You are on page 1of 110
Advanced Building Analysis for Seismic Design with Eurocode 8 20! and 21% November 2012, Singapore (Hands-on 29" November) Prof. Koh Chan Ghee Dept. of Civil and Environmental Engineering National University of Singapore and Er. Choo June Shyan Otte International Pte. Ltd. Singapore ‘Aavanced Bulng Analyt or Siemie Design Copyight © 202, At Righs Reserved | | | | Day 1: Structural Dynamics Fundamentals of dynamic analysis Single-degree-of freedom systems Formulation of equation of motion Effective force due to ground motion Periodic and non-periodic loading Time domain and frequency domain methods Numerical integration methods Site-specific and smoothed response spectrum * Multi-degree-of freedom systems * Mass, damping and stiffness matrices * Eigen vectors and load-dependent Ritz vectors * Mode superposition method / modal analysis Peak response spectrum * Mass participation factors ‘* Modelling of tall buildings by finite element method @ | 12 © Prof G Koh te international Pte Ltd Advrce ina Art fo Seis Deson Coy ©2012 Al Rs Resened Structural Dynamics + Dynamic load - magnitude, direction and/or position varies with time + Newton's Law of Motion: F= ma + mais known as the inertia force + Force > Acceleration -> Change in velocity & displacement with time + Response (e.g. displacements, forces, bending moments) are function of time. + Much more difficult than solving static problems. Examples + Vibration of buildings due to wind and earthquakes + Vibration of bridges and railways due to traffic load + Defense structures against blast and impact + Offshore/marine structures — current, waves. + Floor systems due to machines and human footfall + Robustness of buildings by sudden column removal, etc. 13 Types of Dynamic Analysis + Deterministic — Loads are fully described (may be highly irregular such as earthquakes) @ * Complete time histories can be determined for given dynamic load. * Fora prescribed input, the response is uniquely determined. * For seismic analysis, it is not sufficient to consider one earthquake record as input. + Stochastic -- Loads are random & defined in a statistical sense. = Earthquakes are defined in probability form. = Hence responses can only be obtained in probability form. * This is more rational because of uncertainty in earthquake but is more difficult to implement, 14 © Prof. G Kon (tte Intemational Pie Ltd e @ ‘Advanced Bulang Analysis for Seismic Design Copyright © 2012. Al Rights Reserves Single Degree-of-Freedom (SDOF) Systems Simplest representation: mass-spring-dashpot system Spring - represent Displacement elastic stiffness u(t) — i P(t) Applied dynamic force “Dashpot” ~ to account c it for energy loss Equation of motion: mii +cu +ku = p(t) Vibrations Caused by Support Motion (u,) E.g. floor vibration due to machine, or Total di ground motion due to earthquake or traffic Relative displ. u(t) Fixed Reference’ 16 © Prof G Koh (ote International Ple Lis Advanced Builing Anaysis fr Seismic Design Copyright© 2012, All Rights Reserved ia Effective Force Equation of motion: Damping force and elastic force depend on relative (not total) motion SS . \ Fixed u(t Ga cut+ku =0 Reteence , Hele) Z Inertia force = Mass x Total Accel => miitcu+ku =Cimiig) “Effective force” due to support motion (inertia-driven). Not a physical force (unlike wind Load, for example). u(t) | | | | | | 1 1 ) @ eS 17 Damping To account for eneray loss (in a simplistic way): Thermal effects of repeated straining Opening and closing of microcracks vvv Friction at connections, between structure and non-structural elements + In practice, difficult to model energy loss exactly + Need to resort to experimentalifield observations * Different damping models: viscous, hysteretic, Coulomb friction, fractional derivative, etc. Most commonly used in linear viscous damping: c Damping ratio ¢ = —— ping S* ome Natural frequency (inrad/s) o = ¥k/m Naturalfrequency(in Hertz) f = o/(2n) Naturalperiod(ins) T = 1/F 18 © Prot G kon. Cote intermational Pte Lid © ‘Advanced Building Analysis for Selsic Design Copyright © 2012. Al Rights Reserved Estimate of Damping Ratio (4) + Free vibration — use the decay rate of amplitudes to estimate damping ratio. + Plot In(peak displacement) versus cycle and do linear fit * Other methods: Half-power (bandwidth) method, Nyquist plot method, random decrement method, ete © Experimental Determination of Damping Cyclic loading test (e.g. rubber pad) Compute dynamic stiffness (K,) & hysteretic loss factor (Y) K,=P,/d, sdlvia re Wh aa du lof Sarees E,= max. elastic energy stored =05P,d, Viscous damping ratio (at resonance): o05y Deformation hysteretic hoy 10 ePetceran a one orator Pi sate loss | ‘Advanced Building Analels for Seismic Design Copyright © 2012, Al Rights Reserved Typical Damping Ratios for Buildings Serviceability | Welded steel, prestressed concrete, RC with | 1-3% (Damping can be _| slight cracking lower for low-level vibration, e.g. for [RC with considerable cracking 35% human-comfort consideration) _| Boltediriveted steel 57% Ator near yield | Welded steel, prestressed concrete without | 6-7% point i complete loss in prestress bel { us T Rc structures, prestressed concrete with 710% wre 7 complete loss in prestress she Bolted/riveted steel, or wood structures with | 10-18% nailed/bolted joints For structural systems with passive dampers or base isolation, damping is usually high and depends on the materials and devices. Periodic Loading > Harmonic ~ a pure sine (ideal) P(t) es aay > Complex - e.g. most M&E equipments, regular foottall Prof. G Koh Perlod | 1412 (ote international Pte Lia e -10 ‘Advanced Building Analysis for Seismic Design Copyright © 2012. Al Rights Reserved Response due to P(t Harmonic Loading miu+cu+ku = p,sin(ot+$) Excitation Phase angle frequency For ©=0, the solution is Transient bog wl Ss Example for Harmonic Response Steady-state amplitude (p) For 5% damping ratio, initially at rest 10 5 o 1 2 3 4 Steady-state displacement: Uss = p sin( of -0) Steady - State Amplitude (p) D ‘Amplification = D = Yynamic Ampiieation =. ~ Static Displacement = pp /k 114 © Prof G Koh Cte International Pte Ltd ‘Advanced Builing Analysis for Saemie Design Copytght © 2012. Al Rights Reserved Dynamic Amplification (D) 10 - Resonance 94 8) c= 005 74 6 . ‘Amplification at resonance is inversely 5+ Se proportional to damping ratio 44 lAtw=0: D=| 26 Frequency ratio by s Periodic Loading but not single harmonic * Expand dynamic force in Fourier series: Porod pit) =a, + 5a, cos(niat) +>: b, sin(nat) om + Coefficients tend to diminish for higher modes. * For each harmonic term, obtain the steady-state response. + For linear systems, superposition is applied. i + Can be extended to non-periodic ioading by Fourier Transform, 1-16 ©PI!C koh Otte International Pte Lis ® \@ ‘Advanced Buldng Analysis for Seismic Design Copyright ©2012. AllRights Reserved Non-Periodic Loading > Impulsive eg. blast, impact, collision Short duration (compared to structural period) > General e.g. earthquake, wind, traffic 200 400 600 800 200 400 60) 800 Tima(sec) Time(sec) Response due to P(t), Impulsive Loading Area = J * Alarge force acting over a short duration a + Response depends on impulse magnitude: At at impulse = p(t) at Principle of impulse and momentum: impulse = Change in momentum > Initial velocity 1-18 (te Intemational Pte Lia © Prof G Koh ‘Advanced Building Analysis for Seismic Design Copyright © 2012. A Rights Reserved Vibration Response due to Impulse Free vibration after impulse: u(t) = exp(-Cot)sin(o,t) Mop 08, os Bo a 04 ool * If T= 1 and force duration is extremely small > unit impulse + The response is called the unit impulse response, /({), or known as the “Green function’. nit) = ew, cot)sin(ost) : Response to General Loading Analytical approach: A general dynamic load can be represented as a sequence of impulses. PO) Duhamel integral a t u(t)=| p(t) A(t—1) dr oa Lo pepo for one impulse A This is mathematically elegant but computationally inefficient. 120 i © Prof G Koh (te International Pte Les ‘vanod Bung Analy for Semi Design Copy ©2012. ARs Resoved Frequency Domain Method * Basis for spectrum analysis, e.g. for ground motion & wind velocity + Convert time-domain problem into frequency domain + Extension of Fourier series method — decompose time signals to harmonics of various frequencies + Use an efficient algorithm called the Fast Fourier Transform (FFT) + If need to recover to time domain, use inverse FFT. + Beware of constant term in FFT formula/subroutine in order to represent the energy correctly, particularly when used to compare with vibration criteria + To check, apply Parseval's Theorem: Total energy in time history = Total energy in FFT + Advantage ~ allows averaging in magnitudes (but loss in phase angles) + Extension to non-periodic signals, but fied to add “zero tail’, or use the exponential window method. 421 Example of FFT _ Lhe eta ‘Time (3) * Time history of acceleration ‘ecleraton (miss) + FFT based on 1 3 | record of 4s (1024 i e data points) zw | -* \sialiocabasbvabic Frequency (Hz) + Average FFT based on 45 records of 4s, each FFT Amalitude § od boceth ha Frequency (Hz) | “ 1.22) © Prof G Koh Otte interational Pte Ltd ‘Advanced Building Analysis for Seismic Design Copyright © 2012. Al Rights Reserves Time Domain Integration Methods + Time-stepping, or step-by-step method * Numerically integrate equation of motion in an approximate way. + E.g. Newmark method, Wilson method, HHT method, Central Difference method, Euler method, Runge-Kutta method, etc. + Advantage: Can be extended to nonlinear systems. + Disadvantage: Time consuming for MDOF systems if used in full analysis. + Modal analysis would help by decoupling the full system to multiple SDOF equations. vo » @ Au ol ye! tek tO ela 7 Response Spectrum ™ Pe b Analysis (RSA for seismic analysis, but < _ often misunderstood + This method aims to obtain peak (absolute max.) response for a prescribed earthquake time history. + SDOF - Equation of motion in terms of relative displacement (structure relative to ground): mi+ ca+ ku=—mu, * Divide by m: wo. Kk G+ su mom Original Displaced ground — ground position position 4.24 ©PrfC.G Koh te inteatora Pre Li © ‘Advanced Building Analysis for Seismic Design Copyright © 2012. Al Rights Reserved Structural Parameters Needed + Since c =Jkim oe o k vkim b= 5 the equation of motion becomes Minus sign can be ignored / for earthquake problems Ui+ 2gou+ ou= lg + Only two structural parameters are needed (2) Natural frequency (w) or period (T = 217/ w) and (b) Damping ratio (6) + Elastic force (base shear) is givenby f,=ku=mou f 2 mo Definition of Response Spectrum Given an earthquake ground motion, we can obtain the peak displacement (relative to ground): Sa(T,6)=max| u(t) | Structural Dampin, ral Period If we plot peak displacement versus natural period (7) for a given value of damping ratio, we have the displacement response spectrum. Similarly, by getting peak velocity (relative) and peak acceleration (total), we can obtain velocity response spectrum (S,) & acceleration response spectrum (S,) 126 © PIofC G Koh (tte International Pte Lid ‘Advanced Building Analysis for Selsmic Design Pseudo Spectral Acceleration Copyright © 2012 Al Rights Reserved + In getting the base shear, we need w? multiplied by displacement. + Let's define Pseudo Spectral Acceleration as (o S,(7,6)= 0° 8,(T,6)--- + This is close to the true spectral acceleration for small damping \ Venax = S,(T,S) + Max. base shear: Divide by structural weight (W= 7.4 bollt Que! Vonax = m0 max|u(t) mo” pak U mg): we Max Base Shear ~ Coefficient x Weight | + Hence max. (peak) base shear can be written as Vinee = Cs W/ where. ic. Sa(T,5)/9 * This dimensionless value is called the base shear coefficient or seismic fesponse coefficient. + This also shows that seismic force is Equivalent Lateral Force Method: estimate C,, e.g. in American codes: inertia (weight)-driven Simplified formulae are proposed to uBc-85: C,=ZIKCS ZIC f C= UBC-91 8 Re UBC-97: Gl 3h 1BC-2006: C, =Sps (4) © Prof G Koh 1.28 ‘te International Pte Lid ‘Advanced Bulg Analysis for Selsmic Design Copyright © 2012. Al Rights Reserved Example: El Centro Earthquake (18 May 1940) | NS Component ~ Time history a , it- ee a | 18: | E te oof — a, rma se Ref: http://nisee berkeley.edu/elibrary/Data/E0001 129 | Constructing a “Site-Specific” RS Curve spor , u(t) | Define damping ratio © System | [__e9.0.05 "J | Define natural period T |} Solve Equation of Motion Tec t+ 2c 0+ @7u =i, \ \ i \ g Displacement history u(2) a(t) Select an earthquake acceleration record as input, Obtain peak value i.e. max | u(t)| eg. El Centro NS Component qT ' his gives 1 data point: ' S,(T,¢) | prmemmneer eaecaey [ieriermne— Repeat for all other T values | — 130 eProtc G Koh (te International Pte Lia ‘Advanced Building Analysis for Seismic Design Copyright © 2012. Al Rights Reserved Response Spectra for El Centro Earthquake (NS) 100 50 20 ‘Damping ratio ¥ 10 0% (top curve) 2 2% 5 Ce ° 410% (bottom) . 3-way \ log plot S,=08, @ os S,=0S, gor 005 “OT 0205 10 2050 Natural Period T (sec) 131 | Recorded Ground pnb fe peo | Motions: yl Ce 1070-88 sonFaacice | | ~— a is gia te Es Earthquakes vary in magnitude, duration, | | | ! | frequency content, etc. Ref: "Dynamics of Structures” by A.K. Chopra, 2012 1-32 © Prot G Koh tte International Pte Lis © ‘Advanced Building Analysis for Seismic Design Copyright ©2012. A Rights Reserved Smoothed Design Response Spectrum + Use of a single earthquake record is suitable for checking purpose (particularly in time domain analysis). + But earthquake motion is random in nature & does not repeat itself. + Assingle earthquake record gives rise to jagged appearance of peaks > peak response is sensitive to the natural period of structure + Smoothed response spectrum is needed. + For design purpose, the response spectrum should be obtained from an ensemble of possible earthquake motions + Statistics are used to find the mean & standard deviation, e.g. mean plus one standard deviation would imply 15.9% probability of exceedance. + Furthermore, for ease of use, certain portions of the spectrum are linearised or represented by simple equations. 