Download as pdf
Download as pdf
You are on page 1of 28
a 5S Numerical Evaluation of Dynamic Response PREVIEW Analytical solution of the equation of motion for a single-degree-of-freedom system is usually not possible if the excitation—applied force p(t) or ground acceleration di(¢)— varies arbitrarily with time or ifthe system is nonlinear. Such problems can be tackled by ‘numerical time-stepping methods for integration of differential equations. A vast body of literature, including major chapters of several textbooks, exists about these methods for solving various types of differential equations that arise in the broad subject area of ap- plied mechanics. ‘Th literature includes the mathematical development of these methods; their accuracy, convergence, and stability properties; and computer implementation. Only a brief presentation of a very few methods that are especially useful in ‘dynamic response analysis of SDF systems is included here, however, This presentation is intended to provide only the basic concepts underlying these methods and to provide a few computational algorithms, While these would suffice for many practical problems and research applications, the reader should recognize that a wealth of knowledge exists ‘on this subject. 5.1 TIME-STEPPING METHODS For an inelastic system the equation of motion to be solved numerically is mit + ott + felu, i) = poy 6.) 185 158 Numerical Evaluation of Dynamic Response Chap. 5 subject to the initial conditions (0) 20), ‘The system is assumed to have linear viscous damping, but other forms of damping, including nonlinear damping, could be considered, as it would become obvious later. This is rarely done for lack of information on damping, especially at large amplitudes of motion. The applied force p(t) is given by a set of discrete values pj = ptt), i= 0 to 1 (Fig. 5.1.1). The time interval Ans lisa — te (5.1.2) is usually taken to be constant, although this is not necessary. The response is deter- ‘mined at the discrete time instants , denoted as time é; the displacement, velocity, and acceleration of the SDF system are wi, ii, and ii, respectively. These valves, assumed to be known, satisfy Eq. (5.1.1) at time i: mils + ett + fs): = Ps 613) where (fs); is the resisting force at time i; (fs); = kuy for a linearly elastic system but would depend on the prior history of displacement and velocity ifthe system were inclas> tic, The numerical procedures to be presented will enable us to determine the response ‘quantities ir... 1i41, and iizgy at the time instant i-+1 that satisfy Eg. (5.1.1) at time i-+1: mir + ctiys + FS) = Pint Gray % es Figure S.1 Notation for tine sepring metods ‘See. 5.2 Methods Based on Interpolation of Excitation 187 When applied successively with £ = 0, 1,2,3,..., the time-stepping procedure gives the desired response at all time instants 7 = 1,2,3,.... The known initial conditions provide the information necessary to start the procedure. ‘Stepping from time 7 to i+1 is usually not an exact procedure. Many approximate procedures are possible that are implemented numerically. The theee important require- ‘ments for a numerical procedure are (I) convergence—as the time step decreases, the ‘numerical solution should approach the exact solution, (2) stabiity—the numerical solu- tion should be stable in the presence of numerical round-off errors, and (3) accuracy—the ‘numerical procedure should provide results that are clase enough to the exact solution, Although these are important issues, they are mentioned only briefly in this book. Three types of time-stepping procedures are presented in this chapter: (1) methods based on interpolation of the excitation function, (2) methods based on finite difference expressions of velocity and acceleration, and (3) methods based on assumed variation of acceleration. Only one method is presented in each of the first (wo categories and two from the third group. 