2. Some Retrofit Options for the Seismic Upgrading of Old Low-Rise School Buildings in Mexico Arturo Tena-Colunga, M.EERI An analytical study regarding the seismic upgrading of typical 60's and 70's designs for public school buildings in Mexico is presented. Some schools with these designs were moderately damaged during the 1985 Michoacén Earthquake in Mexico City. The damage was primarily observed in their longitudinal direction where existing slender RC columns have their weak axis, In addition, these columns are confined and shortened by masonry walls that do not run all the story height. These walls are supposed to be non-structural components, however, they experienced shear cracking during the quake due to the distress of the confined columns. Some school buildings were retrofitted after the Michoacan Earthquake adding post-tensioned bracing systems composed of prestressed high-slenderness steel strands (tension-only bracing systems), a retrofit option that is economical. In fact, there is an interest on assessing the effectiveness of this retrofit scheme in other regions with different soil conditions, as for example, the hard soils of the Mexican Pacific Coast. Therefore, the post-tensioning retrofit scheme used for the school buildings in Mexico City was also evaluated for hypothetical locations in the Mexican Pacific Coast. Another option that seems economical for the seismic retrofit of old school buildings in the Mexican Pacific Coast is the use of base isolators. Then, a retrofit plan using lead-rubber bearings was also evaluated. Acceleration time-histories recorded in the Mexican Pacific Coast during the 1985 Michoacan and the 1995 Manzanillo Earthquakes were used to assess the effectiveness of the studied retrofit schemes. Records in Mexico City for the 1985 Michoacén Earthquake and postulated ground motions for a Mg = 8.1 earthquake in Mexico City were also used. The effectiveness of each retrofit scheme is discussed through the comparison of the seismic behavior of original and retrofit structures using a comprehensive set of analyses. INTRODUCTION Typical low-rise school buildings were severely affected in Mexico City during the 1985 Michoacén Earthquake. The most common structural system used nationwide for school buildings before the 1985 earthquake consisted of RC ordinary moment resisting frames. These buildings usually have one bay in the transverse direction and several bays in the longitudinal direction (Fig. 1), having slender columns with their weak axis oriented in the longitudinal direction, which’ makes these schools very flexible in this direction, Infill unreinforced masonry (URM) walls are provided in both directions, but are assumed to be non-structural in the longitudinal direction. In the longitudinal direction, the columns are confined and shortened by these masonry walls that do not run all the story height (Fig. 1) The severe shear damage that the URM walls of many of these schools suffered during the Centro de Investigacion Sismica, Carretera al Ajusco # 203, 14200 México, DF, MEXICO 883 (@arthquake Spectra, Volume 12, No. 4, November 1996884 A-Tena-Colunga 1985 Michoacan Earthquake showed that all walls (longitudinal walls included) participated in the structural response. Because of the damage presented in school buildings in Mexico City during the 1985 earthquake, several techniques have been studied to retrofit these structures. Among them, the use of post-tensioned bracing systems composed of prestressing cables was a popular solution. This system was initially introduced for this purpose in Mexico by a design firm. Previous studies have been carried out to assess the effectiveness of this system in the seismic upgrading of existing school buildings in Mexico City (Miranda, 1990) and in the Mexican Pacific Coast (Miranda and Bertero, 1990). Both studies concluded that this upgrading technique significantly increases the strength and stiffness of these structures, and, if adequately implemented, the system will improve the performance of school buildings during future earthquakes. In addition, Miranda and Bertero warned that, when using this retrofit technique, special attention has to be paid to the following aspects: a) changes in dynamic characteristics of the building and the frequency contents of expected ground motions, and b) verify that the increment in axial loads on the original columns does not endanger them under the expected levels of deformation. These studies have been very valuable, however, they ignored the participation of the original URM walls, assumed to be non-structural, in the seismic response of the structure. Damage presented in these URM walls during the 1985 earthquake suggests that the walls contributed in resisting lateral forces. The URM walls not only modify the strength but also the stiffness and dynamic characteristics of both the original and retrofitted school structures. Recent studies (Teran-Gilmore et al, 1995) have been conducted on the seismic rehabilitation with post-tensioned braces of old Californian frame buildings infilled with URM walls, where a design strategy is presented to make the URM walls contributing to the stiffness and strength effectively, avoiding severe damage to the masonry walls. The present study also addresses the effect of including the masonry walls in the modeling and shows that a post- tensioned bracing system improves substantially the dynamic behavior of the original school buildings, The study confirms the importance of using moderate prestressing forces to insure the elastic behavior of the braces and to inhibit that braces would become slack during the dynamic loading, specially for school buildings in the Mexican Pacific Coast. The use of high prestressing forces so the post-tensioned braces may yield in tension at relatively low drift angles have been recently studied and discussed by Pincheira and Jirsa (1995). Another option that seems economical for the seismic retrofit of old school buildings in the Mexican Pacific Coast is the use of base isolators. Up to now, there are only four base- isolated structures in Mexico, all of them in or near Mexico City in hard-soil conditions: a) a school building and a church using a sliding isolation device composed of steel marbles, b) the press machine for the Reforma newspaper using an isolation system based on pendular action and, c) a highway bridge distributor using rubber bearings. Base isolation is attracting the interest of many Mexican researchers and practicing engineers, but its development has been slow, as most projects have been focused for Mexico City. Mexico City is not the best place for base isolation as the lake-bed zone has soft soil condition with dominant site periods that vary from 1.0 to 3.5s, and important soil settlements, make base isolation unattractive. In addition, the design spectrum for the hard soil conditions in Mexico City has relative low pseudo-accelerations, then, the