EARTHQUAKE ENGINEERING RESEARCH CENTER SHAKE @ COMPUTER PROGRAM FOR EARTHQUAKE RESPONSE ANALYSIS OF HORIZONTALLY LAYERED SITES by Per B. Schnabel John Lysmer H. Bolten Seed @ Lomputer program aistributed by NISEE / Computer Applications Report No. EERC 72 - 12 December 1972 College of Engineering University of California Berkeley, Californiaa TABLE OF CONTENTS Intreduetion Theory 2.1 Propagation of harnonte shear waves in a one~ digensional systen. 2.2 Difference between rock outcrop and base rock motions, 2.3. Transient sotion: Description of Progmn SHAKE System and Operation Documentation 4.1 Computer equipment. 4.2 Storage requireseat 4.3 Runtime. 5.1 Organization of input data. 5,2 Initialization card. 5.3 Run option card. 5.4 Data cards and explanatory notes for the various options. Example Run 6.1 Selection of soil system and input notion. 6.2 Input data for che analys 6.3. Computer output from the snalys! “s Page 10 2 2 a3 B 1 as 6 38 36 38 43age BS Program Identification and Abstract oe ToL Program Identification. 5 7.2 abstract, ss 8. Seurce Listing for Program SHAKE? 38 Acknowledgements 8 References 81. dutRopucrroN Several nethods for evaluating the effect of local #o{l conditions on ground response during earthquakes are presently available. Most of the methods are based on the assumption that the main responses in @ soil tion of shear waves from the depostt are caused by the upward prop underlying rock formation, Analytical procedures based on this concept in~ corporating nonlinear so{l behavior, have been shown to give reculte in good agreanent vith field observations in # number of cases. Accordingly they are finding increasing use in earthquake engineering for predicting responses within so{l deposite and the characteriatice of groumé surface ottons. ‘The analytical procedure generally involves the folloving ste Deteraine the characteristics of the motions Likely to develop ia tthe rock fornatton underlying the afta, and eelect an accelero- gram with these characteristics for use in the analysis. The saximm acceleration, predominant period, and effective duration are the most important paraneters of an earthquake motion. Empirical relationships between these parameters end the d{etance from the causative fault to the site have bees established for different magnitude earthquakes (Gutesberg and Richter, 1956, Seed et aly 1969, Schnabel and Seed, 1972). 4 deaign notion vith the desired characteristics can be selected from the strong motion accelerograns that have been recorded during previous earthquakes (Seed and Larisa, 1959) or from + 1964)» artifictally generated accelerograns (Housner and JensinDetermine the dynanic properties of the soil deposit. Average relationships between the dynanic shear moduli and dacping ratios of sotis, as functions of shear strain and static properties, have been established for various soil types (Hardin and Drnevich, 1970, Seed and Idrise, 1970). Thus a relatively staple testing program to obtain the static properties for use tn these relationships vill often serve to establish the dynamic properties with a sufficient degree of accursey. However more elaborate dynanic teating procedures are required for special problems and for cases involving soil types for which empirical relationships with static properties have not been established. 3. Compute the response of the soil deposit to the base-rock motions. ‘A one-dimensional nethod of analysis can be used if the soil structure is essentially horizontal. Prograns developed for performing this analysis are in general based on either the solution to the wave equation (Kanai, 1951; Matthiesen et ale, 1964; Roesset and Whitean, 1969; Lysmer et al., 1971) or on a lumped mass simiation (Idrise and Seed, 1968). More irreguler soll deposite may require a finite element analysis, ta the following sections the theory and use of a computer progran based on the one-dimensional wave propagation method are described. The progres can compute the responses for a design motion given anywhere in the tex, ‘Thus accelerograns obtained from instrunents on soil deposits can be used to generate new rock mations which, in tur, can be used as design 71). The motion for other soll deposits, Fig. 1 (Schnabel t al progran also incorporates nonlinear oil behavior, the effect of the elas tleity of che base rock snd systems with variable damping.2. aBORY ‘The theory considers the responses associated with vertical propagation of shear waves through the Linear viscoelastic systen hom in Fig. 2. The ayaten consists of N horizontal layers which extend to infinity fo the horizontal direction and hae a halfspace as the bottom layer. Each layer 4s homogeneous and teotropic and ia characterized by the thickness, hy mass density, p, shear modulus, G, and damping factor, B- 2.1 Propagation of harponie shear vaves in s one-dinenstonal systen. hows in Fig. 2 Vertical propagation of ehear waves through the 9; vill cause only horizontal displacenente: ws uae) @ which must satiofy the wave equation: atu, ge, ae eats +n @ ae 8 at" Bator Harmonie displacenents with frequency w can be written in the form: wGe) = UG) + et ° Substituting Eq. 3 into Eq. 2 results in an ordinery differential equation: (6 + su SE = otv @ which has che general solution uC = Beth 4 rete o ia which @ 2 co}aa 40 $193433 SNOILOW GNNOYS NO SNOILIGNOD OS 1¥907 ONILN4WOD HOS 3HNGIIONd JO NOUVINISTGTY TVWSHOS 16 vonow yoy 2508, ' v wouow, unos papiosay» Layer Coordinote Propagation Properties No System Direction Sy Bu Pa Yew ee Incident wove Fig 2 ONE-DIMENSIONAL SYSTEMwhere k is the complex wave nusber and G* 1s the coupiex shear modulus. ‘The critical damping ratio, 6, is related to the viscosity n by an = 208 Experinents on many soil materials indicate that G and 8 are nearly constant over the frequency range which 1¢ of main interest in the analysis. Tt is therofore convenient to express the complex shear sodulus in terms of the eritteal dasping ratio instead of the viscosity: oF = G+ fun = c424a) oy hare G# can be assuned to be independent of frequency. Equations 3 and 5 give the solution to the vave equation for harmonic notion of frequency ulayt) = Bet COtet) 4. pert (lmue) @ whore the first term represents the incident wave travelling in the nega- tive x-direction (upwards) and the second term represents the reflected vave travelling in the positive x-direction (downwards) Equation 8 6 valid for each of the layers tn Fig. 2, Introducing & Local coordinate system X for each layer, the displacenente at the top and bottom of layer = are: ut 0) = E+ Rett o sthahe agen = Gyr ettatag se e840) The shear strees on a horizontal plane ts: tan so Hen Eber aor by Bq. 8 Gxt) = tho8 (EM ~ Fe @ az and the shear atresses at che top and botton of layer = ar gio) = ak,ok (2, -Fee an tk, th, TgOehye tok le BE re Pelt an Stresses and dlsplacenents must be continuous at ali interfaces. Hence, by Bq. 9, 10, 13 and 14: Baa as) Enea a6 Subtracefon and addition of Eqs. 15 and 16 yield the folloving recursion formulas for the euplitudes, E.,, ad F,,, of the incident and reflected wave in layer tl, expressed in teras of the amplitudes in Layer an te, 4 aa, 3 a Py, * ByGeage * ede arae as) woere a, 18 the complex tapedence ratio x, pgs? a, = Bea + GE as Fane Peta men which again is independent of frequency.At the free surface, che shear stresses aust be tero. In aldition, ng. 12 with 1 and X; equal to sero gives Ey > Fy —-L.esy the asplitudes nf the incident and reflected waves are always equal at a free surface, Beginning with the surface Layer, repeated use of the recursion formlas qa. 17 and 18 leeds to the following relationships between the suplitudes tm Layer mand those in the surface Laye: By eg) 2, (20) Fat fyCo) By an the transfer functions ¢, and f, ate staply the amplitudes for the case 5, =F, + 1, and can be determined by substituting this eondttion inte the above recurafon formulas. Other transfer functions are easily obtained from the e, aad f, functions. The transfer function A, . between the displacements at level and m is defined by Aa 7 Gly snd by aubsticuting Eqs. 9, 20 and 21: eg(o) + £,(0) 2 FEO a Based on these equations the transfer function A(u) can be found between any two