Seismic Safety and Vulnerability Mitigation of School Buildings

Transcription

1 Seismic Safety and Vulnerability Mitigation of School Buildings C.Z. Chrysostomou, N. Kyriakides Department of Civil Engineering and Geomatics, Cyprus University of Technology, PO Box 50329, 3603 Limassol, Cyprus A.J. Kappos Department of Civil Engineering, City University London, UK Department of Civil Engineering, Aristotle University of Thessaloniki, Greece L. Kouris, V. Papanikolaou Department of Civil Engineering, Aristotle University of Thessaloniki, Greece E.G. Dimitrakopoulos, A. I. Giouvanidis Department of Civil and Environmental Engineering, Hong Kong University of Science & Technology E. Georgiou Technical Services of the Ministry of Education and Culture of Cyprus, Nicosia, Cyprus Abstract The vulnerability of existing buildings to seismic forces and their retrofitting is an international problem. The majority of structures in seismic-prone areas worldwide are structures that have been designed either without the consideration of seismic forces, or using previous codes of practice specifying lower levels of seismic forces. In Cyprus, after the three earthquakes that occurred in 1995, 1996, and 1999, the Cyprus State, acting in a pioneering way internationally, has decided the seismic retrofitting of all school buildings, taking into account the sensitivity of the society towards these structures, which house the future generation of the society. In this paper the details of the over 10 year ongoing retrofitting programme of the school buildings of Cyprus are presented and two representative school buildings are chosen for further detailed investigation as far as their retrofitting is concerned. Non-linear analysis is conducted on calibrated analytical models of the two buildings in order to derive fragility curves. These curves are used to perform benefit-cost and life-cycle analysis and evaluate the effectiveness of the retrofitting programme. Keywords: Seismic retrofitting, Schools, Fragility curves, optimum retrofit level 1 INTRODUCTION Cyprus, being located on the boundary of two tectonic plates, the Eurasian and the African ones, has high seismicity. The boundary between the two plates is located in the sea on the south-west of Cyprus and is the source of a large number of earthquakes. Historical reports and archaeological findings show that in the period from 1896 to 2004 Cyprus more than 400 earthquakes 5 of which were of magnitude higher than 5.6 and have caused limited fatalities but severe damage to the building stock. Despite the recorded history of destructive earthquakes, the first seismic design measures in Cyprus were imposed after 1986 and the first seismic design code was introduced on a voluntary basis in 1992 and was made compulsory through a law in On January 1 st 2012, all previous standards were withdrawn and were replaced by the Eurocodes. Therefore, the majority of structures have been designed without any seismic provisions, which increases their vulnerability to seismic loads. Schools, which are a subset of the building inventory, belong to a very sensitive category of buildings. The Cyprus State, acting in a pioneering way internationally, has decided the seismic retrofitting of all school buildings, taking into account the sensitivity of the society towards these structures, which house the future generation of the society. Any loss of life due to

