n" The Static Condensarion Method in this case does not alter the original proble~n, and ::bus res:Jlts in an equiv:llem eigenprob1em without introducing any error. In the general case, that is, the case involving the condensation of degrees of freedom to which the discretization process has allocated mass, the reduced :nass and damping mmrices are obtained using transformations analogous to eq. (13.10). Specifi.cally, if [M] is the m:lSS marrix of the system, then the reduced mass matrix is given by
11l {Yp] KE. = !(Yp)'"[1l' [M] [1l {]i,} oW" {oYp}TI1l T IC] 11l {Yp}
V= l{ypJ'[1l [K]
([322)
03.23)
(13.24)
The respective substitution of [kJ, [Nfl. and It] from eqs. (D.IO), (l3!7j, and (13. IS) for the product of 'he 'hree cennal matrices in eqs. (13.22), (13.23), and (13.24) yield
(13.25)
( [3.26)
oWe
Thes~
( 13.27)
last three equatio:l.s express the potential energy, rhe kinetIC energy, and the vlrwal work of the damping forces in terms of the independent coordinates {Yp]. Hence the matrices [i;.;, [1\1], and lCl may be inrerp,eted, respectively, as the stiffness. mass, and damping matrices of the structure corresponding to lhe independent of freedom {yp}.
L.~reewdegreeofwfreedom
V;{]
= [1l T [M][1l
(13.17)
Find lhe natural frequencies and modal shapes for the shear building shown in Fig. i3.2; then condense the first of f:eedom and compare the reslJiling values obtained for natural frequencies and modal shapes. The stiffness of each story and the mass at each floor level are indicated i!1 [he figure.
Example 13.2.
where (T] is the transformation matrix defined by eq. 03.9)_ Ana:ogousiy, for a damped system, the reduced damping matrix is given by
SoiuriofL
[Cl=[1l T [C][1l
where (C] is the damping matrix of the system.
(13.18)
Calculation of Naturnl Frequencies and Modal Shapes: The equation of morion in free vibration for this structure is given by eq. (9.3) with the force vectOr {n = (O). namely,
(a)
372
373
W /21T, so that
j, = 0,96 cps
k, "" 10,000 lb/in.
f; = 3.18
j,
cps cps
26~,8
k1
30,000 lb!ln
okr
Fig. 13.2 Shear building for Example 13 2.
The modal shapes are then dete!TTI~ned by s'Jbstituting in tum each of the values for the natural frequencies into eq, (b), deleting a redundant equation. and solving the remaining two equations for two of the unknowns in terms of the third, As we mentioned previously, in solving for these unknow:1s it is expedient to set [he first nonzero unknown equal to one. Performing these operations. we obtain from egs. (b) and (c) the following values for the modal shapes:
where the matrices lMJ and [I<] are given, respectively, by eqs, (9,A) and (9,5), Substitt:ting the corresponding numericai values in eq. (a) yields
Y"
1.00,
= 3.91,
Yn= 1.00
Y2J = 3.338
40,000 , 10,000
Q,QooJ'
01 iY'} r0 1
y, Yo
Y" =6 It,
Condensation of Coordinate
Yn=  2.025
~ 'lo r
0,
Upon substituting Yi = Y; sin wt and canceling the factor sin wt, we obtain
r40,000  25",'
1 
10,000
r  [0,000 I ' 20,000  50",' 10,000, i  10,000 10,000 100w lY, LOJ
l Y'J' 0
' y,. f
(b)
 10,000
 10,000
 l~oooJ
10.000
wh:ch for a nontrivial solution requires that the determinant of the coefficients be equal to zero. that is,
1
o.
(d)
o
SOw'
10,000 10,000  100w'
~O
,  [0,000
20,000
10,000
[T] = [0.25 OJ
__ [ 17,500  10,0001
The expansion of this determinant rest:lts in a third degree eqeation in terms of u.i having the following roots:
and from eq, (13,9)
(.V~=36.1
II<] 
_ 10,000
10,000j
(e)
wi =4000 wi = 1664,0
(c)
1n =
l~
.0.25 O~]
(f)
374
375
(KJ.
Hence
rk' = [0,25
"
I 01.
I 40,000 I  10,000
l 0
IomO,
 10,0001
10,000
o 1 ,0,25 0 1
!O,OOO!
10,00C
0 IJ :
! 7,500
20,000
 10,000
J .0
Ii
Corresponding modal shapes are obtained from eq. {g) after substituting the numerical vaiues for w~ or w~ and. soiving (he first equalior: for Y3 with Yl :=; 1. We then obtain
0
IJ
Y" = 1.00,
Y:H = 1.56,
Y" = 1.00
Yn
,
[K] =
=
0.33
l 10,000
which checks with eqs. (e). The reduced mass matrix is calculated by 5ubsli wI:ng matrix [Tj and irs transpose from eq. (f) into eq. (13.17), so that
, [0.25 I 0 1
. jy,t
I,' Y,
(M!
= lO ~
Y" = ! ,00,
which reselts in
, _ r51 ,6 [041], 0
The condensed dynamic problem is then
Y"
1.00,
:516 .
0 IOOj h:
0:, 1 [,)" + [
17,500  iO,OOO1
 10,000
10,000
[Y'] = [OJ'
Yj,
:.0
The r.aturnl frequencies and modal shapes are cecemlined from the solUTion of the following reduced eigenproblem:
For this system of only three degrees of freedom, the reduction of one coordinate gives natural frequencies lha( compare rather well for tje first two modes (eqs. (h) and (c)]. However, experience shows that static condensation may produce large errors in the calculation of eigenvalues and eiger..vec:ors obtained from [~e reducec system" A generai :ecommendadon given by users of this method is (0 aSSUr:1e that the static condensation process results in an eigenproblem that 9rovides acceptable approxima(ions of only about a thLrd of the calculated eigenvalues (natt:ral frequencies) ap.d eigenvectors (modal shapes.).
l
Equaring to
1(0
II7,5005J.6w'
10,000
10,000:
lO.OOOlOOw')
fy,]_'Ol ,y)i\oi
(g)
Example 13.3. figure 13,3 shows a uniform fourstory shear building. For [his structu::e, determine (he foHowing: (a) the natural frequencies and corres ponding modal shapes as a fou:degreeoffreedom sysrem, (b) the na~ural frequencies and modal shapes after static condensation of coorc.ina:es Yl and YJ.
;;.
Solutioli.'
(a) Natural Frequencies and r"v1odaJ Shapes as a FourwDegreeof.Freedom System; The stiffness and the mass r:1atrices fe: this StflJCWre are :espectively
wi = 403.3
from which
(h)
2 J
0
~~
/1
cps
(K] = 32735
I
2
I
01
a
0
0 I
:1
(al
376
377
For all
s;ori~a:
EJ
k
1
r0.2280
+
Y,
0.5774 0.5774 0
0.6565 0.2280
0.5774
0.42851 0.5774
m
k
f 1
k
lo.d
,
(d)
0.6565 .
ri
(b) Natural Frequencies and Modal Shapes after Reduction to TwoDe greeof Freedom System:
To reduce coordinates y, and y~, we first, for convenience, rearrange the ness matrix in eq. (a) to have the coordinates in order Yi' Yi, Yz, y,,:
stiff~
E~rH!lp!e
:3.3.
[~m"[ :
 1
0 I 2 1
1
2 0
(e)
0 I
1:
"
and
1 0 01 o 100
1M] = 0 0 1 0
Applying GaussJordan elin:ination to the first two rows of the matrix in eq. (e) results in (bl 0 0 1 0 0 0 0  0.5  0.5
327.35
o0
0.5
 163.70 163.70
0 1J
(f)
Substituting eqs. (a) and (b) into eq. 00.3) and solving the correspondlng eigenvalue problem (using Program 8) yields
163,70
WT = 39.48, w~
327.35,
(J)~ = 768.3,
and
(J)~ =: 1156.00
[11 ~
corresponding to the natura] :requencies
and
1,00 cps
__ [
IK]  _ 16370
(e)
(h)
I, =  = 2.88 cps
2tr
J.,>
w,
i~
= W:;
Wi!.
441 ..
cps
r1o.5
1,==5.41 cps
21T
378
379
0,25]
1.25
(i)
Exump1e 13.4. The snear building of Example l33 is subjected to an earthquake mod on at its foundation. For design purposes, use the response spectrum of Fig, 8. to (Secllon 8.4) and delermine the maximum horizonrai displacements of the strucmre at the !eve! of the floors. Solution: The panicipatior, factOr of a shear buiic.ing given by eq. (I ! AD) as
N
wil~
N stories is
;63.70
U)
; 36598
{k) (I)
(a)
where nl J is the mass at jth ncor and <P the jlh element of the nonnalized ith eigenvec:or, (a) Response Considering Four Degrees of Freedom: The subslitution into eq. (a) of !he corresponding numerical results from Example I),) gives
[119
!O
dGm. The eiger,yeClOrs for ehe fourdegreeoffreedom system are calculated for the first mode by eq. 03.8) l'.S
Y,
05 0
0,5
Y,
1
Or
0,5
y,
o
i
J 1.4380'
rO,,~ ,
05552
r, =
 1.890,
r,
05775,
r,=
0.2797,
and
r,=
0,1213
(b)
06723)
l Y4
l 0,6723 ;
(m)
0.4380
The spectra! displacemems corresponding to the values of t~e natllra] frequen~ cies of th!s butldmg [eq. (c) of Example 13,3] are obtained from the response spec<rum, Fig, 8.10, as
Sm
14.)2,
So,
3240,
SD)
1.433,
and
SCA
0.969
(e)
Y'1
y,
y~ ,:
and for the second mode
~
y; f ~ j
j
I' 0,2]90
0,4380 0.5552
1
0,3528
The maximum displacements at the floor levels relatIve to [he displacement at the base of the bt.:i:ding are ca!CUlaled using eq. (11A1), namely, (d)
to obtain
0.6723 J
l/ll1)a~~
: y, ':
~: 1
J,
0.5
0.5
I, r:
O k
~'I
6.274 rn.
illm;.v. =
t.1/h:lJ;
11.65 in,
")"m
= I 5,64 in,
and
07056'
0,6128 1
, ,
0,0464
0.7056  0,6128
= ! 7.81 tn
or
(b) Response Considering the System Reduced to Two Degrees of Freedom: The natural frequencies, calcu!med from eq. (k) in Example 13.3, are
1 y;
. Y,
~: 1
J,
I
0.3528
0,7056
~
/,
(n)
and
!21T~
1.011 cps
0.0464  0.6128
f,
~,
12 7T = 3.044 cps
(e)
380
381
Upon introducing, into eq. (a), the corresponding eigenvectors given in eqs. (m) and (n) of Example 13.3, we obtain [he participation factors
case, the equations of free motion may be wriuen in partitioned matrix form as (13.28) The substitution of {y} = {Y} sin eigenproblem
Wi!
T, T,
== 
1.884 0.492
[0
The values of spectral displacements corresponding (e) can be read from Fig. 8.10:
the frequencies in eq. (13.29) where w~ is the approximation of (he ith eigenvalue which was calculated in the preceding step of the process. To start [he process one takes an approximate or zero value for the first eigenvalue wT. The following three steps are executed to calculate the ith eigenvalue and the corresponding eigenve,cror {y}, as well as an approximation of the eigenvalue of [he next order w;+r: Step l. The approximation of is inrroduced in eq. (13.29); GaussJordan elimination of the secondary coordinates {Y,,} is then used to reduce eq. (13.29) to
SD' = 14.16
SD2
= 2.913
Use of eq. (d) gives the relative maximum displacements at the level of the floors as
w;
w;
"'"''' = j (1.884 x
",." =
= 5.864 in
j (1.884 x
14.16 x 0.4380)' + (0.4920 X 2.913 X 0.7056)' = 11.73 in 14.16 X 0.5552)' + (0.4920 14.16
X X X
",." = j (1.884 X
2.913 2.913
X X
0.0464)'
= 14.81
in
The tirst equation in eq. (13.30) can be written as
{Y,}
(13.30)
",=, =
13.3
)(1.884
0.6723)' + (0.4920
= [T.]{ Y,}
(13.31)
DYNAMIC CONDENSATION
{y}, = [T,]{Y,}
A method of reduction that may be considered an extension of the Static Condensation Method has been proposed (Paz 1984). The algorithm for this method Starts by assigning an approximate value (e.g., zero) to the first eigenvalue w~, applying dynamic condensation to the dynamic matrix of the system [Dr] = [K]  w~ [MJ, and then solving the reduced eigenproblem to detennine the fIrst and second eigenvalues wi and ~. Next, dynamic condensation [""1] to reduce the problem and is applied to the dynamic matrix [D 2 ] = [K] calculate the second and third eigenvalues, w~ and wj. The process continues in this manner, with one virtually exact eigenvalue and an approximation of the next order eigenvalue calculated at each step. The Dynamic Condensation Method requires neither matrix inversion nor series expansion. To demonstrate this fact, consider the eigenvalue probkm of a discrete strucLUral system for which it is desired [Q reduce the secondary degrees of freedom {yJ and retain the primary degrees of freedom {yp}. In [his where
(13.32)
IT)
= [it.]]
l
[I]
and
{y},=[{~'} \ {Y,} J.
(13.33)
wi
Step 2. The reduced mass matrix [MiJ and the reduced stiffness matrix [k;] are calculated as
[M,]
= [T,]T[M] [T,}
(13.34)
and
[K,]
= [D.] + w; [M.]
( 13.35)
where the transformation matrix [T,] is given by eq. (13.33) and the reduced dynamic matrix [D,] is defined in eq. 03.30).
.382
Structures Modeled as Shear BuHdlngs from which, by eqs. (13.30) and (13.33) (13.36)
383
[[Kil
1S solved [0 obtain an improved eigenvalue wi, its corresponding eigenvector {Yp};, and also an approximation for rhe ne)\;[ order eigenvalue W;il_
This threestep process may be applied iteratively, That is, (he value of wi obruined in step 3 may be used as an improvec approximate value in step i to cowip. a further improved value of in SEep 3. Experience has shown thilt one or twO such iterations will produce vInually exact eigensolU!!ons. Once an eigenvec[Qr {YpL for the reduced system is found, the ith modal shape of {he system is determined as {Y}i = [T,J{ l'ph using eq. (13.32).
IT 1= 1 I
l
o
0.5 0.01
0.5 0.5
0
1
wi
nne
[D,]=
r 327.35 l 16367
 163.67] J63.67;
Slep 2, The recuced maS$ and stiffness matrices, eqs. (13.34) Dod (13.35),
Example 13.5. Repeat Example 13.3 of Section 13.2 using the Dynamic Condensarion Method.
Soiwion: The stiffness ma[rix and the mass matrix with the coordinates in the order Yb Y::" Y2, Y~ are given, respecrively, by eqs. (e) and (b) of Example i 3.3. SubsricU[ion of these matrices Into eq. {t 3.29) results in the dynamic matrix for the sys[em:
~O.25 1.25j
and [K,]
[Dj )
1
= [D,l + wi [M ,] = l 
.,..
327.35 163.67
163. 67 1 163.67j
=
327.35
(a)
: 654.70 
Sup 3. The ,olulion of the reduced eigenproblem [[K,]  w' [M,l] {Y,} yields
{OJ
 32735
327.35 
wi
w; = 4039
and
wi = 365.98
Step 1. Assuming we have no ini[ial approximation of by setting w~:::::: 0 and substituting this value inm eq, (a):
w?
we Start step
lD,l ~ 1:327:35
0
r ,
These values for w~ and ~ may be improved by iterating the calcularl0ns. that is, by totroducing w;:::::: 40.39 imo eq. (13.29), This substitution results tn
654.70
0
654.70
~
I
(b)
r
(D.]
614.31 0
0 614.31 ,327.35
 327,35
327.35
6)4.,0
0
O!
327.35
 327.35
0
286.96
 327.35
0
0
0
0.5
... " ..... "." ...
I~
.0
0.5
327.35
163.61. 163.67 J
H;'~~J
i  163.67
[j
1
0
 174.44
~~i~l:~l
112.53 '
384
385
from which
[T,]
= l~
0533 r .0.533
0
I
Li12
 !.I 12
00
 LlI2
0 0
from which
433.27 36388
 363.88'  396.74 j
and _ : [D,] =
l
[M ,] =
and
."
[D.:d
l _162.67
_.
[M,J
1.236,
2.236,
[K,]
..
= [D,] + wi [M,] =
1
:
{Of yields :or
wi = 39.48, wi = 360.21
and corresponding eigenvectors
(e)
= 328.61
An iteration is perfonned by introducing w~ he following:
= 328.61
 327.35  327.35
r
wi
0.69 35 1
{Yp j,=l0.6171j
(d)
The same process is now applied to me second mode, starting by substitudng into eq. (13.29) the approximate eigenvalue = 360.21 calculated far the second mode in eq. (c). In this case we obtain
[
D,i =
[ n'.,
0

0
326.09 327.35 327.35
0

3~7.35
326.09 0
0  126
 327.35 0  327,35 327.35 l [DJ = :I'''_'''''''"~'_'''' .. 3 ............~: .. ,:: .. <:+~: .. 94:9 :.. ........ 0'.....'.1. .)L.., .) 327.3" _.4
I
c
294.49 0
0 294.49
 327.35
 32.86
~ ~
o
0
2.004  1004
~~~.~~~~!::~j
0.0  1004
i
I
386
from which
and
1.004 0,0 .,
[
and
[I),]
:~331.14
" 328,62]
Hence
I l
r Y, Y,
, Y
387
=: 004 0I 0(Y1
[1.004 0,0
I
[
r 0 ~7661 = \
 0,)766,
Y, ,
lO
5789 0. 0,5766 0.0  0.5766
328,62
329,88
1
!
(i)
and
,"
1659,78
The eigenvalues and eige:1VeclOrs (t). (h), and (i)l caicula[ed for (he first twO modes in this example US!:1g dynamic condensation are virtually identka! to the exact solution derermined in eqs. (c) and (d) of Example 13.3. It should be noted lhat normalizarion of the eigenvectors is not needed in eqs. (h) and (i) if the reduced vectors art normalized with respect to [he reduced mass of [he system, that (s, if a reduced eigenvector {Yp} satisfies the normalizing equation
\  05766,
Therefore, from eqs. (c), (d). and (e) we have obtained for the first two eIgenvalues
wi = 39.48
and corresponding eigeEvectOrs
and
wi = 327.35
0,5766\ 05766)
(f)
{f}, = (n {Yp)
(g)
we see that
{Y}:[M] {Y};= I
The eige!lveccors of the sysrem are then computed using eq. (l3.32) as foUmvs:
. Y, , Y j,Y"
Hence
'Y']
II
1
i
thus demonstradng that (he eigenvector {Y}; is nonnalized with respect to the mass matrix of the system [M] if {Yp} is normalized wirh respect to [M].
