You are on page 1of 19
The Mexico Earthquake of September 19, 1985— Design Spectra for Mexico’s Federal District E, Rosenblueth, M.BERI, Mario Ordaz, FJ. Sanchez-Sesma, and S.K. Singh We describe the methods used to obtain seismic design spectra adopted for different zones of Mexico's Federal District in 1987 Building Code. Paper exposes the two approaches followed in the study, deterministic and probabilistic. The assumptions adopted are presented and justified. Several aspects that require detail- ed scrutiny are pointed out. INTRODUCTION This study was undertaken in order to propose the seismic design spectra for the 1987 Mexico City Building Code. Two approaches were used: a deterministic analysis and a probabilistic one The deterministic approach used catalogs of Mexican earthquakes of this century. For large earthquakes the catalog’s coverage includes last century. Four representative events were selected under the assumption that they are the most dangerous earthquakes one can reasonably expect during the next 150 years. For these events, the Fourier amplitudes of accelera- tions at the University City (CU) were estimated from semiempirical rela- tionships and applicable seismological theory. The station at the Univer- sity City is the one with the largest number of earthquake records in Mexico City’s Valley. From the estimated Fourier spectra at CU the Fourier amplitudes yere calculated for representative sites of the Federal District using the one dimensional theory of shear wave propagation in viscoelastic layered media. From focal data of the selected earthquakes their duration was estimated and through the aplication of random vibration theory the expected acceleration response spectra for 5% damping were obtained for the (ER, FISS, MO) Instituto de Ingenierfa, UNAM, C.U., México 04510, DF and Centro de Investigacién Sfsmica, Camino al Ajusco 203, México 14200, DF. (SKS) Institute de Geoffsica, UNAM, C.U., México 04510, DF and Instituto ae Ingenterfa, UNAM, C.U., México 04510, DF. This paper appeared, in Spanish in a slightly different version, in the Proceedings of the 7th National Conference on Earthquake Engineering, Queretaro, Mexico, November 19-21, 1987. 273 ‘©Barthquake Spectra, Vo. 5, No.1, 1989 274 E. Rosenblueth et al. locations selected. These results and the computed spectra for the 1985 Michoacan earthquake, at the sites where records are available, allowed construction of the design spectra proposed for inclusion in the 1987 Bullding Code. These are simple shape envelopes of computed response spec- tra with modifications to account for the expected duration of ground shaking at each site, and for inelastic behavior of structures. In the probabilistic approach, the information from the catalogs was combined with that from tectonically similar regions in order to increase the data base. The non-Poissonian nature of the generating processes was accounted for when that assumption was significant. Design spectra were computed for the combination of all relevant events in such a way that, for each site, the spectrum was associated with a fixed expectation of the present value of the losses. Results were compared with the proposed code spectra. It was found that, in general, the adopted spectra were consistent with the probabilistic computations. EARTHQUAKE CATALOGS The present analysis is based on the catalogs of the Institute of Geophysics of UNAM and on information of the National Seismological Ser- vice. They list magnitudes and focal coordinates; for events between 1806 and 1899 such values are inferred from the area in which the shocks were felt; for 1899 to 1987 the information has instrumental basis. Mexican earthquakes have been divided in four groups (see Fig. 1): 1 Local earthquakes (L) 2 Continental plate earthquakes (A) 3 Intermediate depth earthquakes (N) 4 Subduction earthquakes (G) LOCAL EARTHQUAKES In the North American plate, under the Valley of Mexico and in fact along the Trans Mexican Volcanic Belt, there exist several systems of seismogenic geological faults. The principal systems (see Fig. 2) are: A, faults essentially running NE; B, faults with SE orientation, and C, faults with roughly EW direction. In what follows we will use three magnitude scales according to how the data were published: ML, Ms and Mv. Mi, the local magnitude, is based on the trace of a standard seismograph placed at a standard distance from the source. Ms is computed from the amplitudes of surface waves. And Mw, which is the only scale based directly on physical quantities at the source, is obtained from the seismic moment. In general, ML practically coincides with Mw up to magnitudes of about 6.3, while Ms differs little from Mw In the magnitude range of about 5.8 to 8.0. Outside these ranges, Mi and Ms are almost systematically lower than Mv. (see Idriss, 1985.) The maximun magnitude recorded in this century is ML = 5.1. Given the alternancy of motions of fault systems A and B, the information of respect- Design Spectra for Mexico's Federal District 275 Guerrero Acomboy es Figure 1 - Considered earthquake groups and their possible occurrence regions. Figure 2 - The Valley of Mexico and its fault systems. 