Seismic Vulnerability and Loss Assessment for Buildings in Greece
Summary (5 min read)
1 Introduction
- The numerical studies involved in the development of the aforementioned ‘hybrid’ fragility curves included modelling and analysis of a large number of building types, representing most of the common typologies in S. Europe.
- The R/C building models were analysed for a set of carefully selected accelerograms representative of different ground conditions.
- Several earthquake damage (and loss) scenario studies appeared wherein some of the most advanced techniques have been applied to the urban habitat of European cities (Barbat et al.
2.1 Buildings Analysed
- Using the procedures described in the following, analysis of several different R/C building configurations has been performed, representing practically all common R/C building types in Greece and several other S. European countries.
- Each of the above buildings was assumed to have three different configurations, ‘bare’ (without masonry infill walls, RC1 type), ‘regularly infilled’ (RC3.1) and ‘irregularly infilled’ (soft ground storey, usually pilotis, RC3.2 type).
- R/C buildings with medium level of seismic design (roughly corresponding to post-1980 codes in S. Europe, e.g. the 1985 Supplementary Clauses of the Greek Seismic Codes) and reasonable seismic detailing of R/C members; e.g. RC3.1HM (high-rise, moderate code), also known as - Moderate code.
- Even if this is not necessarily the case in all cities, differentiation between RCixN and RCixL, as well as between RCixM and RCixH is difficult, and judgement and/or code-type approaches are used to this effect.
2.2 Inelastic analysis procedure
- For all Low, Moderate, and High code R/C buildings inelastic static and dynamic time-history analyses were carried out using the SAP2000N (Computers & Structures 2002) and the inhouse software DRAIN2000, respectively.
- In total 72 structures were addressed in this study, but full analyses were carried out for 54 of them (N and L buildings were initially considered together, as discussed previously, but different pushover curves were finally drawn, see section 2.3).
- To keep the cost of analysis within reasonable limits, all buildings were analysed as 2D structures.
- Using the DRAIN2000 code, inelastic dynamic time-history analyses were carried out for each building type and for records scaled to several PGA values, until ‘failure’ was detected.
2.3 Estimation of economic loss using inelastic dynamic analysis
- From each analysis, the cost of repair (which is less than or equal to the replacement cost) is estimated for the building type analysed, using the models for member damage indices proposed by Kappos et al. (1998b).
- Due to the fact that the cost of the R/C structural system and the infills totals less than 40% of the cost of a (new) building, the above relationships give values up to 38% for the loss index L, wherein replacement cost refers to the entire building.
- In R/C frame structures (RC1 and RC3 typology), failure is assumed to occur (and then L=1) whenever either 50% or more of the columns in a storey ‘fail’ (i.e. their plastic rotation capacity is less than the corresponding demand calculated from the inelastic analysis), or the interstorey drift exceeds a value of 4% at any storey (Dymiotis et al. 1999).
- This set of failure criteria was proposed by Kappos et al. (2006); they resulted after evaluating a large number of inelastic time-history analyses.
2.4 Development of pushover and capacity curves
- A resistance curve (wherein resistance encompasses both strength and ductility), also called curve, is a plot of a building’s lateral load resistance as a function of a characteristic lateral displacement (typically a base shear vs. top displacement curve) derived from inelastic static analysis.
- Some typical pushover curves and their corresponding bilinear versions (derived on the basis of equal areas under the curves) are given in Figure 3; as shown in the figure, the equal areas are calculated up to the point where the first significant drop in strength (usually about 20%) occurs in the ‘complete’ pushover curve.
- Some example curves were shown in figure 3 for R/C frame buildings designed to old codes (L); shown in the figure are (from top to bottom) the cases of infilled, pilotis and bare building, respectively.
- It is noted, though, that a ‘global type’ analysis that cannot fully capture local failure to R/C members due to interaction with infill walls, in principle can not yield a reliable ultimate displacement for the structure; more work is clearly needed in this direction.
2.5 Derivation of fragility curves
- One possibility for deriving probabilistic vulnerability curves is in terms of macroseismic intensity (I) or PGA; it is recalled herein that as long as a certain empirical relationship between I and PGA is adopted, the two forms of curves (in terms of I or PGA) are exactly equivalent.
- In the common case that Lact is available at one or very few points the scheme should be properly adapted by the analyst.
- It is worth noting that the ratios Lact/Lanl calculated for the Thessaloniki 1978 data were reasonably close to 1.0 when the entire building stock was considered, but discrepancies for some individual building classes did exist (Kappos et al., 1998b).
- Different sets of fragility curves are plotted in Fig. 5 (full and dotted lines), the difference lying in the way empirical data were introduced (cf. w1, w2 factors in equation 4).