layers in the system. Hence, if che motion 1s known in any one layer tn the system, the motion can be computed fa any other layer. ‘The anplitudes, E and P can thu be computed for all layers in the aysten, and the strains and accelerations can be derived from the displace nent function, Accelerations are expres:ease) = BBs = atceat Cg neTAO)) an and strains by ye Bee scent ton) _ pertttore), ay 2.2 Ratio between rock outerop motions and base rock motions. If the amplitudes of tho incident and reflected wave couponerts, Ey and Fy, in the elastic halfspace, Fig. 38, are knom, the notions in the halfspace with the sof] system removed, Fig. 3c, are easily computed, The shear stresses are zero at any free surface; thus Fy = Fy. and the incident wave 1s coupletely reflected with a resulting amplitude 2y at the free surface of the halfepace. The amplitude of the incident wave in the halfspace 1s independent of the properties of the systen above it since the reflected wave {6 completely absorbed in the halfspace and doxs not contribute to the incident wave. The incident vave component, Ey, 1s therefore aqua! in all systens shown in Fig. 3. ‘The vatlo between the base motion, uy» and the motfon, wis ot che free surface may be computed from the transfer function: uy eye) + fy aod + he aaa 2s) ‘The transfer function between the sotion at the surface of the deposity ty, and the motion at the free surface of the halfspace is: 8) : 1 §2@ + Rw IE the halfepace 1s the rock formation underlying a soil deposit, Eq. 25 shows the ratio between the motion in the base rock snd in the out~ cropping rock. The ratio between the amplitudes of the base rock sotionTa SUBAVT ONIddOYDLNO HLIM WALSAS TWNOISN3WIG-3NO ¢ O14 Bujddossino a90dsy10H Bujddosjno wi 49K04 parkiouo waishs 2 @ © a(ttg sia) =n ty!‘end the outcropping rock motion de alvays Jess than 1, with einioum values ‘at the resonance frequencies of the deposit. Transfer functions for the deposit veed in the example, (Sect. 6), are shown in Fig, 4, The asplitude of the base rock motion is only 65% of the amplitude of the rock outcrop motion at the fundanental frequency of the deposit. This difference 10 a fonction of the tnpedance ratio between the deposit and the rock end of the damping in the deposit. ‘The differences ta the couputed responses resulting from the use of a rigid base, relative to the use of an elastic base, depend also on which frequencies fare dominant in the rock motion, Rock motions with frequency deminance near che resonant frequencies of the deposit will be considerably more affected than notions with frequency dominance between the resonance frequencies, see Fig. 4. ‘The effect of the elasticity of the base rock is, therefore, not only « function of the impedance ratio between deposit and rock and of the damping in the aeposit, but also of the frequency distribution of the energy tn the rock notion relative to the resonance frequencies of the deposit. An approxtnation for the free eurface motion for one of the layers fn che system, Fig. 3b, may be obtained in the sane way as for the halfepace, provided the incident wave cosponest in the outcropping layer ad in the Layer within the system are equal--t.e. B+ BL. This te approxinately the case when the properties of Layer m and all layers below are equal ia the two systens and when the impedance, 9, V,, is of the sane order of sagni- tude a6 for the halfspace, This te the case for example, in sedimentary rock layers overlying a crystalline rock base. Yor @ more arcuate solution, ‘the motion in outcropping layers must be computed in a separate system from the motion in the halfepacSNOILONNS H34SNVHL & 81d 985/, Kouanbasy s » £ z T T T T ————————— N'\y ggpjins [10S Oy HOOT 8s0E voossidwy2.3 Transient potions ce barnonte ‘The expressions developed above are valid for steady 2! notions. The theory can be extended to transient motions through the use of Fourier transformation. A digitized seisnogran with a equidistant acceleration values, i, (Jtt), sync], can be represented by @ finite eum of harmonic aption where uj, 9*0,...+s0/2 are the equidistant frequenct re a) land b, designates the complex Fourfer cocffictents: et So a2 E awe 2 gee” ow io io and each term in Eq. 27 49 a hernonic motion oseillating with frequency If the series in Eq. 27 represent the motion in a layer a, a nev series representing the notion in any other layer n, 1s obtained by ch tere in applying the appropriate amplification factor from Eq. 22 to the series ‘the representation of a discrete motion vith ite Fourfer transform gives an exact representation of the motion at the discrete points t= J-tty nol. Cyelte repetition of the motion with the period T= arte 32015 dmplied in the solution, The solution applies, therefore, to an infinite train of identical accelerograns rather than the given single accelerograa. For systems with damping this is not of any significant consequence since che individual accelerograne can be separated by a quiet zone of zeros causing che responses from one eycle to damp out before the beginning of the next cyele. the Fourier Transformation can be performed in several vays. The SHAKE prograa utilizes the Fast Fourler Transform algorithm developed ty Cooley and Tukey (1965), which is faster by a factor n/logn over the corventional pethod. This technique computes all values in the series similtencously. ‘The vethod requires that the number of terns in the series be sose power of 2, A typical analysis using an acceleration record of 800 terms with tine- step O€ © .02 sec. will use 1024 values 1a the Fast Fourier Transform, with all values between 400 and 1024 eet equal to 0. This will satiefy both che requirenents of a quiet zone after the acceleration record and that the total nunber of tems must be a pover of tuo. \CRIPTION OF PROGRAM SHAKE Program SHAKE computes the responses in a system of homogeneous, visco- elastic layers of infinite horizontal extent subjected to vertically sravelling shear waves, The system {e shown in Fig. 2. The program is based fon the continuous eolution to the wave-egquation (Kanai, 1951) adapted for ise with transient motions through the Fast Fourfer Transform algoritha (Cooley and Tukey, 1965). The nonlinearity of the shear modulus and damping 4s accounted for by the use of equivalent Linear soil properties (Iérise and Sead, 1968, Seed and Idrtes, 1970) using an iterative procedure :o obtain values for sodulus and damping compatible with the effective strains in each layernn ‘The folloving assumptions are Implied tn the analysis: 1, The sotl system extends infinitely in che horizontal éizectton. 2, Bach layer dn the systen te completoly defined by its velue of shear modulus, eritical damping ratio, density, and thickness. ‘These values are independent of frequency. 3. The responses in the system are caused by the upvard propagation of shear waves from the underlying rock formation. 4, The shear vaves are given as acceleration values of equally spaced tine intervals. Cyclic repetition of the acceleration tine history ds implied ia the solution. 5. The strain dependence of modulus and damping is accounted for by an equivalent Linear procedure based on an average effective strain level cosputed for each Layer. ‘The progran ie able to handle systems with variation tn both moduld and damping and takes into account the effect of the elastic base. The motion used as a basis for the analysie, the object motion, cen be given in fany one layer in the system and new motions can be computed in any other layer. ‘the following set of operations can be performed by the program: Read the input motion, find the maxtnun acceleration, scale the values up or dovn, and compute the predominant period 2. Read date for the aotl depoett and compute the fundamental period of che deposit. Conpute the maximum stresses and strains in the middle of each sub- Layer end obtain new values for modulus and damping compatible with a apecified percentage of the marinus strain.A. Compute new motions at the top of aay sublayer inside the system or outcropping from the system. Print, plot and punch the motfons developed at the top of any aublayer. 6. Plot Fourier Spectes for the motions. 