26 E. Rosenblueth et al. ive motion rates is lacking. Even if these were known, the use of relations to predict the maximun magnitude would be limited for, according to Mooser (1987), the seismogenic zone under the valley is relatively thin Although the magnitude of local earthquakes is likely to be as large as 5.5 (Mooser, 1987) it was decided to select M. = 4.7 with focal distance of 11 km for the deterministic analysis. This was dictated by the fact that each local event affects only a small zone; for magnitudes close to the maximum possible, the return period for any location of Federal District is much larger than the value corresponding to the whole region. (In contrast, nonlocal perturbations affect the basement of fhe valley with comparable intensities everywhere. ) CONTINENTAL PLATE EARTHQUAKES In the remaining of the North American plate the slip of seismogenic faults is due essentially to two phenomena: the displacement of this plate with respect to the oceanic plates, owing mostly to the drag and push of magmatic convection currents, and to the bending induced by the friction of oceanic plates. The earthquakes that most seriously threaten the valley are produced at the intersections of principal faults or close to the end of those growing steadily. The most dangerous candidate site is the East end of a couple of faults defining the Acambay graben (Mooser, 1987). We se~ lected that place as representative. A Mv = 7.0 earthquake occurred there In 1912 and the site has not produced significant shocks since. The verti- cal slip rate in Acambay is nearly 0.4 mm/year, which suggests a return period of about 1500 years for magnitude 7.0. In this light we assigned to earthquakes of this type a magnitude Mw = 7.0 and a focal distance of 80 kn. INTERMEDIATE DEPTH EARTHQUAKES Recorded magnitudes of interplate earthquakes in the Cocos plate de- crease with the distance from the subduction zone. This can be explained in terms of the tensions induced by the gravity and by the drag caused by the convection currents. These tensions are greater close to the subduction zone. The variation of magnitude with distance from the coast can be ap- proximated by the curve in Figure 3. Given the depth of Cocos plate beneath the valley it is found that the most dangerous earthquake of this group has Mv = 6.5 and a depth of 80 km. SUBDUCTION EARTHQUAKES Subduction earthquakes have historically generated the most severe consequences in Mexico City. The maximun magnitude recorded since 1800 is Ms = 8.4 for the 1932 Jalisco earthquake. However, the earthquakes origi- nating near the coasts of Guerrero and Michoacén produce the most violent motions in the valley. These events have had magnitudes systematically low er than Me = 8.4. An analysis of the zones that can slip in a single great event shows that the most violent shock that one can reasonably expect in Design Spectra for Mexico's Federal District 27 Figure 3 - Variation of magnitude Ms and focal depth H vs. distance to the coast De for intermediate depth earthquakes. the Guerrero seismic gap, West of Acapulco, will have Mw = 8.2. The dis- tance from the Valley to the rupture area is very uncertain. Probabili- ties of 0.25, 0.5, and 0.25 were assigned to R = 130, 248, and 280 km, res- pectively. The average computed time between great events is 59.3 years. The last such earthquake occurred in 1911. FOURIER SPECTRA ON FIRM GROUND Expected Fourier spectral ordinates on firm ground for all nonsub- duction earthquakes were obtained from the w~ model (Boore, 1983). The model produces Fourier amplitudes of spectral accelerations, A(w), in terns of source and path characteristics: A(w) = C Mo S(w,We) PCW, wm) exp(-wR/20B) /R a where C = constant depending on average radiation pattern, on free surface and site effects and on average path conditions; Mo = seismic moment in ergs; S(w,ue) = source spectrum, (Aki, 1967; Brune, 1970); we = corner frequency = 2nfe; fe = 4.9 x 10° B(Ac/Mo)”~; B = shear wave velocity (km/s); a¢ = stress drop (bars); P(w,wa) = high frequency filter (Boore, 1983); um = 2nfm; fm = maximum frequency; R = focal distance, Q = quality factor that controls attenuation (in general, a function of frequency). 