2 their seismic vulnerability will have unbearable consequences. The total number of school buildings in Cyprus is 660. Of these 26 were demolished and replaced by new ones at a cost of about 31 million Euros and 280 were retrofitted at a cost of 140 million Euros. The rest were designed after the enforcement of the seismic codes and thus do not require any intervention. As of today about 90% of the school buildings of Cyprus are seismic resistant. The effectiveness of this programme was evaluated in a research project funded by the Cyprus Government and the European Regional Development Fund. This was done through a pilot application in which non-linear analysis is conducted on the pre- and post-retrofitted buildings to assess their performance to various damage states. Subsequently, benefit-cost and life-cycle analysis was conducted based on local economic loss data to evaluate the effectiveness of retrofit and propose the optimum retrofit level of future strengthening. 2 GUIDELINE FOR THE ASSESSMENT OF THE SAFETY OF SCHOOL BUILDINGS Following the decision for the retrofitting of the school buildings, which was a repercussion of the 1999 earthquakes of Turkey, Athens and Cyprus, a Technical Committee was formed to develop a guideline for the Assessment of the Safety of School Buildings (Chrysostomou et al. 2000). The most complete set of guidelines at the time were the FEMA 273 (1997) document and evidently the concepts and methodologies included formed the basis for the development of the guideline for the retrofitting of the school buildings, the main provisions of which are outlined below. The methodology followed was to establish the existing condition of the structure, select a safety level against which the structure was checked and make a decision for retrofitting measures to be taken, considering at the same time economic and social factors. In case that retrofitting was needed, strengthening measures were taken until the safety level was satisfied. The safety level selected was that of Life Safety. In this level, conforming buildings may suffer extensive damage both in structural and non-structural elements. They may need repairs in order to become operational and the repair may be economically infeasible. The thread to human life in buildings satisfying the requirements of this safety level is low. The associated Peak Ground Accelerations (PGA) for the five seismic zones found in the dated Cyprus hazard map for three design life return periods (10% probability of exceedance) are shown in Table 1. Table 1 Peak Ground Accelerations per Seismic Zone (Cyprus Seismic Code, 1994) Peak Ground Acceleration, PGA Seismic Zone (g) Design Life Design Life Design Life 50 years 20 years 10 years 1,2, For the calculation of the seismic force the importance factor was taken equal to 1.5 while a behaviour factor equal to 2.0 and 1.5 for infilled reinforced concrete frames and load-bearing masonry structures, respectively was assumed. Both the equivalent-static and response spectrum methods are allowed to be used for the analysis of the structures. In both cases, infill walls should be modelled either with two equivalent diagonal bars, or surface finite elements. Nonlinear methods are also allowed to be used. Limits were set on the mean concrete cylinder strength and the number of core samples to be taken, as well as on the yield strength of the reinforcement. Full details on the methodology used is given in Chrysostomou et al. (2012).

3 3 COLLECTION OF DATA FOR THE RETROFITTING OF THE SCHOOLS IN CYPRUS In order to assess the effectiveness of the programme, a data collection was performed, in order to obtain the details of the building inventory and the methods that were used for the retrofitting of schools. This was a very laborious task since to obtain these data one had to look into the records kept in paper form by the Ministry of Education and Culture. It was impossible to check every single folder; therefore the data presented here cover the majority of the schools but not all of them for the period up to the end of April Since vulnerability and cost/benefit analysis will be used for the assessment, a number of parameters were defined and collected during this operation, which will facilitate the assessment. These are: 1) the number of buildings each school consists off, since they may consist of different structural systems and may have different ages, 2) the year of construction, including the starting and finishing date of the construction, which is related to the design code used and consequently the design accelerations, 3) the structural system of each school, 4) the height of each building, 5) the type of intervention, which may include seismic retrofitting, strengthening, or refurbishment (such as changing of tiles, painting, upgrading of electrical and mechanical installations, etc.), or even expansions to cover new needs of the school, 6) method of intervention based on the guideline presented in the previous section, which includes jackets, structural walls, carbon-fibres, steel elements, or mixed methods for RC buildings, and construction of ring beams, replacement of masonry elements, strengthening or replacement of roofs for masonry structures, 7) the acceleration that was specified for the design/retrofitting of the structures, 8) the service life that was used as described in Table 1, 9) the remaining life of the structure, 10) the material properties of the structure before any intervention, 11) the results of chemical analyses, if any, 12) the retrofitting cost that covers only the cost of the intervention on the structural system, which in this case it was very difficult to be identified since it was mixed with the cost of other simultaneous interventions in the school building, and 13) the replacement cost, which is defined as the cost for building a new school. 4 ANALYSIS OF THE DATA As of early 2012, 280 schools have been retrofitted with 20 being in the process of being retrofitted, while another 26 were demolished and replaced. The remaining of the schools were designed after the enforcement of seismic codes. As it was mentioned before, it was not possible to have access to all the folders kept by the Ministry of Education and Culture, therefore, the data presented below refer to the recorded data for 117 schools all over Cyprus. Even for these schools, the 13 parameters mentioned in Section 3 were not always possible to be recorded, since they were missing from the records, therefore the results presented below correspond to a subset of these schools. As far as the construction material is concerned, 68% of the school buildings examined are made of reinforced concrete, 22% is a dual system consisting of reinforced concrete and masonry, and 10% are made of masonry. Regarding the period of construction, 73% were built before 1986 (gravity load design only), 12% between 1986 and 1992 (seismic force was taken as equal to 10% of the base shear force), and 15% after 1992 (modern seismic codes). The limited number of school building designed to modern seismic codes made the need for retrofitting imperative. In 50% of the buildings upgrading, strengthening and maintenance operations were undertaken while only for 8% it was required, in addition to the above, to make expansion of the school and upgrading of the electrical and mechanical installations. Another very important parameter that had to be decided was the selection of the design life (Table 1), which corresponds to the assumed remaining life of the structure after the retrofit and