7. compute, print and plot response spectra for notions. 8, Compute print and plot the amplification function between any two sublayere 4. Increase or decrease the time interval without changing the pre~ dominant period of duration of the record. 10. Set a computed aotion as a new object motion, Change the acceler: tion level and predominant pertod of the object motton. 11. Conpute, print and plot the stress or strain tine-histery in the niddle of any sublayer. hese operations are performed by exerctatag the various avatlatle options da che progeas. A List of these options 1s given in Section 5, Required 4, SYSTEM AND OPERATION pocUMENTATION 4.1 Computer equipsent ‘The program has been developed on a CDC 6400 computer using FORTRAN LV Language. the CDC 6400 hes a 131 k core memory snd uses 60 bit words. The progeem has been run without modifications on CDC 6600, 7600 and UNIVAC 1108 computers, and with afnor modifications on ISH 360 and 370 computersa 4.2 Storage requirements ‘The program requires approximately 50,000 octal words of storage excluding the blank common X, The additional storage is a function of the maximum nunber of terns used ia the Fourier Transforn as shown tn Table 1. Table 1. Storage Requirements Munber of | Langth of | Field length cerns, array X ‘octal ° ° 50,000 512 3220 37,000 1024 6420 65,000 2068 12820 102,000 4096 25620 134,000 aisz 51220 220,000 4.3) Runeine ‘The runtime te a function of the munber of term, m, used in the Fourter Transformation and of the susber of sublayers in the deposit. The etme tavolved in the Fast Fourter Transformation is proportional to slogan; all other operations are approximately proportional ton. In the computation of strain compatible sofl properties, the tine vill also increase in proportion to the suaber of sublayers. For the example run, Sect. 6, the approximate run times on the CDC 6400 fare shown tn Table 2.16 Table 2. Bunts umber of Tine Run tine ters | interval, sec. | sec. siz 06 4s 028 02 80 2068 son 70 REQUIRED INPUT paTA 5. Organization of input data. wing 4s a description of the operations performed by the aifferent options, the required format for the taput data, and explanations of some of the input paraneters. The input format is also described at the beginning of the main program, see Heting sect. 8.3. ‘The various options can be executed and repeated in any logical sequence. ‘The operations in an option will be performed on the data given or computed in the progeaa when the option fe called, and cha date may ba changed at any. cine during the execution by repeating the option vith ney data. For exaaple, in order to compute new mations in a soll deposit, (Option 5) object sotton (Option 1), soll profile data (Option 2), specification of location of object motion (Option 3), dynamic soil property-strain relation (option 8), aad strain iterations (Option 4—if etrain compatible properties are desired), ust precede Option 5. Soll responses for a new (additional) sell deposit may be obtained by repeating Options 2, 3, 4, and 5, The last-read gol deposit may be subjected to a new earthquake by repeating Optiors 1, 4, and 5.5.2 Ini: rr 1s lon card (15,F10.0) Cole. 1-5 MAMAK Maximum number of terms to be used in the Fourier Traneformation in any of the problens to be run. Must be a pover of 2 such as 512, 1024, 2088, ete 6-15 SKO Coefficient of earth pressure at rest for sand layers. If blank the value is set equal to 0.45. May be left blank f all layers are clay. After the initfalization card follows one run option card. 5.3 Rup option card (15) cons. ° 1 2 3 ° 10 a a 3B Mu a5 16 “5 KK Run of option stop, ne more data Yead input motion, and set as object notion read soil profile data assign the object motion to a specified sublayer iterate to obtain atrain—compatible sotl properties Compute new notions at the top of specified Gublayers, print maximum accelerations and punch acceleration tine history print or punch acceleration tine history of Object motion or any specified couputed notion Spatfy object motion or set the motion in any specified sublayer as new object motion fread relations between éynanic soil properties and strain compute response spectra for ary specified notion Increase time interval in sottens decrease tine interval in sotiens plot Fourier Spectrum of object notion Compute and plot Fourier Spectrum of motion in any specified sublayer plot acceleration tine history of object motion or sny specified computed motion compute and plot amplification function between any two specified subleyers compute and plot stress or atrein history in the middie of any specified sutlayer. After the run option card follows the data set for the selected option16 $4 Dats cards and explanatory notes for the various options Option 1. Bead Input Moeton @ @ @ ® 6) o Let tnd Operations performed, Acceleration values are re from card: ‘The sequence of the carda 1s checked ‘The maxisun acceleration value in the record is found. The acceleration values may be scaled either by a specified factor or to a specified maximum acceleration Tratling zeros are added to, the record to obtain sufitetent Jength on the quiet zone () and a total number of values veh are a power of 2 The higher frequencies in the record are renoved and the naxinum sccelevation in the modified record 1s found-optional, ‘The notion is set as the new object motion. Data Carde Card (215,710.0,546) cols. 1-5 Nv Mumber of acceleration values to be read from cards 6-10 a) —Nunber of values to be used in Fourier transform Must be a poxer of 2. 11-20 ve? rime interval between acceleration values (see) 22-50 TETLE(L) [dent fteation for earthquake, card (3710.0) cols, 1-10 xF Muleiplication factor for acceleration values. Used only if WAX is 0, left blank otherwise, 11-20 MAK Maximum acceleration value to be used. ‘The acceleration values in the record will be sealed to give maxinus acceleration = WUX, unless XF ts lefe blank. 21-90 maK() Maxtoun frequency to be used in the calculations. Acceleration amplitudes fat all frequencies greater than FMAX are Set equal to 0. ded and consecutive card, Acceleration record. (SF9.6,17) Cole. 1-72 X(Z) 8 acceleration values. (e's) T19 K Card nusber, Warning vill be given for catda not in sequence.oy © uv Explanatory notes for Option 1. he acceleration values between NV ond MA are eet equal to 0. in the program. Cyclic repetition of the motion ts taplied in the Fourier transfors and a quiet zone of 0."s or low values are necessary to avoid interference between the cycles. For sost problens a quiet zone of 2-4 seconds is adequate vith longer tine required for profiles deeper than shout 250 ft and/or damping values less than about 5 percert the predominant period of the earthquake record can be changed by attoring the tioe interval At from that originally assigned to the acceleration record. If the original record has time interval Oe, and corresponding predominant period Ty, @ nev predominant period 7, ts obtained by changing the time interval to Feaquenctes above 10-15 c/see carry a relatively anall asomt of the energy in earthquake sotione, and the amplitudes of these frequencies can often te set equal to 0 without causing any signiftcent change £5 ‘the responses within a sofl system. Table 3 shows the maximum accelers- clons and etraine in the soil system used in the example run, sect- 6) computed for the Pasadena motion with tine interval 0.02 sec and @ maximus frequency of 25 c/ase. Results are also shown for the sane notion with all amplitudes above 5 e/sec set equal to 0, The difference in maximum accelerations was lese than 6.5% and in maxtmun strains less than 0.7% im the two ‘The diéference in response spectral values wan less than 1% for periods sbeve 0.2 sec and less than 10% for periods from .0 to 0.2 eee.a9 sto" acto" ort « sree veo ot seco" sez" oot e aco" 610" 9 6620 srco" os u 050° 5050" z 69y0 ovo" z Q 68600" 78600" z 1050 050" w - waco" taza" 9 950 sso o e 06c0" 60" Y 6650) 0030" oz v sect’ zee" « 660° 8560" “ v 2100" szeo0* 6 2360) 160" o 2 peeyo ¢ | v08/9 ez = z pee] gz = 735 sounsesang [~—F Fae powara33a Trepese mapa] EAT oun wo soyouanbaag 304 seupeaag pus suoyaexoTo22y snore, Yo 130538 ve tae1» tn the computation of responses in deep soft eysteas fron « motion given near the surface of the deposit, errors in the higher frequencies will be amplified and may couse erroneous results. To avoid this source of error, ‘the auplitudes of all frequencies above 10-20 c/sec. may be set equal to 0., since these frequencies generally are of Little interest and do act affect the response. Several runs should be performed with different amounts of the higher frequencies renoved £0 investigate the effect on the response and to ensure a stable solution. Renoval of the higher frequencies in a notion has « smothening effect on the acceleration tine Matory as show in Fig. 5 for @ segment of the Pasadena motion, In this case the maximum acceleration for the aodified and original motion were approxinately equal, but the maximum accelerations may deere! Je or increase with the removal of the higher frequencies depending on tthe shape of the acceleration curve near the maximum value. Option 2. Read Data for Soil Deposit. erations performed (2) The properties of the soll deposit are read fron cards. (2) The sequance of the layer cards is checked. (2) The Layers are eubdivided into sublayers-—optional. (4) Effective pressures in the middle of each eublayer are computed. (5) The fundamental period of the deposit ts computed. Dat Let Card (315,646) Cole, 1-5 MSOIL Sof deposit number. Can be left blank. (8) usher of Layer carde to be read including card for halfepace, There is one card for each layer whose. properties are individually specified. © 6-20 1198 AMOUSIH ~ SAIL NolLvy31390V 3HL NO SHIONSNDZYA YBHOIH BHL 30 103413 S Sls spuoses us aul os se | Pees Presacols, 11-15 37-51 TDNT(2) 2nd and consecutive cards. hhalfepace. (315,6F10.0,85. cole, 1-5 10 u-15 16-25 26-35, 36-45 46-55, 56-65 66-75 16-80 K rere max a) xo 2 yst@? actor’ BEAC For the elastic half 6 ») cay «© 20 o, Number of f{ret subserged eublayer Identifseation for soil profile. one ecard for each layer including 0) Layer nusber, The layer cards most be in sequence vith the surface layer ae layer 2. Nove that the number pf layers aay be S the umber of subleyereS, Sotl type > clay 22 sand 3 > rock Number of aublayers in layer K. The KER layer Will be divided {nto NLN sublayers of ehlekness = HL/KLN. Layer thickness (ft.) Initial estimate of shear aodulus (ktps/eq.ft.) Not necessary if VS is given. Initial estinate of critical dasping ratio (@ecinal) « Uuntt wetghe (kips/eu. fe). initial estimate of shear vave velocity Ge/ece). Not necessary Sf GHOD is given. Factor for shear sodulus ‘Clay - F, = undrained shear strength (etps/sq. £0.) factor sodifying the average carve read in wider Option 8. Set F, = 1. for no change. Rock - Fy = Shear"wave velocity for low strain Valuesin thourande of ft./eee. Sand - Factor modifying the standard danping curve read im under Option &, For exaaple, a fector of 1.2 Increases each and every value by 20 percent. Jo, oil Layer card number ML, it is suffictent to give values for K, GMop or VS and W. suaximun total nunber of sublayers iseluding the base 1s 20.@ Explanatory notes for Option 2. With ths wave propagation method che responses can be computed in a honogencous layer of any thickness. A soll deposit will, hovever, have varying properttes not only due to the variation in the soil itself but also due to the differences in the strain-level induced during shaking. Since the soil deposit must be represented by a set of homogeneous layers, cach with a constant value of modulus and damping, the thickness of each layer must be Limited based on the variation in the soil properties. for a fairly uniform deposit, a sublayer ehickness snereasing from about 5' at the surface to 50-200" below 100" depth should give sufficient accuracy. Accuracy may be checked by making « trial run and comparing resulte with a subsequent run where more layers and/or sublayers are used. The divieion of a layer into aublayers ts for conventence t avoid punching of several cards with the sane properties, and all sublayers are treated a2 separate layers in the following computations, The sublayers are nusbered consecutively starting at the top of the soit deposit and the halfapace 4a counted as the last Layer and the Last sub- Layer in the deposit Computations of shear moduli for the different soil types are based on the following expressions: clay 6, = K+ Fe sand 6, =k, + 1000+ (on? . 2, eis Br Fock G, = Ky +p * (1000- ¥,)7/2000, R@ 2 where K © strain fimetion given in Option 8. factor given a9 dnput (FACTOR) p = masa density in kips/eu. ft. oh, = mean effective pressure (psf). ‘The strain function for clays, Kj, gives the average relationship between G/S, and strain for saturated clays. Whtle the wndrained shear acrength of the clay, S,, t# normally used in this sedulus-strain relation, the factor for clay, Fj, should be given a value vhich gives the correct modulus: rain relation; thus F, 1s not necessar{ly equal to Sy, Ue the modulus of the clay 18 found from seisntc Savesttgations, the value of F, should be set to G,/k, where K, is the value for 207° percent strain in the curve given in Option 8. ‘the modulus snd dazping are dn general used as inieial valuss on the First iteration for ene computation of strain mmpacible properties, but they can also be used directly to compute the responses for the values given, by omitting Option 4, Typical values of the modulus for strong shaking are of the order of 500 kips/sq. ft- near the surface increasing to 3000 kips/eq. ft. at 100-200" depth for sand, 500- 2000 kipe/aq. ft. for clay with values as low as 50-100 kipe/sq. ft» for soft clay. Usuelly 3-5 iterations are sufficient to obteis strain compatible values within a 5-10% error Linte. ‘The results ore not highly sensitive to errors in the damping ratio and values selected between 0.05 to 0.15 vill usually give strain coupatible values with 2 to 3 {terations.2 Option 3. Agate Object Motion to a Specified Sub: ‘The object motion 4s asetgned to the top of one sublayer in the soll depoate. Daca Cards ist card (215) Cole. 1-5 IN Nugber of sublayer where object motion 1s S-10 INT ype of oublayer ‘0 ~ Outcropping©®) oubtayer 1 = sublayer within profile Explanatory notes to Option 3. *) sea Section 2.2. Option 4, Obtain Strain Compatible Sofl Properties. operations perforsed (a) Paraneters for the iceractons are read from cerd. (2) Yaxtmm atraine, stresses and times for the maxtea are comput in the middle of each sublayer. (3) Effective strains are obtained from the maximn strains and used to compute new soil properties. (d) the operation ta repeated until strain-compatible sotl properties are obtained within a given error limit or until » specified ‘aximun nusber of iterations 1s reached. (5) The fundanental period of the deposit is computed after the final (5) A get of soft data cards with the new strain compatible properties 4s punched--opt ional.2 Date Carde tet Card (215,2710.0) @ () © cols, 1-5 x5? Set equal to 1 for punched set of sofl daca cards with the soil properties after final Heeration. Leave blank if punched cards are not wanted. 6-10 rmx? 11-20 enn) Maxtsum acceptable difference between the Last-used modulus and dasping values and the atrain-coupatible values (percent). Ratio between effective strain and maximum strain (dectzal)- Maxinus number of iterations. 2i-30 pmo) Explanatory notes for Option 4 ‘The most tine consuming part of the computations is to obtain strain compatible soil properties. A set of eoil data cards vith strain~ compatible properti ‘aay save computer or punching tine tf additional computer rune are to be nade subsequently. ‘The iterations stop vhen the specified maximum nusber of iterations (MUX to reached or when the difference between the sodulus and denping used and the atrain-compatipie modulus and damping values ie less than the acceptable difference (ERR). Usually 3-5 itere- tlona are sufficient to obtain an error of lese than 5-102. The values given as "new values" in the final dteration are used in all computations folloving Option 4, and the actual error is lees then the error values given in the final iteratton. ‘The effective strain 42 used to compute nev soil properties. The ratlo between the effective and the maxtnun strain has bees enpiriecally found to be between 0.5 and 0.7. The responses, hovever, are not highly sensitive to this value and an estimate betveen 0.55 to 0.65 is usually adequate, with che higher value appropriate for giving more uniform strain histories.25 eon a @ o “ Compute Notion in Specified Operations performed The acceleration tine history 1s computed at the top of specified the saxinua acceleration and tines for maxina are printed for the computed motion ‘The computed acceleration tine histories aay be punched—~optional. 