278 E. Rosenblueth et al. Adopted parameters appear in Table 1. Their assessment is described as follows TABLE 1 Selected parameters for Fourier spectra” Earthquake ‘ 8 Mo fo Me B Te (km) (ergs) (Hz) (bars) (g/cc) (km/s) (sec) Acambay 7.0 80 3.5x10% 0.10 100 3.1 3.2 30 Local 4.7 11 14x10 0.80 50 25 2.3 7 Normal Gis | 802 7 Oxide Oo 200 alae 30 * In all cases C = 3.74, Q = 200, fm = 8 Hz (Eq. 1) For the intermediate depth earthquake, which can be considered a nor- mal faulting event, the records of July 8, 1964 and October 24, 1980 earthquakes were available. Therefore, the parameters of theoretical spec- trum were selected such that satisfactory fit with both observations was achieved No record was available for the Acambay type event. Parameters used were those of the preceding type, save for a smaller stress drop because of the shallower depth. A satisfactory agreement was found between the theo~ retical spectrum and the one computed for the July 12, 1974 local earth- quake. Again, the assumed stress drop is small. Semiempirical relationships were used to represent the subduction type earthquake (Castro et al., 1987). These expressions predict expected spec- tral ordinates of ground acceleration in terms of magnitude and distance. Results are displayed in Fig. 4. Curves represent expected Fourier spectra against frequency. FOURIER SPECTRA FOR OTHER LOCATIONS Using available information (Marsal and Mazari, 1959; Comisién Hidro~ légica de la Cuenca del Valle de México, 1966; Mazari, personal communi cation; Mooser, 1987; Jaime et al., 1987; Jaime and Romo, 1987; Suarez et al., 1987) stratigraphic profiles were determined for sites of the transi- tion and soft zones in the valley. For each site the transfer functions that relate incoming motion at the basement with the surface response were computed using the Thompson- Haskell matrix method (Haskell, 1962; Aki and Richards, 1980) of wave prop- Design Spectra for Mexico's Federal District cm/s 10 ~-~= Acomboy 7 —-— Normot —— Local 001 ° Guerrero 0.001 oor on v 10 Frequency, Hz 00 Figure 4 - Fourier amplitude spectra of accelerations for selected events. ' zt 4h p i\ sie 210 s tt === Site 105 = | il sile 064 Frequency Hz Figure § - Transfer functions for three sites of the Valley. 279 280 E. Rosenblueth et al. agation in viscoelastic layered media.Only vertically incident SH waves were considered. As an example, Fig. 5 shows the transfer functions com- puted for three sites. These results do not take into account the effects of oblique incidence, of surface wave propagation, nor of nonlinear behav- jor. Their influence has been qualitatively considered in the proposed de- sign spectra. RESPONSE SPECTRA For the Building Code provisions the acceleration response spectra for 5% of critical damping are required. These were computed using random vi~ bration theory (Boore, 1983; Boore and Joyner, 1984). This requires the Fourier amplitude spectra and the equivalent duration of ground motion. The duration assessment was done on the basis of the observed values for earth- quakes recorded at CU. Table 1 gives the equivalent duration of the assumed events, Te. Computed response spectra are displayed in Figs. 6-9. They in- clude a reduction factor of 0.4 that will be discussed below, The method used for computing response spectra from Fourier amplitudes of ground accelerations was calibrated through direct comparisons. For var- ious events recorded in several sites of the valley, response spectra were computed from the records and from the Fourier amplitudes (Ordaz and Rei- noso, 1987). The agreement was satisfactory showing that random vibration theory provides an efficient way to generate accurate estimates of response spectra. This had also been confirmed by Romo et al. (1987) Design spectra are based on simplified envelopes of the computed accel- eration spectra. The 1987 Building Code reduces these linear spectral re~ sponse accelerations in two steps. One is a flat coefficient of 0.4, inde- pendent of the structure’s natural period. The second one is the reduction factor Q’. The latter is equal to Q for periods exceeding Ta (where the first break in the envelope spectra occurs) and is obtained through linear interpolation for shorter periods. Q is the nominal reduction factor. For zero period, Q’=1. Selection of the 0.4 reduction coefficient is based on the following considerations. 