( Y
Y,
Y,
I
=
\0,6562
I 04283 I
,06562)
05780
13.4
f 0,2283 f 0,4283
l05780
0,6562 J
(h)
,y"
The dynamic condensation method requires the application of elementary operations, as it [s routinely done 0 solve a Unear system of algebraic equations, using [he GaussJordan elimination process. The elementa..ry operations are required (0 [fansform eq, (13.29) to rhe form given by eq. (13.30). However, the method also requires the calcuiation of the reduced mass matrix by eq, (13,34). This last equation involves the mulriplication of three matrices of dimensions
388
389
equal to the toral number of coordinates 1n the system, Thus, for a system defined with many degrees of freedom, the calculation of the reduced mass matrix [M] requires a large number of numerical operations, A modification (Paz 1989), recently proposed, obviates such large number of numerical operations. This modificatlo:l consists of calculating the reduced stiffness matrix [k} only once by simple elimination of s dispJacemer.{s in eq. (13.29) after seuir.g wi, = 0, thus making unnecessary the repented calculation of [k} for each mode using eq. (1335), Furthennore, :t also eliminates [he time consumed in calculating the reduced maSS matrix V~fj using eq. (13.34). Ir. the rnodified method, the reduced mass matrix for any mode i is calculated from eq. (1),)5) as
_ I w, _ 
Now, the reduced mass matrix (ki;] )s calculated from eq. (D.37), after substitution in this equation of [K,J from eg. (a) and [D,] from eg. (d). as 1M,J
. ]K,]  [D,ll
(I)
Ther. the reduced stiffness and mass ma!rix from egs. (a) and (f) are used ro SOlve the reduced eigenprobJem
[M,J
= , [[K]
[D;]J
wi
39.46,
w~
= 363,67
(g)
where [.K] 1s [he reduced stiffness multi;.;., already calculated, and [i\] is the dynamic matrix given in the partitioned matrix of eq, (1330). As can be seen, the modified algOrithm essentially requires, for each eigenvalue calculated. only the application of the GaussJordan process to eliminate s unknowns in a linear system of equations such as the system in eq. 03.29). Example 13.6. sation method, Repeat Example 13.5 using the modified dynamic conden
IY'] = 10433591
I Y,
,0.66424)
(h)
The eigenvector for the first mode, in tenns of rhe original four coordinates. is then obtained from eg. (13.3 i) as
Solution: The initial calculations of the modified method are the same as ,hose in Example 13.5. Thus, from Example 13.5 we have
to
(i)
The combination of eqs. (h) ar.d (i) gives the eigenvecwr for the first mode as
(a)
0.25'
1.25 J
[, l
y;
y.
Y}
0.23110
0.43359
(b)
(c)
0.58514
0.66424
(')
J,
wi = 40.39. wi = 36598
=. l
(d)
Analogously, for the second mode, we substitute = 363.67 from eq. (g) into eq. (a) of Example 13.5, to obtain the fonowing matrices after reducing the first two coordina~es;
wi
and
[1\1
[0.533
.0533 0533j
r 445.43 l 368.20
(e)
 368201
 40454 J 1248
[1',]
1.1248 1.1248
390
391
The reduced mass matrix (l~l:d is then calc'Jlated from eq. (13.37) as
13.5
PROGRAM PROBLEM
12~REDIJCTION
OF THE DYNAMIC
/'112 = "
163.671 16J67j
l
c  445.43
 368.2]1
404.54
368.20
= lO.5624
The compurer program described in this section reduces by SUllie condensation or by dynamic condens:'Hion the stiffness and mass matrices of a structurat system and solves {he reduced eigenpcoblem. The user has the option of selecting eiLher of these twO :nelhods, Example 13.7. Fo: tht four~5rory shear building shown in Fig. 13.3, use 12 (0 de(errnine the nawral frequencies and modal shapes after reducing the sYi;tem :0 coordinate:; Yl Gnu Y4 using [he following methods: (a) Sta[ic condensmion, (b) Dynamic condensi.\t:on, (c) Exact soiu~io.:l as <1 system wi(h four degrees of freedom.
P~ogram
SOhlfl'On.: The execu:ion or "Program 12: Condensation," requires the previous preparation of :1 file modeling Ihe structure, that is, a file storing the sliffness and mass matrix of the sys{em. For this ~xarnple. this file can be prepared by either cxccudng Progr:.m 7 to Madej {he structure as a shear building or execUling [he Auxiliary Program Xl, which directly Stores in a file the stiffness and mass mallict';s fram inpur data.
Then, for the second mode, [he solution of the corresponding reduced eigenproblem give::
319.41,
1.1248
iY')=i 0.61894)
~ y~
l
0.63352
o 1[
0.61894)_'
 0.63352J
[ 1.1248 1.1248
1 0.01640)
0.696181
and
0.69618
0.61894
 0.01640  0.63352 ;
,Q~~")orTDl
XltQJ,:'?C
;)S.j)CE.t:1
r.COOCQ( .QC
;171S~.(n
Q:Joao!;.OQ
)171~E.C)
,
)~ns.c}
ii.S~"102.:H
, .031)CJl:>CO
.n"llS;;:'~)
H7}'>E.C;
The results obrained for rh:s example using the modified method, although [Q the exaCt solution as those obtained in Example D,S by the direct application of the dynamic condensation method. Table 13.1 shows and compares eigenvalues calculated in Examples 13.5 3!ld 13.6 with the exact solution obcained previously in Example DJ.
l!.lOOO(.0[ COOOOE,OO
.COCOO.(lO
' . U;;OOO".,)[
OQOOO(.OC
".O(lQCOE.OO
.,00000.00
.00002.OC .OOC(HH':.OC
.,\10000;:;.010
TABLE 13.1
Comparison of Results in Examples 13.5 and 13.6 Using Dynamic Condensation and Modified Dynamic Condensation
"VIISER 0. 02GR,:ns Ot
tRZE:X:~
Eigenvalue Mode
I
SUMlER
os
i'f\l.J;"Y Coo?tHN;'.1'SS
m:...
CCIWEtGATW:. :NO;;;<;:
Exact Solu6on
Dynamic CondenSation
Error %
Modified Method
...
~~
i'f(ll1.;R!tS COO/WIn,:;,;::;:
Error %
0.05
2.42
FHEQ.
39,48 327.35
39.43
327.35
0.00
0.00
39.46
CUT!'VT
;I
R:;S'j~T::;:
319.41
392
393
2IG2tlVtW& #
J;:tQZtNALllE
:i<2Ql:ENC'{ Ie ".5.'
). Q4~
As expected, results given by the computer for Example 13.7 J.gree with calculations for the same structure obtained previously using static condensa[ion ir: Example 13.3 and dynamic condensatlon in Example U.S.
~6{)C:.Q2
13.6
SUMMARY
~UHeER
0.'2
c:sc~e2S
Ot fRze:::OM
ND~ tI(,~
'I I.
1
INOU
n;;csx"
Oll1'ptr, RSSUt.,S,
fR!SQUENCY (C.?S.)
LGJO
(L22B219
(;.517311
0.6%167
The reduction of unwanted or secondary degrees of freedom is usually accom~ plished in practice by the Stalic Condensation Method. This method consists of determining, by a parliH! Ga'Jss~Jordan elimInation, [he reduced stiffness marr:x correspondbg to the primary degrees of freedom and the transformatio:1 matrix relaring the secondary and primary degrees of freedom. The same transformation matrix IS ased in an orthogonal transformation [0 reduce the mass and damping matrices of the system. Stmic condensation introduces errors when applied 1O the so1:Jtlon s(rJctural dynamiCs problems. However, as is shown in this chapter, the appiication of the Dynamic Condensation Method s;Jbs!antiaUy reduces or eliminates these errors. FunhemlOre, the Dynamic Conde:1sarion MeLhod cO:1verges rapidly ro the exact solution when iteration is applied.
PROBLEMS
13.1 The stiffness and m:::.ss matrices of a cenain Slructu:e are given by
10
O.65n3E~C) >;n:nE~O~
 1
0 OOOOOE_OO
D.
Oi)CD()r~CC
 .J21::r/2+0) 'J,OC:)OCE+OO
O.oCOOC+OO
O.6S47}:,:+03
~,);"';;'::wC:'
. :n;nE~in
O.SS4iJlkJ)
.32731!;'0}
HA7?:::X"
O . OOQ{)Ce.JO
_)1n'll;'o~
[K]
OO:)OOlt.CQ
C, J:t7J'l;:+O)
=[ =
6  3
3 J2
2 1
J
~l
'I>f< =
l,
8J
r cO 0 0 2
OOOOI o 0 (J 0 a0 3 a
~~ss
0.0:;00:: ..>00
O.HjOOCE~O!
0,(::10002.00
j . Q('HlOCE.OO
OOOOOAlO
C.o:)OOO.OO
O. COCCJf: .. OO
,).OOO()~E~O~
0.100002+01
G.CO:liJ02~OO
O.OOOC(:\:>OO :).lCOOD .. rH
(a) "{.Is\: the S(a~i(: Concensation Method w determine the transformation matrix ur.d the reduced stIffness ,lOd mass matrices correspcnding [0 lhe e1iminatior. of lhe first ~wo degrees of freedon: (the massless deglees of freedom). (b) Determine the naturr;.l frequencies and ccrresponding normal modes for 1he
EIC<:"VALi.!ES,
13.2
reduced system. Repe<it (a) and (b) of Problem ;3.[ for u structure having stiffness matrix as indicated in IhaL problem, but mass matrix given by
656S~
v.O(HDO 0.S'(7)5
C.55tS4 O.22Wl 0
:::nJsc . ns)
OA:::a~n:L6%S4 O.S77);'~().22~O!
394
395
13,3
Determi::e the rtllH.:ra) frequencies and modal shape 0: the system in Problem 13.2 in terms of irs four original coordinates; find the errors \0 the rwo modes Obtained in Part (b) of Problem 13.2.
"tn in/sec?
i
400k
'\
1
1
13.9
L~__ . ____
!:J '"
Tim, (,"O'
0.25
Fig. PlJ.9.
Consicer the fiveslO:Y shear building of Fig. P13,4 subjected at it:> foundation 0 !he [ime.acce!er::.tion excit<:tion depicted in Fig. P13,9. C'se static conden:,'alIon of the coordinate:; YI, )':1. and )'... and determine: (a) The two natumi frequenc"I\';$ and corresponding mocal shapes of the reduced $ysrem. (b} The displacements at the tloor levels considering twO modes. [c) The shear forces in the coh.:mns 0; the struCture also cOns:ldering two modes. Solve Problem ]3,9 using (he Dynamic Condens<l[lon Method.
"The stiffne>.$ and mass
ma~rices
1949{\~) \'~
13.1Q
13.11
Fig. PlJA.
{AI] "'"
r 0.906 0.294
~K] ~ 10';0.2940.318
I
0.424] 0.176
288  3
1556'.\
6>14
(a) Deleff:)ine the transformut;on mfnrix and (he reduced stiffness and mass matrices corresponding to the star:c condensation of the coordln;:tes YI, YJ, ;;!:ld )'4 as indicated in the figure. (b) Determine the natt;ral frequencies and normal moces for tile reduced sys:em Obtained in Part (a} (c) Use rhe results of Pa~ (b) to determine the modal shapes, des.;ribed in Ihe frve original coordinates, corresponding to the two lowest frequencies. 13,5 Use the results obtained in Problem 134 to deterrnine the maximum displace" ments relative [0 tbe foundation for the struc~ure shown in Fig. Pl3A when subjected 0 an earthquake for which the response spectrum is given in Fig. of Section 8.4. 13.12
 8
304
16 644 SO,OOO,
Calcu!ate the fundamental nml;(a! frequency of the system after redcction of the first coordiJ1<.Uc by the fo:low~ng methods: (a} Static condensa,ion. (b) Dynamic condensa:.ion. Also obtain (he ~.mural t'requelicles as a [hree~degreesoffreedom system and compare results for the fundamental trequency. Repeat Problem 13,11 using the Modified Dynamic Condensation Method aod com pure results wi[h [he SOlution obtained with no con:knsatlon.
ruo
13,6
ese rhe results In Problem 13.4 to determine the maximurr. shear forces in the sl:)nes of the building i:1 Fig. Pl3A when subjec~d to an earthquake for whIch the response spectrum is giYen in Fig. 8.10 of Secrion 8.4.
Repeat Problem 13.2 using the Dynamic Condensation Method, Repeat Problem 13.4 using the DynamiC Condensation Method and compare results with the eXllC! solution.
13.7 13,8
PART III
Framed Structures Modeled as Discrete MultidegreeofFreedom Systems
14
Dynamic Analysis of Beams
.,
In this chapter. we shall swdy the dynamic behavior of struCtures designated as beams, that is, struclures that cJ.:TY loads thar nre mainly transverse to the longitudinal d~rectjon, thus producing nexural stresses and lateral displacemems, We begin by establishing the static characterisrics for a beam segment~ we ~he:l inrroduce (he dynamic effects produced by [he inertial forces, Two approximate methods are presented to take into account the inercial effect in the Structure: (1) the iumped mass method in which the distributed mass is assigr.ed to poine masses, and (2) the consistent mass method in which the assignment to point masses includes rotational effects, The latter method is consistent Wilh [he stadc elastic deflecrfons of the benm, In Chap(ers 21 and 22. rhe eXact theory for dynamics of beams considering (he elastic and inertial distriblHed properties will be presented. In these chapters, [he mathemanca: relationship belween [he exact solution and the stiffness and the ccnsisrem mass coefficients will be shown.
399
400
401
14.1
Consider a '.mifonn beam segment of crosssectional moment of inertia 1, length L. and mererial modulus of elasticity E as shown in Fig. 14.1. We shall establish the relation between static forces and moments designated as Ph P2, P;, and V! and the corresponding linear and angular displacements 8), B:!. 0:.. and 0.1 at the ends of the beam segrr.ent as indicated in Fig. 14.1. The relation thus obtained is the stiffness matrix for a beam segment The forces P; and the displacements 0; are said to he at the nodal coordinates defined for the heam
{.,
I~
segmenl
The differential equation for small transverse displacements of a beam, which is well known from elementary studies of strength of materials., is given by
(bl
M(x)
(14.1)
Idl
in which iv1 (x) is the bending moment at a section of the beam and Y IS the transverse displacement. The differential equation (]4. 1) for a unifonn !Jearn segment is equivalent to
pix)
Fig. 14.2 Static deflection cu::ves due to a t:11!r displacement at one of the nodal coordlnaxes
(14.2)
since
dM(x)
dx
and
Vex)
(14.3)
ment at noda1 coordinate j whHe 3:1 other nodal coordinates are maintained at zero displacement Figure 14.2 shows the displacement curves corresponding to a unit displacement at each one of the four nodal coordinates :or a beam segment indicating the cOlresponding stiffness coefficients. To determine the expressions for the stiffness coefficients kif> we begin by finding the equations for displaced curves shown in Fjg. 14.2. We consider the beam segment in F1g. 14. I free of loads [p (x) = 0], except for the forces P" P,. P" P, applied at the nodal coordinates. In this case, eq. (14.2) IS reduced to
=p(x)
in which p{x) is the beam load per unit length and V(x) is the shear force. We state :1rst the general definition of stiffness coefficient which is designated by k,}, that is, k'j is the force at nodal coordinate i due to 3 unit displace
(14.4)
(14.5)
(14.6)
I
Dynamic Analysis at Beams
403
402
in which C:, Ch CJ and C~ art coostanrs of inregrcrion ro be evaluated using boundilfy conditions. For example, to de(ennine {he function ifil (x) for the curve shown in Fig. 14.2(a), we make est of the following boundary conditions:
at x:;;;;;
0 yeO)
.nd
dy(O) = 0
dx
dy(L)
( 14.7)
(H }:
L y (Ll = 0
:lnd
eLt
( 148)
be l:.sed to detef!TIlne expressions for the stiffness coefficients. For example, consider lr.e bean~ in Fig. i4.2(b) \vo;ch is in equilibrium with the forces producing the displacement 02 = ; .0. For (hiS beam in the equilibrium position, we assume tho.t <I viru!a! clisp:[lcemenr equal to the deflection curve shown in Fig. 142(a) taKes place, We (hen apply the. principle of vinual work which states thm, for an elastic system in equilibrium, (he work done by rhe external forces is equal to the wo:k of the intem[l\ forces during !he virtual displacement. 1n orde; to apply lhis priilciple, we nOle that the exremul work WE is equal to l!le product of Ihe force k:l displ<lceci by 0 1 = I, that is (14; I) This work. as sUltcd a:;ove. is eqL:al to ,he work performed by the elastic forces during the vlr;~a! displacernenL Considering [he work perfol!1":ed by [he ':Jendirl.g m0f!lea:, we ob:ain for the Jnremal work
Use of these conditions in eq::L (14 ..5) and (14.6) resulrs in an algebraic sysc:~m of four equations to determIne the conSLants C j , C2 ('3. and C4 _ The 5u'JSrlmtton of [hese consw.tits lOW eq. (i4.6) resulls in the equation of the detlecled curve for the beam segme;j( in Fig. 14.1(a) U
!
('4.90)
IV,
in which t/JI (x) is used instead of y (x) to correspond to the condilion 0 1 = 1 imposed on {he beam segment. Proceeding irl analogous fushioe, we obtain for the equatio.'1s of the deflected curves in the other cases depicted in Fig. 14,2 the following eqt:ations;
["'M(.r)de
j()
(14.12)
ill which M(x) is the bending moment Jl ~ec[ion x of the beam and dB is the relative angldar disp!aceme:it of this section.