2 seceleration ting Matortes say alao be printed or plotted (Option 6, 7 and 14) (8). Date cards lst Gard (1515) le. 1-75 USC) Array showing the numbers of the sublayers at. the top of which the motion 4s to be computed. Maximum of 15 locations. 2nd Care (1515) aed «@) oe Gola, 1-75 LIS(I) Array apectfying types of above sublaye (0 = outcropping (©) sublayer 1 - sublayer within profile cara (1515) cots, 1-75 175(1)(®) Acray with aode of output for the computed notions: 0 max. acceleration value only printed. 41> punched cards giving acceleration tine Ristory in addition to the printed maximum acceleration valu Explanatory notes for Opeton 5 ‘The acceleration tine histories can be printed or plotted through the use of Option 7 where a specified motion 1s set as the new object motion. Subsequent use of Options 6 and 14 give respectively 4a printed and a plotted output of the acceleration tim: history of the motion. See section 2.2.6 option 6. Print or Punch Object Motion. operations performed (2) Maximus acceleration and tine at hich maxtmun oceurs are found. (2) the object motion 4s printed-—optional. @) The object motion 4s punched on cards--optional. ee Card (25) Gol. 5 K2 Selects mode of output. K2=0 Max, ace. only 1 Punched output 2° Printed and punched output. Option 7. change Object Hotton. operations perforsed (2) A-motion at the top of a specified cublayer can be set as the nev Object motion and printed or punched (Option 6) or plotzed (Option oF used for subsequent computations—oprional (2) The tine etep in the object motion can be changed--optional. (2) The acceleration Jevel in the object sation ean he chanzed: optional. 4 Cards Let Card (215,2710.0) con 1-5 UL Musber of sublayer. Use 0 $f object poston originally assigned 1s to be retained(®), 6-10 LIL type of above sublayer D'outeropping (©) aublayer 1 = sublayer within profile 11-20 xP Multiplication factor for acceleration values--1, for no change. 21-30 pmvEW New timestep” Explanatory notes for Option 7 (a) The acceleration level and tinastep can be changed ether on the notion originally set as the object motion, or on the computeda notion which {s set #5 the nev object sotion through Option 7, (b) A change in tine interval will change the predosinent period of the sotton. If the tine interval snd predominant period of the ortginal motion are dt, and T, respectively, a new predosinant a period T, 1s obtained by changing the tine interval to 2 (e) See section 2.2. Strain ‘operations performed (2) gfective atrain values with corresponding valu and noduli ere read from cards for damping (2) Parameters are computed for interpolation of sodulus and damping values using a Linear senilogarithaic relation between the given values. (3) The relatsonshép between the dynanic properties and the strain ts Daca Cards Let card (315,710.0,20A5) Cols, 1-5 NSOILT Muaber of different soil or rock tyres to be read, Maximum 4.(@) ow) 10 wee) ser equal to 1 for plot of curves. arts wm) umber of stratan unit to be plotted. 16-25 sc Maximum value of the ordinate ia the plotting. 26-60 Utle or identi#testion data. Next follows two sete of carda for each soll or rock type. The first ues in each logerithatc 1 modulus parameters (c) and set gives the relationship between the st the effective straings the second set give the relation between the critical danping ratios end the effective strains, Typical data to shown on page 40.Firet Sett Ase Card (15,75.0,1146) cols. 1-5 NV(L) _Nunber of strain values to be read. Maximus 20. 6-10 FeL(i) mutttpiteation factor for shear-sodylye parameter. Used for plotting only. © 12-76 1D(L,1) Identification for firat date set. Used for plotting only. 2nd and consecutive cards (6710.0) cols. 1-60 X(L,1) Effective strain values in percent beginning vith the lovest value. @ values per card with maximum of 20 values, Consecutive cards (8F20.0) cols, 1-80 Y(L,1) Values of the shear modulus paranc corresponding to the strain valu above. Eight values per card with naximua of 20 valuee. - piven Second Set: The input format for the second set 1 identical to that for the first jet with values of eritical damping ratios in percent instead of the values for the sheer nodulus paraneter. Bel for option 8. ory nots (e) Three different soft or rock types can be used in the program as doserited in Option 2, The relationships between effective strains and the dynamic properties mst be read in the sane sequence as the sot] type using the notation: clay Send Rock (b) The values for the shear sodulue parameter and the damping can be plotted against che effective etrains, If plotting 4s epectfied (WPL = 1), values for the shear sodulus paraneter and damping are29 computed for a spectfied number of effective strains (NN) in each logerithate untt. The computed values should be sealed (FPL(L)) to obtain good representation of all curves on the sane plot. The scaled values and the corresponding effective strains are also printed. (c) The values are used to compute the shear modulus for the different soit types. The relattonshsp for sand and clay used in the program 4s based on the expressions given by Seed and Idriss (1970): 5.) ciy 0) +E a Sind KO) = apg EIT ‘The relationship used for rock 1e che scaled ratio between the shear modulus at lov effective strain (10~* percent) and the shear modulus at a apectfied effective strain: eck yy) = Sore 2000 gCy10"*) Compute Response Spectra Operations performed (2) The notion te computed at the top of a specified sublayer. (2) Tives for maxina ta the acceleration, velocity and displacement apactra are comuted and printed. (2) Acceleration and carde--optional. locity apectra may be plotted and/or punched on et Card (215) Cols, 1-5 LLL Sublayer number. Use 0 if the response spectre fare to be computed for the object mtion. 10 LTL_ Type of sublayer. 0 outeropping sublayer 1 = aublayer within profile. ‘The reoponse spectra are computed for the sotfon at the top of the eublayer, May se left blank ££ LL1 is 0.30 2nd card (515) Col. 5 RD Total number of damping values to be used. Maximum 6 values. 10 KP Set equal to 1 for punched output. 15 KAY Select plot and punch option: (= plot and/or punch velocity spectrum 1 = plot and/or punch acceleration spect-um 2 = plot and/or punch acceleration and velocity Spectrum 20 KPL Set equal to 1 for plot of spectra accorditg to RAV. ALL spectra computed since last plotting will be plotted together. 25 KPER Select periods to be used {a the computations: EFER = 0 9 steps from 0.1 sec to 1. sec 5 steps from 1. eee to 2. gee 4 steps from 2. sec £0 4. ace RPER 1 18 steps from 0.1 sec to 1. sec 40 steps from 1. gee to 2. sec 8 steps from 2. sec to 4. wee PER = 2 38 steps from 0.05 vec to 1. see 20 steps from. sec to 2. sec 50 steps from 2. sec to 5. see RPER = 3 Logarithafe increnente with 10 steps in each fog. unit from 0.1 to 5. KPER = 4 Logarithate inerenent with 25 steps in each Jog. unit from 0.05 to 10. 