1. Modern building codes recognize as extremely conservative a design aimed to keep structures within the linear range. Taking advantage of structural ductility, significant reductions can be achieved without important damage. Moreover, ordinary structures have great reserves over their capacity computed in a conventional way. For narrow band earthquakes, as those of high intensity in the lake zone of the valley, ductility reductions are more pronounced than for wide-band events, which are typical of strong earthquakes on firm ground. For the latter the theoretical reduction factor is about 1/u, where # = ductility factor (Newmark and Rosenblueth, 1971). The reductions computed using the EW component written by the 1985 Michoacan earthquake at a site in the lake zone are much more drastic. This is especially true for periods close to that for which the maximum elastic spectral ordinate occurs (Avila and Meli, Design Spectra for Mexico's Federal District 281 1987). When the ductility factor of the structure under consider~ ation exceeds about 2 the reduction factors Q allowed in the code are intended to approximate # in a conservative manner, With small- er values of w, Q underestimates 4 in a systematic and quite appre- clable way. In fact, the elastoplastic response spectra for w= 2 and 4 = 4 around periods close to the maximum elastic ordinate roughly coincide with the elastic design spectra reduced by 2.5 Thus a reduction factor of 0.4 together with the nominal value of Q lead to the values observed for the EW component of the SCT rec- ord. This situation does not apply for structures with = 1. In such cases it seemed justified to preserve the reduction factor because even the systems for which the code specifies Q = 1 have the capacity of developing some ductile behavior. In fact, the ductility required to reduce the spectral ordinates from 1g to 0.4g is 1.6 in the case of the record mentioned. Also, real structures, particularly the most common short-period ones, have substantial reserves of capacity to resist lateral loads, in the sense that conventional methods of analysis seriously underestimate this ca~ pacity. Coefficient 0.4 is intended, in part, to take these re~ serves into account These results correspond to elastoplastic single degree of freedom systems with stable hysteretic cycles. The situation is less favorable for systems that undergo deterioration in strength after a great number of alternating load cycles. However, the va~ lidity of the comment still holds regarding the greater efficiency of ductility for this type of earthquakes than for those of wide band. The new building code establishes especially conservative strength reduction factors for structures that can suffer signif- icant deterioration. Rather than a flat reduction factor of 0.4 applied to the linear response spectra with 5% damping, it would have been clearer to use inelastic design spectra dependent on ductility, sensitivity to load repetitions, reserve strength, and structural periods. This would have avoided apparent contradictions, such as the use of the reduction factor for structures with nominal Q = 1. Nevertheless, it was felt that the format adopted in the code was preferable be- cause it implied a less drastic departure in its presentation from the 1976 Code and from the 1985 Emergency Regulations. 2. It is of interest to compare our approach with SEAOC guidelines (Structural Enginners Association of California, 1986). In these guidelines the reduction factors with respect to elastic spectra for 5% damping vary between 1/3 and 1/12 if certain requirements are fullfilled. These are similar to the ones of the 1987 Federal District Building Code, that allows reductions of 1 to 1/4 with response spectra already reduced to 40%. Therefore, the reductions with respect to expected elastic spectral ordinates with 5% damping range from 1/2.5 to 1/10. On the other hand, the departure spectrum in California is the expected one plus one standard deviation, whereas for the Federal District the initial spectrum is the ex- pected one. This roughly compensates for the 20% difference in the reduction factors 282 E. Rosenblueth et al. 040 —— Guerrero — Michoocén.l9B5 ooo Constant exceedance 032 rote 2 Locot 024 Acomboy 0.16 008 Figure 6 ~ Response and design spectra; lake-bed zone, Ts=3.97 sec. Loco! 016 ---- Acomboy o —-— Normot — — Guerrero —+— Michoocén S012 Constant exceedance rate 0.08 0.04 Figure 