133 Example: Newmark-Hall Design Spectrum for PGA = 0.2g 2c0f 4 2 j200 me _ oe a Ep ai 2 ‘ fi 4 5 Bea. Eee A : 5 . é Fey: * e888 134 © Prof CG Koh (tte international Pte Ltd ‘Advanced Building Analysis for Selsmic Design Copyright © 2012. A Rights Reserved Multi-DOF Systems + SDOF model provides mathematical basis and understanding. + SDOF model gives an approximate solution -- accuracy depends on whether the motion can be adequately described by a single shape + More refined analysis would require MDOF systems. Finite element method (FEM) is usually adopted for its versatility. * Proper structural modelling by FEM is essential. + Equations of motion: mii + ou+ ku = pit) Stiffness matrix: Same as static problem * Mass matrix: Consistent mass approach — same shape functions are used in eS} computing mass matrix & stiffness matrix. v > Lumped mass approach — divide the mass equally among the element nodes for translational DOFs; no mass for rotational DOFs. 1.35 Damping matrix . pies establish analytically. + CAssuméiproportional (classical) damping matrix, e.g. Rayleigh damping: c=aym+ak ° \ / Rayleigh damping \ A ssitvess Z- proportional Can be related to damping ratios of eceoniora aa, any two modes as follows q eo; 0; |{ao fo 2|11o; o;jlas |g; : 5 Mass proportional damping (a, = 0) Natural frequency w, 138 © Pref G Koh (te International Pte Lis @ é ‘Advanced Building Analysis for Seismic Design Copyright © 2012, Al Rights Reserved Effective Forces due to Ground Excitation | Let u,, u,and u, be the relative displacements ms U3 of floors with respect to the ground. P Ky, C3 mi + cu + ku im, “ ka, = Doglt) =~ 12} g(t) oh \ | | ™ i i im m, | Ground acceleration ky. Cy | Original Displaced position position 137 Mode Superposition Method for MDOF Linear Systems | + This is a time domain approach (not frequency domain) + Convert physical domain into modal (eigen) domain fa (ee i + Modal solutions are superposed to give total response. ~~ Ce 7 | + This is aiso called modal analysis (though this term sometimes refer to just the eignvalue analysis). ~ + Usually low modes are sufficient > Advantage: allow exclusion of high modes + Since superposition principle is applied, this method is applicable to linear systems only (strictly speaking). + Need to solve eigenvalues and eigenvectors (alternative: Load- dependent Ritz vectors) 1-38 Prof G Koh Otte International Pte Ltd ‘Advanced Building Analysie for Seismic Design Copyright © 2012, All Rights Reserved Eigenvalue (or Modal) Analysis * This gives the modal frequencies (Ww, , W., ...) and mode shapes (9,, 6,,...), which are needed to compute other modal properties. 0004 008m +0019 200089 ™m mm 7 Mode 1 Mode 2 ors w31.4 m Mode 3 w= 46.1 rad/soc + The amplitude of mode shape is arbitrary. The mode shape can be scaled by any non-zero constant. Hence modal properties are not uniquely defined (depending on the scale used). 138 Importance of Eigen Solutions * The natural frequencies and mode shapes are important characteristics of structures in dynamic analysis. e + They should be obtained first to check the structure — physical (design) and numerical (computer): Y — Fundamental period should be reasonable (e.9. T~0.1N second for N-storey building) | ¥ Identify disconnected or improperly restrained members & rectify the computer modelling problem Y Identify local modes and suppress them | Y — Severity of torsional mode (use animation) 140 | © Prof @ Koh te Intemational Pte Lis ‘Advanced Buiding Analysis for Selsmic Design Copyright ©2012. Al Rights Resenved Effect of Asymmetry Non-symmetric structure — torsion present in fundamental mode | ‘Symmetric structure — translational modes do not have torsion Mode 3 Mode 1 Mode 2 144 Key Steps in Mode Superposition Method (1) Solve eigenvalue problem (corr. to undamped free vibration): » mu+ku=0 € > natural frequencies (w;, Ww», ...) and mode shapes (4, , 4, ....). (2) Transform physical coordinates (y) to the eigen-space by letting | 142 | © Prof CG Koh Otte international Pte Lid ‘Advanced Buling Analysis for Selmi Design Copytght © 2012. Al Rights Reserved (3) Assume proportional damping matrix, e.g. Rayleigh damping c=amtak (4) Use orthogonality properties of mode shapes, ie... 7 4, ko, 0) for mzn > Modal equation of motion (scalar, not matrix) MnGn + Cnn + Kin = Pa (t) Properties (5) Each modal equation is solved as a SDOF system. e@ (6) Computed modal solutions q,(t) are then superposed to give displacement response history (see equation in Step 2). (7) Response quantifies such as shear forces & moments are calculated from their relationship to displacements via structure/element stiffness matrices. 1-44 © Prof CG Koh (tte Intemational Pte Lid Modal @ © ‘Advanced Bulding Analysis for Seismic Design Copyright ©2012 AlRighs Reserved For seismic analysis, the excitation is due to ground motion. The modal equation of motion becomes M.Gn+Crdn + KG, =—, mv li, (t) where r is an influence vector relating the nodal DOF to ground motion. E.g. For horizontal-x ground motion, the coefficients are 1 for all horizontal-x DOF (in the same direction) and zero for all other DOFs: Us Ground displacement — 145 + Often ignore the minus sign of ground acceleration and divide by modal mass: a 2 F Gn + 28nOnGn + On In = Yn Halt) * Mode superposition method transforms a MDOF system to sum of independent SDOF systems (via mode shapes) + Usually contribution of high modes can be neglected. + The main advantage is that the user can include only selected modes, say first J modes out of a total of N modes. For seismic analysis of real buildings, normally J<< N. + Equation for each mode can be solved very efficiently + Overall this is faster than direct analysis of full system 146 © Prot G Koh. (tte International Pte Lid ‘Advanced Bulding Analysis for Selec Design Copyright © 2012. Al Rights Reserved Three Ways to Solve Modal Equations Each modal equation can be solved by: (2) Response history analysis in time domain (via step-by-step method) (b) Response spectrum analysis (suitable for estimating peak response, widely used for seismic design) (c) Frequency domain analysis (with the help of Fast Fourier Transform) Do not confuse (b) and (co) Method (a) gives response time history of each mode considered (q,). The total response history is obtained by superposition via mode shapes: 2 where Jis the number of modes u(t)= z In(t) Which are included in the analysis 147 Adding Peak Values of Different Modes If only the peak response is needed, Method (b) can be used. + The peak response value of each mode considered is obtained e the same way as in response spectrum analysis of SDOF system. + For each mode, the corresponding mode shape is used to obtain the corresponding peaks at different DOFs. + But the peak values for different modes do not necessarily occur at the same time. + Furthermore, response spectrum analysis gives maximum value without + sign. + Addition of peak values from different modes (“Absolute Sum Rule’) tends to be too conservative. E.g. base shear: J, ypeak _ S$ 1v, [Pek combining modes 1 to J * Alternative combination rule is required to give a more reasonable estimate of the total peak response from modal peak values (to be explained later) us g ol 1.48 c © Prof G Kon Otte International Pie Ltd J BQ enced Bag Artif Sea Design Copy © 202.8 igs Reserved Ritz Vectors vs. Eigen Vectors + Numerical effort required to compute eigen vectors can be time consuming for large systems. + Eigen vectors correspond to free vibration mode shapes, without paying any regard to the spatial distribution of loading. * Many eigen vectors that are orthogonal to loading do not participate in dynamic response. + Hence the convergence may not be efficient in terms of the number of modes required to reach certain mass participation ratio. Alternatively, it is possible to compute a set of stiffness & mass orthogonal Ritz vectors with less modes than eigen vectors in convergence. a + Reason: Load-Dependent Ritz (LDR) vectors are generated by taking into account the spatial distribution of dynamic load. 1-49 Load-Dependent Ritz Vectors + For earthquake: -Me G,(t)=E G(t) + G(t) can be expanded into a Fourier series. Hence consider Mi+ Ku=F sin(ot) leading to Ku= EF +0°Mu * This cannot be solved directly because of unknown frequency. + Use perturbation. First neglect mass: Ku, = F + This gives a trial displacement vector u,. + The error arising from neglecting inertia force is F, =Mu, + Thensolve Ku, = F, + This gives the first LDR vector uy. 1-60 @ Prof CG Koh Otte International Pte Ltd ‘Advanced Building Analysis for Seismic Design + Subsequent vectors u, are derived corresponding to errors arising from treating dynamic problem as static: K Copyright © 2012. Al Rights Reserved Faster Convergence with LDR FE = Mu,, + The vectors are made orthogonal with respect to mass and stiffness matrices, Tmbero! | Fee Vivaton tide Shapes | Lost epedet is Veco Nectors [Displacement | Moment_| Displacement [__Nloment + Example: couse | ame | cours | sur i 1 (2any (22m) _| oa, (92) 7 | eemma | ame | comsar | ates qeay | cz | aon | (ee \" paoonnemdles i ¥ (046) (85) (008) 35) y Y a ne _— ee | (i | eomea | — se | ones | — sen (ay | 4 | em | 09 7) | amass se 5 Ritz vectors vs. 9 eigen 209 | (29 cones | sett vectors, to achieve 0.0% error. 8 nenvltaen Ne: Harber ie pate ae percentage er. 151 Direct Analysis of MDOF Systems (For Linear and Nonlinear Systems) + Solve the matrix equations of motion for the full system directly in time domain, usually by numerical integration methods. + Commonly adopted: Step-by-step (or time march methods) e.g. Newmark method, Wilson method & Huber-Hughes-Taylor method + Computationally expensive for large systems. + Applicable to nonlinear systems e.g. large displacement, nonlinear material, damage, and non-classical damping, * Good judgement is required in the finite element modelling of structures. + Step-by-step methods are used to solve the dynamic equations. + Solution accuracy depends on the time step (relative to the natural period of the system) — the smaller the more accurate. + Some methods are conditionally stable, i.e. solution will blow up if time step is not small enough, e.g. Central Difference Method. = move Gur black saudysis + Some methods introduce numerical damping, e.g. Wilson-6 method, which would be good to damp out inaccurate high modes @PIoIC Koh « wee \0 ‘wt ee de « 182 Otte International Pie Ltd svances Big Arab fo Sela Dstn cape 2012 Ante Ramee Step-By-Step Methods + Solve equations of motion one step at a time in an approximate way by making certain assumption of the solution in each time step. + Newmark-8 Method: (2) Constant (Average) Acceleration Method - Assume acceleration is constant in each time step. (b) Linear Acceleration Method: Assume acceleration varies _ linearly in each time step, i.e t a(t) =a, +(a,., -a, fort, h-tefinement (most commonly adopted) + Alternatively, higher order elements are used to see change in results > p-tefinement + Another possibility is rearrangement of FE mesh by shifting nodes to regions where finer mess is required (e.g. stress concentration) without increase in DOFs > rrefinement (‘reshaper’) PH EE ae Original mesh —_—_A-refinement p-refinement refinement 1.87 Imposing Kinematic Compati + Finite elements interface with one another through the nodes ity + Main unknowns are displacements at the nodes (which could include rotations for frame and shell elements) + Compatibility is imposed between elements at the nodes, but not necessarily along the edges between nodes. + Sometimes it is necessary to achieve compatibility by imposing kinematic relations — * This is particularly useful for mismatched meshing between two different elements, or simply to reduce DOFs. 1-88 © Prof G Koh Otte International Pte Ltd ‘Advanced Building Analysis for Seismic Design Copyright ©2012. Ad Rights Reserved Kinematic Constraints (1) * Finite element mesh can affect the solution considerably. + Rectangular elements are more accurate than arbitrary shaped elements. * From coarse mesh to fine mesh, mesh transition would generally mean non- rectangular elements. + The process of defining transition elements can be tedious, and mesh distortion decreases accuracy. Kinematic Constraints (2) + Alternatively, kinematic constraints can be imposed to enforce displacement compatibility between elements where nodes do not coincide — known as “line constraint’. + This eliminates the need for mesh transition elements. + This approach is also useful for capturing interaction of different structural objects, e.g. curved shear wall supporting a curved ramp. Displacement ‘compatibility enforced here 1-60 | © Pref C.G Koh Otte International Pte Lid ‘Advanced Bulking Analyss for Seismic Design Copyright ©2012. Al|Rights Reserved Kinematic Constraints (3) + The finer mesh is constrained to the coarse mesh. + Shape functions of the coarse mesh is used to evaluate the displacements. at the nodes of the fine mesh. * Displacements at the 3 fine mesh nodes (#5, 7 & 9) are “constrained” to move linearly as defined by displacement shape between the 2 coarse mesh nodes (#3 & 4). + Fori=8,7&9: uy; =N3Ug+NgUg v; =N3V¥3+NgVg 4 4 9 10 7 8 2 Seee 8 coarse fine * Another example: Inclined @ floor connected to walll, but i ead their nodes do not coincide. * If kinematic constraint is not imposed on the elements, there will bé gap or overlap between the slab and wall along the interface. Kinematic Constraints (4) | 1-62 © Prot CG Koh ‘ote International Pte Lis ‘Advanced Building Analysis for Selsmic Design Copyright © 2012. AlRights Reserved 1D, 2D or 3D Elements? + For buildings, bending-action members are best modelled by frame elements (which have rotational DOFs) + Depending on the size of opening, shear walls with openings may be modelled effectively by frame elements with rigid ends. + If 2D or 3D elements are needed, it is often sufficient to use up to quadratic orders in most problems; higher order elements do not necessarily work better. + Incompatible elements could be better than compatible elements in some problems, e.g. to mitigate shear-locking as encountered in 4-node quadrilateral (2D) & 8-node hexahedral (3D) elements. 