5.2 METHODS BASED ON INTERPOLATION OF EXCITATION A highly efficient numerical procedure can be developed for linear systems by interpo- lating the excitation over each time interval and developing an exact solution fora linear system using the methods of Chapter 4. Ifthe tie intervals are shor, linear interpolation is satisfactory. Figure 5.2.1 shows that over the time interval 4, < ¢ < fists the excitation function is given by Ap, ploy = wt (52.10) where Api = put = pi (2.1b) and the time variable x varies from 0 to Af,. For algebraic simplicity, we Gist consider systems without damping; later, the procedure will be extended to include damping. The > Actual Pst > | Interpolated: pts) % Figare S21. Notation for linearly es interpolated excitation. 158 Numerical Evaluation of Dynamic Response Chap. 5 equation to be solved is . 4p. mit ku = py + Be (622) The response u(z) over the time interval 0 < t < Ay is the sum of three pans (1) free vibration due to initial displacement u; and velocity ti, at + = 0, (2) response to step force p; with zero initial conditions, and (3) response to ramp force (Ap, /A4,)€ with zero initial conditions. Adapting the available solutions for these three cases from Sections 2.1, 4.3, and 4.4, respectively, gives % nme 2 yy BR (E_ sinont wer) =mvcosoue + Hsinoye + BA concn) + AE (SRA) (5229) and i@) an, Fi cosane + sinaye + 9% 11 cosayn ae Hsinage + cosayr + F sinaye + PU nt) (5.2.36) Evaluating these equations at r = Ag gives the displacement u,, and velocity iy, at time i +1 1 =H cos(en, At) + sin) 7 Apt + Ett —cox(a, any) + Pay, at, —sin(ay At) (5-249) a Ah Met =, sin(ay At) + costem 1) n oy +B simian a1) + 92 — costwn At 5.2.4b) k Kk Oy ty ‘These equations can he rewritten after substituting Eq, (5.2.1b) as recurrence formulas: Minn = Aus + Bi) + Cry + Doin (5.2.5a) thy) = Alu; + Bid, + C'D; + D' Dis (5.2.5b) ‘These formulas also apply to damped systems with the expressions for the coeffi- cients A, B, ..., D? given in Table 5.2.1 for under-critically damped systems (i, £ <1). They depend om the system parameters w, f, and Z, and on the time interval Are a, Since the recurrence formulas are derived from exact solution of the equation of motion, the only restriction on the size of the time step Af is that it permit a close approximation to the excitation function and that it provide response results at closely spaced time intervals so that the response peaks are not missed. This numerical provedure is especially useful when the excitation is defined at closely spaced time intervals—as for earthquake ground acceleration—so that the linear interpolation is essentially perfect, If the time step ¢ is constant, the coefficients A, B, ..., D’ need to be computed only ‘once. ‘See, 5.2 Methods Based on Interpolation of Excitation 159 TABLE 5.2.1 COEFFICIENTS IN RECURRENCE FORMULAS (¢ < 1) wp a1 teneo a) ‘The exact solution of the equations of motion required in this numerical procedure is feasible only for linear systems. It is conveniently developed for SDF systems, as shown above, but would be cumbersome for MDF systems unless their response is ‘obtained as superposition of modal responses (Chapters 12 and 13). Example 5.1 ‘An SDP system has the (llosting propenies: m = 0.2533 kip-ec*in, & = 10 kipsfin Ty = 1 se (ony = 6283 radsec), and ¢ = 0.08. Determine the response u(t) of this, system to p(t) defined by the hal-cycle sine pulse force shown in Fig ES.L by (a) using piecewise-linear interpolation of p(t) with At = 0.1 sec, and (b) evaluating the theoretical solution, Solution 1 Initial calewations e881 = 9.9691 