2.6 Fragility curves in terms of Sd
- The aforementioned fragility curves in terms of PGA were also used to derive additional curves, this time in terms of Sd, necessary for fragility assessment using the HAZUS approach (FEMA-NIBS 2003).
- The procedure adopted was to transform the median PGA values to corresponding median.
- Sd values, using an appropriate spectrum and either the fundamental period of the ‘prototype’ building, assuming that the equal displacement rule applies, or using the capacity spectrum approach (for short period buildings).
- It is noted that the convenient equal displacement approximation (inelastic displacement demand approximately equal to elastic demand) is a valid assumption for medium-rise and high-rise buildings, but usually a crude one for low-rise buildings.
- For the present application of the methodology it was decided to use the mean spectrum of the microzonation study of Thessaloniki (Anastasiadis et al., 2001) since the derived Sd-based fragility curves were primarily intended to be used for the Thessaloniki risk scenario (Pitilakis et al. 2004).
3.1 Overview of the methodology adopted
- For URM buildings, apart from the Thessaloniki 1978 earthquake data (used for R/C structures, see section 2), the database from the Aegion 1995 event (Fardis et al. 1999) was also utilised.
- The first step for the utilisation of these two databases was the assignment of an appropriate intensity (or corresponding PGA) for the area they refer to.
- These databases were used for the simple, purely statistical, procedure described in section 3.2, and were extrapolated to lower and higher events using nonlinear analysis in the hybrid approach described in sections 3.3 and 3.4.
3.2 Purely empirical approach
- A purely empirical approach (similar to that used by other researchers, e.g. Spence et al. 1992, Lagomarsino & Giovinazzi 2006), was first adopted by the authors for deriving fragility curves in terms of intensity for URM buildings.
- The database includes a total of 5740 buildings, 1780 of which (31%) are unreinforced masonry ones, and most of the remaining buildings are reinforced concrete ones.
- Details of the processing of the database are given in Penelis et al. (2002), where the reasons are discussed why economic damage indices (ratio of repair cost to replacement cost) and post-earthquake tagging of buildings (‘green’-‘yellow’-‘red’) had to be combined in interpreting the Thessaloniki data.
- Empirical curves were first derived using the aforementioned databases and an exponential type of statistical model and they are reported in Kappos et al (2006); albeit useful, they are not deemed as sufficiently reliable, since data for only two intensities were available.
3.3 Nonlinear analysis and capacity curves
- It is well known that the nonlinear response of unreinforced masonry (URM) buildings is not easy to model, mainly because the frame element (beam-column) commonly used in the case of R/C buildings is generally not amenable to modelling URM buildings.
- The difficulties are increased in the case of dynamic analysis where the inertia forces should not be concentrated at the diaphragm levels (which is the rule for R/C buildings).
- These two main categories are further subdivided into single-storey, two-storey and three-storey buildings.
- Using the same procedure as for R/C structures (section 2.4), capacity curves have been derived for one, two, and three storey URM buildings, belonging to the types M1.2 (‘simple stone’ URM buildings) and M3.4 (URM buildings with R/C floors).
- The corresponding parameters for these curves are given in Table 5.
3.4 Hybrid fragility curves
- The hybrid methodology described in previous sections was used to calculate vulnerability curves for URM buildings in terms of spectral displacement.
- Instead of using semi-empirical interstorey drift values (the HAZUS approach), the AUTh group (Kappos 2001, Kappos et al. 2006) has suggested expressing the damage state thresholds in terms of the basic parameters of the capacity curve (yield displacement and ultimate displacement, both referring to a bilinearised capacity curve); this proposal is shown in Table 6.
- The aforementioned procedure cannot be directly integrated within the hybrid approach.
- Using the hybrid procedure, damage histograms were constructed for the URM building classes of interest; among these histograms, the ones corresponding to the Sd values assigned to the Thessaloniki and Aegion earthquakes consisted of actual loss values, while the rest were derived by the scaling procedure described previously.
4 Region-specific fragility curves
- Hence, a critical step in applying such curves to a specific study is to make them region-specific, i.e. dependent on the characteristics of the representative ground motions in the cities studied, which can be quite different from those used for deriving the PGA-based curves (and also Sdbased hybrid curves that involve assuming a specific spectral shape, see section 3.4).
- The mean acceleration spectrum of the 16 records of Fig. 2, normalised to a PGA of 1.0g, is illustrated in Fig. 12, together with the mean spectra derived from the Grevena and Düzce microzonation studies (Pitilakis et al. 2010) and the Greek and Turkish Code design spectra for soil types that are typical for the two cities.