163 Frame Element + Frame element- most useful elements for structural analysis, accounting for bending, shear, axial and torsional deformations. + Some software allow non-prismatic sections @ * 6 DOFs per node ( “member end release” possible), e.g. setting an end moment to zero. + Use (Hermitian) cubic shape functions NA * Truss and spring element — axial vy deformation only. Nr + Truss element can be obtained from 1\s frame element by end moment zp Soot release cc fu, + Gap element — compression only + Hook element — tension only + Gap &hook elements are nonlinear. iobal 2-node frame y element 164 © Prof CG Kon (te International Pte Ltd ‘Advanced Building Analysis for Seismic Design Copyright © 2012. A Rights Reserved Joint Modelling (1) + To be closer to reality, include the effects of joints by using frame elements with rigid ends. * Inclusion of rigid zone effect stiffens the system and increases the natural frequency. Rigid zone Rigid ends @ =node 1-65 Joint Modelling (2) B= 0 for no rigid zone (span from centre line to centre line) kinematic constraints). @ B= 1 for full rigid zone (rigid for full width of intersecting member) 1 i | 1 | ‘Shear ! ibe Wasi Deformed shape i i Rag me od Rigs i Rigid zone ! Constitutive equations are established ee ant I accounting for the rigid ends (which are ~ 1 i — Centre tines —*} 1-66 OPI G Kon (Otte international Pte Ltd ‘Advanced Buling Analyss for Seismic Design Copyright© 2012, Al Rights Reserved Rigid Diaphragm (1) + In-plane deformations in floors of buildings are usually negligible due to the strong axial rigidity of floor system. « Floor can thus be assumed to translate horizontally & rotate about a vertical axis as a rigid body. + Kinematic condensation by “master-slave node concept”: 3 master DOFs in-plane per floor. * Qut-of-plane deformations are not affected. + Reduce DOFs significantly & save computational time. us] [1 0 -y)fum fl us|=|0 4 x {lum om e| lo o 4 llem 2 | x & y are coordinates of slave node relative Master to master node. node 20-storey frame building No. of joints = 480 Total DOFs= 480.6 = 2880 Reduced to | 20x3 + 480x3= 1500 DOFs it Hy 3 master 3 out-of-plane DOFs DOFs per floor _ Per beam-column node Lee | DOFs can be further reduced if static or dynamic condensation is used. 1-68 © Prof G Kon Otte Intemational Pte Lid ‘Advanced Building Analysis for Seismic Design Copyright © 2012. A Rights Reserved Caution in Using Rigid Diaphragm + Caution! Truss action will not be modelled if horizontal chord members are __. | part of rigid floor diaphragms. + Use “semi-rigid” diaphragm which does not impose in-plane constrain but only facilitates the definition of floor system for lateral load analysis. 169 Modelling of Floor Systems There are various ways to model floor systems (and similarly for wall systems) One approach is to extend beam theory (1D) to plate bending theory (2D). Plate bending accounts for out-of-plane deformation but not in-plane deformation. On the other hand, membrane accounts for in-plane deformation but not out-of-plane deformation « Shell = Plate Bending + Membrane Shear deformation should be accounted for thick shell or thick plate. 1-70 PICS Koh ‘Ote Intemational Pte Lis ‘Advanced Building Analysis for Selemie Design © Pref @ Koh Shell Element Plate element — bending action only (1 displacement + 2 rotation DOFs per node). Membrane element — membrane action only (2 displacements + possibly 1 “drilling” DOF per node). Shell element = Plate + Membrane (5 or 6 DOFs per node). Drilling DOF is preferably included to allow rotational compatibility with frame elements. Drilling DOF 4-node shell element “Membrane Slab” In ETABS, slab with membrane properties can be used to transfer floor load on floor to beams & columns. The membrane slab (or virtual membrane) does not resist the load. The concept of tributary loads is used as a simplifying assumption for transferring the loads. The dividing lines resemble yield lines. Full uniform load 172 Copyright © 2012. Al Rights Resowed Otte Intemational Pte Lis “ 4 ‘Advanced Busing Analysis for Selomic Design Copyright © 2012. AlRighs Reserved Beam-Slab Compatibility By default, the centre lines of beam and slab coincide. If one wishes to capture the interaction of beam and slab more accurately in the analysis, one way is to impose displacement compatibility by ‘rigid constraint” option (provided in some software). ‘As such, there is no issue of effective width of T or L-beam This way also allows non-prismatic beam where neutral axis is not on a straight line. Nevertheless, design should be consistent with the analysis, e.g. axial force in beam would be significant and should be accounted for. Rigid constraint ‘on nodes | and J 1-73 v vvv © Prot CG Koh Shear Wall Behaviour Shear walls are often used to describe RC walls that are designed to help the building resist lateral loads (shear forces to the building; hence the name “shear wall’). The name is somewhat misleading. Under normal circumstances: ‘Shear walls deform in bending mode primarily Shear deformations are rarely significant Only very low shear walls (hiw ratio <1) fail in shear They are designed to resist the combined effect of axial, bending and shear 174 (te imernational Pte Ltd Advanced Bling Anal for Seem Deson Copyigh ©2012. ligt Resened Modelling of Wall Systems (1) + Plane stress elements can be used to model walls. But this does not account for out-of-plane bending. + Shell elements are preferred, which also allow beam-wall interaction through rotational DOFs (compatibility). + For good accuracy, it is necessary to use sufficiently fine elements for each panel but this is not practical for high- rise building with many walls. [|_| FE mesh for 1 one panel 178 Corrected Panel Element * Classical 4-node plane stress element — using single element to model shear wall can lead to significant error, particularly for tall walls (over-stiff). * Correction is needed. In ETABS, each panel element is automatically divided into 2 sub-elements. A force method is used to compute panel forces, and i internal forces are in exact equilibrium with applied load regardless of wall | shape. A ETASS Panel Element on 08 |S exeeeseeeeReEeernEE | ed ae 'SAPZ000 Shell Element] = B¥os u Og # Eo 2 3a 2 gt @ 8" 02 2 of | o 2 4 6 8 1 Aspect Ratio (hiw) 1-76 © Pref CG Koh Cte Interational Pte Ltd ‘Advanced Bulding Analysis for Selsmic Design Shear Wall with Openings Big openings — frame elements Small openings — Medium openings — piers & spandrels Copytight © 2012, Al Rights Reserved (full) panel elements — @ —— 17 3D Solid Element For very thick shell, it may be necessary to use solid elements (e.g. very thick floors and deep foundation). To capture bending behaviour, more than one layer of solid elements is usually required. Much more computational time than 1D and 2D elements. 4-node tetrahedral elements give constrain strain & do not perform well particularly for capturing stress gradients. 20-node (serendipity) 8-node solid element Tetrahedral element 178 © Prot G kon (te International Pte Lta ‘Advanced Building Analysis for Seismic Design Geometric Effect Vertical load on laterally displaced columns & walls can cause additional moments that, in turn, increase lateral displacements and also alter axial forces in general. ‘To account for this geometric effect, it is necessary to impose equilibrium & re-compute structural stiffness matrix based on the deformed position. This can be obtained at element level by k= Kelastic + Kgeometric + Iflinear shape function is used, geometric stiffness is 1 =1) Compression P has negative 1 1] effect on stiffness matrix + For frame element, cubic shape function gives a better approximation. Copyright © 2012. Al Rights Reserved A — Pp _ = @ 179 P-Delta Analysis of Building + First-order analysis assumes small displacements and rotations > equilibrium is based on un-deformed (original) configuration > + Second-order analysis takes into account displacement (if significant) @ caused by loading. Equilibrium is updated based on deformed configuration. * By using geometric stiffness matrix, P-A effect can be included consistently in both static and dynamic analyses. * In most cases, the lateral displacements are small (24 order effect). Axial forces on vertical members for buildings remain approximately the same after doing P-A analysis once. No iteration is required + In-cases where axial forces do change significantly due to relatively large A, iterations are required, ie. repeat P-A analysis based on updated stiffness until convergence. This is called “P-A and large displacement’ in the software. * Iterative solutions are more accurate but more time consuming. * Apredominate load case is used to compute stiffness matrix accounting for P- A effect, and this is used for all other cases. © Prof G Kon 1-80 te Inerational Pte Ltd ‘Advanced Building Analysis for Seismic Design Copyrights © 2012. AllRights Reserved Linear & Nonlinear Material Models 8 8 S S “4 “| Linear Elastic Linear Inelastic “Deformation “Deformation “a Nonlinear Elastic Force Force Deformation Deformation Cracked Section for RC Members + To avoid nonlinear analysis, nonlinear material behaviour due to crack is often accounted for by applying cracking factors to the gross moment of inertia. Se ‘ACI 316 Strength Serviceability Beams 0.35 0.50 Columns 0.70 1.0 Shear walls — Uncracked 0.70 1.0 Shear walls — Cracked 0.35 08 Slabs 025 0.36 + Based on above, drift & deflection obtained for serviceability = 0.7 x drift & deflection obtained in individual load case of strength analysis (without load factor) + For post-tensioned slabs and beams, cracking factor can be increased by ~30% (engineering judgment, not in code)" *Ref. JF. Howvillour, V.B. Patel, and K.A. Young, “Modeling Assumptions for Lateral Analysis", ACI ‘Special Publication, SP-240, V40, pp. 73-100, 2006. 1-82 OP rot. CG Koh ‘Ote international Pte Lis ‘Advanced Builing Analysis for Seismic Design Advanced Building Analysis for Seismic Design with Eurocode 8 20 and 21% November 2012, Singapore (Hands-on 29" November) Prof, Koh Chan Ghee Dept. of Civil and Environmental Engineering National University of Singapore and Er. Choo June Shyan Otte international Pte. Ltd. Singapore Day 2: Seismic Analysis and Design + Characteristics of earthquake motion + Seismic design philosophy + Earthquake design code Eurocode 8 + Equivalent lateral force method + Modal response spectrum analysis + Combination rules for peaks * Comparison with IBC-2006 and UBC-97 + Key principles of seismic design * Effects of structural configuration and irregularity + Inelastic response and ductility + Deflection/drift amplification + P-Delta effect and geometric stifiness + Response history analysis (linear and non-linear) + Performance based evaluation and design + Nonlinear analysis procedures + Force deformation relationship + Capacities of structural elements + Performance level definition + Capacity design with nonlinear procedure Prof. G Koh 22 ‘ote intecratonal Pte Ltd Copyright © 2012. A Rights Reserved. ‘Advanced Building Analysis for Seismic Design Copyright ©2012. Al Righis Reserved Main References = “Eurocode 8: Design of Structures for Earthquake Resistance”, EN1998, 2004 (mainly Part 1) * International Building Code IBC2006, International Code Council, USA, + ASCEISEI 7-05, “Minimum Design Loads for Buildings and Other Structures”, 2005, ‘American Society of Civil Engineers (mainly Chapters 11 and 12) * Uniform Building Code 1997, International Code Council, 1997 (mainly Volume 2, Chapter 16) = ASCE/SE! 41-06, Seismic Rehabilitation of Existing Buildings, 2006 + “Dynamics of Structures: Theory and Applications to Earthquake Engineering’, A.K. Chopra, 4 Edition, 011 = ETABS and SAP2000 reference manuals, Computers and Structures, Inc. Disclaimer The course material is intended to illustrate the concepts and methodologies involved and is not to be used as an official reference for analysis or design The lecturers accept no responsibility or liability caused by incompleteness or & inaccuracy. Class participants are advised to refer to the relevant documents (e.g. references above) for full details and accuracy. 