wy = yy 78 = 6.275 sinp At = 05871 coswp At = 0.8095 180 pokips 10 Numerical Evaluation of Dynamic Response Chap. § 1 10sin (nr /0.6) 8.66 Piecewise linear 06 Sabsttuting these in Table 5.2.1 gives ae 8129 3.5795 B= 8 09067 17559 interpolation | 1, see Figure ES C =0012% © _D = 0.006352 C019 Di =0.871 2. Apply the recurrence equations (5.2). The resulting computations are summa- rized in Tables B5.la and FS.1b. TABLE E5.1a NUMERICAL SOLUTION USING LINEAR INTERPOLATION OF EXCITATION nm om 09 0.0000 0.0000 01 5.0000 0.0618 02 86602 0.1070, 03 10.0000 0.1236, 04 8.6603 0.1070 05 5.0000 0.0618 96 0.0000 0.0000, 07 0.0000 0.0000 08 9.0000 0.0000 09 0.0000 0.0000 3. Compute the theoretical response. Dpest 0.0318 0.0550 0.0685 Bi 0.0000 0.0888 2782 0403 04290 0.1783 0.2735 ~0.6767 0.8048 96272 0.0000 0.9354 3.0679 42558 47318 19336, 3.0159 =74631 38765 69177 25171 Au 0.0000 0.0288 0.1809 05150 og218 12109 1am 0.7346 oar 08160 0.0000 00318 02274 016336 1.1339 14896 1.4480 0.9037 0.0579 01577 10432 Theoretical 0.0000 0.0328 02332 O87 1.1605 saay 14814 ones 0.0593 07751 12718 Equation (3.2.5)—valid for 1 < 0.6 see, Eq, 224) modified appropriately—valid for 1 > 06 sec, and the derivatives of these ‘two equations are evaluated foreach t; the results are given in Tables ES.ta and ES.1b. 4. Check the accuracy of the numerical results ‘The numerical solution based on piecewise linear interpolation of the excitation agrees reasonably well with the theoretical solution. The discrepancy arises because the half-cycle sinc curve has been replaced by the series of straight lines shown in Fig. ES.1 approximation would be closer to the hall-eycle sine curve, and the numerical solution ‘would be more accurate ‘With a smaller Ar the piecewise linear Sec. 5.3 Central Diference Method 161 ‘TABLE E5.1 NUMERICAL SOLUTION USING LINEAR INTERPOLATION OF EXCITATION 6m Cm Dip Aa: 4 Bi iy Theoretical i 00 0.0000 0.0000 0.9354 0.0000 0.0000 0.0000 0.0000 0.0000 01 5.0000 08544 1.6201 0.1137 0.0318 0.7071 09358 09567 02 6602 1.4799 18707 08140 0.2274 23192-30679 3.1383, 03 10.0000 1.7088 1.6201 -2.2679 0.6336 3.6708 «4.8558 «4.9674 0 86603 1.4799 09384 —a0SKR 1.1339 35771 A738 4.8408 05 5.0000 0.8544 0.0000 5.3320 1.4896 1.4617 1.9336. 1.9783. 06 0.0000 0.0000 0.0000 5.1832 1.4480 -22799 30159 —3.0848 07 0.0000 0.0000 0.0000 -3.2387 0.9037 ~S.6418 -74631 -7.6346 08 0.0000 0.0000 0.0000 —0.207 0.0579 6.7103. —R8765 — 9.0808, 09 0.0000 0.0000 0.0000 2.7124 -0.7577 -52295 69177 —70771 1.0 0.0000 12482 “25171-25784 5.3 CENTRAL DIFFERENCE METHOD This method is based on a finite difference approximation of the time derivatives of displacement (j., velocity and acceleration). Taking constant time steps, At; = At, the central difference expressions for velocity and acceleration at time # are =u 2a; tu nF Substituting these approximate expressions for velocity and acceleration into Eq, (5.1.3), specialized for linearly clastic systems, gives 63.) Ar (632) In this equation u, and uj. are assumed known (from implementation of the procedure for the preceding time steps). Transferring these known quantities to the right side Teads to we me 2m [cart as) [ae as E-aa a fg) = where (ane * Zar 635) 162 ‘Numerical Evaluation of Dynamic Response Chap. § and , me an aen-[aip-as hol ak eo ‘The unknown u,», is then given by (637) ‘The solution ia) at time i + 1 is determined from the equation of motion atime i without using Bq. (5.L4), the equation of motion at time i-+ 1. In pancular, the elastic ann damping forces can be computed expicily using known displacements, and i, and velocities w; and d;_1. Such methods are called explicit methods. ‘Thus up and u_; are required to determine u,; the specified initial displacement uo is known, To determine uy, we