3rd card (6710.0) Cole. 1-60 7LD(I) Values of critical damping ratios ix decimal £0 be used in the spectral analysis. ND values must be given. Option 10. Increase the Tine Interval Operations performed ‘The time interval is increased,bata car: let card (15) ass 1m) accor for increasing time interval. Must be a power of 2. cou Explanatory ates for Option 10 (@) The Fourier Transformation of a given acceleration time history consists of a series of harsonic motions With the harmonic motions given, acceleration values can be computed for any value of the time, ©, and a new acceleration tins history can be generated with a time interval different from the original. Suppose, for example an acceleration record te given with 2048 values anda timestep At = 0.01 sec. Through Option 10 with LPR = 2 a new record with 1024 values and timestep 0.02 sec Je generated. The acceleration values in the two records are identical at all times n+ ,02 gec., n= 1,2..,1026, The new record has a maxiaun frequency of 25 ¢/sec. compared to 50 c/sec. in the original records, and frequencies from 25 c/sec. to 50 c/sec. are lost in the operation. Increasing the time interval reduces the computer tine as shova under sect. 4.3. For computation of maximum accelerations a tine Anterval of 0,02 sec. will generally give adequate accuracy while a tine interval of 0.04 sec. may be sufficient for the computation of the stresses and strains in a deposit.2 ‘The atfference in maxisun accelerations and strains resulting from the use of different tine intervals are shown in Tables 4 and 5 for the example rin, The effect may be sooeuhat higher for earthquakes with lover predominant periods and for atiffer sok eystons. Optdou 11. decrease the Tine Interval erations performed ‘The tine interval 1s decreased. Data Cards lee card (15) cot. 1-5 rH‘) Factor for decreasing the tine interval; must be a power of 2. Explanatory notes for Option 11. (a) See explanation co Option 10. Through Option 11 @ nev tine history is generated with the tine interval reduced by @ pover of 2. Compared with the usual Linear interpolation, thts method has the advantage of not dntroducing additional frequencies to the motioa. Option 12. Plot Fourier Spectrum of object Motion Qperstions performed (1) The Fourter Spectrum of the object motion is plotted. (2) The spectrum may be snoothe Let Card (315) cols. 5 Kl Select for plotting (O-= Store spectrum for later plotting. Max. of 2 epectra can be plotted together 1 ~ Plot ali spectra stored since last plotting. Nuaber of tines the spectrum is to be smoothed. ais Number of values to be plotted--naximun of 2049. 6-20 xsu(®» Table 4. Incerval on Computed Maximum Strain Z Depen _| ae = Ot ‘be = 02 be = 08 35 ore 00725 00725 13.5 029 -9 aa 25. 0392 039, 0390 36 0287 0287 0285 32 00982 00982 00981, n 0508, 0508 0505 90 0350 0349 0348 uo 4.0320 .0320 006Table 5. Effect of Tine Interval on Maxtmm Acceleration. —_ Waximun Acceleration ae = OL ae 02 ° 097 0972 0367 7 .0960 0958 0956 20 0598 0600 0590 30 0356 10553 0548 “2 0508 0508 0498 6 047, 40470 0462 ® 0317 ons 0318 100 0238 0238 0242 120 0181 0178 ove 33 ‘Explanatory notes to Optton 12. (2) The expression used to smooth the spectrum ts: where A, {8 the acceleration amplitude for the 48 frequency. Option 13, Plot Fourter Spectrun ©) of comuted Motions Operations performed (2) The motions at the tops of the spectfied sublayers are computed. (2) The Fourter Spectra for the computed notions are plotted and printed. (2) The spectrum may be snoothed--opttonal. ea cards Ast card (515) cole, 1-5 LLC) Sublayer nusber, io ia) ‘Type of sublayer: ‘0 = Outcropping®®) eublayer 1 - Sublayer vithis profile. 1s ay Select for plotting: (O'= Store spectrum for later plotting: wax, of 2 spectra can be plot together 1 = Plot all spectra stored since lest plotting. 16-20 uxsw(ay“*) Nuaber of times the spectrum 18 to be smoothed. 21-25 LLL) —_— Number of values to be plotted. Max. of 2049. 2nd Card (S15) As for Card 1 for a second notion. A blank card must be used {f only one spectrum 1s to be computed. Explanatory notes for Option 13 (a) See option 12. (b) See section 2.2. (©) See section 2.3.x e Time Opeton 14) ory of Object Motton' Qperations performed The tine history of the object motion is plotted. Data carde Let caré (215) cole. 1-5 NSKIP Number of values skipped in the plotting. 0 = every value is plotted L = every second value 4s plotted 6-10 Number of values to be plotted. Mex. of 2049. values. Explanatory notes to Option 14, (a) The tine biatory of a computed motion can be plotted ty setting this motion as the object sotion through Option 7. Option 15. Compute Anplifteation Spectrus Qperations performed (2) The amplification epectrus between any tvo gublayers in a given soil aysten 40 computed. (2) The maximun amplification and the corresponding perlod are printed. (G) The amplification spectrum may be plotted and printed--optional, Data Cards Let Card (515, F5.0,8H6) cole, 15 Lin”) © LIN Type of Firat subjgyer (0 ~ outcropping”? eublayer 1 - sublayer vithin profile 12-15 our? wusber of second sublaver. 20 LOT Type of second eublayer (0 outcropping sublayer 1 ~ aublayer vithin profile. Number of frst sublayer. @ 3 Select for plotting: ‘0 = Store spectrum for later plosting. Maximua of 8 spectra can be stored. 1 - Plot all spectra stored since Inst plotting.37 26-30 DFA. ywency steps. The amplification factor 1 computed for the first 200 frequencies with interval DPA c/sec. beginning at 0. 32-78 AMP(I) Identit: ‘ton. Explanatory notes to Option 15. (a) The anplifteation factors are computed from the firet sublayer to the second. (b) See section 2.2. Conpute Stress or Strain History in the Middle of specified Sublayare. operations performd (2) The etress and/or strain tine history in the middle of any two specified sublayers are computed. (2) The computed tine hetories may be plotted or punched on cards: Data Car Lee Card (SI5,F10.0,5A6) cols, 1-5 LLC) Sublayer number. The stress or strain history 4a computed on the middle of the sublayer. 10 1L05(1) Select type of respon 0 - strain 1 = aerest 1S LIPGH(1) Set equal to 1 for punched outpet. 20 LLPL(L) Set equal to 1 for plotting. 24-25 LNV(L) —Nusber of values to be plotted; maximum of 2049. 26-35 SK(1) Seale for plotting—{.e. aaxinun value of ordinate. If blank, the larges: value in the response ie ect aa the maxinum value of the ordinate. 37-65 D(1,) — Identifteation. 2nd Card. As for Card 1 for second eublayer. Use blank card if only one response 1s to be computed.38 6. Eons amt 6.1 Selection of soft aystes and input motion ‘4s exanple problen is chow in Fig. 6. Maximm accelerations, stresses ‘and etraine in the soll deposit and response spectra for the surface accelere- tlons are wanted for @ magnitude 7.4 earthquake occurirg 100 miles from the sire. Based on the relations given by Seed and Iériss (1970), the sofl systen The factors used fer clay are show on Fig. 7 vas selected for analys equal to the undrained shear atrength in kips/sq. ft. The factors for wand are estinated from relative densities and content of gravel. ‘The motion tn rock for a magnitude 7.4 earthquake 100 niles from the causative fault {e eotinated to have maximum acceleration of .02g and a predoninant period of 0.65 sec (Selnabel and Seed, 1972; Sead et al., 1969). ‘Anong the available strong notion records, the Pasadens record from the 1952 Kern Coty earthquake seeus to have characteristics most similar to those desired. The uaguitude of the earthquake was 7.7, the racord vas obtained cove 75 niles from the fault, che maximum acceleration was 0.0575 and the Jc. Modification of this record to give a maxiaum predominant pertod vas 0.65 acceleration 0.02g gives the desired characteriatice for the notion in an outcropping rock formation near the exanple site.38a Depth Soil Deposit Sond Op 45% Y Cloy Su +250 pst Cley Su 750 pst Y Cloy Sur lISOpst Sond with some gravel - Y clay sexzocopet tou 100 tiles Sit 50 — Earthquoke rognitude: 7.4 we Li Rock Vs #8000 ‘Yee Fig. 6 EXAMPLE PROBLEM38 Depth Soil type Factor o- 2 0.7 1 0.25 1 0.75 1 15 5o- 2 1.25 = ' 4 Wotton in outropping rock - ' 2.0 Posadenc record trom the 1952 Kern County foo— | ————————_earthquate scared 10 O.02q maximum cee _ \ 225 eration Halfspace Vs = 8000 "sec Fig. 7 SYSTEM USED IN THE ANALYSIS OF THE EXAMPLE PROBLEM9 6.2 Input data for the analysis,40aSe43 6.3 Computer output from the analysts.45 HOEEEILEGE HEUREESUREEEEEEE46a48 g Ptoog ae EERE A ae co gt ae a“930By —32356 7. PROGRAM IDENTIFICATION AND ABSTRACT 7.1 Program Identification 1. Program title: Vertical propagation of shear waves through horizontally layered soit/rock eysten. 2. Progres name: SHAKE. 2. Wetter Per B. Schnabel, Research Assistant John iysner, Associate Professor of Civil Engineering. 