23 Characteristics of Earthquake Motion + Sudden movement of ground that releases elastic energy stored and generates seismic waves. + These waves radiate outward from the source (“focus”) & shake the @ ground. + A faultis a large fracture in rocks, across which the rocks have moved. {t can be up to thousands of km long and tens of km deep. + There are many types of earthquakes which have different characteristics (duration, frequency content, depth, etc) Sea ‘ oe Australian oe te —™ —_ Subduction Strike slip 24 ©PIIC G Kon Otte International Pte Lid Global Ref: © Prof .G Kon ow ‘Aavanced Bulsng Analysis for Selec Design Copyright® 2012. All Rights Reserved Plate Tectonic Movement a ee ee hitto-//sideshow.ipl.n: imbh/alvimages/global, ‘Ote International Pte Ltd ‘Advanced Bulding Analysis for Seismic Design Copyright © 2012. Al Rights Reserved. Sumatra Earthquakes M=9.3 27 Earthquake Magnitude and Intensity Magnitude + Richter Scale is a measure of earthquake magnitude (M) by measurement of ground motion (via seismometer) + E.g. Ground movement is 10 ym for a quake of M = 1 at 100 km from the epicenter, and 100 ym for M=2, etc. + One magnitude higher > Amplitude of ground shaking is an increase of 10 times in displacement and 31.6 times in energy + Energy (in Joules) = 104(4.8+1.5M) Intensity + Intensity measures the strength of shaking at a certain location. + Intensity becomes weaker away from the epicenter (attenuation) due to wave scattering and energy loss. + Attenuation relations are empirical in nature. © Prof ¢ @ Koh 9 ‘Aavanced Bulking Analysis for Selec Design Copyright © 2012. All Righis Reserved Modified Mercalli Intensity (MMI) This is a commonly used scale for measuring the earthquake intensity. |. Not felt except by a very few under especially favorable circumstances, Il, Felt only by a few persons at rest, especially on upper floors of buildings. Delicately suspended objects may swing lll, Felt quite noticeably by persons indoors, especially on upper floors of buildings. Many people do not recognize it as an earthquake. Standing motor cars may rock slightly. Vibration similar to the passing of truck. Duration estimated. IV. Felt indoors by many, outdoors by few during the day. At night, some awakened. Dishes, windows, doors disturbed; walls make cracking sound. Sensation like heavy truck striking building. Standing motor cars rocked noticeably. V. Felt by nearly everyone; many awakened. Some dishes, windows broken. Unstable objects overturned. Pendulum clocks may stop. Vi. Felt by all; many frightened. Some heavy furniture moved; a few instances of fallen plaster. Damage slight. 29 Modified Mercalli Intensity (MMI) - Cont’d Vil. Damage negligible in building of good design and construction; slight to moderate in well-built ordinary structures; considerable damage in poorly built or badly designed structures; some chimneys broken. Noticed by persons driving motor cars. Vill. Damage slight in specially designed structures; considerable in ordinary substantial buildings with partial collapse. Damage great in poorly built structures. Fall of chimneys, factory stacks, columns, monuments, walls, Heavy furniture overturned IX. Damage considerable in specially designed structures; well-designed frame structures thrown out of plumb. Damage great in substantial buildings, with partial collapse. Buildings shifted off foundations. X. Some well-built wooden structures destroyed; most masonry and frame structures destroyed with foundations. Rails bent XI. Few, if any (masonry) structures remain standing. Bridges destroyed. Rails bent greatly. XIl. Damage total. Lines of sight and level distorted. Objects thrown into the air. 240 @ Prof CG Koh (te international Pte Lis ‘Advanced Building Analysis for Seismic Design Copyright © 2012 Al Rights Resoved. Very Approximate Relationship for Peak Ground Acceleration (PGA) Richter | PGA(g) | Duration MMI PGA (a) Magnitude (sec) V 0.03 5.0 0.09 2 Vv 0.03 -0.08 55 0.15. 8 vi 0.08 - 0.15 8.0 0.22 us vil 0.15 -0.25 65 0.29 18 vill 0.25 -0.45 7.0 0.37 24 XK 0.45 - 0.60 75 0.45 30 x 0.60 —0.80 80 05 34 xt 0.80 - 0.90) 8s 05 37 xi 20.90 * The above gives very rough indication mainly based on California earthquakes. There were exceptions, e.g. 1989 Loma Prieta earthquake e@ magnitude was 7.1 but PGA was 0.65g MMI is based on observed damage which depends on not just PGA but also many other factors, e.g. structural design, soll condition and construction quality. an Japan Meteorological Agency (JMA) Scale + JMA scale is a measure used in Japan to indicate seismic intensity, in units of shindo (ji) -- literally “degree of shaking”. Historically determined by human perception (like MMI) but in early 1990s moved towards instrumental seismic intensity. * Computation from measured accel. time histories to JMA scale is complicated, taking into account 3 directions, time duration and frequency content. * No simple and linear relationship between JMA scale and PGA. Next table provides only approximate correlation + See graphical illustration of JMA scale at htto:/Awww.ima.go.ioima/en/Activities/intsummary.pdf 2412 © Prof G Koh (tte Iniermational Pie Lia ‘Advanced Buldng Analysis for Seismic Design Copyright © 2012, Al Rights Reserve. | sMascale | roa | People RC buildings | | Trem} | aposoos [MEE | petits pop | | | [lavesae | 2006 Fe boty some pose in the [revasze | oasmit | bute Felt by most people inthe bulding. 21a}/15:24 | oors-oce ms | FSreY mos Peers | Fat by mos people inthe bung 31/2534 | ogeo2smAt | Fee ee ee | ‘Many people are frightened. some | 414)/3544 | 025-080? | peopletry to escape fram danger i Mos peopl io expe roma sloner/4s-49 | o80-s4omee | dagsesome pepe cut pf : any el conraly Ceaslonaly, cracks ae formed in walls of lass | Most sleeping people awake. | | 1 earthquake-esstant buildings. | cational reac re fermedinwa | 15054 mye | Mamypecle are coneratiy | eossbeams an lls of ess earthauake- Samper/s054 | 140-2sompt | tered an nee a ee eee I yee ‘resistant buildings have cracks in walls | | ‘Occasional, walls and plas of es earth: | . resistant uly are destoyed and even hy Slower 55:59 | 250-35 mst | Deelto keep sanding ee ee eee inal, cossbeans ana plars ‘Occasionally less earthquakeesistant bucings supper/6o.ea | 315-00 | basse tokeepstancingand to | clips. n some cases even highly earthqule- | ‘move without eawing ‘esstant buldings ser damage to wall and | pila. | Dm Greaterthan | Thrown by the shaking and ‘Occasional, even highly earthauake resistant [265-07 | amish | Impossile to move at wil | bulings ae severely damaged and lean. From Source to Building | + Earthquake —type, magnitude, geology, depth, duration, frequency content, distance and attenuation > Ground motion parameters + Site/soil effects — local soil properties that affects ground motion frequency with possible amplification > Ground type / site classification + Structural dynamics of buildings — structural system, configuration & | properties (e.g. mass, damping & stiffness) > Analysis and design + Soil-structure interaction — usually not considered except for buildings with large basements or foundation systems (not easy to model) surace Epicentre /“ waves it ey ee \ Continental crust (bedrock) Oe 214 © PrOtC G Koh Otte Intemational Pte Lia ‘Advanced Building Analysis fr Seismic Design Copyright© 2012, Al Righls Reserved Intent of Seismic Design Codes * Seismic codes are mainly about “survivability” (life safety) and not “serviceability” except for critical structures™ in major earthquakes. + The intent is to have the strength and stability so that the structure will not have life-threatening collapse in major earthquakes. + Expensive to design structures in elastic range in major earthquakes. Thus, damage and large deformations are usually allowed. + Seismic code provide “reasonable” but not complete assurance of protection of life (particularly for major earthquakes). + Level of protection can be increased by increasing the design lateral force, ductility (energy dissipation capacity) and redundancy. + Each major earthquake has taught new lessons. Thus seismic codes are regularly revised (e.g. every 3 years in USA). For critical structures (e.g. hospitals and fire stations), higher requirement of design force is imposed via importance factor. 248 Typical Earthquake Design Philosophy Intensity [Non-structural | Structural Level Damage Damage Minor No No Moderate | Yes No Major Yes Yes, but no collapse (Ref. SEAOC “Blue Book”) * Main reason for allowing damage is to reduce the otherwise huge economic cost required to make all buildings undamaged in moderate and major earthquakes. + Most important of all, catastrophic collapse must be avoided. 246 © Prof G Koh Otte Intrational Pte Ltd ‘Advanced Building Analysis for Seismic Design Copyright © 2012. Al Rights Reserved Eurocode 8 Design of Structures for Earthquake Resistance EC8 contains six parts. Part 1 (EN 1998-1): “General Rules, Seismic Actions and Rules for Buildings’. Provides a template for code. Each member state has to produce its own Nationa! Annex to define Nationally Defined Parameters (NDP) and seismic hazard map. Part 3 deals with assessment and retrofiting of buildings puch over analysis, Parts 2, 4, 5 and 6 deal with other types of structures (bridges, tanks, foundations, towers, etc). | EN1998-1: General rules, seismic actions and rules for buildings + Section 1: Scope, normative references, assumptions, principles and | application rules, terms and definitions, symbols and units. + Section 2: Basic performance requirements and compliance criteria applicable to buildings and civil engineering works in seismic regions. + Section 3: Rules for the representation of seismic actions and for their combination with other actions. + Section 4: General design rules relevant specifically to buildings * Section 5: Specific rules for concrete buildings + Section 6: Specific rules for stgel buildings + Section 7: Specific rules for composite steel-concrete buildings + Section 8: Specific rules for timber buildings + Section 9: Specific rules for masonry buildings. + Section 10: Base isolation of structures 248 © Prof G Kon Otte international Pte Lid ( user 7 fol» ‘Advanced Bullting Analysis for Seismic Design Copyright © 2012. Al Rig EC8 ~— Fundamental Requirements (1) No-Collapse Requirement (NCR) + To withstand the design seismic action without local or global collapse, thus retaining its structural integrity and a residual load bearing capacity after the seismic events. + Design seismic action is expressed in terms of: (a) reference seismic action associated with a reference probability of exceedance, Pycp, in 50 years (or reference return period Tyg) and (b) importance factor y, of building. + Recommended Pyca= 10%, which corresponds to Tyca = 475 years (to be decided by the National Authorities) + Reference peak ground acceleration corresponds to the reference return period Tygp of the seismic action for the no-collapse requirement. ECB EC8 — Fundamental Requirements (2) (ea Damage Limitation Requirement (DLR): + To withstand a seismic action having a larger probability of occurrence than the design seismic action, without the @ occurrence of damage and the associated limitations of use, the costs of which would be disproportionately high in comparison with the costs of the structure itself. + The seismic action for DLR has a probability of exceedance, Ppy a, in 10 years and a return period Tp, p. * Recommended Pp,= 10%, which corresponds to Tp,g = 95 years (to be decided by the National Authorities) 220 © Prof G Koh (ote Intermational Pte Lis ee ‘Advanced Bulding Analysis for Seismic Design Copyright© 2012. All Rights Reserved. Design Earthquakes in UBC & IBC 1Be2006 (For Comparison) UBC 4997: Maximum Capable Earthquake (MCE) — 10% probability of exceedance in 100 yrs (1000-yr return period) Design Basis Earthquake (DBE) — 10% probability of exceedance in 50 years (475-yr return period) IBC 2008: Maximum Considered Earthquake (MCE) - Most severe earthquake considered by the code: (a) Probabilistic MCE — 2% probability of exceedance in 50 years (2475 years return period) (b) Deterministic MCE - based on characteristic earthquakes in the region (ASCE 7-05 Ch 21) Design Earthquake (DE) - 2/3 of MCE © Prof CG Koh (Otte International te Lis Earthquake Probability & Return Period If P, is the probability that an earthquake ground motion of a given intensity** | will be exceeded in any given year, then the (mean) return period is T= 1/ P, If a building is designed for N years, the probability that an earthquake ground | motion of the given intensity will be exceeded during its lifespan is P= (1-4) =1-en(-2) if Nand T are large T) T) IfN = 50, T= 475, then 4 (a7, 50 ) =0.1=10% If N = 50, T = 2475, then + Pop = 1-[1-—4—| =0.02 = 2% 2475, The ground motion intensity corresponding to a return period (say 475 years) depends on the seismicity of location. E.g. Areturn period of 475 years could correspond to ground motion of 0.4g in high seismicity region and only 0.05 g in low seismicity region 222 ‘Aavancea Building Analysis for Seismic Design Copyright © 2012. Al Rights Reserved. Earthquake Probability & Return Period Exceedance Time span | Return Period | Reference in Probability (Py) (N) ) Codes 20% 10 years 45 years 10% 10 years 95 years DLR in Ec8** | 20% 50 years 225 years 10% 50 years 475 years NCR in EC8**; DBE in UBC97; DE in IBC2006** 5% 50 years 975 years 10% 100 years | 950 years | MCE in UBC97 5% 400 years | 1950 years 2% 50 years 2475 years MCE in IBC2006 ** Based on the recommended value of 10% for Poue and Pyyoa in ECB. + Approximate equivalence EC8: Seismic Zones + Local hazard is described in terms of the reference peak ground acceleration on type A ground, denoted as age. + Seismic zone maps are to be provided in National Annex. + Design ground acceleration on type A ground: y= Aga X Importance Factor (y,) + Corresponds to reference return period Tyor chosen by the National Authorities.(475 years recommended). where the importance factor depends on the consequences of failure. + Low seismicity: areas where a, < 0.08g or a, S < 0.1g (where S = soil factor) + Very low seismicity: areas where a, < 0.04g or a, S < 0.1g (where S = soil factor). © Prof G Koh 2.24 Otte International Pte Ltd ® @ © ‘Advanced Bulking Analysis for Selsmic Design Copyright © 2012. All Rights Reserved Ece Seismic Hazard Assessment Hazard is usually described in terms of a single parameter, i.e. peak ground acceleration (PGA) on type A ground. Earthquake events are random in nature, with many uncertainties associated with intensity, location, duration, attenuation, etc. Probabilistic Seismic Hazard Analysis (PSHA) is usually adopted to determine PGA for a given return period at a give site. ‘Seismic hazard map of Europe for PGA with 10% probability of exceedance in 80 years hitp:Wleurocodes jrc.ec europa euldoc/EUR23563EN. pdf 2.25 Importance Class of Buildings Importance Importance Class Buildings Factor y, v Of vital importance for civil protection, e.g ta hospitals, fire stations, power plants, etc Of importance in view of consequences uM associated with a collapse, e.g. schools, 42 assembly halls, cultural institutions, etc. " Ordinary buildings, not in other categories 1.0 Of minor importance for public safety, e.g. ae agricultural buildings, etc. : Design ground acceleration is increased for operation-critical buildings (x1.4) and large-occupancy buildings (x1.2). 