(14.9b)
For the virtual displacement t.:.nder consideralion, [he transverse deflection (0 the bending momejl( through (he clifferen{ial equation (14.1). Substitution of rhe second deriva::ive ,p;' (x) of eq (14.9b) Imo eq. (;4.1) results in
04.13)
(l4.9d)
The angular dcf1ectio,1 dO produced during (his virtual displacement lS relnted to the resulring transverse det1ectio:i of (he beam l/II (x) ':::y
Since tj;j (x) is (he de!lection corresponding to u unit displacement I, [he displacement resuhlog from an arbitrary displacement o\> is tJ;; (x) 0,_ Analogously. the deflection resulting from nodal dispiacements ~, 0), and 04 are, respect" ively, if2 {x) Oz, if1 (x) 8) and t/14 (x) Dol' Therefore, the total deflectlon y (x) at coor~ dinare x due ~o arbitrary displacements at lhe nodal coordinates of the beam segment :s given by superposition .us
at
dO d,
or
...~
cI',p, (.rl
ljJ;' (x)
dB =
,p;' (x) dx
(14.14)
(;4.iO)
The deflection equations which are given by eqs. (l4,9) and which CO!Tespond to unit displacements at the r,odal coordinates of a beam segment may
Equating lhe external virtual work w'r from eq. (t 4.11) with the internal virtual work W, from eq. (14.12) of(Of ""ng M(x) and de from eqs. (14.13) and 04.(4), respec,ively, finally gives the stiffness caefficiem as k"
J,
(14.15)
404
405
In general. any stiffness coefficient associated with beam t1exure, therefore, may be expressed as
kif
t J i'
('L
(x)
(14.16)
= k21 (51 + kZ2 o~ + k2) 0) + k74 o.~ P, :;; k31 01 + k n 02 + k33 o} + k;4 04
(14.17)
since the interchange of indices requires only an interchange of the :wo factors i' (x) and !' (x) in eq. (14.16). The equivalence of kij :;;; kji is a panicular case of Betti's theorem. but it is better known as Ma:tweU's reciprocal theorem. It should be pointed out that althoogh the deflection functions, eqs. (14.9), were obtained for a uniform bear:1, in practice they are nevertheless also used in detennining the stiffness coefficients for non!;niform beams. Considering the case of a uniform beam segment of length L and crosssectional momen[ of inertia I. we ;nay calculate any stiffness coefficjent from eq. (14.16) and the use of eqs. (l4.9). In particular, the stiff"ess coefficient k" is calculated as follows. Fro;n eq. (l4_9a), we obtain
kij;::; kp
P:,=k:.IO i +k~101+k430;+k.w(j!l.
p,l
_I k"
k"
1 s, 1
~" I
K~J
l:
(14.18)
J
(14.19)
= [k]{o}
tf"(' (x)
+L L'
6x
in whiC!1 {P} ar:d {8} are, respectively. the force and [he displacement vectors at the nodal coordin~tes of the beam element and LkJ is the bearT: element stiffness r:1lltriX. The use of eq. 04.16) in the manner shown above to determine the coefficient kl2 will result in the evaluation of all the coefficients of the stiffness ;natrix. This result for a uniform beam segment is
(
' ,
,
+ L'~ ,I i \ L + Idx L~I
12x114
P,
6
6x 1
l~:
I
l
6L
0,
6L
~~: j
4L
0,
Ii" 0,
(14.20)
or in condensed notation
IP)
= (kJ{o}
(14.21 )
6EI = ...... , L
Since the stiffness coefficient k f; is defined as the force at the nodal coordinate 1 due to unit displacement at the coordinate j, the for~es at coordinate l due to successive displacement Ob 02, 6: and 04 at the four nodal coordinates of the ream segment are given, respectively, by k, J 01> kr:. fh., k); ~> and k!4 0,_ Therefore, the total force PI at coordinate 1 resulting from these nodal displacements is obtained by the superposition of the resulting forces, that is,
406
407
7~
~..
,&
;'i'j)
,+ .+
3
&:,
G)
4~ 0
Ei"O'Nm'
L = 1m
Fig, Vt3 Cantilever beam divided imo three beam segments with numbered sy"tem nodal coordiflU(e::.
The first step in obraining the system stiffness matrix is to divide (he structure into elements. The beam in Fig. 14.3 has been divided into three elements which are numbered sequemially for idenLificarion. Tne second step is to identify the nodes or joints between elements and to number consecutlve ly those nodal coordinates [har are not constrained. The conscra:ned or fixed ncdal coordinates are [he laS[ to be labeled, All {he fixed coordinates may be given (he same label as shown in Fig. 14.3. In the presen[ case, we consider only two possible displacemems at each node, a vertical defiecrion nod an angular displaceme:1t. The cantilever beam in 14,3 with its three elements ,esults in a total of six ffee nodal coordinates and twO fixed nodal coordinates, :he latter being labeled with {he number seven, The third step is to obtain sys[ematically [he stiffness marrix for each element in the system and to add [he element stiffness coefficients appropriately to obtain the system sriff:1ess ma[rix. This method of assembling the system stiffness matrix is called [he direct method. In effect, Z!.:ly stiffness coefficient k'j of ~he system may be obtained by adding togerher the corresponding stiffness coefficients associmed with [hose nodal coordinates. Thus, for example, to obtain the system stiffness coefficient k:n, ir is necessary to add [he stiffness coefficients of beam segments In and 11::. corresponding to node three. These coeffjcients afe designated as kW and kW, respectively, The upper i:1dices serve to identify the beam segrnenl, and [he lower indices to !ocme the appro?riare stiffness coefficients in the COITes~ ponding ele:nent stiffness matrices. Proceeding with (}le example in Fig. 14.3 and using eq, (14.20), we obtain the following expression for the sIiffness matrix of beam segmem &, namely
For rhe Decm segme:1t &,. the element nodal coordinc[es numbered one to fout coincide With the assignmen: of system nodal coordinates also numbered] to 4 as may be seen in Fig. 14.3. However. for the beam segments .& and & of this beam, rhe assignmem of element nodal coordinates numbered 1 to 4 does not coincide wirh the assigned system coordinates. For exam?le, for element & (he assigned system coordinates ~s seen in Fig. 14.3 <lre 7, 7, [, 2; for eleme:u lA, 3, 4, 5, 6. In [he ?rocess of assembling the sysrem sriffness, coefficiems for element wlll be correctly a\locU[ed to coordinates I, 2, 3, 4: for ele:ne.'1t .& to coordinates 7, 7, 1,2; and for element &" [0 coordinates 3, 4, 5, 6. A simple way to indicate (his aHocatio~l of coordinates, when working by hand, is :0 write m the lOp and on the right of the elemenr stiffness ma~rix [he coordimne nu:nbers CO:TcSfxmdi:1g [0 the system nodal coord mares for the element as it is indicated in eq, (14.22) for element &. The stiffness matrices for elemems .& and fA with the corresponding indication of $ys~em nodal coordinates are, respectivelY,
m.
12
rk(!)]
=
6
4
12
6
2 6
7
7 (14.23)
i 0'
6 6
 12 6 2
12
6
:J
6 6
2
4
and
,:'''1 
'"~ [
12
6
4
5 12
3
4
6 6
12 6
12 6
(l4c24)
2
12
3
Cc
6  [2
6
(422)
 6 2
12 6
Proceeding systematically [0 assemble the system stiffness matrix, we [ra05late each entry in (he element stiffness matrices, eqs. (l~.22), (14.23), and (14,24), to the appropriate location in the system stiffness matrix. For instance, l2 X 10" should be translated to the stiffness coefficienr for element .ill, location at row 3 and column 5 since these are the coordinares indicated at right and top of matrix eq. (14.24) for this entry. Every element stiffness coefficient translated [0 its appropriate location in the system stiffness mmrix IS added to the mher coeHiciems accumu~rHed at that loca!ton. The stiffness coefficients corresponding w columns or rows carrying a labei of a fixed system nodal coordinate {seven in the present example) are simply disregarded since :he constrained nodal coordinates are not unknown quantities. The as~
kW
408
409
sernblage of the system matrix in the manner described results in a 6 X 6 matrix, namely
this example
!''''.:........1
i11A
24
0
0  12
6
2
0 0
Uniform
nil =,
8
2
0
6
14
[kJ = 10'
12 6
0 12 0
8
6
0
0
6
12
6 2
(14.25)
 12
6 2
6
4
6
Equation (14.25) is rhus the system stiffness matrix for the cantilever beam shown in Fig, 143 which has been segmented into three elements. As su:::h, the system stiffness matrix relates the forces and the displacements at the nodal system coordinates in the same manner as the element stiffness matrix relates forces and displacements at the element nodal coordinates.
Ai
~I
Im,xl
'
'8
r'i
14.3 INERTIAL PROPERTIESLUMPED MASS
J
I
L
Genera!
The simplest method for considering the inertial properties for a dynamic system is to assume that the mass of the structure is lumped at the noda1 coordinates where translational displacements are defined, hence the name lumped mass method. The usual procedure is to distribute the mass of each eJement to the nodes of the element. This distribution of the mass is determined by statics. Figure 14.4 shows, for beam segments of length Land distributed mass m(x) per unit of length, the nodal aHocation for uniform, triangular. and general mass distribution along the beam segment. The assemblage of the mass mauix for the entire structure wUl be a simple matter of adding the contributions of lumped masses at the nodal coordinates defined as translations. In this method, the inertial effect associated with any :otational degree of freedom is usually assumed to be zero, although a finite value may be associated with rotational degrees of freedom by calculating the mass moment of inertia of a fraction of the beam segment about the nodai points" For exampie, for a uniform beam, this calculation would result in determining the mass :nomem of inertia of half of the beam segment about each node, that is
Fig. 14..4 Lumped masses for beam segments with distributed mass. shown in Fjg. 14.3 1n which onlY translational mass effects are consIdered, the mess matrix of the system would be the diagooaJ matrix, namely
[M]
rm
a
my
1
I
0
2 3 4
(14.26)
m,
in which
ml
aJ6
I5
InLl ,m~
2 :22
;)
where in is the mass per unit length along the beam. For the cantilever beam
mS=2
inLJ
410
411
tr J)
rm,
The inertial force 11 (x) per unit of length along the beam due to this acceleration is then given by
iM] ~
,",0m,
0J
II (x) ;
14.4
It is possible to evuluu[e [he mD,Si> coefficients corresponding w the nodal coordinates of a beam element by a procedure similar to the determination of element stiffness coefficients. First, we defJ ne [he mass coefficient rn!; as rhe force at nodal coordinate i due to a unit acceleration at nodal coordinate j w!1ile all Other nodal coordinates are maintained at zero acceleration. Consider the beam segment shown in Fig. 14.5(a) which !las distribuled mass m(x) per uoir of length. In the consisrent mass method, it is assumed that the deflections resulting from unit dynamic displacements a[ the nodaJ coordinares of the beam elemen[ are giVen by the same functions 1/11 (x), ifi2 (x), 1/11 (x), and ifJ4(X) of eqs, (14,9) which were obtained from static consideratlons_ If the beam segment is subjected to a unit nodal acceleration at one of the nodal coordinates, say '~> the transverSe acceleration developed along the length of the beam is given by the second derivative with respect [0 time of eq. (14.10). In this case, with 8; = 8) = 8,,::::; 0, we obmin
8:: = 1,
j, (x)
(14.29)
Now to determi:1e the mass coefficient {fin, we give to (he beam in Fig. 14.5(b) a virtual displacement cOITesponding to a unit displacement a( coordinate I, 0, = 1 and proceed (Q apply the principle of v!rtual work for an elastic system (extema! work equal to imemal virtual work). The virtual work of the external force is simply (:'4.30) since (he only external force undergoing virtual displacement is the inerrial force reaction !finD, wirh 01 = 1. The virtual work of the internal forces per unit of leng;h along [he beam segment is
(14.28)
,
~
mi.>;}
L
,..,
t=
1
t ,
and for The entire beam
(.J
('
(14.3 I)
Equating the external and internal vinuai work given, respectively, by eqs. (14.30) and 04.31) rescirs in
ll/IZ=
(bJ
L
"
(14.32)
Fig. 14.5 (a) Beam element wirh dimiDuled mass shOWing four aod,1 coordinates. (b) Beam element supporting inenial toad due to ;)cceleration 5z "'" t. undergoing vinuai displacement OJ = L
which is the expression for (he consistent mass coefficient mn. In general, a consistent mass coefficiem may be caiculated from
m" =
L""
(14.33)
412
413
It may be seen from eq, (14.33) that mij = mj; since the interchange of the subindices only results in an interchange of the order of the factors ;,fI, (x) and if'; (x) under the integral. In practice, the cubic equations {14.9) are used ir. calcu!a:ir.g the mass coefficients of any straight beam elemenL For the special case of the beam with uniformly distributed mass, the use of eq. (I433) gives the foHow:ng relation betweer: inertial forces and acceleration at the nodal coordinates:
7~
210 Kg
2HJ Kg
I )
r P,
P2
156
54
13L
8,
3:2
0;.
( 14.34)
210 Kg
21~.K.,._",&=2,,_2_'.. ~ Kg
Fig. 14.6 Lumped masses for Example 14.1.
mL
420
P, P,
27.. 54
 !3L
156
3L' . 22L
4L'
8,
When (he mass matrix, eq. (14.34), has been evaluated for eaer. beam element of the structure, the mass matrix for the entire system is assembled by exactly the same procedure (direct method) as described in developing the stiffness matrix for the system. The resulting mass matrix \"lin in general have the same arran2:ement of nonzero terms as the stiffness matrix. Th~ dynamic analysis using [he lumped mass matrix. requlres considerably ~ess computational effort than the analysis usjng the consistent mass method for the following reasons. The h.:.mped mass matrix for the system results in a diagonal mass matrix whereas the consistent mass matrix has many off diagonal terms whkh are called mass coupling. Also, the lumped mass matrix contains zeros in its main diagonal due to assumed zero rotational inertial forces. This fact permits the elimination by static condensation (Chapter 13) of the rotatior.al degrees of freedom. thus redUCIng the dimension of the dynamic problem. Nevertheiess, the dynamic analysjs using the consistent mass matrix gives results which approxlmate bette: to the exact solution compared to the IUr.1ped mass method for the same element discretization.
example LIm, m::;:;: 420 kg/rn into eq. (14.34) gives the CO:ls:stent mas.s matrix [M~2)1 for any of the three beam segments as
0
3
54
13
4
 13 1 1 3
156
22
4
[M:."J =
22
5L
:, 13
13 ..3
156
22
22
12 ,
1
(a)
"J 4
Example 14.1. Determine the lumped r.12.SS ar.:d the cor.sis;:ent mass matrices for the cantilever beam in Fig. 14.6. Assume uniform mass. iii = 420
kg!m.
The assemblage of the system mass :natrix from the element mass matrices is car:ied out in exactly the same manner as the assemblage of the system stiffness matrix from the element stiffness matrices, that is, the element mass matrices are allocated to appropriate entries in the system mass matrix. For tl:e second beam segment, this allocation corresponds to the first four coordinates as indicated above and on the right of eq. (a). For the beam segment &, the appropriate ailocation is 3, 4, 5, 6 and for the beam segment 8, 7,7, 1. 2 since these are the system nodal coordinates for these beam segments as indicated in Fig. i4,6. The consistent mass matrix [Me] for this example obtained in this manner is given by
0
4
5 0
Solution: (a) Lumped Mass Matrix. The lumped mass at each node of any of the three beam segments, into which the cantilever beam has been divided, is simply half of :h; mass of the segr.1cnt. In the present case, the lumped mass at each node is 210 kg as shown in Fig. 14.6. The lumped mass matrix ~Md for this structt:re is a diagonal matrix of d:mension 6 X 6, namelY.
[M d =
312
0
54
13
312
13
01
0
8
13  3 0
0
3
0
2
!
[Mol
54
 13
54
13
I3
(b)
0 54 13
8
13
 3; 4
22 4
r420
420
210
0j
0
0
156
 22
3
J5
(b) Consistent Mass Matrix, The consistent mass matrix for a unifonn beam segment is given by eq. (14.34). The substitution of numerical values for this
We :'lote that the mass matrix [iWe ] is symmetric and also banded as jn the case of rhe s:iffness matrix for this system. These facc:s are of great importance
414
415
in developing computer prog!"nms for strucClral analysis, since it is possible to perform the necessary caicu!.2.lions storing in the computer only the diagonal elements ::md lhe elements (0 one of the sides of the main diagonal. The
maximum m:.mber of no:1ZUO elements in any row which are required to be stored is referred to us the bandwidth of the matrix. For lhe ma~rix eq. (b) the bandwidth is equat ~o four (NBW 4). In this case, it is necessary to Store a total of 6 x 4 = 24 coefficier.rs, whereas if the square matrix were [0 be srored, it would require 6 X 6 36 storage spaces. This economy in swring spaces becomes more drar:1:lttC fOf structures with a large number of noda! coordinates. The dir:leoSloli of rhe bandwidth is di:ectly related to :he largest difference of the nodal coordinates assigned to any 0: the elemc:nts of the structure. There:ore, It is imporrnn{ to number the system nodal coordinates so as to r.'1inimize (his differenCe.
Pix. ()
dis~
14.5
DAMPING PROPERTIES
Damping coefficients are defined in a manner entirely parallel to the deflni[Lon of [he stiffness coefficient or the mass coefficier:r. Specifically, the damping coefficient Ci' is defined as the force developed at coordinate i due to a unit velocity at j. If [he damping forces cisrributed in [he StfUcmre :::ould be determined, the damping coefficients of the various structural elements would then be used in obtaining the damping coefficient correspor.dir.g to the system. For example, Ihe damping coefficient C;j for an element might be of the form
c,=
r'
wrirten direcl:y. In ger.eral, however, loads are app:ied at poims other than r.oda! coordinate:>. Ir. addition, the external load may :r:clude [he action of distributed forces. !n {his case, [he load vectOr corresponding (0 the nodal coordinntes consists of the eql,ivaiem or generalized forces. The procedure to de{emline the equivalem r.odaJ forces which is consistent with the derivation of the sciffness matrix and ~he consistent nU1SS malrix is to assume (he validity of the smtic deflection functions, eqs. (14.9). for the dyr'.amic problem and use the principle of virrua! work. Consider rhe beam elemem in Fig. 14.7 when subjected to an arbitrary distributed force p{x,t) which is a function of position along the bearr as weI! as a :unction of ;::me. The equivalent force PI llt coordinate 1 may be found by giving a virtual cisplacement 0] = 1 at this coordinate and equating the resulrir:.g exrernal work and imerr;a! work during this virtual displacement. !n (his case, the external work is
t;
( 14.35)
Wc~P,8,
where C (x) represems the dis(ributed damping coefficier:[ per unl[ length, If the element damping matrix could be cakulated, the darr.ping matrix for the entire S[f1.lCfUre could be assembled by a supe:position process equivalent to rhe direct stiffness matrix.. In practice, the evaluarion of the dnmpjng property c(x) is impracticaole. For this reason" the damping is generaHy expressed 1n terms of damping ratios obtained experimentally rmher than by a direct evaluation of the damping :natrix usir.g eq. (14.35). These dampir:.g ratios are evaluated or estimatec for each natural mode of vibration. If (he explicit expression of the e.amping matrix [C] is needed, it may be computed from the specified reiative damping coefficienLs by any of the methods described in Chapter l2.
=P,
(14.36)
since 8, !. The in;:ernal work per unit of length along {he beam is p (x, t) if;1 (x) and (~e total internal work is then
W, =
r
f
(14.37)
Equating exterr;al work, eq, (14.36), and ir.(ernal work, eq. (14.37), gives the equivalent nodal force as
P,(l)=
r' J,
p(x,l)ifi,(x)dx
(14.38)
2.S
14.6
EXTERNAL LOADS
\Vhen the dynamic loads acting on the StruCture consist of concentrmed forces and mOments applied at defined nodal coordinates, the loae. vector Clln be
(14.39)
416
MuJtideg:eeof~Freedom
Systems
Nixl
417
f;
Fig. 14.8 Beam segment subjected to external distributed load showing equivalent nodal forces,
14.8 and detennine the Example 14.2. Consider the beam segmenr in element nodal forces for .1 uniform disuibmed force along the Jeng:h of !he
beam given by
Solution: Introduction of numerical values into the displacemenrs functions, eqs, (14,9), and substitution in eq, (14.39) yield
P,(f)=200C (13x'
Fig. 14.9 (a) Beam element loaded with arbitrary distrib<.::(ed axial force. (b) Beam element acted on by noda: forces resulting from displacement 02 I, undergoing a
,
virtual disphcement
o! = 1.