4 had to be defined in order to choose the appropriate PGA for the analysis of the structure. It was found that 79% of the schools were designed for a 20 year remaining life, 19% for a period of 50 years while only 2% for a period of 10 years. Hence, based on this graph, the majority of the school buildings will need to be re-assessed in about 10 to 20 years from now. The strengthening cost per school for nearly 40% of the schools was less than Euros whereas for 20% of them it was over Euros. The mean value of the strengthening cost was Euros. Regarding the retrofitting methods used in the case of R/C schools, R/C jackets were used on columns (79%), followed by jackets on beams (17%) and the rest of the jackets concerned jacketing of external infill walls. When structural walls were used as a method of retrofitting in the majority of the cases (80%) new ones were constructed while in 16% of the cases existing columns were converted into walls, and in only 4% of the cases external infill walls were partially replaced by R/C walls. Specifically for R/C columns, in 74% of the cases they were strengthened using R/C jacket, while in 14% of the cases FRPs were used. In the rest of the cases either the shear or the column reinforcement was increased. For the case of beams in 31% of the cases R/C jackets were used, while in 50% of the cases FRPs were used, either in the form of wraps (7%) or in the form of strips (14%) with the rest 29% not identified specifically. 5 DESCRIPTION OF ANALYSED BUILDINGS To assess the vulnerability of the school buildings and the efficiency of the retrofitting methods used, two representative school buildings were selected (a two-storey reinforced concrete and a single-story unreinforced masonry (URM) one) to determine their vulnerability and assess the effect of alternative retrofitting methods. This requires sophisticated analyses to be conducted to capture their non-linear behaviour under seismic loading. For increased reliability of the analytical models, these were set-up independently in two different packages ANSYS and SAP2000, both well-known and widely used for both research and design purposes. The models included line (beam-column) and shell elements, as described in the following. The RC building was approximately 200m 2 in plan (20m 10m) with RC frames at 3m spacing providing the resistance in the short direction. In the long direction 2 lines of columns are present connected only through the slab. A skylight extending to a height of approximately 500mm below the slab was left open to enhance the lighting of the building. RC members were modelled using line elements, while slabs were modelled with shell elements. During the retrofitting of the building steel truss members were introduced to strengthen the opening and provide frame action in the long direction as well. Columns are placed in two lines, one on each side of the building in the long direction. The initial dimensions of all columns were More than half of them were increased in area ( ) using RC jackets for retrofitting purposes. The masonry building is of Π-shape in plan and consists of thick load bearing walls (approximately 600mm thick). The compressive capacity of the walls was approximated at about 3MPa based on the type of the stone and test results from similar buildings. The building was modelled using two approaches, a shell element model (Fig. 1-right) and an equivalent frame model; the roof was modelled independently as a truss member. The reliability of the simpler frame model was checked on the basis of the dynamic characteristics of the two models. 6 MODAL ANALYSIS RESULTS Initially the RC building was modelled using ANSYS. Masses were distributed on slabs and reduced secant stiffness (10% of uncracked) was assumed for columns and beams to account for the effect of cracking. The 1 st mode of vibration corresponds to a period of 0.76s whereas the 2 nd mode period is 0.69s. The 1 st mode shape corresponds to vibration of the building along its long