specialize Eq (5.3.1 for i = 0 to obtain fiy= (538) Solving foe u; fom the first equation and substituting in te second gives an? 2) =~ Artin) + Sig (639) ‘The initial displacement uo and initial velocity uo are given, and the equation of motion at time 0 (= 0), ‘mito + cttg + kup = Po provides the acceleration at time 0: fig = panei bee 63.10) ‘able 5.3.1 summarizes the above-described procedure as it might be implemented on the compute. The central difference method will “blow up. presence of numerical round-off i the time step chasen is not short enough. The specific requirement for stability is ‘This is never a constraint for SDF systems because a much smaller time step should be chosen to obtain results that are accurate. Typically, At/T, < 0.1 to define the response adequately, and in most earthquake response analyses even a shorter time step, ‘ypically Ar = 0.01 to 0.02 sec, is chosen to define the ground acceleration iig(?) accurately. Sec. 5.3 Central Difference Method 163 TABLE 53.1 CENTRAL DIFFERENCE METHOD 1.0 Ininal catewlarions 210 Caledaions fr tn 21 =m maui—bu 22 23° If required: 4 ao 3.0 Repetition for uhe nex time step [Replace | by i + 1 and repeat steps 2.1, 2.2, and 2.3 for dhe nent tame step. Example 5.2 Solve Example 5.1 by the central difference method using Ar Solution 110 Initial catewations m= 02533, 0.1592 “ant 220 Coleulations for each time step Pi ~ S31 + 40.660, 2a fi = py aus bu 22 wis 30 Computational steps 2.1 and 22 are repeated for i = 0, 1, 2, 3, ... leading to ‘Table E5.2, wherein the theoretical result (om Table E5,1a) is aso included. 164 Numerical Evaluation of Dynamic Response Chap. 5 ‘TABLE E5.2 NUMERICAL SOLUTION BY CENTRAL DIFFERENCE METHOD i tnt eB RY} Bq. 22))- Theoretical 0.0 0.0000 0.0000 0.0000 0.0000 000.0828 0.1 5000 0.0000 0.0000 5.0000 ase 0.2332 02 6602 0.0000 O.1914 1644190620368 03 10.0000 i914 06293. 3089 1.1825 1.1605 04 Roos 06293 1.1825 41.3001 1380815281 05 Som) 1.1825 1.5808 402619 15412 LB 06 0000) 15808 «15812288809 OBIT ——a.92as 07 000m 154120914 06456-00247 0.0893 08 0000 O94 00247 23.4309 —a868 0.7751 09 a0 00247-0896 358598 1376 1.2718 10 0.0000 08968 13726 334058 —12940 1.2674 5.4 NEWMARK'S METHOD 5.4.1 Basic Procedure In 1959, N. M. Newmark developed a family of time-stepping methods based on the following equations: tags = ty (YD ANN + ADL (54.10) y+ (Adis + [05 prAN?] i + [BAO] — 6.4.18) ‘The parameters and y define the variation of acceleration over a time step and determine the stability and accuracy characteristics ofthe method. Typical selection for y is $ and <<} is satisfactory from all points of view, including that of accuracy. These two cquations, combined with the equilibrium equation (5.1.4) at the end of the time step, provide the basis for computing u,41, ai=1, and iy, at time #-+ 1 from the known u;, 4, and i; at time i, Iteration is required to implement these computations because the unknown i. appears in the right side of Eq, (54.1) For linear systems itis possible to modify Newmark’s orginal formulation, how- ever, to permit solution of Eqs. (54.1) and (6.1.4) without iteration, Before describing this modification, we demonstrate thet two special cases of Newmatk’s method are the well-known average acceleration and linesé acceleration methods, 5.4.2 Special Cases For these two methods, Table 5.4.1 summarizes the development of the relationship between responses ust. ist, and ii at time # +1 to the corresponding quantities at time f, Equation (5.4.2) describes the assumptions that the variation of acveleration over Sec. 5.4 Newmark’ Method 165 TABLE 5.4.1 AVERAGE ACCELERATION AND LINEAR ACCELERATION METHODS ‘Average Acceleration Linear Acceleration he 4 4 4 ' ‘ ~ cr! — $i +H) Wile) = ty + Gig — id (54.2) ie) = + Hse 