design of conventional structural systems is economical in these zones, so base isolation seems to offer no substantial advantages. However, base isolation is very attractive for other zones in Mexico classified as high seismic risk regions, for example, most of the Mexican Pacific Coast. There are many school buildings as the ones described in this paper in the Mexican Pacific Coast that need to be seismically upgraded. Base isolation might be a sound and economical technical solution. Therefore, a retrofit plan using lead-rubber bearings was also evaluated for the old school buildings for typical accelerograms recorded during recent strong earthquakes in the Mexican Pacific Coast and inSeismic Upgrading of Old Low-Rise Schoo! Buildings in Mexico 885 Mexico City, The study confirms that base isolation is effective for the retrofit of existing school buildings in hard soil conditions, such as the Mexican Pacific Coast and the hill zone of Mexico City. However, base isolation is inappropriate for the retrofit of structures located in the sof-soils of Mexico City’s lake-bed region. The details of these studies are discussed in following sections. DESCRIPTION OF THE SUBJECT SCHOOL BUILDINGS Plan views and elevations of the school buildings under study are depicted in Figs. | and 2. Both buildings are found in the lake-bed region of Mexico City, in Santa Maria district, within five blocks one from each other. These schools are typical designs of school buildings of the times in Mexico. School EPI (a 1960's design) is composed of two three-story buildings 9m tall (story height of 3m) with typical bay width of 3m in the longitudinal direction (E-W) and 9m in the transverse direction (N-S). Transverse infill walls are provided every three bays and are made of unreinforced hollow concrete masonry units (UHCMU). Longitudinal "non-structural" clay URM walls do not run all the story height as they have window and door openings, shortening the columns (Fig. 1). The main structural system consists of moment frames with 30x50 cm rectangular RC columns oriented in the transverse direction, 25x65 cm (axis A), and 15x65 cm (axis B), 30x65 cm (axes | to 24) rectangular RC beams, and 10 cm thick RC slabs as floor system. Longitudinal reinforcement is the same for all columns and the supplied steel ratio is p=0.034. Longitudinal reinforcement provided at the top and bottom of the beams is symmetric and varies from p=0,0044 for the top story beams to p=0.0088 for the first story beams (P=Piop=Pbottom). No information on the transverse reinforcement details was available @Oe0e8 TTT Building A Building B a) Plan view Building A Building B b) Elevation axis A ‘a_i am mi p= a. Building A ‘Building B c) Elevation axis B Figure 1. Plan view and elevations for school EP!886 AvTena-Colunga School EP2 (a 1976 design) is a four-story building 12.4m tall (story height of 3.1m) with typical bay width of 3.5m in the longitudinal direction (E-W) and 8.7m in the transverse direction (N-S). Transverse infill UHCMU walls are placed every two bays. Transverse outer walls are reinforced concrete walls, Longitudinal clay URM walls do not run all the story height either, and they shorten the columns (Fig. 2). The main structural system consists of moment frames with 25x50 cm rectangular RC columns oriented in the N-S direction, 25x50 cm rectangular RC beams, and 10 cm thick RC slabs as floor system. Longitudinal reinforcement is the same for all columns (p=0.046). Transverse reinforcement consists of 2E#3 @ Scm at the columns ends in a 50 cm length and 2E#3 @ 25 cm in the remaining length. The longitudinal reinforcement supplied at the top and bottom of the beams is symmetric and varies from p=0.0057 for the top story beams to p=0.0095 for the first story beams (p=piop=Pbottom)- Transverse reinforcement consists of 2E#3 @ 10cm at the beams ends in a 50 cm length and 2E#3 @ 20 cm in the remaining distance. e 8) oo | c) Elevation axis B Figure 2. Plan view and elevations for school EP2 For both buildings, the specified strength of the concrete (f.) was 200 kg/em? and the yielding strength of the reinforcement steel (fy) was 4200 kg/cm? The assumed compressiveSeismic Upgrading of Old Low-Rise School Buildings in Mexico 887 strength of the masonry (fq) was 15 kg/cm? for the clay units and 20 kg/cm? for the UHCMU, according to the masonry provisions of the 1995 Mexico's Federal District Code (NTCM-85, 1995). No specific soil-mechanics studies are available, however, the natural period for the Santa Maria site is about 1.3 seconds according to the isoperiod chart contained in the seismic provisions of the 1995 Mexico's Federal District Code (NTCS-95, 1995). GROUND MOTION RECORDS Artificial acceleration records were generated for the Santa Maria site (SANTA) for the 1985 Michoacdn Earthquake. These artificial accelerograms were obtained according to a procedure proposed by Ordaz et al (1992) to generate acceleration records at any site in Mexico City based upon the strong motion data recorded in more than 100 stations in the last seven years and the accelerograms of a reference station (usually CU station, a rock site) for the earthquake of interest. The artificial accelerogram for the E-W direction for the 1985 earthquake and their corresponding response spectra for 5% equivalent viscous damping are presented in Fig, 3. Peak ground accelerations were 0.12g for the N-S direction and 0.19g for the E-W direction, with a representative duration of 70 seconds In addition, typical accelerograms for the Mexican subduction zone recorded during recent earthquakes were selected for the present study. A set of thirteen stations was selected, nine corresponding to sites within the Mexican Pacific Coast and the remaining four to sites in Mexico City. The stations in Mexico City correspond to the Mg=8.1, September 19, 1985 Michoacan earthquake. Two stations are in the lake-bed zone, one of them is the well-known Ministry of Communications station (SCT) and the other one is the station at the pumping facility at Tlahuac (TBOM). The remaining stations in Mexico City correspond to the “transition” zone (near-hard soil conditions) station at Viveros (VIV) and the hill zone station at Tacubaya (TACY). The stations at the Mexican Pacific Coast correspond to different earthquakes, and most of them are found in the Guerrero gap. Three stations (UNIO, CALE and PAPN) correspond to the 1985 Michoacan earthquake (Mg=8.1), three stations (VIGA, MSAS and SMR2) for the Ms=6.9, April 25, 1989 earthquake and two stations (LLAV and PARS) for the M.=6.1, May 30, 1990 ‘earthquake. The remaining record (MANZ) corresponds to the M.