2.26 © Prot G Koh (te inernational Pte Lis ‘Advanced Builing Analysis for Seismic Design Ground Types jaama;—tasaapiion Vane type (mis) * To account for the effect |—|—Trsekor other rook Ike $800 of local soil conditions on Rese cnet ground motion B [Deposits of very dense sand, | 360-800] >80 | >260 + Shear wave velocity (Vs 20) aravol or vory et clay. is the average of the top | ~E—|Brap doposts of ao ease] ETOH 30 m for layered soils. rmediumdonee sand sul Sepoais offooae we mealan | ais aio v= V6Tp » Tesheclontos sll raf predaminanty soft rm Eoneslve sl E [Reel prot sonanng Ta Zurace sui layer wih “aluee of type ¢ ord and {hotness varying between about sm and 20, uncertln by stiffer material with v, Soom. Si] Bepostercanaising a ayer <0 cr) : Teast Om ick or sot | inca Possible large Glaytato wth high v0) amplification or ea Hele lex (P40) and liquefaction, requiring gh water con special study SF | ereoraltveclaye,erany Strorool pote ct Ide shove. LEcs | Elastic Response Spectrum (S,) Building response $= Soil factor 2.580 S,/ a, © Prof G Koh n= damping correction factor (1 for 5% damping) O 5.5 Ground type Ss Ta (s) Ts) To (8) A 1.0 0.15 04 2.0. B 4.2 0.15 05 2.0 c 1.15) 0.20 0.6 2.0 D 1.35 0.20 08 2.0. E 14 0.15 05 2.0 ‘Type 2 — for low magnitude earthquakes of M,<5.5 Ground type s Te) T8) Te A i) 0.05 0.25 12 B 1.35 0.05 0.25 12 c 15, 0.10 0.25 12 D 18 0.10 0.30 12 E 16 0.05 0.25 12 M, = Surface wave magnitude which is a scale of earthquake based on Rayleigh surface waves travelling in top layers 2.29 4 Ecs Type 1 Ground type Spectrum — S, la, . Period T(s) Type 2 | Sar Spectrum Ground type Saliag 5 230 © PIofc G Koh Otte International Pte Lid ‘Advanced Bulding Analysis for Seismic Design Copyright © 2012. Al Rights Reserved Design Spectrum _, Se [ees | + To take advantage of ductility, the elastic spectrum is reduced to the design spectrum by a behavior factor q. * Design spectrum S,(T): 4 O 4 + DCM = Ductility Class Medium 1.5 Lateral force method Ku =F B. Linear dynamic analysis > Modal spectrum analysis Mii(t)+ Cu(t)+ Ku (t) = F(t) C. Non-linear static analysis > Pushover analysis Ku+Fy, =F D. Non-linear dynamic analysis > Time history analysis Mii(t)+ Cu(t)+ Ku(t)+ F(t), = F(t) 235 ECB (A) Lateral Force Method * For buildings with elevation regularity and have fundamental periods of vibration T, smaller than 47, and 2 s. = The response is assumed to be governed by the fundamental mode @ (higher mode contributions are not significant) + Base shear is the single most important quantity for seismic design. * Base shear in horizontal direction: F,=S,(7,):m-A where S,T,) = design spectrum value at T; T, = Fundamental period of vibration in the horizontal direction considered, obtained by eigenvalue analysis or empirical formula. m = total mass of building above foundation or a rigid basement. d = correction factor = 0.85 if T,s2T, and more than 2 storeys, or 1.0 otherwise. Note: Am is to represent the “effective modal mass” of fundamental mode. 2:38 © Prof Koh Otte Intertional Pie Ltd ‘Advanced Building Analysis for Seismic Design Copyright © 2012. Al Rights Reserved. ECB Approximate Fundamental Period (T, ins) T,=C, H°!4 where H = height in m Structure type c Moment resistant space steel frames 0.085 Moment resistant space concrete frames 0.075 Eccentrically braced steel frames 0.075 All other structures Alternative formulae are given in EC8 Sect. 4.3.3. (A quick rule of thumb is 0.1 s per storey.) 0.050 2.2. 237 Summary: Base Shear by Ece Lateral Force Method = Seismic Soil Earthquake investigation type 1 or 2 hoes | Ground type. Parameters — ae [US ta ToT neatly Design spectrum a al sign Reference IF, = S4(T,)-m- A] ground }-—4 Re fa accel (a,) PGA (a5) Importance. Behaviour yp factor (q) | Building 1st period (T,), mass (m), corr. factor (A) © Prot G Koh 2:38 Otte International Pte Ltd ‘Advanced Building Analysis for Seismic Design Copyright © 2012. Al Righs Reserved. pera a [cs] Distribution of Seismic Forces Horizontal seismic force F; acting on storey i Sm, zs, where s;= displacement at storey i of mass m; in the fundamental mode shape. Fah: ‘The fundamental mode shape may be assumed to be linear along the height of the building. Then where z;= height of mass m, above the base. Storey shear induced at Level x is > ae 2.39 Ee (B) Modal Response Spectrum Analysis * 3D structural model is usually used. * All modes of vibration contributing significantly shall be taken into account. = Number of modes depends on effective modal masses: (2) Sum of effective modal masses 2 90% of total mass, or (b) Include all modes with effective modal mass > 5% of total mass. * Altematively, «>3.Jnumber of storeys and 7, <0.2s 2 1 Din ™ Effective modal mass for n-th mode is M, Raed , $n where ©,,,= coefficient at the ith floor of the n-th mode shape (obtained by eigenvalue analysis of the structural dynamic model). = Effective modal mass reflects the contribution of a particular mode of vibration to the base shear. = Sum of effective modal masses for all modes = Total mass of building. 2-40 @ © Prof G Koh (ote ntermaional Pte Lid ue ‘Advanced Building Analysis for Selsmic Design Response of Individual Mode For mode n: Copyright © 2012. All Rights Reserves. ce = Based on the natural period (7,= 217/ w,), obtain the value (S,), from the design spectrum: * Base shear due to mode nis Fay = Sa(Tn) Mp = Do this for all modes that are needed to satisfy the mass participation requirement. Distribution of Seismic Forces in Each Mode In each mode, distribution of seismic force is determined by the mass distribution and corresponding mode shape (which could reverse in sign) Example: © Prof. G Koh Faz Fog Fra | pL iam Fa Mode 2 Mode 3 242 (ote international Pte Lis ‘Advanced Builéing Analysis fr Seismic Design Copyright © 2012, Al Rights Reserved Combination of Modal Responses = Note that the response obtained for each mode is the peak (absolute max.) quantity without + sign. = To combine contributions from all J modes considered, it would be too conservative to simply add the modal peak values (Absolute Sum Rule). = Two common combination rules: > SRSS (Square Root of Sum of Squares, or vector sum) Fy =[DFe | > CQC (Complete Quadratic Combin 1)... Correlation coefficients | kok oe 12 between modes /& j ~ Fy {Du Loja 1 Posh, ] important when two frequencies are close to (within 10% of) each other e@ = The above equation is for base shear as an example, and the same combination rules are similarly applied to other response quantities. = When in doubt, use CQC rule since SRSS is a special case of CQC. 2.43 Be careful in computing the peak value of one response quantity from the peak values of other response quantities. + E.g. to compute storey drift between level 2 and level 3 from the horizontal displacements of levels 2 and peak _ |, peak, peak Agi = ug — Up < + The above computation is wrong since the u, and Us peak values do not necessarily happen at the same time. e Combining Peak Values — Beware! un(0) + Furthermore, the peak (absolute max.) values for u, and u, have lost their + peak peak signs due to combination “204 tule. 244 © Prof G koh ‘Otte nterrational Pte Lia ‘Advanced Building Analysis for Seismic Design Copyright® 2012. Al Rights Reserved Combining Peak Values — The Correct Way + Peak computation has to be done for each mode: n-th mode aK n-th mode ak n-th mode \peak (As speak — (us peak — (uy yk / rthmode ypeak nethmoda) , ¢n-thmode where level 2: (u2 pea =max_| q(t) Ix(o Ye signs are Same peak at same time “> captured in level 3: (y,ithmede peak _ may j g(t) ™thmodey,. (gr-thmade ys mode shape t iyg2nd mode E.g. for mode 2:[ u(t) "POE (fy Mnmede, pithmode 2nd mode at) max absolute value = peak ‘2nd mod (g?me meee (g2nd mode y, ‘Then apply combination rule to combiné the modal peak values of A, by e.g. SRSS: geek = {[(agtstoe pron 4. 1( 2nd mode perky? Na 245 Vertical Elastic Response Spectrum OsT 5 m horizontal or nearly horizontal pre-stressed components; beams supporting columns; base-isolated structures. 246 OPI CG Kon Cte international Pte Lid ‘Advanced Building Analysis for Seismic Design Copyright © 2012. Al Rights Reserved. . . . Ece Load Combination in EC8 — + Load combination involving seismic action: Lewy Aca Diva Ms where + G,, = nominal value of permanent action j (dead load), * Acq = design seismic action (corresponding to the reference retum period multiplied by the importance factor) + Q\; = nominal value of variable action i (ive loads) * Uq1 = quasi-permanent coefficient given in Normative Annex A1 of EN1990 that EQ load is specified at strength design level + Horizontal seismic action is described by two orthogonal components assumed as independent and represented by the same response spectrum. + To combine the effects of the two horizontal components (and also the vertical component if necessary), use _ (2) SRSS rule, or (b) Percentage combination rule: 100% for one direction + 30% for the other direction(s) 247 [kes] Non-Linear Methods Both Method A (lateral force) and Method B (modal response spectrum) are linear methods. = Non-linear method may be used as an alternative. Nonlinear material models shall be used, bilinear force-deformation relationship as a minimum. {C) Non-linear’static (pushover) analysis - EC8 4.3.3.4.2 > Constant gravity loads and monotonically increasing horizontal loads. > Atleast 2 vertical distributions of lateral loads: () Auniform pattem based on lateral forces proportional to mass. (i) A modal pattern in accordance with Method A or B. {D) Non-inear time history (dynamic) analysis - EC8 4.3.3.4.3 > Use at least 7 EQ time histories and take the average of response quantities as the design value in the relevant verifications of 4.4.2.2 (To discuss later) 2.48 ©PrfC.G koh ‘tte Internaoral Pe Lid ‘Advanced Building Analysis for Selsmic Design copy ©2012 A te Resend Prior to 2000, three building codes were used in USA: Uniform Building Code (UBC) ~ West Coast Standard Building Code (SBC) - South East National Building Code (NBC) - East Coast and Midwest vvve * Among the above, UBC was the most comprehensive in seismic requirements (mainly based on efforts of California). * UBC uses seismic zone factor (Z = 1 to 4) to classify seismic risk based on the anticipated max ground acceleration. * In 1994, the three groups formed the “International Code Council” to combine these codes > International Building Codes (IBC) * Major change in IBC: Ground accelerations are specified on a local basis by seismic risk maps (no longer use seismic zone factor). * For technical requirements, IBC 2006 edition makes reference to ASCE 7-05 “Minimum Design Loads for Buildings and Other Structures” (Ch.11 onwards for seismic design). * Latest: 1BC2012 with ASCE 7-10 (but software are not ready yet). 2.49 Ground Motion Parameters + Severity of earthquakes — Code-defined earthquakes evolve over years, €.g UBC 1997: Maximum Capable Earthquake (MCE) — 10% probability of exceedance in 100 yrs (1000-yr return period) Design Basis Earthquake (DBE) - 10% probability of exceedance in 50 years (475-yr retum period) IBC 2006: Maximum Considered Earthquake (MCE) — Most severe earthquake considered by the code (a) Probabilistic MCE - 2% probability of exceedance in 50 years (b) Deterministic MCE - based on characteristic earthquakes in the region (See ASCE 7-05 Ch 21) Design Earthquake — 2/3 of MCE Characteristic periods of ground motion on site — this depends on soil condition > Site Class A-F in IBC2006 2-50 © Prof G Koh ‘Otte international Pre Ltd ‘Advanced Builing Analysis for Seismic Design Copyright® 2012. Al Righls Reserved Example: Map of PGA (in g) for 2% probability of [wee | exceedance in 50 years (MCE) in Western US This is Peak Ground Acceleration Map, based on uniform firm-rock site conditions defined as a site with average shear wave velocity of 760 m/s in the upper 30 meters of the ctust (Site Class B in 1BC2008). Seeeasccascanecs 5 0 ay raw ov ow ae Source: hitp:/ wiotl2008/1128/pdH/OF 08-1128 v1.1 pdt 251 IBC Spectral Response Acceleration (S,) IBC 2006: + Probabilistic MCE S, = 5%-damped acceleration response spectrum having a 2% probability of exceedance in 50 years (return period 2475 years) * Deterministic MCE S, = 150% of largest median 5%-damped spectral response acceleration computed for characteristic earthquakes on all known active faults within the region (subjected to some minimal values depending on site class) Design Earthquake Ground Motion — two-third of MCE ground motion This is the basis for earthquake design Design Response Spectrum — Gives spectral acceleration value (S,) versus building period (7) with 5% damping based on design earthquake 2-52 © PIC G Kon Otte Intemational Pte Ltd Advanced Bung Analysis for Seismic Design Copyright © 2012 Al Rights Reserved [ic Example: Maps of spectral response acceleration for 2% probability of exceedance in 50 years (MCE) in Western US S=S,at0.2s S,=S,atis These seismic hazard maps are based on Site Class B See Contour Maps in Chapter 22 of ASCE 7-05. 253 IBC | Seismic Ground Motion Parameters | = Mapped Acceleration Parameters S, and S, are determined from 0.2- | s and 1-s spectral response from the seismic maps. |e * Site Coefficients F, and F, are needed to obtain adjusted MCE spectral | response acceleration parameters: * Design spectral acceleration parameters are given by Sos = 2/3°Sys_and Sp, = 2/3*Sin For Site Class F which covers soils vulnerable to seismic loading, special analysis is needed using site response analysis in accordance with §21.1 of ASCE 7-05. 