4 He) = iy FE + Ean —) 643) vty + ia 4) digs =H + Bia + 644) tle) Ss + die + Bing +H) we) =u) tive + + lin — ih) GAS) mutha MEG FH) gm dE COO (Bins +4) 8) «time step is constant, equal tothe average acceleration, or linear. Integration of itt) ss Eq, (5.43) for the variation i(r) of velocity over the time step in which r = At substituted to obtain Eq. (5.44) for the velocity si; at time i + 1. Integration of i(z) gives Bq. (5.4.5) for the variation u(z) of displacement over the time step in which + = Ar is substituted to obtain Eq. (5.4.6) for the displacement u41 at time i-+1. Comparing Eqs. (5.4.4) and (5.4.6) with Eq. (5.4.1) demonstrates that Newmark’s cquations with y = {and 6 =} ate the same as those derived assuming constant average acceleration, and those with y = + and f = } correspond to the assumption of linear variation of acceleration. 5.4.3 Noniterative Formulation ‘We now return to Eg. (5.4.1) and reformulate it to avoid iteration and to use ineremental quantities: Au; erm Oty yy — A, i (6.4.7) el — Pi 6.48) While the incremental form is not necessary for analysis of linear systems, its introduced because it provides a convenient extension to nonlinear systems, Equation (5.4.1) can dpi 168 Numerical Evaluation of Dynamic Response Chap. 5 be rewritten as, (an? Aly = (ANH +780 Am Au = (AN + SP + plan? aii, (5.4.9) 2 ‘The second of these equations can be solved for je a t= Fane! Bas (54.10) Substituting Bq. (5.4.10) into Bq, (5.4.99) gives y x y sive giganZasa(i-Z)n wan Fartu— bint aii F 64.10) Next, Bqs. (54.10) and (5.4.11) are substituted into the incremental equation of motion: mits +e ii: +k Au; = Ap, (54.12) cobiained by subtracting Eq. (5.1.3) from Eq. (5.1.4), both specialized to linear systems with (fs); = ku; and (fs)i41 = kis). This substitution gives Kau = Apr (5.4.13) where Perey bans Soot 6414) and , loo, 1 y aii ant (gagmt he) [soma ar (S-")<]a (64.15) With & and A; known from the system properties m, k, and c, algorithm parameters and 8, and the ti; and ii, at the beginning of the time step, the incremental displacement is computed from Ab du, = SP uns 64.16) Once Au; is known, Ait, and Aiiy can be computed from Eqs. (5.4.11) and (5.4.10), respectively, and u+1» 15 and ii, from Eq. (5.4.7) ‘The acceleration can also be obtained from the equation of motion ati, Piss ~ cttigs — kui a 64.17) rather than by Bqs. (5.4.10) and (5.4.7). Equati ‘the computations [see Eq. (5.3.10) In Newmark’s method the solution at time ( + 1 is determined from Eq, (5.4.12), ‘which is equivalent to the use of the equilibrium condition, Bq, (5.1.4), at time i= 1 ‘Such methods are called implicit methods. ‘Table 5.4.2 summarizes the time-stepping solution using Newmark’s method as it ‘ight be implemented on the computer. (5.4.17) is needed to obtain ii to start Sec. 5.4 Newmark’s Method 167 ‘TABLE 5.4.2 NEWNMARK'S METHOD: LINEAR SYSTEMS Special eases (1) Average acceleration medhod (y (@) Linear acceleration method ( 1.0 Initia eatewlations ~ city — ho 1 fig = Been = tte 1.2. Select ar 20 Calculations for eack tine step, ¢ 21 A= Bp tae + bit Ay 22 au = AP y iy = oan - 28 nag t Loa Fane! al BM 25 igs ast Ale tt = Ai igs = ts + Bi, 24 aia 30 Repetition forthe nest time step. Replace ¢ by 1+ 1 and implement steps 2.1 10 2.5 for the rent time step Newmark's method is stable if rn TB 4 and f= } this condition becomes, (5.4.18) ‘This implies that the average acceleration method is stable for any A‘, no matter how large; however, it is accurate only if Ar is small enough, For y = 3 and f= }, Eq, (5.4.18) indicates thatthe linear acceleration method is stable if at Fs 05st However, as in the case of the central difference method, this condition has Fttle signif icance in the analysis of SDF systems because a much shorter time step than 0.5517, must be used to obtain an accurate representation of the excitation and response. 