=7.6, October 9, 1995 Colima earthquake, and it was recorded at the power plant at Manzanillo. The strongest component of the selected acceleration records and their associated acceleration response spectra for 5% viscous damping are depicted in Fig. 3. RETROFIT PROJECTS WITH POST-TENSIONED CABIES School buildings EP] and EP2 were retrofitted adding post-tensioned bracing systems as depicted in the elevations of Figs. 4 and 5, according to the design of an engineering firm, The cables are connected at the foundation and at the roof only. Special RC mats were built to anchor the cables at the roof and the foundation. The post-iensioned cables used in school EPI are composed of six 1/2" 4 prestressed high-slenderness steel strands protected from weathering with a grouted steel tube. In school EP2, braces are made with eight 1/2" > prestressed high-slenderness steel strands protested with a grouted PVC tube. Yielding strength of the steel strands (f,) is 13200 kg/cm?. The prestress applied to each cable was equivalent to 20% of the yielding strength of the strands for school EP1 and to 21.3% for school EP2. This prestressing level is lower than the range of 33 035% of yield strength reported by Miranda (1990) and Miranda and Bertero (1990). Then, it is of interest to evaluate the effectiveness of such a moderate prestressing force in the cables when subjected to strong ground motions recorded in hard soil conditions, such as the ones recorded along the Mexican Pacific Coast.A.Tena-Colunga 888 Spi00ar uOT}BLa[aD9e pszDaT9g “Eg (8) qorad (8) aWLL (8) doraaad srezto ost oor os 0 srezto eg HO B pm ~y 0°0 = [;-——_____—~s—j_0°0 a Jot 2 ® | s-a ‘svsn we Me & = : yo & oz a P0- = 00 e 3 \ e = -— oo On = ® | s-n ‘vo wixs=a | & a = oz f(s ¥0- el ae 00 . 2 |; ——_—-#j] 00 1 F) or & | S-N ‘zans w& :%S=2 & a yo Sie oz ey W0- a rreeey 0°0 2 85 w FE natalie] org BE ot & ® | w-a ‘nava j we EKS=3 & — ny So 8 wf — 7 ¥0 z ef + + a oo eenrenyngie 00 oT & | a-a ‘aTvO = & [| S-N ‘WOSL estore ee ee oz oe ee _ = af oc 7 a ¢ a. g i ay oo oT 7 [ernest —— & = ® | a-a ‘los j & & = = oz ; g 5 00 z ee a oo & oF | —_+eseietir LS e i ®& [a-a ‘VINVS vo = oz fe tenet eee] v0- oo 7 ¥'0 v0- 00 vo v0- oo v0 v0- oo v0 vo- 00 v0 v0- 00 vO vo- oo vO (3) gov (3) Ta90v (8) Ta9v (2) TaOV (8%) THOV (8) TaOV (9) THOVismic Upgrading of Old Low-Rise School Buildings in Mexico 889 Building A Building B 4) Elevation axis A Building A Building B b) Elevation axis B Figure 4. Retrofit scheme for school EP! i i | a) Elevation axis A ») Elevation axis B Figure 5. Retrofit scheme for school EP2 3-D Elastic Analyses Three-dimensional elastic analyses using ETABS were done to define dominant frequencies and modes of response for different idealizations of the original and upgraded structures with post-tensioned cables, that is, including or not the longitudinal URM walls as structural members. Results are summarized in Table 1. Original schools EP1 and EP2 have short natural periods for the N-S direction because of the contribution of the infill walls in the transverse direction, regarding of the modeling. On the other hand, it can be observed that the natural periods for original schools EP1 and EP2 for the E-W direction are reduced dramatically when the longitudinal walls are included in the modeling, leading the structures away of peak responses for the SANTA site (Fig. 3). If longitudinal walls are considered non-structural’, the structures are very flexible in the E-W direction, having high E-W natural periods because of the reduced moment of inertia of the columns in this direction. If longitudinal walls acted indeed as non-structural, then, school EP2 could have had a resonant response as the natural period for the SANTA site is around 1.3 seconds (Fig. 3)890 A.Tena-Colunga Table 1. Dynamic characteristics of school buildings EPI and EP2 Model Direction] Period ‘Modal Mass (%) 6 Ns Rotation Original EPI-A, NS 0.18 | 74.83 0.00| 12,40 no longitudinal walls E-W 0.70 0.00_| 90.76 0.00 Original EP1-A, NS 017_| 76.06 oor 11.20 with longitudinal walls E-W 032 0.00__| 83.76 0.00 Retrofitted EP1-A (cables), |_N-S 0.17 | 78.16 0.02 952 no longitudinal walls E-W 0.27 0.00 91.10 0.07 Retrofitted EP1-A (cables), N-S 0.17 78.06 0.12 9.67 with longitudinal walls E-W 0.23 0.10_| 90.47 0.00 Original EP1-B, N-S 0.17 87.64 0.00. 0.00. no longitudinal walls E-W 0.68 0.00] 90.60 _| 0.00 Original EP1-B, N-S O17 87.50, O11 0.10 with longitudinal walls E-W 0.38 0.03 90.14 0.17 ‘Retrofitted EP1-B (cables), N-S 0.17 87.92 0.00 0.00, ‘no longitudinal walls E-W 0.26 0.00 90.86 0.00 Retrofitted EPI-B (cables), |__N-S 0.17] 87.93 0.16 0.04 [with longitudinal walls E-W 0.22 0.13 | 90.27 0.01 Original EP2, N-S 0.12 85.60 0.00 0.00 no longitudinal walls E-W 1.23 0.00 88.56 0.00 Original EP2, N-S 0.12 85.54 0.00 0.00 with longitudinal walls E-W 0.63 0.00_|_88.35 0.00 Retrofitted EP2 (cables), N-S 0.12 77.40 0.00 0.00 no longitudinal walls E-W 0.32 0.00__| 88.53 0.00 Retrofitted EP2 (cables), N-S 0.12 77.40, 0.03 0.01 with longitudinal walls E-W 0.28 0.00 88.95 0.00 Translational mode shapes are practically pure for both directions for either idealization, although coupling increases when the longitudinal walls are included in the modeling because walls of axes A and B are not identical (Figs. 1 and 2). The exception is the coupling observed for the N-S mode shapes of building A of school EPI because the distribution of the transverse walls and the location of the stairs lead to additional stiffness and mass eccentricities (Fig. 1). It can be concluded that the contribution of the longitudinal walls helped the structures to move away from resonant response in soft soil conditions such as the ones found at the SANTA site. However, the negative effect is that these walls do shorten the columns and a short-column brittle failure could develop under these conditions. Fortunately, this undesirable mechanism did not develop in these specific structures, but it was observed in other similar building structures in Mexico City during the 1985 Michoacan Earthquake. Natural periods for the N-S direction for schools EP! and EP2 are practically unaffected by the retrofit with the post-tensioned cables (Table 1). On the other hand, it can be observed that the natural periods of schools EP] and EP2 for the E-W direction are reduced dramatically with this retrofit, particularly for the idealizations where the longitudinal walls are not included (Table 1). These retrofit scheme lead structures away of peak responses (Fig. 3) for the SANTA site. The stiffening effect of the post-tensioned cables is reduced when the longitudinal walls are considered "structural" (Table 1). Coupling is reduced in the E-W translational mode shapes with the retrofit when the longitudinal walls are included in theSeismic Upgrading of Old Low-Rise School Buildings in Mexico 891 modeling (Table 1). Therefore, it can be concluded, from the modal response viewpoint, that the use of post-tensioned cables is an effective solution when stiffening is recommended to take a flexible low-rise structure away of resonant responses and/or peak dynamic responses, as happens in soft soil conditions such as the Mexico City’s lake-bed region (ie., stations SANTA, SCT and TBOMB, Fig, 3). The stiffening will also