5 direction (Fig. 1-left) whereas the 2 nd mode is along the lateral (short) direction. Fig. 1 Fundamental modes of school buildings: R/C building (left) and URM building (shell elements) The same analysis was repeated on the simulation frame created on SAP2000, which can also be used for non-linear analysis and provides simpler modelling capabilities. The results of the modal analysis for the 1 st and 2 nd modes of vibration gave periods equal to 0.80s and 0.67s, respectively. The comparison of the corresponding results from both simulation frames clearly shows excellent correlation. The masonry building was analysed for different values of masonry strength and ground conditions. Based on comparisons with in-situ measurements (section 7), the best fit was found for the shell-element model assuming low masonry compressive strength (2.85 MPa) and firm ground conditions; in this case the first mode (Fig. 2) period was 0.26 sec and the second mode 0.23 sec. Differences in the periods of the shell and frame elements were small for the lower modes; in general the frame model gave slightly lower periods. It is important to note that, mainly due to the absence of a rigid diaphragm at the roof level, the modes are of local nature, each mode corresponding to substantial deformation of one part of the building only (Fig. 1-right). 7 IN SITU RECORDINGS Results from the in-situ recording on the two selected buildings using an accelerometer network are reported here. In the case of the masonry building, the accelerometers were placed on the lintel over the windows, to monitor the vibration of the stone walls next to the selected windows. These walls were found in the analysis to vibrate at a higher amplitude compared to the rest of the walls at the side of the building. Ambient vibration recordings were conducted the results of which are shown in Fig. 2. The recorded 1 st and 2 nd periods of vibration of the building are 0.26s and 0.23s respectively, coinciding with the analytically estimated values reported in the previous section. In the case of the RC building, 3 accelerometers were used. One was placed near a column of the 1 st floor whereas the remaining 2 were placed on the corner frame, one at each side of the building in the short direction (front and back). The analysis of the ambient vibration recordings was conducted on DASYLab. Figure 3 shows the recorded frequencies of the 1 st (1.28Hz, period 0.78s) and 2 nd (1.50Hz, period 0.67s) modes of vibration of the building.

6 Amplitude (g) H z E E-08 Power Spectrum Density 5.0E E E E E E Time (sec) Fig. 2 Power Spectrum Density in period terms for the masonry building z H0.5 1 z H8.2 1 E0.0 0 E5.2 0 E0.5 0 E5.7 0 E0.0 1 E5.2 1 E0.5 1 E5.7 1 E0.0 2 E5.2 2 E0.5 2 g 0.26 sec 0.23 sec Fig. 3 Fourier Spectrum of the recordings on the RC building Table 2 shows a comparison of the modal analysis results of the two analysed RC frames with the in-situ recorded frequencies. It is obvious that the dynamic characteristics of the analytical models (both RC and un-reinforced masonry) are very close to the ones of the actual building. This validates the analytical models, which can therefore be used for non-linear analysis of the buildings and determination of their vulnerability. Table 2 Comparison of the analytical and recorded dynamic characteristics of the RC building 1 st mode 2 nd mode frequency (Hz) frequency (Hz) ANSYS SAP In-situ recording DERIVATION OF FRAGILITY CURVES In order to assess the performance of the selected buildings, analytical fragility curves were derived and used to perform benefit-cost and life-cycle analysis. These curves represent the probability of exceedance of a certain damage state for a given seismic hazard level. For the selected RC building, the derivation of these curves was conducted using the probabilistic approach to accommodate the variations in the geometric and material properties of

7 P[ds>=dsiPGA] P[ds>=dsiPGA] buildings of the same type. 15 simulation frames were created based on the simulation values in Table 3. It was assumed that the retrofit measures (R/C jackets) were designed using the postretrofit values in Table 3. Time-history analysis was conducted to estimate the structural response. In order to address the uncertainty in seismic hazard each frame was analyzed for 7 different records and the top displacement at each damage state was recorded. The three damage states included in EC8-Part 3 were selected and a fourth was added for the collapse (failure) of the building. Each record was scaled until all 4 damage states were attained by the simulation building. The recorded top displacements (7records 15simulation buildings=105 values per damage state) were transformed to spectral displacement (S d ) values and then to PGA values using the appropriate response spectrum of the area. The average and standard deviation of the PGA values for each damage state were fitted into a logarithmic distribution and fragility curves were drawn for each damage state. To assess the effectiveness of retrofitting, the curves were derived for the pre- and post-retrofit building. The derived curves are shown in Fig. 4 (note the different scale on the horizontal axis). Table 3 Simulation values for the probabilistic study Probabilistic Pre-retrofit Post-retrofit Parameter Average St. Deviation Average St. Deviation f cm f y s l 30Φ 6Φ 40Φ 6Φ DL SD NC Fail DL SD NC Fail PGA (g) PGA (g) Fig. 4. Fragility curves for the pre- (left) and post-retrofit RC building In the case of the masonry building, similar fragility curves were derived using the incremental dynamic analysis procedure up to collapse of the building. Five damage states were used based on the values of yield and ultimate rotations (Kappos et al. 2006) as shown in Fig. 5. The material strength and ground conditions were treated as variables. Due to the aforementioned local character of the modes, which result in localisation of damage in parts of the building, two scenarios were initially used for the definition of the damage states. The first assumed that each damage state was reached when even a single masonry wall attained this damage state, which is the most conservative approach. In the second scenario a damage state is reached when at least 20% of the walls have reached the specific damage state which is a less conservative scenario. The derived curves for the 2 scenarios showed that both scenarios resulted in unrealistic curves for some damage states, and a third scenario was introduced in which at least 10% of the masonry walls reached a certain damage state (an average scenario). As in the case of the R/C building, fragility curves were derived for both the pre- and postretrofit building. The building was retrofitted using two alternative methods: first by introducing an R/C beam (lintel) at the top of the masonry walls at the perimeter of the building in order to connect the walls and provide some degree of diaphragm action at the roof; second, by assuming full diaphragm action (this could be the case if a stiff steel truss was used at the roof level, an