168 Numerical Evaluation of Dynamic Response Chap. 5 Example 5.3, Solve Example 5.1 by the average acceleration method using A Station 10 Initial cateutations O41 se, m=0253 k= 10 =O tig =O chy 11 ig = BSS 1a oreo wes ts bate et tomes a * Gn 1h ax Ln te= 148 ab =n = 05068, 2.0 Calculations for each time step BA Mj = Api ay + bis = Apr + 10.45, + 0.5066 Abi _ Ab: 145 ain = By = 200; — 24 a ‘4 cam, — atin) 2, an Dm 25 wis met Aus, 100( Am, ~ 0.1%;) ~ 2 i Mls ty j= A ‘30 Repetition for the nex time step. Steps 2.1 to 2S are repeated for successive time seps and are summarized in Table E5.3, where the theoretical result (from Table ES.18) is also included: Bxample 8:4 Solve Example 5.1 by the linear acceleration method using Av = 0.1 se Solution 1.0 Initia calowations m=02533 k= 10 592 w=0 ig=0 pad = cig — 11 iy = P= Stoo 12 ara. os Om a’ * ant 6 14 a= Sms 13k 166.8 6h snd b= an ~ Sem 0767, OOHLAN NOLLVHaTZ00¥ JOVEAY AB NOLLNIOS WOWSWNN. €'sa TTeVL 169 170 Numerical Evaluation of Dynamic Response Chap. 5 20 Calculations foreach time step 2A AA = Ap) bail + bil = py + 15.68, +0.7679%), 22 An, = Sh = SP 23 by = Aft = 304u; ~ 3a; - 0.05%). 24 ily = = Bi = 6OU( Aa, — 0.109) — 3 2S ig = uit Any egy = ti + At ig tip i 300 Repetition for she next time step. Steps 2.1 t0 2.5 are repeated for successive time steps ‘and are summarized in Table ES.4, where the theoretical result (from Table ES.1a) is also included. 5.5 STABILITY AND COMPUTATIONAL ERROR 5.5.1 Stability Numerical procedures that lead 1 bounded solutions if the time step is shorter than some stability limit are called conditionally stable procedures. Procedures that lead to bounded solutions regardless of the time-step length are called unconditionally stable procedures, The average acceleration method is unconditionally stable. The linear ac- celeration method is stable if A/T, < 0.951, and the central difference method is stable if At/T, <1 ‘The stability eviteria are not restrictive (1e., they do not dictate the choice of time step) in the analysis of SDF systems because Ar/T, must be considerably smaller than the stability limit (say, 0.1 or less) to ensure adequate accuracy in the numerical results. ‘Stability of the numerical method is important, however, in the analysis of MDF systems, Where itis offen necessary fo use unconditionally stable methods (Chapter 15). 5.5.2 Computational Error Error is inherent in any numerical solution of the equation of motion. We not discuss error analysis from a mathematical point of view, Rather, we examine two of the important characteristics of numerical solutions to develop a feel for the nature of the errors, and then mention a simple, useful way of managing error Consider the free vibration problem (0) it + kw 1 and #0) =0 for which the theoretical solution is ees 65.) more] — 1f0r'r —PSTSO- 91948-00000. eIIE — BIZVT—LzeBO-SOBTEEI- 00000 vassee 629 ES'O- OTF LET- 00000 1NEE~ ODL Grs7Z —&SSP'Y— OTISO- B6ITSe~ 00000 SHNVE~ O00 90 EE0r'9I- Fo6L'y- ESTO] suzVz- O000'S~9zhSEE- OOS SO LITLE 9EEDT- TSE EBSD TOB'E- AcE TI- HBR. YO LIST 96C00~ FGHO PRL ZB © LoEEI- LET ZL —OODOT_ 0 I6IS I= UoHeT —ELGEO ——GLPTON RECT 6gET cM TO 900s PERT —«HORTO- GrLSIE © SOHE —COGHLI_ —OONS TO 066.1 S66 OEVO 0000'S GOD'S COD O00 m (ez das) Gz dag) Tdos) (ETdng) (rds) Czdas) dy Tag) ronan, in In inv ayy 49 ™ (HIS NOLWEST300¥ HVSNIT AG NOLS WORBWNN 993 TT8VL a 172 ‘Numerical Evaluation of Dynamic Response Chap. § ‘This problem is solved by four numerical methods: central difference method, average acceleration method, linear acceleration method, and Wilson's method. The last of these methods is presented in Chapter 