2-54 © Prot G koh (ote interational Pte Ltd ‘Advanced Building Analysis for Seismic Design Copyright © 2012. All Rights Reserve. [asc | Site Classification _ Rough equivaience Ecs 1BC2006 * The same three soil parameters as ground 7 WO types defined in EC8 are used: shear wave velocity, SPT and undrained E c shear strength. Cc D Do E + ASCE 7-05 Table 20.3: ES: 8S, | __F [1 ft/s=0.3048 mis, 1 psf=0.0479 kPa) ‘Shear wave velocity | SPT (blows Undrained shear (fils) — average of top per foot) ‘strength (psf) 100 ft ‘A. Hard rock >6000 NA NA B. Rock 2600 0 6000 NA NA C. Very dense soil & soft rock 720010 2500 >60 2000 D. StF soi 600 to 1200 75t050 | 1000t0 2000 E. Soft clay soil <600 <15 <1000 "Any profile > 10 fof sol with plastily index >20, moisture content 240% and unerained shear stength<500 pt. F Soils requiring site response | See Sect. 20.2.1 (e.g iqueiabe sols, very high lastly clay, analysis (Sect. 21.1) peats, highly organic clays, very chick clays) 2.55 . ee IBC Site Coefficients: F, & F, F, (ASCE 7-05 Table 11.4-1) SheCiass | S)5025 | S,=08 | S,=075 | S,=40 | S,2125 A 08 08 08 08 08 8 10 10 40 40 10 c 12 12 1 10 10 D 18 14 12 1A 10 E 26 17 42 08 08 F See Sec 11.47 for sespecfc ground motion procedure F, (ASCE 7-05 Table 11.4-2) Site Class | _S,<0A : 03 [| S04 | S208 a 08 08 08 08 08 8 10 10 10 10 10 c 1 16 18 14 13 2 24 20 18 16 15 E 35 32 28 24 2a F See Sect. 11.4.7 for site-specific ground motion procedure Use linear interpolation for intermediate values. 2-56 © Prof G koh Otte international Pte Lis ‘Advanced Building Analyeis for Seismic Design Copyright © 2012. Al Rights Resawved. iBc | Summary on Getting Sp, and Sp, Soil investigation [Ssen— Seregrrs Obtain Site Class A-F Obtain Site Coeffs F,, F, from Table 20.3-1 from Tables 11.4-1 & 2 Sus =F, a Ss Sm = Fy S1 Design a — a Sor® 2/3 Sin Spectrum IBC Design Response Spectrum for IBC2006 ‘Sp; = Design spectral response acceleration parameter at 1-s period Sps = Design spectral response acceleration parameter at “short” period (T,) a T,= Spr! Sps i Ty = 0.2T, T, = Long-period transition period (geology dependent) Sos n g Spectral Response Acceleration (g) S z 07% ‘Ts; 4.0 qT, Period T(s) [Similar in shape to elastic spectrum in EC8; T, ranges from 4 to 16 s for USA, whereas 2 or 1.2 s is used in EC8.] 258 PIF G koh (tte itrnatonal te Lis ‘Advanced Bulling Analysis for Selsmic Design Copyright © 2012. Al Rights IBC eve. Example of Design Response Spectrum 048 -——-——--—— 2.69 . . IBC IBC-Permitted Analysis 7 Analysis. Seismic Design ASCE 7-05 | Type of Procedure Category" Analysis | (A) Equivalent | B, C — all structures Sect. 12.8 Linear statics Lateral Force | D to F — certain conditions apply; see ASCE 7-05 Table 12.6-1 (B) Modal BtoF Sect. 12.9 Linear dynamics by Response mode superposition, | Spectrum combination rules | (approx) for peak | (C) Seismic BtoF Ch. 16 Linear or nonlinear Response dynamics in time History domain ** For Category A, need only design for lateral load of 1% dead load (§ 11.7) [EC8 includes static pushover analysis as an alternative, whereas in USA this is covered in other documents e.g. ASCE41-06 mainly for performance-based evaluation and seismic rehabilitation. ] © Prof @ Koh 260 ‘Ote International Pte Ltd ‘Advanced Building Analysis for Seismic Design Copyright © 2012, Al Rights Reserved (A) Equivalent Lateral Force Procedure [se] Base shear is computed as a fraction of structural weight: V=0,W where C, is the seismic response coefficient (or base shear coefficient) and Wis the effective seismic weight. [Compare with EC8:] EC8: F, =S,(T,)-m-A * S,in EC8 has the unit of acceleration, whereas C, in IBC2008 is dimensionless (unless mass is used instead of weight). * 182006 has no correction factor A. 261 Seismic Response Coefficient (C,) |~—! V=C,W where C, =Sps (4) Sps= Design spectral response acceleration parameter at short period 1 = Occupancy Importance factor (1 for occupancy category | and Il, 4.25 for Ill and 1.5 for IV) R= Response modification factor depending on seismic force-resisting system, to account for ductility of system. R ranges from 2.2 to 8.5 See ASCE 7-05 Table 12.2-1 [Compare with EC8] « [has the same idea to have larger design force for operation-critical and large-occupancy buildings, but has only 3 categories with different values. * Similar to q in EC8, Ris to reduce the design force by using ductility. The values are summarized in a big table — ASCE 7-05 Table 12.2-1. But the values of R and q may be different for the same structural system * Note that / and R are accounted for here in the base shear equation, unlike EC8 where the factors are included in the design spectrum. 262 © Prof G Koh (Otte nirnatonal Pte Ltd ‘Advanced Building Analysis for Seismic Design Copyright © 2012, Al Rights Reserves. 1Bc | Summary on Getting C, Value (Equivalent Lateral Force Method) Soil investigation 8,517. Seismic maps | Site Class A-F_|—— Site Coefs F,, Fy Occupancy seas) Structural System Cat LIV 7"LYos: 201 Table 12.2-1 Occupancy 7 Importance [++ IC; = Sps (4) RAOMICH Factor! | / R i ; e Seismic Design Detailing Cat A-F requirements 263 IBC Bounds on Seismic Response Coefficients C, value computed is subjected to some upper & lower bounds (Ref. ASCE 7-05 §12.8.1.1) @ = Upper bound: 1 ple for TT, = Lower bound: C, = 0.01 C,=05 5, 2 for $;>0.6g [EC8 has a lower bound B-a, in the design spectrum but no upper bound] 2.64 © Prof G Koh ‘Otte International Pte Lia ‘Advanced Building Analysis for Seismic Design Approximate Fundamental Period (T, in s) Copyright © 2012. All Rights Reserved. Tz =C; hy — where h, = height in m (or in ft)* [isc | Structure type Cc; x Steel moment-resisting frames 0.0724 08 (0.028)* Concrete moment-resisting frames 9.0466 0.9 (0.016)" Eccentrically braced steel frames 0.0731 0.75 (0.03)* All other structural systems 0.0488 0.75 (0.02)* There are other approximate formulae (e.g. 0.1 s per storey) see ASCE 7-05 § 12.8.2.1 Distribution of Horizontal Seismic Forces Horizontal seismic force applied at Level x is ws hk @ y= ; Diy wi) Ay where W, = effective seismic weight at Level x fh, = height from base to Level x k = 1 for T50.5 8, 2 for T22.5 s, else by linear interpolation Storey shear induced at Level xis = The Fi [Compare with EC8: The mode shape is linear or quadratic or in between, whereas EC8 assumes linear mode shape only] © Prot G Koh Base shear (V) 265 2.66 tte International Pte Ltd ‘Advanced Building Analysis for Seismic Design Copyright© 2012. All Rights Re (B) Modal Response Spectrum Analysis + The procedure (including combination rules) is similar to EC8. * Effective modal mass is called modal participating mass. For i-th mode, ak .m, $4] 85 + Number of modes to be included should account for at least 90% of total mass. Ref: ASCE 7-05 §12.9.1 * Check: If base shear computed using this method is less than 85% of the base shear computed using the equivalent lateral force procedure, forces shall be scaled up by 0.85(V/V)). Ref: ASCE 7-05 §12.9.4. [EC8 has no such check. UBC97 has similar check but the factor is 0.9 or 1.0 depending on the irregularity.] UBC 1631.6.4 287 . . ee IBC Seismic Response Coefficient For each mode (/): + Based on the natural period (7; = 2rr/ w,), obtain the spectral acceleration from the design response spectrum: Saf € (Sai + Modal base shear is GS) IV; =(C,); Wi] where |(C,), = —2 2a| ~ and W,= M,g is the modal effective weight for the ith mode. 268 © Prof CG Koh ‘Otte International Pte Lis © ‘Advanced Builing Analysis for Seismic Design Summary on Getting C, Value for Each Mode Copyright® 2012. All Rights Reserved la (Modal Response Spectrum Analysis) Soil investigation 8.8.7. Seismic maps | Site ClassAF |——4{Site Coeffs F,. F, | Sos» Sot Occupancy Structural System Gattlv | , For each mode Tapa on / sas : 7; Occupancy PtSae Importance (C,), 224 R04, Ca Factor / Peemig | Seismic Design Detailing Cat A-F requirements 269 Summary on Response Spectrum Analysis Modal analysis Natural frequencies Mode shapes Modal participating masses Other modal properties Decide # of modes to include (min 90% contribution) For each mode IBC Obtain (S,),from Design Spectrum T Sa) 1 ly, = Sak te ee ‘Compute modal base shear W, Compute seismic forces & other response quantities... I For all modes considered, use SRSS or CQC to compute total base shear Vasa If Vasa < 0.85V ease, then scale up Vasa to match Compare with Vyaie by Equivalent Lateral Force Method © Prof. Koh 2-70 Otte interational Pte Ltd Advanced Buiting Anaysis fr Seismic Design Copyright © 2012. Al Righis Reserved Load Combination in IBC-2006 a + Load combinations involving seismic action: >» Comb $= 1.2*D+1.0*E+ L+0.2*S with E = £,+E, > Comb 6 =0.9°D+1.0*E +1.0°H withE = E,~E, where D=deadload, L=live load, S= snow load, E,,= horizontal EQ load, E, = vertical EQ load = 0.2*S,*D H = load due to lateral earth pressure, ground water pressure, or pressure of bulk materials + Horizontal seismic action is described by two orthogonal components assumed as independent and represented by the same response spectrum. + To combine the effects of the two horizontal components (and also the vertical component if necessary), use Percentage combination rule: 100% for one direction + 30% for the other direction(s).. [Unlike EC8 (or UBC-97), SRSS rule is not stated in ASCE7-05 as an orthogonal combination rule.] (C) Response History Analysis ASCE 7-05 Chapter 16 + Use at least 3 actual earthquake records consistent with those that are consistent with the Maximum Considered Earthquake (in terms of magnitude, fault distance and source mechanisms). + factual records are insufficient, use simulated ground motion records. + For 2-D analysis, scale the ground motions such that the average value of 5%-damped response spectra is not less than the design response spectrum for period range 0.2T to 1.57. + For 3-D analysis, use at least 3 pairs of ground motion acceleration components. For each pair, SRSS is computed for scaling purpose. + The average of the SRSS spectra from all horizontal component pairs should not fall below 1.3 x design response spectrum by 40%, for period range 0.2T to 1.5T. SRS COG¥ISDRS a, @ Prof 6 Koh Scale wp bith x b if te international te Lis Cc + If building is assumed to be linear in the analysis, response parameters are multiplied by (Z/ R) for each ground motion analyzed, where = importance factor and R = response modification coefficient. This scaling is for member design and detailing purpose. + Ifmax base shear is less than code-specified minimum value (lower bound), scaling-up is required for forces to match value) of scaled response parameters is used. + Ifat least 7 ground motions are analyzed, average value can be used. + Ifnonlinear model is used to account appropriately for nonlinear * Ifless than 7 ground motions are analyzed, the worst case (maximum | hysteretic behaviour of structural components, then | (a) dead load + at least 25% live load should be used, | @& \ (b) scaling by //R or Cy/J is not required for response parameters, (c) allowable drift limit is increased by 25%, and (d) design review by an independent team of PE is needed. | 273 | . . IBC Summary on Response History Analysis Select N pairs of horizontal For each scaled pair | ground motion (N23) Run response history analysis ® | For each pair of ground motion Compute response spectra for X & Y motions, and use SRSS to combine * Multiply response quantities by (Z/ R) q * Multiply displacements & drifts by (CG, /Z) Compute average SRSS of all pairs If max base shear Vays < code-specified lower bound, then scale up member forces Compare with Design Response Spectrum (DRS) in the range of 0.2T to 1.57. YT ifthe SRSS spectrum < 0.91.3xDRS, scale up both X & Y ground motions For all ground motion pairs: Take worst response if N <7 Take average response if N= 7 274 © Prof G Koh (ote international Pte Lis ‘Advanced Bulling Analysis for Seismic Design Copyright © 2012. Al Rights Reserved. usc UBC97 ~ Basis of Many Seismic Codes + UBC97 contains extensive seismic provisions and forms the basis of IBC and seismic codes used in many countries. + Caution! Earlier UBC versions (‘94 and before) adopted only “Allowable Stress Design (ASD)" + UBC$97 allowed both ASD and “Load and Resistance Factor Design" (LRFD) but the earthquake loads are specified for LRFD (i.e. strength level instead of service level). + UBC uses seismic zones (1 to 4) for simplicity and use near-zone factors in Zone 4 for places near earthquake faults. [Comparison] » EC8 uses reference PGA to define hazard level, and !BC2006 uses high- resolution seismic risk maps. + Design Basis Earthquake has the same probability as the NDR Earthquake used in EC8, corresponding to return period of 475 years. Design Earthquake in 1BC2006 is defined as 2/3 of MCE which has return period of 2475 years 275 | e UBC (A) Equivalent Lateral Force Procedure Base shearis V=C,W where W= effective seismic weight and C, = base shear coefficient. e Seismic response coefficient to Seismic importance account for local soil conditions & ~ , -* factor (1 or 1.25) regional seismicity Structural system coefficient to account for ductility Li" —_1+.. Fundamental (UBC97 Table 16-N) ~~ period 2561 * Upper bound: ©. = 5" 0.8ZNy 1 = Lower bound: C,=0.11C,! and for Zone 4 onlyCs = R [Compare] R values are similar to 1BC2008, but there are some differences due to change in defining seismic risks and research/field observations of structural behaviour under earthquakes. 1BC2006 has more sub-systems than UBC97. 