- This approach is quite general but very convenient for deriving region- or site-specific analytical fragility curves for a building stock in a specific area (regardless of whether the appropriate ‘target’ spectrum is defined from a microzonation study or a seismic code).
5 Development of earthquake scenarios
- Two types of scenarios can be developed using the analytical tools presented in the previous sections.
- The fragility models developed by the AUTh group originate from repair cost considerations, hence it was relatively straightforward to use them for economic loss assessment purposes.
6 Concluding remarks
- This chapter has tackled a number of issues relating to vulnerability and loss assessment, with particular emphasis on the situation in Greece and S. Europe.
- It is noted, though, that values of the variabilities proposed in HAZUS should not be adopted blindly if the analytical procedure used is not the one based on the ‘capacity spectrum’.
- Of particular practical relevance is the simple procedure suggested in section 4, based on the area under pseudovelocity spectra, for adapting fragility curve sets developed for a specific ground motion (be it a spectrum or a set of accelerograms) to the ground motion that is (more) representative of seismic hazard in another geographical area.
8 References
- Vulnerability assessment and earthquake damage scenarios of the building stock of Potenza (Southern Italy) using italian and greek methodologies.
- Pitilakis, K. et al. (2011) Development of comprehensive earthquake loss scenarios for a Greek and a Turkish city: Seismic hazard, Geotechnical, and Lifeline Aspects.
Did you find this useful? Give us your feedback
Citations
29 citations
Cites background from "Seismic Vulnerability and Loss Asse..."
...HAZUS sets the collapse threshold at a maximum interstory drift of 0.0500 and 0.0333 for C2L and C2M buildings, respectively, while Kappos [40] at 0.02 for reinforced concrete buildings in Greece, which demonstrates that generic thresholds may not be suitable....
[...]
...0333 for C2L and C2M buildings, respectively, while Kappos [40] at 0....
[...]
17 citations
Cites background from "Seismic Vulnerability and Loss Asse..."
...The GMPE is a model proposed by Ambraseys et al. [60], which has been used to assess the seismic vulnerability of RC buildings in Greece [61]....
[...]
...The earthquake scenarios, aftershock sequences, and building models are presented in the second section of this article and are based on the test site of Thessaloniki (Greece)....
[...]
...The models are considered representative of three-dimensional structural systems in Greece and are used to estimate losses [46]....
[...]
...This value of the inter-story drift corresponds to the collapse threshold for RC buildings in Greece [55]....
[...]
...Silva et al. [73] proposes 5 · 10−5 and 10−4 as tolerable annual rates of collapse leading to loss of human life for low and moderate earthquake-prone countries, such as Greece (10−4 )....
[...]
8 citations
Cites background from "Seismic Vulnerability and Loss Asse..."
...More recently, methodologies for vulnerability (fragility) curves have also been applied to estimate seismic damage and the direct economic loss of buildings at community and municipal levels [38,39]....
[...]
Cites background from "Seismic Vulnerability and Loss Asse..."
...0333 for C2M buildings, while Kappos [29] at 0....
[...]
...HAZUS sets the collapse threshold at a maximum interstorey drift of 0.0333 for C2M buildings, while Kappos [29] at 0.02 for buildings in Greece....
[...]
References
615 citations
"Seismic Vulnerability and Loss Asse..." refers methods in this paper
...It is also different from the empirical approach used by other researchers within the RISK-UE project (Lagomarsino & Giovinazzi 2006)....
[...]
...A more refined procedure based on the vulnerability index method (Lagomarsino & Giovinazzi 2006) was also used....
[...]
...A second interpretation of the available data using the vulnerability index approach (Lagomarsino & Giovinazzi 2006), re-assigning the intensities of Thessaloniki and Aegion to 6.5 and 7, respectively (based on comparisons with the Italian data), and finally using beta distributions for the…...
[...]
...According to the RISK-UE building typology matrix (Lagomarsino & Giovinazzi 2006), single-storey and two-storey buildings of the same material (stone or brick) should be grouped into a single category (M1....
[...]
...A purely empirical approach (similar to that used by other researchers, e.g. Spence et al. 1992, Lagomarsino & Giovinazzi 2006), was first adopted by the authors for deriving fragility curves in terms of intensity for URM buildings....
[...]
302 citations
"Seismic Vulnerability and Loss Asse..." refers background or methods in this paper
...3) are such as to warrant lumping them in one single class (RC4) for vulnerability assessment purposes (Kappos et al. 2006)....
[...]
...Instead of using semi-empirical interstorey drift values (the HAZUS approach), the AUTh group (Kappos 2001, Kappos et al. 2006) has suggested expressing the damage state thresholds in terms of the basic parameters of the capacity curve (yield displacement and ultimate displacement, both referring…...