7 - Response and design spectra; firm ground Design Spectra for Mexico's Federal District 283 0.32 2. uP 0.24 Loco! === Acomboy —-— Normat —— Guerrero “~~ Constant exceedance rate 0.08 Figure 8 - Response and design spectra; transition zone, Ts=0.87 sec. 0.40] — — Guerrero Michoacén,1965, Constant exceedance rate Figure 9 - Response and design spectra; lake-bed zone, Ts=2.08 sec. 284 E. Rosenblueth et al. In a short range of structural periods of practical interest the spectrum obtained for local earthquakes is slightly above that adopted for design. This apparent error on the unsafe side is justified in terms of exceedance rates, as we will see. Moreover, the apparent strength of very rigid structures under short duration earthquakes exceeds the strength predicted by conventional analyses (Meli and Avila, 1987). This is supported by experience in the Hill zone of the Federal District, where the short period, rigid, old Colonial buildings subsist. DAMAGE DISTRIBUTION AND ITS IMPLICATIONS. The following sources of information were taken into account 1. For local historic earthquakes, very stiff Colonial buildings not designed to resist seismic action, have only experienced minor cracks under magnitudes of up to 4.7 and focal distances as low as 7 km. 2. The distribution of damage in houses and buildings for the 1957.7.28, Me=7.7, earthquake (Rosenblueth et al., 1958). 3. For the 1979.3. 14, 7.6, Petatlan earthquake, the distribu- tions of damage in pipelines (Direccién General de Construccién y Operacién Hidraulica, 1979), and damage in buildings (Meli et al., 1973) 4. For the 1985.9 earthquakes, the distribution of damage in pipelines (Direccién General de Construccién y Operaci6n Hidraulica 1985); the preliminary damage survey (Mell et al., 1985); the equal damage level contours (Secretaria General de Obras del DDF, 1987); the equal damage index contours (Iglesias et al, 1987); and the more accurately computed forces to which some selected buildings were subjected (Avila and Meli, 1987). It was concluded that the design spectra originally proposed in the 1985 Emergency Code were satisfactory, excepting two areas in which the damage level 1s much higher than that of the surrounding zones. Various sources of information confirm this situation. In fact, greater effects of surface waves are expected in these places. Such waves were not considered in calculating response spectra. It has been argued that higher design base shear coefficients should be used in these areas (Iglesias et al., 1987). This was accepted by the Subcommittee for a small portion in the transition zone, but not for the rest, which is Included in the lake-bed zone. This was decided in the light of the various conservative provisions already contained in the code for this zone. Detailed 2D and 3D soil dynamic anal- ysis, still pending, should lead to a more precise microzonation. PROBABILISTIC ANALYSIS OF LOCAL EARTHQUAKES Because there are many potentially seismogenic faults, the generating process was assumed to be Poisson. The following exceedance rate-magnitude Design Spectra for Mexico's Federal District 285 relation was adopted, da if M< Mu ACM) = (2) 0 otherwise (Cornell and Vanmarcke, 1969), where A = A(M) is the exceedance rate of magnitude M, Ma is the completeness threshold magnitude, Aa = A(Ma), M=M-~- Ma, Mu = Mu - Ma, Mu is the maximum magnitude, and @ is a param- eter. As mentioned before, Mu is 5.1. To evaluate the prior distributions, we considered the world-wide average of intraplate earthquakes. We used a diffuse prior distribution for As. By incorporating data in the catalog, we obtained the posterior densities using the method proposed by Rosenblueth and Ordaz (1987a), but doubling the number of events per unit area to take into account its concentration on the southwestern part of the valley. The valley was divided into ten concentric circles to account for the variation in focal distances. The focal depth was assumed to be 11 km. We used the approach described for calculating the expectations and coeffi~ cients of variation of 5% damping spectral ordinates. PROBABILISTIC ANALYSIS OF CONTINENTAL PLATE EARTHQUAKES We assumed that an event of this kind could occur anywhere inside the two tectonic provinces, Trans-Mexican Volcanic Belt and the Southern Sierra, and that the generating process was also Poisson. Equation (2) was used to describe the variation of A with M. The exceedance rate per unit rea was assumed constant in both provinces. Mu was assigned a prior Gauss- jan distribution with expectation of 6.8 