276 © Prof G Koh (te Intemational Pte Lid ‘Aavance Bling Analysis for Selamie Design Design Response Spectrum for UBC97 2.5C, Spectral Response Acceleration (g) Seismic Zones and Site Profile Type Copyright © 2012, All Rights Reserved. UBC + Cand C, are called seismic response coefficients. + They depend on seismic zone, proximity to active seismic source & soil i-profile type + Values given in UBC97: Tables 16-Q & 16-R [Compare: UBC97 does not have constant- IBC-2007.] displacement part for long period, unlike EC8 and Fundamental Period of Building T(s) Zone factor = effective PGA in g for return period of 475 years 2a7 uBC Zone _| Seismic Zone Factor (2) Approx. MMI 4 0.075 Vtovi 2A 0.15 vil 2B 02 3 0.3 vill 4 0.4 Vill and higher Site profile type S, to Sz are the same as site classification A to E in 1BC2008. [1 #V5=0.3048 mvs, 1 psf=0.0470KPa] | Shear wave velocity | SPT Undrained shear (tvs) strength (psf) > Hard rock 35000 NA NA Rock 2500 to 6000 NA NA Very dense soil & soft rock 71200 0 2500 >50 32000 = Stif soll profile 600 to 1200 15t0 50 1000 to 2000 = Soft soil profile <800 <15 1000 ‘Sol requiting site-specific evaluation OPI CG Koh 2.78 (ote Inirational Pte Lia ‘vanced Building Analysis fr Seismic Design Copyright® 2012. All Rig Seismic Response Coefficients Table 16-0: C, ‘Soll Profile Type | _ Z=0.075 20.18 z02 20.3 Z=04 Ss 0.08 012 0.16 024 0.32N, Se 0.08 016 020 0.30 0.40N, Se 0.09 018 02a 0.33 0.40N, So 012 022 0.28 0.36 aan, | Se 019) 0.30 034 0.36 0.36N, S Require site-specific geotechnical investigation and dynamic site response analysis Table 16-R: C, Soll Profile Type | _2=0.075 250.15 202 203 04 Ss 0.06 012 016 024 0.32N, | Sp 0.08 O15 0.20 0.30 0.40N, Se 013) 025 032 0.45 O.56N, @ So 018 032 040 04 OG4N, | Se 0.26 0.50 0.64 0.84 0.96N,, S Require site-speciic geotechnical investigation and dynamic ste response anal 279 uBC Correlation with IBC2006 + IBC uses seismic risk map with finer resolution than seismic zone factors used in UBC, in defining the spectral acceleration (S,) values. + Ignoring the finer resolution (particularly for seismic zone 4), the following & correlation is noted 2 Spy in 1BC2006 = C, in UBC97 Spg in IBC2006 = 2.5 x C, in UBC97 + For soil type B, C, and C, are the seismic zone factors (Z) except for zone 4 where near source factor has to be considered, and for zone 1, rounded up to 0.08 + Eg. for Zone 2A, Z = 0.15 (i.e. PGA for 10% in 50 years = 0.159), C=C, = 0.15 for soil type B > Spy iN IBC = 0.15 Spg in IBC = 2.5 x 0.15 = 0.375 + For other soil types, C, and C, vary to account for soil amplification (which is. accounted for in IBC via site coefficients F, and F,). 2-80 © ProfC.G Koh Otte International Pte Lid ant utr ats Ser Dn comnewu nRgeRac | UBC Near Source Factors i | + For seismic zone 4 only — to account for proximity to known fault (after | 1994 Northridge and 1995 Kobe earthquakes). | Seismic Seismic Source Definition Kets Description ‘Max. Moment Slip Rate, SR ype Magnitude, M (rmmlyear) A Faults capable of producing large magnitude events ‘and that have a high rate of seismic activity Leu) eee Me70 SRe5 5 | Allfauts other than Types A & M<70 SR>2 Mz65 SRe2 © | Faults not capable of producing large magnitude oe ae earthquakes & a relatively low rate of seismic activity Table 16-S: Ny Table 16-T: Ny Soismic | Closest Distance to Known Seismic | Closest Distance to Known ‘Source ‘Seismic Source ‘Source Seismic Source @ Type [s2km [ sam | 240km type [s2km | 5km | 10km| 215km ~ a Eee t0 A |e] Ke 2 | 10 5 eae] 1.0 1.0 Bf ie pte | 10 41.0 c 10 10 19) ¢ 10 [| 10 | 10 | 10 281 | | | uBC Occupancy Category Occupancy ‘Occupancy / Functions Seismic Importance Category Factor, / 1. Essential faciities | Must remain operative immediately 4.28 following an earthquake e.g. emergency | treatment areas & fire stations. | 2. Hazardous House toxic or explosive substances 4.28 facilities | | 3. Special occupancy | All structures with an occupancy greater 4.00 structures than 5,000 persons, schools, ete | 4. Standard ‘Those not in Gat 1, 2, 3 or 5 (most 7.00 occupancy buildings will fall under this Cat) structures 5. Miscellaneous | Refer to code 7.00 structures Ref: UBC-97 Table 16-« [Compare: 1BC2006 has only 4 categories and / = 1.00, 1.25 & 1.50.) © Prof G kon 2:82 Otte international Pte Ltd ‘Advanced Bulling Analysis for Seismic Design Copyright © 2012. Al Rights Reserved. UBC Summary on Getting C, Value in UBC-97 (Equivalent Lateral Force Method) Soil investigation Zone Factor Z| Seismic zone , maps i J Site Profile Type |_}| Seismic Response |__| Zone 4: Seismic | Sa- Se TL Cosfis C,, Cy Source Type : Near Source | | Cceupancy | Factors Nj, N, Seismic ‘ 1 Structural System acon Table 16-N Detailing requirements iso 2.83 UBC Bounds on Seismic Response Coefficients C, value computed is subjected to some upper & lower bounds @ 2.5C,1 + Upperbound: = Cs =p C,=0.11C, 1 * Lower bound: co. 282M! + Another lower bound for Zone 4 only: $ R 204 © Prof @ Koh ‘Otte International Pte Lio ‘Advanced Building Analysis for Seismic Design Copytight @ 2012. Al Righls Reserved UBC Distribution of Seismic Forces * Distribution of seismic forces is linear except that F an additional force may be applied at the top t level (to account for higher mode effects). = Horizontal seismic force at Level x is F,=0.07TV <0.25V if T>0.7s [Compare] * EC8: Linear distribution without additional force at top. * 1BC2006: Distribution can be between linear and quadratic, without additional force at top. 2.85 Load Combination in UBC-97 we » Load combinations involving seismic action: >» Comb §= 1.2*D+1.0*E+f, *L+h,*S > Comb 6 =0.9*D£1.0*E where D= dead load, L= live load, S= snow load, f, depends on public assembly usage, f, depends on roof configuration, and E = p*E, +E, where E,, = horizontal EQ load E, = vertical EQ load = 0.5*C,*/"D (at least 0.2D for zone 4) = redundancy factor (ranges from 1.0 to 1.5, requires fairly elaborate procedure — see UBC-97 §1630.1.1) + Horizontal seismic action is described by two orthogonal components assumed as independent and represented by the same response spectrum + To combine the effects of the two horizontal components (and also the vertical component if necessary), use (2) SRSS oF CQC tule, or (b) Percentage combination rule: 100% for one direction + 30% for the other direction 2-86 © Prot CG Koh, (te international Pte Lid ‘Advanced Bulling Analysis for Seismic Desion Example: 20-Storey RC Building Analyzed by CS! software ETABS* v9.7, * Software of Computers & Structures Inc., Berkeley, CA; received “Top Seismic Product of the 20" Century Award”. © Prof. Koh Copyright © 2012. A Rights Reserved. Plan view for Storeys 5 ~ 20 Plan view for Storeys 1-4 287 2-88 (te International Pte Lts Advanced Biking Aral fr Sm Design Copier © 2012 Al Rats Resenes a Modal Mass Participation (% ode [Period @) | Ux] Sumux Sumuy_[Re(Torsion)| Sum Re [tesa | “asa 30] 0908 pote too ses oe $e ttt ag eT <—[oas sae SB [oe [Tena Shs [os ass [esa ier tba Tass i63 ooze 08 on [oa 08 sz ee S08 Ts ite at ws i t ee Bs | 43 T 88.4 ‘B47 it te ae i a3 ces 6 st 55 iF a3 aro if ar er i a =7 28 ar na | a oar tr we os @ z so ere e a Set a | oe Sit ar 2 se [oar fa $s 131 a oa Sst te} sa | 29 994 977 02 98.3__| Use 30 C30 99.4 979 417 90.0 _] modes | 209 Case 1: “Zone 3” Earthquake Consider “PGA based on 47-year return period = 0.39 * Shear wave speed V, 2 = 500 m/s ~ very stiff soll + Importance factor = 1 — ordinary building Ecs | "Reference ground acceleration, aye = 0.39 * High seismicity region: Spectrum type 1 «From Table 3.2: S=1.2, T,=0.15, T,=0.5, Tp=7 * Since T;>2To, no correction factor is needed for effective mass, ie. A= 1.0 .0 uBc97 * Seismic zone factor Z= 0.3 * Site profile type Sc * From Tables 16-Q and 16-R: C,=0.33, ©,-0.45 * T.= C,N(2.5C,) = 0.548, Ty= 0.27, = 0.109 18¢2006 © Site class © * Parameters are to be obtained from detailed hazard maps. Otherwise, convert from UBC97: Spg = 2.5C, = 0.825, Sp, = C,= 0.45 = F,=1.0, F,= 1.3, S5=1.2375, 8, =0.5192 | + T.= Sp1/ Spg = 0.546, To= 0.27, = 0.109 2-90 © Prof G Kon Otte international Pte Ltd ‘Advanced Buiding Analysis for Seismic Design Case 2: “Zone 4” & Near Source Earthquake Consider * PGA based on 475-year return period = 0.49 * Shear wave speed v, 49= 500 m/s (same soil as Case 1) = Importance factor = 1 (ordinary building) Ece = Reference ground acceleration, agg = 0.49 = Very high seismicity region: Spectrum type 1 * From Table 3.2: S=1.2, 7,#0.15, T,=0.5, Tp=2.0 = Since T,>2T:, A= 1.0 uUBCS7 * Seismic zone factor Z = 0.4 * Suppose the location is about 5 km from known seismic source type A (large earthquake) Copyright © 2012. Al Rights Reserved. 0.40°N, = 0.48, C,= 0.56*N,= 0.896 747, To= 0.27, = 0.149 «Parameters are to be obtained from detailed hazard maps. Otherwise, convert from UBC97: Sys = 2.5C, = 1.2, Sp, = C, = 0.896 1.0, F, So =1.8, 8, T,= Sp:! Sps= 0.747, To= 0.2T, 201 Example Input to ETABS for Zone 4, X-direction FS ee cea F xoeemr cecnavon ans © veomtnas it Fo © Peomen: . (1 @ ventaes re RI Fe serra oo oe , oo ars feet [wm csnindinia'isis| UBC 1997 jiisite (aes i —— Cnmateeaty meee 4s. | + ECB looks similar in shape but generally deviates in values. 2.93 Comparison of Base Shears for g=R=3 [Note 1] Lateral Force Method [2] Response Spectrum Method [3] Caset:Zone3| Fy, Fy Fy 14} Fox Fy F(a 18C2006 13,335 | 14554 | 19,739 | 13,962 | 12,949 | 19,042 uBce7 13,396 | 14,555 | 19,741 | 13,963 | 12,950 | 19,741 j61 | ECs 13,336_| 14.555 | 19,744 13,085 __| 12,290 | 47,952 Case 2: Zone 4 iec2o06(7) | 30,541 | 30.541 | 43,191 | 22,881 | 21,710 | 36,743 (5) | uscs7 (7) | 30.252 | 30252 | az7e3_| 22,881 | 21,711 | 42,783 (6) | ECB 17,782 | 19406 | 26,321 | 16356 | 15,363 | 22,440 [1] The q and R values are taken as the same for fair comparison and are assumed to have considered structural systems and irregularity according to the respective codes. [2] Lateral force method should not be used in this case as the building is not regular in elevation. The results are listed for comparison only. [3] Use CQC rule to combine modal responses. [4] Use SRSS rule to combine F,, and Fy [5] Scaled up to match 0.85x base shear obtained by lateral force method (I8C2006). [6] Scaled up to match 1.0 x base shear obtained by lateral force method (UBC97). {7} 1BC2006 and UBC97 have significantly higher results than ECS in this case, because near-zone factor increases the seismic force, and also lower bounds apply. 2.94 OProtC.G Koh Otte Intemational Pte Lid | ‘Advanced Building Analysis for Seismic Design Copyright © 2012. Al Rights Reserved Storey Shears for Case 2 0 cenctatfore ag vr vse tt Face se-uic at Foxe ws ~ SHECE Lat Force 18 ~ “SPECS Lat FORE” 6 | retiree srearun 4g __ =e 1oceep specu | “URC Resp Spectrum “@-UBC Resp Spectrum uf cet -EOUREG Sect | 2+ az | 5 0 | 10 2 8 3 |— 8 6 — 4 a} 2 ° : — © 0 $09 1000 ano 2000 25800 20000 35,000 "9 5800 19800 35000 29600 25000 20000 35/00 Storey Shear in X-Direction {kN} Storey Shear in direction (kN) = Response spectrum results for 18C2006 and UBC97 have been scaled up. ® Even if the base shear results are the same, the seismic force distribution are different resulting in different storey shears. 2-98 Key Principles of Seismic Design + Light - Avoid unnecessary mass (seismic force is inertia driven) + Simple & symmetric - Simplicity makes modelling and prediction of seismic behaviour more accurate. Lack of symmetry leads to larger torsional effects. + Uniform distribution of lateral stiffness — Abrupt change causes concentration of high inter-storey shear. * Torsional resistance and stiffness — To resist torsional effects. + Multiple defence lines — Ductility and high degree of statical | indeterminacy are desired. + Good balance between members and connections — Weakest link theory applies. Connection details are important. 2.96 © Prof G Koh Otte Intemational Pte Ltd ‘Advanced Builing Analysis for Seismic Design Copyright© 2012. Al Rights Reserves. Buildings with Irregular Configurations PPOVS ot =a are | YZ w > ni ae ms RS Ge eee rec TORSIONAL FORCES eee fe wo stress concentration unbalanied Fesistance le i | balances resistance: ae) oe) eocentty round motion cutter of mage and resatnce 2 297 Buildings with Abrupt Changes Abrupt changes in lateral resistance ig @ Conan eon renitorsletenestng Abugt canges nse Sanepanse n 2.98 © Prot CG Koh (Ote International Pte Lis ‘Advanced Building Analysis for Seismic Design Copyright © 2012. Al Rights Reserved. Effects of Soft Storey A disproportionate amount of the building's drift is concentrated on the soft storey > vulnerable to collapse Lesson learnt: Good detailing is essential to ensure that structural members and connections behave in a ductile manner when undergoing cyclic loading of large deformation 2-100 © Prof .G Koh (te inermational Pte Ltd e ‘Advanced Bulling Analysis for Seismic Design Copyright © 2012. All Rights Reserved. Seismic Detailing Requirements Depending on the extent of ductility achieved, some structural systems are classified as special (high ductility), intermediate & ordinary. Good detailing is essential to ensure that structural members and connections behave in a ductile manner when undergoing cyclic loading of large deformation. Concrete by itself is a brittle material and has poor ductility. Its weight is also a disadvantage (unless lightweight concrete is used) , as compared to steel structures. For concrete structures to behave in a ductile manner, two important concepts for detailing are continuity and confinement. Stee! of permitted grades is relatively ductile as a material For steel structures, attention is on’ensuring that the structural frame will behave in a ductile manner. In particular, beam-column connections are designed to have sufficient plastic capacity. 2401 Nonlinear Analysis Much more difficult and time consuming to do than linear analysis. Response is not linearly proportional to loading Superposition principle does not apply in general. Upon removal of external loads, the final state is not necessarily the same as the initial state. The sequence of loads matters. Different loading sequences do not necessarily give the same response. To obtain the solution, load has to be applied incrementally (even for static problems), not in one step. 2-102 © Prot G Koh Otte Inerational Pte Lid ‘Advanced Building Analysis for Seismic Design Copyright © 2012. ARights Reserve. Nonlinear Static Pushover Analysis + Main purpose: Performance evaluation of structure from linearly elastic behaviour to first crack and finally to collapse > Capacity. +The capacity is then checked with Demand due to a given or design earthquake + This also provides insight of possible damage mechanisms & weak spots so as to improve the overall design. » Demand for a 23 “given EQ a5 By Be 35 a8 oe Spectral Displacement (Deformation) 2.403 Nonlinear Dynamic Analysis + Apply gravity load — this analysis can usually be linear. + Consider only horizontal components of earthquake loads. + Runa nonlinear step-by-step analysis. + Typical time step = 0.01 or 0.02 s. + Ateach step, compute demands -> compute component DIC ratios > compute limit state usage ratios. If all usage ratios are <1, the performance requirements are satisfied. Pros and Cons of NL Push-Over Analysis versus NL Dynamic Analysis Pros: + Ituses a response spectrum rather than a number of ground motion records. + Static analysis is simpler & faster than dynamic analysis Cons: + Itis approximate, using static analysis to represent dynamic loads and cyclic deformations. + Itdoes not work well for tall or complex structures. 