16,675io 101
examp.le, kc;r:; 1s the vertical force at the left end. If we now to this deformed beam a unit displacement 0 1 = 1. the resulting ex[ernal work is
p) (Il
200
" L
= 100 sin
:Ot
Or
P,(r)
=200fo'x'~t
(14.40)
since 0= 1. The internal work during this virtual displacement is found by con.~idering a differential element of length dx taken from the beam in Fig. 14.9(b) and shown enlarged in Fig. 14.10. The work done by the axial force N(x) during the virtual displacement is
(14.41)
where Dc represents the relative displacement experienced by the nonnal force N (x) acting on the differentia! eiement during tbe virtual displacemen:. From Fig. 14.10, by similar triangles (triangies I and H). we have
~=d~(x) diff, (x) dx
418
419
called the consistent geometric s(i.f/ness matrix, In the special case where the
aXial force is constant along the length of the beam, use of eqs, 04.44) and 04.9) gives rhe geometric stiffness equation as
'
P, P,
p)
36
3L
 36
1 i
.y.
3'OL
P,
l
3L
4C'
 3L
 L'
3L
36
3LJ L" :
Ii,
36
3L
3L  3L 4L'
(14.45)
~~~f
w,lxl
1~==~======:j
Fig. 14.10 Differemial segmem of deflected beam in Fig, 14.9.
The assemblage of the sys.tem geometric stiffness matrix car: be carried out exacrly in the same manner as for rhe assemblage of the elastic stiffness matrix. The result:ng geometric Stiffness matrix will have the same configurallon as the elastic stiffness rr.mnx. It is customary to define the geometric stiffness matrix for a compressive axial force. In this case, the combined stiffness maLrix (K"J for :he structure is gi\'en by
(14.46)
in which (K]ls [he assembled eifistic sdffness mauix for the structure and [KcI (he corresponding geometnc stiffness matrix" Example 14.3. For the cantilever beam in Fig. 14" II, detennine the system geometric matrix when an axial force of magnitude 30 N is applied at the free end as shown. ir. this figure.
or
15 = d.p, . d.p, dx
dx
(x) .
dx
15. = .p,
in which t/I{ (x) and I/I~ (x) are the derivatives with respect to x of the corres~ ponding displacement functions defined in eqs. (14.9), Now, substiruting 8~ into eq, (}4.41), we have
( 14.42)
Solurion,' The subs[irutlOn of numerical vatues into eq. (l4AS) for any of the three beam segments in which the cantilever beam has been divided gives the element geometric matrix
3 36 36 3 3 36 3 36  3 ' 3 41 3 1
Then integrating this expression and equating (he result to the external work. eq. (14.40), finally give
kGn
[Kcl
"
n
6
"
(14.43)
In
,~
(14,44)
6::,
)~
&,
)~
iO
6::,
1m
N =. 30 Nl"WtQflS
lm{lm
Fig. 14,11 Cantilever beam subjec(cd
j
In the denvatior. of eq. (14.44), if ;s assumed that rhe normal force N(J;) is
icdependent of time. \Vhen the displacement functions, eqs. (14.9), are used
420
421
since in this eX2mple L = 1m and N= 30 N. Use of the direct method gives the assembled system geometric matrix as
r !
[Kc'
72 0
0 36
8
3
0
0 0
3 1
72
0
l'j
3
0
0
36
3  1
n
!
In practice, the solution of eg. (14.51) or eg. (!4.52) is accomplished by standard methods of analysis and the assistance of appropriate computer programs as those described in this and the following chZl:Jters. \Ve illustrate these methods by presenting here some simple problems fo~ hand calculation. Example 14.4. Consider In 14.12 a :miform beam with the ends fixed against translation Of rotation. In preparatJOn for analysis, the beam has beer. divided into four equal segments. Determine the first three natural frequencies and corresponding modal shapes. Use the lumped rr:ass method in order to simplify l;,e numerical calculations.
So[urion: We begin by numbering sequentially the nodal coordinates starting with the rotational coord:nates which have to be condensed in the lumped ffi8.SS method (no inert:a! effect associated with rorationai coordinates), con tinuing to numbe, the coordinates associated with translation, and as~igning the dummy last D'Jmber 7 to any fixed nodaJ coordinates as shown in Fig. 14.12. The stiffness matrfx for any of the beam segments for this examo:e is obtained from eq. (\4,20) as .
w
 36 3
36 3
3
14.8
EQUATIONS OF MOTION
In the previous sections of this chapter, the distributed properties of a beam and its lot:d were exp:essed in tenns 0: discrete quantities at the nodal coor~ dina[es. The eq:lJtions of motion as funct:ons of these coordinates may then be established by imposing conditions of dynamic equilibrium between the inertinl forces {FI(r)}, damping forces {FD(I)}, elastic forces {F,(I)}, and the external forces {F(t)}, th2.t is,
(14.47)
5
7
4
7
[K]
6
4
12
The forces on the lefthand side of eq, (14.47) are expressed in terms of the system mass matrix, the system damping matrix, and the system stiffness matrix as
{FI (In = eM] {FD(t)}
(14.48)
6
!2
6
2
 6
n~
!
( 14.54)
6
4 J 1
5 2
With the aid of the system nodal coordinates for eact: beam segment written
= [C] {y}
(14.49)
(14.50)
SubstItution of lhese equations into eq. (14A7) gives the differential equation of motion for a linear system as
[lvl]{y)C[C]{j}7[K]{y} {F(t))
(14.51)
at the top and on the right of the stiffness matrix, eg. (14.54), we proceed to assemble the system stiffness matrix using the direct method. For the beam segment lA, the corresponding labelS are 7. 7, 4, L Since the label 7 which corresponds to fixed coordinates should be ignored, we need to trans]2.te only
7
fn addition, if the effect of axial forces is considered in the analysis, eq. 04.51) is modified so that
[M) (Y}+[C]{j} +[KJ{y}
{F(I)}
4
I'
"
2'
&
(1) I
.&.
1m
""
,
I
&
1m
3' "
&
7~
(1452)
EI,iil/
ilm
0) I
1m~~
in \vhich
(j ~.53)
Fig. 14.12 Fixed beam divlded in fo~r elements with indication of system nodal coordinates (Exampie 14.4).
422
Framed Structures
f>.~.odeJed
423
the lowest 2 X 2 submatrix on ~he cighr to locarions given by the combinatio:;'L of row indices 4, 1 and column i:1dices 4, I; for the beam segment 11::., we mmslnte [he 4 X 4 elements of matrix eq. (14.54) to the system stirfness mu!rix to rows and columns designated by comb~r.atton of indkes 4. r. 5, 2 as labeled for this element 2nd so forth for the other two beam segments. The assembled system stiffness matrix obtained in (his m<lflner is
r
[T] =
0.214
0 I 0
0 0
(14.59)
r~
[K]
0
2
0
6
8
2
6 0
6
01
6
0
l
{ 14.55)
0 0
EI
8
0
6
0
0
24
lj
f1
0
6 0
12
0
~2
As an exercise, [he reader may check eq. (13.7) for this example by simply performi:1g the mutrix multip:icurions
 12
0
24
 12
6
24 The lumped mass method applied to this example three equal masses
of mag:li[Ude iii at each of [he three translatory coordinates as indicated in Fig.
The reduction or condensation of eq. (14.55) is accomplished as explained in Chapter 13 by simply performing (he GaussJordan elimi:1a[ion of the first three rows since, in this Cfise, we should condense these first ~hree coordiames. This elimlrHlrion reduces eq. 04.55) to the foHowing mmrix:
[i4J =mlo
rIO 1 0
01
( 14"60)
0
t
 0.214 0.858
0.214
...... " ..
 0.750
0
~o 0 I j
0 0
0
[A]
!O
0
1 ,0
0 0
0
0.750
 12.00EI
The natural frequencies and modal shapes are found by solving the undamped
18.86EI  12.00EI
(14.56)
15.00E1
12.00EI
= {a}
=
(14.6l)
l..O 0 0
S.l4El
Ass:im:ng the harmonic solu(jon ({y} ~ {a} sin Wi), we obtain ([K] 
Comparison of eq. (\4.56) in partition form with eq" (13.1!) permits the
identlfi:::ation of the reduced stiffness matrix [7:, so that
{OJ
(14.62)
{K]
 12.00
(\4.63)
is.OO
(14.57)
5.14  12.00
znd
"'12
m(2 inn
inn m/2
(d
iii
ro/2 iiia
iiil2
[TJ = L
",12_
0.214 0.858 0.750  0.214 1
(14.58)
~m2
(bl
Fig. 14.13 (a) Lumped mas<;e:; for unifcrm beam segmentf>_ (b) Lumped masses at system nod:\! coordinate:).
424
425
SubstitutlOn in this last equation and (14.60) yields 18.86  " 12.00 5.14
The normalized modal shapes which are obtained by division of the elements of eqs. (l4.69~ by corresponding values of (m,= 100 kg for this example) are arranged In the columns of the modal matrix are
/Y;';;;;;
1=0
(H.64)
A:
0.0707 0
0.0707
0.0562 i
 006071 00562 J
(I ;1..70a)
in which
EI
The roots of the cubic equation (14.64) are found to t>e
(14.65)
!he modal shapes in terms of the six origmal coordinates are then obtained ~y eq, (J 3.8) as
[ <P]
(14,66)
= [71[ \tiJ.,
0.0301
A, = 1.943,
Then from eg. ([4,65)
A, = 13.720.
and
,13
= 37.057
r
[<P] =
0.0594
1I
0
0.0594 0.0431 0.0793 0.043 [
0.1212
00301 0.0707
:""l
0.0457 0.0562  0.0607: 0.0562"
(14.70b)
w:: = 3.704~;
Ellm
(14,67)
w) = 6.087,;
The first three natural frequencie.s for a uniform fixed beam of length L = 4m determined by the exact analysis (Chapter 21) are
WI
Example 14.5. Determine the steady~state respOIl$e for the beam of Example 14.4 when subjected to the harmonic forces F[ Fo, sin
wt
wt
F2 = Fez sin
a<ld
(l4.68)
The first two natural frequencies determined using the three~degreesoffreedom reduced system compare very well with the exact values. A practical rule in condensing degrees of freedom is to condense those nodal coordinates which have the least inertial effect; i:l this problem these are the rotational coordinates. The modal shapes are determined by solving two of the equations in eq. (14,62) after substituti:1g successively values of Wit lVz, w} from eqs, (14.67) and conveniently setting the first element ror each modal shape be equal to one. The resulting modal shapes are
The modal equations {uncoupled equations) can readUy be Wrltusmg the results of Ex~mple 111.4. In general the nth normal equation is given by
t~n
Sol~tian:
(l4.71)
1.01 LOS
LOO"
in which
( 14.69)
N
426
427
2.,
. ~ S;fl I.l/l
= "~~~
W"
(v
Prsin
w[
The calculaticns required in eq. (14.72) a:e conveniently arranged in Table 14. L The deflections at the modal coordinates are found from the tr2.flsformation
{y}
~
[IPJ {z}
(l4.73)
 L207 X 10 y, = ., 4,094 X 10 
6 6
w(
The minus sign in the resulting amplitudes of motion simply indicates that [he motion is 180 0 Ou~ of phase with (he applied harmonic forces.
14.9
', Y2
0.0594
0
0.0431
0.0301  0.1212
 0.0457
0
0.0457 0.0562  0.0607 0.0562
Y J
I" 00594
O.OJOI
0.0707
Y, Y, , Y,
or
0.0793
0
 0.0707
O.Odh ~
0.0431
The central problem to be solved using the dynamic stiffness method is to dercnnl:1e the d;splacements at the nodal coordinates. Once these displacements have been determined, it is a simpk mauer of sUbstitt:ting the appropriate displacements in rhe condition of dynamic equilibrium fer each element to calculate the forces at the nodal coordinates. The nodal element forces {P} may be obrained by adding the tnen:ial force {PI}, the damping force {P D }, the e:astic force {P s}. and subtracring the nodal equiva1ent forces (P E }. Therefore, we may write
Y,=
Y,
Y3 =
Y~
 1.818 X IO'rad
3.524 X 1O 6 rad 1.207 X IO'm
or
{Pi ~ [m] (8)
T
[el (ill
+ [k]{B)
(Pel
(14.75)
=
Ys=  4.094 X lO b m
In eq. (14.75) me inertial force, (he damping force, and the elasdc fcrce are. respectively,
(PI)
Y,
TABLE 14.1
3.329x 10'm
= [mliB)
~
(P D ) {P,}
Z,,=
 5.200 10 ,5 l.500 10 , 0.048 10'
[el (B)
(14.76)
Mode n
w,
P" =
= [kl {B}
2
3
367.2 70.7
. . _  _..
 135
where [m] is the element mass matrix.; [e] (he element damping matrix; (kJ the element stiffness matrix~ and {8}., {8}, {8} represer.ts, respecti ve)y, lhe displacemer:t, veiocity, and acceleration VeClOrs ar the nodal coordinates of the element Aitho:lgh the deterrnin2:ion of element endforces at nodal coordinates is
42B
429
given by eq, (14,75), commercial computer programs onlY use the forces due to the elastic displacements, that is, the elew.ent end forces are calculmec as
(P}
= [k]{8}
2
121.2 201.0 664.3  322.2 1347.;' 322.2
3
1947.9 481.4 1392.4 587.0 ....
4
 382.3  5870 1880.4 1292.6
Un:ts
N Nm N
This simpiified approach may be justified by the mathematical model in which the inertial forces. damping forces, and external forces are assumed to be acung directly at the nodes. Therefore, it is important in the process of digitizalion :0 se!ect beam elements relatively short as to approximate the distributed propenies with discrete concentrated values at the nodes between the elements. The detennination of the element nodal forces is :llustrated :n the follOWing example. ExampJe 14.6. Determine the element nodal forces and moment'l for the four beam segments of Example 14.5.
P, P, P,
,038.3
481.4
~
p,
Nm
w 3000 fadl
To cornpiete this example, we substitute the numerical valees of sec, m = i 00 kg 1m, and E! = lOs (N  m J ) and obtain
6
4
12
6
2
Solution: Since in this exampie damping ;s neglected and there are no external forces apphed on the beam dements except those at the nodal coor~ dinotes, eg. (14.75) reduces to
{P}
6
7.5
6
2
which then gives
Pi
6 4
0
 120.7
6
I ,  261.6
(ld.77)
\'"
The displacement functions for tbe six nodal coordinates of the beam in Fig. 14.12 are given by eg. (14.74). These displacements are cenainly also the displacements of the dement nodal coordinates. The identiflcation for this example of corresponding nodal coordinates between beam segments and syste:n nod2.: coordinates is
P,
P, =
(oj,
0 0 y,
:"YI
J~: l , , I j'
8}
)'s
y,
{O}"
Yf
y, , y,
l y, ,
0 0 0
0 0 0
i Y5
0 0
(14.78)
where {o} i is the vector of nodal displacement for i beam segment The substitution of appropriate quantities into eq. (14.77) for the frrst beam segment results in
, 1
: P, i 00 . P: = in 0 0 0 P, 2 .0 0 I P, , L0 0 0
12
The nodal element forces found in this manner for all of the four beam segments in this example are given in Table 14.2_ The results in Table 14.2 may be used to check that the dynamic conditions of equilibrium are satisfied in each beam segment. The free body diagrams of the four elements of this hearr, are sbown in Fig. 14.14 with inclusion ofnccal inertial forces. These forces are computed by multiplying the nodal mass by the corresponding nodal acceleration .
y,
Y: J
( ui)
sin Wt
6
4
12
6
2
4
+1
6
12 6
12 6
6
j: l ~: \"'.
121.2
MMM~
664.3 1347 1038.3 1948 1392.4 :3a2.3
Dyn~!nic
1880A
Fig. 14.14
43C
Mu:llidegreeof~Freedom
Sys:ems
431
14.10
PROGRA~
13MODELING STRUCTURES
100 00 10C.OO
AS BEAMS
The computer program presented in this section calculates the stiffness and mass matrices for a beam and stores the coeft1cients of these macrices in a fiie, for future use. Since the sriffness and mass matrices are symmetric, only [he upper triangular porrion of [hese mat:ices needs to be stored, The progrrtm also stores in another fi!e, named by rhe user, the general information on the beam, The :nformation s[Ored in these tiles is needed by programs which perform dynamic analysis such as calculation of natural frequencies or de[ennination of [he response of the strUCtLlre subjected to exremal excitation.
fixed~ended
o .to
!)
10
1"0.00 iOO.M
Example 14.7. Determine the stiffness and mass matrices for the uniform beam shown in Fig. 14.lS, The following are the properties of the beam: Length: L = 200 in Crosssec~ion moment of inertia: 1 = 100 in ~ Modulus of elasticity: ~ 6.58 E6 Iblin' Mass per unit length: fit = O. I 0 Ob" sec:! lin lin)
?~:H:::.05
O.OOOOi!>OO
1.OS2EZ.O$
~l.
:>.H!i.<1!:.O;
.,
o. GOOC5;~GO
}:6as.o~
1 ,'i;]j:t!,;;6 (,1.6),::_05
,
.,
s
l,~n:ie~o~ 6noz~:p
0.0000=:.00
v.coooe,,:n
;;.Jl5SE.Ct.
1.~7ne:dl6
1.5J;;20:.C5
O,vQOC_CC
C,~Qcc:::.eo
S192i>G5 1.$):::(;5;.0
C.;)O'.i~Ii':.ct; D00n~C0
O. Ot:l:tl:;:. 00
.0<;28"2.n
.,
:.;
.O{)0~'t.:/) .il.~S't,t;..;
:\7$2Z.36
1.26H:::.OS
O. COOO.OO
c,5nn.:;5
.42S.sr:)1 1.7)$,,,.00
L1t~3!i>N'
.6120.0,
.0523:;.OS
lfi
The beam is divided in four segments of equal length as shown 14,15. This division results in a total of five joints of which two are fixed, thus giving a tota1 of four tixed coordinates (the two ends of the beam are fixed for translation and for rot3lion).
Solwion:
3.71no:.oo C .0000:::.00
O.00COE.OO
1. '81Ol';.C2
.
7.73B1';:'00
~8.923i>+Ol
o.oooos.OO
G. OOOOg.oo
ooocz.:.CQ
O.