8 intervention not used in the actual building). The derived fragility curves for the retrofitted masonry building for the three scenarios are shown in Fig. 6; it is worth noting the sensitivity of the damage thresholds to the definition of the damage state. Fig. 5 Damage levels for the masonry building 1.0 P(D DS PGA) Direction Χ-Χ DS1 DS1 DS2 DS1 DS3 DS4 DS2 DS2 DS3 DS4 0.5 DS3 0.4 DS4 0.3 Lower limit 0.2 Middle limit 0.1 Upper limit Fig. 6 Fragility curves for the retrofitted masonry building for the three damage state scenarios PGA 9 BENEFIT-COST AND LIFE-CYCLE COST ANALYSES The feasibility of a retrofit/strengthening programme can be properly assessed only if its cost is taken into account and it is evaluated in the light of the benefits resulting in the future due to the strengthening of the buildings. This can be done in a relatively straightforward and quick way using cost-benefit analysis, and in a more systematic (and time-consuming) way by carrying out life-cycle cost analysis; both methods were implemented in the framework of this project. In benefit-cost analysis the ratio of benefit (B) to cost (C) is determined by dividing the present value of the future benefits (reduction in losses due to strengthening) with the cost of carrying out (today) the strengthening. The adopted methodology was that used for Greece by Kappos & Dimitrakopoulos (2008), with the following modifications: 1) The fragility curves that form the basis for calculating damage (and future losses) were those derived in the frame of this project for typical schools in Cyprus (section 8). 2) The economic data introduced in the analysis were those for Cyprus, wherever available. 3) An ad-hoc software (COBE06) was developed (in Excel) for calculating B/C ratios. The aforementioned, tailored to Cyprus buildings, methodology was implemented (utilising the in-house developed software) to carry out several B/C analyses for the different types of school buildings, including a sensitivity analysis to some key parameters like the time frame (or planning horizon ) of the strengthening programme (20 and 50 years), the relationships used for estimating seismic hazard (3 different relationships were used, see Kappos & Dimitrakopoulos, 2008), the discount rate (1.5% and 5%), and, importantly the way human life was accounted for in the estimation of benefits; human losses were estimated using the well-known Coburn & Spence (2002) model. Fig. 7 shows the results of a typical benefit/cost analysis for RC school buildings based on the