and standard deviation of 0.25. The distributions beginning with a noninformative prior for Aa, were up- dated with the corresponding data. Each province was divided into five regions with focal distances ranging from 123 to 392 kn. PROBABILISTIC ANALYSIS OF INTERMEDIATE DEPTH EARTHQUAKES It was assumed that the area in which these events can occur is lim- ited by the locations of epicenters of earthquakes identified as normal faulting, and by the inferred position of the subducted Cocos plate under the North American plate. The prior densities of Aa and 8 were determined as for the previous types The seismogenic area was divided into ten regions. In each region a prior Gaussian distribution was assigned to Mu, with standard deviation 0.3 and expectation 6.8-8.0 depending on the depth of the subducted plate: the 286 E. Rosenblueth et al. greater the depth, the lower the expectation. The focal distances varied from 96 to 254 kn. PROBABILISTIC ANALYSIS OF SUBDUCTION EARTHQUAKES The large and great earthquakes of this kind are generated in a few well-delimited zones, so a Poisson idealization is not justified. We assigned a gamma distribution to the interoccurrence times of events with M = 7. For M< 7 we adopted the Poisson hypothesis and Equation (2) for A(M). For greater magnitudes we adopted a parabolic variation of -da/dM, symmetric with respect to 0.5x(Ms + 7) (This functional form is a modification of the one by Youngs and Coppersmith (1985) for the charac- teristic earthquakes). In accordance with Nishenko and Singh (1987), the Mexican subduction zone was divided into twelve regions. To evaluate the prior distributions of parameters Aa and B we used the method proposed by Rosenblueth (1986) assuming as tectonically similar the regions forming the Circumpacific Belt. For each region we took Aa propor- tional to 8LS, where L is the length of the region, S is the slip rate and 6 is the seismogenic fraction of S. To obtain the prior distributions of the maximum magnitude we used three semiempirical formulas (Bonilla, et al., 1984; Chen, 1984; Singh et al., 1980) which predict the expected Mu as a function of maximum rupture length, which was assumed to be the length of the region itself plus the lengths of the adjacent ones. Results from the various formulas were combined with the approach described by Rosenblueth and Ordaz (1987b). The corresponding expectations ranged from 8.01 to 8.33; the calculated standard deviation was 0.2. These distributions were updated using the subduction earthquakes data from the catalog and the method pro- posed by Rosenblueth and Ordaz (1987a). EXCEEDANCE RATES OF STRUCTURAL RESPONSES Consider earthquakes generated by a Poisson process. Let a = a(T,Mi) be the spectral ordinate at a given site with natural period T due to an earthquake of magnitude Mi originating in the i-th seismogenic source. Let Ai = A(Mi) be the exceedance rate of magnitude Mi for earthquakes from source i. Then, the exceedance rate of a due to the i-th seismogenic source is also Ai. The combination of Poisson processes is also a Poisson process (Parzen, 1964) with exceédance rate v(a) = EA (3) The expectation of the present value of the number of events producing exceedances of a, assuming only one limit state exists, is v(a)/z, where 7 is the discount rate. This ratio is the quantity which has to be multiplied by the expected losses in one event to obtain the expected present value of the losses (Rosenblueth, 1976). Design Spectra for Mexico's Federal District 287 For non-Poisson events whose interoccurrence times possess the density function p(-), the expected present value of the number of ocurrences is _ PV(a) = Pr (A> a) (4) a Summation of the same quantities for each region gives EN(a), the expected present value of the total number of events that would produce spectral ordinates higer than a. We will call equivalent exceedance rate the product of EN(a) and 3: n= 7 EN(a) (5) This quantity, like the exceedance rate, measures the expected present value of the total losses. Since the contribution of all regions except those corresponding to the Guerrero coast is relatively small, it seemed Justified to treat only the latter as non-Poissonian. To acount for the sensitivity of spectral ordinates to focal distance, we assigned this distance three possible values, as in the deterministic approach. We took the spectral acceleration as lognormally distributed with standard deviation of the natural logarithm equal to 0.25 and median equal to the value predicted for the deterministic coastal earthquake