QCCO~CC lJaH:~CC
.,
42e6:::~Ot
.7331S.00
".000::'0:.00
0.00002+00 0.0000::.;)0
0.0000;:;:.00 O. CjOCS.oO
6,
~?g:;;<:>Ol
.j,1Hil't."'C
.,
. nal.GO S2S6E.Gl
5 sa60:~Cl ().CC()().CO
GJ,:m:iU;l.
DA7A:
Example 14.8.
Ne,,~
Of nxo COO)l.D1NA'l'ES
l>'C.o4
t;,,6Sl>r;OOO
v.cC()L(J$ Of 5;1..,>,5",;:::::1':::,>
[he natural frequenctes and modal shapes, (0) the response to concentrated force of WOO lb suddenly applied at the center of the beam for 0.1 sec and removed 14.16. Use time step of integration .1: = 0.01 sec. linearly as shown in
F{t)
C .00
O.GO
SO .00 lUG.OO
" .GO
'1 5D .CC;
,OC o co
0,00
1ooo!):)
c..
~(i)
&,
fA
G)
(3)
61
ffi ~
CD
0
a
D.1
f5D"+50"\50"+50"~
432
433
Solution: (a) During the execution of Program to r:1odeI the beam in Problem 14.7, a named "SK" has been created to store the data necess.::.ry to execute Prograa: 8 to calculate natural frequencies and corresponding modal shapes. The execution of Program 8 gives the fonowing output:
me
C .554
l':~CJ7
c. ):s
t7""i.N
Output Results
1.. 1;'<3
'H"S.Sii
().5';.4
:;
.C~ '1
14.11
7.:n
1) .l.?3S1
c. 0C5.:;2
~C.(j0:;55
C.J56Dii
LOC::;~O
~LC;;CQv
1?JSJ
.o.nne
, nest
~5l57
C.C~2~?
n'ila:
~?CSl ~S~S,
, .,
[j
:Wf42 :;)155
D.%7S.
(
;)::1:
:;,00030
0)"''''
0.
:;.0652
~S:"S'
:.D2Z9:
J;B5
.,
~;:
,0:3::3:;
.OCCJC
0.02$,,:
~J.0"';;52
c. :; .. %5
C, :/47$8
C .04758
c .O7~;'7
." . ~$:;H
Example 14.8. The cantilever beam shown in Fig. 14.17(a) is subjected to the timeforcing function shown in Fig. 14,17(b). applied at the free e!1d of the beam in the y direction. Use the program COSMOS to: (1) generate a finite element modei of [he beam with len twodimensional beam elements, (2) calculate the ten lowest narural frequencies of the beam. and (3) detenni:1e the response in tenns of displacements, internal forces, moments. and stresses z.t time = 0.009 sec. Use 50 time steps at interval Lil 0.006 sec .
w;
",so;o:\d'~ '1=0..3
P:::: O.!l ":>sd/i,;'
Fit)
(al
Gfu"\'VI7;':'ZO:';;'~
:::l'DE..X
o~~~
D 0.\
(b)
o .i.e:;o
iJ .20;)()
:CiJiJO.CC
J.G)
Fig. 14.17 (a) Cam:lever bea;n for Example 14.8. {b) Loading function.
435
Soituion: COSMOS:
(10) Set options for frequency calcuLation to eXfract 10 frequencies using the Subspace Iteration Method wirh a r:1uximuf:i of 16 irerations, and run the frequency analysis;
A.."1ALYSIS > fREQ /BUCK > f.._FREQ;JSNCY
V:EW, 0, ), 1, D
(2) Define the XY plane", Z = 0:
GEOMETR't ) GftID > PLANE:
JLFREQUENCY, 0, 0, 0" !)
l,
S,
15,
0,
0.
O.
0,
1805,
0,
:'E06,
ANALYSIS
R_FREQUENCY
FREQ/BGCK
>
R~FREQUENCY
PLANE, Z,
0,
CURVES ) CR!?CORO
Q,
f.::eq'lency
10 i
;#
CRPCORD, 1, 0, 0, 0, 10, O.
0, 0 1
2
?recr.:ency
(Kad !sec}
Frequency
(cycles /sec) 1. 50770e: + 001 9.96317e+OOl
2. 497<13e + 002
Period
(seconds}
using BEAM2D:
SGROUP,
,
3
5
5 7 8
MPROP
EX,
30E6,
DE:::.JS,
0 3
9 10
::.. OlOISe T 002 6 . 26005e + 002 1 "56918e + ::;83 .73528e+ 003 3 . 36'.)02e j 003 .66891e + 003 5 _SCO",Se 1 003 7. 65376e + :::03 B" 10530e + 003 1 Q4S00e + 004
C;.22006e002 I. ::;0370e002
4,Q0412e003 3,62085e003 ;. 85720eCC3 1.34575e003
2 .76178e
5" 35560e
r
002
7 .43080e
+ CO2 + 002
1.l4224eC03
8,20938e004 7.7509ge004 o.C1263e004
(7) Generate mesh of 10 beam elements with two nodes along curve (Fig. 14.18 shows the beam modeled with 10 bea:n segments, 11 nodes):
MESH:::NG > PARAM_!4.ESH > I'CCR
PD_.ATYPE,
2,
0,5. 0.25, 2
(13) Define dynamic forcing function and apply as force at node II in the Y direction:
AKALYSIS
;'.N.!;r~YSIS
j
M_,CR,
1, 2, 10, :
POST~DYN
!?C~Cl1RVES)
PD_CtJR'IYP
PD,~cu:rI'Yp,
>
1, 0, 0 POST_DYN
CURVES:>
P:::'LCURDEF
OISPL~u~TS
;>
~ND
PD_CtJRDEF,
... , 0,
0, 0,1,
1000(}, 1, 10000
P.L,
0,
NODES
11. 1,
NHERGE
C.OOOl,
5
CO~?I'ROL :> ACeYVE ) AC1'S2T ACTSE'r, 'Ie, 1 LOAD Be > STRDCTUR.P.L ) FORCES F.ND, 11, FY, L :1, 1
>
FND
1,
0,
0,
7
:)
10
Fig. 14.18 Cantilever beam modeled with 10 beam segments (11 nodes).
PliAL'(s:::S ) PO:S?~DYN ) PD_PRlr;:T, 1, 0, 0, 0, J:..l"f.~I.,ySIS ) ?OS'f_DY~ > ?D_?LOT, 1, SO, 1, 0 ANft.:':iSIS ~ POST_DY~ >
0,
2.,
1,
50, 1
PO_PLOT
OL:Tl?UT
OUTPUT
> NR2SP
436
437
! 2.S
U
t~.: t ,
I ,
9.S
It
11, .2
,
)./
;U '\ I V \ 1/
Ii
i
/ '\
\ \/
,
\L't . \
\IJ
i
TABLE 14.3 Forces and Stresses at the Nodes of the First Three Elements of the Cantilever Beam Modeled in Fig. 14.18 for Time Step 15 (t= 0.09 sec)
i ,
i
,
&.1')$
,
I ,
GAiS
,
v;;.~.'7~'2;:;';
:
'J:;"C.DO;;CE.(D
;
C~$~V.)C)vS~C)
c.7i).i.E;.. j~
I';.C:;;O~jr.:
0,::)Cn.cc
C,7%1;:;:')~
\M$;S$)~Q.::DJ()S~G0
C.vi)OCE;'JG
, iLl
I
9,12
G.le
,
G.2l
I !
1$,24
:~:~~.8154ii>CS
W~/$~)~').50?OZ.C7
C.C;5S",>!)7
, j
(LV
\'''~''
$150E04 0 0
.iH5n~c~ (\)C~2'C!,;
Hs~C
.,)000,,")0 0,
C(l()C:;;~CO
(~~
iSS;
~'J, OOOQ::>O~
{) .OCJGJS,()O
G.4(j~'fZ.07
\'t~0.Gj"C2<;')
H~~,1Snt~O~
0,5';46.05
(~;t/Stl",O."5S6S*07
CJ. 8"O~r::~()~
0.(lD0Q:;:C~
'I,
OOOO::~CO
(i{v ,. $",)
~G.
o.seMs.~o~
(Xt:S~!~').1':~':!E.;);S;;v:;.x~C.1c~n_~7 S;t,,i,.r."~,404~S47
14.12 SUMMARY
In this chapter, we have formulated the dynamlc equations for beams in reference to a discrete number of nodal coordinates, These coordinates are translational and rotational displacements defined at joints hetween structural elemer::ts of the beam (beam segments). The dynamic equations for a linear system are conveniently written in matrix notation as
[M] {j} + [C] {y} + [E] {y}
1, :;:. liN
>
R..STRESS
(17) Activate Xl' plot information for Y displacement at node 11 as a function of time, and plut this function (see Fig. 14.19):
D:=SPLAY
>
XypLOTS
:>
AC'TXYPOS'I'
\JY, 1:, :;'2,
{F(!)}
ACTXYPOST,
1, T.n1E,
>
DISPLAY ) XY?LOTS
XYPLO'L
.L
XY?LOT
LASER...J:::::r, 150, 0
(18) Cse an editor to list from tbe output file the forces and stresses at the nodes of the beam elements: TabJe 14.3 shows the internal forces and the stresses at the nodal coordinates for the first three beam segments calculated at time step 15 (t 0.09 sec).
where F(t) is the force vectOr and [M], aod [E] are, respectively, the mass, damping, and stiffness matrices of the structure. These matrices are assembled by the appropriate superposition (direct method) of the matrices determined for each beam segment of the stmcture. The solution of the dynamic equations {i.e., the response) of a Unear system may be found by tbe modal superposition method, This method requires the determination of the natural frequencies w.~(n::: 1, 2. 3, '"', IV) and the cones ponding norma! modes \vhich are conveniently written as the columns of the nodal matrix [<Pl. The linear transformation {y} [<PJ {zj applied to the dy
"
I ,
Dyna.mic Anaiysis of Beams
439
438
numie equations reduces them to a set of independent equat;ons (uncoupled equations) of the form
143 For {he beam in Problems 14.1 and l4,3 use slacie condensation to eliminale the [Ota~ional degrees of freedom. Find the transformation matrix nnd the reduced stiffness l!.nd mass matricc.:>.
De:err:line the ::.a:uml frequenci~$ and corresponding normal modes using the reduced stiffness anc mass matrices obta)ned in Problem 14.4,
where (" is the modal damping ratio and P,,(t) :i,;"F,(;) is the modal force. An alternate method for determining the response of Hnear systems (also valid for nonlinear systems) is the numerical integration of the dynnmic equa(ions. Chapter 20 presents {he stepbystep linear acceleration method (with a modificarion introdttced by Wilson) which is an efficient method for solving :he dyr.amic equ<1rions. A computer program is r:lso cescribed for the dynamic analysis of beams. Tnis progrr:m performs the task of assembling and storing in a flIe the stiffness anc mass matrices of the system, T:1.ese matrices are subsequemly used by olher programs to calculate nawoal frequer.cies Of the response of the beam to external excitation.
14.7
De(ermi::.e [he nalt;:al frequencies and corresponding normal modes using the :educed $[~tbes:; U:1C mass matrices obtained in Problem 14.3. Delermine the geometr~c s:iffness matrix for the beam of Problem 14J when ;( ~s subjected :0 a co:::;(.:::( lensile force of 10,000 Ib as shown in Fig. PI4,g,
y
4
J.l.8
El "'" ,cf{!b~in,71
Q ..
3.86 {lbJif'l.)
3~
~~~~~~~x zt,o,XX) !b
Fig. P14.8.
14.9 Perform .:>tatic condens<Hion (0 reduce the geometric stiff:1ess matrix obtained in Problem l4,K Eliminate (he rotational coordinates. Use results from Problem 14.4 and 14.9 and determine the n:Hural frequencies a"d corresponding normal modes for the beam shown in Fig, P14,8.
PROBLEMS
14.1
A uniform benm of flexural sriffncss 1 10it (lb in 1) and length 300 in h.D.s one end fixed and the other Simply sl,pponed, Determine lhe system sriffness r.,atrix considering rhree beam segu:.ems and rhe nodal coordinaces indicatec in Fig. P14.1.
y
Use (he results of Problems 14.5 and 14,9 and determme the natural frequencies and corresponct:,g norma! modes for [he beam shown in Fig, P!4.8. Deren:"'J:ne the stiffness matrix for;) beam segment in which the flexural stiffness has a !inear varimion as shown in Fig. Pi4.12.
5
EI'1<
I
Fig. PI4.!.
14.2
~~~~~~_x
3''"
tOOin.i,oo,n.+,oo,n....,,1
Ass:Jming thar the beam show;) ;n Fig, P14,1 carries a uniform weight per unic 1e:1gth q = 3.86 Ibfln, determine the system mass matrix corresponding to the lumped Class form:.tlal:Ofl.
Determine the system mass m,t(rix fOf ProbleF.l 14,2 using the consistent mass method. For the beam in Problems 14,1 :lOO 14.2, use static conde:1sation !O eliminme the massless degree,) of :reedom. Find the transformation matrix and me: reduc.
Cil======i,2EI
,'1'
~
14.13 14.14 Determine the lumped mass m:.itrix for a beam segment in which [he mass has a linear disrriburion as shown tn Fig. P i4,13.
I~~~L
Fig. P14.12 .
14.3 14.4
Determine [he consistem mass matrix for the beam segment shown in Fig,
PI4,13.
440
441
figure. Eliminate t.1e fm<ltion<.ll coordinates by srntic condensation (Problem 14.5). ]\;eglect damping in the system,
2mo (lb/inJ
14,19
Determine the natural frequencies and corresponding normal modes for the beam shown in F!g. P!4.1: (<1) condensing the three rotational nodul coordinates: (b) no condensing coordinates. (Use lhe consistent mass method.) Derermine the response for the beam shown in fig. PI4.15. Neglect camping. (a) Do not condense coo::dinates: (b) condense the three romtionaj coordinates. Repeat Problem l4.20 assuming 1.0% damping in all the modes. Determine the steadystate response for [he beam show::; In Fig. P14.18 whe::; SUbjected to a harmonic force as shown in the figure. Do not condense coordinates, and neglect damping in the system. Repeat Problem 14.22 assuming that the damping is prGponionai [0 stiffness of the sys:em where the constant of proponionalir), Cle = 0.2. Solve Problem 14.22 after condensing the three rotational coordi::ares. Repeat Problem ]':'.24 assurr.:ng 15% damping in all the modes. Derermine the steadystate response for the beam shown in Fig. P14. i5. Do no: condense coordinates. 'and neglect damping in the system.
14.20
2
4
fL3
14.21 14,22
Fig. P14.13. 14.15 The uniform beam shown in Fig. P 14.15 is subjected (0 <l constant force of 5000 Ib suddenlY applied along the nodal coordinate 4. lise rhe results obwined in Problem :4.6 to determine the response by the modal superposirion method. (Use only the two modes left by the static condensation.)
1"" lQ9(lbin:'l q "" 3.86\1b{in.)
r
100 11"1."1t100 in."1I
Fig. P14.15.
14.16
Solve Problem 14.15 using the results obtained in hoblem 14.7 which are based on the cor:sistent mass :orrnulauon.
14.17 Solve Problem ]4.15 using the results obtained in Probletr. 14.9 which includes the effect of the axial force in the stiffness of the system, 14.18 Detennine the steadystate response for the beam shown in Fig. P14.1S which is acted upon by a harmonic force F{f} 0;.7 5000 s;n 30t (lb) as shown in the
i
100 in. _ _I100
4
in.~;
/
__
s
Fig. P14.18.
443
15
Dynamic Analysis of Plane Frames
15.1
The inciusion of axial forces in the stiffness macrix of a flexural beam segment requires the determination of the stiffness coefficients for axial tonds. To derive the sriffness matrix for an axiaHy loaded member, consider in Fig. 15.1 a beam segmem acted on by the axial forces PI and p~ producing axial displacemems o! and 8::. at the nodes of the element. For a prismatic and Gniform beam segmem of length L and crosssectional A, it is relativeiy simple ro obtain the stiffness. :e!ation for axiai effects ":Jy [he apptication of Hooke's law. in relation to [he beam shown in Fig. lS.l. the displacements 0 1 produced by the force PI acting ::.t node i while node 2 is maimained fixed (Ol "" 0) is given by
!;i~
P,L
AE
(15.1)
From eq. (15.1) and [he defiol:ion of ~he stiffness coefficient ktl (force at node 1 to produce a unit displacemer:t, oJ), we obtain
P! AE == Ot L
. II
(15.2a)
The equilibrium of the beam segment acted GPon by [he force ko requires a force at the other end, namely The dynamic u:1alysis using lhe stl ffness marrix melhod for structures modeled as beams was presented in Chapter 14. This method of analysis when applied to beams re(!uires the calculatio:1 of element matrices (stiffness, mass. and damping matrices), the assemblage frorr: these matrices of the corresponding system matrices, the formation of the force vector, and the SOlution of the resultant equations of motion. Thes.e equations, as we have seen, may be solved in general by the modal superposition method or by numerical integrat~on of the differential equations of motioo. l!J (his chapter and in the following chaprers, [he dynamic analysis of structures mode~ed as fromes is presemed, \Ve begin in this chapter with the analysis of structureS modeied as plane frames and with the loads acting in the plane of the frame. The dynamic analysis of such strucwres reqllires the inclusion of the axial effec[s in the s:iffr:ess and mass matrices. It aiso requ~res a coordinate transfonnation of the nodal coordinates from eiement or local coordinates to system or global ::oordinates. Except for the consideration of axial effects and the need [0 transform these coordinates, the dynamic analysis by the sdfffless method when applied to frames is identical ra the analysis of beams as dis::ussed in Chapter 14.
442
AE
L
Analogously, the ocher stiffness coefficients ure
(l5.2b)
( 15.2c)
and
AE
L
(lj.2d)
Fig. 15.1 Beam elemenr with nod3.1 aXial loads Pl. Pl , and corresponding nodal displacemems 0:, 02.