9 Cost (million ) hazard relationship that was deemed more appropriate for Cyprus and accounting for the cost of human life ( 500,000). It is clear that in this case retrofit of all types of schools is the appropriate choice, since B/C ratios are well above 1. On the contrary, if the cost of human life is ignored in the analysis, B/C ratios are clearly below 1 and strengthening is not (economically) recommended years 50years Nurseries Primary Secondary Lyceums 0 Nurseries Primary Secondary Lyceums Fig. 7 Benefit/cost ratios for RC (left) and URM buildings In URM schools, application of the light strengthening scheme (lintel) results in negligible B/C ratios (close to 0), as seen in Fig. 7-right; although to a certain extent this is due to the fact that out-of-plane failure through separation of orthogonal walls at their interconnection (a failure mode that is supposed to be prevented by continuous lintels) cannot be captured in the present analysis, it is apparent that the addition of just a top lintel is not a satisfactory scheme. On the contrary, addition of a rigid diaphragm (e.g. steel truss), without substantially increasing the mass of the building (as would be the case if an RC slab was added) was found to lead to B/C ratios are well above 1 when human life was included in the analysis (but, again, close to 0 when neglected). The additional information that can be obtained through a full life-cycle cost analysis is the determination of the optimum level of strengthening, which cannot be estimated from a standard benefit-cost analysis. Various levels of strengthening are considered in this, more involved, method, starting from lighter and less expensive methods and going to the heaviest (and costliest) methods. Fragility curves are derived for each level of strengthening and the total life cycle cost is determined as the sum of the initial cost of strengthening plus the cost of the expected future losses during the lifetime of the buildings. The optimum scheme is then the one that corresponds to the minimum life cycle cost. Details of the method can be found in Kappos & Dimitrakopoulos (2008) W 0.2 w/o % 20% 40% 60% 80% 100% Retrofit Level (%"Full Retrofit") Fig. 8 Life-cycle cost analysis of secondary school buildings (RC) The above methodology was applied to the various types of school buildings in Cyprus

10 Cost (million ) (nurseries, primary, secondary, lyceums). Fig. 8 shows the results for secondary school buildings (RC), for the cases with ( W ) and without ( w/o ) the cost of human life. It is seen that the optimum retrofit level is around 0.50, i.e. 50% of the cost of the heavy jacketing scheme that was described in section 5; again, if the cost of human life is ignored, strengthening is not required. Finally, for masonry buildings, the analysis was found to be very sensitive to the definition of damage states (consistently with what was mentioned previously with regard to B/C ratios); for the conservative definition (Fig. 9), the recommended retrofit level is 100% (full strengthening with a rigid but light diaphragm), whereas for the least conservative definition, the recommended retrofit level is 0 (i.e. no strengthening) W 1.00 w/o % 20% 40% 60% 80% 100% Retrofit Level (%"Full Retrofit") Fig. 9 Life-cycle cost analysis of secondary school buildings (URM, lower bound damage thresholds) AKCNOWLEDGEMENT This project ΑΕΙΦΟΡΙΑ/ΑΣΤΙ/0609(ΒΙΕ)/06 is funded under DESMI of the Research Promotion Foundation of Cyprus and by the Cyprus Government and the European Regional Development Fund. The authors would like to acknowledge also the contribution of Ms. Elpida Georgiou in the collection of data for the school retrofitting programme. REFERENCES ANSYS Academic Research, Release 13.0 Chrysostomou C. Z., Demetriou Th., Georgiou E., and Karydas A. (2000). Safety assessment of school buildings: Goals and targets of retrofitting: Strengthening methodologies and safety upgrading, Technical Manual, Ministry of Education and Culture, Nicosia, Cyprus Chrysostomou C. Z., Kyriakides N., Kappos A. J., Georgiou E., Vasiliou O., and Milis M. (2012). Seismic Retrofitting of School Buildings of Cyprus, Proceedings 15 WCEE, paper 3266, Lisbon, Portugal Coburn, A. & Spence, R. (2002) Earthquake protection (2 nd edition), Wiley, Chichester. Cyprus Seismic Code (1994), Cyprus Association of Civil Engineers and Architects, Nicosia, Cyprus DASYLab, National Instruments, Release 9.0 FEMA 273. NEHERP (1997) Guidelines for the seismic rehabilitation of buildings, Federal emergency Management Agency report no.273, Applied Technology Council, Washington, USA. Kappos, A. J., Panagopoulos, G., Panagiotopoulos, C. & Penelis, G. (2006) A hybrid method for the vulnerability assessment of R/C and URM buildings, Bull. of Earthquake Engineering 4(4), Kappos A.J., and Dimitrakopoulos E.G. (2008): "Feasibility of pre-earthquake strengthening of buildings based on cost-benefit and life-cycle cost analysis, with the aid of fragility curves", Natural Hazards, 45(1), SAP2000, Computers and Structures, Release 14.0