reduce considerably lateral drift deformations. The effectiveness of post-tensioned cables is not clear for hard-soil conditions from the modal response viewpoint, as higher responses are expected in these regions in the short period range (i.e., stations MANZ, UNIO, CALE, PAPN, Fig. 3). Limit Analyses Limit analyses were done to assess lateral load capacities for the different idealizations of the original designs and the retrofitted structures with post-tensioned cables. Several failure mechanisms were studied for each idealization in either direction. Critical failure mechanisms are summarized in Table 2. It can be observed that the critical failure mechanism for all models in the E-W direction is brittle in nature, described by the flexural collapse of the base columns at their top and bottom. However, the base shear capacity associated with this mechanism is relatively high, even when the contribution of the longitudinal walls is neglected, except for the original EP? building where its ultimate strength without the transverse walls is moderate. Longitudinal walls considerably increase the ultimate lateral-load capacity of the schools, however, it is uncertain that these walls would develop all their effective shear capacity as it was assumed in the analyses. Table 2. Failure mechanisms for the different idealization of the school buildings Model Failure Mechanism Dir | vw [Original EPI-A, no long. walls [Base columns E-WJ 0.28 (Original EPI-A, long. walls Base columns E-W]_0.54 Retrofitted EP1-A, no long. walls| Base columns, yielding of cables E-W] 0.94 EP1-A, original and retrofitted Transverse walls overturning N-S | 0.29 (Original EPI-B, no long. walls | Base columns EW] 0.29 (Original EPI-B, long. walls Base columns EW] 038 [Retrofitted EP1-B, no long. walls} Base columns, yielding of cables EW 1.14 EPI-B, original and retrofitted _ J Transverse wails overturning NS [0.29 Original EP2, no long. walls [Base columns EW] 0.17 Original EP2, longitudinal walls _|/Base columns. E-W{ 0.31 [Retrofitted EP2, no long. walls _| Base columns, yielding of cables E-WI_0.76 EP2, original [Transverse wails overturning N-S [0.42 EP2, retrofitted ‘Transverse walls overturnins N-S [0.59 In any case, the ultimate base shear capacities for the original school buildings under study satisfy the requirements of the building codes in effect at the time of construction. Original school EP1 would also satisfy current seismic provisions of Mexico's 1993 Federal District Code (NTCS-95, 1995) for all zones and the 1990 Guerrero State Code (RCGS-90, 1990) for zone C-I. The minimum base shear capacity for this structure should be 0.27W (NTCS-95, zone III and RCGS-90, zone C-1), taking into account its type, location, natural period and a response modification factor of two allowed for this structure. Original school EP2 would only satisfy NTCS-95 and RCGS-90 if the longitudinal walls would be completely effective, as minimum base shears of 0.30W (NTCS-95, zone III) and 0.27W (RCGS-90, zone C-1) are required for this structure. The use of post-tensioned cables considerable increases the strength of the school buildings in the E-W direction, as it was observed before (Miranda,892 ‘A Tena-Colunga 1990, Miranda and Bertero, 1990), although their inclusion does no inhibit the ultimate story column failure mechanism. However, due to the considerable increase in strength, it is likely that the structure would remain in the elastic range when subjected to strong ground shaking. in soft-soil conditions, and might remain elastic when subjected to strong ground shaking in hard-soil conditions along the Mexican Pacific Coast. For the N-S direction, the critical failure mechanism is associated to the overturning strength of the transverse walls, The strength in this direction is also high and satisfies the strong provisions of NTCS-95 for zone III (0.26W) and RCGS-90 for zone C-I (0.27W) for both original schoo! buildings in this direction. Additional strength in the retrofitted school EP2 is because two interior transverse URM walls were replaced by RC walls with reinforcement ratios py=py=0.0030. Nonlinear Dynamic Analyses Two-dimensional nonlinear dynamic analyses were done to estimate deformation and strength demands of the original and retrofitted structures for the E-W SANTA ground motions presented in Fig 3. Detailed studies for all models were conducted only for this record and are reported in detail elsewhere (Alvarez-Ruiz, 1995). The DRAIN-2DX software (Prakash et al, 1992) was used to do these analyses. Frames A of the school buildings (Figs. 4 and 5) were modeled for the idealizations discussed in this study. Gravitational loads due to dead loads and reduced live loads were included in the modeling. Results from these dynamic analyses are summarized in Figs. 6 to 8. The variation of the axial force in the post-tensioned cables with respect to time is depicted in Fig. 6 for the schools under study, whether or not the transverse walls are included in the modeling. For all models, the initial prestressing force is affected by the gravity live loads, but this prestressing is not lost at any stage. Therefore, the lateral stiffness in the retrofitted ‘schools remains unchanged through the duration of the seismic excitation (SANTA). Overall, the variation of the axial load in the cables is smaller when the walls are included in the modeling (buildings EP1-A, EP1-B). However, there can be instances where the variation of the axial loads could be higher when the walls are included, for example, the case of school EP2. This phenomenon is probably caused by the frequency content of the ground motion (SANTA) that might lead the EP2 model with the walls to respond to higher forces and deformations. Peak dynamic story drifts and story shear column indices for the different models of the original and retrofitted buildings EP1-A and EP2 are depicted in Fig. 7. In this figure, OM stands for the original structures including the walls in the modeling, O for the original structures neglecting the contribution of the walls in the modeling, RM for the retrofitted structures with the cables including the walls, and R for the retrofitted structures neglecting the walls. For the drift curves, (a) stands for the allowable RCDF-93 code drift limit for structures that may have non-structural elements separated from the structural elements (8=0.006), and (b) for the allowable RCDF-93 code drift limit where non-structural elements should be separated from the structural elements (6=0.012). It can be concluded from the drift curves depicted in Fig. 7 that the retrofit with the post-tensioned cables is very effective as the maximum dynamic drifts are well below the more strict allowable limit prescribed by RCDF-93. Thus, the structures remained elastic as it was confirmed with further analysis of the data, where no hinging was observed in any structural element at any time-step. For models RM, story drifts (5<0.0012) at the walls levels suggest that they should remain uncracked, On the other hand, it can be observed that the distortions for the original structures (0) are extremely high if the walls are not included in the modeling For these cases, the peak dynamic drifts surpass considerably the code limits and lead one to believe that the structure should have collapsed. This fact was confirmed with the mapping ofSeismic Upgrading of Old Low-Rise School Buildings in Mexico 893 the dynamic hinging where a first-story mechanism was formed at the time of maximum dynamic drift and shear responses for all models under this idealization. Also, it can be observed from Fig. 7 that drifts are considerably reduced when walls are included in the modeling for the original structures. For school EPI-A, the story drifts under this modeling (OM) are within the more strict allowable drift limits of RCDF-93 code, and the size of these deformations (6<0.001) lead one to believe that the masonry walls should have not cracked, However, it can also be observed that the presence of the walls demands higher deformations along the length where the columns are not confined by the walls, evidencing the dangerous side of the short-column effect (Figs. 4 and 7). For school EP2, story drifts are within the maximum allowable limit of RCDF-99 code (6=0.012), but surpass the limit when non- structural elements are not properly separated from the structural ones (6=0.006). It is expected that the masonry walls should have considerably cracked under these levels of deformation. In addition, damage should be expected in non-structural elements, and in the structural elements as well. Columns and beams hinging were recognized for this model in several time-steps. It can be noted from the maximum dynamic story shear indices curves depicted in Fig. 7 that: (1) Original columns are subjected to smaller shear forces for the retrofitted models, and (2) For the models where walls are included (OM and RM), shear forces are generally reduced along the height of the walls and the deformations are low. However, shear forces attracted by the columns along the height where they are free to deform are much higher, evidencing the dangerous side of the short column effect. Some dynamic hinging was observed for the OM idealization in these zones, Also, for the original structures (0), it can be deducted from Fig. 7 that base shear capacities for the modeled frames correlate well with those computed from limit analyses. These frames (A) should carry about half the base shear reported in the E-W direction (Table 2). Some conclusions can be addressed from the peak dynamic story drifts and maximum story shear columns indices curves depicted in Fig. 7: (1) the retrofit with post-tensioned cabies is very effective, leading the structures to remain elastic; and, (2) the contribution of “non-structural” walls may have saved the schools from the structural collapse during the 1985 Michoacan Earthquake, because they lead the structures away from high responses for SANTA site as a consequence of considerably reducing their natural period (see Table 1 and Fig. 3). However, the short-column effect may trigger the collapse of similar structures not properly detailed’ to resist the increased shear forces and deformations, especially when stiffening of the structure is not helpful because of the frequency content of the ground motions of a site. This might be the case of structures located in hard soil conditions Hysteresis curves at the zones where the columns are free to deform for original and retrofitted idealizations of school EP2 are depicted in Fig. 8. It can be confirmed that: (1) the original structure could have responded to high deformation and strength demands that should have lead the structure to collapse (model 0), (2) the contribution of non-structural walls saved the columns from failure (model OM), although some short column yielding was detected at the first story and, (3) the retrofit with the cables leads the structure to a reduced elastic response, hence, illustrating the effectiveness of the retrofit for these structures when subjected to ground motions typical of soft soil conditions, where stiffening of the structure is advised to lead structures away of resonant responses (model RM). Similar results were obtained for schools EP!-A and EP1-B when subjected to the SANTA record. In fact, the effectiveness of the retrofit for soft soil conditions was confirmed with nonlinear dynamic analyses of models R and RM for schools EP1 and EP2 subjected to the SCT and TBOMB records. These analyses suggested that an initial prestressing force of about 20% the yielding strength of the cables was enough to prevent, by a large margin, that the braces would become slack when subjected to typical strong motions of the lake-bed region of Mexico City.894 ATena-Colunga 025 peepee O85 preerprerrprerpeepett EP2, walls, EP2, no walls © 20 [Ni AHO) & 20 Fanti _ iV _ AMAL so os bbl 0.15 0 10 20 30 40 80 60 0 10 20 30 40 50 60 0.25 7 (085 - - » EPL ~A, walls EPi-A, no walls 4 oe | ‘| 0.20 0.20 leah S “li Se NM ois Gelb ole Medd fg 120): 30) 1060 O10 20 90 40 50 60 0.25 pry 0.25 - - 1 EPI-B, ‘walls 1 EP1-B, no walls a: ah : j 0.20 indiana mindeavetnmon a 0.20 , - > nena Mra { > Nie allio og Cbd pag Geb ed 0 10 20 30 40 50 60 0 10 2 30 40 50 60 TIME (s) TIME (s) Figure 6. Variation of the axial load on typical cables SCHOOL EP1 AL 1 Fs |e 2 '- 1 so Bijoj) i'--f i | Fy Mediwistsuundll Ls uf 0.000 0.004 0,008 = _SCHOOL EP2 _ a ae . Bk a it | bo Et op le | Fy lft Coie 0.000 0.025 0.00 © 0.10 0.20 DRIFT, & DRIFT, & V/Wy V/Wy Figure 7. Peak dynamic drifts and column shear indicesSeismic Upgrading of Old Low-Rise School Buildings in Mexico pi0se1 MA-VLNVS ‘2dF [90yos 1OJ seaino sisaiaysAY *g sINTIy so'0- g ‘LaTaa 9 ‘Ladd 9 ‘LaTad 8 ‘Lada 8 ‘Lala af} 00°0 T0°0-10°0 00°0 = t0"0-T0°0 00'0 To‘0-10°0 00°90 §=—- F0"0-T0 00°0 T0'0- 2 - : = = —— | j | j | | i | i | | if | P| | Ivao1D. IN | 2N eN j Nj 7 ~ WH TEGO “2dai 100HOS soo 00"0 $0"0-S0'0 00°0 $0°0-S0"0 0°0 0°0-S0"'0 00°0 $0°0-S0"0 00°0 ‘so'0- a ee oe = — | / | { i Ig [ IN : | en | eN WO TACOW ‘2dd TOOHOS so'0 00°70 §=— $0"0-S 00 00°0 = $0"0-S0"0 00°09 §=— 0"0-S0"0 00°0 = $0"0-S0"0 00°70 = $0'0- | : : | : — as = | Tvaor9) | 1N; | an | 8N, WN: 0 Ta0W ‘edad TOOHOS896 A.Tena-Colunga To assess the effectiveness of the retrofit with post-tensioned cables for typical school buildings in the Mexican Pacific Coast, additional analyses were carried out for retrofitted models "R" of school buildings EPI and EP2 subjected to the following ground motion records: UNIO, CALE, PAPN, VIGA, MSAS, SMR2, LLAV, PARS and MANZ (Fig. 3) The vertical component of the acceleration records were not considered, as they were not available. The initial prestressing force applied to each cable was the same as for previous analyses, this is, 20% for school EP1 and 21.3% for school EP2. The reason of keeping this relatively low