444
445
The stiffness coefficients as given by eqs. (15.2) are the elements of the stiffness matrix relating axial forces and displacements for a prismatic beam segment, that is,
gives half of the total mass of the beam segment allocated at each node. Then for a prismatic beam segment, the relation between modal axial forces and modal accelerations is given by
( 15.6)
(15.3)
The stiffness matrix corresponding to [he modal coordinates for the beam segment shown in Fig. 15.2 is obrained by combining in a single matrix the stiffness matrix for axial effects, eg. (15.3), and the stiffness matrix for flexural effects, eg. (14.20). The matrix resulting from this combination relates [he forces P; and the displacements 0; at the coordinates indicated in Fig. 15.2 as
where m is the mass per unit of length. The combination of the flexural lumped mass coefficient and axial mass coefficients gives, in reference to the modal coordinates in Fig. 15.2, the following diagonal matrix:
P, P, P,
p,
AL'II 0 EI
 L3
symmetric
12 6L 0 12 6L
4L2
0  AL' II 0 0
0  6L 2L'
AL'II 0 0 12  6L 4Ll
0, 0, 0, 0,
Os 86
(15.4)
P, P, Po P, P5 P,
filL
2
0
81 82 8, 8, 85 86
(15.7)
P5 P6
To calculate the coefficients for the consistent mass matrix. it is necessary first to detennine the displacement functions corresponding to a unit axial displacement at one of the modal coordinates. Consider in Fig. 15.3 an axial unit displacement 01 = 1 of node 1 while the other node 2 is kept fixed so that 02 = O. If u = u (x) is the displacement at section x, the displacement at section x + dx will be u + duo It is evident then that the element dx in the new position has changed in length by an amount du, and thus, the strain is duldx. Since from Hooke's law, the ratio of stress to strain is equal to the modulus of elasticity E, we can write
du P (x)
15.2
dx
AE
(15.8)
The detennination of mass influence coefficients for axial effects of a beam element may be carried out by any of two methods indicated previously for the flexural effects: (1) the lumped mass method and (2) the consistent mass method. In the lumped mass method, the mass allocation to the nodes of the beam element is found from static considerations which for a unifonn beam
(15.9)
P, k"~j:======KO::"'~"';S"~========~JP, ~~=J.d:~ I=
du
"k"
b\\"'J f+dx
+ du
Fig. 15.2 Beam element showing flexural and axial nodal forces and displacements.
446
447
in which C is a constar.t of integration. Introducing the boundary cor.ditions, = 0 and H = 0 ar ;c ;;;:;: L. we obtain [he dispiacement funcrion U I (x) corresponding to .'! unit displacement OJ as
u = ~ a: x
as shown in Fig. 15.4. Hence the inlcmal work for element dx!s obtained fro!":l eqs. (tS_l3) and (l5. :4) as
dW, AEut (x)u{ {x.)dx
u, (x) = l 
(! 5.LO)
W,=
Analogously, the displacement funCtion
me::1t
HZ
B1 = I is
I)
WI~=
J,
f' AE!i((x),,;Gr)dx
(15.l5)
k"
(15.16)
The application of the principle of virtual work results if: a general express~ ion for ~he calculation of [he stiffness coefficients. For example, consider the beam in Fig. i5.3, which is in equilibrium with [he forces PI k 1j ar.d P2 = k;n at its two ends, Assume that a virtual Jispiacement S: 1 takes place. Then, according to the principle of virtual work, during this virtual displacement, the work of the ex'[ernul anc imemal forces are equal. The exterr.al force k21
perfonns [he work
In general, [he stiffness coefflcie:1l k'j foc axial effects may be obto'ined from
k" =
Using eq. (1).17), the reader may check the results obtained in eq. (15.3) for
'C
r'
(lS.l?)
a unlfolm beam. Howeve:, eq. (lS.J7) could as welt be used for r:onuniform beams 1n which AE would in general be a function of x. Ir. practice, the same dis;:>1acement U (x) <l:1C u~ (;I':) obmined foc a uniform beam are also used in eq.
j
or
05.12)
(l5J7) for a nonumform member. The displncemem y(x,t) at any section x of a beam element c:.te (0 dynamic modal displacements, 8 1 (r) and Dz ([), is obtained by superpos:tioo. Hence
( lS.l8) in which '" (x) and ",(x) are given by eqs. (l5.10) and (IS.ll).
slnce 8z ;;:;; I. The internal force P (x) a( any sec[ion x is obtained from eq. (i5.8) as
P (x)
AEu! (x)
([5.l3)
Now consider the beam of F:g. 15.5 while ur:.dergoing a unit acceleration, 51 (t) ;;:: 1 which by eq. (15. J 8) ;esults in an acceleration at x given by
iii (x,r);;:;;
HI
(X)SI (I)
in which !<: (x) = dUt Idx.. The relative displacement of elemem dx during [his virtual displacement is
or
du..,=dx

duz dx
(15.l4)
~ll{t)"1
m\l ...............
i=:;=:::
CD
============:;;Ji+
i
,, 'i I
.. ""
Mll
=
Fig. 15.4 Displacement a[ node 2 (~ '.~ l) of a beam element S:.lbjected to axial dis~ placement at node 1 (0 1 = I).
m n+L.. __ _
Fig. 15.5 Displacemem at .l.ode 2 (0; erarion llt node (81 (:) = 1].
aC(;el~
448
449
since 81 (t) = 1. The inertial force per unit length along t::'e beam resulting f:::om this unit acceleration is (15.19) where tnex) is the mass per unit length along the beam, Now, to determine the mass coefficlent m!h we give to the beam shown in Fig. 15.5 a virtuai disp!acemeot ~ = l. The only external force doing work during this: virtual displacement is (he reaction m2> This work is then
and
m I 2=m:!=
"L
Jo
m 1
x\/x\
\L,
'jdx=6
mL
(15.25)
In matrix form. the axial inertial force relationship for a unifoml beam may
be written as
(15.26)
Finally., combining the mass matrix eq. (14.34) for flexural effects with r.q. 05.26) for the axial effects, we obtain the consistent mass matrix for a uniform beam element in reference Lo the modaJ coordinates (t) shown in Fig. 15.2 as
or
(15.20)
since 02 L The internal work pe: unit length along the beam pei'fonned by t;,e inertial force Ii du:ing this virtual displacement 1S
P,
P, P,
fl40
niL
symmetric
]56
( ~,
1 p,
r=
0 0
70
22L
0
54
0
13L
2
8;: 8)
140
0
420
P,
, P, ,
156
3, 3,
Ii,
(15.27)
!3L  3L
0 22L 4L2..;
J, m
"
or, in general,
In,!
"'
Im
it
j"J
(15.23)
The application of eq, (15.23) to the SPecial case of a unifonn beam :esults in
mll:
_ (L _ { X ihL mllTJ dx  3
\ "
,2 _
( 15,24)
Similarly,
m::1;=
lfiL
The stif:;Jess matrl;; for the beam element in eq, (l5.'~) as \veIl as the mass matrix in eq. (15.27) are in reference to nodal coordinates defined by coordinate axes fixed on [he beam eiement. These axes are called local or element axes while the coordinate axes for the whole st:11cture are known as global or system axes. Figure 15.6 shows a beam element wim nodal forces P j, P1",", P (, referred to the local coordinate axes x, y. Z, and p;. Pl, .... P(, re:erred to giobal coordinate set of axes X, Y. Z. T..'1e objective is to transfonn t;,e element matrices (stj:fness, mass, etc.) from the reference of local coordinate axes to the global coordinate axes. This transfonnation is required in order that the matrices for ail the elements refer to the same set of coordinates; hence, the
450
451
y
y
I~:
t,;
( Pi
r cos e
: sin
P,
0 cos 8 0
sin
f)
0 0 0
cos
f)
0 0
0
0
,, 1
P,
P, P,
0 0 0
0 0 0
0
0
sin
(J
PJ
p~
(IS.3I)
 sin 0
cos fJ 0
J\ .
L })6
or in condensed notmion
{PJ
r.Z
= [1] (10,)
(15.32)
Fig. 15.6 Beam element showing nodal fOr(:e~ Pi in loc.)! (;.r, y, z) .)nd nodal forces ia globaI coordino:e axes (X. Y, 2;.
marrices become compatible for assemblage imo the system ma[rices for the
structure. We beglo by expressing [he forces (P I' P?. PJ ) in termS of the forces (Ph Pz> i\). Since (hese (Wo sets of for:::es are equivalent. we obtain from Fig. 15.6 the fol1owing relationships:
p,
=
in which {P} and {.o} nre, respectively, [he vectOrs of (he element nodal forces in local and global coordinates and [1] is the transformation matrix given by [he square matrix in eq (15.31). Repea~lng {he same procedure, we obtain the relarion between nodal dis~ pla::.:emems (Oh Oh ".,66) in local ::.:oordinates and the componems of the nodal displacements in global coordin::ues (51) 82, ... , 86), namely
sin 8
cos
o
o
cos
p cos 8 +
sin 0
(15.28) The Erst two equations of eq. {[5.28) may be wrinen in matrix omarion as
o o o o
or
o o o
0 0 0
sir. cos
e
e
 51_""1 {j
o o~1 ~,
l
8~
u
8,
(1533)
JI
(15.34)
(15.29)
"OW,
the substirution of (P) from eq. (15.32) and (0) from eq. (15.34) into the snffness equation referred to ioeal axes iP} = [Kj {o} results in
Analogously, we obtain for rhe forces on the other node [he relationships.
or
Ps=
P
sin 8+
cos I) ( 15.30)
0')
(15.35)
where [7]! is the inverse of matrix {T] However, as the reader may verify, the
452
M!Jltjdegree~:)f~Freedom
Systems
2
453
transformation matrix [1] in eq. (15.31) is an orthogonal matrix, that is, [11' ~ [1]7. Hence
(15.36)
6
'"
~
in which
( 1538)
Fig. 15.1 Plane frame for Example i5.:.
is the stiffness m2.trix for a beur:l segment in reference to the global system of coordinates. Repeating the procedure of transfonnation as applied to the stiffness matrix for the lumped mass, eq. 05.7), or consistent mass matrix, eq. (15.27), we obtain )0 a similar manner
&
0
0
0
0
0
01
0
in which
(1539)
[T.J
0
~'
~
0 0
I
L
a:ld for element
0 0 0 0 0 0
0 0
0
,/2
is the r:lass matrix for a beam segment referred to global coordinates and [1] is [he matrix of the transfonnat~on given by the square matrix in eq. (15.33).
&.
with
e= if IS
Example 15.1~ Consider in Fig. 15.7 a pJane frame having two prIsmatic beam elements and three degrees of freedom as indicated in the figure. Using the conSistent mass formulation, determine the three natural frequencies and normal modes corresponding to this discrete model of the frame.
Solution: The stiffness matrix for element by eq. (15.4) is
The mass matrix in local coordinates for eitber of the two elements of this frame from eq. (15.27) is
140
[Mil = [M,J
0
: 70
Symmetric
156
r 600
[K,1
~
2200
40,000
12
600 40.000
[K,l
1000
0
12
600
20,000
600
600 0 0
0 0 54 c 0 1300
0 140
1300
30.000
12
156
 2200 40,OOOJ
600
40,QOOJ
The element stiffness and mass matrices in reference to the global system of coordinates are. respectively. calcu;.red by egs. (15.38) and (15.39) For el
454
455
4
0306 0,294
 0,424
4
0306
2
Symmet:ic
The system stiffness and mass matrices are assembled by the direct method. As was mentioned before, j( is expedient for hand calculation of these matrices [0 indicate [he corresponding system nodal coordinates ac the top and right of rhe e;c:rnc:m matrices. We thus obtain the system stiffness matrix as
0,424
40,000
0306
1 : I
0,424 40000J
20000 0,424
[k] ~
424 0. 0,;76
1
J
and
~he
system
:TIaSS
matrix
:\$
148
2 Symmetric
[,\1] ~
l288 
8
]556
8 304
644
1556 644 80,000.1 The natural frequencies are found as ~he roots of (he characteristic equation
8
1556
148
1556
62 8
8
62
148
2
,~O.O{)O _
'[K] 
w' [MJ'
1556
which, upon substituting the values given for this example, yields 906  0,288w' 10' '294 + O,008,,}
:424  1.556u/
Lb.,
I
[1(,]
1
0,600
424  L556w~1
176  o.644w~1
Symmerric 0,012
40,000
80,000 
10'
iii
0,600
~ 0600 
o
2
SOu.ll
0,012
",'; 6385,
and the nmural freqt.:..encies are
w; = 976.6,
31.24 rad/sec,
~ = 4211.6
ar.d
4 4 Symmetric
0,600
w, ~ 25,26 radlsec,
W, ~
and
140 1 U;1,J ~I ~ , 70
Or
156
2
40,000
C 140
2200
3
4
I,
4,02 cps, J,
4.97 cps,
and
I, =
10.33 cps
o
54
o
~
The normal modes are given as [he nontrivial solution of the eigenp:"oblem
([K]  ",' [M]) (al = (0)
1300 30,000
0 0
0 [300
456
457
Subst.ituting w~ = 638"5 and setting all = LO. we obtain the first mode shape as
D8
l.00}
where
o
P,=
which is normalized with the factor in this example, the nodal applied forces are
(b)
F;
The normalized eigenvector is then
100.000 lb,
F, = 0,
F; = 0
H1
{,}=
f<h,,"
, "j
't, + 638.Sz,
2180
i, + 976.6z, = 498
z) + 4211.6:::; = 5830
The solutions of these equations are of rhe form
P,
(c)
f<py,}=jv
l33J
Zf=
yields
=
0.0218 0.00498
[]
l'
z,
z, =
Z,
05(0 (1 ~.
cos 31.25061)
cos 64.8970t)
(d)
= 1.384(1
Example 15.2. Determine the maximum displacement at the moda~ coor~ dinaEes of the frame in Fig. is.? when a force of magnitude 100,000 tb 1S suddenly applied at nodal coordir.me 1. Neglect damping.
Solmion: From Example 15.1, the natural frequencies are Wi = 25.26 fad/ sec, W:: ~ 31.24 rad/sec, and W;; &:k90 fad/sec; and the modal matrix is
0.0218 0.00498
['"] =
[<p]{d
which results in
0. 0583
I  0.0527
L
y, = 0.1577  0.0744 cos 25.26,  0.00254 cos 31.2St 0.0807 cos 64.9t
y,
000206 0.00341
0.1455
+ 0,1800 cos
25.26r  0.00105 cos 3L25t  0,0333 cos 64.9r 0.00174 cos 31.25, + 0.0022
COS
y,
=
0.000475
+ 0 cos 25.26r
64.91
(e)
458
459
The maxiroum possible displacemenrs at the nodal coordinates may then be estimated as [he summation of the absolute va!ues of the coerficienL<; in [he above expressions. Hence
:"_>'SS/LEH7i.h
SECT. :::W.::r<.T:;>,
4.Z0Q
too.co
100.00
(; .OC
0: .00
}'.'IT.;,., =03177 in
y, = 03598 in ",,1'1';
(l)
1I
15.4
Program 14 serves to determine [he stiffness and [he mass matrices for the plane frame and to store the coefficients of these matrices in a file. Since the stiffness anc masS marr.ces are symmetric matrices, only the upper triangular ponion of these matrices needs to be stored. The program also stOres in another file, named by the user, the general information on the f~ame. The informarion srored in these files is needed by programs which the user may call to ?erform dynamic analysis of the frame, such as determination of natura! frequencies or calculation of the response of the structure subject to external eXCitation, Example 15.3. Use Program 14 to determine the stiffness and mass trices for the plane frame show:! in Fig. 15.7,
Soli/rion:
ma~
2:. S4cn.cs
._ !BOGE.(:S
;;;0('.:.05
4.?4~H:~~
. SS::;;;2 .. ;)?
7. "9?;;E.')::
1 J400;;;1);I
S:.>SI;2.0J
H40E:~O}
<1.9999E,J<J
5S3~2,J:l
!f9H2,C.:.
Example 15.4. For the slruc:ure shown in 15.7 which was modeled in Example 15.3, determine: (a) naCJral frequencies and modal shapes: (b) the response to a force of magnit'Jde [00,000 Ib sudcenly applied at nodal coordinate 2, The solurion reqtdres the USB of files prepared during rhe exeCUlion of Program 14 lO model the StnIC!l/re as a plane frame followed by rhe execmion of Programs 8 and 9 ,'0 de!ennine natural frequencies and to obtain the response using che modal superposition merhod.
Solution:
cr
O.
.JO:rtrS
seAl~
NJ: ]
NE.t
NC~G ~lE.i)7
WM82R Of'
ELEHEN'tS
JO!~
X~COO!l?!:;ATE
YCOOP.o!t:;:rE:
':ON<:ttl'!'PA,20
)o'~5S
.;onn
0.00
7C.7t
C _GO
o .ona
c. Goe
c
OllB)
D.DS27J
, (JonC'
70.71 7:1.11
tLOC~%
G_
ilOl0~
nIL II
c coo
G;;~ .. :
:}0l,,~
1.0S:H:' 0.:124.!5
G.
460
461
(3) Define keypoints at the end of the two beam elements in Fig, 15.7:
GECM:STRY :; POI!\'1"S
> Pi:'
IJsing BEAM2D:
"1;)"';
'?1~S
fl''''l
O1~AV:':N"!ONAL
M??.O?
EX, lS7, DENS,
Q. 7
ReOKS?
L L S, 6,. lO{l, 3. 0, 0, C, 0, ::;
RCONST,
2,
~:.AX:. '.'1::(.\:1;
1'.A..'<; Ace
:f:ESHING ;.
7.cH 6.,!75
tl,~ll!l
lee2.
1,
2.
1. 2,
1. :
O.
~3.;;
:: .001
(9) App!y constraints in all degrees of freedom at nodes 1 and 3 and at node 2 for transiation Z and mtation X and Y:
L,OAJ_BC
:>
15.5
STRUCT(J!<'Jl.L
:>
DISF=.KN'TS
:>
:JND
DND, :, AL, 0, 2, 3
ON;:), 2,
UZ,
0,
2,
EX RY
Example 15.5. Use program COSMOS to model the plane frame showa in Fig, 15.7 with two beam elements and calculate the fi:st six natural fre~ quencies. Use the opdon for consistent mass matrix formulation.
Solution: MOS:
0,
.0JC:.)MPR=:SS
(11) Set options for f:equency calculation to ex.tract three frequencies using Subspace Iteration Method with a maAjnum of 16 iterations and the use of consistent mass formuiation, then run the frequency analysis:
.';';~ALYSIS
> FE2Q!B:)CK
:>
P,,FREQ<JSNCY
P.._FRSQ:r:SNCY,
1,
3,
S,
16,
0,
C,
C,
0,
1E5,
1..2>6,
0,
0,
462
463
LIST >FREQLI5T
0.5,
0,25,
FREQUE;/\,CY
t;UHBER
ri<.EQU.\lCY
:REQl:ENCY
(CYCLES/SEC) 0.34915:42+01
?ER!OD
(SECONGS)
(RA:l/SEC)
(15) Define the fo:'cing function and apply it as a force at node 3 b the X direc!:on:
ANALYSIS
ANAL.YSIS ),
!?D_CUFl:DEF, ?OST_0Y:>1
1 2 3
4
C "2193783+02
0 .2306635+02
0.2864:::87+00
O.36?1124E+Ol
(;.6541844E+Ol
5 5
8.8893843+01
o. 177S86H>,.02
~
.2153(:4,312+02
CCR'JES
PO_CURDEF
1,
1,
0,
100008,
0.2,100000,
0.2l,
8.5, 0
CON'.:RO~
;.. .",.CTIVB
.:<.CTST
AC'!'S'I'z 1'C,
1
:>
i..:ASERJE'T
LOADSBC ). STRUCTURAL
E'ND, 3, FX, L 3, 1
FCRCES ) F'ND
(Fig. 15.8 shows the p!o' for the structure modeled as a plane frame wi[h tOml of four beam elements.)