for each period of vibration. This is a conservative assumption since magnitude 8.2 is close to the maximum magnitude. For the code purposes we selected the equivalent exceedance rate associated with a maximum spectral ordinate of 0.4 g -the selected design ordinate- at a site with predominant ground pepiod of 2.08 s. The equiv- alent exceedance rate turned out to be 7 x 10 ~/year. This is the constant equivalent exceedance rate associated with the acceleration spectra in Figs. 6-9, where they are compared with the deterministic and design spec- tra. Although the local deterministic earthquake predicts higher spectral ordinates than the code provisions for a range of structural periods, this ordinate corresponds to an extremely low equivalent exceedance rate. CONCLUSIONS The seismic design spectra adopted for different zones in the 1987 Mexico's Federal District Building Code were obtained using both a deter- ministic and a probabilistic approach. Results from the two approaches are consistent. The working assumptions are made explicit. Additional research efforts are clearly required in (a) the assessment of maximum magnitudes, (b) the distribution of interoccurrence times, (c)detailed studies of source processes, (d) the wave propagation from source to site (including site effects), (e) nonlinear response of structures, and (f) soil-structure interaction. In the light of the present state of art in earthquake engineering, the design spectra adopted are reasonably conservative. They also reflect the consensus of a number of experts in various fields. 288 E. Rosenblueth et al. AKNOWLEDGEMENTS Thanks are given to F Mooser and G Suarez for their valuable coop- eration. F J Chavez-Garcia, E Reinoso, L E Pérez-Rocha, M Suérez, E Mena, and M A Bravo colaborated in different parts of the study. Their help is greatly appreciated. This work was supported by Departamento del Distrito Federal and Mexico’s National Council for Science and Technology (CONACYT) REFERENCES Aki, K. and Richards, P.G. (1980). Quantitative seismology. Theory and methods, W. H. Freeman, San Francisco, CA. Aki, K. (1967). Scaling law of seismic spectrum, J. Geophys. Res. 72, 1217-1231. Avila, J. and Meli, R. (1986). Respuesta inelastica ante el acelerograma del 19 de septiembre de 1985, Memorias V Congreso Nacional de Ingenie~ ria Estructural, Veracruz, México, AS.01-A5.11, Avila, J. and Meli, R. (1987). Andlisis de la respuesta de edificios tipi- cos ante el sismo del 19 de septiembre de 1985, Instituto de Ingenieria UNAM, in press. Bonilla, M.G., Mark, R.K., and Lienkaemper, J.J. (1984). Statistical rela- tions among earthquake magnitude, surface rupture length and surface fault displacement, Bull. Seism. Soc. Am. 74, 2379-2412. Boore, D. M. (1983). Stochastic simulation of high-frequency ground mo- tions based on seismological models of the radiated spectra, Bull. Seism. Soc. Am. 73, 1865-1894. Boore, D. M. and Joyner, W. B. (1984). A note on the use of random vibra- tion theory to predict peak amplitude of transient signals, Bull. Seism. Soc. Am. 74, 2035-2039. Brune, J.N. (1970). Tectonic stress and the spectra of seismic shear waves from earthquakes, J. Geophys. Res. 75, 4997-5003. Castro, R., Singh, S. K., and Mena, E. (1987). An empirical model to pre- dict Fourier amplitude spectra of horizontal ground motion in Ciudad Universitaria, Mexico City from coastal earthquakes, Earthquake Spec- tra, this issue. Chen, D. (1984). Some empirical formulae for hazard analysis in China, Proc. IV Norld Conference on Earthquake Engineering 1, San Francisco, CA, 473-180. Comision Hidrolégica de la Cuenca del Valle de Mexico (1966). Boletin de Design Spectra for Mexico's Federal District 289 Mecanica de Suelos No. 5: Julio 1963 - junio 1966, Secretaria de Recur- sos Hidraulicos. Cornell, C.A. and Vanmarcke, E. (1969). The major influences on seismic risk, Proc. IV International Conference in Earthquake Engineering, San- tiago de Chile, Chile. DGCOH (Direccién General de Construccién y Operacién Hidraulica), (1979, 1985). Unpublished manuscripts. Haskell, N. A. (1962). Crustal reflection of plane P and SV waves, J. Geophys. Res. 67, 4751-4767. Idriss, 1.M. (1985). Evaluating seismic risk in engineering practice, Proc. XII Internacional Conference of Soil Mechanics, San Francisco, Cal. Iglesias, J., Jara, M., Aguilar, J., Mota, 0., Teran, A., Gomez, A., Or- tega, R., Gonzalez, E., Guerrero, J.J., Juarez, H., Paniagua, J.P., and Sordo, E. (1987). Estudio de las intensidades del sismo del 19 de sep- tiembre en la Ciudad de Mexico, Universidad