prestressing force was to investigate if this level of prestressing was enough to keep the cables in the elastic range, even for strong ground motions typical of stiff soil conditions. It was expected beforehand that much higher axial forces would develop in the post-tensioned cables for structures in hard-soil conditions of the Mexican Pacific Coast, as the natural periods of the retrofitted structures (Table 1) are much closer to the dominant periods of the site (Fig. 3) The acceleration records of stations UNIO and MANZ were critical for the response of schools EP1 and EP2 for hard-soil conditions. The variation of the axial force in the post- tensioned cables with respect to time is depicted and compared in Fig, 9 for schools EP1-A and EP2 when subjected to the following critical records: SANTA (soft soil, Mexico City), UNIO (Fock site, Guerrero state) and MANZ (hard soil, Colima state). It can be observed that although an initial prestressing force of 21.3% is enough to keep the post-tensioned cables of school EP2 of becoming slack for all the considered ground motions, a 20% initial prestressing force is inadequate for preventing the post-tensioned cables of school EP1-A of becoming slack for the MANZ record. In fact, the analyses predict that the cables of school EPI-A would become slack for a 20-second time frame, this is, about 75 complete cycles. Under these conditions, it is likely that the post-tensioned cables may fracture. On the other hand, all columns of schools EP1-A and EP2 remained elastic when subjected to all records (MANZ included), but a mild nonlinear response for the beams was detected for school EP1- A.under the MANZ record. Therefore, a successful retrofit plan with post-tensioned cables for old schoo! buildings in the Mexican Pacific Coast would require a more careful planning of the initial prestressing forces in the cables, and a good estimate of the ultimate capacity of the original beams and columns. For these studies, nominal capacities for RC beams and columns were computed according to the equivalent concrete stress block of RCDF-93 and an elastic- perfectly-plastic behavior of the longitudinal steel reinforcement. Overstrength due to the confinement of the transverse reinforcement and/or "real" behavior of the steel reinforcement was not included because of the lack of enough information on the reinforcing detailing of these elements (embedment lengths, lap splices, etc), particularly school EPL Additional analyses were done to find a successful retrofit plan for school EP1-A to survive gracefully the action of the MANZ record. An initial prestressing force of 32% the yield strength of the cables was selected for this purpose. There was no need to jacket the columns, but local jacket reinforcement at ends of the top story beams was considered to hold the additional stresses transmitted by the post-tensioned cables to the beams. The variation of the axial force in the post-tensioned cables of school EP1-A with respect to time with an initial prestressing force of 32% is depicted in Fig. 10 when subjected to the MANZ record. It can be observed that this amount of prestressing was enough to prevent the cables of becoming slack, although the design is tight. There was no evidence of beam or column hinging at any time-step, and the overall response for the retrofitted school EP1-A was elastic always. However, it seems reasonable to think that a safer but still economical retrofit design with post-tensioned cables for old school buildings in the Mexican Pacific Coast would include RC jackets in the columns that carry additional compressive axial forces due to the prestressing force applied by the post-tensioned cables. Also, some local jacket reinforcement at the beam ends in those beams connected to the joints where the post-tensioned cables are anchored would improve the retrofit scheme.897 ozt oor TF oor ZNVW ‘V-Tda soqqvo yeord4} uo peor [exe oy} Jo UOTFEIIEA (8) SWIL (s) aWIL 08 09 oF “6 eandig | 1 ZNVA ‘eda | Couietia al Lusi OINA ‘V-Tda OINA ‘zda L Vu Seismic Upgrading of Old Low-Rise School Buildings in Mexico oor og 09 oor NVS ‘V-1da so VINVS ‘eda Borba suutuud ro- | oo v0 zo 0 - 80 ro so 90 ro- oo ro zo fo 47 vo vo v0 /d ro- oo ro z0 eo Qt v0 v0 v0 i/d898 A.Tena-Colunga 08 os o4 > os oz on 0.0 oa 1 1 i Creer Ot aan foo a0) 00 tao TIME (a) Figure 10. Variation of the axial load on typical cables for a 32% prestressing force P/P RETROFIT PROJECTS WITH BASE ISOLATORS Base isolation could be an economical retrofit solution for old school buildings in the Mexican Pacific Coast. Base isolation has been extensively used in retrofit projects worldwide before (Kelly, 1993). Then, a retrofit plan with base isolators was also analyzed. Lead-rubber bearings (LRB) were initially designed for hypothetical locations in the Mexican Pacific Coast. The stiffness and strength properties used for the LRB were based upon recommendations available in the literature (Skinner et al, 1993). The design of the isolators was done with 3-D time-history analyses using bidirectional input with the help of the 3D-Basis software (Nagarajaiah et al, 1991). The two horizontal components for the recorded ground motions during the 1985 Michoacan earthquake (stations UNIO, CALE and PAPN) were used for the design of the isolators. According to Clark et al (1993), severe events should be used for the design of the superstructure and isolators, whereas minor events should be used to check that non-structural damage may not occur during low levels of earthquake shaking Base isolators for the school buildings were designed to have an isolator below each column and for an effective isolated natural period greater than 1.5s. For school EPI, a total of 50 circular LRB isolators (36 for building EPI-A and 14 for building EP1-B), 45 cm in diameter and 40 cm in height, with a lead core of 7.7 cm (EP1-A) and 7.2 cm (EP1-B) in diameter were needed for effective isolated natural periods Tjy=Tyy=1-71s. The yield strength of the LRBs was 10 percent of the total weight of the structure (V=0.10W) and their post yield stiffness was equal to 10% of their initial elastic stiffness. The maximum allowable isolator displacements for dynamic stability were Xy=Yq=15 cm. The final design for school EP-2 was almost identical, but here only 26 LRBs with a lead core of 8 cm in diameter were needed. The remaining dimensions and global definition of the LRBs are the same as for school EPI Nonlinear Dynamic Analyses Nonlinear dynamic analyses for the base-isolated projects for schools EP! and EP2 using bidirectional input were conducted using the 3-D Basis software (Nagarajaiah et al, 1991). The fourteen stations depicted in Fig. 3 were considered. Note that in Fig. 3 only the strongest component for each station is plotted, attending to space constraints. Analyses were conducted considering that the E-W components corresponded to the long direction of the buildings and the N-S components to