[...]
...This set of failure criteria was proposed by Kappos et al. (2006); they resulted after evaluating a large number of inelastic time-history analyses....
[...]
...…specific information regarding the type of masonry (stone or brick), therefore the assumption that all URM buildings constructed before 1940 were stone masonry and all the rest brick masonry, was adopted, based on historical evidence on types of masonry construction in Greece (Kappos et al. 2006)....
[...]
...Empirical curves were first derived using the aforementioned databases and an exponential type of statistical model and they are reported in Kappos et al (2006); albeit useful, they are not deemed as sufficiently reliable, since data for only two intensities were available....
[...]
218 citations
155 citations
150 citations
"Seismic Vulnerability and Loss Asse..." refers background in this paper
...…damage (and loss) scenario studies appeared wherein some of the most advanced techniques have been applied to the urban habitat of European cities (Barbat et al. 1996, Bard et al. 1995, D’Ayala et al. 1996, Dolce et al. 2006, Erdik et al. 2003, Faccioli et al. 1999, Kappos et al. 2002, 2008, 2010)....
[...]
Related Papers (5)
Frequently Asked Questions (9)
Q2. What have the authors stated for future works in "Seismic vulnerability and loss assessment for buildings in greece" ?
A classification scheme that is deemed appropriate for the building stock in this area has been proposed, aiming at an adequate description of the R/C buildings that currently dominate the built volume, without neglecting the case of URM buildings, which due to their higher vulnerability are often an important contributor to the future losses. The Sd-based curves take into account the spectral characteristics of the motion but further research is needed in several points such as the case where the capacity spectrum method does not result in a solution, or the equal displacement assumption is not valid. The type of assumption made for the functional form of the fragility curve is also a key one, but the current trend world-wide seems to be towards adopting the lognormal cumulative distribution function ; the determination of damage medians and the variabilities associated with each damage state can be done using the procedures described in HAZUS, or the alternative ones suggested herein. Regarding the two different types of fragility curves that can be used, PGA-based curves offer a number of advantages, but also ignore, to an extent that depends on the spectral characteristics of the motions considered for deriving the fragility curves and their relationship to the characteristics of the scenario motions, the possibly lower damageability of motions with high PGA and spectra peaking over a very narrow band and/or with very short duration ( both these characteristics are more or less typical in strong motions recorded in Greece ).
Q3. What is the key feature of the recent among these studies?
A key feature of the most recent among these studies, including the one presented here for Thessaloniki, is the use of advanced GIS tools that permit clear representation of the expected distribution of damage in the studied area and visualisation of the effects of any risk mitigation strategy that can be adopted on the basis of the scenario.
Q4. What is the key idea of the hybrid approach to seismic vulnerability assessment?
It is within the scope of the work envisaged by the AUTh research group to improve the methodologies for assessing the vulnerability of both common and monumental structures, using damage information from past earthquakes in combination with nonlinear analysis of carefully selected representative structures.
Q5. What is the common type of URM building in Thessaloniki?
The curves presented herein refer mainly to simple stone masonry and brick masonry buildings, with sufficiently stiff floors to provide diaphragm action, such as reinforced concrete floor slabs or vaulted floors, which are by far the most common URM building types in Thessaloniki, as well as in the rest of Greek cities (see also Penelis et al. 2002).
Q6. Why is the pushover curve used for the vulnerability assessment a necessity?
It is emphasised that due to the fact that the pushover curves used for the vulnerability assessment are bilinear versions of the actually calculated curves (see Fig. 3), a necessity arising from the fact that bilinear behaviour is considered in reducing the elastic spectrum to an inelastic one (or an equivalent elastic one for effective damping compatible with the energy dissipated by the inelastic system), the ‘ultimate’ capacity generally does not coincide with the actual peak strength recorded duringthe analysis.
Q7. What was the first step for the utilisation of these two databases?
The first step for the utilisation of these two databases was the assignment of an appropriate intensity (or corresponding PGA) for the area they refer to.
Q8. What is the key idea of AUTh’s hybrid approach to seismic vulnerability assessment?
Despite the different type of analysis used in each case, the hybrid component was used for both types of buildings and in both cases the key empirical parameter was the cost of repair of a damaged building; this is a particularly useful parameter, but reliable data is not always available on it, which means that other parameters (structural damage indices) could certainly be explored within the broader frame of the hybrid approach.
Q9. What type of buildings are considered to be low seismic?
Regarding the level of seismic design and detailing, four subclasses could be defined, as follows:- No code (or pre-code): R/C buildings with very low level of seismic design or no seismic design at all, and poor quality of detailing of critical elements; e.g. RC1MN (medium-rise, no code).-