Auténoma Metropolitana. Mé- xico. Jaime, A., Romo, M. P., and Ovando, E, (1987a). Caracteristicas del suelo en el sitio SCT, Instituto de Ingenieria, UNAM , internal report. Jaime, A., Romo, M. P., and Ovando, E. (1987b). Caracteristicas del suelo en el sitio Central de Abasto Oficinas, Instituto de Ingenieria, UNAM, internal report. Jaime, A., Romo, M. P., and Ovando, E. (1987c). Caracteristicas del suelo en el sitio Central de Abasto Frigorifico, Instituto de Ingenieria, UNAM, internal report. Jaime, A., Romo, M. P., and Ovando, E. (1987d). Caracteristicas del suelo en el sitio Tlahuac Bombas, Instituto de Ingenieria, UNAM, internal report. Jaime, A., Romo, M. P., and Ovando, E. (1987e). Caracteristicas del suelo en el sitio Viveros, Instituto de Ingenieria, UNAM, internal report. Marsal, R. and Mazari, M. (1969). El Subsuelo de la Ciudad de Mexico/The subsoil of Mexico City, School of Engineering, UNAM , México. Meli, R., and Avila, J. (1987). Sismo de San Salvador. Octubre, 1986, Instituto Mexicano del Cemento y el Concreto, 25, 31-49. Meli, R., et al. (1979). Evaluacién de los efectos del sismo del 14 de marzo de 1979 en las edificaciones del Distrito Federal. Instituto de Ingenieria, UNAM, Internal report. Meli, R., et al. (1985). Efectos de los sismos de septiembre de 1985 en las construcciones de la ciudad de México. Aspectos estructurales. Institu- to de Ingenieria, UNAM, internal report. 290 E. Rosenblueth et al. Mooser, F. (1987). Riesgo sismico en la cuenca de Mexico, Congreso Nacio- nal de Ingenieria Sismica, Querétaro, Mexico, Invited lecture. Newmark, N. M. and Rosenblueth, E. (1971). Fundamentals of Earthquake Engineering, Prentice-Hall Inc., Englewood Cliffs, N.J. Nishenko, S.P. and Singh, S.K. (1987). Conditional probabilities for the recurrence of large and great earthquakes along the Mexican subduction zone, Bull. Seism. Soc. Am. 77, 2095-2114. Ordaz, M., and Reinoso, E. (1987). Uso de teoria de vibraciones aleatorias para la determinacién de los espectros de diselo del Reglamento para las construcciones del Distrito Federal, Memorias del VII Congreso Na~ cional de Ingenieria Sismica, Querétaro, Mexico, A155-A167. Parzen, E. (1964). Stochastic processes, Holden Day Inc., San Francisco, CA Romo, M.P. and Jaime, A. (1986). Caracteristicas dinémicas de las arcillas del valle de México y respuestas del analisis sismico. Primera etapa, Instituto de Ingenieria, UNAM, internal report. Romo, M. P., Jaime, A., and Ovando, E. (1987). Analisis de la respuesta sismica del suelo en el valle de México, Instituto de Ingenieria, UNAM, internal report. Rosenblueth, E. (1976). Optimum design for infrequent disturbances, J. Struct. Div. ASCE, 102, 1807-1825. Rosenblueth, E. (1986). Use of statistical data in assessing local seis- micity, Earthquake Engineering and Struct1. Dyn. 14, 325-337. Rosenblueth, E., Marsal, R., and Hiriart, F. (1958). Los efectos del te- rremoto del 28 de julio y la consiguiente revisién de los criterios pa~ ra disefio sismico de estructuras, Ingenieria, XXVIIT. Rosenblueth, E. and Ordaz, M. (1987a). Use of seismic data from similar regions. Earthquake Engineering and Struct1. Dyn., in press Rosenblueth, E. and Ordaz, M. (1987b). Maximum earthquake magnitude from a fault, ASCE Eng. Mech. Div. Speciality Conf., Buffalo, NY. Ruiz, S.E., Esteva, L., Mendoza, E., and Acevedo, W. (1985). Analisis de sismicidad en la central termoeléctrica Lazaro Cardenas, Instituto de Ingenieria, UNAM, internal report. Secretaria General de Obras del DDF (1986). Estudio estadistico de inten- sidad de dafios en el Distrito Federal ocasionados por los sismos de septiembre de 1985. First Int. Symp. Earthquakes and their Effects in Urban Areas. Mexico. Singh, S.K., Bazén, E, and Esteva, L. (1980). Expected earthquake magni- tude froma fault, Bull. Seism. Soc. Am. 70, 903-914 Design Spectra for Mexico's Federal District 291 Structural Engineers Association of California (1986). Lateral force requirements. Seismology Committee, SEAOC. (4986). Tectonic interpretation of the Trans- Suarez, G. and Singh, S.K. 155-160. Mexican Volcanic Belt. Discussion, Tectonophysics, 127, Suarez, M., Sanchez-Sesma, F.J., Bravo, M.A., and Lermo, J. (1987). Carac- teristicas de los depésitos superficiales del valle de México. Memorias VII Congreso Nacional de Ingenieria Sismica, Querétaro, México, A61- ATA. Youngs, R.R., and Coppersmith, K.J. (1985). Implications of fault slip rates and earthquake recurrence models to probabilistic seismic hazard estimates, Bull. Seism. Soc. Am. 75, 939-964.

You might also like