Example 15.6. Use [he program COSMOS w obtain th::: displacemenr resoonse at node 3 of the frame IT.odeled in Exampte 15.5 when If is st!bjected to ; force of magnitude 100,000 lb applied suddenly for 0.2 sec at node 3 ~n [he X direction. Also determine ::he forces and s[resses on the beam elements modeling this frame.
Solurion: The following commands are implemented in COSMOS after the execu~ion of the commands described in Example 155:
CONTRe:.. ) LPtS2RJET,
DEVICES
> Lfo,$2RJc.:T
150,
~he
plot
ob~a:ned
)7t..
v
,,,, , , 5. . . .
15&80
.
,
r
'r~'
:
i
Fig. I5.S Modeled pla:;e frame structure of Example 15.5.
,",.
2&flli&
Jg$$iJl
4{j(ih<)&
,: I
IL 15
,:,
, :<
'.5
464
465
P;)_P;:;!NT
?D_PRINT,
:, 0,
0, 0,
0, 1, 1, :"00, 1
(20) Detennine the stresses on the beam elements and use an editor to obtain from the output file (EXlS.OUT) a print of the forces and stresses at seep #40 corresponding to time ( = 0.20 sec,
.A..:.'1AZ"YSIS
tCSTRESS
ANALYSIS
'?D~PLO'l',
lr
POS't_DY:i
CONTKOI., ) uTILITY
>
SYSTE.>1
1,
lCe, L
:SDIT EX:.5.0UT
OU!put
R_DYNAI'1::L :::
15.5 SUMMARY
The dynamic analysis of plane frames by the stiffness me~hod requires the inclusion of the axial effects in the system matrices {stiffness, mass. etc). It aiso requires a transfonnation of coordinates rn order to refer all the element matrices to the same coordinate system, so that the appropriate superposition can be applied to assemble the system matrices.
TABLE 15.1 Forces and Stresses at the beam elements of Example 15.6 at step "# 40 corresponding to time t=O.20 sec.
(19) Ob,ain a plot of the displacement response at node 3 in 'he X and Y directions:
DIS2LAY ,.
".CTXY?LOT, ACTXY?LGT,
XY_?LO~S
,. ACTXYPLOT
3, 3, P 1, 10, ..!., 0, 0, 3X 3y
L
2,
::..:; 0
.J
;: i
~28<)e04 0.!;2e~~r;~
(Fig. 15.10 shows the plot for the dispJacement response at node 3 in the X and Y directions.)
')'o;)
~C:)7;;;+C'\ 8Y;;)::~OQ
.4S)7<:~Q4 JUlC~:'Ot
!(S",r,.OOQC">00 ?Gcro;:,t;!,l
?!:"O.:2~9~O"i 0,!;41S~~""
:~SfSS:
'/~~O
(M;:i5~i~
'.4
~.J2
i',u,
\ i,D,
"
?".~,:, <.7S9::~JS
;
""r }c&c::~~a J.cci;tl"";.;;D xs,CC03Ct';;C o.co:cs; .... ro
x;:~ .}5~S::+?! JC'J2~GC ~r76E'~S
.<'7590
_=H2::;~~"
:?!Ai
~~35n:;;_:~
'.~:;'S;;;~:::~
~.~OCC;::~OD
S 24
.!S
,.,a
(L~fl
, ~~~~:::
'\
V "
r==:t\ /
j
V
"'..
Vs,,j
>;~2204
:Ms!Sn"J.(j00~2:_t)
i'
~/
l~ 1 _.
\r::.~G
JCGC::~:;O
o.:);;{:c;::.:::o
(Y:.j$;:)~~."t~l;1i:~C}
c..Hn~",
",~
(7~~C7~/.:.,)~t
?JtCG;::.Q,'
SN'.x "~.~(:';"E:<)~
.ns.;::.:;q
5211t:.t4
S"l:1 ".511tSO{
,
0.4
'\
[/
r
;/
'"
" j.
:. 1
.:,
,:,
, .3 (LiS
I'L3S
.: 4
,:;
Q.45
~.$""
.3!!J;a"~4
C.J!!W;'>J;
r.:)C;;GF;.(lI)
y.sz~
:'O;;CB';
('00JS~C:) t~3~S""Ji
(YsiS';; .. C.CCOC;;<Hl
(~~~/s:;;;",.S1seS"'O; s:~~X~" ~9SES;;S
'1~,,).:rcJ2~().O
X:;;~.3~9i::;;~;JS
T! "E
i'i"r'C70;:t_~"I~C
:ccn;''}:J
)CC}2C;;i
Fig. 15.10 Displacement response in the X and Y directions at node 3 for Examp:e 15.5.
.
. ~~S"'2';<,
466
467
The requjred mmrices for consideration of axial effects as well as the matrix reouired for the trilnsformatlon of coordinates are developed in this c~aptef. A
15.5
~o
Th:s program is organized foilowing the p,mero of the BEAM program of {he preceding chapter.
Deterrntne the max.~mum response of the frame shown in Fig, PIS.; when subjected (0 ihe triangulnr impulsive [ood (Fig. PL5,5) along (he nodal cootdi~ naie 2. Use results of Problem 15.3 to ocrain rhe r:",odal equarions and use the appropriate response spectrum fO find maximum mOGtll response (Fig. 4,5)_ Neglect da:nping in {he system.
F(tl !
1.0'1: i
PROBLEMS
T~e following problems are imended for hand calculation, though it is recom
mended that whenever possible solutions should also be obtajned using Pro2ram 14 w model [he structure as a plane frame, Program 8 to determ:ne ~atural frequencies and modal sh:pes, and Program 9 to calcuiate the response using modal superposi[ion me(hod.
15.1 For the plane ffame shown in Fig. P1.5.: determine the system stiffness and mass matrices. Base the analysis on the four nodal coordinates indicated in the figure. Use consiSlem mass methoc.
Fig. P1S.5.
3
2 ....,..:4'+ict__ Fltl
15.6
Derermine the !lreooYS((lle response of the fn:tme shown in Fig. PlS.l when subjected to hnrmonic forcl':: F(I) = 10 sIn 30f(Kip) along nodal coordinate 2. Neglect damping in the system. Repeat Problem J5.6 ass:Jming thm the damping is proponio:1ul to the stiffness of the system, (C1 = au (KJ, where (iij = 0.2. The frame shown b Fig. PIS.S i:> acted upon by (he dynamic forces shown b ~he figure. Determine {he equivalent nodal forces corresponding to each mem~ ber of (he frame. ncdal forces which were ca!cu!,Hed io Problem ] 5.8.
15.7 15.8
15.9 Assemble [he system equivalent nodal forces {FJ from eqUivalent membet
15.10 15.11 Determine the narurflJ ftequencies and corresponding normal modes frame shown in Fig. PIS.S.
fOf
the
Fig. PlS.I.
15.2 Use the results obtained in Problem 15.: in perfor:ning the Srlltlc condensation to eli:ninate ~he rotational degree of freedom al the suppOrt to de~ennlne the ~ransforma[ion matrix and the reduced stiffness and mass ml:ltrices. Determine the natural frequencies and corresponding normal modes for the reduced system in Problem 15.2. 15.4 Determine the response of the fl1lme shown in Fig. P15.1 when it is acted upon by a force F (f) 1.0 Kip sudde:lly applied at nodal coordinate 2 for 0.05 sec. 15.12
Determine [he ~esponse for [he fro me shown in Fig, P15.t l{a) wilen subjected to Ihe force FJ(r) (Fig. 15.Il(b)] aCling along nodal coordinate I. Assume S% damping in all the modes, Determine the steadystate response of the fra:ne in Fig. P15.LO acred upon harmonic fotce F; (!) 10 cos SOf(Kip) as indicated in the tlgure. Negle:::t damfJing in the system Solve Problem 15.1 t usmg 19), !Xeglect damping.
step~byM!;[ep
15.13
15.3
; >
, " ~ ,
!p
15.14
Determine rhe response of [he frame shown in Fig. PIS,} whet: acted upon by the force F(f) {depicted in Fig. P1S.14 applied ar nodal coordinate 2. Assume 10% damping in all ~he modes. t,'$e modal superposition method.
468
16
iO f(rHKip)
50 in.
A"'10in 2
12 '" 104 iksH q"" 386 X 10 5 Klin. t"" 100 in4
45"
Fig. PIS.S.
2
100 Ib/ft
w "" HXlib/ft
E""
1.1)
X loJ Ksi
I
1 "
A"" 8 in
fFig, PIS,IL
(.,
Fig. P15.14. 15.15 Find the response in Problem i5,l4 using stepbystep linear acceleration method (Program 19). Assume damping propodonal .to stiffness by a facror an = 0,01.
In Chapter 15 consideration was gi yen to the dynamic analYSIS of the plane frame when subjected to forces acting on the plane of the structure. When the phwar strc.lctural system is subjected to loads applied normally to its plane, ~he structure is referred to as a grid. This S[fUcture can also be treated as a special case of the threedimensional frame to be presented in Chapter 17. The reason for considering the planar frame, whether loaded in its plane or norma! to its plane, as a special case, is the jnunediate reduction of unknown nodal coordinates for a beam element, hence a conside,rable reduction in 13e number 0: unknown displacements for the struclUral system. Vl~en analyzing the planar frame under action 0: loads in the plane. the possible components 0: joint displacements that had to be considered were translations in the X and Y directions and rotation about the Z axis. However. if a plane frame is loaded nomal to the plane of the structure, the components of joint displacements required ~o describe the displacements of a joint are a translation in the Z direction and rotations about the X and Y axes, Thus treating the planar grid struCture as a special case, it wiH be necessary to consider only three components of nodal displacements at each end 0: a typical grid member.
469
470
471
16.1
For a beam element of a grid, the local orthogonal axes WIll be established
s:.lcn that rhe x detines the longitudinal centroidal axis of the member and the xy plane wi:1 coincide with rhe plane of the structural system, which wijj be
defined by rhe X~Y plane. In [his case, the l axis will define the minor principal cuis of the cross section w~lle (he y axis wiU define {he major axis of the cross section. It wiil be assumed ~hat the s!:ear center of (he cross section coincides with rhe centroid of the CiO$$ secrioo. The grid member may have either a vunnble or constant crOSS section along its length. The possible nodal displacements with respect to the local or to the global systems of coordinates are identified in Fig. 16.1. It can be seen that the linear displacements along the z direction for local axes and along the Z direction for the global sys:em are idenricnl since the tWO axes coincide. However, in general, rmational components at the modal coordinates differ from these twO coordinate systems. Hence, a mmsformarion of coordinates will be required [0 m:msfonn the element rr.nrrices from [he local w rhe global coordinates,
=
du
dx
dS
T
( 16.1)
ax
y
i,
\
"
0,
6.
.....'
~
5,
\a)
0,
in which 8 is the angular displacement, T is (he tOrsional moment, G IS :he modulus of elasticl(Y in shear, and j is the torsional constant of l:.:le cross section (polar moment of inertia for circular sections). ,As a consequence of the analogy between eqs. (15,8) and (16.1)., we can nte . the fOHowing results already obtained for axial effects. The displacement r:n~nor.s for :he ~orSlonQl effec:s are the same as the corresponding functions gJVlng _the dispt<lcerr.ems for axiai effects; hence by analogy to eqs. (15.10) and (1),1!) and in reference to [he nodal coordinates of Fig. 16.2, we obtain
:v
8, (x) = 1
[I
'I
x\
(16.2)
"
X
and
"
5,
0
'.
x
5,
0, (x)
(16.3)
in which (he angular displacement func~ion 8 1 (x) corresponds to a unit anoular cisplacemem 8 1 = 1 at nodal coordin<!te I and B2 (x) corresponds to the displace
It;,
z.,
Ito)
Fig. 16.1 Components of nodal displ;."lcemcnts for a grid member. (a) Local coordin:ue system, (b) Global coordinare system,
472
MuJtjdegre~ofFreedom
Systems
473
mem function resulting from a unit angular displacement 8;. = 1 at nodal coordinate 2. Analogous to eq. (15.17), the stiffness influence coefficients for torsional effects may be calculated from
k;;
a grid structure. In reference to [he local coordinate system indicated in FlO'. 16,I(a), the stiffness equation for z: unifonn member is then e
Jo
r'
fG8:(x)e;(x)dx
(16.4)
in which 8; (x) and tt.;. (x) are the derivatives with respect to x of the displacement functions 81 (x) and (J? (x). Also analogous to eq. (15.23), the consrs!ent
mass matrix coefficients for torsional effects are given by
r
P, P,
.TGL'IE!
0
Symmetric
4L' 6L
0
E!
 LJ
on in which 1m is the polar mass moment of inertia, per unit length along tbe beam etement. This moment of inertia may conveniently be expressed z.s the product of the mass in per unit Jength times the radius of gyration squared, e". The
radius of gyr.,;.tion may. jn tum, be calculated as the ratio lolA. Therefore 1 the mass polur moment of inertia per unit length 1m is given by
( 16.6)
m;;~
r'
(
~
0  .TOL'IE!
12
0
JGL'IE!
0
P,
P6 )
!,,8,(x)8;(x)dx
(16.5)
0 0
2L' 6L
 6L
4L' 6L
12
12)
(16.9)
or in condensed fonn
{Pi
[KI{o}
(16.]0)
in which 10 is the polar morne:1t of jnert~a of the crosssectional area and A the crosssectional area. The application of eqs. (16A) and (16.5) for a uniform beam yields the stiffness and mass matrices such that
r P!
P,
1
filL
140!o1A
Symmetric
4L'
fr,) = fG
)r"
and
I llf~'1
 I
1jl
P" p,
P,
,=420
o
70lc iA
22L
156
o,!
(16.7)
140loiA
o
o
P, )
3L'  DL 13L 54
or in concise notation
(: 6.8)
in which llit is given by eq, (16,6). and T;, are torsional moments at the ends of the beam segments shown in Fig. 16.2 as Pi and P2
(16.12)
~n which [.iW,J is the mass matrix for a typical unifonn member of a grid structure,
;me lu~ped mass aUocatio!l to the nodal coordinates of a typical grid member IS oba!:led from static consideratiol1s_ For a '.lfliform member having a distributed mass along its lengtb, the nodal mass is simply onehalf of the total
474
Framed
Struc~!Jre:s
Mode!ed as Discrete
~Jl!_liidegree~ofFreedom
Systems
475
p~nding
rotational mass 1",L. The marrix equatIon for the lumped mass matrix corres~ to [he torsional effects is LlJen
([6.13)
The combination of the lumped [orsional m2.SS matrix from eq. (l6,D) wi[h [he flexural mass matrix for a typical member of a grid results in the diagonal matrix which is the lumped mass mnrrix for the grid elemem. This matrix, relating forces and accelerations at nodai coordinates, is given by the following
y
equation:
rhJA)
o
( 16.[4)
Fig. 16,3 Compmreuts of the noual moments in tocnl and globrtl coordinates.
o
Of
p~
briefly
(16.15)
?j=
p(:,=
P. . sio
Pf,
fJ+P S C05 8
(l6.16b)
in which rM J is, in this case, the diagonal lumped mass matrix for a gnd
member.
The identical fomI of these equadons with those derived for the rransformarion of coordinates for nodal forces of un element of a plane frame, eqs. (15,28) and (15.30), s~ould be nored. Equations (16.16) may be wriHen in matrix
nO(a[ion as
16.6
TRANSFORMATION OF COORDINATES
The stiffness matrix, eq. (16.9), as well as the consistent and [he lumped mass matrix in eqs. (l6.11) and (16.14), respec"vely, are in reference [Q the local sysrem of coordinates, Therefore, it is tle:::essary to transform the reference of these matrices to the global system of coordinates before their assemblage in the correspondi!1g marrices for the structure. As has been indicated, the z axis for the loca~ coordinate system coincides with [he Z axis for the global system. Therefore, &..e only step left to perform is a rotarion of the coordinates in the xy plane. The corresponding ma(rix for this transformation may be obtained by establishing the relationship between components of the moment at the nodes expressed in these two systems of coordinates. In reference [0 Fig. 16.3, these relations when written for nDde (i) are
sin if
cos
0 0
I
o o
o
cos $
 sin 8
o o o
o o o
sin f)
0 0
0
(l6.L7)
cos
f}
o
or in shon notation
(16.18)
P:;
(16.16a)
in which {P} ana {P} are, respectively, the vectors of the nodal :orces of a typical grid member in iocal and global coordinates and [11 the Transforml!rion
476
477
matrix. The same transfonnation matrix [71 serves also to transform the noaal components of the displacements from a global to a local system of coo:x:linDres. In condensed notation, this :elation is given by
{o}
h",,125hJ. 10 [n. 2
[1] (iii
(16.19)
vlhere {a} and {8} are. respectively, the component'i of nodal disp{acement in local and globai coordInates. The substitution of eqs. (16.18) and (16.19) in the stiffness relation eq, (16.10) yields the element stiffness mat:ix in reference to the global coordinate system, that is,
[1]{P'
L =60 in. 1== 100 lr..4 J"'"200 in.4 m"'"' 10 lbsec"2/ 1n ,2 c""30x lOOps? G 12X 10 6 psi liff"" 125 lb sec z
;;_'4
F3
50001b
"
[K][1]{/J}
or, since
111
{,o}
(1]T[K][1]{8)
If we define
we obtain
[kJ
as
[KJ = [TfiK]
(16.20)
Fig. 16.4.
{16.21)
(T,] =
fA with
e=(f' is simply th
[K,J
: . e un.' matnx
{P} =
in which
[,Wi {S}
(16.22)
= [T,j'[Kj [T,]
I 0 0 0 0
J.\ with
e= 90"
0
is the transfonned mass matrix, Example 16.1 Figure [6.4 shows a grid structure in a horizontal plane consisting of two prismatic beam elements with a total of three degrees of freedom as lndicated. Detennine the namral frequencies and corresponding mode shapes. Use the consis[enr mass formulation. The stiffness matrix for elements I or 2 of the grid in local coordinates by eq. (16.9) is
[T,]
[1
0 L 0
i
0
I 0 0 0
0 0 0
0
0 I
0
I 0
0
0
O.
lj
11
so that
[K,l = fT:1T[K:]
[T,]
r
[K,J = [K,J = 10'
40
0
200
40
0 0
I
L
0 0 40
0
0
5
0.167
0
0 100
5
 0.167
200
5
0
100
s
0
40 0
0
0
5 0.167
= iOti
5
 0.167
200
5
0 5 100 0
2 0 40 0 0
40
0
3 5 0 0.167 5 0
4 100 0
4 0
sl
5 200
0
5
0167
5
T
4
4
3 4
478
479
fK t ]
0
and
[K1 J is
5
[KJ
= 10'
l
c
240
0
5
w; = 396.35,
[hen
UJI
! 0,402,
and
= 23,866
5 240 0.333 5
19.91 rad/sec,
w" =:
101.99 rad/sec,
and
Wj
!S4A9 rad!sec
I !