the short direction ("x-y" quake) and vice versa ("y-x" quake). Peak dynamic responses for these base isolation designs when subjected to the considered pairs of acceleration records are summarized in Tables 3 and 4Seismic Upgrading of Old Low-Rise School Buildings in Mexico 899 Table 3 summarizes the results for school building EP1-A. It can be observed that the proposed design is adequate for all the ground motions typical of hard-soil conditions of the Mexican Pacific Coast (stations CALE, PAPN, UNIO, VIGA, MSAS, SMR2, LLAV, PARS and MANZ). The design is tight for the acceleration records associated to severe events (stations CALE, PAPN, UNIO and MANZ), and the epicentral record SMR2 for a moderate earthquake, as the last record has a strong pulse associated to a period range that affects the isolators (Fig. 3). The isolators work for some records related to moderate earthquakes (stations VIGA and LLAV), as the peak dynamic displacement is higher than the yielding design displacement (Xy=¥ y=1.36 cm or Xy/X4~0.090) while in others remained elastic due to the small amplitude ‘of the records (MSAS and PARS). Maximum relative displacements between the roof and the floor level supported by the isolators are low. The proposed design for the isolators works also for the ground motion records typical of hard soil conditions in Mexico City for a severe earthquake (stations VIV and TACY). However, is clear that a more economical and optimal design could be done for these zones as the isolators barely yield and the maximum dynamic base shear is smaller than the global base shear assumed for the initial design. On the other hand, it can be confirmed that base isolation is not a good technical solution for the seismic retrofit of school buildings in the lake-bed zone of Mexico City. For the present study, the dynamic instability of the isolators is quite clear when subjected to representative ground motion records for the lake-bed zone (SCT, TBOMB and SANTA), as marked with an asterisk in Table 3. Base isolation seems unfeasible and unreliable for this zone of Mexico City, because important soil settlements are also present, besides the important dynamic amplifications that base isolators could be subjected, as clearly shown in Table 3. Table 3. Peak dynamic responses for the isolation project, school EP1-A (W=1918 Ton) ‘Max relative roof [ Peak LRB isolator] Base Shears | V, | V, Event] Station | Quake | displacements wrt | displacements (Ton) win isolators (cm) isolators ‘Axmax | A¥max | Xm | ¥¥u | Me | Y 09/19/85] CALE | y-x [| 0.38 | 0.15 | 0.654 | 0.318 | 223.8[ 165.5] 0.117] 0.086 PAPN | y-x | 0.27 | 0.12 | 0.220 | 0.189 | 150.4] 123.2] 0.078] 0.064 UNIO | yx | 043 | 017 [ 0871 | 0401 [2515 rr 0.094 04/25/89] VIGA | y-x | 0.30 | 0.12 | 0.220 | 0.135 159.4] 127.9] 0.083] 0,067 MSAS | x-y 0.17 | 0.08 | 0.064 | 0.060 | 92.8 | 87.2 ]0.048]0.045 SMR2 | y-x 0.39 0.17 [0.918 | 0.729 [234.1] 194.270.122 0.101 05731790] LLAV | xy | 038 | 0.14 | 0.385 | 0.292 [1975] 151.2[0.103]0.079 PARS | y-x | 0.09 | 003 | 0.027 | 0.020 | 47.1 | 31.7 [0.025] 0.017 10/09/95 | MANZ | x-y 0.42 0.19 } 0.951 | 0.709 [237.9] 195.1] 0.124] 0.102 09/19/85 | SCT xy 1.37 0.37 | 4.611* | 2.270* [806.0 | 378.9] 0.420] 0.198 TBOM | y-x 0.69 0.29 [/2.079* | 1.413* [413.3 | 325.77 0.215 [0.170 VIV xy 0.23 0.10 [0.149 | 0.112 [133.4] 112.5]0.070] 0.059 TACY | y-x 0.22 0.10 | 0.136 | 0.088 | 127.9] 108.7] 0.067 | 0.057. M,=8.1 [SANTA] x-y 4.23 2.26 | 16.51* | 16.40* } 2463. | 2470. [1.284] 1.287 * Dynamic instability, as the maximum isolator design displacement is exceeded900 A Tena-Colunga Table 4 summarizes the peak dynamic responses for the isolation project of school building EP2. Similar observations as the ones discussed for Table 3 can be drawn for Table 4 It can be concluded that base isolation is a sound technical solution for the seismic upgrading of old school buildings in the Mexican Pacific Coast. However, the application of base isolation for this purpose will depend in economical factors as well, as it was shown that the use of post-tensioned cables can be a sound technical solution too, and this retrofit system is economical. Table 4, Peak dynamic responses for the isolation project, school EP2 (W=1530 Ton) Max relative roof [Peak LRB isolator] Base Shears | V, | Vy Event | Station | Quake | displacements wrt] displacements on) | Hy | isolators (em) a Avmax | Xm | ¥¥u | Ye 1% 09/19/85] CALE | y-x | 1.58 | 0.06 | 0.621 | 0.342 [184.2] 131.1]0.120] 0.086 PAPN | y-x | 1.15 | 0.05 | 0.208 | 0.170 |119.2/139.5|0.078 [0.091 UNIO | yx | 186 | 006 | 0.902 | 0.444 |206.7| 140.1] 0.135] 0.092 04725789] VIGA | y-x | 1.51 | 0.04 | 0.192 | 0.119 [154.3] 90.6 | 0.101] 0.059 MSAS | xy | 0.75 | 0.03 | 0.071 | 0.062 | 79.9 | 67.2 [0.052] 0.040 SMR2 {yx | 1.53 | 0.06 | 0.894 | 0.743 [188.7] 153.0] 0.123] 0.100 05/31/90] LLAV | x-y | 1.74 | 0.06 | 0.363 | 0.305 [135.4] 121.3] 0.088[ 0.079 PARS | y-x | 0.42 | 0.01 | 0.030 | 0.028 | 42.3 | 32.1 [0.028] 0.021 10/9795] MANZ | x-y | 1.62 | 0.07 | 0.524 | 0.679 |189.2| 138.6) 0.124] 0.091 ‘Oss | SCT | xy | 454 | 0.14 | 4.090" | 2.045" [504.7 | 253.1] 0.3691 0.165 TBOM | y-x | 271 | 0.10 | 1.995*| 1301" [326.4] 243.4] 0.213] 0.159 viv | xy | 105 | 0.04 [0.138 | 0.109 [1167] 91.2 | 0.076] 0.060 TACY | yx | 0.73 | 0.04 | 0.090 | 0.094 | 87.2 | 84.4 [0.057] 0.055 pT PSANTAT y-x_[ 15.98 | 0.78 | 16.89*] 13.03" | 1553. [1.314] 1.015 * Dynamic instability, as the maximum isolator design displacement is exceeded SUMMARY AND CONCLUSIONS An analytical study on the seismic upgrading of existing three-story and four-story school buildings at Mexico City using post-tensioned bracing systems or base isolation was presented. The effectiveness of each retrofit scheme for different soil conditions and regions of high seismic risk of Mexico was discussed, For the retrofit plan using post-tensioned cables, the influence of "non-structural" URM walls was discussed through the comparison of the seismic behavior of original and retrofit structures using a comprehensive set of analyses. The results of these analyses permit one to conclude the following: (1) The masonry walls (commonly assumed as non-structural) were responsible for the survival of the school buildings located at the Santa Maria district during the 1985 earthquake, as they lead the structures to shorter natural periods avoiding resonance with the SANTA site, in addition of increasing their lateral strength. This was also possible because the short columnSeismic Upgrading of Old Low-Rise School Buildings in Mexico 901 effect was not triggered perhaps because of an appropriate shear confinement provided in that region. Miranda and Bertero (1990) reported good confinements for the columns of this class of buildings before, so this assumption for school building EP1 might be also correct. Nevertheless, for retrofit projects using post-tensioned braces, separating properly the transverse URM walls from the columns seems