15. The numerical results obtained using ‘At =0.17, are compared with the theoretical solution in Fig. 5.5.1. This comparison shows that some numerical methods may predict that the displacement amplitude decays, ‘with time, although the system is undamped, and that the natural period is elongated or shortened Liner acelersion "Tegal Average ateleration Le \e Wilson WT, Figure $5.1. Fie vibes slain, luton by fur numerical methods (4 /T, = 0.1) the thesis Figure 5.5.2 shows the amplitude decay AD and period elongation PE in the four numerical methods a5 a function of As/T,; AD and PE sre defined in parts (b) and (©) of the figure, respectively. The mathematical analyses that led to these data are not presented, however. Three of the methods predict no decay of displacement amplitude. Wilson’s method contains decay of amplitude, however, implying that this method in- twoduces numerical damping in the system; the equivalent viscous damping ratio E is shown in part (a) of the figure. Observe the rapid increase in the period error in the central difference method near Ar/T,, = 1/r, the stability limit for the method. The ‘central difference method introduces the largest period error. In this sense it isthe least accurate of the methods considered. For A1/T, less than its stability limit, the linear acceleration method gives the least period elongation. This property, combined with no ‘amplitude decay, make this method the most suitable method (of the methods presented) for SDF systems, However, we shall arrive at a different conclusion for MDF systems because of stability requirements (Chapter 15). ‘The choice of time step also depends on the time variation of the dynamic exci- tation, in addition to the natural vibration period of the system. Figure 5.5.2 suggests that Ar = 0.17, would give reasonably accurate results. The time step should also be AD See. 5.5 Stability and Computational Error 173 ® @ ue “ or os © Witon's method om 0.08 04 me 03 0.06 03 c mR 004 02 e on fi om 'S on Cent sixence x /rkemer tin a acceleration erage acceleration ole de jot 09 O02 03 oa ait, " ® on 02 AIT, Figure $52. (@) Amplitude dovay versus A/T, (b) dfnition of AD and PE; (6) pio elongation vers A/T short enough to keep the distortion of the excitation function to a minimum, A very fine time step is necessary to describe numerically the highly iregular earthquake ground acceleration recorded during earthquakes; typically, Ar = 0.02 sec is chosen and this dictates a maximum time step for computing the response of a structure to earthquake excitation, One useful, although unsophisticated technique for scleting the time step is to solve the problem with a time step that seems reasonable, then repeat the soltion with a slightly smaller time step and compare the results, continuing the process until two successive solutions are close enough, ‘The preceding discussion of stability and accuracy applies strictly to Linear systems. ‘The reader should consult other references for how these issues affect nonlinear response analysis. 174 Numerical Evaluation of Dynamic Response Chap. § 5.8 ANALYSIS OF NONLINEAR RESPONSE: CENTRAL DIFFERENCE METHOD The dynamic response of a system beyond its linearly elastic range is generally not amenable to analytical solution even if the time variation ofthe excitation is described by a simple function, Numerical methods are therefore essential in the analysis of nonlinear systems, The central difference method can easily be adapted for solving the nonlinear equation of motion, Eq. (5.1.3), at time i. Substituting Eqs. (5.3.1), the central difference approximation for velocity and acceleration, gives Eq. (5.3.2) with ku replaced by (5). ‘which can be rewritten to obtain the following expression for response