2-104 © Prof G Koh (te International Pte Ltd ‘Advanced Buuling Analysis for Selamie Design Copyright® 2012. All Rights Reserved. NL Force-Displacement Relationship Force (F) Strain Ultimate hardening strength Ductile limit First yield 5 Strength Initial loss stiffness”. Residual strength is Deformation (D) ** Hysterisis loop 2405 Strength Based Design (For Non Earthquake Load) Estimated force capacity (nominal strength) F,, Force demand atfactored load. | _,/*~ “Zetimated force capacity for Force demand design OFn at working load + Check demand versus capacity at component level | ie. Force demand at factored load < Force capacity for design + Use linear analysis to calculate demands — usually ok because structure should remain essentially linear. 2-106 © Prot G Koh (ote Interrational Pie Lid ‘Advanced Building Analysis for Seismic Design Copyright® 2012. All Rights Reserved Earthquake Load is Much Larger é Force demand can greatly exceed design force capacity Check deformation demand @ D + Earthquake design philosophy permits structures to behave in inelastic manner, so as to reduce the design lateral force (for economical reason). 2-407 Deformation Based Design Computed Deformation capacity F | deformation (usually based on demand available ductility) + Check deformation demand < deformation capacity for all components. * Deformation demands must be calculated using nonlinear analysis, because the structure can yield. 2108 @ Prof G Koh Otte international Pte Lid ‘Advanced Builing Analyls for Seismic Design Copyright® 2012. All Rights Reserved Components with Brittle Behaviour 2 Little or no ductility; thus not work Such components must be and capacity Performance Based Design * Conventional design codes do not explicitly establish a performance level. It is assumed that if the design satisfies the code, the building will not collapse in strong earthquakes. + Performance based design seeks to provide reasonable assurance that a design will indeed satisfy a specific performance level. + Most commonly used levels are, according to ASCE41: (a) Immediate Occupancy (IO) - little or no damage (b) Life Safety (LS) — some damage, few or no injury (©) Collapse Prevention (CP) — more damage, no collapse EC8 Part 3: The corresponding levels are the limit states of Damage Limitation (DL), Significant Damage (SD) and Near Collapse (NC). © Prof G Koh designed using force demand deformation approach does 2-109 | 2410 ‘ote Inecratonal Pte Lis ‘Advanced Building Analysis for Seismic Design Copyright © 2012. Al Rights Reserved Deformation Capacities for Different Performance Levels F lo tS cP + 10 —Allows little inelastic behaviour + LS -—Allows substantial inelastic behaviour + CP Pushes the limits of the ductile capacities of the components 24m Key Steps of Performance Based Design * Choose the performance level and the design loads + Define demand-capacity measures: Drift, plastic hinge rotation, shear strength, etc + Get deformation and force capacities (C) Strength: concrete and steel codes Deformation: ASCE 41 (FEMA 358), experiment + Compute deformation and force demands (D) based on structural analysis + Compute D/C ratios and check D/C < 1 for all components. Capacity Desian + Some components can safely be allowed to yield; others should remain elastic. + Instead of relying on analysis to tell us, itis better to decide in advance which components can yield and which must remain elastic. + Then design the yielding components to have sufficient duetility, and the elastic components to have sufficient strength. 2412 © Prof G Koh Otte international Pte Ltd ‘Advanced Building Analysis for Selsmic Design Copyright © 2012. All Rights Reserve. Capacity Design Frame Structure * Specify the components that can yield. Use deformation D/C. —- Beams in bending - Columns at base and at roof + Remaining components should not yield. Use strenath D/C. ~ Columns at except at base and roof ~ Connections ~ Beams and columns in shear ~ Foundations Shear Wall + Specify the components that can vield. Use deformation D/C. ~ Flexural hinging at base of wall ~ Shear in coupling beams + Remaining components should not vield.” Use strength D/C. ~ Bending in wall at all levels except the base ~- Shear in wall at all levels, especially in hinge region -- Foundations ats Effective F-D Relationship Initial curve, for monotonic load F Degraded curve after cyclic loading D * Current practice is to use a degraded relationship, allowing for a reasonable amount of cyclic loading. 214 © Prof G Koh (Otte International Pte Lia ‘Advanced Building Analysis for Seismic Design Copyright © 2012. Al Rights Reserved Practical F-D Models ASCE-41 PERFORM-3D F No strength Hardening stiffness Initial stiffness D D There are so many uncertainties that the, best we can hope to do is to capture the main aspects of nonlinear behaviour — piecewise linear approximation PERFORM-8D is more general than ASCE-41 2116 Typical ASCE 41 Capacities LS capacity: a ri CP capacity: 10 capacity er eo usually at or usually at onset of of CP capacity close to ductile visible damage NN limit F | 4 D + ASCE 41 gives capacity values for a wide range of components (under “acceptance criteria” in ASCE 41). + Alternatively, experimental results can be used. 2116 © Prof G Koh (te Intemational Pte Lia ‘Advanced Building Analysis for Seismic Design Copyright ® 2012. Al Rights Reserved Steel vs. Concrete in ASCE 41 D capacities are D capacities are multiplies of Dy F_ multiplies of (D-D,) Expected strength” Expected strength ~ Up to 6% Up to 10% hardening hardening & i 247, Steps in Static Push-Over Analysis 1. Choose loads. 2a. Apply V load. 2b. Then apply H load and obtain push-over curve. Vertical load + _— ¢ 44 ' t ’ ———--L_______+A ; ' 1 | 3. Using a response spectrum, estimate the “performance point’ (target | displacement) -- see next slide. 4. Compute limit state usage ratios at this point. If all ratios are <1, performance is ok. 2118 Prof. G Koh ‘Otte International Pte Lis Advanced Building Analysis fr Setsmic Design Copyright® 2012. ll Righis Reserved Performance Point by Coefficient Method 4. Get factors C,, Cz and C5 from empirical H formulae. Scale A by C,xC;xC; 3. Given S,, get H 2. Given period T,, and hence D get S, from 5% response spectrum 4b. Stiffness = K,. --] 8, Hence get T, 1a, Estimate A. Construct bilinear approximation. 5. Compare scaled A with Step 1. Iterate until they are equal 2419 Vertical load | | Capacity Spectrum > UT j | de . £2 8 sa a zp Roof displ, A SDOF, ° 3 8 8 8 £ 3 & Spectral displ, Sy | 2-120 | © Prof G Koh Otte international Pte Lid | ‘Advances Bulding Analysis for Selsmic Design Copyright © 2012. All Rights Reserved. Capacity Spectrum * Capacity spectrum obtained from push-over analysis. * Consider first mode only. + Vj and A,oo¢ are converted to spectral values by — Avot FG rot where , Eanal Le be 2 Em & s effective mass for first mode mass participation factor for first mode 94 rot = roof-level amplitude of first mode shape 2421 Different component types: 1. Beam (bending / shear) 2. Shear link 3. Column | 4, Beam-to-coliimn zone; Connection panel zone 5. Diagonal Brace + For each component type, we need -- Reasonably accurate NL F-D relations -- Reasonable deformation and/or strength capacities i | 2122 | eProt CG Koh (te International Pte Lis ‘Aavanced Building Analysis for Seismic Design Beamin Bending | Transverse load oo ALeLe i Clear span + End zones are assumed to be stiff and linearly elastic * Inelastic behaviour can occur in the clear span region Model 4 Use exact beam theory > Nonlinear C moment-curvature (M-\) relation. But curvature demand is very sensitive to the hardening, (zs -{t ee] M, 8 aM, | [ky ky |La9, moor2 MI Kr Kadlae] Finite element model — the result is sensitive to the finite element mesh. 2.123 Model 3: Plastic Hinge Model + Itis assumed that all inelastic deformation is concentrated (lumped) at zero-length plastic | hinges. + Beam between hinges is elastic. + The deformation measure for DIC is hinge rotation. Rotation (8) 2124 © Prof G Koh (te Intemational Pte Ltd Copyright© 2012, Al Rights Reserved @ ‘Advanced Bulking Analysis for Seismic Design Copyright © 2012, Al Righis Reserve. Model 4: Plastic Zone Model + Inelastic deformation ocours in finite- Plasticzone yg M, 6, ’ . 8 length plastic zones. 4 Ny + The plastic zone length is chosen so ( » that the actual M-w curve, which may IT require experimental calibration. Elastic 4 Plastic zone length A reasonable length of plastic zone seems to be 0.5 x beam depth + The deformation measure for D/C can be: e ~- Average curvature in plastic zone -- Rotation over plastic zone (= average curvature x plastic zone length) 2425 | Model 5: Chord Rotation Model End moment M, or (M)) -- Elastic stiffness = 3EI/L, (or 3EV/L) + Usual assumption is L, = L = L/2; thus elastic stiffness = 6EV/L Endrotation The uae but most restrictive 8, or (8) We 2126 © Pro CG Koh ‘Otte International Pte Lid ‘Advanced Building Analysis for Seismic Design Copysight © 2012. Al Rights Reserved, ASCE-41 Deformation Capacities 10 capacity t Moment + The above curve is for components of all types. + ASCE 41 gives capacities for many different components. LS capacity cp capacity Rotation or Curvature * For beams & columns, ASCE 41 gives capacities only for the chord rotation model. 2427 ASCE-41 Capacities - Chord Rotation Moment plastic rotation Rotation from CLC! 8, = yield rotation Performance Level 10 Ls cP Steel Beam 6,/0,=1 9,/=68 | 0,/8, RC Beam Low shear High shear © Prof CG koh 2128 tte international Pte Lis ‘Advanced Bulding Analysis for Seismic Design Beam in Shear + Steel beams: -- Rarely yield in shear. - One exception is shear link in eccentric braced frame (EBF). + Concrete beams: ~- Usually brittle in shear. -- Hence usually design for strength. ood + Ifinelastic behaviour in shear is allowed, it requires a shear hinge. * The deformation capacity can be shear displacement across the shear hinge, or shear strain (= shear disp! J member length). ® Copyright © 2012. All Rights Reserved, TV AI y\ 2-129 Shear Link in EBF L Rigid-plastic i shear hinge « \ U/ Elastic beam segments (add end zones if needed) End rotation essentially = shear strain M, *) + Plastic shear strain for yield in shear (with adequate stiffener) + 10= 0.005 rad, LS = 0.11, CP=0.14 © Prof G Koh 2130 (Otte International Pte Ltd avace Bulg Aalal for Setami Deion copy 6202 its Resene Shear in Concrete Beam Plastic hinge End S zone End | Elastic | Shear strength Rigid-plastic shear section, if inelastic hinge, if inelastic shear is not shear is allowed allowed + If astrength section is used, compute the shear force and hence the strength DIC ratio. + Ifa shear hinge is used, compute the shear displacement and hence the deformation D/C ratio. + Chord Rotation and Plastic Zone models can also be used. 2431 Column Element Models + Basic models same as for beams: -- Chord rotation Plastic hinge; Plastic zone & -- Finite element or exact theory (not preferred) = + But more complicated: --M strength depends on P, and vice versa (P-M interaction) — Moments act about two axes (P-M-M interaction) ~- Hinge behaviour after yield is complex -- Also, shear behaviour is more complex than beam. + P-M interaction: -- Plasticity theory (strictly speaking, not applicable to concrete because crack is not plastic). ~-Alternative: Fiber Section model (more computational time) 2132 © Prof G Koh Otte international Pte Ltd ‘Advanced Building Analysis for Seismic Design Conytight © 2012. All Rights Reserved P-M Interaction Surfaces Steel RC Inelastic if ‘on surface Inelastic if ‘on surface M @ M Elastic if inside surface Elastic if, inside surface 2133 Column Deformation Capacity + Because of P-M interaction, column hinges or plastic zones have both axial and bending deformations. ie + However, the deformation D/C measure is usually based on bending deformation only, not on axial deformation (which is usually small * One complication is that the bending ductility is smaller if the axial compression (P) is large. Hence the rotation capacity of a hinge depends onP. + Asecond complication (for RC columns) is that the bending ductility is smaller if the shear force (V) is large. Hence the rotation capacity of a hinge can also depend on V. 2134 | OPI CG Kon (Otte international Pte Ltd ‘Advanced Building Analysis for Seismic Design Copyright© 2012, All Rights Reserved ASCE-41 Concrete Column Capacities Rotation Capacity Same as beam At low shear 0.75 x beam At high shear: 20% = smaller capacity + ASCE 41 gives capacities only for the chord rotation model + Fora concrete column, the chord rotation capacity depends on the axial and shear forces, as shown above. 2138 Shear in Concrete Column Shear Shear Strength Strength Compression Tension Hinge rotation + Shear strength is affected by axial force and hinge rotation. * Use strength section if inelastic in shear is not allowed. * Use shear hinge if inelastic in shear is allowed. * Chord Rotation and Plastic Zone models can also be used. 2196 © Prof G Koh Otte international Pte Lid

You might also like