optimal, so they will no longer participate in the response of the retrofitted structure, as they are not longer needed. (2) The post-tensioning retrofit method improves substantially the dynamic behavior of old schoo! buildings, particularly when they are found in soft-soil conditions such as Mexico City’s lake-bed region. For these zones, an initial prestressing force near 20% the yielding strength of the cables is safe enough to prevent the cables of becoming slack and to avotd nonlinear response of the original columns and beams, For the hard-soil conditions of the Mexican Pacific Coast, the study shows that the use of post-tensioned braces could be effective but a more careful design of the initial prestressing forces is needed to insure the elastic behavior of the braces, to inhibit that the braces would become slack during the dynamic loading, and to avoid overstressing the original elements. For this region, an initial prestressing force of 30% to 35% the yielding strength of the cables seems more reasonable to insure a correct work of the cables when subjected to strong ground shaking. This result correlates well with the study of Miranda and Bertero (1990). However, these levels of initial prestressing have to be further evaluated by including in the analyses the influence of the vertical component of the considered acceleration records, which may require even higher initial prestressing in buildings located near an active fault. The present study suggests that there was no need to reinforce the original columns. However, a safer but still economical retrofit design with post-tensioned cables for old school buildings in the Mexican Pacific Coast would include RC jackets in the columns that carry the additional compressive axial forces due to the prestressing force applied by the post-tensioned cables. In addition, the retrofit scheme would be complete by including some local jacket reinforcement at the beam ends in those beams connected to the joints where the post-tensioned cables are anchored. (3) Base isolation is very effective for the retrofit of existing school buildings in hard soil conditions, such as the Mexican Pacific Coast and the hill zone of Mexico City. However, base isolation is inappropriate for the retrofit of structures in the soft-soils of Mexico City's lake-bed region. The use of base isolation for retrofit projects of old school buildings in the Mexican Pacific Coast will depend on economic criteria, besides careful designs for the isolators taking into account the local seismicity of the Mexican Pacific Coast. The use of static design procedures for base isolation projects in the Mexican Pacific Coast is currently under study, based upon the current design spectra for Guerrero state (RCGS-90) and the procedure proposed in the 1994 UBC code. It has been preliminary found that the parameters proposed by the 1994 UBC code to define the base isolation design displacement (D) needs to be modified to achieve reliable and safe designs of base isolators using the static procedure. Therefore, for the moment, base isolation design projects for the Mexican Pacific Coast should be based upon dynamic design procedures until some design recommendations for the design of base isolated structures will be incorporated in Mexican design codes. ACKNOWLEDGMENTS Financial support of Secretaria General de Obras del Departamento del Distrito Federal, Mexico City, Mexico is acknowledged. Appreciation is extended to José Luis Alvarez-Ruiz for computing some results and drawing some figures presented in this paper, as a part of his BSS. thesis work directed by the author, Eduardo Pérez-Rocha is thanked for generating the artificial acceleration records for the Santa Maria site, Prof. Enrique Del Valle and anonymous reviewers are thanked for their critical review of the original manuscript.902 A.Tena-Colunga APPENDIX I: CONVERSION TO SI UNITS conv. To Multiply by Ton KN 9.81 kg/om2 Pa 0.0981 REFERENCES [1] Alvarez-Ruiz, J. L. (1995), "Estudio Analitico Sobre el Comportamiento Sismico de la Reestructuracién de Planteles Escolares por Medio de Cables de Acero de Presfuerzo," B.S. Thesis, Facultad de Ingenieria, UNAM, September. (In Spanish) (2] Clark, P. W., A. S. Whittaker, I, D. Aiken and J. A. Egan (1993), "Performance Considerations for Isolation Systems in Regions of High Seismicity,” Proceedings of ATC 17-1 Seminar on Seismic Isolation, Passive Energy Dissipation, and Active Control, San Francisco, California, pp. 29-40. Kelly, J. M. (1993), "State-of-the-Art and State-of-the-Practice in Base Isolation," Proceedings of ATC 17-1 Seminar on Seismic Isolation, Passive Energy Dissipation, and Active Control, San Francisco, California, pp. 9-28. Miranda, E. (1990), "Upgrading of a School ‘Building in Mexico City," Proceedings, Fourth US National Conference on Earthquake Engineering, Palm Springs, California, May, Vol. 1, pp. 109-118. Miranda, E. and V. V. Bertero (1990), "Post-Tensioning Technique for Seismic Upgrading of Existing Low-Rise Buildings," Proceedings, Fourth US National Conference on Earthquake Engineering, Palm Springs, California, May, Vol. 3, pp. 393- BI 4 15 6 Nagarajaiah, S., A. M. Reinhorn and M. C. Constantinou (1991a), "3D-Basis: Nonlinear Dynamic Analysis of Three-Dimensional Base Isolated Structures: Part IT," Technical Report NCEER-91-0005, National Center for Earthquake Engineering, State University of New York at Buffalo. NTCM-95 (1995), "Normas Técnicas Complementarias para Disefio y Construccién de Estructuras de Mamposteria" Gaceta Oficial del Departamento. del Distrito Federal, México, February. (In Spanish) [8] NTCS-95 (1995), "Normas Técnicas Complementarias para Disefio por Sismo," Gaceta Oficial del Departamento del Distrito Federal, México, February. (In Spanish) [9] Ordaz, M., R. Meli, C. Montoya-Dulché, L. Sanchez and L. E. Pérez-Rocha (1992), "Data Base for Seismic Risk Assessment in Mexico City," Proceedings, Simposio Internacional Sobre Prevencion de Desastres Sismicos, CENAPRED, México. [10} Pincheira, J. A. and J. O. Jirsa (1995), "Seismic Response of RC Frames Retrofitted with Steel Braces or Walls," ASCE Journal of Structural Engineering, Vol. 121, No. 8, pp. 1225-1235. [11] Prakash, V., G. H. Powell and F. C. Filippou (1992), "DRAIN-2DX: Base Program User Guide," Report No UCB/SEMM-92/29, Department of Civil Engineering, University of California at Berkeley, December. [12] RCGS-90 (1990), "Reglamento de Construcciones del Estado de Guerrero, Disefio por Sismo," México. (In Spanish) [13] Skinner, R. I., W. H. Robinson and G, H. Mc Verry (1993), An Introduction to Seismic Isolation, First Edition, John Wiley and Sons, England. [14] Terén-Gilmore, A., V. V. Bertero and N. Youssef (1995), "Seismic Rehabilitation of Framed Buildings Infilled with Unreinforced Masonry Walls Using Post-Tensioned Steel Braces," Report No UCB/EERC-95/06, Earthquake Engineering Research Center, University of California at Berkeley, June. (7