[M,]
= [M,J =
'
0 0
0
20,570
1886
1250 0 0
0  15,430
oj
11 i4;
:886
223
1114 0
20,570
1250
1
0
 15,430 1114
0
1114 77
l
since
2500 0
0
7~ I
LOOO
[a J
1.000 1.000 0
 1.0001
LOOO
1.000 i
7,765
1886j
L 54,285
1886
213
The eigenVectors are conveniently no:malized by divldir.g [he columns of the :nodal matrix, respectively, by [he factors
[M,)=[M,J
i{a,}T[Mj{:;'J ~ 97475
,{a,}' [/,1,] {tl,}
214.81
[T,)
and analogously
= (I]
3
4 4
2
20,570 0 2500
0
so that
 1886
15.430
0
1250
0
1114l
188~
[M,J =
0  1886
15,430
0 lli4
0
1114
Q, 2
223
!Il4
0 1250
0
20,570
1886
0
0
0
77
0 2500
223
"j
[<PJ = lO
3
4
,,I
I
0.1026
0.1026 5.5691
0,4655
0.4655
 083201
0.8320
64603
4
4
Example 16.2.
when subjected
[0
Detemline rhe response of the gJld shown in Fig. 16.4 a suddenly applied force F) = 50001";) as indic8.(ed in the
figure.
SOiUlio,,: The r:aturaI frequencies and modal shapes for this Structure were calculated ie Example 16,1. The modal equation is given i!1 genera! as
From
lM,]
23,070
[M,]
o
23,070
:886]
 1886
0
1886
 1886
446
where
The natural frequencies and mode shapes are obtained hom the solution of [he eigenproblem
([K,]
q,,,,F,
480
481
and Fi the external forces at the nodal coordinates which for this example are F, = F, 0 and F3 ~ 5000 lb. Hence. we obtain
analyslS, such as determination of natural frequencies or caIcu1ations of the response of the structure subjected to external excitation.
z, + 396.352,
Zl ~ 23,86623
~
27846
323.01
Example 16.3. For the grid frame shown in Fig. 16.4 and analyzed in the previous examples, (a) use Program 15 to mode! this structure, (b) use Program 8 to calculate :he natura: frequencies and mode s:,apes, and (c) use Program 9 to determine the response to a constant force of 5000 lb suddenly applied for 0,1 sec as indkated i<J the figure,
So/ulion,'
27846 " =    ( 1  cos 19.911) 396.35
Problem Data:
Modulus of elasticity:
E = 30 X 10' psi
G ~ 12 X ! 0' psi
ih
323.01
= 10 lb sec' !inlin
I, = 125 'n'
{y)
 0.1026 0.4655 0,4655
[<P]{z)
1=200 inl
{~
0.1026
5.5691
100 in'
A = 10 in'
and finally
y, = 10 3 (
.
19.91t
",l/:~33ll:
NL!"~3:;:R
t:,;>.snc:::'Y'{
:)t:{)t,t.!'s
Cl' "W10,>"(
2E:r,7
16.7
>:':;C'O:i<;)IK::'7S
,
S"lUt
l'IRS:r
v0;:~"7
J.Cj
!>J
,co
Program 15 calculates the stiffness and mass matrices for a grid frame and stores the coefficients of these matrices in a file. Since the stiffness and the mass m2.~rices are symmetric, only the upper triangular portion of these matrices :leeds to be stored, The program also stores in another file, named by the user~ the general information 0:1 the grid frame. The information stored in these files is needed by programs that rhe user may call to perform 2. dynamic
3.00 0.::0
6~
0, )c
:;,00 0,00
.cc
stttm~
:0;;:';'
!'~".SSi
L:::t.~,r;
5SC':":OX
i'iSC7rCN
Ai\]';",
7J"S!Ce:;L
COt':S?"''''':"
:;"r;l',TZA
:0. CCj
10. c:::
::;:; ,;;;:
:r:;: .YO
: c .CO :0 J
::.;'c ,.){o
ZC;;,00
482
463
~AX_
:l!S?:".
."AX. V=='C.'C.
,~"X
. .;CC"
0.001.1;1
o.c}Crif
0.0)09
G
sn~
COil:
o.c%~oc
);):,J?
" "
Tne results given by (he compurer :;ompare very closely to those obtair::ed in Examples 16.1 and 16.2 by hand calculations. For example, the displace" men~ given by [De computer (not printed) ror nodal coordinate 3, a rime 0.1 sec, is ?l(r=0.1)=0.0568 in while the hand cakulation from Example 16,2 is
~Oi)o!!:.oa
O,JCC:;~O(l ~:1!1<J .. GB
S, {JJ0Ge.C6
:)0002.0
s. COOCE~C6
5 COOOE.06
S.00002.%
;11lS,05
0.00;)(,,;;'00
Ci)CCSCO
.~asn;.o}
J.)OIlE.04
1..seS7+i}j
y, (r = 0.1)
= 0.04 
0.0568 in
f.lC::Nl.lAU)2S,
16_8
) .li3
stCi2NVCTORS B~ ROO1S
2~.
S8,
Example 16.4. Use t:,e p.ogram COSMOS to determine the first SlX natural frequencies for the grid frame shown in Fig. 16.4. Model the strucwre using BE;\yi3D with a mesh of six elements.
wO,0010;'
C.OO~S6
0.0n01
0.0$5;;9~O"G::J)OC (LC6~6C
Q.OOMii
c.eeen
Solution: COSMOS:
VIE1C!?AR " 0, 1,
VIE'I~
PL;'.N2,
roSlCE :, :::OO:U:. ; ht1:t:I!$
Z,
2LA..~E
;>c:o:n:::
(3) Gene:ate keypoints at tr_e three joints of the grid frame and a fourth keypcinr at the principal plar.e of the c~oss section of the beam elements:
GEON2TRY :> POINTS Fr, 1, 0, 0, 0
?T, 2, 60, 0, 0 P'l', 3, 0, 60, 0 PT', 4, 0, 0, 10
> ?'1'
464
485
GEOMETRY ) CURVES
CRL:;:::H::, 1, 1, 2 ::~RLINE, 2, 1, 3
CRL::>rE
)
, 4:
) EGROU?
l,
B~f!;113D,
EGROU?,
MPROP,
::, ,SX,
3086, GXY,
12.E6,
DEI,,'S, 1.0
PROPSETS ) RCONST
RCONS?, 1,
1.,
10,
10,
100,
leO,
3.36,
},36,
IJ.
5
.~
0,0, zOOrc', 0
(8) Generate a mesh of six beam elements with two nodes along curves and 2:
NESHING )PARk"L~ESH
r
;. l! __CR
peeR, 1, 2 ( 1
:3
6,
1.,
Fig. 16.5 Mathematical model for the grid frame of Example J6.4.
1, 80,
:>
1, 0.C001,
15
ANALYSIS
MESfL!NG NO;)ES ,. r;COH?RESS
>
FREQ/BUC::: ;. :ECfRSQUE:NCY
?_?::.EQGE!JCy'
KCOMPRESS, 1,
(10) Apply constrainl' in all degrees of freedom at nodes 7 and 14, and transiations along the axes X, Y and rotation about the Z axis for all other frame of nodes of the beam elements, (Fig, 16.5 shows the model for the Example 16.4):
DND, 7, .;:.." 0, 7, 1
FREQUENCY
Kt:r1BER
FREQUENCY [RA.D/SEC}
?:REQUE~CY
PERIOD
(SECONDS)
1
DND, 14, .;".:.."
uN!), 9, UX,
0 1963908E+C2
O.7779C73E+D2
(C'fC:::":SS/SECi O.312S555E+Cl
O.123S078~+02
0.3199328+8C
I}.
0, :4., 1
:;, 14, 1, UY, QZ
2 3
S07703501
O.108592SE+D3
C.17:283092+02
C.5786004EOl
It should be obserled that only the first natural frequency calculated by COSIV10S using stx elements per beam provides a value dose to that determined in Exampie 16.3 uSlng a single element per beam. Example 16.5. Use the program COSMOS to obtain the response of me grid frame of Example 16.4 subjected to a suddenly applied force F = 5000 lb z:s shown in 16.4.
0,
H>Q05,
lS0:Jo,
486
487
Solution.' The following commands are implemented in COSMOS following the commands to solve Exarnple 16.4;
([4) Define the type of dynamic analys~s:
p,~ALYS1S
.,
~"$64
/'
/
"
:L&48
\ \ \
I
e,$lZ
/
/
! I I
'"
:;
pcs'r._.oy~
?D_,:',TYP
PD_A_TYPE,
2,
6, 50,
0, 0.01,
C,
0.5,
0.25, 2.
j
, ,
A
>ll,~15
/'
if. the
&.C:6
"rJ .el:?
/9.(148
G.GG~
\ \ \
7iI
,
>
,:,
i
,'L!5
, .,
i
/.
iL2S
"f r .:, :,
.:i
AC!'SE'I
Fig. 16.6 Dispb,cemem response in [he zdirection foc node I for the grid frame of Example 16.5.
LOAD_Be
FNC,
>
1, FZ,
(20) Set range between 0 and 0.5 sec for (he time axis and between  0,08 ar:d 0.08 for the UZ axis:
D1S:'LAY
:>
XYPLOTS
>
XYRA.;~GE
PD_PRnn, 1,
}l.NA:::'YS!S )
PD_PLO~,
0,
a,
0,
1, I, SQ, 1
j
ZR.A.NGS,
1,
0,
0,5,
0.08,
0.08
?OST_DYN ~ OUfPUT
REPRINT
PD_PLO'!'
1, SO, 1,
:>
a
:>
(21) Obtain a print of the displacement piOL (Fig. 16.6 shows rhe displacement response at node l in [he Z direction.)
CONTROL:> DEVICES;. LASERJET l
ANALYSIS
?D_~RESP,
NRESP
LASERJE>r,
50" :::
POS~~DYN ~
K_DYNA~lC
at
(22) Use an editor to list from [he ompUl me, EX16.0lJr, the forces and stresses at the nodes of the fi::s[ three beam elemems at [he time for srep No. 15 (r 0.15 sec). Table 16,1 reproduces the output forces and stresses for the first three beam eieme:1ts of (he grid frame for Example 16.5.
16.9 SUMMARY
This chapte: has presemed the dynamic analysis of pJane frames (grids) supporting loads app:ied normally [Q lts plane. The dynamic aoalys:s of grids
(19) Activate XY plOt :nformatloo for Z displacement at node 1 as a functiO:1 of rime, and plot (his function:
DlS?GAY > XYPLO~S > ACTXYPOST ACTXYP8ST, 1, TIME, ~Z, 1, 12, 1, 0, iN
DISPLAY XYPLOT5 ~ X:PLOT XY?LOT, 1
, The
p!ogr<tn~
386CO!'\F.EXE
the s.ystem.
488
489
TABLE 16.1
Forces and Stresses at Step 15 ((=0.15 sec) for Beam EI ements 1, 2 and 3 of the Modeled Grid Frame for Example 16.5
S":'?ZSSRS
~;o0:::: ~:?nc;:
'is,,C. H ?$r~N
>~~C.OOCCdC
:;:";5+:;'';
O.Gjc:,z+t)~
C. D()CCS.GQ
.:::?~:i::::.(lS
1YS!SE ~
~.
COOO,> ;'0
m:iS~i"~
.4&%:::C:
as(so:;
'!sd),'S2'tn:N
'i~~C.I);;DD'>CJ
5197Z.('>"
O.CO!)CS.(;C'
;y":&:),,J.i::>QCf:<GJ C,')Ovr;S.1)C
C~~;,k'=+.49~4S~,j >::<;~S<:
fig. PI6.l .
r~~t
. GJQn:~()O
)~<l0;;OO
GCFCO',>Oo
~99SSN
7" ..
':5~:
}'.~ ~
.G~41S<)'
::r.61r.9z>OS
J.CJC~E.cn
{?/A)
~:),C0002.()G
CYDCOS'C:,
(: {l{lO:;::C}
:~.")S~Q4
'Js,c. c:)Sr~c,;
S't,,)
;; .\lClO"S)C
. OC()C,>OO .2:>2,;0;. OS
:Ys!Ss;,,().:lOC~E:.Oti :1:tfS~)~:::
.:322.,.!)S
39t::J:>J1
16.4 16.5
Determine the natural frequencies and corresponding normal modes for the grid anHlyzed in Problem 16. L Determine the response of the grid shown in Fig. P;6.1 when acted upon by a force F(t) = 10 Kip sudden:y applied for one second at the nodal coordinate 3 as show;: 1:1 the figure. Use results of Problem 16,2 to obtain the equation of thot:o:; for the :::ondensed system. Assume 10% modnl damping.
requires the inclusion of torsional effects in the element stiffness and mass matrices. It atso requires a transformation of coordlnates of the element matrices previous to the assembling of [he system matrix, The required matrices for torsional effects are developed and a computer program for the dynamic analysis of grids is presented. This program is also organized along the same pattern of the programs in the two preceding chapters for the dynamic ana:ysis of beams and plane frames
16.6
16.7
Use results from Problem i 6.4 to solve ?robiem 16.5 on the basis of the three nodal coordinah:;s as indicated in Fig. PI6.L Determine the steadystate response of the grid shown i:1 Fig. P 16.: when subjected [0 harmonic force F(l) = 10 sin SOl (Kip) along noda: coordinate 3. Neglect damping in the system. of the system, [C] =oo[K], where 00=0.3.
16.8 Repeat Problem 16.7 assuming that the damping is proporrional to the s:iffness
16.9 Determine the equivalent nodal forces for a member of a grid loaded with a dynamic force, P (t) = Pr/(c), unifor:nly disrributed along ils length. Deter;nlne the equivalent :1odal forces for a member of a grid supporting a cOGcentroted dynamic force F(:) as shown in Fig. PHilO.
PROBLEMS
The following p:obiems are intended :or hand calculation, though it is recommended chil[ whenever possible solutions 5hould also be obtained using Program 1.5 to model the structure as a grid frame and Prog,ams 8 and 9, respectively. to cz.!cuiate natural frequencies aGd to solve for the response. 16.1 For the grid shown rn Fig. P16.1 determine the system stiffness and mass matrkes. Base the am:.:ysis 0::> the three nodal coordinates indicated in the figure. Use consistent mass method. Use static condensf:lrion to eliminate the ~otational degrees of freedom a:1d detem:ine rhe transformaricn matrix and the reduced stiffness and mass rr.atrices :n Problem l6.1.
Dere~mine
16.10
16.2
16.3
Fig. P16.10.
490
Mu!tideg(eeCrf~Freedom
Sys!ems
The following problems are intended for computer sollaion using Program 15, to model the Structure and Program 8 :0 determine natural frequencies and nodal shapes and Program 9 to caiculate the response. 16.U Determine the nawrol frequencies and corresponding nonnal modes fer [he grid shown in Fig. Pl.6.L 16.12 De(ermine {he response of :he grid shewn in Fig. PI6.! when acted upon by the force depicted in rig. P16.12 flc(ing along nodal cooroinate 3. Neglec[ damping in (he system, 16.13 RepeJ.[ Problem 16.12 for 15% damping in all the modes, Use modal super position me[hod, 16.14 Repem Problem 16.12. Use ~tepby:>(ep lineur accelera[ion method, Program 19. Neg:ec! damping,
Fitl {Kipli
17
Threedimensional Frames
'~i
10 9
,
:
: : , ~!!!
0.1
0.2 0.3 OA
0.5 O.S
l.0 '[(sec)
Fig. P16.12.
The stif:ness method for dyncmic analysis of frames presented in Chapter 15 for plane frames and in Chaprer 16 for grids can readily be expanded for the analysis of three~dimensional space frames. Although for the plane frame or for the grid there were only chree nodai coordinares at each joint, the threedimensional frame has a total of six possible nodal dispiacements at each unconstrained joint three translmions along the x, y. Z axes and three rotations about these axes. Consequently. a beam element of a threedimensional frame or a space frame has for its two joints a total of 12 nodal coordinates; hence the resulting element matrices will be of dimension 12 x i2. The dynamic a:1alysis of rhrcedimensional frames results in a comparative ty :onger computer program in geneml, requiring more input data as well as substantially more computational time. However, except for size, [he analysis of three~dirnensional frames by the sliffiless method of dynamic anaiysis is identical to [he analysis of ?lane frames or plane grids,
17.1
Figure 17.1 shows a beam segment of n space frame with "itS 12 nodai coordinates numbered consecutively. The convention adopted is (0 label first
491
492
Threedimensional Fral'nes
493
the three translatory displacements of the first joint fonowed by the three rotational displacements of the same joint. then to continue with the three translatory displacements of the second joinl and finaHy the three rotational displacements of this second joint The double arrows used in 17.1 serve to indicate rotational nodal coordinates; hence, these are distinguished from translational nodal coordinates for which single arrows are used. The stiffness matrix for a threedimensional uniform beam segment is readily written by the superposition of the axial stiffness matrix from eq. (15.3). the torsional stiffness matri, from eq. (16.6), and the flexural stiffness matrix from eq. (14.20). The flexural stiffness matrix is used twice in forming the stiffness matrix of a threedimensional beam segment to account for the flexural effects in ~he two principal planes of the cross section. Proceeding to combine in an appropriate manner these matrices, we obtain in eq. (i 7.1) the stiffness equation for a uniform beam segment of a threedimensional frame, namely
P,
z
Fig. 17.1 Beam segment of a space frame showing forces <:.r:ct nodal coordinates:.
di:;;:l~;}cemer:ts
E'
L
at 1he
P,
P,
0
0 0
0
y;0 0
l2
J2 1"
Syffiffie:rie
p,
0 61,
GI
L
P,
0
6El"
in which II. and It are, respectively, the crosssectional moments of inertia with respect to the principal axes labeled as y and:: in Fig. 17, Land L, A, and J are respectively the length, crosssectional area, and torsional constant of the beam element.
,y0
0 0
P, ...
p,
P,
0
EA
T ,.
0
0
17.2
0
FA
0 0
L
0
J21,
L'
a
0
p,
0 0
0
f)El,
70
2EI,
[.
bE(,
0
0 0 0
P"
0
0
 Gl
L
0
0
0 0
0 0
The lumped mass matrix for the uniform beam segment of a threedimensional frame is simply a diagonal matrix in which the coefficients corresponding to translatory and tOrsional displacements are equal to onehalf of the total inenia of the beam segment while the coefficients corresponding to flexurnJ rotations are assumed to be zero. The diagonal lumped mass matrix for the unifonn bearr. of distributed mass in and poiar mass moment 1m = ihh/A of inertia !.Xr unit of length may be written conveniently a.~
IQ/A
P"
,y0
6EJ,
0 0
:P,&
Гораздо больше, чем просто документы.
Откройте для себя все, что может предложить Scribd, включая книги и аудиокниги от крупных издательств.
Отменить можно в любой момент.