4, Organdzation: Geotechnical Engineering Department of Civil Eogineering University of California Berkeley, California 94720 Dete: December, 1972. Version: 2 Source language: FORTRAN IV 7.2, Abstract The progres computes the response in a horizontally Layered soil rock system subjected to transient, vertical travelling cheer vaves. The pethod is based on Kenai's solution to the wave equation and the Fast Fourfer ‘eansform algoritha. The motion used as basis for the analysis can be applied to any Le with elastic base and with 1s in the eystem, Syste veritable damping in each layer can be analyzed. Equivalent linear soil properties are used with an iterative procedure to obtain soil properties compatibie with the etraine developed in each Layer. A varied set of operations of interest in earthquake response analysis can be performed.35 8, SOURCE LISTING FOR PROGRAM SHAKE.36375839606«263646566Cl686970n2m8%516n8naLa83a aah aah3s86 aoea7 AGRIOMLEDGE ENTS ‘the development of the computer program SHAKE vas conducted as part " sponsored of a study of "Sof1 and Foundation Response During Earthquakes! by che National Setence Foundation. The authors are most grateful for this support and for valuable auggestions from Professor W, N. Houston tn seviewing the sanuscrtpt.88 References (4965) “An Algoriehn for the Mactine Calcula- Vol. 19, No. $0, Cooley, J. We and Tukey, J. Wi tion of Complex Fourter Series," Mathematics of Computation pp. 297-301 Gutenberg, B. and Richter, C. F. (1956) "Earthquake Magnitude, Intensity, Energy and Acceleration (Second Paper)" Bulletin of the Seismological Society of Anerica, Vol. 46, No. 2. Hardin, B, 0, and Drnevich, V. P. (1970) "Shear Modulus and Danying tn Sofie: 1, Measurenente and Paraneter Effects, 11. Deeign Equations and Curves,” Technical Reporee UKY 27-70-CE 2 and’3, College of Engineering, Univeraity of Kentucky, Lexington, Kentucky. Wousner, G. W. and Jennings, P. C. (1964) "ceneration of Artifictel Earth quakes,” Journal of Engineering Hechanice Division, ASCE, No. 90, February, pp. 113-150. Iériss, I. M. and Seed, H, B. (1968) "Seientc Response of Horizontal Sotl Layers," Jourmal of the Sof] Mechentes and Foundations Divisfon, ASCE, Vol. 94, Nov SM4, July, pp- 1003-1031, Kanal, K. (1951) "Relation Between the Nature of Surface Layer and che Anplitude of Bathquake Motions,” Bulletin Tokjo Earthquake Research Institute. Lysuer, J. Seed, H. B. and Schnabel, P. B. (1971) "Influence of Base-Rock Characteristics on Groumé Response,” Bulletin of the Seissological Society of Averica, Vol. 61, No. 5, October, pp. 1213-1232. Matthiesen, R. B., Duke, C. M.y Leede, D. J. and Fraser, J.C. (1964) "site Characteristics of Southern California Strong-Motion Earthquake Stations, Part Two," Report No, 64-15, Dept. of Engineering, University of Californ: Lo Angeles, August Roesset, J. M. and Whitman, R. V. (1969) “Theoretical Backgrout for Anplifi- Cation Studies," Research Report No. RS9-15, Soils Publication No. 231, Massachusetts inetitute of Technology, Canbridge. Sctmabel, P. B. and Seed, H. B. (1972) “Aecelerations tn Rock for Earthquakes ‘in the Western United States," Report No. BERC 72-2, University of California, Berkeley, July. Schnabel, P. B., Seed, H, B., Lysner, J. (1971) "Modification of Setsnograph Recorde Zor the Effect of Local Soli Conditions," Report No. EERC 71-8, University of California, Berkeley, Decenber. Seed, H. B. and Idrias, I. M, (1970) "Soil Modul! and Danping Factors for Dynatie Response Analysie," Report No. EERC 70-10, University cf California, Berkeley, Decenber. Seed, H. 8., Idriea, 1. M. and Kiefer, F. W. (1969) "Characteristics of Rock Motions During Earthquakes," Journal of the Soil Mechanies and Foundations Divieten, ASCE, Vol. 95, No. SMS, Septenber.SUPPLEMENT TO COMPUTER PROGRAM SHAKE* by T. Udaka and J. Lysmer Septenber 1973 Suggested corrections are shown franed on the attached segments of subroutines SHARIT, EARTHQ, CXSOIL, MOTION, STRE, UTPR, RESP, STRAIN, REDUCE, FFT, RFFT and RFSN of program SHAKE. ‘The purpose of these changes are: 1, To decrease the execution tine by up to 50% depending on the type of problem to be solved. 2, To redefine the complex odulus fron G* = G(1+248) zo ch = cai 2p?4i 2-8). (Tis change only influences subroutine CXSOTL.) Input and output formats are unchanged by these correcttons and response values vill differ only slightly from these tn the prbltahed* test example, see page 16. "SHAKE 4 Computer Progran for Earthquake Response Analysis of Horizontally Layered Sites,” by P. B. Schaabel, J. Lyeper, and H. 2. Seed, Report Nov HERC 72-12, Earthquake Engineering Research Center, University of California, Berkeley, Decesber 1972.PROSRAM S4A APRIL L372 JUNE 1972) CHuvez = 2PTION 7 CORRECTED = SUSROUTINES STEPG ann FFT Aa TED TO UNIVAC COMPUTERS eMSEs 1972 Cra Vee ~ SUARIUTINES DRCTSP_£vO STRAIN CravOe = SUSRIUTINGS SHAKITSEARTHOS CASOTL +MOTIONyST#T ySTAAIN REDUCE SUTPR RESPSFFT «FFT AND arshe EXEPUTION TIME REDUCED Hy UP To Sn DERCENTs. COWPLEX SnEa® MODULUS CHANGED FUOM'G(1,eTe2s@RETA) TO GCL en2,0neTae2+ 192.0927) SRT (ieeseTAre2)) bEeFInttTiuas TN3uT MOTTON DRJECT MOTION woTLow weaa IV #204 cagos MuTEIN USED AS BASIS FOR COMPUTING NEW MOTIONS IN a SOT, PROFILE fOAPUTED MOTION = WOTIIY COMPUTED ANYWHERE TN a GIVEN SOTL, SOT. PRUFTE FROM G BIVEN OBJECT WOTIOVSUBROUTINE SHaKIT(XyaXeabySeINV) . THIS ROJTIVE CauuS TOE DIFFERENT SEAUEVCES OF OPERATIONSIS DIFFERENT OPERATIONS Cav 3E PERFORMED aS LISTED SELOW, INTEGER TITLE YT? COMPLEX ty ax GCOMBLEX Ge Ve PLUSs MINUS DIMENSZON'8525¢19)5 a3SP2 19) ,aRSCa(19) DIMENSION LLG3he LT{3) #LNSW C3} DIMENSION LULA y LLG5(9) #LLBCH (2) ve PL (2) ¥S8H (2) aLAV U2 DEMENSLON 4(3y9) 6A (342701 ¥4R(24550) 950701 sTNVCTOD AIMEMSION LUStE3)s LTSCSI4 LOBC15) HLPL3) bingwsto. RAGMAR STATUES) ¥DTs Mey Mae DF aHX CONEY SO7Las TOUT (6) QL C207 voc (20) geLCT (20) om (20) 98 (20) ¥AF 2D) Goma: /S21L Ry Fac 20) 4 wi (201+ T8129), DEPTH (20) 4 WEIGHT (20) conaoy 7S21Cc7 MSOTLyamL Cumwos *eSuTLs gl200+ VERDI PLYSt2006 wrwus i201 Como yecog iuc9.11), 142049) Comet iey DATE TeLave son uaTa (aBsrs(tye 1s101 70" TIME «SHIW SEC +6HONDS 4 7H6H / en 7) 09 74 Tw 1MFOLD Th oxcr) = TTT OXe ewe Ls 73 IN'e NEW EAiNt nce ign AF sDTy oTvEW be eMPDLD AxUeTD aX) 77 CONTI SUF = eseaseot) To toy . a LS19 PRINT 1o194KK. Tra TeWeteaXect) ania} ose 31a CALL FFT XSW, INV Sea tFERR | c Cees ceeene By PRINT 135148 READ 10000 1FR CALL Nee (Fee Kyax trist ven) awd =6 CALL FFT xe Midy INy p Sos IFERR he ¢ Ceo eee taoe ——E 14 PRINT 147648 READ 100Gs NSKI9+ NXy NSW fe FT FF TF tvieLee) we 7ST TF (WsGreenes) ANF 2089 AN 2 FAW NSKIE DO135 Tels sve SKIP Vered Tek) = FLOAT (Le1 aut 136 CouTrWeE MB Iye noRss f= Lew AR(NPLMY = SEALOXCEDD Need BRCNB aay ae ATMASCCCTD 13 contr WE TF (ASK124E041) 59 TO 193 DU136 Toe 1yNN yNSKI® Newed an (Pant AACN ATD BFeTiemeeTWeSeirER a) | DO TSI T= 1.5 a1 OCW yt) = THTle ty 0.132 T= onl ID(WPeD) = INNTiT=5) TP (nSayea.0)) TN(UPy1) = T8LAvK 132 CONTI WE IF GSayr9e11 92 TO cL CALL PLOTUIB, Wy Tr de QaSTSe 19+ Ove 24 81 Yes 69 To intSUBROUTINE EAPTHOGKY AXy Ady Sy INVD es (eesesenenes ZEEUSy SCALFS THE VALUES) FING MAXIMUM VALUE AND VBA ICUS PARAMETERS. AnD TUARSEEt ThE "ITION INTD THE FREQUENCY DOMAINS INTEGEE TITLE COMPLEX Xy Ax DIMENS TON” x® (8) DEMENSION X(300) 4 AX(342701 y84(24550198(70 5 THV(70) COmwCN f69/ MEMO ,we 2 TIME(S 50%, MA y MMR DE, ME (owe eer nro zene Pte = 6.28 PEAD 1001, RVy MAy OT, TITLE a Y 30 KUL) = xCLaaAF Xaax = MO XE Tvey = FLOAT CxvaxeL pepT BUTNT 2014 y Hy Tin Xp AF XMEX, c Co THIS 7uUTIVE GERDS THE MOTION IN THE TIME DOMAIN, £00S TRALLING © och StF TOR, ag Vp SST PERE ID | KUT = oe c c © 2PSOVE FREQUENCIES 289VE FHAK AND FIND MAX. ACC. OF © FEC = 9. SKX = 0. PTD 73 Ts rth SITE PRERT GOTO Wa} NeurencuT +i xy coat 229 RE CSTR Sxxe SKK + xReXA SPX = SPK + FREGHXARKE Pett) = xa) FREQ = FREO + OF 33 CONTINUE SEX’ SFX/SHX FEUTSAEPLS= FCO | PRINT 29054 SFX Le tevexecteraeoy of TRH CHUL IK a VSG TE CEU HORTR Ey REET no 72 T= LeMfoud Te x)= axa PRINT” 20014 3a FMASUBROUTINE CXSILINLD c Cesterenenensersanacons cen seesaueeneunessenttantenntesenenteceeecage sans C THIS PUTING CALCULATES THE COMPLEX SOIL PROPERTIES AND TRANSFER CFUNE TIONS Fra THE Lavees © c ML NUMBER CE SOTL LaveRs c et KATEO OF COTTICAL DAMPING € ot SHEAR MODULUS c e Density c 6 = COMPLEX SHEAR vooULUS c v COMPLEX SHEAR WAVE VELOCITY ¢ PLUS = COMPLEX TRANSFER FUNCTION c minus MPLEX TRANSFEP FUNCTION c © CODED 3Y PEP B SCHNAREL oCT 1972 c Cone senacneneenseraseaunninseneeetuneeeeseneneaddaenenges tneeees tencunee c COMPLEX Gy Vy PLUS, MINUS) MU COMMON /S7LLAY TONTU6) sBL{20) ¢GL(200 yFACT (209 4H( 209, 812019 3F(20), COMMTL /ESOIL/ GI20)4 VE200, PLUS( 20), MINUS(2O}, med GINAGeZ.