at time i + I: fuss = Bi Gel where tare Gas 562) and me 2m as Tat 7 bar 1 Pi at 663) ‘Comparing these equations with those for linear systems, itis seen thatthe only difference is in the definition for 4,. With this modification Table 5.3.1 also applies to nonlinear systems, ‘The resisting force (fs), appears explicity as, it depends only on the response at time i, not on the unknown response at time ‘+1. Thus itis easily calculated, making the central difference method perhaps the simplest procedure for nonlinear systems. Although attractive in this regard, the method is not popular for practical or research applications because more effective methods are avaiable. 5.7 ANALYSIS OF NONLINEAR RESPONSE: NEWMARK’S METHOD In this section Newmark’s method, described in Section 5.4 for linear systems, is extended {0 nonlinear systems. Although not as simple as the central difference method, it is pethaps the mast popular method because of its superior accuracy, ‘The difference between Eqs. (5.1.3) and (5.1.4) gives an incremental equilibrium equation: m Ail, +e Aly + (Afs)i = Bp: 67.) ‘The incremental resisting force (Afs)i = Wace Au 672) where the secant stiffness (K))zc, shown in Fig. 5.7.1, cannot be determined because 1.1 is not known. If we make the assumption that over a small time step Af, the secant ‘Sec. 5.7 Analysis of Nonlinear Response: Newmark’'s Method 175 fe eh 8 “igure $7.4 stiffness (Kj sec could be replaced by the tangent stiffness (k))r shown in Fig. 5.7.1, then Eg, (57.2) could be approximated by (Ofs)s = (kde Au; ($73) Dropping the subscript T' from (Jr in Eq. (5.7.3) and substituting it in Eq. (8.7.1) ives «mbit; +e, +k Au, = Ap, (74) “The similarity between this equation and the corresponding equation for linear systems, Eq, (54.12), suggests that the noniterative formulation of Newmark's method presented cater for linear systems may also be used inthe analysis of nonlinear response. All that rceds to be done is to replace k in Eq, (S.4.14) by the tangent stiffness k; to be evaluated at the beginning of each time step. This change implies that step 1.3 of Table 5.4.2 should follow step 2.1. For nonlinear systems step 2.5 and Eq. (5.4.17) would give different values of ii and the latter value is preferable because it satisfies equilibrium at time i+ 1. “This procedure with a constant time step Ar can lead to unacceptably inaccurate resulls. Significant errors arise for two reasons: (1) the tangent stiffness was used instead of the secant stiffness, and (2) use of a constant time step delays detection of the transitions in the force-deformation celationship. First, we consider the second source of eror,ilustrated by the force-defoemation relation of Fig. 57.22, Suppose that the displacement at time i, the beginning of atime step, is uy and the velocity si; is positive (.e., the displacement is increasing); this is shown by point «. Application of the previously described numerical procedure forthe time step results in displacement uj. and velocity 41 at time i + 1; this is shown by point b. If ij, is negative, then at some point &/ during the time step, the velocity became zero, changed sign, and the displacement started decreasing. In the numerical procedure, if we do not bother to locate b, continue with the computations by starting 176 Numerical Evaluation of Dynamic Response Chap. 5 fe Numerical Exact @ © Figure 812 the next time step at point b, and use the tangent stiffness associated with the unloading ‘branch ofthe force-deformation diagram, this procedure locates the point c at the end of the next time step with displacement u,,2 and negative velocity. On the other hand, if the time instant associated with 6’—when the velocity actually became zero—could be

You might also like