*OUt TPO CTT #30 TIT BLUE EL GRFALHGLCT)#(1 0-2 eBLUTDABLCLDD SU sCwLXICPEAL, CT MAG) 1 CONTI NUE DS 20 = 1yNd seer Mu» CSORTC> C1 Rare CLI/6 00) FLUSCT) = (Le # Muls2e MInUS(TY= CL = MUI7 25 2 contre PETURN, ENDSUBGQUTENE MOTIONENLsINGINTSLLGLTs XeAX) THIS ROUTINE CALCULATES THE MOTION IN ANY TWO SOIL LAYERS OR IN ROCK FROM MOTION GIVEN IN ANY LAYER OR IN ROCK nn NUMBER OF SOIL LAYERS EXCLUDING ROCK .N NUMBER OF LAYER WHERE OBJECT ENT = ma TEON “TYPE If £90 OUTCROPPING LAYER LLC) = NUMBER OF LAYERS WHERE OUTPUT MOTIGN IS WANTED Max 3 LAYERS, LO» MOTION Tree 0 = GUICROPPING Laver 1 = LAYER WITIN PROFILE xo OBJECT MOTION 2x0) © OUTPUT MOTION CCOEO BY PES B SCHNABEL OCT 1970 INTEGER LLG LT (3) INTEGER TITLE COMPLEX A4(3) COMPLEX Xp Ax COMPLEX” “Gy Vy PLUS MINUS COMPLEX Ey Fy EEy FFs Ay EXy AINy {PLZ DIMENSION’ x(300)y £X(34270) S70), 1Nv (70) WON JEDY WENN ME 2, TI TLECS) yOTs Hay MUA, DIESER COMMON /9UTLAY TDNT CG) y BL (20}y GL {20%¢ FACT (2005 HU20)4R 20) mamwon fe $ 2 PLUS(20} 5. MINUS (20) COMMONTERCUT/ NCUT; NTERO BF 20) IPI = CMPLK (+) 6.289 Do 20. L= 1ys te OF axtusny TFTRCUT-€0-MFOLOI Go TO-25 DO 30. TENZER Gy MEOLD BR (Ly T)SCHPLXCO ay On) 2otoNTINUE FEO = Oe ete SCRE a} =r feed. FREQ = FREQ + OF = FREQHIPIZ OO gL k= LyM TP (K-NESIN) GO TO 192SUBROUTINE STRTC UT) NLsDGMAX)PEMUL 9Xp AXe Ady SF5INV) THES POUTENE CALCULATES STRATN TN THE MIODLE SF EACH LAYER AND FIND NEW SCIL PROPERTIES COMPATIBLE WITH THE STRAINS COMMON /SCTLOG/ $(91 2095 AS14200 85(9520)4 NVC) Duvoy Zest 20by ViZ0), PLUSIZO)» MINUS (20) DISENSTON THAC(ZOD MAX (201, STRI20) DIMERSTIN XC ON)y AXUBy O4)y AALZ 128) ySEULOD, EAVCLDD © 00 43.1 = Ly HESLD CAC ty = REALCKUTDD 43 NZI) = ATMBGOCUTDD boon" t= dymecuo PR(Qs1) = AR T9726 LaK(3en = axt2yD Pr2=6.283, LPT 2a WOLXUO4y P12) Gt = 32.2 Ze Diem FEO = 0. xU) = oy FE = GT/UIprasviK)) BE WUK)/2aat 2/2 /V KD =e aNCUL FPEO = FREQ + OF EX = CEXP(FFEQREED XCD) = CAC, TOHEX = AKU, LIZEXIAPR/EREG Ex = exery Fo exlaeT rex Fos ax(3ytivex AXC2y1)2 PLUSCKOHE & MENUSCOFE EXC3sTIs OLUSURI#E # 4 INUSCK DAE 20 CONTINUE ENAxLK) = 4 PFUPCUTSFO.NFUCDY GO -VO-ZP Br ize Tr=hzer ayMecLO. X(IDISC¥PL CO. 00d taz Camttsue ¢ © OETERMINE MAX. STOAIN BY INVEATING FCURTER TRANSECEM OF STRAIN © otnto THe Tiwe ommarn eS CREE RW Ex Hay KH NOMARTSUBROUTINE UTPA(KKgD PTH LSeK2yLMy LT9Xp AK Ady Ss ENVD THIS F0UTENE TRENSFEGES THE VALUES TN AXCLM, 1 INTC TRE TIME DOMAIN, INKO by PRINTS AND PUNCHES OUT THE RESULTS. © o. SEX = Os Sxx = 0 CTEENSFORM” VALUES IN X OR IN AX TNTO THE TEMEDOMATN 07 24 1 = LyMectD TF (L5-£2.0) co Te 241 save = xi) XDD = axCtsy7) ER(LS)1) = SAVE Bel xe = CasscxtTy) SXx= SAX + XARA, SPX = SEX + XBSPOFQERE FREQ = FREQ 6 DF 24 CONTINLE: SEX = SFX/SXX ‘ on [EE PUNCH 209050406) yJe1y 8) ot (ko e800 21 PAINT 2019, (380009 J = teBe hoe om wernt « aA EPCOS PPC PYT i] Treen) Eat 09 27 'T = 1,MFOLO SEVE = AX(LSs1) AUS) = UD) 27 x1) 2 save RETURNSUBFOUTINE PESP(LLyLNgLSpNNy eA Xe Ay Sy INVD c Carre wesc eee cece treet nee n eee erees C THIS PanG2aM FEEDS OFTA Fra RESPONSE SPECTRUM ANALYSIS. NECESSARY SUBPTUTINES —DECTSP, CMPMAXy PLOT € ¢ AN = FESPTHSE SPECTRUM NUMBER c nO = NUMBER OF DAMPING VALUES ¢ Kay = SWITCH c £00 ACCELERATION SPECTRUM ¢ FO 1 VELOCITY specTaum c FO 2 ACC. AND VEL. SPECTRUM c Kt = surtew c FO 1 PLOT FESPONSE SPECTRA ACCORDING TO KAV c Ke = SwITcH € FQ 0 NO PUNCHED AuTOUT € NEO. PUNCHED CUTPUT ACCCROING TO KAV © x FOURIE® TRANSFCRM OF CAJECT MOTION c a FOUR TEP TRANSFORM OF COMPUTED MOTIONS. c Dw PERICG STEPS © nM NUMBER OF EACH STEP © r PESIOOS WHEFE PESPONSE IS TO BE COMPUTED c aor TU) = ,oo1 Cc SAVE VALUES THox IN AE OF ALT = LesFOLD Neue ty = BEALOMET DD AQQD) = ETMAGOTND tr (5.Fo.0) GO 19 12 XCD) = ax 9 ED LL CONTINUE TEENSFOOM VALUES TX O@ AX INTO THE TIME OOMEIN ER PERT GTN SaTPERYs=201 Ee aaIh ERT, 27 Tet Te (NN.GE.S) RNS 0 Ma ewe D2 1st T= 195 IDL TDGANGT) = TETLECED pe 132 t= 6yit ToCoMyt) = TONTCE=5) EE (US269.0) TO(NNyI) # TSLANK 132 CONTINUEs-a SUBFOUTINE STRAING Lhe LGSp LPCHy LPL sLNVsSK»Xy AXy AA WN ySy INV) THIS SUBROUTINE COMPUTES STRAIN AND/OR STRESS TIME-HISTORY AT THE TCP CE ANY LAYE2 FOR ECCELEGATION HISTORY KROWN IN ANY LAYER TWO RESPONSE HISTORIES ARE COMPUTED IN ONE RUN a INTEGER TITLE, TP CONDLEX xy Ax COMPLEX Gy Vy PLUS, MINUS COMPLEX Evy EEy Ay Hy IPL2y AE, AF EX, DIMENSION’ ast S10) DIMENSION FEC, APCD DIMENSION KCL) 5" AKC yL}y AACZeLIyS (Lhe INVELD DEMENSION LLC21q LGS(D1y LOCH(2) y LPLED) » SK(2 J yLNUC2) COMMON /SOTL AY LONT(G)y BL {202 GL ( 20) 4 FACT( 20) 9H 201 y Ri 20) » BEA201 Commo /SPILB/ FAC (ZOV4 WL(20}y TP(20}y DEPTHIZO I+, HELGHTE 20) COMMOW FCSOILS G(20)4 Vi20)y PLUS(Z0) + MENUS(20) COMMON FEOF MEOLDGMA2sTITLECS 4 OTs MA's M¥Ay DPGMX Cows ser6/ TOLs yy Ti2049) (oe ct wcotnee ¢ 1PI2 © CMPLX(0046.283) Gt = 32. BxC2y1) "= 0. axGyLI = OF FREQ'= 0. DATA (43S1S(11, T1410) /6H TIME y6HIN SEC y BeoH , T= Gt/IPI2 c € _ STASTING AT The SURFACE THE STKALN I'S COMPUTED SUCCESSIVELY OCWNWARO FE ecuente © > Ky Tee OS 09 2 1 © 1yMFOLO 26x = kT) post = aye TF (LUAU). Ee.) GOTO 3 TE(NCHTAFO« ELD) O15 38 00 34 Ten7eacy MELD KULTI=CMPLKL Oey 00) 36 CONTINUE 33 CONTE UF CALL SF Sv x OF 37 Tat Pere PECL 281-1) SE AL(KCLI A100. 32 BR(LA2mI) = AIMAGIX( 1) 34100 3 CONTINUE InveSsTFERR»-2)2 SUBROUTINE PEOUCECTFR) Xy AXy LL} ¢ ¢ CAs teat ease cere eee eee ee eee eee € THIS ROUTINE INCREASES TIME INTERVAL AND REDUCES NUMBER OF VALUES c c c TF = DIVIDING FACTOR ON LENGTH OF FRECORD c MULTIPLICATION FACTOR ON TIMESTEP c MUST BE & POWER OF 20 € or TIMESTEP IN SEC. € oF FREQUENCY STEP IN C/SEC. © Me = NUMBER OF PCINTS USED TN FOURIER TRANSFORM ¢ x FOURIER TRANSFOR OF CAJECT MOTTON ¢ BK FOURIER TRANSFGRM OF COMPUTED MOTICRS. c c C CODED BY PE A, SCHNABEL DEC. 1970. © McDLFIED SEPT. 1971 c c ee ee rs c INTEGER TITLE JMVCN /EG/ MEDLOMAZ,TITLE(S)¢DTy MA 4 MMAy OF, MX Covey FeecUT/ VCUTsRZER™ CaMOLER Xp DIMENSION’ XC 68), AK(3y 64)y LU(BI Fl = .5/07 FLOAT UD FR) oTare Mas TER wae 2 wn272 N= NFotD #1 09.12 1 = MFOLOWN xO) = 0 po 12 b= 13, Tr (LLCLI-LE20) G0 To 12 Ax(LyTD = Oy 12 conTINuE YFOLO = WFCLD #1 F2 = 5/01 PRINT 1000¢F L9F2,0T MA rua = FLoATOMeD Hx = (ALOGLOCEMAI/ALGGIO(261)-16 Le (vt CT. 2m eC Mx) MX = XL TETRCOT-LESPFOUOY GO-TO 15 CUT=4FOLD 15 CONTINUE, araSUBROUTINE FET (AsMyINVy SyTFSETSIFERED DIMENSION @C1)y INVCL D4 SULD ANC 3) ¢HC3)4NDC3) 9 WC 2) 9 MRE 2) 932) EQUIVALENCE (NLAWCLI I, (NZENC2D Dy (N3yNEBID mTewex0(24 MTT) TO___TF (1SaSTTFSETD=11 61046105 20 610 LxOU CLD gC 2) g M(B) 4-2 ST =MAX0(2¢ MT) IF (41-20) 630,630,620 620 Tree G0 To 600 630 T FER 30 TFERP=1 PRINT 1000 stop 1000 FORMAT(31M --- ERROR IN FOURIER TRANSFORY 40 LereRs9 rc c Nis 2eeT n2e2eeu N5ezeeua. Te CLPSET) 5030470 50 Wx=NLeN2ENa ENSNX be 60 B(2AT-1) 8 @#T-1) 7N 60 AC 2eTya=80 2410 76N TO NBL) aNI#2 NPC2)=NEU LY*N2 NPU3)=NPL219N3 DD 330 108143 TLsNPC3}-NP UTD) Ts tbe Nxou SUaetOTINE FFFT (2pm, INV, $, TFERR TFSETD LGiso. nrot=aeem Not a= 2eN TOT FNantnr De Lo Ta2,hT9r2¥2 Uden} 00 20 tetynror2 21) 2A TES CALL FFT (pbs TM 4 Sy TPSET) TFERRD MOVE LAST MELE OF ACS1S DOWN ONE SLOT AND AOD ACN) AT ACTTCH TO GIVE ARRAY FOR ALPRIME AND AZPRIME CALCULATIONSURFROTEN FSR TR TS PERE, TEST 70 TUS TRVT SOT t t ie i n i 1 Fave 2) ek Ce euneli=2 (Nt=1) at T1244 O70 te3yATM 242 reo.se 8) sete atest) DP 69 TeLyAT ary? KgetTOT 24 21 E(KBde BUKB=2) Dee a) 28 (KB=LD RInen Ir 2 Prerrer Labse26s/en SimcoeL) casey DG 50 fe1ynt(COMPARISON BETWEEN ORIGINAL AND NEW RESULTS ‘The redefinition of the complex modulus c* from of = ca +21) here § is the fraction of eritical damping to the daproved value ot = ca-26* + 218 AB) slightly changes the response values computed by progran SHAKE, The following table shows the influence on maxtaum accelerations through the profile, see page 50 of the original repore*. Dapeh 0.0 7.0 20.0 30.0 42.0 62.0 20.0 100.0 120.0 Getginal Max. Ace. 0.09377 0.09259 0.05934 0.08487 0.04666 0.03195, o.o2az3 0.01793, New Max. Acc. 0.09878, 0.09758 0.03942 0.03540 0.04667 0.03140 0.02364 0.01763 all other differences observed vere of sintlar or snaller relative magnitude.